still crossing the quality chasm: a mixed-methods study of

STILL CROSSING THE QUALITY CHASM: A MIXED-METHODS STUDY OF

PHYSICIAN DECISION-MAKING WHEN TREATING CHRONIC DISEASES

by

CHRISTOPHER C. LAMB

Submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Weatherhead School of Management

Designing Sustainable Systems

CASE WESTERN RESERVE UNIVERSITY

May, 2018

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Christopher C. Lamb

candidate for the degree of Doctor of Philosophy*.

Committee Chair

Kalle Lyytinen, Ph.D., Case Reserve Western University

Committee Member

Adrian Wolfberg, Ph.D., Case Western Reserve University

Committee Member

Yunmei Wang, Ph.D., Case Western Reserve University

Committee Member

J.B. Silvers, Ph.D., Case Western Reserve University

Date of Defense

January 16, 2018

*We also certify that written approval has been obtained

for any proprietary material contained therein

© Christopher C. Lamb, 2017

All rights reserved.

Dedication

I dedicate my dissertation to my family for their unwavering support over four

years and the many times I was away from home.

v

TABLE OF CONTENTS

Table of Contents .................................................................................................................v

List of Tables ..................................................................................................................... ix

List of Figures .................................................................................................................... xi

Abstract ............................................................................................................................ xiii

Chapter 1: Introduction ........................................................................................................1

Chapter 2: Literature Review ...............................................................................................5

Overview of U.S. Healthcare Quality and Cost .............................................................7

Evolution of the Relationship between Physicians and Patients: How We Came to

Shared Decision Making (SDM) ...................................................................................9 History of SDM......................................................................................................10

Facilitators of SDM................................................................................................14

Theories Related to Clinical Decision Making ............................................................15 Decision Theory .....................................................................................................16

Power .....................................................................................................................25 Traits ......................................................................................................................28 Organizational Context ..........................................................................................34

Summary of SDM Decision Theory ......................................................................42

Summary ......................................................................................................................42

Chapter 3: Chronic Diseases ..............................................................................................45

Hemophilia ...................................................................................................................46

Primary Immunodeficiency (PID) ...............................................................................49

Analysis #1: MEPS ......................................................................................................53

Analysis #2: IDF ..........................................................................................................56 Purpose ...................................................................................................................56 Methods..................................................................................................................56 Results ....................................................................................................................57 Discussion ..............................................................................................................57

Analysis #3...................................................................................................................57 SPARCS A: PID ....................................................................................................57 SPARCS B: Hemophilia ........................................................................................59

Analysis #4...................................................................................................................60 Truven Data: PID ...................................................................................................60 Truven Data: Hemophilia ......................................................................................63

Summary ......................................................................................................................64

Chapter 4: Research Framing.............................................................................................66

vi

Theoretical Framing .....................................................................................................66 What is known?......................................................................................................66 What We Do Not Know .........................................................................................67

A Tentative Theoretical Framework ......................................................................68

Research Purpose .........................................................................................................70

Research Plan ...............................................................................................................70

Research Synopsis and Question Development ...........................................................71

Research Design...........................................................................................................74

Integration of Results ...................................................................................................77 Philosophical Lens .................................................................................................79

Summary ......................................................................................................................79

Chapter 5: Study 1: Patient-centric Vs. Physician-driven Decision Models in the

Treatment of Hemophilia ...................................................................................................80

Methodology ................................................................................................................81

Sample..........................................................................................................................82

Data Collection ............................................................................................................83

Data Analysis ...............................................................................................................84

Findings........................................................................................................................85 Findings related to Decision Theory ......................................................................86

Findings related to Power Balance ........................................................................89

Findings related to Traits .......................................................................................92

Findings related to Organizational Context ...........................................................95

Results ..........................................................................................................................95

Discussion ....................................................................................................................98 Research Gap .......................................................................................................103 Implications..........................................................................................................104

Limitations and Future Research ...............................................................................105

Conclusion .................................................................................................................106

Chapter 6: Study 2: Does Physician’s Decision-Making STYLE Effect Patient

Participation in the Treatment Choices of Primary Immunodeficiency?.........................108

Research Gaps ............................................................................................................109

Design ........................................................................................................................110 Measurement Operationalization .........................................................................110

Constructs of Interest .................................................................................................113 Physician’s Decision-Making Process Effects on Patient Participation ..............117

Mediating Effects of Treatment Approaches to Patient Participation .................117 Physician Traits ....................................................................................................118

vii

Methods......................................................................................................................121 Sample..................................................................................................................121 Data Screening .....................................................................................................122

Multivariate Assumptions ....................................................................................123 Exploratory Factor Analysis ................................................................................123 Confirmatory Factor Analysis ..............................................................................124 Mediation .............................................................................................................126 Interaction (Moderation) ......................................................................................127

Multi-group ..........................................................................................................128 Controls ................................................................................................................131 Hypotheses Results ..............................................................................................131

Discussion ..................................................................................................................136

Summary of Major Findings ......................................................................................137 Expected Findings ................................................................................................139

Unexpected Findings ...........................................................................................140 Observations that Require Future Research .........................................................142

Limitations .................................................................................................................143

Future Research ...................................................................................................144 Ethical Assurances ...............................................................................................145

Conclusions and Implications for Patient Participation .............................................145

Chapter 7: Study 3: A Qualitative Study of Immunologists that Treat PID and the Factors

that Effect Shared Decision Making with their Patients ..................................................147

Post Hoc Reanalysis ...................................................................................................148

Post Hoc Methods ................................................................................................148 Post Hoc Findings ................................................................................................150 Summary of Post Hoc ..........................................................................................152

Research Method of the Qualitative Part of the Third Study.....................................153

Findings......................................................................................................................155

Findings Related to Dual Process ........................................................................156 Findings Related to SIT and Agency Theory ......................................................159 Findings Related to Traits ....................................................................................162 Findings Related to Organizational Context ........................................................164 Findings Related to Feedback ..............................................................................166

Cross-Referencing Findings from Post Hoc and QUAL Interviews .........................168

Discussion ..................................................................................................................170

Limitations, Implications, and Future Research ........................................................171

Chapter 8: Integrative Framework ...................................................................................173

Meta-inferences..........................................................................................................175 Rational Decision Making ...................................................................................176 Patient-Centrism ..................................................................................................176

viii

Traits ....................................................................................................................177 Bias/Nudging .......................................................................................................178 Organizational Context ........................................................................................178

Implications for Practitioners .....................................................................................179

Chapter 9: Discussion, Implications, Limitations, and Contributions .............................183

Theory ........................................................................................................................184

Practice .......................................................................................................................186

Other Research Domains ...........................................................................................187

Limitations .................................................................................................................187

Conclusion .................................................................................................................189

Appendix A: Study 1 Interview Protocol and Questions .................................................190

Appendix B: Invitation Template ....................................................................................194

Appendix C: Study 2 Constructs......................................................................................195

Appendix D: Post Hoc Constructs ...................................................................................197

Appendix E: Study 3 Interview Guide .............................................................................198

Appendix F: Mixed Methods Tutorial .............................................................................202

Research and Analysis Methods ................................................................................202

Mixed Methods ....................................................................................................202 Qualitative Methods .............................................................................................203

Quantitative Methods ...........................................................................................205

References ........................................................................................................................210

ix

List of Tables

Table 1. Theories of Factors that Predict SDM .................................................................. 7

Table 2. Physicians Responsibilities when Practicing SDM (Godolphin, 2009) ............... 9

Table 3. Role of Patients in the Competing Models of Healthcare Decision-Making ..... 12

Table 4. Summary of Biases ............................................................................................. 23

Table 5. Physician Characteristics that Influences the Decision Process ......................... 29

Table 6. MEPS 2014 Item Data Summary ........................................................................ 55

Table 7. Summary of PID Inpatient Admissions and Costs (2008–2010) ........................ 62

Table 8. Summary of Hemophilia Inpatient Admissions and Costs (2008-2010) ............ 64

Table 9. Summary: Quantifying the Quality Gap ............................................................. 65

Table 10. What We Know Concerning SDM Implementation ......................................... 67

Table 11. What We Do Not Know Concerning SDM Implementation ............................ 68

Table 12. Purpose and Description of Mixed Methods .................................................... 77

Table 13. Quality Standards Used for Each Method ........................................................ 78

Table 14. Study 1: Sample Details .................................................................................... 83

Table 15. Study 1 Codes ................................................................................................... 85

Table 16. First Study: Summary of Findings .................................................................... 97

Table 17. Second Study: Summary of Hypotheses and Results ..................................... 115

Table 18. Sample Characteristics .................................................................................... 122

Table 19. Reliability and Validity ................................................................................... 125

Table 20. Descriptive Statistics and Correlations Table of Variables ............................ 126

Table 21. Mediation ........................................................................................................ 127

Table 22. Interaction Effects ........................................................................................... 128

Table 23. Multi-Group Invariance Test by Pathway ...................................................... 128

Table 24. Invariance between White and Non-white Physician Groups ........................ 129

Table 25. Invariance between Ph.D. and Non-PhD Physician Groups ........................... 130

Table 26. Influence of Physician Years of Practice ........................................................ 131

Table 27. Study 2: Expected vs Unexpected Results ...................................................... 138

Table 28. Post Hoc Abbreviations .................................................................................. 149

Table 29. Pathway Values for Figure 13 ........................................................................ 150

Table 30. Indirect Effect: Mediation of Rational Decision Making through Patient-

centrism ........................................................................................................................... 151

Table 31. Indirect Effect: Mediation of Heuristic Decision Making through Patient-

centrism ........................................................................................................................... 151

x

Table 32. Influence of Race on Patient Participation ..................................................... 152

Table 33. Influence of Immunologist Sex....................................................................... 152

Table 34. Study 3: Sample Details .................................................................................. 154

Table 35. Cross-Referencing Study 3 ............................................................................. 169

Table 36. Strengths and Weaknesses of Qualitative Methods ........................................ 204

Table 37. Strengths and Weaknesses of Quantitative Methods ...................................... 206

Table 38. Model Fit Thresholds ...................................................................................... 208

xi

List of Figures

Figure 1. Healthcare Decision Making in Transition .......................................................... 8

Figure 2. ACO Measures of Patient Care ......................................................................... 13

Figure 3. CAHPS Items of SDM ...................................................................................... 13

Figure 4. Responses to Faces ............................................................................................ 31

Figure 5. Medicare Incentive Measures (CMS, 2016) ...................................................... 38

Figure 6. Chronic Care Model (Wagner et al., 2001) ....................................................... 39

Figure 7. Theoretical Framework ..................................................................................... 69

Figure 8. Theoretical Explanations for Predicting SDM .................................................. 71

Figure 9. Overview of Study Design and Results ............................................................. 76

Figure 10. Theory Exploration ........................................................................................ 103

Figure 11. Hypothesized Model ...................................................................................... 115

Figure 12. Second Study: Structural Equation Model Result ......................................... 138

Figure 13. Model of Post-Hoc Analysis of Second Study .............................................. 149

Figure 14. Theories that Predict SDM ............................................................................ 156

Figure 15. Theoretical Framework from Chapter 4 (Revised) ....................................... 174

Figure 16. Summary of Key Findings............................................................................. 185

xii

Acknowledgements

I wish to thank committee members, Dr. Yunmei Wang, Dr. Adrian Wolfberg,

and Dr. J.B. Silvers for agreeing to serve on my committee and who graciously shared

their insights and constructive criticism. I especially thank Dr. Kalle Lyytinen, committee

chair, for his mentorship throughout my doctoral program. I also thank Marilyn Chorman

and Sue Nartker for their incredible support. I also thank Ryan Dagenais for his dedicated

assistance. Lastly, I thank those parties and individuals, both known and anonymous, for

participating in and contributing to my research.

xiii

Still Crossing the Quality Chasm: A Mixed-Methods Study of Physician Decision-

Making When Treating Chronic Diseases

Abstract

by

CHRISTOPHER C. LAMB

Overall healthcare spending in the U.S. is in the trillions and more than 15% of GDP, yet

outcomes rank below the top 25 in most quality categories when compared to other

OECD countries. The majority of spending is directed toward small patient populations

with chronic diseases. Within the context of access to insurance coverage and a certain

level of health literacy, experts believe increased patient–physician shared decision

making (SDM) should result in better care and lower cost. However, the study of the

physician’s role in facilitating SDM is limited. By understanding what factors predict

when physicians will implement SDM during the treatment of specific chronic diseases,

we can begin to understand the dynamics that most influence behaviors and offer

recommendations to improve certain aspects of healthcare in the United States. A

sequence of three studies was completed by interviewing or surveying 369 physicians

who treat hemophilia and primary immune deficiency (PID). The study used dual process

theory to explain the relationship between patient-centered care and SDM within a wider

framework of power balance, patient/physician traits, and organizational context. These

studies were supplemented by an analysis of 1) survey of 33,162 individuals across the

xiv

U.S.; 2) 25 million hospitalization records from New York State comparing two (2) five-

year periods; and 3) data from 200 million individual-level, de-identified enrollment data

and health insurance claims across a continuum of care (both inpatient and outpatient).

The first study qualitatively explored decision making between hemophilia

physicians in the U.S. and U.K. and found U.S. physicians to be more patient-centric and

less rule-based. The second study quantitatively tested the relationship between

slow/rational vs. fast/intuitive decision making by U.S. physicians treating PID and SDM

as mediated by patient-centric care; results showed a statistically significant relationship

between slow/rational decision making and SDM. The third study analyzed decision-

making by U.S. immunologists when treating PID and found a stronger association

between rational/slow decision-making, patient-centered care, and SDM but the process

is bounded by anchoring bias, power balanced by health literacy alignment, and

organizational context related to time and coordination. Overall, results show that

rational/slow decision making as predicted by dual process theory is important to achieve

SDM for patients with complex (i.e., multiple comorbidities) chronic diseases. However,

the impact of health literacy, trust, and reimbursement is much different than what is

reported in the literature. In addition, the study found evidence that information is

anchored or framed in a way that both biases and “nudges” patients as predicted by

behavioral economists. A new integrated SDM model is put forth as a result of this

research.

Combining the physician survey/interview data with an analysis of individual

surveys, hospitalization data, and insurance claims demonstrates a significant opportunity

to improve the quality of care in the U.S. through better decision making.

xv

Keywords: physician decision making; shared decision making; dual process theory;

bounded shared decision making; patient alignment; primary immunodeficiency;

hemophilia; patient-centric approach; bias; decision theory; patient participation;

physician perspective; nudge bias

1

CHAPTER 1: INTRODUCTION

I think that some people with low health knowledge just accept whatever

the doctor says and they just do it. Whereas sometimes people that are very

literate challenge you, which I enjoy, I like the challenge. And then there's

people that are very smart but really don't understand the science of what

they're talking about, and they'll argue with you with things that make no

sense, so those are the most challenging I think. (Study 3; Participant 06-

RS)

Some people don't wanna know, and they, 'cause they don't want the burden

of decision making. You still turn it into a combined decision, or collective

decision, though, because you don't want them to feel is that they're totally

left out. You want them to be involved. (Study 3; Participant 04-TH)

Crossing the Quality Chasm: A New Health System for the 21st Century, a report

prepared by the Institute of Medicine (IOM), highlighted the urgent need to address

quality gaps in the U.S. (Richardson et al., 2001). These gaps include treatment of

chronic diseases which consume 86% of healthcare resources and are commonly

associated with reduced lifespan and co-morbidities (Anderson & Horvath, 2013; CDC,

2017). More pointedly, 5% of these patients consume half the costs (Eapen & Jain, 2017).

Despite spending more per capita, the U.S. ranks below the top 20 countries in most

quality categories (Berwick & Hackbarth, 2012; Keehan et al., 2011; Schneider, Sarnak,

Squires, Shah, & Doty, 2017). To close the quality chasm, the IOM report had two

recommendations: (1) set patient-centric goals for improving the U.S. healthcare system,

meaning that patient preferences and values should guide clinical decisions, and (2) target

chronic health issues to costs (Barry & Edgman-Levitan, 2012; Gerteis et al., 2014).

One method to address both recommendations—and the subject of this paper—is

the implementation of shared decision making (SDM). SDM is a defined term in the

medical community and it relates to a process whereby patients and their healthcare

2

providers come to conclusions about patient treatment jointly (Friesen-Storms, Bours,

van der Weijden, & Beurskens, 2015). It represents a significant opportunity to improve

outcomes and lower expenses (Barry & Edgman-Levitan, 2012; Lee & Emanuel, 2013).

Physicians who incorporate SDM into their decision-making process have the potential to

provide more accurate diagnoses, improved care and better patient-perceived health

(Chambers et al., 2016; Murray, Pollack, White, & Lo, 2007; Seeborg et al., 2015).

Chronic diseases offer an excellent model to study SDM because the complex and

challenging dimensions of the disease require difficult and ongoing treatment decisions

(Abolhassani et al., 2013). Chronic diseases are those that last more than a year and

whose complexity includes multiple co-morbidities which interact in difficult to predict

and emergent ways (Parekh & Barton, 2010). However, despite several developed and

defined SDM models, implementation of SDM with chronic care is limited and little is

known about how it can serve as a useful and practical way for doctors and patients to

interact (Charles, Gafni, & Whelan, 1999; Couët et al., 2015; Légaré et al., 2014;

Stevenson, Barry, Britten, Barber, & Bradley, 2000). Some models describe the factors

that influence SDM in chronic care treatment (Légaré et al., 2013). However, these

factors have not been examined and tested within the context of specific chronic diseases

(Noonan et al., 2017). Senior U.S. policymakers have advocated for more research to fill

knowledge gaps about interventions that will benefit people with multiple chronic

conditions (Parekh, Goodman, Gordon, Koh, & Conditions, 2011; U.S. Department of

Health Human Services, 2010).

A sequence of three studies involving 369 physicians was completed to address

this knowledge gap and identify which factors most influence decision-making when

3

treating hemophilia and primary immune deficiency (PID). Hemophilia is a rare clotting

disorder that affects 10,000 people in the U.S. and is considered one of the top five most

expensive diseases to treat (Brooks, 2017). PID is a rare genetic immunological disorder

with an incidence of 32,000 diagnosed patients, although the prevalence is thought to be

much greater (Bousfiha et al., 2013; Kobrynski, Powell, & Bowen, 2014). It too is one of

the most expensive diseases to treat due to preventable hospitalizations (Gardulf et al.,

2017; Menzin, Sussman, Munsell, & Zbrozek, 2014). Hemophilia and PID represent part

of the 50% of U.S. healthcare spending driven by 5% of the population. Both are good

examples of complex chronic diseases where there is the most opportunity to improve

cost-effective care.

These studies were supplemented by an analysis of 1) survey of 33,162

individuals across the U.S.; 2) 1,425 U.S. PID patients; 3) 25 million hospitalization

records from New York State comparing two five-year periods; and 4) data from 200

million individual-level, de-identified enrollment data and health insurance claims across

a continuum of care (both inpatient and outpatient). These studies help quantify the

quality chasm and the associated cost implications.

There is no published research demonstrating an integrated model that predicts

successful implementation of SDM for chronic disorders and the potential impact on

quality of care. Therefore, theory and data were needed to understand the unique

perspectives of specific chronic diseases and the relationships between factors that

encourage physicians to use SDM in their routine practice (Alston et al., 2012). The

results from this research identified key influences which encourage and discourage

physicians to adopt SDM and should support recommendations on how these influences

4

can improve care and reduce cost. Primary influences include rational/slow decision

making and bias/nudging as predicted by dual process theory, power imbalance mitigated

by patient-centric care and health literacy, as well as organizational context related to

time and coordination of care. Recommendations include physician training on cognitive

phycology, better use of decision aids and improved systems to coordinate care between

physicians. These results can be applied more generally to other chronic disorders.

The remainder of this dissertation is organized as follows. Chapter 2 contains a

review of published literature on the clinical decision-making process. Chapter 3 presents

an analysis of patient survey data and hospitalization records to better understand the

quality gap. Chapter 4 presents the research design and purpose of the three studies.

Chapters 5–7 provide a detailed account of each study and the findings. Chapter 8

integrates the results of the analysis and three studies and discusses the implications and

contributions of this research. Finally, Chapter 9 summarizes the limitations of this

research and potential directions for future studies.

5

CHAPTER 2: LITERATURE REVIEW

There is much commentary about what ails American healthcare. In 1985, Garrett

Hardin famously wrote about the tragedy of the commons: an economic theory whereby

within a shared-resource system individual users acting independently according to their

own self-interest behave contrary to the common good of all users by depleting that

resource through their collective action (Hardin, 1985). Hardin bemoans the “U.S.

medical commons” and the “trend towards commonizing medical costs.” However,

Hardin got it wrong on two points: first, he inaccurately asserts that socializing medicine

will diminish containment of cost; and second, his notion that the psychology of life-

saving emergency acute care inexorably encourages extravagance. The fact is that

socialized single payer systems are far more cost-effective and chronic care—not acute

care—is what is driving exorbitant costs (Coghlan, 2017; Squires & Anderson, 2015).

Others more recently attribute America’s healthcare problems to a lack of access

to basic care and claim that universal coverage will eliminate health inequalities and

improve life expectancy (Ansell, 2017). Solutions include “Medicare for All” which

adherents assert will result in a simplified system that standardizes care (Fram & Freking,

2017). However, universal coverage does not address the issue that most health care is

chronic and complex. While getting more young and healthy people to buy into or be part

of an insurance pool may lower average costs, it will not address the fundamental

problem that small patient populations with chronic diseases are driving exorbitant

expenses and poor outcomes.

Too often the healthcare debate in the U.S. is about financing (public/private or

single vs. multi-payer systems) or access to coverage. However, chronic care is the real

6

problem and the U.S. has the highest percent of complex chronic care patients, meaning

those with multiple comorbidities, amongst its international peers.1

This problem is only

getting worse as baby boomers age and must be urgently addressed before costs

overwhelm business competitiveness and government expenditures. As quoted by Warren

Buffet, "The ballooning costs of healthcare act as a hungry tapeworm on the American

economy" (Hiltzik, 2018). Recent increases in overall healthcare expenditure were almost

entirely driven by increases in multiple chronic care prevalence and the high cost of

caring for these very complex patients (Gerteis et al., 2014). These increases are driven

by healthcare services prices and intensity (Dieleman et al., 2017). Shared Decision

Making (SDM) is an approach to treatment which may improve cost-effective care.

Models and confirmatory data are needed to better understand the circumstances under

which SDM can be optimally implemented.

This chapter reviews the literature related to the research question: what predicts

physician implementation of SDM. First, there is an overview of the U.S. healthcare

system and the evolution of the relationship between physicians and patients. Second,

there is an overview of clinical decision-making and theories which attempt to explain

SDM (Table 1). Using Davis’ criteria, the list of relevant theories in Table 1 were chosen

not only because they are “interesting” but because they are often cited by significant

1

Long term disorders are the main challenge facing health care systems worldwide. For example, 70% of

health service spending in the U.K. is on long term conditions Barnett, K., Mercer, S. W., Norbury, M.,

Watt, G., Wyke, S., & Guthrie, B. 2012. Epidemiology of multimorbidity and implications for health care,

research, and medical education: a cross-sectional study. The Lancet, 380(9836): 37-43, UKHCDO. 2012.

UKHCDO annual report 2012 & bleeding disorder statistics for the financial year 2011/2012.. In 2014,

68% of the U.S. population over the age of 65 have two or more chronic conditions Squires, D., &

Anderson, C. 2015. US health care from a global perspective: spending, use of services, prices, and health

in 13 countries. The Commonwealth Fund, 15: 1-16..

7

researchers in the field and there is recent scholarly work with supporting data (Davis,

1971).

Table 1. Theories of Factors that Predict SDM

Research Question Theory Explanation

What factors predict

SDM?

Dual Process Rational/slow thinking, bias, and nudging

Social Identity

Theory/Principal Agency

Dilemma

Power balance and information asymmetry

Trait Demographics, experience and trust

Organization Policies, rules, feedback, reimbursement,

and coordination

After a review of theory, there is a summary of two chronic diseases that are good

examples to examine SDM: hemophilia and PID.

Overview of U.S. Healthcare Quality and Cost

The motivation for this research is the suboptimal quality of care to cost ratio in

the U.S., especially when compared to countries of similar socioeconomic status

(Berwick & Hackbarth, 2012; Keehan et al., 2011). In 2015, overall healthcare spending

in the U.S. was $3.2 trillion, or 17.8% of the $17.9 trillion GDP.2

The costs are expected

to increase to 19.9% by 2025 with a significant percentage (37%, or $1.18 trillion) from

government expenditures related to chronic care (CMS, 2015, 2017). The U.S. healthcare

2

2015 is the last reported year of actual data on healthcare spending in the U.S. The vast majority (86% or

~$2.7 trillion) of U.S. healthcare spending is directed toward treating small patient populations with

chronic diseases but only 56% of chronic patients in the U.S. receive the correct or recommended care for

their illness CDC. 2017. Center for disease control and prevention, Vol. 2017: CDC, Gerteis, J., Izrael, D.,

Deitz, D., LeRoy, L., Ricciardi, R., Miller, T., & Basu, J. 2014. Multiple chronic conditions chartbook.

Rockville, MD: Agency for Healthcare Research and Quality (AHRQ) Publications, McGlynn, E. A., Asch,

S. M., Adams, J., Keesey, J., Hicks, J., DeCristofaro, A., & Kerr, E. A. 2003. The quality of health care

delivered to adults in the United States. New England journal of medicine, 348(26): 2635-2645..

8

system was ranked 37th of 190 countries according to the WHO in 2000 and dropped to

50th of 55 according to Bloomberg (Du & Lu, 2016; WHO, 2000). For most health

outcomes the U.S. does not rank in the top 35 countries (OECD, 2015).

In the early 20th century, public health advances successfully addressed infectious

diseases and poor nutrition which resulted in improved life expectancy. Subsequent

scientific discoveries and technological advances have made acute illness, such as heart

attacks and strokes, survivable. Now in the 21st century, chronic disease has become the

most important challenge (Gerteis et al., 2014).

Figure 1. Healthcare Decision Making in Transition

(Source: Toro, 2012)

One solution to advance chronic care—and the healthcare system as a whole—is

to implement shared decision making (SDM). The potential for SDM to improve difficult

clinical situations is well documented and demonstrated (Lee & Emanuel, 2013;

O'Connor, Llewellyn-Thomas, & Flood, 2004; O’Connor et al., 2007; Schoen et al.,

2007; Veroff, Marr, & Wennberg, 2013). For example, SDM with decision aids helped

reduce the cost of knee replacement surgeries by 20% through better patient education

(Arterburn et al., 2012; Oshima Lee & Emanuel, 2013). However, despite documented

evidence of success, it is unclear what predicts physician implementation of SDM.

9

Evolution of the Relationship between Physicians and Patients: How We Came to

Shared Decision Making (SDM)

SDM is defined as an “approach where clinicians and patients share the best

available evidence when faced with the task of making decisions, and where patients are

supported to consider options, to achieve informed preferences” (Elwyn et al., 2010). It is

the process of integrating patient goals and concerns with medical evidence to achieve

high-quality decisions (Alston et al., 2012; Friesen-Storms et al., 2015). Godolphin

(2009) summarizes the steps necessary when practicing SDM (Table 2).

Table 2. Physicians Responsibilities when Practicing SDM (Godolphin, 2009)

Step Description

1 Develop a partnership with a patient.

2 Establish or review patient preferences for information.

3 Establish patient’s role in decision-making and the existence of any uncertainty about

the course of action to take.

4 Discuss the patient’s ideas, concerns, and expectations.

5 Identify choices and evaluate the research evidence in relation to the individual patient.

6

Direct the patient to evidence, taking into account points 2 and 3, above, framing

effects. Help the patient to reflect upon and assess the impact of alternative decisions

with regard to his or her values and lifestyles.

7 Negotiate a decision in partnership and resolve conflict.

8 Agree upon an action plan and complete arrangements for follow up.

The term “shared” implies joint decision making, but who decides the final

treatment decision? Physician decisions are related to medical diagnosis and an

understanding of the unique characteristics of each patient (Légaré & Witteman, 2013),

especially in cases of great uncertainty (Sox, Higgins, & Owens, 2013). Patients are fully

capable of making final decisions due to a combination of physician endorsement,

uncertainty, and contextual factors (Frosch & Kaplan, 1999; Hunink et al., 2014; Mulley,

10

Trimble, & Elwyn, 2012; Samaan et al., 2014). However, some patients want to be

involved without making the final treatment decision, whereas other may not want any

involvement (Say & Thomson, 2003). SDM, therefore, is not about making shared

decisions but rather integrating patient-physician input into a mutually agreed upon plan

that optimizes values, preferences, cost, and outcomes.

History of SDM

Physician decision-making has been described as informed, paternalistic, or

shared (Friesen-Storms et al., 2015). These should be seen as a continuum of how much

influence is exerted by a patient or physician. Patients range from passive to active

participants in the decision-making process; meaning some patients prefer physicians to

paternalistically make decisions while others demand a decisive role (Kaba &

Sooriakumaran, 2007). Research confirms that, while there are some exceptions, there is

a clear preference on the part of patients for SDM (Kiesler & Auerbach, 2006; Légaré &

Thompson-Leduc, 2014).

Informed decision making is when physicians simply provide information that

allows patients to make all decisions. A famous example is Steve Jobs, Apple CEO, who

insisted on nine months of natural therapy to treat his cancer; unfortunately, many believe

this decision delayed effective therapy and led to a premature death (Isaacson, 2011).

Informed decision making is not applicable to chronic disorders due to disease

complexity and uncertainty of care (Friesen-Storms et al., 2015).

Prior to 2000, paternalism was the dominant decision-making approach

(Rothstein, 2014). Patients were provided minimal information about their condition and

expected to receive and follow orders (Rothstein, 2014). The introduction of “evidence-

11

based medicine” (EBM) as coined by Gordon Guyatt in 1991 was an attempt to

modernize the paternalistic model (Bensing, 2000; Djulbegovic & Guyatt, 2017; Guyatt,

1991). Evidence-based medicine (EBM) “is the conscientious, explicit and judicious use

of current best evidence in making decisions about the care of individual patients”

(Sackett, Rosenberg, Gray, Haynes, & Richardson, 1996: 71). It is a “bottom-up approach

that integrates the best external evidence with individual clinical expertise” (Rousseau,

2012: 38). The practice of EBM was part of a movement to standardize care and reduce

variability in treatment approaches used by different physician specialties (Charles et al.,

1999; Timmermans & Berg, 2010).

After 2000, there was a shift to a patient-centered decision-making paradigm

otherwise known as SDM (Charles et al., 1999; O'Hare, Rodriguez, & Bowling, 2016).

As described in

Table 3, this shift was caused by the limitations of EBM and the recognition that

there may not be a “best” decision but rather an optimal decision based on patient values

and goals (Sturmberg & Martin, 2013: 47-49). Unlike many acute conditions like the flu

or a broken bone, chronic diseases are usually not “cured” but managed over the course

of a person’s life. Chronic disease management is complex and unpredictable and the

physician-patient interplay is crucial. Complexity results from unstable cases where

standard medical techniques cannot be applied and the “process of inquiry” is most

important: listening and exploring possible explanations (Schon, 1983). This approach is

consistent with the theory of Adaptive Health Practice, derived from Adaptive

Leadership, which recommends that physicians be mindful of not “substituting technical

12

interventions for adaptive work” with patients (Heifetz, 1994; Sturmberg & Martin, 2013:

665; Thygeson, Morrissey, & Ulstad, 2010).

Table 3. Role of Patients in the Competing Models of Healthcare Decision-Making

Paternalistic SDM

Objective Obtain relief from ailment. Make choices to alter future

probabilities of wellbeing

Information regarding (a)

medical status and (b)

medical therapy

(a) High but imperfect

(b) Low and imperfect

(a) High but imperfect

(b) High and imperfect

Capabilities Limited to observation and

feeling. Process treatment

information if packaged

appropriately.

Significant within therapies

Complex cognitive structures

involving therapy attributes,

risk, and efficacy

Preferences Not easily accessible.

Require physician effort.

Largely homogeneous within

therapies.

High motivation to express

preferences.

Heterogeneous within

therapies.

Role in physician interactions Provide information Provide and seek information

Role in decision making Passive but involved Active and engaged

Timing Crisis-induced Need- and desire-based

Motivation Prolong life, pain-free if

possible

Superior quality, not

necessarily quantity, of life

Note: Reproduced from “Toward Understanding Consumers' Role in Medical Decisions for Emerging

Treatments: Issues, Framework, and Hypotheses,” by J. Singh, L. Cuttler, and J. Silvers, 2004, Journal of

Business Research, 57(9), p. 1058. (Singh et al., 2004).

Research and official U.S. policy, enacted in the 2010 Patient Protection and

Affordable Care Act, state that SDM is best applied to chronic health issues (Friesen-

Storms et al., 2015; Frosch et al., 2011). Accountable Care Organizations (ACOs),

integrated payment and care delivery systems used by Medicare, Medicaid, and

commercial payers have developed sets of quality measures and associated financial

incentives and penalties tied to the use of SDM. (See Figure 2.) In addition, the SDM

consumer assessment of healthcare providers and systems (CAHPS) survey is listed

13

under the “Patient/Caregiver Experience” category (ACO-6) (CMS, 2016). (See Figure

3.)

Figure 2. ACO Measures of Patient Care

(Source: CMS, 2016)

Figure 3. CAHPS Items of SDM

(Source: CAHPS, 2012)

14

While healthcare systems are rated by patients according to their success implementing

SDM, the ratings are not linked to individual physicians. Therefore, it is very difficult to

assess physician performance and SDM.

Facilitators of SDM

By understanding what influences implementation of SDM, optimal decision-

making strategies or designs can be applied to complex chronic diseases where it is most

needed. However, despite reports suggesting the U.S. healthcare system excels at SDM,

attempts to implement and coordinate SDM into practice are scarce and poorly

understood (Desroches, 2010; Schneider et al., 2017).

Facilitators of SDM are thought to be education, appropriate training on SDM,

and improved interpersonal skills with patients (Belanger, Rodríguez, & Groleau, 2011;

Légaré et al., 2013). Barriers to SDM are thought to be conflicting practitioner roles

within healthcare systems, power imbalances between patients and practitioners,

inadequate resources, physician payment structure (quantity vs. outcomes), and time

constraints (Belanger et al., 2011; Légaré et al., 2013). Other considerations include bias

as predicted by cognitive psychologists and physicians’ relationships with stakeholders

within the larger healthcare system (Chapman, Kaatz, & Carnes, 2013; Croskerry, 2014).

The factors influencing patient participation—and SDM by extension—have been

explored on a general level, but have not been explicitly studied in the context of specific

chronic health issues (Longtin et al., 2010). Scales have been developed to measure

whether SDM is preferred by patients but do not capture the factors that predict SDM

(Giguère, Labrecque, Njoya, Thivierge, & Légaré, 2012). SDM has been measured by the

levels of perceived comfort, pain management, perceived anxiety levels, and involvement

15

of family and friends (Barry & Edgman-Levitan, 2012); but again, not the factors that

predict these perceptions. The issue is not whether SDM can be implemented but what

facilitates or predicts implementation.

As a caveat, the suspected or known consequences of SDM are also poorly

described in the literature. For example, the U.S. has been in the midst of a prescription

opioid epidemic since 2000 (Baker, 2017). The epidemic was an unforeseen consequence

of a patient-oriented approach to pain management (Manchikanti, Helm, & Janata, 2012).

The overzealous prescribing of opioids was the result of a self-reported pain scales using

standards set by the Joint Commission (Baker, 2017). The goal is to find a balance in the

decision-making process—using SDM—to provide optimal care; in other words, to avoid

under-treating and over-treating patients.

In summary, SDM is an approach that adopts the advantages of paternal and

informed approaches to care. The interpersonal aspect of this approach is best utilized for

the treatment of chronic diseases which are complex and not curable. Despite perceived

benefits, factors that predict implementation have not been adequately studied or

documented in published literature. Exploring clinical decision-making theory may

provide insight into specific factors that may facilitate or discourage SDM by physicians.

Theories Related to Clinical Decision Making

Clinical decision making is a “contextual, continuous, and evolving process,

where data are gathered, interpreted and evaluated to select the optimal choice of action”

(Tiffen, Corbridge, & Slimmer, 2014). Dual process theory has been used to explain

clinical decision making. Dual process theory is the latest iteration in a long history on

decision theory; however, it may only a partial explanation. Therefore, the following

16

sections review the literature on decision theory, the movement from classical to dual

process theory, as well as alternative theories that attempt to predict implementation of

SDM.

Decision Theory

Decision theory is the study of strategies for optimal decision-making. The best

application of decision theory to the medical community is an ongoing debate: algorithms

vs. personalized individual interviews. Some argue against the use of algorithm-based

diagnoses over traditional physician-patient in-depth interviews—specifically, recipe-like

approaches that may hinder physicians’ creativity and flexibility, often overlooking

subtle clues that can lead to the correct diagnosis (Groopman, 2008). Given that 75% of

chronic care clinical diagnostic failures are likely attributed to errors in physician

thinking, a better understanding of the medical decision-making process is necessary

(Graber, Franklin, & Gordon, 2005). Unfortunately, various studies have analyzed

characteristics of decision-making styles in a variety of settings but with mixed results

(Calder et al., 2011; Coget & Keller, 2010; Kaplan, Greenfield, Gandek, Rogers, & Ware,

1996). Further study is needed to extend decision theory into chronic care treatment

particularly since it is dominated by a mix of uncertainty, evidence-based

diagnosis/treatment and patient values (Groopman, 2008; Montgomery, 2005). This

extension of decision theory could then serve as a template for use by the medical

community for optimal SDM implementation (Croskerry, 2009a).

Classical Decision Theory

Classical decision theory is “the collection of axiomatic models of uncertainty and

risk (probability theory, including Bayesian theory), and utility (utility theory, including

17

multi-attribute utility theory)” (Beach & Lipshitz, 2017: 85). It is a method of formalizing

the alternative decision options based on existing conditions (North, 1968). However,

classical decision theory depends on two elements: an ideal decision setting and a

decision maker who can make an optimal choice after carefully evaluating all options and

possible outcomes (Beach & Lipshitz, 2017). Chronic care does not have either of these

elements; the healthcare system works best for patients with single conditions

(Grembowski et al., 2014).

Dual Process Theory for Decision Making

Dual process theory, otherwise known as prospect theory, developed from a long

history of thinkers who challenged classical decision theory (Barrett, Tugade, & Engle,

2004; Kahneman, 2011; Tversky & Kahneman, 1992). It explains how people choose

from among sub-optimal decision options based on context (Tversky & Kahneman,

1986b) and weighing costs and benefits depending on pre-existing preferences and biases

(Scott, 2000).

Dual process theory integrates two processes of thinking that can be applied to

clinical diagnoses and decisions (Croskerry, 2009b; Evans & Stanovich, 2013; Pelaccia,

Tardif, Triby, & Charlin, 2011). It categorizes how physicians think into two modes:

heuristic thinking (“System 1”) and rational thinking (“System 2”) (Djulbegovic, Hozo,

Beckstead, Tsalatsanis, & Pauker, 2012; Durning et al., 2015; Stark & Fins, 2014). The

concept of System 1 and System 2 is a modern, more robust version of Dewey’s type 1

and type 2 thinking. Dewey (1925) described “lived” experience (type 1), which is “the

result of a minimum of incidental reflection,” and “secondary” experience (type 2), which

is the result of “continued and regulated reflective inquiry” (Dewey, 1925). Simon

18

rejected Dewey’s perfect rationality and developed the concept of bounded rationality,

which takes into account environmental complexity, cognitive limits and the task-at-

hand; people satisfice by choosing what is sufficient. Bounded rationality explains how

satisficing decisions are made in the absence of information, time, and resources (Simon,

1996). Gigerenzer extended Simon’s concept of “fast and frugal” heuristics; humans

apply algorithms to achieve near-optimal decisions (Gigerenzer & Goldstein, 1996).

Heuristic decision making (System 1) is an approach that relies on experience

(Croskerry, 2009b; Kahneman, 2011). Heuristics are “strategies that ignore information

to make decisions faster, more frugally, and more accurately than more complex

methods” (Gigerenzer, 2015: 109). Physicians simplify information by forming standard

approaches to treatment-based clinical experience, otherwise known as “illness scripts”

(Borrell-Carrió, Estany, Platt, & MoralesHidalgo, 2014; Campbell, 2013). Furthermore,

physicians use “intuition” or mental cues and primers to access information from their

memory (Kahneman, 2011) However, intuition is nothing more or less than recognition

(Simon, 1996). In short, heuristic decisions are mental shortcuts based on experiences,

irrespective of supporting data (intuitive reasoning). While the ability to identify patterns

quickly may have benefits in certain settings, heuristics or fast thinking by physicians

carries certain risks, including misinterpreting conceptual relationships and introducing

bias into decision making (Campbell, 2013; Ely, Kaldjian, & D'Alessandro, 2012;

Lovallo & Sibony, 2010).

Experts have a long-standing debate between expert intuition and algorithms.

Some argue that algorithms outclass expert intuition by removing the bias and

subjectivity that comes with expert opinions, especially when involving uncertainty

19

(Kahneman, 2011). Others favor experts because algorithms cannot be applied to

dynamic, flexible real-world instances (Kahneman, 2011). Overall, algorithms may be

better than human reasoning because people rarely detect subtle valid cues in noisy

environments (Kahneman, 2011). However, expert intuition may have an advantage

when applied to a predictable environment through prolonged practice; a “less-is-more”

effect of intuitive thinking for redundant and predictable tasks (Gigerenzer, 2015). In

either case, expert intuition or heuristics, have the same risks and biases associated with

System 1 thinking (Campbell, 2013; Ely et al., 2012; Lovallo & Sibony, 2010).

Rational decision-making (System 2) is a slow and effortful process of problem-

solving by conscious analysis (Durning et al., 2015; Kahneman, 2011). This style of

thinking requires physicians to meticulously consider available options and variables

(Campbell, 2013; Croskerry, 2009b). Physicians may analyze different factors, such as

the ratio of harm to benefits, especially when there is no clear or standard procedure

given a unique or complex patient circumstances (Djulbegovic et al., 2012). Dual process

theory suggests that System 2 thinking is better able to integrate ambiguous data,

literature, and statistical algorithms (Kahneman, 2011).

Uncertainty

The presence of uncertainty plays an overarching role in the physician decision

process. Uncertainty refers to the risk that is incurred as a result of lack of definitive

answers or solutions to a problem (Jones, 1992). Physicians are usually confronted with

three types of uncertainty when making clinical decisions: limitations of medical

knowledge, the physician’s perception of the gaps in his or her medical knowledge, and

the tolerance of uncertainty (Jones, 1992). The limitations of a physician’s medical

20

knowledge are often regarded as one of the strongest influences on the decision-making

processes because of the need to use clinical skills and judgment despite incomplete

empirical support (Flynn, 2003). In some cases, physicians may have experience that

contradicts research evidence, thereby increasing their uncertainty (Timmermans & Berg,

2010: 163). Physicians have to cope with the reality that there will be gaps in their

knowledge and competencies, underscoring the importance of remaining updated and

properly trained (Flynn, 2003). Furthermore, physician tolerance to uncertainty could

affect how they treat patients; high tolerance to uncertainty may increase a physician’s

willingness to deviate from standard protocols to accommodate the patient’s lifestyle and

preferences (Flynn, 2003). Due to multiple co-morbidities, uncertainty is inherent to

chronic care and requires complex management, decision making and coordination

(Whitson & Boyd, 2016). Dual process theory suggests that greater uncertainty requires

System 2 thinking.

Bias

Physician decision making is susceptible to biases listed in Table 4 (Silvers,

Marinova, Mercer, Connors, & Cuttler, 2010). Bias consists of flawed evaluations of

initial information. For example, a physician may treat a new patient with the same

methods and drugs as previous patients because of similar symptoms and mentioned

insurance coverage (anchoring, availability, money-priming, and status quo bias).

Training as a resident may drastically affect how physicians think and potentially

reinforce bias: initial exposure to real patients is often the link between knowledge and

experience (Patel, Kaufman, & Arocha, 2002; Patel, Yoskowitz, & Arocha, 2009). One

21

method to combat bias is for physicians to practice meta-cognitive thinking by analyzing

how they reach conclusions (Klein, 2005).

Perhaps the most important bias related to SDM is how word framing can alter

decision making. For example, suppose patients have to decide whether to have a

hypothetical drug administered. If they are told, “This drug has a 95% survival rate,”

people are more likely to agree than when told, “This drug has a 5% death rate.” Words

have power and verbal primes impact decision making (Sapolsky, 2017: 93).

Recent studies suggest biological support for word framing bias. When patients

receive medication with a positive expectation, there is an increase in activity of the brain

area known to be involved in pain relief (regardless of whether the pain relief is from a

narcotic or placebo). Alternatively, when medication is delivered with a negative

expectation, there is reduced brain activity in this area (Ofri, 2017: 80).

Libertarian paternalism is a method of encouraging individuals to make choices

which are in their best interests while maintaining their freedom of choice (Aggarwal,

Davies, & Sullivan, 2014; Thaler & Sunstein, 2003). Physicians usually have an opinion

on the best treatment for their patients and are prone to “frame” information to nudge the

decision process in favor with aligning to their decisions (Aggarwal et al., 2014).

However, healthcare involves an immense knowledge set that patients do not normally

have access to, nor do they fully comprehend all the information (Walls, 2014).

Therefore, libertarian paternalism’s goal is to supplement the patient decision towards

effective treatment (Schiavone, De Anna, Mameli, Rebba, & Boniolo, 2014).

Both biological and behavioral theories support the hypothesis that how doctors

frame a treatment has profound effects on how patient receives, interprets and

22

experiences that treatment. One method to combat bias is for the physician to practice

meta-cognitive “slow” thinking, or to study how they reach conclusions (Klein, 2005).

However, we should consider when a bias is a negative attribute and when it is a “nudge”

which may serve as encouragement for advantageous decisions (Sunstein & Thaler,

2008).

In summary dual process theory can be a useful method for understanding

decision making and the potential for predicting SDM with chronic care. However, it is

likely that dual process theory only partially explains how clinical decisions are made.

The next section reviews alternative theories that have the potential to predict when SDM

will be implemented.

23

Table 4. Summary of Biases

Bias Definition/Description Source

Anchoring

When one piece of information is heavily relied upon. It involves forming a starting point for

initial approximations and then adjusting according to additional information. However, the

adjustments are often insufficient, especially in the presence of new significant information.

(Eraker & Politser, 1982;

Sunstein & Thaler, 2008;

Tversky & Kahneman, 1986a)

Availability bias

The tendency to predict the likelihood of an event according to the vividness and ease of which

relevant instances can be recalled. Recollection does not guarantee accurate information,

therefore judgments could be influenced by exposure to irrelevant or insignificant sources.

More experienced physicians may be more susceptible to availability bias than less experienced

physicians.

(Eraker & Politser, 1982;

Mamede et al., 2010; Sunstein

& Thaler, 2008; Tversky &

Kahneman, 1973)

Certainty effect Decisions involving “certain” options are overweighed. (Eraker & Politser, 1982;

Stanovich & West, 2008)

Confirmatory bias The tendency to look for, notice, and remember information that fits with our preexisting

expectations by selectively accepting and ignoring information. (Groopman, 2008; Klein, 2005)

Expected utility The value when calculating the risk vs. reward. Situations with similar options change based on

the context (down-payment vs. loan shark example).

(Stanovich, 2011) (Stanovich &

West, 2000)

Feelings

As long as the situation is one that is either interesting or important to the decision maker,

positive affect facilitates systematic, careful, cognitive processing, tending to make it both more

efficient and more thorough, as well as more flexible and innovative

(Isen, 2001)

Focusing illusion Overestimate the negative effects and impact of an illness and neglect the natural resilience and

numerous positives in our lives. (Groopman & Hartzband, 2012)

Framing Decisions are influenced by how they are presented.

(McNeil, Pauker, Sox Jr, &

Tversky, 1982; Stanovich &

West, 2000)

Herd mentality People are heavily influenced by the actions and opinions of others (Sunstein & Thaler, 2008)

Illusory correlation To perceive two events as causally related, when in fact the connection between them is

coincidental or even non-existent (Klein, 2005)

24

Money-priming

First, people reminded of money are less interpersonally attuned; They eschew

interdependence. Second, people reminded of money shift to professional, business, and work

mentality. They exert effort on challenging tasks, demonstrate good performance, and feel

efficacious. (see “Priming” below)

(Vohs, 2015)

Outcome bias Judging a decision with a positive outcome as superior than a decision which causes a negative

out. (Stanovich & West, 2000)

Overconfidence

Excessive self-assurance which leads to the overestimation of personal abilities and limitations,

which can to a disinterest in decision support or problems gathering accurate information.

Characteristics of overconfidence include attitudinal and cognitive aspects.

(Berner & Graber, 2008; Ely et

al., 2012; Klein, 2005; Sunstein

& Thaler, 2008)

Priming

When someone is exposed to a factor their mind uses for identification. Often involved with

System 1 thinking, priming causes the brain to recall relevant information depending on the

trigger(s) involved.

(Sunstein & Thaler, 2008)

Representativeness

heuristic

Where people judge the probability or frequency of a hypothesis by considering how much the

hypothesis resembles available data. (Sunstein & Thaler, 2008)

Status quo bias

The tendency to continue a course of action regardless of the advantages associated with

change. More options increases status quo bias. Includes the endowment effect and loss

aversion.

(Kahneman, Knetsch, & Thaler,

1991; Sunstein & Thaler, 2008)

25

Power

Power imbalance between a physician and a patient is thought to be an important

influence on clinical decision making (Topol, 2015). Social Identity Theory (SIT)

predicts that the level of imbalance is important and patient advocates claim that health

literacy is the way imbalances can be mitigated (Wald, 2015).

SIT suggests that people tend to categorize themselves and others depending on a

collection of individual roles (Ashforth & Mael, 1989). Defining social roles as part of a

performance to reinforce status and identity has been long recognized in the medical

community when, for example, doctors use an elaborate personal front such as white

coats or stages such as large desks (Goffman, 1959). Power dynamics are a fundamental

aspect of human relationships and the physician-patient dyad is no exception (Fiske,

Dupree, Nicolas, & Swencionis, 2016; Mirowsky, 2017). According to SIT, physicians

are likely to categorize themselves and colleagues as highly educated when compared to

patients. Asymmetries of information may explain why physicians are slow to adopt

SDM (Tapscott, 2010: 195). Their interactions with patients, who are likely in a different

category, affect the physician decision process depending on the categories associated

with the patient. These categories, or identities, lead to a power imbalance in the

physician-patient relationship and likely a reduction in SDM.

While the balance of power in the physician-patient relationship can affect the

overall quality of care, some believe that physicians can take measures to increase patient

empowerment which will lead to improved patient outcomes (Bravo et al., 2015). These

interventions include a patient-centered approach to care, counseling, coaching, and

addressing patient values; all of these actions are represented in SDM (Bravo et al.,

26

2015). Experienced clinicians are usually aware of power dynamics and often serve as

“guides” to patients (Hibbard, 2017; McCormack, Thomas, Lewis, & Rudd, 2017).

Nevertheless, it is clear that clinicians often use authority to “nudge” patients to do what

is best (Sunstein & Thaler, 2008).

Health literacy is a method proposed by health advocates to reduce the power

imbalance between a physician and a patient. Patient education and understanding of

medical information (“health literacy”) plays a role in the patient’s capability to make

decisions with the physicians. Information asymmetry, as described by the “agency

dilemma,” deals with the study of decisions where one party has more or better

information than the other and creates an imbalance of power (Eisenhardt, 1989).

Information asymmetries have been studied in the context of principal–agent problems

where they are a major cause of misinforming or misleading communication (Christozov,

Chukova, & Mateev, 2009).

Patient education as a source of information and empowerment is predicted to

mitigate the agency dilemma and reduce power imbalance (Nguyen, 2011). Health

literacy is seen as a necessary condition that enables SDM (Noonan et al., 2017).

Rosenthal, a Harvard-educated physician and New York Times reporter, emphatically

argues that “we must become bolder, more active and thoughtful about what we demand

of our healthcare and the people who deliver it” (Rosenthal, 2017: 329). This framework

suggests that what patients need is good data on their condition and treatment options

(Tapscott, 2010: 179). Health literacy is the ability of a patient to read and comprehend

medical information. Having this ability enables patients to contribute meaningful

additions to the physician-patient relationship and the decision process (McCormack et

27

al., 2017). According to the Picker Institute, patients can make progressively more

informed decisions with clear, specific and high-quality information (Barry & Edgman-

Levitan, 2012). The ability to make progressively more informed decisions is especially

important for patients with chronic disorders which require decisions that have lifelong

implications and consequences (Montori, Gafni, & Charles, 2006). However, one

limitation is that the data necessary to inform patients is becoming more difficult to

access and decipher (Hathi & Kocher, 2017).

One major resource likely to enable patient health literacy is the use of decisions

aids; tools that simplify information on treatments, risks, and benefits (Stacey et al.,

2014). Decision aids include pamphlets, references, or simplified graphics and have been

documented to improve options and lower costs (Arterburn et al., 2012; Légaré &

Witteman, 2013). Using decision aids, patients are more inclined to discuss and comply

with treatment regimens when they understand the goals and consequences involved in

their care (Friesen-Storms et al., 2015). There is a considerable amount of scholarly work

on the importance of decision aids to facilitate SDM (Butler, Ratner, McCreedy, Shippee,

& Kane, 2014).

The collection of professional or personal contacts and networks of patients

affected by a disease can enable SDM. Being part of a community has tremendous

benefits to patient engagement, psychological health, and physical health compared to

those “bowling alone” (Blumenthal-Barby, 2017; Putnam, 1995). The internet is a

platform to share information, deliver care and guide communities with similar chronic

disorders. Shared platforms generate vast databases to learn, collaborate and teach

(Tapscott, 2010: 184). With an interactive system in place, patients and physicians in the

28

network can be exposed to the collective experiences of patients, such as the methods of

improving day-to-day life with a chronic disease (Frost & Massagli, 2009). Patients are

part of a community, enhanced by the Internet, which disseminates information about

their disease and offer a psychological support system. (Sugawara et al., 2012). Examples

of these patient communities include the National Hemophilia Foundation and Immune

Deficiency Foundation for hemophilia and PID. The benefits of information through

these networks can extend to passive participants in virtual communities (i.e., forums),

known as “lurkers” (Leimeister & Krcmar, 2005). However, one limitation is that online

communities may not have a uniform impact on patient empowerment because of

differing levels of patient literacy and physician paternalism (Petrič, Atanasova, &

Kamin, 2017).

In summary, understanding power balance dynamics is crucial to understanding

SDM. SIT predicts that if there is a power imbalance (i.e., greater physician influence),

the likelihood of successful SDM will be lower. Health literacy is seen as the prerequisite

for successful SDM and achieved by improving literacy through knowledge, aids, and

networks. It is considered a necessary condition to balance the power relationship

between a physician and a patient as predicted by the Agency dilemma. A patient-

centered approach in combination with health literacy may explain the mechanism

through which power can be balanced and SDM is achieved.

Traits

Trait theory uses personal characteristics to predict influence. These include

demographic (sex, age, education), and experience (intelligence, knowledge) (Eysenck,

1963). Trait theory predicts that decision making is influenced by both physicians’ and

29

patients’ background and sociodemographic characteristics (Kaplan et al., 1996).

Participants in the medical dialogue bring with them all of their personal characteristics

which affect patient-provider communication (Cooper-Patrick et al., 1999; Cooper &

Roter, 2003; Cooper et al., 2012). The following traits are thought to most influence

physician decision making: age, gender, race, experience, trust, culture, and family

(Hawley & Morris, 2017).

Physician Traits

The effects of many physician traits have been described in the literature (Table

5).

Table 5. Physician Characteristics that Influences the Decision Process

Factor Citation

Age (Hajjaj, Salek, Basra, & Finlay, 2010)

Gender/Sex (Delaney, Strough, Parker, & de Bruin, 2015)

Race/Culture (Lawrence, Rasinski, Yoon, & Curlin, 2015; Lin & Kressin,

2015; Paradies, Truong, & Priest, 2014)

Experience (Marinova, Kozlenkova, Cuttler, & Silvers, 2016)

Trust (Mondak, Hibbing, Canache, Seligson, & Anderson, 2010;

Nannestad, 2008)

Physician age is thought to increase intuitive and paternalistic decision-making

behavior. As a doctor gets older there is an increased likelihood of intuitive/paternalistic

decision making (Hajjaj et al., 2010). It is suggested that larger age differences between a

patient and a physician reduce patient-centric care (Peck, 2011).

Physician’s gender as a factor for influencing the decision-making process has

mixed results but favors the hypothesis that female physicians are more likely to show

empathetic behavior when treating patients. Some studies conclude that female

30

physicians have more participatory styles related to empathy and information sharing,

adopt a more “patient-centered” behavior or spend more time with patients showing

greater concern to patient psychosocial factors (Bertakis, Franks, & Azari, 2002; Janssen

& Lagro-Janssen, 2012; Kourakos, Fradelos, Papathanasiou, Saridi, & Kafkia, 2017;

Mazzi et al., 2016; Roter, Hall, & Aoki, 2002). However, one study reports that patient

participation is a function of many complex factors and not linked to gender (Street Jr,

Gordon, Ward, Krupat, & Kravitz, 2005).

Race

There is extensive literature on the influence of race (Burgess, Van Ryn, Crowley-

Matoka, & Malat, 2006; Cooper-Patrick et al., 1999; Cooper & Roter, 2003; Cooper et

al., 2012; Lawrence et al., 2015; Schulman et al., 1999; Snipes et al., 2011; Traylor,

Schmittdiel, Uratsu, Mangione, & Subramanian, 2010). White physicians are more likely

to have different prescribing behaviors (Lawrence et al., 2015). Non-white physicians

have been reported by their patients to be less participatory (Kaplan et al., 1996). Studies

have shown that there is less conversation between physicians and patients if their races

are different (Lin & Kressin, 2015). Therefore, it is hypothesized that white physicians

are more likely to implement SDM.

Minority groups are less likely to be screened for and treated for breast cancer

than whites. One recent report documented that Black women have a 40% higher breast

cancer mortality than white women in part due to differences in diagnosis and

comorbidities (DeSantis, Ma, Goding Sauer, Newman, & Jemal, 2017). Patient

characteristics can be both explicit and non-explicit. For example, nonverbal sensitivity

(NVS) relates to how non-verbal communication is correlated with patient-centered

31

performance—racial and gender NVS can have a significant impact on health (Cooper et

al., 2012). While the importance of non-verbal expressiveness has been long documented

(Goffman, 1959), its application to medical environment is now being considered and

efforts made to identify non-verbal cues that impact practice. For example, racial bias has

been associated with patient-centered visit communication and patient ratings of care

(Cooper et al., 2012). In Figure 4, when there is an implicit preference for Whites on the

Compliance IAT, the responses to the top set of pairings is faster than for the bottom set

(Roter, 2017).

Figure 4. Responses to Faces

Cultural factors that influence decision making include patient perspectives of

decision roles and outcomes, understanding risk factors (numeracy), perceptions of

discrimination, and trust in providers. Religious and spiritual culture play a stronger role

for certain minority cultures (Mead et al., 2013). Hawley and Morris (2017) found that

approximately half of Latina and African American patients use prayer to deal with

32

illness and believe God can guide and inform patients to provide protection from harm.

Ethnic minority patients have been found to be “less verbally expressive” and may appear

less assertive (Kourakos et al., 2017; Paternotte, van Dulmen, van der Lee, Scherpbier, &

Scheele, 2015). It is hypothesized that physicians are less likely to implement SDM with

minority cultures.

The effect of family and community can be different between groups because of

alternative perspectives on the role of others in decision making, including community

involvement (Mead et al., 2013); Latina women are more likely to defer treatment

decisions to family members than White and African American women (Maly,

Umezawa, Ratliff, & Leake, 2006). The type of patient (e.g., frail) may also influence the

role of families (Holroyd-Leduc et al., 2016). In the context of life-enhancing procedures

such as growth hormones for short children, parents of the patient have a strong influence

in the treatment decisions (Marinova et al., 2016; Silvers et al., 2010). However, the role

of families may depend on whether the type of patient and procedure are life-enhancing

or life-saving; clear influence of patient family is when treatment is for a child (Larcher,

Craig, Bhogal, Wilkinson, & Brierley, 2015).

Experience

Physicians’ experience refers to the cumulative knowledge and frequency a

physician has treated patients in designated disease categories. This can include the

influence of education (MD or MD/ Ph.D.), number of years practicing medicine, and

participation in clinical trials/research. Experience may play a role in decision making by

increasing the value of patient medical criteria and decreasing the value of patient

preferences (Marinova et al., 2016). Experienced physicians may use intuition to make

33

decisions based on patterns, not patient input and thereby decrease the likelihood of SDM

(Marinova et al., 2016).

Trust

Some scholars suggest that SDM relies on a trusting patient-physician

relationship. Theories and research on trust advance the notion that trust may be “hard-

wired into our brains by evolution” (Nannestad, 2008). Personality researchers suggest

that interpersonal trust is an important component of personality with a strong inheritable

component (Sturgis & Smith, 2010). It is well-documented that trust improves patient

compliance, satisfaction, and outcomes (Cook et al., 2004; Schoenthaler et al., 2014). The

literature describes that trust is built over time, but the rate in which it is built is unclear

(McGuire, McCullough, Weller, & Whitney, 2005). Some suggest that physicians prefer

to engage with patients to build trust and honesty which improves diagnosis and care by

limiting uncertainty (McGuire et al., 2005). There is an association between lack of trust

and adherence to medication (Bauer et al., 2014). Physician trust erodes due to patient

actions such as non-compliance (Kramer & Cook, 2004). Physicians also track non-

verbal cues from patients as a means of determining trustworthiness (Kramer & Cook,

2004: 89). Recent research has shown that the use of decision aids may increase trust

thereby facilitating the shared decision-making process (Nannenga et al., 2009). Overall,

research suggests a key role for trust when disclosing information and patient-physician

decision making (Bansal, Zahedi, & Gefen, 2016; Cole, Kiriaev, Malpas, & Cheung,

2017). It is hypothesized that trust is a necessary condition for SDM.

Separate from individual characteristics, some consider that the disease defines

the patient. Disease is the focus of the technologic and market-driven medical system

34

while illness and the socio-cultural aspects of medicine have blurred into the background.

A “disease-based” approach to health care views individuals as “cases” and undervalues

the sociocultural and humanistic aspects of patient care (Green, Carrillo, & Betancourt,

2002). The distinction between disease and illness has been well described. Whereas

disease defines a pathophysiologic process, illness is defined by the complete person—

physical, psychological, social, and cultural (Eisenberg, 1977). Illness represents an

individual's unique and personal experience of being unwell (Helman, 1981). The more a

physician has a disease focus, the less likely there will be SDM since there will be less

focus on unique patient preferences and values and more focus on standardized or EBM

approaches.

In summary, for some scholars, the key to patient involvement in SDM is to

understand the interplay of certain physician and patient characteristics on clinical

decisions—background, culture, family and trust (Carman & Workman, 2017; Hawley &

Morris, 2017). Some physicians claim that getting to know patients through “slow

thinking” using techniques from behavioral psychology can mitigate negative bias

associated with patient traits (Ofri, 2017).

Organizational Context

An organization’s operational and structural context can influence how and why

physicians engage in SDM (DeMeester, Lopez, Moore, Cook, & Chin, 2016). Physicians

and patients interact within a larger healthcare system that consists of policies, rules,

feedback, coordination of care, feedback and reimbursement. Rules determine how much

time a physician can spend with a patient. Feedback is an organizational process that

could influence future behavior. Coordination refers to patient care between two or more

35

providers. Reimbursement relates to how care is paid and who must approve treatment

costs. SDM is most likely to be optimal if aligned with organizational context

(DeMeester et al., 2016).

Policy

Physicians must take into account national policies and the structure of healthcare

system (Flynn, 2003; McMurray, Pullen, & Rhodes, 2011). The selection of one

treatment approach over another can be based on various factors such as efficacy, safety,

and cost (Liras & García-Trenchard, 2013; Peyvandi, Rosendaal, O'Mahony, &

Mannucci, 2014). The physician’s institution is a part of a larger healthcare system,

which is part of a larger macro system, which includes regulatory agencies, patient

advocacy groups, and accepted standards of care. The system as a whole is an

intermingling cluster of entities and policies that impact the boundaries within which

decisions are made. Policies are thought to reduce decision making between a patient and

physician and limit SDM options (McMurray et al., 2011).

Rules (Time)

All organizations have rules and one rule that can significantly influence SDM is

time. Lack of time is a common influence on physician decision making for chronic care

(Légaré et al., 2012; Légaré et al., 2013). Time constraints may limit the ability of

physicians to comply with preventative service recommendations (Yarnall, Pollak,

Østbye, Krause, & Michener, 2003). Schattner and Simon (2017) state that 9% (approx.

67 minutes) of the physician workday is meeting with patients. However, it is unclear

what the right amount of time is needed for patients: American physicians use more time

than British and German physicians and, according to one study, feel they spend

36

sufficient time with each patient often using more time than allocated for scheduled visits

(Konrad et al., 2010). According to the 2017 Great American Physician Survey, less than

8% of U.S. physicians believe there is not enough time to educate their patients

(Physicians Practice, 2017).

The timing of care can also affect SDM. The initial meetings likely have more

participation than subsequent check-ups. Follow-up examinations may involve minimal

changes if current treatment is satisfactory.

The alignment of physicians and SDM must fit into clinical practice and not

increase the time physicians spend during appointments (MedPAC, 2010). Specialists are

thought to be more successful in implementing SDM because of this alignment and when

decisions are most useful to a patient: cancer treatment vs. cancer screening (Medicare

Payment Advisory Commission, 2010: 192).

In summary, if chronic care requires more physician time, then the amount of

interaction may be as important as other factors. Adequate time for discussion is thought

to facilitate involvement in SDM and provide opportunities for relationship building

which is important for effective communication (Joseph-Williams, Elwyn, & Edwards,

2014). Slow vs. fast thinking may be related to the amount of time allocated to decision

making.

Coordination

Coordination of care is the organization of patient care activities between more

than two participants (McDonald et al., 2007). Effective care requires coordination

because patients with complex chronic diseases need all of their healthcare providers to

understand their unique context. For example, patients with PID have specific

37

gastrointestinal needs and procedures to avoid infections, whereas patients with

hemophilia will need extra precautions when visiting a dentist. Misinformed healthcare

providers—in other words, not coordinated with the patient’s needs—can lead to

disastrous results, nullifying any benefit of SDM. Poor coordination of care is thought to

be a barrier to SDM due to sub-optimal information flow between physicians.

Conversely, good coordination of care supports SDM (Joseph-Williams et al., 2014).

Effective coordination of care helps patients avoid the damage and cost of adverse

drug interactions, unnecessary or duplicate tests and procedures, and conflicting

information from multiple providers (McDonald et al., 2007). Furthermore, care

coordination is best applied to chronic conditions because they are “maintained” over a

lifetime rather than cured or fixed (McWilliams, 2016; Peikes, Chen, Schore, & Brown,

2009). The benefits of care coordination are recognized and incentivized by Medicare

(similar to SDM); they measure coordination of care via ACO scales (Figure 5).

38

Figure 5. Medicare Incentive Measures (CMS, 2016)

Multiple models, such as the chronic care model (Figure 6), have been developed

to illustrate the elements involved to manage chronic illnesses (Parekh et al., 2011); such

elements include the physicians and their team, healthcare entities and systems, patient

communities (Baker, Day, & Salas, 2006). Other models make similar improvements for

the care of high-cost, small populations of patients with chronic conditions by moving

away from the fee-for-service model for a more quality-oriented approach to care. One

model, CareMore, focuses on optimizing care for the sickest patients and maintaining the

wellness of the majority to reduce costs (Hostetter, Klein, & McCarthy, 2017). Another

model used by ChenMed, a physician-led practitioner in Florida, focuses on the elderly

with multiple chronic health conditions (Tanio, 2014).

39

Figure 6. Chronic Care Model (Wagner et al., 2001)

Until recently most SDM models were focused on the patient-physician dyad, yet

care is mostly delivered through healthcare teams. It is thought that SDM interventions

must take into account inter-professional dynamics particularly given the complexity of

chronic care due to multiple co-morbidities. SDM should be analyzed across a larger

spectrum of care contexts. This can reduce silos, improve integration of services, and

enhance continuity of care. Therefore, research needs to accept the importance of

multiple actors (D'Amour, Ferrada-Videla, San Martin Rodriguez, & Beaulieu, 2005;

Légaré, Ratté, Gravel, & Graham, 2008; Légaré & Thompson-Leduc, 2014; Marshall,

Haywood, & Fitzpatrick, 2005).

40

Feedback

Feedback is an organizational factor that increases SDM by encouraging

physicians to consider their approach to patient care (Légaré et al., 2014). In the case of

Medicare ACO measures, feedback and auditing methods may make physicians consider

incorporating SDM by offering incentives similar to employee reviews. If patients do not

feel that they participated in the decision process, they respond that the physician did not

incorporate SDM, thereby reducing compensation. Feedback is thought to be an effective

method for increasing the likelihood of SDM.

Effective feedback methods can help improve how the physician and healthcare

organization meet the patient’s needs (NORC, 2014). Physicians and patients agree that

patients can evaluate healthcare providers based on infrastructure, staff, organization, and

interpersonal skills; however, both patients and physicians agree that patients cannot

evaluate the technical skills (Rothenfluh & Schulz, 2017). Feedback from the patient

allows physicians to confirm the efficacy of current treatment, whether they maintain the

regimen or make changes (Schiff, 2008). Furthermore, feedback is a possible method for

avoiding misdiagnoses by updating physicians with additional information; patients

reporting reactions and symptoms provides crucial warnings that physicians can

recognize (Schiff, 2008). Peer feedback influences the way physicians treat patients,

which reduces over/under treatment (Yeh et al., 2015). In summary, feedback is thought

to increase SDM.

Colleagues

The literature of colleague influences on the physician decision process is scarce.

It describes physicians seeking aid from colleagues when the decision is out of their

41

expertise, reconfirming knowledge, or an issue of morality (Rothman, 2017); for

instance, getting clarification for laboratory professionals (Hickner et al., 2014). The

approval of colleagues influences to physician decision making if any information of the

decision is available for colleagues to view (Godager, Henning-Schmidt, & Iversen,

2016). However, the extent of colleagues influencing the decision-making process has

little representation in the literature for the treatment of chronic diseases.

Reimbursement

Reimbursement of healthcare costs can be categorized into voluntary healthcare

insurance plans, managed care plans, and government-sponsored healthcare programs

(Casto, Layman, & Association, 2006). Approximately 40% of physicians think that there

is too much government and third-party (i.e., insurance companies) interference to

effectively and efficiently treat their patients (Physicians Practice, 2017). This view

supports the claim that insurance reimbursement is a major impediment to

implementation of SDM.

Depending on the healthcare system, the cost and method of reimbursement can

significantly influence a physician’s decision process (Flynn, 2003; Scalone, Mantovani,

Borghetti, Von Mackensen, & Gringeri, 2009). Insurance companies do their own cost-

effectiveness analysis to determine which treatment approach is the most suitable for a

patient based and what is covered under the policy (Wilson et al., 2014). For example,

when treating hemophilia, reimbursable costs are one of the main factors affecting the

decisions of physicians and patients regarding treatment (Lee et al., 2008; Scalone et al.,

2009). Reimbursement for treatment is thought to be a common barrier to SDM. The goal

of the insurance organization is to cover expenses at the lowest cost. Insurance companies

42

have criteria set for which brand, type, or method of treatment is covered (Barua et al.,

2015). Unless the physician contests the insurance and overrules the reimbursement

decision, the physician may be presented with a set of predetermined options of care for

the patient. Some states have introduced new systems to reduce the financial burden of

healthcare to insurance companies through “reinsurance” and thereby minimize coverage

issues (Johnson, 2017; Pear, 2017). However, the extent to which insurance dictates

clinical decisions is thought to be an impediment to SDM.

Summary of SDM Decision Theory

In summary, clinical decision making can be explained by different theories: dual

process, SIT and Agency, traits and organizational context. Dual process theory suggests

that cognitive thinking explains decision making. SIT and agency theory stress the role of

power imbalances resulting from role expectations and information asymmetry. Trait

theory focuses on the potential for bias associated with personal characteristics.

Organizational context represents overarching influences from the larger healthcare

system such as policies and reimbursement. How these different theories contribute to a

larger theoretical framework when treating chronic diseases is the object of this study.

Summary

The literature overwhelmingly supports SDM as an appropriate method of

improving the quality of care and reducing the cost of care for chronic diseases

(O'Connor et al., 2015; O'Connor et al., 2004; O’Connor et al., 2007; Stacey et al., 2011).

No published literature exists about factors that influence physicians to incorporate and

exercise SDM when treating hemophilia or PID, although some literature does address

support for and benefits of patient choice (Elwyn et al., 2012; Friesen-Storms et al., 2015;

43

Légaré & Witteman, 2013). One approach to explaining chronic care clinical decision

making and SDM is dual process theory; however, dual process cannot fully explain the

use of SDM. Other theories from the literature may provide alternative or complementary

explanations include power balance/agency, physician/patient traits and organizational

context.

Both the context and method by which decisions or choices are structured,

referred to by behavioral decision researchers and cognitive psychologists as “choice

architecture,” is crucial to understanding SDM as a method that encourages patients to

participate in decisions which are advantageous to their health (Kahneman, 2017;

Sunstein & Thaler, 2008). Examining physicians’ perspectives are imperative but poorly

understood. If understood, opportunities for structuring and facilitating SDM and

bridging the quality chasm may emerge (Beshears & Gino, 2015). Given the magnitude

of the problem and the impact on society, there is an urgency to understand and address

the variability, complexity, and effectiveness of chronic medical care and create decision

designs that facilitate good decision making that results in better care (Boland Jr &

Collopy, 2004; Hock, 1999: 57).

To achieve a robust understanding of SDM from the physician’s perspective, a

series of three studies were conducted to gather and analyze data and generate theory on

the physician’s role in the decision-making process. Both qualitative and quantitative

methods, “mixed methods”, were used to increase the likelihood of obtaining meaningful

results. The next chapter (Chapter 3) attempts to quantify the quality gaps with chronic

diseases in general and hemophilia and PID specifically. Chapter 4 discusses the research

question development, research methods and the results of each study. The theories

44

summarized in this Chapter 2 were used to generate and test hypotheses related to

decision theory, power imbalance, traits, and organizational context. These theories and

their hypothesized relationships to SDM have not been studied as they relate to specific

chronic diseases such as hemophilia and PID. Furthermore, very little research on these

topics has been from the physician perspective. The proposed research study should yield

better insight into what predicts physician SDM implementation.

45

CHAPTER 3: CHRONIC DISEASES

This study is focused on the treatment of chronic diseases. It is claimed that

effectiveness of SDM is optimized in the treatment of chronic diseases (Friesen-Storms et

al., 2015; Frosch et al., 2011). The goal of treatment for patients with the chronic diseases

is to return their quality of life as closely as possible to that of un-afflicted people. As

cited above, chronic diseases represent the vast expenditure of healthcare resources.

Unfortunately, Western medical training is primarily focused on acute care (Allen,

Maguire, & McKelvey, 2011a). The acute care model is more consistent with traditional

evidence-based medicine and mechanistic protocols for clinical interventions and

management. Compared to chronic care, decision making for acute care has a much

different context (Curreri & Lyytinen, 2017). The combination of longevity of treatment

(more than one year) and complexity (multiple co-morbidities) makes chronic care much

more unpredictable. There is a greater the need for co-created relational care plans and

the need to deal with uncertainty: both consistent with the need for SDM (Allen et al.,

2011a; Allen, Maguire, & McKelvey, 2011b; Letiche, 2008; Lissack & Roos, 1999).

This study explores decision-making processes of physicians who are specialists

in treating hemophilia and primary immunodeficiency. These two chronic genetic

disorders are expensive to treat and are known for their complexity (Dalton, 2015). In the

case of hemophilia, the development of inhibitors or requirement for surgery changes a

complicated disease (replace clotting factor) to a complex disease with significant

uncertainty and no standardized protocol (HFA, 2014). For PID, the disease is inherently

complex because many symptoms fall into more than one category of immunodeficiency

(Al-Herz et al., 2014). Furthermore, PID is manifested by infections, malignancy, and

46

immune dysregulation (Picard et al., 2015). Both diseases are life-long with no cure.

While hemophilia and PID are genetic diseases (not acquired), they nevertheless are good

examples to study since they affect well identified small populations and there are

specialized physicians who treat the disease that are available to survey. In addition, the

researcher has been working with these two communities for 30 years so the ability to get

access to physicians was more convenient.

Hemophilia

Hemophilia is a rare (1:10,000), extremely high cost, life-threatening genetic

disorder with minimal leeway for error and inaction because of the severity of disease if

not treated properly. Approximately $18 billion was spent for the care of hemorrhagic,

coagulation, and disorders of White Blood cells (which includes hemophilia) in 2014

(Agency for Healthcare Research and Quality, 2014a); the average cost per patient in this

category was $14,722, and approximately 78% was spent on prescribed medicines

(Agency for Healthcare Research and Quality, 2014a, f). Hemophilia is a chronic disease,

which further requires physicians to make lifelong, skilled, decisions. At $100,000–

$300,000 per year over a lifetime; it is also one of the most expensive diseases to treat

(Chen, 2016; Eldar-Lissai, Hou, & Krishnan, 2015; O’Hara et al., 2017). The disparity of

cost/outcome for hemophilia when compared to other countries is similar to the overall

disparity between healthcare systems. For example, the U.S. and the United Kingdom

(U.K.) yield similar clinical outcomes, but the cost of hemophilia care in the U.S. is 50%

or more than in the U.K.

Within hemophilia treatment, there are four basic goals at the start of a diagnosis.

First, if the patient is bleeding, the bleeding must be stopped (achieve hemostasis).

47

Second, if possible, future bleeding must be prevented (prophylaxis). Third, all of the

above are exacerbated whenever the patient requires further medical care, such as a

surgical procedure or treatment of infectious diseases (Sox et al., 2013). Fourth, if the

patient develops an antibody to the treatment (inhibitor) such that bleeding continues and

the inhibitor must be eradicated.

The current treatment of hemophilia is replacement therapy using clotting factor

products, which are categorized as either blood-derived (plasma-derived) or made

through biotechnology (recombinant). The advantage of the plasma clotting factor is its

lower cost, whereas the recombinant method eliminates the risk of a viral transmission,

such as HIV (Gringeri, 2011). Although there is less than a 1 in 1.8 million chance of

contracting HIV from plasma-derived clotting factor, the recombinant factor is generally

considered safer, due to its non-biological origins (Goodnough, Shander, & Brecher,

2003).

In recent years, long-acting recombinant factors have emerged in the current

market for clotting factors. Although more expensive, these long-acting agents allow

patients to infuse clotting factor about once a week; standard prophylactic regimens are

three or four infusions per week (Pipe, 2012). Although not yet complete with clinical

trials, therapies involving subcutaneous infusions are making product administration

easier compared to therapies that are administered intravenously (Meunier et al., 2015).

The formation of an inhibitor is one aspect of the disease’s complexity. The exact

cause of inhibitors has yet to be discovered; only a few factors have been found to

increase inhibitor risk, such as ethnicity (DiMichele, 2008). In such situations, the

physician has to approach the patient with specified protocols or adjust depending on

48

reaction to dose or tools (Minno, Minno, Capua, Cerbone, & Coppola, 2010). One

common tactic is to try immune tolerance induction (ITI), or to desensitize the patient to

his or her usual product by continuing treatment for an extended period of time; this

could involve decreasing or increasing the volume depending on patient reaction in the

hope that the efficacy will return (Zaiden, 2014a, b). More complicated treatments are

necessary if the initial treatment is ineffective. If inhibitors continue, bypassing agents

and substitutions are available, such as activated prothrombin complex concentrate

(aPCC), anti-inhibitor coagulant complex (e.g., FEIBA), recombinant clotting factor 7a

(rFVIIa), and rPorcine VIII; these options tend to be much more expensive or pose a risk

of adverse reaction in the case of rPorcine VIII because of reduced platelet counts and

allergic reactions (Gringeri, Mantovani, Scalone, & Mannucci, 2003). These products

avoid the usual coagulation cascade but achieve similar results; however, caution is often

needed because of the aforementioned adverse effects observed in some patients.

A patient’s treatment can be administered at a hemophilia treatment center (HTC),

at a hospital, or at home; self-infusion at home increases quality of life by minimizing

hospital time and expenses (Schwartz, 2013). Although self-care helps patients to manage

expenses and time, professional facilities are better suited to emergency situations and

advice for future treatment (Street, 2012). The treatments can be infused via catheters or

syringes; insertion and removal of the catheter is done by a healthcare professional.

Catheters are implanted for months to years and allow easier access to a vein for

infusions, whereas syringes, although they need precision for each infusion, pose less risk

for infection, let alone bleeding during an initial implantation.

49

Patients are generally associated with an HTC, which employs physicians, nurses,

and social workers who specialize in bleeding disorders. Approximately 150 HTCs that

treat almost all of the hemophiliacs in the U.S. and the U.K. and are closely connected to

one another through research, national and international scientific conferences, and

patient associations (Erikson, Jones, & Tilton, 2012; UKHCDO, 2014). Each HTC has

general standardized treatment protocols within which decisions are made (Srivastava et

al., 2013). Hospitals are part of a larger healthcare system that influences the overall

context in which physicians, HTCs, and hospitals make decisions.

In summary, hemophilia is a genetic disease that requires life-long maintenance,

thereby enabling patients to live full, normal lives. Key decisions for the treatment of

hemophilia involve the severity of the disease, presence of inhibitors, reaction to product

brands and types, patient lifestyles, location of therapy and reimbursement methods.

Primary Immunodeficiency (PID)

PID is a set of chronic genetic disorders resulting from a deficiency in the

immune system, typically exacerbated by increased vulnerability to infection (Costa-

Carvalho et al., 2014; Moens et al., 2014; Younger, Epland, Zampelli, & Hintermeyer,

2015). There are approximately 32,000 treated PID patients in the U.S. (Bousfiha et al.,

2013; Kobrynski et al., 2014). Better screening methods and increasing annual PID

diagnoses suggest that a substantial number of undiagnosed patients exist, particularly in

pediatrics (Ehlayel, Bener, & Laban, 2013; Espinosa-Rosales, Condino-Neto, Franco, &

Sorensen, 2016; Hernandez-Trujillo et al., 2015). Due to difficulty in identifying the

disease, it takes approximately five years from the onset of symptoms for patients to be

diagnosed with PID (Guaní-Guerra et al., 2013).

50

Inherited disorders, such as PID, cause recurrent medical problems for patients

when not treated correctly (Chapel et al., 2014; Hernandez-Trujillo et al., 2015; Jolles et

al., 2015). While manageable and curable under many circumstances, these disorders can

remain undiagnosed or be treated sub-optimally, leading to prolonged or fatal health

complications (Chapel et al., 2014; Condino-Neto et al., 2012). Approximately $28

billion was spent for the care of endocrine, nutritional & immune disorders (which

includes PID) in 2014 (Agency for Healthcare Research and Quality, 2014a); the average

cost per patient in this category was $3,195 (Agency for Healthcare Research and

Quality, 2014a, f); however, this does not reflect the cost to PID patients because the

relatively small population. Modell et al. (2011) have estimated that the annual cost for

an undiagnosed patient with PID (approximately $135,000 annually) is about two times

higher than that for a diagnosed patient (approximately $60,000 annually), which

suggests a substantial societal health care burden (CMS, 2015).

A study performed by a research group for the Immune Deficiency Foundation

approached 10,000 households (27,000 household members) and inquired regarding any

diagnosis for primary immunodeficiency for these household members (Boyle &

Buckley, 2007). This study reported a prevalence rate of 1:1200 population for the U.S.,

bringing the total U.S. PID population to 266,000 patients!

A very different approach was adopted by Joshi, Iyer, Hagan, Sauver, and Boyce

(2009). This research group utilized a cohort-based approach in calculating incidence

rates of PID in the U.S. Essentially, these investigators examined the rate of PID

diagnosis per year in a specific county (Olmsted County, MN) and tried to extrapolate

this incidence to the rest of the country. They calculated an incidence rate of 4.6

51

cases/100,000 population year. It is somewhat difficult to determine prevalence from

incidence rate because survival is a substantial element of any such

calculation. Assuming an incidence in the U.S. of 14,720 cases per year, and utilizing a

mean lifespan for these patients of 50 (assuming that most patients were diagnosed in

early age), one would reach a prevalence estimate of 736,000 patients with PID diagnosis

in the U.S. Of course, lifespan may have been substantially lower; the researchers only

determined that 93.5% of the patients were alive after 10 years. They do, however, note

that diagnosis at an older age (than <12) was significantly associated with mortality.

Thus, the mean lifespan may have been lower, leading to lower numbers for PID

prevalence in the U.S.

At about the same time, the CDC commenced a study on the prevalence and

morbidity of PID in the 2001–2007 period, examining mainly insurance records

(MarketScan databases that compile claims from insurance companies and Medicaid).

This study reported prevalence anywhere from 29.1 to 41.1 patients/100,000 population,

for a total of U.S. PID patient basis of 93,120–131,520 (Kobrynski et al., 2014). The

study determined that prevalence among whites was twice as high as among Blacks or

Hispanics and the B-cell defects predominated. The inherent bias of the CDC studies is

that it is based on health records. Obviously, it would be undercounting incidence of PID

in populations not covered by insurance companies reporting to the databases examined.

Overall, these studies indicate that PID prevalence in the U.S. is substantially

higher than numbers provided by registry-based studies. The approach by Boyle and

Buckley (2007) has been innovative; the assumptions required to calculate prevalence

from the Joshi et al. (2009) study make the very high numbers therein dubious and

52

uncertain. The CDC study by Kobrynski et al. (2014) is certainly a very solid effort and

the numbers reliable and should serve as the current baseline (U.S. PID population up to

131,520 patients), taking into account that it refers only to the period between 2001 to

2007, and that the rate of diagnosis is increasing.

Assuming the incremental care associated with an undiagnosed patient is $65,000

(referenced above) and the number of untreated patients is approximately 100,000, the

cost of missed diagnoses could be as high as $7.7 billion. Alternatively, taking the upper

CDC estimate as the baseline and assuming (a) that 50% of B-cell defects may not

require IgG treatment (very conservatively), and (b) that B-cell defects account for 75%

of all diagnoses, one can derive the estimate of approximately 50,000 patients in the U.S.

that require regular or incidental IgG treatment for their PID disorder. This would put the

cost of missed diagnosis $3.85 billion. Therefore, the likely cost of missed or poor

treatment is between $3.85 billion and $7.7 billion.

In summary, PID is a genetic disease that requires life-long maintenance that—

like hemophilia—enables patients to live full, normal lives. Key decisions for the

treatment of PID involve the severity of the disease, infection history, reaction to product

brands and types, patient lifestyles, and reimbursement methods. The impact of the

failure to properly diagnose and treat is significant.

Estimate of the Quality Gap in Chronic Care and the Implications for Hemophilia

and PID

An analysis of 1) survey data of 33,162 individuals across the U.S.; 2) 25 million

hospitalization records from New York State comparing two (2) five-year periods; and 3)

data from 200 million individual-level, de-identified enrollment data and health insurance

53

claims across a continuum of care (both inpatient and outpatient). These studies help

quantify the quality chasm and the associated cost implications. First, the MEPS data was

analyzed to understand quality gaps from the patient perspective. Second, the SPARCS

database of all New York State hospitalizations was analyzed to understand the cost of

hospitalization associated with hemophilia and PID; both diseases if properly treated

should result in very few hospitalizations. Third, a broad cross-section of U.S. health

insurance claims was analyzed to better understand the overall cost of care.

Analysis #1: MEPS

The Agency for Healthcare Research and Quality (AHRQ) conducts a survey, the

Medical Expenditure Panel Survey (MEPS), to gather data on the U.S. patient population

(Household Component) and insurance (Insurance Component). The Household

Component (HC) collects data from a nationally representative sample of families and

individuals in selected communities across the United States using a panel survey, which

features several rounds of interviewing covering two full calendar years to determine how

changes occur in respondents (Agency for Healthcare Research and Quality, 2018). Prior

MEPS published data estimates that 84% of Americans have at least one chronic

condition, wherein 31.5% of Americans have multiple chronic conditions (Gerteis et al.,

2014). This correlates well with the CDC data of which 86% of healthcare spending is on

patients with one or more chronic conditions (Gerteis et al., 2014).

The 2014 MEPS Household Component consisted of survey results from 33,162

households of which 15,158 participants were at least 18 years of age and have visited a

doctor in last 12 months. Of this sample, approximately 33% believed their healthcare

providers were not listening to them carefully (“listening”), 30% feel their healthcare

54

provider do not always respect what they have to say (“respect”), and 43% feel the

healthcare provider did not spend enough time with them (“time”); these three items were

used as the quality measures to analyze. Specific sub-samples may reflect the healthcare

quality gap. For example, participants that are older than 65 have insurance and those

with excellent perceived health rated their experiences with healthcare practitioners

significantly higher (1 standard deviation) than the average for items involving listening,

respect, and time. Whereas participants with low perceived health and no insurance were

significantly less likely to provide positive responses to the items. Table 6 was recreated

from the MEPS item data summaries, 2014.

55

Table 6. MEPS 2014 Item Data Summary

2014, in

thousands

Listen

(%)

Respect

(%)

Time (%)

Category Population Characteristic Population Applicabl

e

Sample

Always Less

than

Always

Always Less

than

Always

Always Less

than

Always

Total Total 238,870 151,581 66% 34% 70% 30% 57% 43%

Age 65 and over 46,970 37,232 70% 30% 73% 27% 62% 38%

Race Asian/Hawaiian/Pacific

Islandera

14,138 7,602 59% 41% 67% 33% 50% 50%

Black, Non-Hispanic 27,936 15,596 71% 29% 75% 25% 60% 40%

Hispanic 37,018 17,583 62% 38% 68% 32% 52% 48%

Insurance <65, Public only 27,348 16,733 61% 40% 65% 36% 50% 50%

<65, Uninsured 27,229 8,072 55% 45% 61% 39% 49% 51%

65+, Medicare and other public 4,993 3,921 71% 29% 70% 30% 60% 40%

65+, Medicare and private 25,344 20,704 70% 30% 74% 26% 64% 36%

Perceived Health Excellent 60,514 32,740 76% 24% 79% 21% 68% 32%

Fair 24,165 17,546 58% 42% 62% 38% 50% 51%

Good 64,600 42,350 61% 39% 65% 35% 52% 48%

Poor 7,263 6,001 57% 43% 58% 42% 46% 54%

Note: Recreated from MEPS summary tables, 2014 (Agency for Healthcare Research and Quality, 2014c, d, e). The lower responses, “I Don’t Know”

and “Non-Response”, and “Sometimes/Never” were grouped into “Less than Always” from the table; “I Don’t Know” and “Non-Response”

cumulatively represented less than 4% responses in each row. Except for the “Total” characteristic, characteristics (rows) that provided response

percentages within one (1) standard deviation of the mean for the “Always” column were excluded.

*visited a doctor within 12 months

a) Full label: Asian/Hawaiian/Pacific Islander, non-Hispanic

56

In hemophilia3

and PID4

MEPS (2014) samples of 75 participants, there are

similar patterns around income, insurance, perceived health, and the doctor-patient

relationship (Agency for Healthcare Research and Quality, 2014b). There are 32

participants (42%) that had visited a doctor within 12 months prior to the survey; these

patients were qualified to respond to the “listen,” “respect” and “time” questions in the

MEPS. This sub-sample had 15 participants (46%) with poor perceived health and one or

more unsatisfied quality measures.

In summary, the 2014 MEPS survey data confirms that most health issues are

chronic and across three key quality indices (listening, respect and time) there are

substantial gaps: 30–43% of households give their healthcare provider low scores.

Although the numbers are small, the hemophilia and PID patients surveyed by MEPS

(total of 75) show greater quality gaps: more than half have not seen a doctor in 12

months and of those who have seen a doctor, 53% gave lower quality scores.

Analysis #2: IDF

Purpose

Seeborg et al. (2015) sought to gain a better understanding of the drivers of

perceived health (PH) among patients with PID by evaluating the data from the National

Survey of Patients with Primary Immune Deficiency Diseases in America.

Methods

The survey sample of 1,526 patients was nationally distributed and based on the

single, largest database of persons with PID in the world. Patients’ own evaluation of

3

Labeled as coagulation disorders. 4

Labeled as immunity disorders.

57

their health is a simple health metric which links to patient satisfaction and quality of life.

PH was based on the reply to the global question: “How would you describe your current

health status?”

Results

Thirty-nine percent of the patients perceived their health status as poor or very

poor. Patients who were acutely ill and hospitalized in the past 12 months, ones with

limited activity, and chronic diseases, were more likely to have poor PH. Patients with

“on demand” access to specialty care and ones on regular IVIG had higher PH. Patients

not cared for mostly by an immunologist were more likely to have poor PH.

Discussion

The occurrence of acute illness and hospitalization within the last 1 year were

significantly associated with poor PH. Multiple co-morbidities also influenced health

perception as patients with PID and morbidities such as bronchitis, malabsorption, and

recurrent infections were more likely to have poor PH.

Analysis #3

SPARCS A: PID

Purpose

This analysis compares the cost and treatment quality of PID patients from 2000–

2004 (Resnick, Bhatt, Sidi, & Cunningham-Rundles, 2013) to the most recent data from

2010–2014. Resnick et al. (2013) investigated PID in New York State (NYS) using

International Classification Codes (ICD-9).

58

Methods

The frequency and cost of hospitalization with PID in NYS was analyzed using

the Statewide Planning and Research Cooperative System (SPARCS) database, a

comprehensive data reporting system that collects ICD-9 codes patients hospitalized in

NYS. Using ICD-9 codes that classify PID patients (using 279 as the generic root code).

For the identified patients, the demographics, length of state, insurance states, chares,

charges/patients and charges/day were determined. A total of 26,132,412 hospitalizations

were reviewed.

Results

Previously published data (2000–2004) found 2,361 hospitalized patients with

PID. Identified patients were likely to be Caucasian and less likely to have Medicare

when compared to the overall NYS population. The average length of stay was 8.3 days

which was statistically longer than the general hospital population (6 days) (Resnick et

al., 2013). A review of data from 2010–2014 found 611 hospitalized patients with PID.

The racial makeup and Medicare insurance was much more consistent with the general

population. The average length of stay was 6.87 days which was statistically longer than

the general hospital population (5.5 days). The charges/patient (77,936) and charges/day

(11,342) were significantly greater than the general population (40,187 and 7,235). The

total cost PID patient hospitalizations in 2014 was $10.9 million (Resnick et al., 2013).

Conclusion

PID is a rare chronic genetic disease that affects people of all ages, sex and ethnic

groups. If properly treated, PID patients can live a normal life free of immune defects and

other complications. Diagnostic delays and different levels of treatment result in extreme

59

complications which result in unnecessary hospitalizations. The number of

hospitalizations is a surrogate measure of prevalence and cost of undertreatment amongst

diagnosed patients. NYS data suggests a significant improvement (75% reduction) in

treatment between 2000–2004 and 2010–2014. It is currently estimated that there are

32,000 treated PID patients in the U.S. Extrapolating NYS data suggests that

approximately 2,333 diagnosed patients, or 7.3% of the PID population (reduced from

24% in the prior period), are undertreated to the point of hospitalization with a total cost

impact of $181 million to the U.S. healthcare system (substantially reduced from $613

million). This estimated may be understated because NYS has the ninth-best health care

system in terms of access and only captures diagnosed patients (Bernardo, 2017).

SPARCS B: Hemophilia

Purpose

Using International Classification Codes (ICD-9) codes to investigate hemophilia

and the cost associated with under treatment in New York State (NYS). An analysis of

2014 data was performed (most recent available data).

Methods

The frequency and cost of hospitalization with hemophilia in NYS was analyzed

using the Statewide Planning and Research Cooperative System (SPARCS) database, a

comprehensive data reporting system that collects ICD-9 codes patients hospitalized in

NYS. Using ICD-9 codes that classify hemophilia patients (using 286, 287 and 289 as the

generic root codes for coagulation and hemorrhagic disorders). For the identified patients,

the demographics, length of state, insurance states, chares, charges/patients and

charges/day were determined.

60

Results

A review of 2014 data found 2,520 hospitalized patients with coagulation and

hemorrhagic disorders. The charges/patient (54,708) and charges/day (11,026) were

significantly greater than the general population (40,187 and 7,235). The total cost in

2014 for NYS was $137.8 million. While the number of hospitalized patients declined by

15% over the 5 year periods from 2010 to 2014, the charges per patient have been rising

from $44,808/patient to $54,708/patient.

Conclusion

Hemophilia is a rare chronic genetic disease that affects people of all ages, sex

and ethnic groups. If properly treated, hemophilia patients can live a normal life with

minimal bleeds and other complications. The number of hospitalizations is a surrogate

measure of prevalence and cost of under treatment. Extrapolating NYS data suggests that

the cost of under-treatment to the point of hospitalization with a total cost impact of $2.2

billion to the U.S. healthcare system. This estimate may be understated since NYS has

the ninth-best health care system in terms of access and only captures diagnosed patients

(Bernardo, 2017).

Analysis #4

Truven Data: PID

Purpose

Few studies have estimated the morbidity of PID. Data derived from the Truven

Health MarketScan medical claims databases (96,791,177 enrollees from 2001 to 2007)

were used to investigate the incidence and cost associated with undertreatment of PID

(Kobrynski et al., 2014). A comparison was made between previously published results

61

from 2001-2007 and an analysis of data between 2008 and 2010 (most recent available

data) (Truven Health MarketScan® Research Databases, 2010).

Methods

MarketScan databases comply claims from commercial health insurance plans

and Medicaid, documenting diagnoses for hospital care. The frequency and cost of

hospitalization with PID was analyzed using ICD-9 codes from hospitalized patients that

classify PID patients (using 279.0, 279.1, 279.2, 279.8, 279.9, 288.1, and 288.2 as the

generic root codes). For the identified patients, length of stay and cost was used to assess

PID-associated morbidity.

Results

Estimated prevalence 41.1/100,000 for a total U.S. PID patient population of

131,520; of which 32,000 are diagnosed (using ICD codes for immune defects) (Grifols,

2017; Kobrynski et al., 2014). Previously published data from 2001–2007 and 2008–2010

found the proportion of PID persons with hospital admission was higher each year from

2001–2005 with (13.9% to 18.6%) compared to those without a PID diagnosis (7.9 to

8.9%). Persons with PID were approximately twice as likely to be hospitalized (p<.001)

(Kobrynski et al., 2014); their significantly longer hospital stays ranged from 6 to 22 days

depending on age. Having a PID diagnosis for Medicaid patients was also related to the

mean number of admissions per person per year (1.55 versus 1.16). The hospitalization

range of 13.9%-18.6% is lower than the 24% found in New York during the period of

2000-2004. The average cost of hospitalization is estimated at $77,000 per visit for

undiagnosed PID patients; the total cost of treating one hospitalization for each of the

estimated 50,000 undiagnosed patients is as much as $3.85 billion per year.

62

The 2008–2010 Truven dataset reinforces the observations of Kobrynski et al.

(2014). First, the prevalence is steadily increasing as better awareness and screening

methods are developed; Kobrynski et al. (2014) estimated 50.6 per 100,000 in 2007,

whereas it increases from 60.9 in 2008 to 65.7 per 100,000 in the 2008–2010 dataset.

Second, the greater prevalence of PID in female patients than males; however, the gap

has increased from a 3% difference up to a 16% difference in 2010 (58.4% female).

Third, the average length of stay (DAYS) for PID patients is nearly double of the

inpatient population. Furthermore, the average length of stay for children with PID under

1 year of age is five-times greater than the population! Fourth, the cost of treating

hospitalized PID patients is significantly greater than the population, especially in the

case of children under 1 year of age (Table 7).

Table 7. Summary of PID Inpatient Admissions and Costs (2008–2010)

Category 2008 2009 2010

Population Inpatient Admissions 1,039,062 1,061,379 1,081,168

PID Inpatient Admissions 763 816 859

PID prevalence per 100K 60.9 59.1 65.7

PID male (%) 45.48% 42.16% 41.60%

PID female (%) 54.39% 57.84% 58.40%

Population average DAYS 4.0 3.9 3.9

PID Average DAYS 6.9 6.8 7.0

PID DAYS, age group: 1-64yrs 6.4 6.4 5.9

PID DAYS, age group: <1yr 19.5 17.4 25.3

Population average Cost per

Hospitalization $ 15,608.62 $ 16,763.24 $ 17,894.44

PID average Cost per Hospitalization $ 27,284.11 $ 29,311.14 $ 38,828.64

PID average Cost per Hospitalization,

age group: <1yr $ 91,237.26 $ 97,230.51 $ 224,526.06

63

Conclusion

PID is a rare chronic genetic disease that affects people of all ages, sex and ethnic

groups. If properly treated, PID patients can live a normal life free of immune defects and

other complications. Diagnostic delays and different levels of treatment result in extreme

complications which result in unnecessary hospitalizations. The number of

hospitalizations is a surrogate measure of prevalence and cost of under treatment.

Truven Data: Hemophilia

Purpose

An analysis of data between 2008 and 2010 (most recent available data) to

visualize the prevalence and cost of treating hospitalized hemophilia patients (Truven

Health MarketScan® Research Databases, 2010).

Methods

MarketScan databases comply claims from commercial health insurance plans

and Medicaid, documenting diagnoses for hospital care. The frequency and cost of

hospitalization with hemophilia was analyzed using ICD-9 codes from hospitalized

patients that classify hemophilia patients (using 286.0, 286.1, and 286.3 as the generic

root codes for hemophilia A-C).5

For the identified patients, length of stay and cost was

used to assess hemophilia-associated morbidity.

Results

The 2008–2010 Truven dataset provides an overview of hemophilia patients, their

average length of stay in the hospital, and the average cost of the hospital stays. First, the

5

ICD-9 codes that began with 2860 thru 2863 were accepted because the there are no ICD-9 codes in 2008-

2010 of "002.86" or "028.6".

64

prevalence of hemophilia has increased from 30.7 to 32.8 per 100,000; this coincides

with the increase in population. Second, the average length of stay for hemophilia

patients is similar to the population. Third, the average cost of hospitalizing hemophilia

patients is more than double that of the population.

Table 8. Summary of Hemophilia Inpatient Admissions and Costs (2008-2010)

Category 2008 2009 2010

Population Inpatient Admissions 1,039,062 1,061,379 1,081,168

Hemophilia Admissions 291 323 353

Hemophilia prevalence per 100K 30.7 29.9 32.8

Hemophilia female (%) 33% 30% 27%

Hemophilia male (%) 67% 70% 73%

Population Average DAYS 4.0 3.9 3.9

Hemophilia Average DAYS 4.3 4.7 4.5

Population Average Cost per

Hospitalization $ 15,608.62 $ 16,763.24 $ 17,894.44

Hemophilia Average Cost per

Hospitalization $ 40,753.57 $ 65,244.17 $ 41,233.69

Summary

This chapter explored issues surrounding chronic care in general and Hemophilia

and PID specifically. While there is a significant amount of published literature on

quality and cost issues, there is little documentation on the actual issues that influence

quality and quantify the impact. Hemophilia and PID are genetic disorders which if

properly treated should result in a relatively normal lifestyle. Failure to properly diagnose

and treat result in unnecessary hospitalizations which is the highest cost treatment setting,

particularly for chronic care patients which typically have even higher hospitalization

cost. The total cost of hospitalizations is approximately $6 billion in the U.S. and the

product cost is $2 billion greater than the U.K.

65

Table 9. Summary: Quantifying the Quality Gap

Qualitative Observations Quantitative Observations

30-43% of all households give poor ratings on

the quality of interaction.

The total cost of hospitalizations: $6b

estimate (diagnosed + undiagnosed).

53% of the hemophilia and PID patients with

poor perceived health gave poor ratings.

Product cost in the U.S. is $2 billion higher

than the U.K.

Poor PH is low for almost 40% of the PID

population.

The total cost of the quality gap for PID and

hemophilia could be up to ~$8 billion

(~$80,000 per person).

An analysis of a representative sample of the U.S. population suggests that a

significant percentage of households (30–43%)—especially those with chronic care

patients—have a poor interaction with their health care provider: inadequate listening,

respect and time. While the numbers are small, this poor interaction is reflected in the

hemophilia and PID patient populations; almost 40% of the PID population reports poor

perceived health. Using hospitalization as an indicator of under treatment, the total cost

of the quality gap for PID and hemophilia could be up to ~$8 billion or $80,000 per

person. It is possible that a significant percentage of this cost could be avoided with

quality interaction/decision making proper treatment.

66

CHAPTER 4: RESEARCH FRAMING

This chapter presents the framework, research plan, and methods for the research

study.

Theoretical Framing

What is known?

From the review of published research in Chapter 2, following is a summary of

“what we know.” Dual process theory is a relevant and well-documented way of

understanding how clinical decisions are made and the associated biases with different

modes of thinking. SIT and Agency Theory predict that power influences decision

making and health literacy/patient-centric care can balance power between a physician

and patient. Physician and patient traits are thought to be relevant. In all cases, doctors

and patients operate within a wider organizational context that impacts how decisions are

made.

67

Table 10. What We Know Concerning SDM Implementation

KNOWN DESCRIPTION SOURCE

Dual Process 1. Heuristic Decision Making (System 1)

2. Rational Decision-Making (System 2)

3. Biases/Nudge

1. (Croskerry, 2009b;

Kahneman, 2011)

2. (Durning et al., 2015;

Kahneman, 2011)

3. (Silvers et al., 2010)

Power Balance 4. Social Identity Theory (SIT)

5. Principal Agency Dilemma:

Information Asymmetry/Health

Literacy

4. (Ashforth & Mael, 1989)

5. (Christozov et al., 2009;

Noonan et al., 2017)

Traits 7. Sex/Race

8. Experience

6. (Burgess, Dovidio, Phelan, &

van Ryn, 2011)

7. (Marinova et al., 2016)

9. Culture

10. Family

11. Trust

9. Mead et al. (2013)

10. (Holroyd-Leduc et al., 2016;

Maly et al., 2006)

11. (Cook et al., 2004;

Schoenthaler et al., 2014)

Organizational

Context

12. Rules/Policy

13. Time

14. Coordination

15. Feedback

16. Colleagues

17. Reimbursement

12. (Flynn, 2003; McMurray et

al., 2011)

13. (Physicians Practice, 2017;

Yarnall et al., 2003)

14. (Légaré & Thompson-Leduc,

2014)

15. (NORC, 2014; Schiff, 2008)

16. (Rothman, 2017)

17. (Flynn, 2003; Scalone et al.,

2009)

Note: Each numbered item in the description column corresponds with the same number in the source

column.

What We Do Not Know

While much has been studied, there is still much which is not known. The above

potential influences have not been studied in the context of specific chronic disorders. For

68

example, one application of dual process research to medical decision making was

whether to order a diagnostic test or bedside (acute) care. But there has been no formal

study regarding the application of dual process theory to chronic care. Likewise, there is

much scholarly research on power balance, traits and organizational context, but no study

has attempted to understand their relative influence or how these influences can or should

be integrated.

Table 11. What We Do Not Know Concerning SDM Implementation

Unknown Description Source

Dual Process Does dual process apply to

chronic disease?

Depression (Beevers, 2005)

Stereotyping (Burgess et al., 2006)

Power Balance How do power imbalance and

health literacy influence SDM?

Power (Gabel, 2012)

Health Literacy (The Joint

Commission, 2007)

Patient/Physician Traits Are there specific pre-

disposing traits that

meaningfully impact SDM?

(Joseph-Williams et al., 2014)

Organizational Context How do wider organization

influences affect SDM?

(Joseph-Williams et al., 2014)

A Tentative Theoretical Framework

Given the gaps in the literature, no single theory is a sufficient guide for this

research. Therefore, the study will integrate the above four theoretical streams to guide

the research. As a starting point, the below figure is an overview of my proposed

framework (Figure 7) based on the review of the literature. A final framework is

presented in Chapter 7 based on the results of the three studies.

69

Figure 7. Theoretical Framework

Figure contains an illustration from: https://www.canstockphoto.com/doctor-and-patient-

13612490.html

The proposed model can be summarized as follows: the basic interaction between

physicians and patients for chronic care is best explained using slow/fast thinking as

described by dual process theory. As predicted by dual process theory, bias and/or

nudging by physicians impacts how patients respond to physicians. In addition,

physicians and patients bring traits to their interaction which influences how decisions are

made. Furthermore, power inhibits SDM but can be mitigated by patient health literacy

and physician-centric care. All decision making takes place within a wider organizational

context which can have an impact on physician-patient decision making. Trying to

understand which elements within this integrated framework have the most influence is

the object of my research. The lack of an integrated SDM theory is an identified gap in

70

the literature and if developed will contribute to enhancing the use and impact of SDM in

medical decision-making (Groot et al., 2017).

Research Purpose

The purpose of this study is to examine, identify, and analyze factors that

encourage or discourage physicians to implement SDM in the treatment of hemophilia

and PID to determine what predicts SDM. Despite greater spending on healthcare in the

U.S. than any other country, quality and cost outcomes lag far behind other OCED

countries. One potential solution is to implement SDM into the decision processes in the

treatment of chronic diseases. However, there is little evidence of how to successfully

implement SDM despite the stated intentions of physicians to adopt SDM, official U.S.

policy, and some documented SDM models.

Research Plan

The theories described in Chapter 2 were used to plan this study as described in

Figure 6 above. Starting from published theory and data, the factors that might predict

implementation of SDM were explored and tested. The study also tried to understand the

context where SDM is most applicable. Both the factors and context associated with

SDM provide an original and significant contribution to the field of clinical decision

making for chronic care. Figure 8 provides an overview of the potential theoretical

explanations for predicting SDM. The basic plan of this study is to explore and test these

theories using a mixed method design.

71

Figure 8. Theoretical Explanations for Predicting SDM

Research Synopsis and Question Development

The overarching research question is: what predicts physician implementation of

SDM. This question can be subdivided into five parts:

1. What are the drivers of successful implementation and adoption of

SDM?

2. How is bias/nudge introduced into decision making?

3. How do power dynamics affect SDM?

4. What physician/patient traits are important?

5. What aspects of organizational influence SDM?

The first in the series of three studies—a qualitative study—explored the

decision-making process of physicians from the U.K. and the U.S treating hemophilia. By

comparing and analyzing data from the two countries with similar GDP per capita and

status, but with different healthcare systems structures, insight was gained into the

72

decision-making process and styles implemented by physicians in each country treating

the same chronic disease. The purpose of the study was to answer the following

questions:

What are the different factors that influence the decision making of

British and American physicians regarding the treatment of

hemophilia?

How does the larger organizational network affect decision making?

Are there any discernable physician or patient traits that affect decision

making?

How do trust and health information influence decision making?

The primary finding was that U.S. physicians are more influenced by the patient (i.e.,

patient-centric) than U.K. physicians who are more influenced by Evidence-Based

Medicine (EBM). The U.K. physicians tend to make decisions in collaboration with

colleagues as guided by policies (physician-driven) rather than by patient input. A simple

model explaining the relationship between physician decision-making style (fast vs.

slow), patient-centric approach, trust and SDM was developed.

The second quantitative study was used to testing finding from the qualitative

study using hypothesizes supported by published literature. The research questions were:

Do physician decision-making styles affect patient participation in the

choice of treatment tools and protocols?

What is the role of trust and SDM?

What is the relationship between physician decision making and a

patient-centered approach?

Are there certain physician traits that influence SDM?

The primary finding was that patient participation was predicted by physician

decision-making styles for both treatment protocols and tools. It was hypothesized that

73

physicians using a rational “slow” decision-making style mediated by a patient-centric

approach increases SDM as predicted by dual process theory and SIT. It was expected

that physicians who engaged in rational decision-making would have a more patient-

centric approach than those who use a heuristic decision-making style. It was also

expected that rational decision-makers encouraged patient participation in treatment

protocols, whereas heuristic decision-making by physicians decreased participation.

However, it was unexpected to have a positive relationship between heuristic decision-

making and patient participation when selecting tools; in other words, physician use of

heuristic decision-making encouraged participation with the selection of tools, whereas

rational decision making decreased patient participation. Additionally, physician age and

sex have some influence on patient participation with protocols is this specific context;

however, other traits did not have an effect on participation.

The third study included both a post hoc analysis of the second study and a new

qualitative study. The post hoc analysis was used to better understand the unexpected

results from the second study. The second qualitative study explored other factors not

explained by the quantitative model but hypothesized in published literature: power

balance, patient traits, and organizational context. The research questions were:

What encourages or discourages physicians to adopt SDM?

What is the role of power dynamics and how does health literacy affect

these dynamics?

Do certain patient traits influence SDM?

What are the influences from the wider organization?

The results of the post hoc analysis found that immunologists using rational

decision-making and patient-centric approaches increased patient participation. In the

74

post hoc analysis patient participation was split into four DVs and tool feedback is found

to partially mediate participation with protocol management and options. Patient tool

requests were increased by rational, patient-centric decision-making, but unaffected by

the other DVs. The findings from the qualitative interviews suggest that SDM is bounded

by factors other than dual process thinking. SDM is more likely when there is a balance

of power between the patient and physician within permitted organizational boundaries

such as time and coordination of care.

Research Design

This study was an exploratory developmental mixed-methods design. The design

consisted of three sequential studies: (1) qualitative study of decision-making of

physicians in the U.S. and U.K. who treat hemophilia, (2) a quantitative study of the

decision-making style of physicians who treat PID in the U.S., and (3) a post-hoc analysis

of the second study combined with a qualitative study involving interviews with

immunologists that treat PID in the U.S. Figure 9 represents an overview of the research

design flow. Mixed methods are a robust and multifaceted approach that mitigates the

weaknesses of each method when implemented alone (Castro, Kellison, Boyd, & Kopak,

2010). This approach is particularly appropriate because the variables of interest are not

well-defined and are understudied. Furthermore, mixed methods help scholars document

the consistency of findings through triangulation between study methods, data sources,

and the literature as well as extend theory through complementarity (Gibson, 2017).

The study was initiated using a qualitative method because the overarching

research questions were about uncovering factors that influence SDM. The questions of

the first study sought to explore the influences of decision-making. Therefore, a

75

qualitative method was appropriate. Using a quantitative method, the second study used

survey questions to test theory generated from the first study using a different chronic

disorder. Following a post hoc analysis of the second study, the third study questions

explored other factors; therefore, a qualitative method was selected. The mixed method

approach for my overall multiphase design was: QUAL QUAN QUAL (Creswell,

2013b).

My research addressed five research questions over the course of three studies.

Figure 9 organizes in more detail research questions and methods associated with each

question and the overall flow of the study and related findings.

76

Figure 9. Overview of Study Design and Results

77

Integration of Results

The goal of the integration is to bring the findings related to the research question:

What predicts physician implementation of SDM? A systematic review of the three

studies integrated and interpreted the studies’ findings. It was not just a list of the

individual findings but was used to increase understanding of the SDM phenomenon

(Harden, 2010).

The quality of the mixed methods integration is a function of the study’s purpose.

Table 12 lists the purpose, description, and intent for this study (Tashakkori & Teddlie,

2008).

Table 12. Purpose and Description of Mixed Methods

Purpose Description

Triangulation Investigate SDM with different research methods

Completeness Gain complete picture of SDM which is more meaningful

than each of the components.

Developmental Questions for one method emerged from the inferences of

a previous one (sequential mixed methods) and provided

hypotheses to be tested in the next one.

Corroboration/Confirmation Assess the credibility of inferences obtained from each

method.

Expansion Expand and elaborate the understanding of SDM.

Adapted from Table 7.1 of Tashakkori and Teddlie (2008)

Triangulation refers to the use of multiples methods to investigate the same

phenomenon (Greene & McClintock, 1985). This research design intentionally used two

methods (qualitative and quantitative) to assess the phenomenon of shared decision

making in the context of two chronic diseases. This assessment provided a more complete

picture of SDM (completeness).

78

The utility of the findings of this mixed methods study rely on the methods for

each of the qualitative and quantitative studies as well and the combination through the

integration of data. See Table 13 for a summary of the quality standards used for each

method.

Table 13. Quality Standards Used for Each Method

Research

Method

Quality Standard Threshold

Qualitative Credibility

Transferability

Dependability

Persistent observation across two physician groups

Detailed descriptions

Interview saturation

Quantitative Internal Validity

External Validity

Construct Validity

Convergent Validity

Discriminate Validity

Reliability

Co-variance between IVs and DVs

Representative Sample

Instrument and Item pretesting; adapting established

scales

CR > AVE; AVE > 0.50

MSV < AVE; ASV < AVE

CR > 0.70; Cronbach’s Alpha > 0.50; factor loadings

> 0.45

Mixed

Methods

Design Quality Answered research questions across all studies

Interpretive Rigor Based on results from multiple studies

Both methods were used to explore and test results between the qualitative and

quantitative studies (developmental). The first study found the important factors

regarding SDM implementation such as dual process, patient-centered care,

organizational context, trust and physician experience. The quantitative study was used to

confirm how decision styles and patient-centric approach result in SDM and found that

some physician traits are relevant. The third study revealed the bounded nature of SDM

implementation: slow thinking and bias do predict SDM but power, traits and

79

organizational limits limit or bound its impact. Furthermore, both methods were able to

corroborate the results from each of the three studies.

There was also the intent to expand theory related to SDM whereby results from

the 2nd qualitative study (third study) were used to elaborate the findings found in the

quantitative study. This expansion relates to different factors influence SDM and how

these factors can be integrated into a new overall decision-making framework for chronic

care.

Philosophical Lens

This research views reality as socially constructed with the goal of understanding

what meanings physicians give to reality when making decisions (Van de Ven, 2007).

Social constructivism is founded on the idea that conclusions about human functioning

are understood based on the biological and social factors that impact behavior. All

decisions take place in a social context and is a social interpretation of reality. Social

constructivism defines decision-making process as an interaction between a physician

and a patient rather than an individual or intrapsychic process. The process “involves

negotiating, consensualizing, and, is guided by social and cultural factors in defining

what is acceptable” (Cottone, 2001, 2004).

Summary

In summary, a mixed methods research design was used to explore and test theory

related to SDM with the intent to develop a new framework that could answer the

research question: what predicts physician implementation of SDM. Chapters 5–7 review

each of the studies. Chapter 8 provides the meta-inferences derived from all three studies.

Chapter 9 details implications for the findings and suggestions for future research.

80

CHAPTER 5: STUDY 1: PATIENT-CENTRIC VS. PHYSICIAN-DRIVEN

DECISION MODELS IN THE TREATMENT OF HEMOPHILIA

The research started with an exploratory qualitative study initiated in 2015. As

described in Chapter 2, U.S. healthcare quality does not match the cost of care.

Therefore, the first study sought to understand how physicians from countries with

similar GDPs make decisions when treating a similar high-cost complex chronic disease.

In this first study, the decision-making processes of physicians that treat hemophilia were

explored. Hemophilia was chosen both because of the significant disparity of cost and

treatment between the U.S. and U.K. and the researcher’s professional experience with

this chronic disease. This cost difference is consistent with the difference in healthcare

expenses between the two countries. Overall healthcare spending is the U.K. is 50% less

than the U.S. The annual cost of treating a severe hemophiliac exceeds $300,000; in the

U.K. the cost is less than half this amount. Using grounded theory as a systematic

methodology, the goal was the construction of a theory on chronic care shared decision

making through the analysis of qualitative data.

The study started with several questions with the goal to develop operational

definitions, establish priorities, and improve the research design for future studies. The

research questions for this qualitative study were:

“What are the different factors that influence the decision-making of

British and American physicians regarding the treatment of

hemophilia?”

How does the larger organizational network affect decision making?

Are there any discernable physician or patient traits that affect decision

making?

How do trust and health information influence decision making?

81

This chapter presents the study’s design, method, results, discussion, and limitations.

Methodology

An exploratory, qualitative study was conducted based on semi-structured

interviews to compare physicians in the U.K. and U.S. and uncover key influences on the

decision process. The U.S. and the U.K. were chosen for comparison to avoid language

barriers and because both countries have similar first world status including relatively

similar per capital GDPs of $54,000 and $45,000 per annum respectively (WorldBank,

2015). The interview transcripts were analyzed using grounded theory to develop a richer

understanding of physician decision making in the treatment of hemophilia. The

transcripts were coded to extract factors and themes (Corbin & Strauss, 2014). The

methodology was appropriate for this study because physician decision-making in the

treatment of hemophilia is a poorly understood and under-theorized phenomenon.

Interviews were used “to gather rich data from people in various roles and situations”

(Myers, 2013: 119). A grounded theory approach provides guidelines for collecting and

analyzing qualitative data in order to construct theories grounded in the data, “rather than

deducing testable hypothesis from existing theories” (Charmaz, 2006: 2). A systematic

qualitative analysis can generate theory (Charmaz, 2006; Glaser & Strauss, 2017).

Qualitative research focuses on exploring the inner experiences of participants and not on

testing theories (Corbin & Strauss, 2014).

Reliability and integrity issues were addressed based on Silverman’s guidance

(2011). To ensure reliability, the interview protocol was pilot tested on the first

participant. Questions were assessed for clarity, appropriateness and relevance. Revised

interview protocols were created after the initial feedback. Data from the pilot was kept

82

in the analysis. The interviews were recorded and transcribed to ensure data integrity.

Exact quotes from participants were used to support findings.

Analytic induction, the constant comparative method, deviant-case analysis,

comprehensive data treatment, and appropriate tabulations were used to increase research

validity. Conducting counts on the data were used to confirm accuracy and identify

deviant cases (Silverman, 2011). Counts were used to keep track of repeating themes.

Moreover, quantity measurements in the qualitative study were used when possible (e.g.,

event frequency, appearance of key words, and other repeated occurrences).

Sample

Sampling was conducted “on the basis of emerging concepts, with the aim being

to explore the dimensional range or varied conditions along which the properties of

concepts vary” (Strauss & Corbin, 1998: 73). Consistent with grounded theory practice,

the sampling and interview protocol were redefined as the data was collected and

analyzed. A total of 24 physicians were interviewed; 12 with U.K. experience and 12

with U.S. experience. Participants were chosen based on availability. Participants are

described in Table 14.

83

Table 14. Study 1: Sample Details

Category U.S. U.K. Total %

Total 12 12 24 100.00%

Male 10 9 19 79.17%

Female 2 3 5 20.83%

Caucasian 11 10 21 87.50%

Indian 1 2 3 12.50%

10<y<20 2 5 7 29.17%

20<y<30 4 6 10 41.67%

30<y<40 4 0 4 16.67%

40<y 2 1 3 12.50%

All participants specialized in treating hemophilia, with differing levels of

experience and backgrounds ranging from consultancy and teaching to treating and

occupying leadership positions. In the U.S. group, two physicians had more than 10 years

of experience, four had more than 20 years of working experience, four had more than 30

years of professional experience, and two had been working in hematology for more than

40 years. In the U.K. group, five physicians had more than 10 years of professional

experience, six had more than 20 years of professional experience, and one had over 50

years of teaching and consulting in hematology.

Data Collection

Data were collected through interviews conducted on the telephone between May

and September of 2015. Interviews averaged 60 minutes. All 24 interviews were recorded

and transcribed verbatim by a professional service. Interviewees were asked to describe a

series of decision-making experiences in their careers, with the intent of probing different

factors that affected their decision-making in the treatment of hemophilia. Memos were

written immediately after each interview to help capture and analyze the interview data.

84

After each interview was finished, the interview audio file was downloaded onto a

personal computer. The downloaded file was assigned a coded file name, and the original

file on the recording device was deleted. Both the personal computer and the stored

backup are password-protected. All printout transcript papers were stored inside a locked

file drawer. A professional commercial transcription service that follows the precaution

procedure for human subject research was used. All research materials (audio recordings,

video recordings, digital files, transcribed Word files, etc.) will be destroyed no later than

August 2018.

Data Analysis

Data analysis began immediately after the first interview and lasted throughout

the study (Corbin & Strauss, 2008). Simultaneous involvement in data collection and

analysis helped pursue those “emphases as we shape our data collections to inform our

emerging analysis” (Charmaz, 2006: 20). After careful review, the transcribed audio

recordings underwent three stages of coding:6

open, axial, and selective (Corbin &

Strauss, 2014). Codes are the categories of information found in the interview transcripts.

Table 15 summarizes the coding results.

6

NVivo 9 software package was used to code the transcripts.

85

Table 15. Study 1 Codes

Codes Descriptions

Data-driven influence of the literature and other forms of information available to

the physician

Policy-driven standards, regulations, and organizational frameworks in place that the

physician may adhere to in their decisions

Experience-driven influence of a physician’s personal history of similar decisions

Patient-centric patient’s input, collaboration, and trust with their physician in the

decision of their treatment

Physician-driven influence of colleagues on the decision

External factors

Trust

outside influences such as cost and organizational context

Trust willingness of the physician to believe in the information provided

by the patient to follow recommended treatment

During selective coding, two themes emerged. First, physicians from both the

U.S. and U.K. primarily relied on data from published literature when in making

decisions pertaining to the treatment of hemophilia (data-driven). All participants made

decisions that were rooted in evidence-based practice and scientific research although the

U.K. is more EBM and the U.S. is more personalized. Second, U.S. physicians tend to be

more influenced by patient-centric factors than U.K. physicians who tend to make

decisions in collaboration with colleagues and guided by organizational policies.

Findings

All 24 physicians expressed a deep passion towards providing optimal care to

their patients. Their stories provided data on approaches to patients, treatment criteria,

preferences/biases, and organizational factors that influenced their decision making.

Seven main findings had emerged from the data.

86

Findings related to Decision Theory

Finding 1: Both U.S. and U.K. physicians use “slow” rational thinking but U.K.

physicians use more EBM when making decisions

For both physicians in the U.S. and U.K., their training is primarily based on

general medicine. They eventually specialized in hematology based on fellowship,

clinical experience, and personal interest. Physicians in the U.S. and U.K. had mentors

who were able to train and provide the necessary insights on how hemophiliacs are

affected by their bleeding disorder. Training outcomes appeared to be similar. Some of

the physicians engaged in cross-national training between U.K., U.S., and other countries.

Physicians from both the U.S. and U.K. primarily relied on evidence-based

medicine (EBM) in making decisions pertaining to the treatment of hemophilia (data-

driven). They valued slow, rational decision making throughout the diagnosis and

treatment process. Data acts as a boundary to decision making on two fronts: (1) the

literature offers a range of effective treatments based on context and evidence, and (2)

how each physician interprets the literature. Overall, U.K. physicians followed EBM

more than U.S. physicians

According to one U.K. participant, participating in clinical trials was less

important because their findings can be accessed in the literature, which was used as a

justification when making treatment modifications:

We did not participate in any of the recombinant trials. We basically became

aware of it from looking at the literature that the recombinant products were

available.

In the U.S., EBM is important as well. Participant US9 reported following the literature

carefully to ensure being updated with the new treatments:

87

I think as new treatments come along, we want to definitely be open-minded

about incorporating those into our practice. I follow the literature carefully.

I just came back from a big meeting where some of these drugs, including

the Roche drug were talked about quite a bit. There’s no doubt that those

will probably have some impact.

However in the U.S. evaluation of products are more nuanced and based on

individual circumstances. One participant (US3) questioned the improvements that new

products, addressing the scientific basis behind the new long-acting products:

Part of the annual visit is to vote into new products. Sometimes the patients

will bring up the question, "Doc don't you think I'd be a good candidate for

these products?" My own philosophy has been the following. … "I need to

know what your pharmacokinetics are on your current product because if

you are on the top side of the kinetic curve using the current product that

you're on, and if you have an active lifestyle so that you are going to be

active 3 out of 5 or 7 days out of the week, then I need to know how to keep

your clotting factor level over 15% at least so that you can be physically

active and I need to know how long that area under the curve is going to

be." Because the prolonged half-life products most patients don't appreciate

only talks about the prolonged area under the curve. It does not talk about

how long the peak activity is sustained. So when you begin to talk to patients

and you get them to understand that the benefits of the extended half-life

products are only at how long will you remain over 1% of Factor-8 clotting

factor activity, they begin to understand that maybe they're not the ideal

candidate to be on these products because what they really need is to have

over 30% activity so they can go out to soccer practice 3 times a week and

not have a bleed.

Finding 2: Bias/Nudging. Physicians intentionally anchor information in a way

that nudges patients to follow their recommendations.

While physicians believed that final choices are made by patients, they often

provided information to nudge the final decision to one that aligned with the physician’s

policies, evidence, or personal preference. Suggested nudges are accepted because of the

mutual trust between the physician and patient; trust is another finding describes in the

traits section below.

88

One U.K. physician (Participant UK5) explained how they approached the

national switch from plasma-derived to recombinant products, “I think it was to left to the

patient. We just gave them the whole information about the reason to move and

everything and it was completely left it to the patient to make that decision.” Other U.K.

participants (Participant UK9) explained their approach with new patients, specifying that

they generally avoid mentioning plasma-derived products with new patients.

If it's a brand-new patient, usually there are parents involved as well.

Generally speaking for hemophilia A, we would just go for recombinant

product and we'd give them information for recombinant product and

actually we wouldn't give them any information on plasma-derived. In that

space, specifically spoke about plasma-derived, wouldn't mention it.

U.S. participant (US5) specified that patients needed to know how they understood the

product options.

We always had a strong feeling that these decisions involved keeping the

patients and families well informed. In that the patient really had the choice

of what they wanted based on the best information that we could give them

… If you can use cryo rather than concentrate, do that. But of course, it

didn't matter, our patients were already infected … I think that in many

cases there really was no difference. People needed to know the pros and

the cons of how we understood it. I think that's the way you have to practice

medicine no matter whether it's hemophilia or not. Paternalism doesn't go.

During the HIV epidemic, US6 would provide input to patients switching to

recombinant despite the circumstances; the patients were adult and already potentially

exposed to HIV, making the switch potentially irrelevant in their opinion:

Adults that had been previously exposed to plasma-derived products, it

seemed very excessive to ... Not really excessive, but somewhat excessive

to switch to recombinant after they were previously exposed. Especially

when there was evidence that the plasma-derived heat treated was probably

quite safe. If the patient really wanted and specifically asked for

recombinant then we would certainly try to do that, but in general, I think

that ... Used a lot of literature to look at the safety of plasma-derived

products. I think most people were pretty convinced that they were probably

OK with heat treatment.

89

Another U.S. physician noted that after receiving all the important information, patients

can choose the best treatment for themselves:

We always had a strong feeling that these decisions involved keeping the

patients and families well informed. In that the patient really had the choice

of what they wanted based on the best information that we could give them.

Another participant shared that patients can choose whether to receive recombinant or

plasma-derived products for the treatment of hemophilia:

The program officially then and now is still patient-choice. If a patient came

in or the family came in and they wanted a recombinant product and they

were on plasma-derived, they would be switched. We weren’t necessarily

pushing one over the other. A lot of it again was patient-choice.

Findings related to Power Balance

Finding 3: Patient-centric vs Organizational-centric: As predicted by SIT, a

patient-centered approach reduces the physician-patient power imbalance and

facilitates SDM.

When making treatment decisions for hemophilia, U.S. physicians tend to be

more influenced by patient (patient-centric) factors than U.K. physicians who tend to

make decisions in collaboration with colleagues and guided by organizational policies

(organization-driven). As stated by one U.K. physician:

We didn't have the option so that's what happened, the system told us what

to do. The U.K. tender system has worked because all clinicians in the U.K.

[agreed] that all products were equal in terms of safety and efficacy. We all

had to submit information to the people doing the negotiation and we all

had to sign up to it that we all believed, and it had to be all clinicians in the

U.K.

The U.S. method of treating patients is what some referred to as a model of

“mutual participation” which expresses the benefit of a physician aiding the chronically

ill patient to self-treat (Kaba & Sooriakumaran, 2007). U.S. physicians emphasize that

90

patients have life experiences that must be taken into consideration when making

decisions. This method of treatment can be considered SDM. Participant US10 elaborates

on the importance of patient history and experiences:

I guess I'm thinking of the diagnostic puzzles of new patients and working

through diagnosis of new patients more than day-to-day treatment.

Hemophilia is more driven by individual patient's experiences and how they

respond to factor, how they respond to different things. When I use that

term, it's more for the upfront diagnostic.

[Do you let them choose the clotting factors?] There are multiple brands

and generations of factor, the factor that are recombinant or factor that is

plasma-derived. In my experience, I think that I don't have a preference of

one over another. In general, I will lean more to do the recombinant factor

and there are patients that are in certain situations and that may not be the

case but we go through the different factors, what the small differences are

in each of the generations of factor with the family.

A lot of times the family comes because they have a family history and they

have other members of the family using one factor and they know what they

want to use and that's fine but we always leave it up to the patients to decide

if they have preference. Otherwise, we use whatever we think is the easiest

product for them to administer but we always go through each of the

different options and what may be one benefit is over another or that I

always look and family knows that I think they're always equivalent in

efficacy and safety and so in the end, it's their decision.

The U.K. physicians were more likely to treat patients without taking patient input into

account.

The only other patient that this has happened to recently was a chap who's

got severe hemophilia B and he had been taking a plasma product,

Replinine, for years and years and he was our only patient on it, and it's

taken us about five years to persuade him to change. It's taken us a long time

to change his mind about it, but he has now moved over to recombinant.

U.K. decision making was primarily up to the individual treating physician or

group of physicians responsible for the patient. In the case of an inhibitor, a U.K.

physician will discuss the situation with other colleagues:

91

Most of my inhibitor patients, because it’s not that common, you tend to

discuss with colleagues, so I'd say this is a very colleague driven area of

discussion really. There was a national guideline but it doesn't say use this

one, don’t use one. It just says there's a choice, so it doesn't really tell you

what to do. So it’s colleague driven and my local colleagues particularly

didn't want to expose a patient to factor 8 because the whole idea during

that period of time not expose the patient to factor 8.

One U.K. physician mentioned the policies put in place by a board of physicians:

I think in the U.K., we make a decision on treatment on an across the board,

across the country [basis]. We have a national organization in the U.K.

called CDO. Treatment policy is decided by that body. We decided a long

time ago that recombinant factoring should be the treatment of choice for

all of our patients. We're now in a situation where we only use recombinant

Factor VIII and Factor IX for our patients with hemophilia.

Furthermore, the U.K. Physicians were more influenced by mentors and colleagues, such

as the following example from UK8:

We had a patient at the beginning of the week who had surgery on his elbow.

He is a patient, we have some of these, who never turns up, who doesn't ...

We haven't had any blood levels on him for five or six years because he just

doesn't come, or when he does come doesn't let us do any bloods on him,

helps me. Dosing him to get him up to 100 percent, he weighed 75 kilos,

he's a factor-9, and so I sort of said, "Well, we could give him 7,000 but

actually that's far too much for him. I think it's too much for him," so we

gave him less than that. We got 5,500 and that got him up to 105 percent…

Having watched [Colleague 1] and [Colleague 2] do quite a lot of it now,

they all just go, "Well, that's just a bit too much." I don't know. It just felt a

bit too much.

One U.K. physician summarized the difference with the U.S.:

It would be impossible to do something like this in the United States because

the way the United States system works and the ways of advertisement and

the way people have strong opinions that one product is better than the

other, in either efficacy or safety. People were allowed to decide one was

better than the other. If it impacted, there wasn't. That's how it was based.

There was no published evidence and it was not purely based on opinion.

There's a person who's got the strong views, it's the opinion leader and

especially someone who speaks a lot at meetings, and people get the

impression that there is evidence when really there isn't.

92

In summary, U.S. physicians are more patient-centered whereas U.K. physicians

follow organizational rules, policies, and colleagues. Organizational context likely

contributes to this difference.

Finding 4: Information asymmetry is greater with physician with more

experience.

Information asymmetry is most apparent in physicians that participated in

research studies, especially research that resulted in new products and treatment protocols

(thought-leaders). Although physicians with research experience were open to SDM,

thought leaders such as participant US3 would make sure patients knew the difference

between their current products to a new one:

Having an extended half-life product for your trough levels maintained at

1% or 5% for 3 days, that's not going to make any difference in the quality

of life unless you're sedentary. So, I think a lot of the hype and the hope of

these extended products has exceeded the educational level of the patients.

I would say that most of the hemophilia treaters and nurses who take care

of patients don't understand the concept that I just discussed. So, as a key

opinion leader, I think it's important for us to actually try and put this whole

science into perspective. In my mind, the extended half-life products are a

transition product to gene therapy, or a transition product to say the Ace 9-

10.

This pattern of stronger nudging the patient was common through the physicians who

participated in research or clinical trials.

Findings related to Traits

Finding 5: Influence of patient groups: patients are more likely viewed from a

disease perspective rather than individual traits.

Physician decisions may be impacted by the opinion/movement of an entire group

of patients (patient organization) rather than individual patient opinions. Participant UK1

discussed the influence of patient groups, “If a patient group says no to something; you

93

can't force them to have a particular type of treatment.” When the HIV epidemic

occurred, participant US6 recalled infected patients switching products, despite there

being no benefit:

Truly all the hemophilia patients as I recall, they all wanted recombinants.

I mean, why not? On the other hand, we knew, I think, there was

accumulating amount of evidence at that point that the heat treatment or

solvent-detergent or whatever was really quite effective.

Individual patient traits were less important than the views of the entire group:

The patients were allowed to comment on the safety and efficacy issue. This

was the U.K. Hemophilia Society and the patients agreed that there was no

difference in terms of the safety and efficacy in terms of the published

information. Then, but then there was another component and that was the

means of administration. Basically, the kit that came with the product, is

one kit better than the other. Patients like one kit better than another, so

there is that, they were able to, they asked a lot of patients which of those

different methods of mixing the powder and liquid they prefer. Based on

those, they came to a decision that it was actually the doctors were not

involved in that part, which of the factors did the patients prefer in terms of

administration. Interestingly, the patients came and said that one of those

products was preferable to another. They, when they came to that, it was

decided that priory the patients make that decision based on the scores they

gave the different products, the result of that will result in a price difference.

What it was decided before the bidding down of the different companies,

was that the patient preference and they did positive for one of the product,

they said they prefer the administration kit and it came to, it was decided it

would be equivalent 0.3 pence, U.K. pence, benefit…

Finding 6: Physician Trust: interpersonal trust is an important component of

physician’s personality and is often assumed at the beginning of the interaction

with patients.

The trust between physicians and patients inherently effects the decision-making

process because it is assumed and encourages patient input. Participant US9 explained

the importance of trust when making treatment decisions with patients:

I think that there's some families that, that the trust also allows some

families to say, "Look, Doctor X, I understand what you're saying. I

94

understand what you are expecting. I will just tell you that it's just really

difficult for me to do this right now. I appreciate that you're giving me the

best recommendation, but I won't be able to follow that treatment, at least

for now, and maybe in 6 months or a year, I might be able to do it better." I

think that the trusting relationship doesn't automatically mean that the

patients will do whatever I want. I think that the important thing about the

trusting relationship is that there's honesty and respect that goes both ways,

so that with somebody having all kinds of problems with their joints and

they're telling me they're dosing 3 times a week. Then I'm wracking my

brain to try and figure out why they're having all these bleeds they shouldn't

be, that would be more annoying to me to know that they've misled me as

opposed to saying, "Look, I'm not doing the prophylaxis the way you asked.

I understand that that's why I'm having these bleeds and is there some other

way we can do this?" That would be a much better conversation to have

than me trying to figure out why this kid's bleeding or why is this patient

bleeding, when the reason they're bleeding is because they're not giving the

factor but they're not telling me that. The trusting relationship really cuts

both ways. It doesn't automatically mean that they will follow what I say,

but I need them to be honest with me no matter what.

Trust is also built very quickly or assumed, which may contribute to bias and

nudging as well. Participant UK05 describes the time it takes for a trusting relationship to

form:

I don’t know how quickly it gets established. It is difficult to say that

perhaps within the, I have worked in two institutions now, I was at [London]

since 2000 and I came to [Southern England] in 2008. I do know that by the

end of 2008 I knew all my patients and they respected and talked to me after

one or two consultations.

Finding 7: Minority cultures involve family decision making.

As described by US9 above, patient culture and family status play an important

role in the decision-making process. Especially in minority groups, the patient and their

family are fully involved in the treatment decisions and education. U.S. participant

(Participant US12) educated a whole family about recombinants. It was an immigrant

family that never encountered hemophilia; the physician had to educate the family of

safety and cost differences between countries.

95

Findings related to Organizational Context

Finding 8: U.S. Physicians are not influenced by reimbursement methods.

In the U.S., insurance does not influence the physician decision process.

Physicians are often in contact with insurance companies to get coverage for specific

treatments instead of pre-approved options. However, they will not fight the insurance

company to fulfill every patient request such as switching to brands to an equivalent

product. Participant US9 recalled:

If your payer's saying you got to use Xyntha, then you got to use Xyntha."

You know what I'm saying? I will pick my battles. It would not be contrary

to my recommendation to have a patient on Xyntha. It's no problem at all. I

don't see any problem with switching. Just because a family liked Advate

or liked Baxalta, Baxter whatever, that's not a reason for me to fight with an

insurer. Now, if on the other hand, an insurer's saying, "Well, your patient

as an inhibitor and you want to use NovoSeven, but we want them to use

Feiba," that I will fight because I think inhibitor patients generally will need

both drugs. If it goes against my recommendation, I will fight for it and I

will not follow through with a treatment plan that a payer has dictated to

me. If it's, like I said, something that is really not a medical issue, such as

switching from one similar drug to another, because one is on formulary

and one is not, I'm not going to fight that battle. That's not worth fighting.

Results

The findings provide insight into the factors that influence U.S. and U.K. physician

decision making for the treatment of hemophilia (refer to the research questions on

page 80). The findings are summarized in

96

Table 16.

97

Table 16. First Study: Summary of Findings

# Description of Finding

1 Both U.S. and U.K. physicians use slow thinking, but U.K. physicians use more EBM

when making decisions.

2

Physicians that believe final choices are made by patients often provided information to

nudge the final decision to one that aligns with the physician’s policies, evidence, or

personal preference.

3

U.S. physicians tend to be more influenced by patient (patient-centric) factors than U.K.

physicians who tend to make decisions in collaboration with colleagues and guided by

organizational policies (organization-driven).

4

Information asymmetry is most apparent in physicians with experience – participation in

research studies – especially research that resulted in new products and treatment

protocols.

5 Patient groups (patient organization) choices influence physician decision making, rather

than individual patient traits.

6 Trust in patients is assumed by physicians and encourages patient input.

7 Minority cultures involve family decision making.

8 U.S. Physicians are not influenced by reimbursement guidelines. They may agree with pre-

approved options, but will fight for desired alternatives and are allows successful.

Overall, the primary influences on decision making are dual process decision

making, power balance, traits, and organizational context. The physician will determine

the amount of effort (slow or fast thinking) necessary to sufficiently treat the patient

depending on the relevant contexts. One of these contexts is the power balance between

physician and patient: U.S. physicians are more influenced by patients than U.K.

physicians because the U.K. national healthcare system has established a greater number

of nationally standardized procedures with more affordable, yet restricted, treatment

options. Another context involves the physician traits such as experience with research.

Physicians who participated in research studies were less likely to be patient-centric

because of their detailed knowledge—they would offer express frustration with the

influence of insurance and often argue with pharmaceutical companies regarding what is

98

the best treatment. Those who did not participate in research were more often indifferent

to the types of treatment and what insurance covered. Lastly, patient insurance—related

to organizational context—in the U.S. allows for more treatment options; despite

insurance policies, the U.S. physician will override the insurance options if suboptimal to

their desired treatment plan. U.S. hemophilia physicians are more “patient-centric” and

more likely to use SDM than U.K. physicians. The influence of the wider healthcare

organization/system was much less in the U.S. where physicians use a slow thinking style

combined with a patient-centric approach as predicted in the literature: patient-centered

care reduces power imbalance and increases SDM.

Discussion

This study provided insight into the decision processes of U.S. and U.K.

physicians. While published scientific data is a significant influence for both groups,

physicians in the U.S. are more likely to make decisions based on the needs of their

patients whereas the U.K. physicians tend to be more influenced by colleagues and

government policies.

Overall, rational/slow thinking seemed to dominate decision making. This was

less so for U.K. physicians who are more grounded in EBM and organizational policies.

The EBM focus of U.K. physicians falls in line with the literature, although described in

a general context rather than hemophilia (Campbell, 2013).

U.K. physicians are trained to develop clinical and decision-making skills. This

was supported by the first finding in which emphasized the foundational importance of

medical training for developing decision-making skills needed to treat patients with

hemophilia. In addition to formal training, physicians in both the U.S. and the U.K.

99

usually found a mentor who was able to train them and provide the necessary insights to

learn how hemophiliacs are affected by their bleeding disorder. Eventually, physicians

develop mental illness scripts to save time and energy when treating patients. If there is

uncertainty, they return to the literature to reinforce their illness script. If the physicians

from both the U.S. and U.K. rely on the data/literature for their decision process, why do

they have such different approaches to patient input?

The differences in whether the physician is patient-centric or physician-driven

likely relate to the larger healthcare system. Prior published research and study findings

support a more patient-centered and less cost sensitive approach in the U.S., whereas the

U.K. is an organization physician-driven approach (Lee et al., 2008). Kaba and

Sooriakumaran (2007) describe the U.S. patient-physician relationship is described as

“mutual participation;” the patient and physician make decisions as equals (Kaba &

Sooriakumaran, 2007). This approach is the essence of SDM. It might be that in the U.S.,

the physician is encouraged to “sell” their service to the customer. This entails the

physician to appease patient preferences and increase their chance to return as well as

ensure the treatment will be reimbursed. Patient choice is a significant value in the U.S.

system and one advocated by the U.S. patient association, the National Hemophilia

Society.

In addition, it might be that U.S. physicians are patient-centric due to the lack of

available aggregate data. As opposed to the national database in the U.K., each U.S.

physician must rely on the patient’s history and other data they have gathered in their

practice. Not only is the U.S. aggregate data limited and outdated, it is also fragmented

by the healthcare system.

100

The U.K. approach is top-down: treating chronic disorders as a whole rather than

on an individual level. The U.K. physician preference to discussing treatment with

colleagues over patients is not mentioned in the literature. This is facilitated through the

U.K National Health Service and supported by a national tender and patient registry. The

tender keeps the overall cost the treatments lower, albeit still in the $100k annual range

per hemophilia patient, and all services are free through the NHS (Mannucci, Mancuso,

& Santagostino, 2012). The tender system has also designated the most cost-effective

products and filters out more expensive therapies. Although this has lowered cost, it also

limits options. The options are chosen through the country’s physicians, who also have

access to the national hemophilia database; the database contains a vast amount of

information on every hemophiliac in the U.K., all required to participate. It may be that

the U.K. system has relieved the burden of cost and insurance from the physicians so that

they can focus on providing the highest, affordable method of care consistent with policy

and EBM.

It is clear that U.S. and U.K. physicians in hemophilia care approach treatment

decisions in part due to organizational context. The difference may result from the

system’s approach to uncertainty. The U.K. system is essentially removing as many

variables as possible in the treatment of hemophilia by limiting product options, offering

care to everyone through public financing and national guidelines, and keeping track of

all the information of each hemophiliac. U.K. physicians follow preset protocols and do

not need to worry about the individual patient need. U.S. physicians have less

organizational constraints: while insurance companies may attempt to exert control,

physicians having enough influence to get exceptions to the reimbursement policies.

101

The overall impact of influences on physician decision making was analyzed

using abductive reflection. Abductive reflection is the logical inference wherein the

simplest and most likely explanation is sought for an observation (Van de Ven, 2007). It

is clear that decision-making style and patient-centered care affect patient participation

but context matters. There must be a set of conditions that encourage and enable the

physician to implement SDM. These conditions/influences may take the form of personal

preferences in treatment approaches or overarching policies that guide the entire

treatment process.

Based on the above reflection, dual process theory was hypothesized as a possible

explanation for how hemophilia physicians make decisions within the confines of the

overall organizational context. U.K. physicians are free to maintain more of an intuitive

mode of thinking while the U.S. thinks more slowly and rationally based on patient

preferences. As the literature describes, System 1 thinking relies on experience and

illness scripts for treatment (Campbell, 2013). With the protocols in place, built using the

collaboration of the nation’s physicians and database information, they do not have to

analytically approach every single patient. This reserves their time, energy, and resources

for the more chaotic cases such as inhibitor treatment. The U.S. physicians, lacking such

protocols, must develop their own personalized approach over many years by analyzing

every patient uniquely (System 2) until patterns arise to a level safe enough for treatment

(Campbell, 2013).

Each mode of thinking (System 1 and System 2) has an impact on the treatment of

hemophilia. System 1, as implemented by the U.K., enables healthcare to be much more

affordable and yield better results in the population; one could also call this approach as

102

“disease-oriented” (Bensing, 2000). However, the protocols limit the treatment options.

For example, patients with unique pharmacokinetics (PK) may not respond well to

standard EMB techniques and require extra care. This could lead to suboptimal treatment

or multiple rounds of trial-and-error. Not only would the patient need to switch products,

but the expenses of an adverse event could be more than approaching the patient with

System 2 from the start. On the other hand, System 2 approaches have a much lower

chance of error due to the analytical nature of the mode of thinking (Campbell, 2013). By

approaching care on an individual level, unique PK patients can be distinguished and

optimally treated. This also applies to the other patient values such as lifestyle (on-

demand vs. prophylaxis) and product choice (recombinant vs. plasma-derived). Health

literacy, nudging, and trust support this practice.

One difference between U.S. and U.K. physicians was power balance. In the U.S.,

power is more balanced because the physicians use a patient-centric approach and

encourage health literacy in their patients more than the U.K. physicians. Whereas the

U.K. physicians have all the power because all treatment centers and physicians in the

U.K. provide identical/similar options to the patient.

The findings show that U.S. physicians are more likely to treat patients without

the help of colleagues and take each individual patient’s needs into consideration. These

practices may be caused by the lack of universal treatment policies and multiple

reimbursement programs in the U.S. It is also possible that without a universal data-

collection system, U.S. physicians are forced to treat patients based on their knowledge of

individual events. U.K. physicians tend to be more influenced by national policies and

procedures, limiting their ability to make personal and individual decisions. However, the

103

existence of national policies and procedures enhances standardization in the treatment of

hemophilia. However, despite the patient-centricity of U.S. physicians, there is a strong

element of nudging patients to do as they recommend.

Based on the finding and abductive reasoning, Figure 10 shows a model of

relationships that were constructed and could be quantitatively tested.

Figure 10. Theory Exploration

Figure 10 represents a possible model of dual process theory explaining the

implementation of SDM. Within the organizational context, which includes its rules and

reimbursement methods, the physician uses a decision-making style (dual process) to

treat the patient. However, for the decision-making process to be categorized as “shared

decision” likely depends on physician traits, patient-centric approach to treatment, and

the presence of trust.

Research Gap

There is an inadequate understanding of how the combination of these influences

affect hemophilia treatment decisions, and which factors are more important than others.

104

There has been no research that examined the decision-making processes of physicians

regarding the treatment of hemophilia, particularly between the U.S and the U.K. This

study addressed the gap in the literature regarding the decision-making processes

involved in the treatment of hemophilia from the perspectives of American and British

physicians.

Implications

The results of the study should be of interest to the academic community,

governments, funding agencies, and pharmaceutical company marketing departments in

understanding the influences and barriers that physicians experience when making

treatment decisions. Although the treatment of hemophilia has dramatically improved

over the past 50 years, it is likely that cost-effective treatment outcomes can be further

improved. These findings can be utilized to justify consideration of a national tendering

process in the U.S. to reduce cost while still maintaining patient-centered care to meet the

individual needs patients.

In the short-term, the role of physicians in treatment decisions and the interplay

between the physician and the wider healthcare system should be considered. American

and British physicians do not make decisions in isolation. They often have to consider

external factors such as national policies and the structure of healthcare systems. In the

longer term, thinking about how the system can be changed for new products and

protocols and how the process can be improved should be considered, particularly with

respect to the optimization of treatment of hemophilia. The decisions of American and

British physicians in the treatment of hemophilia are influenced by the literature,

105

highlighting the importance of strengthening research and participation in clinical trials.

The proposed model may be applied more broadly to other chronic diseases.

Limitations and Future Research

Limitations of this study—as with many qualitative studies—include the use of a

non-randomized, small sample with mostly participants who were Caucasian men. Given

these limitations, the homogeneity of the sample might not reflect the experiences and

perceptions of all American and British physicians who treat hemophilia. In addition,

each participant was relatively eager to participate, as opposed to those who had declined

for personal or timing issues. The results of the findings reported in this study were based

on physicians who had the time and interest to participate in this study. The open-ended

nature of the interviews may have left some important information gaps due to time

constraints so that some participants did not provide answers that truly reflect the scope

of their experiences and perceptions. In addition, the study population was mostly White

males; increasing the sample size to include a more diverse group of people and countries

would help to mitigate a potentially ethnocentric/gender point of view.

The proposed model may be applied more broadly to other chronic diseases.

Although the findings have shown divergent behavior between U.S. and U.K. physicians,

can the results be applied more generally to other chronic diseases? Do all physicians use

the data as their main influence for decisions? Is this influence similar to other physician

specialties? The question remains, is one method better than the other? Can each system

learn from each other for mutual benefit and improvement? On the surface, it appears that

the U.K. has a more cost-effective system but less SDM. In the case the U.S., the high

prices may encourage research from the pharmaceutical companies to keep pushing for

106

better, innovative treatments. Future research should examine physician decision making

in more detail.

Conclusion

In terms of the decision-making in the treatment of hemophilia for American and

British physicians, most participants tended to be more guided by empirical data and the

literature when making decisions (data-driven) although much more so in the U.K. The

U.S. physicians can be classified as more patient-centric, where most physicians in the

U.K. can be characterized as more EBM. Patient-centric physicians consider other factors

in order to individualize treatment towards a patient’s preference and values. Trust

appears to be necessary for patient-centered care. Because of the tendering process in the

U.K., physicians tend to be less collaborative with their patients in terms of their

decision-making process.

The cost of treating hemophilia A and B patients in the U.S. is approximately $2

billion more (normalizing for population) than in the U.K., with no apparent difference in

clinical outcomes. Understanding this difference within the context of physician decision

making could help professionals develop treatment strategies that could improve cost-

effective treatment for chronic diseases in the future. Policies such as the national

tendering process in the U.K. help to lower the cost of products and benefit the

manufacturers at the same time (McMurray et al., 2011). However, the tender process

operates within a system in which the options of physicians become limited. The findings

indicate that the national tendering policy can lower the costs of treatment products,

allowing U.K. physicians to be less concerned about cost. Thus, there are reinforcing

mechanisms that provide a good outcome, but simply transplanting the U.K. tender

107

system to the U.S. without the corresponding patient-centered approach may not lead to a

better care. The U.S. healthcare system has different stakeholders such as insurance

companies and patients who are already powerful actors in influencing policies and

decisions, making the establishment of national tendering process more difficult.

In summary, this study offered insight into the physician decision process for the

treatment of hemophilia in different national contexts. Furthermore, the results suggest

that slow thinking and patient-centered care are likely related to SDM. Traits also may

play a role. The next chapter, Chapter 6, discusses the subsequent quantitative study that

was conducted to test the relationships hypothesized from this study.

108

CHAPTER 6: STUDY 2: DOES PHYSICIAN’S DECISION-MAKING STYLE

EFFECT PATIENT PARTICIPATION IN THE TREATMENT CHOICES OF

PRIMARY IMMUNODEFICIENCY?

This chapter covers the methods and results of Study 2, a quantitative analysis of

physician decision-making regarding the treatment of primary immunodeficiency (PID).

This study quantified the relationships theorized based on Study 1 results. The key

findings from the first study included the role of rational decision making and patient-

centric care in the U.S. and its relationship to shared decision making. While slow

thinking may be necessary for SDM, a patient-centric approach is the mechanism through

which SMD is explained. As predicted by the SIT literature, patient-centric care reduces

power imbalance by bringing patient values and preferences into the decision process.

However, patient/physician trust must be in place for agreement on a shared treatment

plan. In addition, physician traits such as training and experience have an influence.

These findings could be tested in a quantitative model.

Because the number of physicians treating hemophilia is small and insufficient for

a quantitative analysis, I changed the target study to PID treating physicians. Similar to

hemophilia, PID is a high-cost treatment and all cases are complex due to multiple co-

morbidities. The research questions of this study were:

Do physician decision-making styles affect patient participation in the choice

of treatment tools and protocols?

Does a patient-centric approach explain the mechanism by which style

influences SDD

Does Trust explain when there is SDM?

Do certain physician traits affect patient participation?

109

This chapter presents the design, method, results, discussion, and limitations of the

second study.

Research Gaps

Some research has been done on PID patient satisfaction as well as barriers and

facilitators of SDM. However, there is very little evidence concerning how participation

is predicted by the style of physician decision-making (Gardulf & Nicolay, 2006; Gravel,

Légaré, & Graham, 2006; Légaré et al., 2008; Nicolay et al., 2005). Furthermore, there is

significant documentation that SDM improves diagnosis and treatment (Shapiro, 2013;

Shapiro, Wasserman, Bonagura, & Gupta, 2014). What is not known is whether a

physician's decision-making process affect patient participation when choosing the

protocols and tools. Does the physician’s treatment approach (patient-centricity)

encourage or inhibit patient participation?

The literature describes that trust is built over time, encourages adherence to

medication, erodes from patient actions, and is partially measured by physicians via non-

verbal cues (Kramer & Cook, 2004: 89). Is the physician trust in patient-given

information necessary for successful SDM?

Finally, there is considerable research on the impact of traits but what physician

traits play a role in patient participation?

To address gaps in the literature, and draw on published studies and related

theory, a quantitative study was designed. This study followed the qualitative study that

compared physician decision-making in the U.S. and U.K. for patients with hemophilia

(Lamb, Boland Jr, Lyytinen, & Wolfberg, 2015). The qualitative results showed how

U.S. hematologists are more likely to use “slow thinking” and be more patient-centric

110

than U.K. hematologists. Based on the earlier exploratory qualitative study, it was

theorized that differences in a physician’s decision-making style impact the level of

patient participation when influenced by a patient’s specific preferences and unique,

individual conditions. Patient-centered care is the mechanism through which SDM is

achieved: patient-centered care balances power which gives patients the opportunity to

participate. Trust plays is a potentially moderating role: it must be present for SDM.

Traits also have an impact, but it is unclear which traits matter.

Design

330 U.S. physicians who treat PID were surveyed to measure the effect of

physician decision-making style on patient participation in PID treatment. Theory

generated from the first qualitative study as well as constructs from dual process, SIT and

trait theory were applied to examine the physician decision-making process. The inquiry

linked physician decision-making style (fast vs. slow thinking) and a patient-centric

treatment approach to the level of patient participation in the decision-making process.

The role of trust and specific physician traits were also tested.

Measurement Operationalization

Constructs used measured using modified validated scales. Each item score was

set to a 5-point Likert-type scale ranging from 1 = “strongly disagree” to 5 = “strongly

agree.” The survey consisted of 76 items: 60 were adapted from other scales and 16 were

developed to collect descriptive information (i.e., age, education). Reflective scales were

used for each of the following constructs: decision-making style, approach to treatment,

patient participation with treatment protocols, patient participation with treatment tools,

111

and trust in patient-given information. With reflective constructs, the indicators are

caused by the latent variable (Kenny, 2016).

DMR & DMH

Decision-making style was measured using a scale developed by from Buck and

Daniels (1985). The scale contains 20 statements assessing a person's natural disposition

to use either an analytical (rational) or an empirical (heuristic) decision-making style

(DMH).

Patient-centric

A physician’s patient-centric approach was measured using a Physician-Patient

Orientation Scale (PPOS). The scale measures attitudes towards sharing information and

the “extent to which the respondent believes that patients desire information and should

be part of the decision making process” (Krupat et al., 2000: 51). The survey is open for

use in academic research with the condition that the original author of the survey tool

gives permission for its use before the start of the study (Krupat et al., 2000). The

instrument was originally created to evaluate doctors’, medical residents’, and patients’

views of their roles in medical care (Pereira et al., 2013). This instrument has been

regularly used and tested for validity in the U.S. and abroad (Pereira et al., 2013). Krupat

et al. (2000) determined the 0.73 reliability to be satisfactory for physician respondents.

Trust

Physician’s trust in patients to provide accurate information was measured using a

modified version of Thom’s (2011) Physician Trust in Patient Scale. The original scale

consists of 18 questions is used to measure respondents’ attitudes toward patients in the

context of prescription opioid treatment for chronic nonmalignant pain. Thom (2011)

112

trimmed the scale to 12 items and demonstrated that the items had reliability (i.e.,

construct validity) with a Cronbach’s alpha of .93 for specific self-reported behaviors.

Patient Participation

The following scales were used to measure the physician’s perception of patient’s

participation in the treatment choices: the customer participation scale from Gallan,

Jarvis, Brown, and Bitner (2013), the “Responsiveness to Patient Requests” scale from

Petroshius, Titus, and Hatch (1995), and the “Participation in Decision Making” scale

from Siegel and Ruh (1973). The scales were combined to measure patient participation

with protocols (IOP) and patient participation with tools (IOT).

Patient participation in the choice of treatment protocols was measured using

Gallan et al. (2013) participation scale and an adapted scale from Siegel and Ruh (1973).

Gallan et al. (2013) scale is a four-item tool that measures the degree to which patients

provide information to the physician and were actively involved in decision-making.

Siegel and Ruh’s (1973) scale contains five-items which concern the degree of

participation individuals have in decisions affecting their jobs.

Patient participation in choice of treatment tools was measured by the

Responsiveness to Patient Requests scale created by Petroshius et al. (1995). The scale

from Petroshius et al. (1995) contains three, five-point Likert-type statements that are

used to measure the attitude a physician has about writing prescriptions for medications

that have been specifically requested by patients.

Physician Traits

Participants were asked a series of questions to measure traits and the relationship

to the factors listed above. Physician traits include age, race, sex, and experience (years

113

of practice, education, and involvement with research). No scales were adopted from

other studies because each trait had one question for each.

Constructs of Interest

The independent variable (IV) was the physician’s decision-making process,

represented by two variables: rational decision making and heuristic decision making.

Rational decision making (DMR) involves thoughtfully attending to information, whereas

heuristic decision making (DMH) amounts to relying on simple heuristic rules or

experiences as the basis for a decision. This represents Systems 1 and 2 (fast and slow

thinking) from dual process theory. The dependent variable (DV) is patient participation,

also represented by two variables: patient participation in the choice of treatment

protocols and tools. Patient Participation with Treatment Protocols (IOP) refers to the

patient’s input on aspects of treatment such as schedules and administration methods,

whereas IOT refers to the patient’s input on the choice of treatment products.

The mediator is the physician’s approach to treatment; specifically measuring the

physician’s tendency to include a patient’s psychosocial (non-biometric) burdens in the

decision process; thus, the variable measures the physician’s “patient-centric” approach

to care (APC). APC partially mediates effects of decision making; physicians take into

account the psychosocial characteristics of the patient when making decisions which

reduce the power imbalance with patients (Bravo et al., 2015; Hibbard, 2017;

McCormack et al., 2017). Patient-centricity reduced power imbalances and enables SDM.

Therefore, a higher level of APC should increase patient participation.

114

The moderating variable is the physician’s trust in the patient to provide accurate

information and follow the treatment regimen (TIP). Patient trust must be present for a

physician to incorporate patient input and thereby increase SDM.

Multi-group variables are race, sex/gender, and education. Physician race is

thought to affect patient participation because the literature suggests white physicians

have more patient participation than non-white physicians (Lin & Kressin, 2015).

Physician gender could influence the results because some studies conclude that female

physicians have more participatory styles such as empathy and information sharing

(Bertakis et al., 2002; Janssen & Lagro-Janssen, 2012; Roter et al., 2002). Physician

education refers to the highest degree earned (MD vs. Ph.D.). Although there is no

literature—to my knowledge—measuring patient participation with physicians of

different degrees, Schnell and Currie (2017) suggest that physicians with education from

better medical schools were less likely to prescribe opioids, thereby likely to decrease

patient participation and SDM by extension.

The control variables were physician years of practice and age. Years of practice

refers to the number of years the physician has treated PID. The decision making of a

physician new to the specialty may differ from an expert. Physician age (AGE) refers

their age in years, which may reveal differences between generations of physician

decision making.

Figure 11 provides a visual representation of the variable relationships discussed.

115

Figure 11. Hypothesized Model

See Table 17 for corresponding hypotheses with pathways.

Table 17. Second Study: Summary of Hypotheses and Results

Hypothesis Weight (β) p-value Supported?

H1a: DMR has a positive effect on IOP. 0.159 p < 0.001 Supported

H1b: DMR has a positive effect on IOT. -0.422 p < 0.001

Not supported; the

relationship is negative

H2b: DMI has a negative effect on IOP. -0.104 p < 0.001 Supported

H2b: DMI has a negative effect on IOT. 0.183 p < 0.001

Not Supported; the

relationship is positive

H3a: APC partially and positively

mediates the positive effect of DMR on

IOP.

0.054 p = 0.003 Supported

H3b: APC partially and positively

mediates the positive effect of DMR on

IOT.

-0.134 p < 0.001 Not Supported

116

H4: Trust positively moderates the effect

of DMR on IOP. -.103 p = 0.066

Not Supported, albeit

close to significant

H5: Trust positively moderates the effect

of APC on IOP. .087 p = 0.089

Not Supported, albeit

close to significant

H6: The positive effects of DMR on IOT

are stronger for women than men. (See

comment)

X2 test

p = 0.038

Supported;

Male (β = -0.397, p <

0.001)

Female (β = 0.063, p =

0.746)

H7a: Years in practice has a positive effect

on IOP. N/A p > 0.250 Not Supported

H7b: Years in practice has a positive effect

on IOT. N/A p > 0.250 Not Supported

H8a: Age has a positive effect on IOP. 0.115 p = 0.026 Supported

H8b: Age has a positive effect on IOT. -0.064 p = 0.196 Not Supported

H9a: The positive effects of DMR on IOP

is stronger for white physicians. N/A

Chi-

square

p = 0.358

Not Supported;

Invariant

H9b: The positive effects of DMR on IOT

is stronger for white physician. N/A

Chi-

square

p = 0.167

Not Supported;

Invariant

H9c: The positive effects of DMI on IOP

is stronger for white physician. N/A

Chi-

square

p = 0.872

Not Supported;

Invariant

H9d: The positive effects of DMI on IOT

is stronger for white physician.

N/A

Chi-

square

p = 0.083

Supported;

White (β = 0.264, p <

0.001)

Non-white (β = 0.088,

p = 0.363)

H10a: The positive effects of DMR on

IOP is stronger for physicians with more

education. N/A

Chi-

square

p = 0.244

Not Supported;

Invariant

H10b: The positive effects of DMR on

IOT is weaker for physicians with more

education. N/A

Chi-

square

p = 0.358

Not Supported;

Invariant

H10c: The positive effects of DMI on IOP

is stronger for physicians with more

education. N/A

Chi-

square

p = 0. 872

Not Supported;

Invariant

117

H10d: The positive effects of DMI on IOT

is weaker for physicians with more

education. N/A

Chi-

square

p = 0.113

Not Supported;

Invariant

DMR= Rational Decision Making

DMI = Heuristic Decision Making

APC = Patient-centric Approach to Treatment

IOP = Patient Participation with Treatment Protocols

IOT = Patient Participation with Treatment Tools

Physician’s Decision-Making Process Effects on Patient Participation

It is expected that physicians using a rational decision-making style (DMR) are

more likely to incorporate patient input into treatment and tools; they are more likely to

incorporate the patient’s input alongside their experience with similar cases. Conversely,

it is expected that physicians with using heuristic decision making (DMH) will reduce

participation in the decision process; they should intuitively apply what they believe to be

the best approach based on their observations.

Hypothesis 1. DMR has a positive effect on IOP (H1a) and IOT (H1b).

Hypothesis 2. DMH has a negative effect on IOP (H2a) and IOT (H2b).

Mediating Effects of Treatment Approaches to Patient Participation

The relationship between the IV and DV is partially mediated by a physician’s

patient-centric approach to treatment (APC). Although a physician might have a specific

decision-making style when treating a patient, it is not sufficient to explain patient

participation. Patient-centric physicians are more likely to incorporate non-biometric

patient feedback into their decision process because a patient-centric approach to

treatment is likely to value the patient just as much as the disease when deciding

treatment (Bensing, 2000). Patient-centric physicians will “have to explore patients’

needs from a biopsychosocial model, in which psychological, and social elements are

118

valued as important as the strictly biomedical elements” (Bensing, 2000: 21). Seeborg et

al. (2015) suggests that patient perceived health is a strong indicator of future health;

therefore, addressing the patient rather than the disease enables the physician to better

access the patient’s unmet needs. A patient-centric approach reduces the power

imbalance and facilitates participation. Therefore, APC physicians with a rational

decision-making style are likely to encourage patient participation because they are

reducing the power imbalance by incorporating the patient’s physical and psychosocial

health.

Hypothesis 3. APC partially and positively mediates the positive effect of DMR on

IOP (H3a) and IOT (H3b).

Physician Traits

Age

Delaney et al. (2015) found that as people age, they are more likely to use

intuitive decision making when compared to younger people. Participants from the first

study support this observation since older physicians intuitively focused on the individual

pharmacokinetics of each hemophilia patient. Therefore, it was hypothesized that older

physicians will generally encourage more patient participation.

Hypothesis 4a. Age has a negative effect on IOP.

Hypothesis 4b. Age has a negative effect on IOT.

Gender

Published results are mixed for determining whether gender is linked to patient

participation. Results from Street Jr et al. (2005) show that patient participation is a

function of many factors, excluding gender. Other studies suggest that female physicians

take a more participatory styles, such as empathy and information sharing, than males do

119

(Bertakis et al., 2002; Janssen & Lagro-Janssen, 2012; Roter et al., 2002). The working

hypothesis for this study assumes that men and women physician’s use patient

participation differently. The differences in approach imply that, although rational

decision-making should enhance participation, the impact would be greater for women.

Hypothesis 5. The positive effects of DMR on IOT is stronger for women than

men.

Race

Lawrence et al. (2015) found that, when treating depression, white physicians are

more likely to prescribe antidepressants than black physicians, whereas black physicians

were more likely to refer the patient to a counselor or psychiatrist. Lin and Kressin (2015)

found there is less conversation between physicians and patients if their races are

different, regardless of physician race. It is expected that white physicians will have

greater participation with patients than black physicians.

Hypothesis 6a. The positive effects of DMR on IOP is stronger for White

physicians.

Hypothesis 6b. The positive effects of DMR on IOT is stronger for White

physician.

Hypothesis 6c. The positive effects of DMI on IOP is stronger for White

physician.

Hypothesis 6d. The positive effects of DMI on IOT is stronger for White

physician.

Experience

Years of Practice: Marinova et al. (2016) found that experience plays a role in

decision making by increasing the value of patient medical criteria and decreasing the

value of patient preferences; experienced physicians use intuition to make decisions

120

based on patterns, not patient input. Furthermore, the more experienced physicians are

likely older and may have paternalistic habits from before the patient-centric movement.

Hypothesis 7a. Years in practice has a positive effect on IOP.

Hypothesis 7b. Years in practice has a positive effect on IOT.

Education: Schnell and Currie (2017) found that physicians from better medical

schools were less likely to prescribe opioids to patients, suggesting there is more patient

participation and SDM by extension. Therefore, it is hypothesized that physicians with a

greater level of education, Ph.D. and MD, will have more patient participation regardless

of decision-making style.

Hypothesis 8a. The positive effects of DMR on IOP is stronger for physicians with

more education.

Hypothesis 8b. The positive effects of DMR on IOT is stronger for physicians with

more education.

Hypothesis 8c. The positive effects of DMI on IOP is stronger for physicians with

more education.

Hypothesis 8d. The positive effects of DMI on IOT is stronger for physicians with

more education.

Trust

The moderating variables in this study explain “when” decision-making style (IV)

affected patient participation (DV). If physicians trust the information provided by

patients, they are more likely to share decision-making (McGuire et al., 2005). Physicians

will not likely incorporate patient participation if the information provided by the patient

is not trusted. For example, one physician in Chapter 4 discussed that trust is assumed but

erodes if the patient is not honest thereby reducing patient participation in the treatment

decision. Therefore, trust is expected to positively moderate patient participation.

121

Hypothesis 9a. TIP positively moderates the effect of DMR on IOP.

Hypothesis 9b. TIP positively moderates the effect of APC on IOP.

Methods

I conducted a series of analyses to construct a model that most accurately

represents to data and its relationships. The following sections describe the methods I

used to create Figure 12 and extract the findings summarized in Table 27.

Sample

During a one-week period in late May 2016, a survey was sent to 16,310

physicians (67% male, 33% female) who treated PID in the U.S. The services of IMS

Health, Inc. were used to distribute the survey. Physicians were given a blinded Internet

survey consisting of 76 questions. The administration of the survey through the Internet

contributed to confidentiality for the participants. The population was selected based on

the following criteria: respondents must be physicians who treat PID in the U.S. A total

of 350 (2.1%) completed the survey. The low response rate was likely due to doctor’s

demanding schedules and the high frequency with which they receive survey requests

(Flanigan, McFarlane, & Cook, 2008). Furthermore, the surveys in this study were sent

by email, which may have decreased the response rate. After screening the respondents

based on pre-defined criteria, such as omitting unengaged respondent data, a sample size

330 was analyzed. Table 18 displays a summary of the sample demographics. The low

response rate might be a concern about the representation of the study population.

However, respondents likely represent the population of doctors who treat PID since the

330 physicians treat approximately 55% (17,500 patients) of the PID population.

Responding physicians treat an average of five PID patients per month.

122

Table 18. Sample Characteristics

Data Screening

The data was screened for respondent’s responses and items with unacceptable

results. After screening the data, 20 respondents were removed leaving 330 respondents’

data for analysis. Eight were removed for completing the survey too quickly (less than

three minutes), six were removed for disengaged responses, and one (1) was removed for

answering the dummy question incorrectly. Then five were removed for an abnormal

Cook’s distance (explained in the following section).

All respondents completed more than 90% of the survey; this is ensured because

Qualtrics software has a forced response feature. There were five surveys with an

Respondent Characteristics Number Percent Respondent Characteristics Number Percent

Sample Size (n) 330 100% Gender

Ethnicity Male 244 73.9%

American Indian or Alaska Native 1 0.3% Female 86 26.1%

Asian 61 18.5% Age

Black or African American 5 1.5% <30 0 0.0%

Native Hawaiian or Pacific Islander 1 0.3% 30-39 63 19.1%

Other 21 6.4% 40-49 118 35.8%

Prefer not to say. 2 0.6% 50-59 104 31.5%

White 239 72.4% 60-69 42 12.7%

Specialty 70+ 3 0.9%

Oncology 83 25.2% Midwest 63 19.1%

Allergy 50 15.2% East North Central 37 11.2%

Internal Medicine 41 12.4% West North Central 26 7.9%

Family Medicine 22 6.7% Northeast 80 24.2%

Immunology 19 5.8% Middle Atlantic 56 17.0%

Pulmonology 19 5.8% New England 24 7.3%

Pediatrics 21 6.4% South 121 36.7%

Infectious Disease 11 3.3% East South Central 22 6.7%

Rheumatology 10 3.0% South Atlantic 72 21.8%

Hematology 7 2.1% West South Central 27 8.2%

Family Practice 5 1.5% West 66 20.0%

Otolaryngology 4 1.2% Mountain 24 7.3%

Other 32 9.7% Pacific 42 12.7%

123

acceptable number of missing values (<5%); the missing values were replaced with the

median because they were on an ordinal scale (Lynch, 2007). No outliers were removed,

except those with abnormal Cook’s distance (discussed in “Multivariate Assumptions”

below).

Skewness and kurtosis were measured to ensure the normality of the data. No

items were removed for skewness or kurtosis issues; those with an absolute value greater

than 1.0 had been monitored. Two items were expected to have high kurtosis values (>

3.0); therefore, they were kept and monitored for kurtosis along with three other items

with kurtosis values greater than 2.0 (Sposito, Hand, & Skarpness, 1983). The reliability

of the data was considered adequate because the Cronbach’s alphas were greater than 0.7.

Multivariate Assumptions

A Cook’s Distance test was performed to determine if any respondents’ data were

abnormally influential. Five respondents’ data were removed because their Cook’s

Distance was abnormal (> 0.05) and this may be because these respondents seemed

unengaged within each construct (i.e., answering all DMR items with “2” and APC items

with “4”, etc.). The model was tested for multicollinearity (highly correlated predictor

variables) by assessing the variable inflation factors (VIF). All predictor variables were

found to have VIF < 3.0 for both DVs, therefore, no multicollinearity was assumed (Hair,

Black, Babin, & Anderson, 2010).

Exploratory Factor Analysis

An EFA was performed to determine correlations among the variables. The

maximum likelihood extraction method was used along with a Promax (kappa=4)

rotation method. Twenty-three items were removed using EFA because they had primary

124

loadings lower than 0.5 or cross-loadings with less than 0.2 difference and two removed

items had low Cronbach’s alphas (< .60). Factor loadings represent how much a factor

explains a variable; the greater the number, the stronger the explanation.

The adequacy of the data is acceptable: the KMO is at .892 and is deemed

significant (p = .000); all the communalities are above .30 with the lowest at .31; these

six factors explain 49.02% of the variance. The eigenvalue for the six factors was 1.216

(3.685% variance; 57.748% cumulative). The factors found were: DMR, DMH, APC,

IOP, IOT, and TIP. There are 48 (9.0%) non-redundant residuals with absolute values

greater than .05; although, it is ideal to achieve less than 5% non-redundant residuals,

model fit can be met as long as there is less than 50% non-redundant residuals with

absolute values greater than .05 (Yong & Pearce, 2013). As evidence of convergent

validity, all item loadings are above .50 (Reio & Shuck, 2014). Evidence of discriminant

validity is that there were no strong cross-loadings and all cross-loadings were less than

0.20 (Costello & Osborne, 2005). The Cronbach’s Alphas for all factors are above .70;

therefore, the reliability is acceptable.

Confirmatory Factor Analysis

A confirmatory factor analysis (CFA) was conducted in SPSS AMOS version 23

using the pattern matrix from the EFA. The EFA was input into AMOS and produced

good model fit (CFI = 0.95, SRMR= 0.048, RSMEA= 0.04, PCLOSE= 1). A configural

invariance test was performed and obtained adequate goodness of fit when analyzing a

freely estimated model across two groups (CFI = 0.92, SRMR = 0.05, RSMEA = 0.04,

PCLOSE = 1). A metric invariance test was performed by constraining the two models to

be equal, and then a chi-squared difference test between the unconstrained and

125

constrained models was completed. The models between men and women were initially

found to be not invariant (p = .094). Two items had caused issues with model fit and were

removed to make the model invariant (cPATIENT_2 and cPATIENT_4); these two items

were chosen because they would get trimmed while testing reliability and validity

regardless of other removal options (e.g. DM_RAT_2 and DM_RAT_9). After removal,

the model was invariant, and the model fit had remained adequate (CFI = 0.92, SRMR =

0.05, RSMEA = 0.04, PCLOSE = 1).

The model has reliability as evidenced by a CR greater than .70 for all factors.

Items with the lowest regression weights for DMR and APC were removed until

discriminant validity was achieved; thus, five total items were removed. After removal,

convergent validity was achieved as evidenced by a square root of AVE that was greater

than any of the inter-factor correlations. Table 19 displays the results of the validity and

reliability tests.

Table 19. Reliability and Validity

CR AVE MSV MaxR(H) IOP TIP DMR DMH APC IOT

IOP 0.834 0.502 0.432 0.841 0.709

TIP 0.911 0.596 0.301 0.943 0.549 0.772

DMR 0.847 0.525 0.300 0.957 0.548 0.441 0.725

DMH 0.763 0.521 0.040 0.963 -0.100 0.028 -0.122 0.722

APC 0.771 0.530 0.432 0.967 0.657 0.379 0.539 0.040 0.728

IOT 0.757 0.511 0.124 0.970 0.352 0.115 0.065 0.200 0.235 0.715

Before testing for common method bias, one (1) item was removed for cross-

loading between factors. Common method bias was tested with a chi-squared difference

test between the unconstrained common method factor model and the fully constrained

zero common method factor model. The models, unconstrained and constrained, were

126

found to be substantially different from zero (p=.000), which is evidence that there is

significant shared variance (Chang, Van Witteloostuijn, & Eden, 2010; MacKenzie &

Podsakoff, 2012). Furthermore, a latent marker variable method was used to test for

common method bias; the chi-square difference tests between nested models

(unconstrained, equal, and zero) also resulted in evidence of common method bias

(Williams, Hartman, & Cavazotte, 2010). Therefore, the common method factor was

retained to create common method bias corrected measures. Model fit for the final

measurements (common method bias corrected) are adequate (CFI= 0.967, SRMR=

0.0353, RSMEA= 0.039, PCLOSE= 0.99). The remaining twenty-five (25) items were

imputed to construct the path model.

Table 20. Descriptive Statistics and Correlations Table of Variables

Mean Std.

Deviation

DMR DMH APC IOP IOT TIP

DMR 2.50 .41 1

DMH 2.18 .76 -.237** 1

APC 1.35 .30 .353** -.025 1

IOP 1.28 .38 .220** -.252** .200** 1

IOT .68 .67 -.305** .267** -.288** -.348** 1

TIP 2.77 .65 .269** -.001 .142** .524** -.153** 1

Note. ** indicates p < .010 for a two-tailed test.

Mediation

A path model was constructed from the imputed variables and had bad model fit

(CFI= 0.805, SRMR= 0.0601, RSMEA= 0.264, PCLOSE= 0). The R2 for IOP and IOT

are greater with the inclusion of the mediator; the mediator increased the r-squared of

IOP (.103) and IOT (.137) to .125 and .180, respectively. Adding the mediator caused the

127

model to have inadequate fit (CFI= 0.90, SRMR = 0.04, RSMEA = 0.24, PCLOSE= 0).

To open parameters, meanig to gain degrees of freedom and improve model fit, the

following pathways were trimmed for having no significant effect: AGE to APC and

AGE to IOP. This resulted in a good model fit (CFI = 0.999, SRMR = 0.0147, RSMEA =

0.013, PCLOSE = 0.595). The regression weights and p-values were calculated for the

direct (with and without mediator) and indirect effects of the following relationships:

DMR to IOP, DMR to IOT. Table 21 provides the results (Hayes, 2009, 2013). Both

relationships have partial mediation because all relevant pathways are significant (p <.

050 and p < .001).

Table 21. Mediation

Path

Direct w/o

Mediator Sig.

Direct w/

Mediator Sig.2 Indirect Sig.3 Type

DMR-APC-IOT -0.288 *** -0.175 *** -0.081 *** partial

DMR-APC-IOP 0.105 *** 0.113 ** 0.058 *** partial

Note: **p < .05 ***p < .001

Interaction (Moderation)

The following pathways were used to create standardized interaction variables in

SPSS and added to the model along with TIP: DMR to TIP, DMH to TIP, and APC to

TIP. The model fit is adequate after adding TIP and the new variables (CFI = 1, SRMR =

0.0094, RSMEA = 0, PCLOSE = 0.771). The direct weight, moderator’s weight, and the

interaction variable’s weight were used to calculate the interaction effect. The interaction

effects are not significant for following two relationships: DMR to IOP and ADC to IOP

(Hayes, 2013). Table 22 displays the calculation results.

128

Table 22. Interaction Effects

Path DMR --> IOP APC --> IOP

Moderator TIP TIP

Direct Weight -0.045 0.175

Mod. Weight 0.297 0.297

Interaction

Weight

-0.027 0.029

Interaction

Significance (p-

value)

0.066 0.089

Strength TIP strengthens the negative

relationship between DMR and IOP.

TIP strengthens the positive

relationship between APC and IOP.

Multi-group

A chi-squared difference test was performed to determine if certain groups of

physician affect patient participation differently. The chi-squared difference test was

conducted on the model level (overall difference) and the pathway level (specific

difference) (Byrne, 2004, 2008).

Sex

On the model level, male and female participants are not different (p= 0.442).

However, only one pathway was significantly different (p = 0.038): DMR to IOT (Table

23).

Table 23. Multi-Group Invariance Test by Pathway

Path Invariant? Chi-squared (p) Male β Male Sig. Female β Female Sig.

Model Yes 0.442 N/A N/A N/A N/A

DMR-->APC Yes 0.965 0.269 *** 0.270 ***

DMR-->IOT No 0.038 -0.397 *** 0.063 0.746

DMR-->IOP Yes 0.654 0.086 0.162 0.141 0.176

***p<.001

129

Race

On the model level, white and non-white participants are different (p = 0.061),

which is consistent with the literature. There are two (2) relationships in the model that

are different between races: DMI to IOT (p = 0.083) and AGE to IOP (p = 0.072). These

differences suggest white physicians using heuristic decision making increases patient

participation with treatment tools (β = 0.264, p < 0.001), whereas heuristic decision

making does not affect participation for non-white physician (β = 0.088, p = 0.363).

Additionally, age increases patient participation for white physicians whereas age does

not affect participation for non-white physicians. Table 24 represents the difference tests

and pathways between races.

Table 24. Invariance between White and Non-white Physician Groups

Pathway CMIN P White Physicians Non-White Physicians

Model --> Model 14.912 0.061 β = N/A, p = N/A β = N/A, p = N/A

DMR --> APC 1.426 0.232 β = 0.312, p = *** β = 0.461, p = ***

DMR --> IOP 0.846 0.358 β = 0.16, p = 0.012 β = 0.033, p = 0.783

DMI --> IOP 0.026 0.872 β = -0.21, p = *** β = -0.198, p = 0.06

DMR --> IOT 1.909 0.167 β = -0.132, p = 0.038 β = -0.322, p = 0.003

DMI --> IOT 2.999 0.083 β = 0.264, p = *** β = 0.088, p = 0.363

APC --> IOP 2.076 0.15 β = 0.217, p = *** β = 0.015, p = 0.898

APC --> IOT 0.248 0.618 β = -0.224, p = *** β = -0.167, p = 0.113

AGE --> IOP 3.229 0.072 β = 0.161, p = 0.004 β = -0.056, p = 0.571

***p<.001

Experience

Physician experience, as explained by years of practice and education, provided

interesting results. This section discusses physician education, whereas years of practice

is analyzed in the next section because it was used as a control variable. Overall, neither

education nor years of practice provided significance effects on factors that encourage

130

SDM, but the results of education groups is potentially an exception because of the small

sub-sample of Ph.D. physicians.

Although only 18 of the 330 physicians7

had a Ph.D., participants with Ph.D.s

have very different results. On the model level, Ph.D. and non-Ph.D. participants are not

different (p = 0.510). There are no relationships in the model that are different according

to the chi-squared difference test. However, Ph.D. physicians have a partial mediation

relationship between rational decision making and patient participation with protocols (β

= 0.372, p = 0.09) through a patient-centric approach to healthcare (β = 0.388, p = 0.069),

whereas non-Ph.D. physicians have an indirect relationship of the same pathway. This

suggests that Ph.D. physicians encourage patient participation and SDM by extension,

with or without a patient-centric approach. Additionally, age increases participation for

PhD physicians (β = 0.367, p = 0.055), whereas there is no effect for non-Ph.D.s (β =

0.083, p = 0.108). Table 25 represents the difference tests and pathways between races.

Table 25. Invariance between Ph.D. and Non-PhD Physician Groups

Pathway CMIN Pval Ph.D. Physician Non-Ph.D. Physician

Model level 7.251 0.51 β = N/A, p = N/A β = N/A, p = N/A

DMR --> APC 0.04 0.842 β = 0.372, p = 0.09 β = 0.349, p = ***

DMR --> IOP 1.359 0.244 β = 0.388, p = 0.069 β = 0.086, p = 0.14

DMI --> IOP 0.026 0.872 β = -0.137, p = 0.506 β = -0.235, p = ***

DMR --> IOT 0.846 0.358 β = -0.118, p = 0.575 β = -0.18, p = 0.001

DMI --> IOT 2.505 0.113 β = -0.116, p = 0.554 β = 0.228, p = ***

APC --> IOP 0.07 0.792 β = 0.244, p = 0.243 β = 0.162, p = 0.004

APC --> IOT 0.438 0.508 β = -0.495, p = 0.019 β = -0.215, p = ***

AGE --> IOP 1.557 0.212 β = 0.367, p = 0.055 β = 0.083, p = 0.108

***p<0.001

7

A similar small sample of physicians that participated in clinical studies (16 of 330).

131

Controls

Age

The analysis was controlled for age (AGE). There was no significant effect

towards APC (β = -0.037, p = 0.477) and IOT (β = -0.064, p = 0.196) from age. However,

there is a significant regression weight from age to IOP (β = 0.115, p = .026). This may

indicate that older physicians are more likely to encourage patient input when they are

deciding what protocols for treatment will be administered.

Years of Practice

Contrary to expectations, physician years of practice treating PID did not have

any significant effect on any factor. Table 26 provides the regression weights of each

variable I tested experience with; all relationships were insignificant (p-value > 0.25).

Table 26. Influence of Physician Years of Practice

X Y Std Reg. Pval

Years of practice DMI 0.065 0.415

Years of practice DMR 0.047 0.548

Years of practice APC 0.079 0.285

Years of practice IOP -0.022 0.77

Years of practice IOT -0.049 0.497

Hypotheses Results

Hypothesis 1a. DMR has a positive effect on IOP.

As expected, the DMR physicians are likely to incorporate patient participation in

their decision process regarding protocols.

132

Hypothesis 1b. DMR has a positive effect on IOT.

The analysis provided counter-evidence to this hypothesis. Contrary to

expectations, DMR physicians are not likely to incorporate patient participation in their

decision process in regard to tools.

Hypothesis 2a. DMH has a negative effect on IOP.

As expected, the DMH physicians are not likely to incorporate patient

participation in their decision process regarding protocols.

Hypothesis 2b. DMH has a negative effect on IOT.

The analysis provided counter-evidence to this hypothesis. Contrary to

expectations, DMH physicians are likely to incorporate patient participation in their

decision process in regard to tools. Therefore, I reject the hypothesis.

Hypothesis 3a. APC partially and positively mediates the positive effect of DMR

on IOP.

As expected, a patient-centric treatment approach is likely to mediate the pathway

between DMR and IOP. In this case, the mediation is positive; thus, enhancing the

pathway’s effect.

Hypothesis 3b. APC partially and positively mediates the positive effect of DMR

on IOT.

As expected, a patient-centric treatment approach is likely to mediate the pathway

between DMR and IOP. Although the mediating effect is negative, the negative effect of

DMR on IOT was weakened by APC.

133

Hypothesis 4a. Age has a positive effect on IOP.

As expected, older DMR physicians have a stronger positive effect on IOP (β =

0.115, p = 0.026). Therefore, the hypothesis is supported.

Hypothesis 4b. Age has a positive effect on IOT.

Contrary to expectations, age did not have any significant effect on patient

participation with treatment tools. Therefore, this hypothesis is rejected.

Hypothesis 5. The positive effects of DMR on IOT is stronger for women than

men.

As expected, female DMR physicians have a stronger positive effect on IOT than

men. However, the results provided interesting evidence in that the female physicians did

not have a strong positive relationship between DMR and IOT (β = .063, p = .746),

whereas male physicians had a strong negative relationship (β = -.397, p < .001).

Therefore, the hypothesis is supported.

Hypothesis 6a. The positive effects of DMR on IOP is stronger for white

physicians.

Contrary to expectations, the effect of DMR on IOP is invariant between races;

the chi-squared difference test was not significant (p = 0.358). Therefore, this hypothesis

is rejected.

Hypothesis 6b. The positive effects of DMR on IOT is stronger for white

physicians.

Contrary to expectations, the effect of DMR on IOT is invariant between races;

the chi-squared difference test was not significant (p = 0.167). Therefore, this hypothesis

is rejected.

134

Hypothesis 6c. The positive effects of DMI on IOP is stronger for white

physicians.

Contrary to expectations, the effect of DMI on IOP is invariant between races; the

chi-squared difference test was not significant (p = 0.872). Therefore, this hypothesis is

rejected.

Hypothesis 6d. The positive effects of DMI on IOT is stronger for white

physicians.

The effect of DMI on IOT is not invariant between white and non-white

physicians; the chi-squared difference test was not significant (p = 0.083), confirming the

differences between races. Furthermore, white physicians had a strong, significant

pathway (β = 0.264, p < 0.001) compared to non-white physicians (β = 0.088, p = 0.363).

Therefore, this hypothesis is accepted.

Hypothesis 7a. Years of practice has a positive effect on IOP.

Contrary to expectations, years of practice did not have any significant effect on

patient participation with treatment protocols. Therefore, this hypothesis is rejected.

Hypothesis 7b. Years of practice has a positive effect on IOT.

Contrary to expectations, years of practice did not have any significant effect on

patient participation with treatment tools. Therefore, this hypothesis is rejected.

Hypothesis 8a. The positive effect of DMR on IOP is stronger for physicians

with more education.

Contrary to expectations, the effect of DMR on IOP is invariant between Ph.D.

physicians and non-PhD physicians; the chi-squared difference test was not significant (p

= 0.244). Therefore, this hypothesis is rejected. However, because the sample of Ph.D.

135

physicians is only 18, it is worth noting that the pathway for Ph.D. physicians is

significant (β = 0.388, p = 0.069) whereas it is not significant for non-Ph.D.s (β = 0.086,

p = 0.140).

Hypothesis 8b. The positive effect of DMR on IOT is weaker for physicians with

more education.

Contrary to expectations, the effect of DMR on IOT is invariant between Ph.D.

physicians and non-PhD physicians; the chi-squared difference test was not significant (p

= 0.358). Therefore, this hypothesis is rejected. However, the pathway for Ph.D.

physicians is not significant (β = -0.118, p = 0.575) whereas it is significant for non-

Ph.D.s (β = -0.180, p = 0.001).

Hypothesis 8c. The positive effects of DMI on IOP is stronger for physicians

with more education.

Contrary to expectations, the effect of DMI on IOP is invariant between Ph.D.

physicians and non-PhD physicians; the chi-squared difference test was not significant (p

= 0.872). Therefore, this hypothesis is rejected. Furthermore, the pathway for Ph.D.

physicians is not significant and negative (β = -0.137, p = 0.506) whereas it is significant

for non-PhDs (β = -0.235, p < 0.001).

Hypothesis 8d. The positive effects of DMI on IOT is weaker for physicians with

more education.

Contrary to expectations, the effect of DMI on IOP is invariant between Ph.D.

physicians and non-Ph.D. physicians; the chi-squared difference test was not significant

(p = 0.113). Therefore, this hypothesis is rejected. Furthermore, the pathway for Ph.D.

136

physicians is not significant and negative (β = -0.116, p = 0.554) whereas it is positive

and significant for non-PhDs (β = 0.228, p < 0.001).

Hypothesis 9a. TIP positively moderates the effect of DMR on IOP.

This hypothesis is not supported in the analysis; however, the analysis provided

interesting results. Trust strengthens the negative relationship between DMR and IOP, yet

the relationship is positive when trust is not present. This may suggest that DMR

physicians have lower levels of patient participation for protocols when they trust the

patient’s initial input. It is possible that DMR physicians do not feel the need to

incorporate patient participation if all the information is accurate and reliable from the

start.

Hypothesis 9b. TIP positively moderates the effect of APC on IOP.

This hypothesis is not supported in the analysis. TIP does not significantly

moderate the relationship between APC and IOP.

Discussion

Based on literature and results from the prior research, this study sought to test the

effect of physician decision-making style on patient participation. Patient participation is

often cited as an important factor to improve healthcare outcomes, so a deeper

understanding of factors that foster shared decision-making may help facilitate

improvements in the care of chronic diseases. The aim of the study was primarily focused

on quantifying the amount of patient participation in the decision process explained by a

physician’s decision-making style. As suggested by dual process theory, the results of the

study demonstrate that physicians’ decision-making style has a meaningful impact on the

patient participation in selecting treatment protocols and tools.

137

The study also focuses on the role of patient-centric care as predicted by SIT and

physician traits. Consistent with SIT theory, patient-centered care is the mechanism

through which SDM words. There are also relevant traits which impact decision making.

Summary of Major Findings

The results highlight how different decision-making styles (dual process) have the

potential to improve patient participation for tools or protocols (not both). Rational

decision making leads to participation with protocols whereas heuristic decision making

increases participation with treatment tools.

Social Identity Theory is supported because patient participation is mediated by

the patient-centric approach to healthcare, increasing patient participation likelihood

regardless of decision-making style.

There were a number of findings related to physician traits. Older physicians are

more likely to increase patient participation, yet years of practice do not affect

participation and trust has no moderating effect. Physician race provided different results

in the effects on patient participation; white physicians were more likely to increase

participation than non-white. Lastly, physician experience did not affect patient

participation; physicians with Ph.D.s were more likely to increase participation with

treatment protocols than non-Ph.D.s, but there is no statistical difference, likely caused by

the small Ph.D. population.

138

Figure 12. Second Study: Structural Equation Model Result

Table 27. Study 2: Expected vs Unexpected Results

Expected Results Unexpected Results

Physicians with a rational decision-making

style (DMR) are more likely to incorporate

patient participation for protocols (IOP).

Physicians with a heuristic decision-making

style (DMH) are less likely to incorporate

participation for protocols.

Physician treatment approach plays a role in

encouraging patient participation for DMR

physicians.

White physicians had more patient

participation than non-white physicians.

Older physicians with PhDs had more patient

participation than non-PhD physicians.

DMR physicians are less likely to incorporate

patient participation for the choice of tools

(IOT).

Physicians with a heuristic decision-making

style are more likely to incorporate

participation for the choice of tools.

The physician’s trust in patient-given

information reverses the direct relationship

between DMR and IOP from positive to

negative.

Physician Years of Practice did not

significantly affect patient participation.

DMR for male physicians decreases IOT,

whereas the pathway is not significant for

female physicians.

139

Expected Findings

The results suggest that physician decision-making style significantly effects

patient participation in the decision process. The results provide evidence that different

decision-making styles increase or decrease how much shared decision-making occurs. In

particular, physicians are more likely to exercise shared decision-making for treatment

protocols if they have a rational decision-making style (DMR), as opposed to the

heuristic decision-making style (DMH). This was not a surprise as we had hypothesized

that a rational decision-making style would have a positive direct effect on patient

participation with the choice of protocols (β = 0.105, p < 0.001). It was critical to

understand the fundamental differences between rational and heuristic thinkers to form

the hypotheses. We had considered hypothesizing that a rational decision-making style

would negatively affect participation with protocols (IOP), but ignoring details is the

opposite of what a rational decision maker would do. Although physicians do form

mental shortcuts, those who prefer a rational decision-making style are less likely to miss

a chance to get the patient’s perspective.

Furthermore, consistent with SIT, physicians that adopt a patient-centric approach

(APC) to treatment are more likely to incorporate patient participation in the choice of

protocols (β = 0.053, p = 0.003).

Few of the expected results involving physician traits were supported. Older

DMR physicians have a stronger positive effect on IOP. Female DMR physicians have a

stronger positive effect on IOT than men. The positive affect of DMI on IOT is stronger

for white physicians.

140

Unexpected Findings

The most interesting results of this study come from the unexpected findings.

Patient participation with tools (IOT) has provided, by far, the most results with counter-

evidence in this study. The direct relationships to IOT were significant, yet the opposite

of what was expected; DMR physicians were less likely to incorporate participation (β = -

0.288, p < 0.001), whereas DMH physicians were more likely (β = 0.195, p < 0.001).

This is the opposite of what we expected and resulted in an interesting relationship with

the mediator, patient-centric approach. We hypothesized a patient-centric approach

would positively affect DMR to IOT, but the result was significant negative mediation (β

= -0.134, p = .001). Although it is negative, the effect of APC on this relationship

actually increases the likelihood of participation by dampening the negative effect of

DMR on IOT.

The negative relationship between DMR and IOT is likely related to the overall

process of prescribing treatment tools. The physician will often prescribe a tool but be

uninvolved with the product selection (i.e., brand). When given a prescription, the patient

will go to their pharmacy to obtain the product; the pharmacist often makes the actual

brand selection. The tool may depend on various factors such as availability and type of

health insurance reimbursement.

It may be that physicians choose a decision-making style depending on how well

they know the patient and the extent of their patient-centric their healthcare approach.

New patients would be treated with a rational decision-making style, focusing on

protocols and compliance with protocols; this would involve the physician learning about

the patient lifestyle and background. The new patients may not know which brand works

141

best for them, neither does the physician. Therefore, the physician may be indifferent to

the brand of product as long as the patient follows the protocol. As the physician-patient

relationship matures, the physician would be more familiar with the patient’s lifestyle and

would not need too many to analyze it in detail. However, the patient can now provide

feedback on the brand to treat their PID. Further research would be needed to confirm

this hypothesis.

Gender had provided interesting counter-evidence to Hypothesis 6, “The positive

effect of DMR on IOT is stronger for women than men”. As expected, male DMR

physicians have a significant negative relationship to IOT (β = -0.397, p < .001);

however, female DMR physicians did not have a significant positive relationship (β =

0.105. p = 0.746). The literature suggests that female physicians are more likely to show

empathy than males. Therefore, the negative relationship between male DMR physicians

and IOT is consistent with the literature and supports the hypothesis. However, the

surprising outcome is the insignificant relationship of female DMR physicians and IOT.

This indifference may be caused by the smaller sample size of women (86) than men

(244); the sample size may provide an inadequate measure for women.

Physician traits had some influence on patient participation in specific contexts.

There was significance with physician age effecting patient participation with treatment

protocols. Physicians with Ph.D.s were more likely to have patient participation with

protocols than non-Ph.D. physicians. White physicians have more patient participation

with treatment tools than non-white physicians. Despite the other traits presenting unique

influences, years of practice does not affect patient participation.

142

The physician’s trust in patient-given information (TIP) had the most interesting

result by far. We hypothesized TIP would positively moderate the positive relationship

between DMR and IOP. However, TIP ended up reversing the relationship between DMR

and IOP; the direct relationship between DMR and IOP is positive (as expected), but

including trust in the model caused the relationship to become negative. The reason for

this reversed relationship may be related to what was discussed above; the physician may

not need to discuss protocols if the information has not changed. If a patient’s

information is accurate about his lifestyle, the physician would not need to revise the

protocols. However, if the patient provides inaccurate information, the physician may

need to compensate for any shortcomings.

Observations that Require Future Research

Certain survey items were tested for significance towards the DVs. For instance,

we asked about the physicians considering patient culture in the decision process, but

there was no significance. However, the item was part of the patient-centric factor,

suggesting they cumulatively have significance.

There were also items cut from the final model, such as those measuring if the

physician is disease-centric; meaning they approach the patient based on their disease,

not as a patient (lifestyle, etc.) with the disease. An individual item, “I tend to...Keep the

conversation focused on the disease”, was significant to participation with protocol

management (β = -0.163, p = 0.034) and nearly significant with protocol options (β = -

0.137, p = 0.073); it suggests that disease-centric patients were likely to decrease patient

participation. However, individual items are not sufficient to measure with factors

consisting of at least three (3) items.

143

Limitations

Limitations of the second study primarily relate to the use of a one-time self-

administered survey that assumes participants provided unbiased answers (Babbie,

Halley, & Zaino, 2007). The chosen sample limited potential generalization. The gender

distribution was mostly male (67%) and did not account for the practice setting (e.g.,

hospital vs doctor’s office). Participants’ experience levels, specialty, education, cultural

background, and other factors may have prompted variability in the responses.

No statistical test can assure a bias-free analysis (Podsakoff, MacKenzie, Lee, &

Podsakoff, 2003). The response rate of 2% out of 16,000 physicians was exceedingly

low. Due to the variable nature of the population (different physician specialties), the

study’s overall validity may be in question, even though doctors are treating the same

disease.

Based on the findings, the use of the dual process theory to explain physician

decision-making is limited. Although there was statistical significance between decision-

making styles and patient participation, the correlation coefficients were low. In other

words, the R squared value was less than .20 meaning patient participation is mostly

explained by other factors. Items in the analysis were mechanistically trimmed and as a

result, more than half of the items were removed (35 of 60). Future versions of the survey

must be shortened to no more than 5 items per expected factor to increase the response

rate and decrease the number of disengaged respondents.

Forming a survey of PID physicians from the findings of a qualitative research

study involving hemophilia physicians may have caused the researchers to overlook key

differences with PID (Lamb et al., 2015; Lamb, Lyytinen, & Wang, 2016). Second, the

144

results did not account for differences in physicians’ subspecialty (immunology vs.

oncology), the role of insurance in the physician decision process, and selection bias

(Certo, Busenbark, Woo, & Semadeni, 2016; Lamb et al., 2016); selection bias means the

sample randomization is limited by factors such as group (e.g race or specialty) and time

interval (e.g. summer or winter).

Another limitation of this study is related to chronic care in general. Chronic care

involves treatment and management over a long period of time, therefore a single survey

may not capture the full influence of factors affecting patient participation. A longitudinal

study would be best for future research.

Future Research

Future research can expand on the current findings by revising the model or

expanding the population. The data can be tested again by selecting different independent

variables; one variable, in particular, is trust. The survey can be administered to

physicians from other countries as well as for the treatment of other chronic diseases.

An addition to the survey could be to ask for the number of interactions the

patient has with the physician. It is apparent that—with chronic diseases—the physician-

patient interaction is not on a case-to-case basis, but a relationship that is built over time.

This information would help us observe the data as different groups of physicians and

patients based on the length of their relationships. Furthermore, this could offer insight

into how decisions are made at different stages of the relationship and the amount of

patient participation at each stage.

Multiple gaps in the second study warrant additional research.

145

Ethical Assurances

Prior to any data collection, the project was reviewed and approved by the Case

Western Reserve University Weatherhead School of Management and an institutional

review board (IRB) against the privacy and license requirements of U.S. data, protection,

and privacy rules. During the data collection process, participants were informed about

the purpose of the research, ethical procedures, and issues regarding anonymity.

Participants were also informed that use of the collected survey data was only for the

proposed study. Any specific and identifiable survey data was kept in a locked container

only accessed by the researcher. Outsiders to the dissertation process will not have access

to any specific information provided by participants to ensure privacy and to minimize

potential risk to those participating in the data collection. Participants received a standard

waiver form and assurance that they had the option of withdrawing from the study at any

point.

Conclusions and Implications for Patient Participation

“Let’s give patients the choice” is a frequent mantra in PID communities (Samaan

et al., 2014). Often cited principals of care require patient choice in treatment protocols

and tools (Chapel et al., 2014). However, patient choice must be seen in the context of

shared decision making between a doctor and patient. Using constructs and relationships

derived from the early qualitative study (First Study), a quantitative study was performed

to explore the factors that influence a physician’s decision-making process to include

patient participation in the treatment of PID. The chosen instruments provided a

systematic method for looking at the perspective of physicians by measuring attitudes

relating to shared decision-making. The sample of physicians who treat PID provided a

146

specialized perspective about the factors influencing a subset of doctors who treat a

chronic and rare disease. Structural equation modeling (SEM) is a multivariate statistical

analysis technique that was used to analyze factor relationships. As predicted by the

literature, the results showed statistically significant relationships between a physician’s

decision-making process, treatment approach, and patient participation; whereas

physician’s trust in the patient as well as the physician’s gender play less of a role than

previously stated. A physician-only trait that affects patient participation is age, wherein

older physicians increase participation with protocols; all other traits do not significantly

affect participation.

If patient participation is to be facilitated in the treatment of PID, greater attention

needs to be paid to the decision-making process of the physician. Government

policymakers, health care providers, patient organizations, and drug companies should

consider the types of interventions in which physician decision making can increase

patient participation in the treatment of PID and other chronic diseases. Furthermore,

physicians should reflect on their decision-making processes to improve patient

participation thereby improving clinical outcomes and perceived health.

147

CHAPTER 7: STUDY 3: A QUALITATIVE STUDY OF IMMUNOLOGISTS

THAT TREAT PID AND THE FACTORS THAT EFFECT SHARED DECISION

MAKING WITH THEIR PATIENTS

The results from Study 1 and Study 2 suggest: (1) physicians in the U.S. are more

patient-centric than the U.K.;8

(2) patient-centric care is the mechanism through which

SDM works; and (3) dual process theory is a relevant framework for understanding how

decision theory can be applied to SDM. However, the second study had a low explained

variance for the DVs (R2 < .2) meaning patient participation is mostly explained by other

factors, despite the significance between decision-making styles and patient participant (p

< 0.001; see Figure 12). Both a post hoc analysis of the second study and a qualitative

study of a specific subspecialty, immunology, were used to improve the understanding of

SDM.

Initial interviews to test the interview guide (discussed later in this chapter)

suggested that physician specialty may impact the relevance of significant factors found

in the second study. Focusing on immunology as a specific specialty would allow a more

focused view of the physician decision making phenomenon to better understand other

factors that influence SDM and how these factors can be integrated into an overall

decision-making framework. Therefore, I conducted a reanalysis (aka post hoc) to

analyze a sub-sample of physicians (immunologists) in the survey data of the second

study. This post hoc analysis was combined with a second qualitative study for a new

8

The U.S. performs better than other countries regarding the doctor–patient relationship, SDM, and

chronic disease management Schneider, E. C., Sarnak, D. O., Squires, D., Shah, A., & Doty, M. M. 2017.

Mirror, mirror 2017: The Commonwealth Fund, July 2017. Available from

http://www.commonwealthfund.org/publications/fund-reports/2017/jul/mirror-mirror-international-

comparisons-2017..

148

third study. The post hoc analysis was followed by a qualitative study of immunologists

to corroborate results and expand the SDM theoretical framework.

Post Hoc Reanalysis

A reanalysis was conducted to target the immunologist sub-sample to better

understand potential differences in patient participation. Instead of SPSS AMOS, the

analysis was conducted using SmartPLS software because of the small sample size

described below.

Post Hoc Methods

The sample of the quantitative reanalysis is 72 of the 330 (21.8%) completed

surveys; the physicians that specialize in allergy and immunology. Although the sample

size is quite low, it represents a significant part of the PID treating immunologists and the

treated PID population. The total PID population is approx. 32,000 (Grifols, 2017). The

sample of 72 physicians who participated in the survey treated between ~3,120 to ~9,720

patients per year (43–135 per physician). These physicians treat between 10% and 30%

of the total PID population.

Following the new factor analysis utilizing the same guidelines as the second

study, a new quantitative model was formed using revised factors. Each factor is

represented by at least three items9

and there is sufficient model fit (SRMR = 0.091)

(Iacobucci, 2010). The convergent validity was indicated by the composite reliability

loadings (CR > 0.7) despite the low average variance extracted (AVE < .5) (Wong,

2013). Lastly, the discriminant validity is sufficient according to the Heterotrait-

9

The items that form each factor are represented in Appendix D.

149

Monotrait (HTMT) test (Henseler, Ringle, & Sarstedt, 2015); the Fornell-Lacker criterion

was borderline sufficient because the value between rational decision making and patient-

centrism (0.628) was greater than the square-root of the AVE (.624). Table 28 represents

the abbreviations used in the post hoc analysis. Figure 13 is the pathway model of the

reanalysis, representing the results from the 72 allergist-immunologist participants; Table

29 summarizes the observed values of the path model.

Table 28. Post Hoc Abbreviations

Name Abbreviation

Rational Decision Making DM_R

Heuristic Decision Making DM_H

Patient-centric Approach A_PC

Protocol Participation IOP

Tool Participation IOT

Figure 13. Model of Post-Hoc Analysis of Second Study

150

Table 29. Pathway Values for Figure 13

Pathway Path Coefficient and P-Value

A_PC --> IOP β = 0.756, p = ***

A_PC --> IOT β = 0.951, p = ***

DM_R --> A_PC β = 0.581, p = ***

DM_R --> IOP β = 0.087, p = 0.181

DM_R --> IOT β = -0.331, p = 0.337

DM_H --> A_PC β = -0.138, p = 0.185

DM_H --> IOP β = 0.053, p = 0.997

DM_H --> IOT β = 0.327, p = 0.041

*** p-value < 0.000

Post Hoc Findings

Finding 1: Rational decision making has no direct effect on protocol

participation or tools participation.

The rational decision making style of immunologists does not directly affect

protocol participation (β = 0.087, p = 0.181) or tool participation (β = -0.331, p = 0.337).

Finding 2: Heuristic decision making increases with tool participation but does

not affect protocol participation.

The heuristic decision making style of immunologists directly increases tool

participation (β = 0.327, p = 0.041), but does not directly affect protocol participation (β

= 0.053, p = 0.997).

Finding 3: Patient-centrism positively mediates the effects of rational decision

making on protocol participation and tool participation.

The immunologist using a patient-centric approach positively mediates the

positive relationship to protocol participation (β = 0.243, p = 0.001) which made it

significant and positive. The physician using a patient-centric approach positively

151

mediates the negative relationship to tool participation (β = 0.278, p = 0.000) which made

it significant and positive.

Table 30. Indirect Effect: Mediation of Rational Decision Making

through Patient-centrism

Indirect Pathway Path Coefficient and P-Value

DM_R A_PC IOP β = 0.243, p = 0.001

DM_R A_PC IOT β = 0.278, p = 0.000

Finding 4: Patient-centrism does not mediate the effects of heuristic decision

making on protocol participation and tool participation.

Immunologists with a heuristic decision making style do not have mediated

relationships with protocol participation (β = -0.083, p = 0.217) and tool participation (β

= -0.095, p = 0.219).

Table 31. Indirect Effect: Mediation of Heuristic Decision Making

through Patient-centrism

Indirect Pathway Path Coefficient and P-Value

DM_H A_PC IOP β = -0.083, p = 0.217

DM_H A_PC IOT β = -0.095, p = 0.219

Finding 4: Race of the immunologist does not affect protocol participation or

tool participation.

White and non-white physicians do not affect patient participation differently (p >

0.100). The different pathways are represented in Table 32.

152

Table 32. Influence of Race on Patient Participation

Pathway Difference in Path Coefficient and P-Value

A_PC --> IOP β = 0.101, p = 0.712

A_PC --> IOT β = 0.167, p = 0.550

DM_R --> A_PC β = 0.178, p = 0.501

DM_R --> IOP β = 0.048, p = 0.874

DM_R --> IOT β = 0.061, p = 0.860

DM_H --> A_PC β = 0.243, p = 0.383

DM_H --> IOP β = 0.304, p = 0.314

DM_H --> IOT β = 0.192, p = 0.533

*** p-value < 0.000

Finding 5: Sex does not affect protocol participation or tool participation.

Male and female immunologists do not affect patient participation differently (p >

0.100). The different pathways are represented in Table 33.

Table 33. Influence of Immunologist Sex

Pathway Difference in Path Coefficient and P-Value

A_PC --> IOP β = 0.387, p = 0.274

A_PC --> IOT β = 0.440, p = 0.231

DM_R --> A_PC β = 0.210, p = 0.364

DM_R --> IOP β = 0.638, p = 0.115

DM_R --> IOT β = 0.337, p = 0.419

DM_H --> A_PC β = 0.250, p = 0.445

DM_H --> IOP β = 0.095, p = 0.816

DM_H --> IOT β = 0.015, p = 0.960

*** p-value < 0.000

Summary of Post Hoc

The post hoc reanalysis of data collected during the second study offers a more

accurate portrayal of decision making for chronic diseases such as PID. The

153

immunologists are the specialists responsible for treating a patient with PID throughout

most of the patient’s life, thereby providing better insight of potential SDM for this

chronic disorder. The findings suggest the following: rational decision making indirectly

increases patient participation with tools and protocols through a patient-centric

approach, heuristic decision making directly increases patient participation with tools,

and traits such as race and sex do not impact the model.

In summary, patient participation depends on a combination of decision-making

style and the approach to patient care. Rational decision makers only increase patient

participation if they incorporate a patient-centric approach to healthcare, which suggests

there is a conditional relationship involved. Therefore, the post hoc analysis was

supplemented by a qualitative study to better understand other factors associated with

SDM implementation in the immunologist population. The research questions were:

“What are the drivers of successful implementation and adoption of

SDM?

How are power and literacy balanced through health literacy

Are there patient traits that influence SDM

Are there organizational factors that influence SDM such as

coordination of care and reimbursement?

These questions guide the research to determine “when” SDM is effective, rather

than the results of SDM because the benefits are understood (Noonan et al., 2017). In this

chapter, I present the method, results, discussion, and limitations for the study.

Research Method of the Qualitative Part of the Third Study

The research is based on interviews with 15 immunologists conducted in between

May 1st and July 31st, 2017. I achieved a saturation point with 15 interviews despite

154

suggested samples sizes from the literature of 20 or more participants (Creswell, 2013a;

Morse, 1994); the variation of responses to my questions had diminishing returns,

especially after the first 10 interviews (Mason, 2010). Twelve were men and three were

women. The participants actively treat patients with PID, and their patient samples range

from approximately 100 to 4000, meaning the participants in this post hoc quantitative

reanalysis treat at least 21.6% to 42.3% of the total PID population in the U.S. All

participants are highly educated; all have their MD, five have a Ph.D. The average

number of years practicing is approximately 15 years with a standard deviation of 7

years. The participants were guaranteed anonymity as well as anonymity of their

association. Each interview lasted between 30 to 90 minutes depending on scheduling

and the flow of conversation. I recorded each interview and used a third-party service

(Rev.com) to transcribe the recordings, then coded responses using a multi-stage open

coding.

Table 34. Study 3: Sample Details

Category Amount %

Total 15 100.00%

Male 12 80.00%

Female 3 20.00%

White 11 73.33%

Non-white 4 26.67%

Age: 40s 1 6.67%

Age: 50s 9 60.00%

Age: 60s 5 33.33%

PhD 5 33.33%

North-East 4 26.67%

South East 1 6.67%

Midwest 8 53.33%

West Coast 2 13.33%

155

Reliability and validity issues were addressed based on Silverman’s guidance

(Silverman, 2015). To enhance reliability, the interview protocol was pilot tested (i.e.,

field-tested). A sample of three (3) healthcare professionals that treated PID (non-

specialists) assessed the clarity, appropriateness, and relevance of the interview

questions; revised interview protocols were created with the initial feedback. The

interviews were recorded and transcribed to ensure fidelity of the data. Exact quotes from

participants were used to state findings.

Using grounded theory as the analytic methodology to analyze the interview data,

1,389 initial codes were extracted. The coding involved primarily “identifying, analyzing

and reporting patterns (themes) within data” (p. 79) where the “a theme captures

something important about the data in relation to the research question and represents

some level of patterned response or meaning within the data set” (p. 82) (Braun &

Clarke, 2006). Codes were derived from participant's words and were added or modified

as necessary when new meanings or categories emerged. The subsequent phases

consolidated the raw data into five codes discussed in the findings section below.

Findings

The physicians interviewed acquired their knowledge and skills through a

combination of their education, training, experience, and the literature. This medical

background was similar to the hemophilia physicians in Study 1. Physicians provided an

insightful perspective to their decision-making process and potential factors that

influence their use of SDM.

156

Figure 14. Theories that Predict SDM

Findings Related to Dual Process

Long Diagnosis Period

PID is notorious for long diagnosis timeframes averaging about 5 to 7 years. All

physicians that I asked about diagnosis timelines (7 of 7) confirmed this and clarified the

context to delayed diagnosis. The physicians believed that the delay is caused by

imperfect data, insufficient screening processes of PCPs, and awareness. One participant

discussed the context of the 7-year diagnosis average caused by the regions and an

imperfect research method:

It varies from one region to another. In rural areas, yes. In major cities like

NY, Toronto, LA, no. The moment you start lumping up different regions,

you are not going to solve well what is behind it. I would say that if there is

a delay of treatment I don’t see very much of this in our place. We need to

study it more carefully. Nothing is simple. If you say a delay of diagnosis

in [IA], I say no. if you say overall PID, possibly yes; not because only

knowledge, but progress of the field. We identified [IA] that had infections

for a long time that never had PID, but even if we did, I doubt we would

have managed to label that way because our diagnostic tools are much better

today. Also do you include autoimmunity in that category? Or cancer? The

SDMDual-Process

Power Balance

Traits (Patient/Physician)

Organizational Context

Feedback

157

delay in diagnosis for PID is hard for me to accept 7 years. It was a survey,

not a study. Part of pushing the enzyme issue. We all support it, but it is not

studied.

(CR)

Another participant believes the disease is overlooked in the primary care setting:

Often times that's because people don't get in to see a specialist in primary

immunodeficiency. I think that many patients who talk about these

diagnostic delays will talk about, "Oh, it was this breath of fresh air when I

got to see Dr. X." Well Dr. X was just somebody who's trained in this

process of true pattern recognition, and has the 90 minutes to go through

and do it, as opposed to community-based allergist that's trying to fit this

into an otherwise 20-patient workday. I also think that ... That's one reason

for diagnostic delay, getting to the true specialist. The other is that some of

these diagnoses do evolve over time, so that when you see someone at point

A, the laboratory tests may not necessarily have caught up to what their

history is, and some of that evolution does happen over time as well. So

those two reasons.

(JO)

Lastly, one participant believed the awareness of PID contributes to the delay:

I think, again, another kudos to the advocacy networks like Jeffrey Modell

Foundation, Immune Deficiency Foundation. I mean, they get the word out

to inform people and put placards up in airports, community areas that have

a lot of traffic, to tell people about these conditions. As physicians, we don't

do a good job of that. So people are becoming more aware. But I think

there's still an awareness gap. I do think patients are coming, and I've seen

it frequently. I mean, I just saw a patient who's 67 years old who actually

makes absolutely no antibody whatsoever; none whatsoever; makes no

antibody-producing cells; was actually diagnosed 35 years ago and put on

IgG replacement therapy, but then stopped due to faulty information, and

has been on antibiotics time, and time, and time again, essentially,

continually for 35 years. [How did they get to you?] She ended up seeing a

very good colleague in the community who was like, "Whoa, you've got a

big problem. You need to go to the center where they're used to taking care

of this." So they came over. Gave her her first infusion of IgG, and bridged

her with some antibiotics because she was sick, and then hopefully she's

going to do well. I think she's going to do quite well.

158

Rational/slow thinking

The physicians were more in favor of pattern recognition and experience than

evidence-based medicine. Although they expressed the importance of following the

literature for quality control purposes, they point out the inherent flaws of approaching

patients with a data-driven mindset; many study results are adequate for the population,

but not the individual.

I think that evidence-based medicine is, it's there to provide some type of

quality control and some type of guidance towards where we want to move.

But we always have to understand where evidence-based medicine comes

from. (BG)

I have a general idea of how much gamma globulin I want to give somebody

based on data but I can tell you that individual patients don't respond the

way the median response in a paper, so I can tell you that lots of people will

do fine with a gamma globulin replacement of about let's say 500 milligrams

per kilogram per month and there are other patients with exactly the same

kind of characteristics that may do fine with 400 and others who may need

1,000. (HL)

Bias and Nudging

The physicians were aware of their potential bias but knew there are some

situations where nudging the patient towards a certain treatment pathway is necessary.

For instance, on physician discussed the importance of using encouragement to help

patients choose treatments, rather than forcing options upon the patient:

I guess it's listening to the patient and offering things in a fair and objective

way. I think those are the most important factors. So if I can understand

someone and lay things out fairly and help guide them, 'cause I'm sure that

I'm biased with what I think is right, but I don't want to ever force someone

to do something, because I think it'll backfire. I'd rather encourage them and

tell them why I think they should do something and have them agree and

buy in, otherwise you don't get the compliance and outcome you want.

(RS)

Another physician described a more subtle approach to nudging patient decisions:

159

As a pediatrician, dealing with the issues, that what you did not want to have

is a scenario set up where anyone would perceive blame. So you don't want

the physician to be blamed for whatever is done, you don't want a parent,

either parent, to be blamed or feel blame. And so, what's done is a collective

decision making. Now there's sometimes when the cost of your medical

knowledge, you believe the decision should be in a certain direction. And if

the parent wants, or the patient wants, are counter to that, you try to use, for

want of better words, savvy psychology to help them understand why that

may be a preferred route to what they're thinking. Many people have mixed

perceptions of things or read testimonies that are incorrect because someone

has a grudge on one or the other. And so what you do is you lay out the

perspectives. If they're equally good, you don't add any bias to it.

(TH)

One participant acknowledges bias as inevitable. However, they believe that experience

helps mitigate the issue of bias:

I’m very sensitive to this issue. I would say if I’m unbiased, no way.

Everyone has their own ideas and experiences. Everyone is biased one way

or another. We try to present in an unbiased way. It is just human nature.

You just try and find the best way. I found that you get better over time in

dealing with being challenged by patients and ideas and being open to new

ideas. Experience gives you flexibility. I think it also has to do with egos as

well. I am definitely better than 25 years ago.

(CR)

Findings Related to SIT and Agency Theory

Power Balance

When asked about how their authority effects the patient, many (7/15) participants

stated that their authority is helpful and makes the interaction more comfortable for the

patient. They described the patients coming to them for advice, to validate their

experiences with a specialized and professional opinion.

I think when they come here to us, already they have the highest

expectations because either they can come here because they didn't get the

satisfactory treatment or approach elsewhere or they came here for unique

things we do for newborn screening, or they're just referred to us because

the other part didn't know what to do. (FS)

160

The physicians were open to SDM for treatment decisions, but not for the

diagnosis. In particular, the choice of treatment administration routes are a shared

decision to best fit the patient’s lifestyle.

I would, usually, emphasis to the patient that in order to progress along this

path of diagnosis to treatment, we need to do x, y, z. In order to understand

the problem more clearly. There isn't usually very much of a discussion

about the pro's and con's, and risk, benefit, cost, et cetera. Most of the time,

during that process. There are circumstances where, specifically, cost will

become an issue. (FB)

I mean sometimes somebody would say, "I want to try facilitated

subcutaneous, because I heard about it." That's fine. That's great. If their

insurance will let them have it, we'll get that for them. If somebody's on IV

and wants to go to subcu, that's great too. (JO)

Physicians are aware of the power balance. Some physicians try to mitigate any

intimidation by reading body language or maintaining a humble persona. However, they

try to keep a professional distance to maintain some power in the relationship for the

more difficult decisions.

That's an area that I've thought a lot about and kind of very conscious about,

so I never address an adult patient of mine by their first name, no matter,

I've known people for 30 years, I've gone to their kids' weddings or

whatever, I never ever ever address an adult patient of mine except as Mr.

or Mrs. or Ms. and I do it because I think I need to maintain a certain degree

of professional distance, maybe part of that is to protect myself but part of

that is there are certain times in a doctor-patient relationship when you have

to say to somebody, you have to give somebody bad news or you have to

say to them "I know this is what you want to do but I think this is really

wrong. (HL)

Health Literacy

The physicians mentioned how modern access to medical information is a double-

edged sword, meaning it can be beneficial or detrimental to SDM. To mitigate this risk,

many participants have taken steps to have literature, links, decision aids, and other

health information on-hand that they personally approve. One participant went as far as to

161

correct and manage Wikipedia pages to ensure their patients are exposed to the most

relevant information.

So, there are people who have read x, y, and z on the net and they may

consider themselves to be health literate but they're getting a lot of

misinformation which can really cause problems. Because now you've gotta

sort of undo what they've read and redirect them to what the actual reality

is of the treatment. (MR)

I think it's helpful; in general, I do. I think the internet is a great resource

for people. Problem is when they go to chat rooms and hear weird things

from different people, it means nothing. You have to use reputable sites, so

we actually give out our list of reputable sites for information on immune

deficiency for patients, so they can read at their leisure and look things up.

(RS)

The physicians often (7/15) mentioned that there are no guarantees of patient

compliance, despite their education and health literacy. Some participants used examples

of educated patients causing self-harm through non-compliance, whereas uneducated

compliant patients have better outcomes. Participant CCR stated, “In fact, sometimes

people who are very highly educated go out there and make up their own mind what they

want to do, and it was a pretty dumb decision.”

Patient networks are the personal and professional connections and programs that

assist in providing access to the information and care to the patient population. All

participants of this study praised the good that patient networks, especially those through

IDF and JMF, can provide. Some participants refer their patients to such networks as a

resource for the more personal needs; physicians sometimes do not have details regarding

daily life with chronic disorders.

I feel like those patients feel like they have a plan. That they can go and find

people ... I think humans, by nature, are herd animals. When they find like-

minded people that are going through the same thing, and they don't feel so

isolated, then you don't have the anxiety, depression, and all of the other

issues related to treating a child with a chronic disease or having a child

162

with a chronic disease. They can find like-minded people that have been

through the same things and can help them with the day-to-day things that

I don't necessarily have advice for because I don't live with it every day.

(LF)

Findings Related to Traits

Patient and physician traits provide inherent influence to the decision making

process. For instance, physicians approach patients as individuals or examples of the

disease (e.g. a patient with PID or a PID patient, respectively). Participant TH made this

distinction:

I think it's really good, but it's always keep telling them, take everything

with a grain of salt, because each patient is really their own individual

disease process. We look at each patient as their own individual experiment.

And so what's solely true for this other patient, and because you think the

symptoms are the same, there only telling you a fraction of the process. And

while that fraction may seem to match up, may not be the direction towards

getting the right answer type of thing. So it's very helpful, but it also helps

the patients to get guided to the right physicians to help with the thing. (TH)

Trust

Trust in the patient-physician relationship is built quickly or obtained immediately

for immunologists, rather than built over multiple visits. Some participants credited their

professional distance to the patient; usually by their formal or informal persona:

I think for my personality, and again, coming as a pediatrician, my patients

call me by my first name, and because their parents call me by my first

name, and so some of the kids will start calling me by my first name. And

I've never been pretentious about that issue, and never tried to correct people

on that. And so I think I come across, for myself, less intimidating, and so

there's a trust that builds up because I'm not trying to snow them, I'm not

trying to pull the wool over their eyes, I'm not trying to intimidate them, and

so I think happens is because they realize I'm well educated, many actually

want me to help more in the decision, not realizing that I'm psychologically

trying to help them in the process. They're actually wanting more of my

input. Even physicians I've dealt with. So I talk about the facts from a

specific patient they're dealing with, and we'll talk over the facts, talk over

163

the data. We'll talk over the different ways of potential faults of

management related issues. And then we're done, and this is what's good for

my position and what I do for providing consultative services. Then the

question comes back and says, they'll ask me, "Well what's your gut feeling?

What would you do?" And I'll say, "Okay, well let's see. That's very easy,

this is what I would do." He says, "Good, that's exactly what I was thinking

I wanted to do. And it always feels better to hear it from you.” (TH)

Other physicians recognize that some patients will try to take advantage of trust or need

to earn it depending on their intentions:

I use their noncompliance, or the dishonesty as evidence in a very

professional and transparent way, as to why I feel the way I feel, and then

I'm very clear that I'm going to document that this is my recommendation,

and they don't take it, that they're going against medical advice. That it's

their choice to go against medical advice, and if their child gets sick, they're

consequences for that. I'm very clear about that and because when you set

that expectation, in a patient that has good intentions, they will work with

you. They will understand that they have been at fault, but if their intentions

are good, and there's no secondary gain, then 9 times out of 10, they will

actually comply with you. When there is secondary gain, then I have

protected myself, and told them what the consequences would be and set

expectations. When they fail to meet my trust again, then I can take recourse

to protect the child. [What would be the secondary gain?] "My child is my

proxy-ish" kind of thing. I don't know. Parents like the attention. That's what

we consider secondary gain. For everybody, I don't know what that would

be. (LF)

Culture/Family

Cultural differences between the patient and physician influence the decision-

making process to take an alternative approach to care. One physician points out the

unique differences of many cultures such as Middle Eastern, African American, and

Hmong patients:

And so, for example, I've dealt with individuals from the Middle East, and

so one of the things we learn as being a pediatrician, obviously, is to make

contact with the individual. And so shake their hand, depending on the

severity of the process and things that are going on. Perhaps hold their hand

a while. Usually a mother, or a female, or the child type of thing for that.

Especially a child. Have them sit in your lap, you know younger children,

sit in your lap while you're doing all this and hold them, so that you reach

164

out. Some of the Middle Eastern cultures, you know, it's very offensive for

a male to touch a female, for example. And so you have to learn to know

that you can't use the same context for connecting. [Is that something you

just learned over time?] Part of that was learning, but part of it was also

when having interpreters and others that would help explain the cultural

differences on there. In the U.S., African American tends to be more

concerned about … So the African American, there is more distrust of the

healthcare system. With good reason for a variety of the things that have

occurred. And so you have to develop that trust from the very beginning,

and honesty from the very beginning on there. And establish the fact that

you recognize they're African American. You point out there's specific

items and issues that are more unique towards African Americans than to

Caucasian. And you, again, you create these, not boundaries, but openness

to that where you can generate that trust, you know?! That "I'm not gonna

be perpetrating on you things that are against your will, or that otherwise

would be harmful. That color of your skin is not a barrier to being able to

achieve the healthcare that you need." Hispanics, different cultural things.

This participant continued into an example of the decision-making process incorporating

entire families:

Hmong believe in a lot of tribalism. From Southeast Asia. Gypsies. Gypsies

are very interesting because of being sort of con artists in many of the things,

they're always very distrustful of anyone because they think everyone's

always out to get them. And usually, with the gypsies, you are in a room of

ten people, because they bring in the elders and everybody to all that.

Findings Related to Organizational Context

Coordination of Care

Many participants adopted the role of coordinator of the patient’s care because the

patient-physician relationships are often long-term when treating PID. These

immunologists, with a highly specialized knowledge-base, would assist in health-related

appointments that could have implications related to PID.

… I'm a pediatrician so there I'm much more of a coordinator, but the

continuity doesn't change, I think that's one of the really big things that we

offer in our clinic is from the time that I started and I was the junior person

with two people in the clinic, we decided the way we were going to operate

is that each of us was just going to take new patients, we would split them

up or whatever way it worked but once you had seen the patient that patient

165

was just going to be your patient and whenever they came into the clinic for

a follow-up, if it was my patient I would see that patient, if they were

admitted to the hospital, I would go to see them in the hospital. (HL)

Other physicians prefer not to be considered the coordinator, or a temporary coordinator

at most. Participant CCR was adamant about treatment roles between specialists:

…when it comes down to the therapy that I'm proposing; for example,

immunoglobulin, or gamma interferon, or Rituxan, or antibio, or anything

else, naturally, I'm going to coordinate and make that happen. I'm going to

make sure it happens. If it's a person who has a gastrointestinal condition,

it's not me. I'm not a gastrointestinal doctor. So I'm going to help them see

that other doctor. So I'm going to coordinate on one hand, and do continuity

on the other. Mine's continuity; the things that I have suggested. If I have to

send them to a rheumatologist, or to a pulmonary doctor, their pulmonary

hypertension, or their hematology, I have to send that person over to the

other quadrant for all of that. (CCR)

Rules related to Time

Most of the participants specified that they treat patients over a long period of

time (10/15) ranging from 10 years to a lifetime (5/15). Before receiving adequate

chronic care, which can take approximately five years, the participants compared their

meeting time with patients to the short appointments of primary care physicians.

Participant TH stated, “When I schedule patients, initial visit's an hour, an hour and a half

when I was doing outpatient, and the follow-ups were forty-five minutes to an hour

depending on the needs of things”. They emphasized the need for prolonged meetings to

explore the patient’s illness and lifestyle. The same participant drew concern to the state

of medicine today:

…in the ideal world, as I think you're eluding to, what we'd have are

primary care physicians that would be set up in a scenario where they would

not be having to see fifty patients a day, but be able to see twenty-five

patients a day. (TH)

166

Reimbursement

A common limitation of SDM is the reimbursement method; patients can only

afford a limited selection of treatment options. Physicians are constantly battling with

insurance companies regarding the treatment of chronic patients. Many have incorporated

strategies to “never lose” arguments with insurance, whereas others prefer to pick their

battles.

Yeah, I don't win all the time. The big issue is that the insurance companies

out there are not aware of what the evidence shows in the literature. Most

of the time what they do is they rely on "physician review" and these

physicians have absolutely no understanding of any particular field. …

Most of those cases get approved after a lot of fighting. Even then

sometimes insurance companies are like no we're not doing this. (IC)

Findings Related to Feedback

Performance Reviews

The physicians described the presence of feedback systems, but few found them

useful. One physician states the benefits of extended conversations with the patient,

which allows for direct feedback. This direct feedback increased patient confidence in the

physician:

I don't think I'm an intimidating presence. That's not my style at all. I mean,

I will tell somebody if I think they're making a bad decision, certainly. But,

I think because I really do try to make this a discussion, you know, a lot of

feedback from the patient. I think most of the time they tell me they feel

much more comfortable with the diagnosis, much more comfortable with

the treatment plan, because we've had this conversation. And I do get a lot

of referrals from people who come in and see me specifically. So, I think

they feel like they've talked to somebody who's got a lot of experience, a lot

of expertise. (MR)

However, most of the feedback is not useful or ignored. In most cases, the

feedback comes from the patients in the form of surveys, wherein patients motivated

enough to participate are often displeased with the physician; the conflict may be related

167

to their illness or irrelevant topics. One physician describes the system of feedback and

why they rarely view it anymore:

[Is there any systematic feedback?] There's not any systematic feedback.

Usually ... I've been fortunate enough that no one has lodged a complaint

against me where administration and patient advocacy have had to get

involved. I have gotten feedback when patients have said nice things to me,

nice things about me. I don't get consistent feedback and I think the only

consistent mechanism, by way that the hospital has the patients rate us is

through the Press Ganey surveys. After that incident where I had that

horrible review online, I have stopped googling myself. I'm a caretaker, I'm

a feeler. I take those things really personally because I don't want other

people, and other patients, who I take care of to read something like that

and then lose their trust in me because of something they read on the

internet. These last two years is a perfect example of seeing things on the

internet that aren't really true or not the full side of the story. They rely on

that too much… (LF)

Cost

One aspect of feedback that physicians are lacking is cost. Most physicians do not

have direct feedback regarding cost outcomes to treatment. One physician mentioned that

the insurance companies will occasionally inform them of costs.

[In the 300 patients that you have, is there any way that someone looks at

the cost of treating them over the course of the year and the outcomes? Is

there any way that someone says, "Wow, here's the 300th." And why not?]

You know, what I'll say is that there's no systematic way that is intentionally

done for my individual patients that gives me feedback. I do hear from

payers, from time to time, to say, "This patient on X asthma therapy," again,

not talking about immunodeficiency, but kind of an easier disorder to talk

about because there are just much more metrics in place to track quality and

so forth, "you might consider stepping them down from drug X to drug Y."

And I think that's based on guidelines. It's also based on cost. For sure.

There's no doubt. I mean, let's face it. We all get that. So there are some of

those things. There's also, even within our own system, like, our health plan

covers women and children who are of lower income. It's a medical

assistance, Medicaid-based payer. So we will get feedback on use of

expensive medications and lower-cost alternatives. There are some of those

things out there, but there is nothing that tracks, "Yeah, here's your 300

patients. 250 of them are optimally treated based on these outcome

measures. You're providing as cost-effective care as you can based on the

complexity level of the patient, available resources, et cetera" (NR)

168

Electronic Medical Records (EMR)

The physicians’ opinions are split regarding electronic medical records (EMRs).

One side praises their use for enabling better care coordination between physicians,

whereas the other side of the argument mentions the flaws such as accidentally nullifying

clinical trial participants. Every participant that discussed EMRs described benefits and

issues:

[Are electronic data records a good thing?] Oh yeah. I think so because I

can read everybody's notes. [Any downside to electronic records?]

Sometimes, you need to be careful what you really need to write. For

example, some of the patients, but we do sometimes research testing, of

course after getting consent, it's not ... But some other sub specialist that's

taking care of the same patient for another thing, they ... If parents tell that

to that doctor and they put it down, that's a problem because research data

cannot be in the medical records for clinical care. They happen to me a

couple of times which is difficult because then you have to addend. But, it's

not good to disappear completely. If the medical records wants to go back

and look at it, they can see it. It's going to go to the record that the other

provider will see, but it will be in the records see, there is no way you can

correct it or get rid of it. [In general terms of caring for the patient-] It makes

it very easy. (FS)

Cross-Referencing Findings from Post Hoc and QUAL Interviews

The findings of the post hoc and the QUAL interviews support each other.

Rational decision making increases participation indirectly through patient-centric care is

supported by interview responses. Of the participants, CCR was strictly rational decision

maker that did not incorporate a patient-centric approach. CCR “paternalistically”

dictated decisions for the patient and was adamant that patient lifestyle and care

coordination were not of primary concern; they had a “listen to me or go elsewhere”

perspective. On the other hand, all other participants (14/15) made an effort to coordinate,

manage, and incorporate patient input for treatment decisions; this supports the QUAN

169

result where patient feedback of treatment tools (products) increases overall participation

with protocol options and management.

Table 35. Cross-Referencing Study 3

# Description of Finding Interviews Post Hoc

1 Rational Decision Making Pattern recognition > EBM DM_R has no direct

effect on

participation, but

DM_H does have a

direct effect.

2 Bias/Nudging Physicians provide options they

approve of first.

N/A

3 Power Balance Patients go to the physician with

high expectations because they were

unsatisfied with previous care.

N/A

4 Health Literacy Patient health literacy must align

with the physician.

N/A

5 Trust Trust is assumed. N/A

6 Culture Culture can change the entire

interaction with the patient.

Race does not affect

participation.

7 Coordination of Care Most act as a coordinator of care N/A

8 Rules (Time)

9 Reimbursement Insurance does not affect decisions

or participation.

N/A

10 Performance Reviews No reliable performance feedback. N/A

11 Cost No reliable/consistent cost

information.

N/A

12 EMRs Helpful, yet inconvenient. N/A

The interviews support the notion that decision-making style and approach

indirectly effects protocol participation through tool feedback. Participants that described

the first 1-2 meetings with the patient were to get a better understanding of specific

patient context (history, lifestyle, etc.). This is to prescribe the best initial treatment

regime, thereby minimizing the extent of future changes to the treatment. Therefore,

170

physicians with a patient-centric approach instigate patients to notify their physician of

the treatment efficacy (tool feedback). Furthermore, this reinforces the influence of

timing in the physician-patient relationship and how it applies to SDM; SDM is likely

most implemented in the earlier segment of the physician-patient relationship because

initial treatment is an educated guess, then refined to the patient.

Discussion

According to the findings above, physicians likely implement SDM under a set of

conditions. In other words, SDM is bounded. The physicians treat SDM similarly to the

“nudges” described in Sunstein and Thaler (2008). Nudges are the subtle suggestions in

the decision process; methods or strategies to compel limited responses. Daniel

Kahneman describes nudges as “explicitly paternalistic” because they set the “choice

architecture” (Kahneman, 2017); choice architecture means to set predetermined options.

In the case of SDM, physicians have a set of requirements to meet before

considering shared decisions with the patient. From the findings, there are four (4) main

requirements with subcategories: patient characteristics, organizational context, power

balance, and dual process. Treatment evidence refers to evidence-based medicine and the

proven treatment methods to help the patient. Power balance refers to the interactions

between the physician and patient; an imbalance could lead to patients being non-

compliant or fired in some cases, whereas others might avoid participating because of the

burden of decision-making. The system context refers to factors such as 3rd party payers;

most patients can only afford the treatment that their insurance covers. This category also

includes factors such as regulations and organizational protocols, but this I did not

explore these extensively in this study. The patient characteristics refers to factors such as

171

their health literacy and their alignment with the physicians. Patient alignment was a

consistent necessity with the participants in particular. Physicians would only implement

SDM when the patient knowledge is backed by information the physician approves.

The literature universally praises the potential benefits of SDM for being a

patient-centric approach to care that is best used for the care of chronic conditions.

However, the results suggest that SDM is parallel to nudges. Each physician has their

personal thresholds to implementing SDM. For some of these thresholds, the physicians

try to encourage particular choices such as having decision aids available for patient

learning. Therefore, contrary to the literature, paternalism is not obsolete but has shifted

to satisfy the patient-centric movement of the last 30 years.

This finding may also suggest there is an influence of time in the patient-

physician relationship. Although this study and its survey did not account for which visit

these answers are referring to (i.e., first, second, third visit, etc.), it is possible that the

immunologist sample responded to the survey items based on experiences from different

patient visits.

Limitations, Implications, and Future Research

Although I attempted to address the limitations of the first and second studies, I

acknowledge the possibility of recurring limitations. One limitation of the third study was

the lack of a demographically diverse sample of participants. Future research should

address this issue by attempting to recruit participants in a larger range of demographic

characteristics, despite the possibility that this sample accurately represents the

population of immunologists.

172

Another limitation is that the sample was selected based on availability rather than

randomized participants, thereby exposing the study to potential bias. Except for one

participant, most physicians interviewed had more than 10 years of experience in their

specialty. To overcome this limitation, future research must incorporate larger samples

with a random selection process.

Understanding physician perspectives of SDM and the encouraging factors has

practical implications for the care of patients with chronic illnesses. Therefore, leaders

and departments responsible for setting treatment policies—such as hospital standards or

insurance reimbursement policies—would benefit to encourage SDM. Forming an

environment which allows SDM can lead to better treatment and cost outcomes in the

long-term because of the known benefits of SDM (Arterburn et al., 2012; Légaré &

Witteman, 2013).

In summary, the third study offers a more refined analysis of physician decision-

making regarding chronic conditions. It expanded upon previous findings by discovering

that SDM is bounded by factors such as time, patient characteristics, system context,

power balance, and evidence. It addressed some limitations of the first and second studies

by taking physician workplace into consideration. In the next chapter, I integrate all three

studies to provide a more robust set of results with fewer limitations.

173

CHAPTER 8: INTEGRATIVE FRAMEWORK

Mixed methodologies must reconcile two standards, quantitative and qualitative,

to assess the validity and credibility of the results. An integrative framework was needed

to make meta-inferences about the SDM phenomenon beyond what each method can

explain alone (Tashakkori & Teddlie, 2008). Inferences are used to make interpretations

using a process of both interpreting findings and conclusions related to the research

question. The quality of the inferences should support the transferability of the

conclusions to other chronic disorders.

In chapter 4, an integrated framework of physician decision making for chronic

diseases was proposed. The decision-making process was described as slow or rational

(dual process) with potential for bias and nudging. Other hypothesized influences

included physician and patient traits, power imbalances, and organizational context

(Table 18). This framework has been revised based on the findings. Specifically,

influences related to organizational policy and feedback have been removed because they

were not significant or pursued; therefore, future research is necessary on these

influences.

174

Figure 15. Theoretical Framework from Chapter 4 (Revised)

Figure contains an illustration from: https://www.canstockphoto.com/doctor-and-patient-

13612490.html

The integrated results of the three studies provide the basis for the above

framework (Figure 15). The first study confirmed the decision making styles (rational or

heuristic) of physicians as described in the literature and revealed the patient-centric

approach that U.S. physicians use to treat hemophilia. This study indicated that U.S.

physicians were not influenced by organizational factors such as policies and

reimbursement. The second study determined that patient participation—and SDM by

extension—with the selection of tools and protocols for the treatment of PID depended

on the decision making style and patient-centric approach to care, but participation

increases with protocols or tools, not both. In the post hoc analysis of the second study,

participation with protocols (disease management and treatment options) occurs with

feedback from the treatment tools. Furthermore, patient participation depends on a

175

rational decision-making style mediated through a patient-centric approach. The third

study provided context for when physicians implement SDM. It reinforces the influence

of rational decision making and organizational context related to time and coordination of

care.

This chapter describes the meta-inferences that emerged from the mixed methods

study which included results from 370 physicians who treat two complex chronic

diseases: hemophilia and PID. As described in Chapter 4, Research Design, these meta-

inferences provide a more complete picture of physician influence on SDM which is

more meaningful than each of the components.

Meta-inferences

This section will cover 7 meta-inferences that emerge from the multiphase mixed

methods research. Meta-inferences are “an overall conclusion, explanation or

understanding developed through an integration of the inferences obtained from the

qualitative and quantitative strands of a mixed method study” (Tashakkori & Teddlie,

2008). The purpose of these meta-inferences is to achieve a more comprehensive

understanding of what predicts physician implementation of SDM. The meta-inferences

include:

1. Rational decision making

2. Patient Centrism

3. Bias/Nudging

4. Organizational Context

5. Feedback

6. Traits

7. Power

176

Rational Decision Making

In each study, I used dual process theory to explain the physician decision

process; it partially describes the physician decision process. However, in each study, I

observe the clear reliance on rational decision making (System 2) over heuristic decision

making (System 1). Specifically, rational decision making is prioritized when the disease

is categorized as complex; hemophilia with inhibitors and PID (in general) are complex

chronic diseases. In the first study, the physicians made a transition from intuitive to

rational decision making as patient complexity increased, (i.e., if inhibitors developed);

they would consider which treatment options are best to remove the inhibitor issue.

Although heuristic decision making appeared equally significant to rational (if not better)

in the second study, the post hoc analysis found the in the population of immunologists,

rational decision making was dominant. This was further confirmed in the third study

wherein the participants all investigated the patient’s full history, especially in the initial

meetings.

Patient-Centrism

Across all studies, physicians with a patient-centric approach to care encourage

patient participation in treatment decisions. To clarify, this does not mean physicians let

patients help them diagnose the problem, but that physicians will take patient values and

preferences into consideration for treatment protocols and tools. In the post hoc, patient-

centric approaches to care reverses the negative relationship between rational decision

making and feedback with tools; meaning, physicians with a rational decision-making

style normally do not incorporate feedback, but will if they use a patient-centric approach

to providing care. A similar observation is in the third study, wherein all the physicians

177

were rational decision makers, but only a few maintained a paternalistic (non-patient-

centric) approach. As described in Chapter 6, these participants adopt an approach where

the patient accepts their decisions or gets treatment elsewhere.

Another pattern concerning patient-centrism is the mediating role it plays in

SDM. Patient participation occurs when (not “if”) the physician is patient-centric,

meaning SDM requires certain conditions to be met before implementation. The

qualitative studies reveal that a patient-centric approach to care was required for SDM to

occur because physicians dictate treatment if they did not attempt to incorporate patient

input. On the other hand, the quantitative analyses show patient-centric care to partially

mediate decision-making style and patient participation.

Traits

Physician traits did not seem to affect decisions throughout my research. In the

first study, all the physicians had similar approaches to care and patients. Although a few

outlier cases involved individuals of extraordinary expertise, they also had the similar

rational decision-making style with a patient-centric approach. The second study suggests

a few significant factors such as age and gender but these were disproven by the post hoc

analysis. Furthermore, the third study did not show differences in immunologist decisions

based on traits, either. However, these studies did not investigate physician SDM with

minority groups.

Trust was a recurring theme in the qualitative studies. In the qualitative studies,

trust was considered to be assumed or built very quickly. Based on how the physicians

described it, trust also acted as a condition for physicians to treat patients, let alone

incorporate SDM. In some cases, the physician had to “fire the patient” because certain

178

patients were non-compliant and uncooperative in providing crucial information (i.e.,

stopping treatment because X). However, when tested as a moderator in the second study,

there was no significance. This likely occurred because trust is assumed, not earned over

time. Furthermore, the patients often go to these specialists as a “last hope,” which likely

results in immediate trust.

Bias/Nudging

Throughout the qualitative studies, there are hints of nudging and bias in the

decision making process. Especially in the physician with Ph.D.s, the physician would

passionately advise and push for the treatments they prescribe. In the case of patients,

they all provide the information to the patient with the good news first, therefore priming

them to the preferred option. This adamant perspective extends to disputes with insurance

companies, wherein the physicians would fight insurance to reimburse the patient for

products the physician specifically asked for.

Organizational Context

There multiple meta-inferences related to organizational context: feedback and

reimbursement. Physicians aim to provide the best treatment on the first try, but often

need to revise based on feedback. In the first study, patients need different infusion rates

and products depending on their lifestyles and physical response to the product. If the

patient develops an inhibitor, the physician must reconsider the treatment. In the third

study, the immunologists often referred to themselves as coordinators of care. After the

extensive initial meetings, they would consistently monitor the patient and adjust the

treatment similar to hemophilia. This is also similar to the post hoc, wherein participation

with protocol options and management depended on patient feedback of tools; without

179

feedback, the physician does not need to change anything. Furthermore, the structure of

the relationships better explains patient participation because the explained variance was

about 0.25, whereas the second study was less than 0.15. In the post hoc, the relationship

between physician decision making and participation with protocols was better explained

indirectly through tool feedback. The patients providing feedback about the product

encouraged participation with decisions regarding their disease management and

treatment options.

The minimal cost feedback might relate to the minimal influence of insurance

companies. Across both qualitative studies, these specialized physicians fight insurance

companies to cover treatment costs if the pre-approved treatments do not match the

physicians’ decisions. This is not to say every case requires the physician to contest the

insurance company options; physicians will agree with the insurance option if it is

acceptable for the patient and interchangeable to that the physician preferred. If conflict

arises, the physicians all stated that they would win. Therefore, contrary to expectations

of U.S. physicians, the influence of insurance companies is minimal to non-existent.

Implications for Practitioners

This research study started with the question: what predicts physician

implementation of SDM. Using four theoretical frames, numerous findings and meta-

inferences were identified. These results have potential for healthcare practitioners to

design a better U.S. system to improve cost-effective care.

1. SDM for chronic care requires slow thinking with a patient-centric approach.

Reimbursement should support this cognitive method of thinking. Currently,

reimbursement is either fee for service, meaning billing for a visit or a test, or

180

a quality of care measure such as hospital readmissions. Greater time should

be allowed for physician-patient interaction and physicians should be paid for

this increased effort.

2. Physicians should know the cost of treating their chronic care patients. Most

hemophilia and PID doctors are specialists who treat many patients—often

dozens or hundreds. Physicians should know the cost of treating these

patients, the mean, median and range. Physicians should also have more user-

friendly electronic records which allow outcomes comparisons with their

treatment cohort. Having cost and outcomes data will allow physicians to

know the cost-effective impact of their care. Incentives can then be used to

improve cost-effective care.

3. Feedback is important. Currently, physicians do not have any formal feedback

from patients on the extent to which patients are satisfied with their

participation in the care.

4. Combining #2 and #3 above, a system which provides physicians both cost

and quality data would provide the basis for improving overall care. Such a

system could benchmark physicians.

5. The U.S. should create patient registries similar to what is done in the U.K.

This will provide national cost and outcomes data. Physicians can then be

provided with comparisons between national and local practice. This can be

used as a way to benchmark care and incentives for improvement.

6. SDM is about physician–patient interaction regarding protocols and tool

feedback. Very few physicians dictate the specific tools to be used. Therefore,

181

creating large tenders which reduce drug cost will not hinder SDM and

provide substantial system savings.

7. Coordination of care is imperative for chronic care. Defining the role of

primary care physicians and specialists is needed to provide comprehensive

treatment.

8. Health literacy is less about educating patients than it is about aligning patient

and physician understanding. Developing consistent tools which facilitate this

alignment is necessary for patient compliance. Decision aids should be

developed for hemophilia and PID to facilitate this alignment.

9. Nudging has been demonstrated to work in a variety of contexts such as

retirement planning. The role of nudging for chronic care should be

acknowledged and optimized.

10. Physician-patient health literacy alignment and nudging will likely result in a

“soft paternalism.” The concept of SDM should be revised to better define

sharing.

The results from the three studies considered in the context of the actual quality

gaps presented in Chapter 3. This analysis from several public databases on patient and

hospitalization records show:

1. 30-43% of all households give poor ratings of quality indices. Fifty-three

percent of the hemophilia and PID patients gave poor ratings on quality

indices.

2. Hospitalizations can be used as a measure of poor diagnosis and treatment.

While there has been a substantial improvement in the number. Diagnosed

182

patients hospitalized due to hemophilia and PID, the associated cost is

estimated to be around $3 billion. There is also a cost associated with

undiagnosed PID patients which is estimated to be between $3.5 and $7.7

billion. The total cost of hospitalizations is between $7 and $10 billion.

3. Given that product/tool choice is not a factor in shared decision making,

national wide tenders such as those organized by the U.K. could reduce the

cost of treatment of hemophilia and PID by up to $3 billion.

When combining the results of the analysis of patient/medical claims and

interview with physicians, there is a substantial opportunity to improve quality and

reduce cost. Quality can be improved by improving patient feedback on defined metrics

such as listening, respect, and time. Cost can be measured by looking at unnecessary

hospitalizations. Using slow thinking for the diagnosis and treatment is likely important

to achieve improvement in care and start to close the quality chasm in the U.S.

183

CHAPTER 9: DISCUSSION, IMPLICATIONS, LIMITATIONS, AND

CONTRIBUTIONS

Often medical care is viewed from the lens of acute hospital inpatients in which

informed consent and EBM are most applicable. Most medical care is now related to

chronic disorders where the setting requires a more active patient and a longer window to

make decisions. Adaption of SDM to chronic care is necessary to facilitate a patient-

physician partnership “to muster trust and mutual respect: and “create an environment

conducive to successful patient self –management” advocates for SDM believe that

chronic care requires a certain level of knowledge by the patient, a trusting relationship

with the physician and mutual engagement in setting goals based on values and

preferences Intimate engagement with patients with chronic disorders is necessary to

achieve decisions which are most advantageous to their health. SDM is a radical

departure from traditional medicine and the means through which it can be accomplished

has not been adequately understood. Once understood, the optimal choice decision

architecture can be designed to improve cost-effective care.

The results from this research reveal several important factors that influence the

adoption of SDM and how these factors can be optimized for chronic care management.

It is clear that rational decision making (System 2 thinking) as predicted by dual process

theory is important to achieve SDM, particularly with patients with complex (multiple

co-morbidities) chronic diseases. In addition, the role of health literacy, networks, EBM,

and reimbursement is more constrained than what has been reported in the literature.

Given these findings we can now better understand why some healthcare systems such as

ChenMed and CareMore are showing dramatic success and how some organizational and

184

behavioral changes can be applied more broadly to help cross the U.S. healthcare quality

chasm and build a long-term sustainable system with SDM at its core design principal

(Hathi & Kocher, 2017; Hostetter et al., 2017; Tanio, 2014). However, the impact of how

information is anchored in the decision-making process needs further research.

Theory

The most significant contribution from this research is the extension of dual

process theory as a method of analyzing the physician decision process for the treatment

of PID. All three confirmed relevance of dual process theory to examining the physician

decision process in the context of chronic health issues. However, we can learn from

other theories as well.

Nudging and libertarian paternalism are subtle methods of encouraging, not

dictating, patient choice and are present in throughout my research and the literature

(Loewenstein & Chater, 2017; Schiavone et al., 2014; Sunstein & Thaler, 2008). The

physicians would make suggested treatment options and supplemental literature for

patients, yet acknowledge the patient makes the decision (Aggarwal et al., 2014).

Status, as predicted by SIT, is relevant and health literacy as predicted by Agency

Theory is needed to balance power. Treat theory also has a contribution: certain physician

and patient traits influence SDM. Organizational context also matters. It is the integration

of all four that provides an overall framework.

185

Figure 16. Summary of Key Findings

First Study (2015) QUAL

•Findings

•Physicians from both the U.S. and U.K. primarily rely on data for their treatment decision (data-driven).

•U.S. physicians tend to be more influenced by patient factors (patient-driven) than U.K. physicians.

•Physicians nudge patients towards decisions.

•Physicians consider the need of the patient population.

•Power imbalances occurs with thought-leaders.

Second Study (2016) QUAN

•Rational decision making style (DMR) increases patient participation for protocols (IOP); however, participation decreases for the choice of treatment tools (IOT).

•Heuristic decision making style (DMH) physicians decreased IOP, but increased IOT.

•Patient-centric approach to treatment (APC) positiviely mediates decision making style and patient participation.

•Being a male physician decreases IOT.

•Being an older physician increases IOP.

•Being a white physician increases participation more than non-white physicians.

Third Study (2017) QUAL

•Embedded Post-Hoc of Study 2

•Immunologists with DMR and a patient-centric approach increases patient participation.

•Immunologists with DMH directly increase patient participation

•Patient participation with Tool Feedback increases participation with Protocol Management and Options.

•Physician education increases patient participation. Other traits have no effect.

•Interview Findings

•Bounded SDM:

•Patient Alignment

•Organizational Context

•Trust & Health Literacy

•Power

KEY FINDINGS

•Facilitators of SDM

•Dual Process decision making when mediated patient-centric decision making

•Trust, if maintained

•Inhibitors of SDM

•Bias/Nudging

186

Practice

The findings of this study suggest the need for the development and refinement of

treatment decision aids for diagnostic procedures and recommended treatment protocols.

To our knowledge, there are no decision aids specifically for hemophilia or PID.

Decision aids would assist physicians and patients in deciding on the optimal treatment

approach; providing optimal and timely care benefits the patient both financially and

physically (Modell et al., 2011). Improving decision approaches would help physicians

and patients correctly diagnose root issues quicker, which would significantly reduce the

economic burden of misdiagnosed chronic health conditions (Modell et al., 2011).

Additionally, this paper may help physicians develop a more nuanced

understanding of their own decision-making process. Some may conduct a self-analysis

to improve or understand how they approach patient treatment and may decide to

incorporate patient participation to a greater degree. Of particular importance is the role

of nudging and to what extent it is transparently being used to facilitate decisions that are

advantageous to patient health and are consistent with SDM. Patient literacy is necessary

for decision alignment and must be sufficient such that nudging does not become another

form of paternalism.

System issues related to coordination of care and feedback need to be

implemented. Chronic care is not a dyad but a team of healthcare providers that must be

coordinated to execute SDM. Meaningful feedback loops are needed that will guide

physician in constructive ways to improve SDM behavior.

187

Other Research Domains

The application and benefits of SDM are not restricted to the treatment of

hemophilia and PID. The studies could be used as a template for additional research into

other chronic conditions. The results of the first study suggest that many U.S. physicians

treating hemophilia and PID are implementing SDM; they work with each patient to

develop individualized treatment plans that accommodate treatment to the patient’s

lifestyle. The second and third study focused on enhancing physician approaches to

chronic care with SDM; therefore, the research design is likely applicable to more

common chronic diseases such as cancer. If similar outcomes can be generalized to

chronic care (i.e., from arthritis to cancer), SDM could be the first step to improving cost-

effective care in the U.S., perhaps matching those of other OCED countries.

The application of SDM can be utilized in other professional relationships. For

instance, a lawyer and client have a similar relationship, wherein the lawyer is a

professional coordinating with the client to address legal issues. Future research can

expand into various similar contexts involving professional gaps.

Limitations

The three studies involved physicians that treat hemophilia and PID; the

specialties are unique with different issues of complexity. Therefore, the results may not

be generalizable to other chronic diseases. In addition, the analysis of quality and cost of

care for hemophilia and PID is limited to published findings.

Qualitative data relies on the use of small non-randomized samples with limited

demographic diversity. The grounded theory process mitigates these limitations,

particularly interviewing participants until saturation. The quantitative studies relied on a

188

one-time survey which had a low response rate and poor distribution of certain traits (sex

and traits). Accepted statistical techniques were used to derive meaningful results which,

although not perfect, allowed for statistically significant correlations. Mixed method rigor

was addressed by using consistent research methods for each study, consistent reference

to published research, and matching the results to similar research questions addressed in

all three studies. This research was further augmented by an analysis of patient and

hospitalization data which show substantial quality gaps and significant opportunity for

improvement.

Another limitation is the inherent meaning of shared decision making. All

participants claim to implement SDM to some extent, but the reality is different when

probing for contextual insight. Physicians are often making the decisions via nudges and

prefer aligned, compliant patients over health literate patients. However, there are some

cases where the physician wants to mitigate the risk of losing patients listening too much.

Furthermore, the definition of health literacy was not clearly defined before the research.

The degree to which each physician considers the patient as health literate greatly varied.

The quality of the information available to the patient and their ability to interpret data

also effects what can be defined as health literate.

The issues of cost and quality are a system-wide, complex problem that cannot be

solved with SDM alone. It is also focused on chronic illness and has limited application

to acute care. The impact of SDM also varies from different chronic issues; patients with

PID will have different interactions with the physicians than stroke victims.

189

Conclusion

“Nothing without us” is a frequent rallying cry by patient communities seeking

broader involvement within the healthcare system (Chu et al., 2016). As the paternalistic

approach to chronic care has lost relevance, we continue to explore influences and

barriers to SDM from the physician perspective. In our first study, we found that

physicians are willing to work with patients in forming treatment regimens that are

supported by patient-centered care using slow decision making. The second study

provided an initial picture of the application of dual process theory to the physician

decision process and some factors directly affecting patient participation. However,

neither of these studies provided a clear and complete image of the physician decision

process and the factors that influence the adoption of SDM. The third study expanded the

initial research and provided a more complete picture of SDM. The integrated results

identified significant influences on physician decision-making styles and attain a deeper

understanding of their contributions to the physician decision process. Adopting the

mixed-methods approach formed an integrated theoretical framework that may become a

new way to help implement SDM. An attempt was made to quantify the quality and cost

gaps associated with poor diagnosis and treatment which significant opportunities for

improvement. Expanding this research from individual physician-patient relationships

treating hemophilia and PID to other chronic disease categories may help healthcare to

cross the quality chasm in the U.S.

190

Appendix A: Study 1 Interview Protocol and Questions

1. Introduction and Explanation—read to the interviewee before proceeding

a. Greeting “Hello [name of participant]. Thank you for taking the time to

meet with me today. Your participation is greatly appreciated. Before

getting started, there are a couple things I would like to cover.”

b. Purpose and Format of Interview “As a current student in the Case

Western Reserve University Doctor of Management (DM) program, I am

interested in developing a greater understanding of the factors that

influence physician decision making for the treatment and management of

hemophilia. I will ask you a series of open-ended questions on this topic,

and I will ask one or more follow-up questions as you respond. The

interview will last approximately 60 to 90 minutes.”

c. Confidentiality “Everything you share in this interview will be kept in

strictest confidence, and your comments will be transcribed

anonymously—omitting your name, anyone else you refer to in this

interview, as well as the name of your current organization and/or past

organizations. Your interview responses will be included with all the other

interviews I conduct.”

d. Recording “To help me capture your responses accurately and without

being overly distracting by taking notes, I would like to record our

conversation with your permission. Again, your responses will be kept

confidential. If at any time, you are uncomfortable with this interview,

please let me know and I will turn the recorder off.”

e. “Do you have any questions before we begin?”

Introduction

1. Name

2. Education

3. Current job title and responsibilities

4. Years of practice

5. Involved with research (clinical or otherwise)

191

6. Practice setting/site of care (multiple?)

7. What is the average length of time with the patient?

8. Most have co-morbidities?

Focus in on factors that determine how you make decisions

1. Describe a typical patient

a. How do they get to you?

b. Diagnosis to treatment to maintenance

2. Describe the types of decisions you make?

a. SC vs IV

3. Do you use decision aids (describe/evidence-based?)

4. Do you find that your patients are educated or well educated on self-

management?

a. High levels of self-efficacy (I am confident I can manage my situation)

b. High levers of self-activation

c. Do you think more informed patients result in fewer health resources

and better outcomes?

d. Are you patients actively involved in patient networks (IDF, JMF,

internet networks such as patients-like-me)?

e. Do you have a patient portal where patients can review their health

history?

5. What seeing a patient: look for patterns that match past experience?

a. Tend to quickly assess symptoms and diagnosis or it it’s a slow

painstaking process that has lots of complexities (particularly with co-

morbidities)

b. Tend to spend more time on protocol or treatment

c. Would you describe yourself as patient-centric or evidence centric

(EBM)?

6. If you were to guess: do you as the physician make the final treatment

decision or leave it up to the patient

192

a. Protocols

b. Drugs

7. Does your DM style vary based on the patients’ level of understanding and

interest?

a. Does it vary based on complexity or uncertainty?

8. Do you think your status as a physician or authority figure influences how

patients respond to you in a clinical setting?

a. Is Intimidated or encouraged to share information or

b. More likely to tell you their treatment preferences or express an option

on treatment options?

9. Do you routinely ask about patient preferences: lifestyle and how treatment

will affect patient goals and values?

10. When discussing pros and cons of a potential treatment (protocol or drugs) do

you tend to lead with the pros or cons

11. How much and in what way do cost or reimbursement influence your decision

making

a. Protocol

b. Drug

12. Do patients want to play an active role in their decisions

a. Function of health literacy

b. Function of health numeracy

13. Do you see yourself a continuation of care or coordinator of care

a. Is coordination of care an issue for your PID patients

b. Are patients actively looking for you to coordinate their care?

c. Do your PID patients have problems accessing healthcare services or

getting adequate treatment?

14. Have you ever made a mistake/misdiagnosis?

a. Proper follow-up

b. Diagnostic test

c. Adequate history

193

15. Prefer face-to-face or is phone or internet possible and how often

16. Trustimpact on patient participation

17. Does DM change over time as uncertainty changes

a. Less shared decision making

18. Provider number

19. Are you rated? Do you recall your score?

20. How would you assess or describe the quality of your communication style?

a. Outgoing

b. seeking

21. Do patients tend to speak up or does it depend of the context and your

relationship

22. Impact of the organization on your decision making

a. IOM: patient-centered care: patient perspectives are now being

factored into Medicare value-based payments to hospitals

194

Appendix B: Invitation Template

Dear Dr. ___________,

My name is Christopher Lamb and I am pursuing a doctoral degree in management at

Case Western Reserve University in Cleveland, Ohio.

I am conducting interviews with experts for my qualitative research project on the

physician decision process of immunologists that treat Primary Immunodeficiency (PID).

My questions will focus largely on your experiences at work. The interviews are

conducted over the telephone for approximately 30-60 minutes and will be recorded

(audio or written notes only). You will be provided with an Amazon gift card to

compensate you for your time.

Should you be interested in being a participant in my research, please provide the best

date, time, and contact information to reach you.

The records of this research and your participation will are confidential. I will provide

additional information regarding confidentiality upon acceptance to participate (or upon

request).

If you have any questions, you are welcome to contact the Responsible Investigator, Dr.

Kalle Lyytinen, at 1-216-368 5353 or [email protected]; or the Co-Investigator,

Christopher Lamb, at (339) 440-6061 or [email protected].

Thank you in advance for consideration.

Sincerely,

Christopher Lamb

Ph.D. Candidate

Weatherhead School of Management

Case Western Reserve University

Phone # 339-440-6061

[email protected]

195

Appendix C: Study 2 Constructs

Patient-Centered Approach to Care (APC)

1. cPATIENT_11 I tend to... Ask patients about their quality of life status.

2. cPATIENT_12 I tend to... Ask patients about their psychological status.

3. cPATIENT_13 I tend to... Ask patients about their perceived health status.

Heuristic Decision Making (DMH)

1. DM_INT_1 I tend to...Make decisions instinctively.

2. DM_INT_2 I tend to...Quickly diagnose PID based on prior patient experience.

3. DM_INT_3 I tend to...Rely on instinct for treatment decisions based on prior

experience.

Rational Decision Making (DMR)

1. DM_RAT_1 I tend to...Be very systematic when I go about making a decision.

2. DM_RAT_3 I tend to...Take my time to think through treatment decisions.

3. DM_RAT_5 I tend to...Leave myself time to think through treatment decisions

before I act.

4. DM_RAT_6 I tend to...Carefully work out a treatment plan before making a

treatment decision.

5. DM_RAT_10 I tend to...Learn as much as I can about possible consequences

before making decisions.

Patient Participation with Treatment Protocols (IOP)

1. PROTOCOL_2 Patients tend to... Engage in discussions regarding treatment

protocol.

2. PROTOCOL_3 Patients tend to... Share with me what they understand about

their treatment protocol.

3. PROTOCOL_4 Patients tend to... Share with me what they don’t understand

about their treatment protocol

4. PROTOCOL_11 Patients tend to... Make considerable effort to discuss their

schedule with me.

Patient Participation with Treatment Tools (IOT)

196

1. TOOLS_1 Patients tend to... Influence which brand I treat them with.

2. TOOLS_9 Patients tend to...Choose the brand of IgG replacement therapy they

want.

3. TOOLS_10 Patients tend to... Actively participate in the product choice to treat

their condition.

Physician Trust in Patient Input (TIP)

1. TRUST_1 I trust the patient will... Provide accurate medical information.

2. TRUST_2 I trust the patient will... Let me know when there has been a major

change in their condition.

3. TRUST_3 I trust the patient will... Tell me about all medications they are

using.

4. TRUST_4 I trust the patient will... Follow the treatment plan exactly as I have

provided.

5. TRUST_5 I trust the patient will... Manage their condition with the prescribed

treatment plan.

6. TRUST_6 I trust the patient will... Tell me if they are not following the

treatment plan.

7. TRUST_7 I trust the patient will... Not manipulate the office visit for

secondary gain.

197

Appendix D: Post Hoc Constructs

Constructs of Study 3 Phase 2: Post Hoc

Factor Name Variable Label

Heuristic Decision

Making

dmH_1 I tend to...Make decisions instinctively.

Heuristic Decision

Making

dmH_2 I tend to...Quickly diagnose PID based on prior patient

experience.

Heuristic Decision

Making

dmH_3 I tend to...Rely on instinct for treatment decisions based on

prior experience.

Patient-centric centP_2 I tend to... Encourage patients to extensively learn about

their condition.

Patient-centric centP_7 I tend to...Learn the patients’ culture and background.

Patient-centric centP_9 I tend to... Focus treatment decisions on patient

preferences.

Patient-centric centP_11 I tend to... Ask patients about their quality of life status.

Patient-centric centP_12 I tend to... Ask patients about their psychological status.

Patient-centric centP_13 I tend to... Ask patients about their perceived health

status.

Protocol

Participation

ioP_3 Patients tend to... Share with me what they understand

about their treatment protocol.

Protocol

Participation

ioP_5 Patients tend to... Play a key role in organizing a treatment

plan.

Protocol

Participation

ioP_11 Patients tend to... Make considerable effort to discuss

their schedule with me.

Rational Decision

Making

dmR_2 I tend to...Rarely make a decision without gathering all the

information I can find.

Rational Decision

Making

dmR_4 I tend to...Only make treatment decisions when all the

information is gathered and available.

Rational Decision

Making

dmR_5 I tend to...Leave myself time to think through treatment

decisions before I act.

Rational Decision

Making

dmR_6 I tend to...Carefully work out a treatment plan before

making a treatment decision.

Rational Decision

Making

dmR_10 I tend to...Learn as much as I can about possible

consequences before making decisions.

Tool Participation ioT_4 Patients tend to... Request medications they've heard

about in social media.

Tool Participation ioT_5 Patients tend to... Request medications they've heard

about from other patients.

Tool Participation ioT_10 Patients tend to...Choose the brand of IgG replacement

therapy they want.

198

Appendix E: Study 3 Interview Guide

Introduction and Explanation—read to the interviewee before proceeding

a. Greeting “Hello [name of participant]. Thank you for taking the time to meet

with me today. Your participation is greatly appreciated. Before getting started,

there are a couple things I would like to cover.”

b. Purpose and Format of Interview “As a current student in the Case Western

Reserve University Doctor of Management (DM) program, I am interested in

developing a greater understanding of the factors that influence physician decision

making for the treatment and management of PID. I will ask you a series of open-

ended questions on this topic, and I will ask one or more follow-up questions as

you respond. The interview will last approximately 60 to 90 minutes.”

c. Confidentiality “Everything you share in this interview will be kept in strictest

confidence, and your comments will be transcribed anonymously—omitting your

name, anyone else you refer to in this interview, as well as the name of your

current organization and/or past organizations. Your interview responses will be

included with all the other interviews I conduct.”

d. Recording “To help me capture your responses accurately and without being

overly distracting by taking notes, I would like to record our conversation with

your permission. Again, your responses will be kept confidential. If at any time,

you are uncomfortable with this interview, please let me know and I will turn the

recorder off.”

a. “Do you have any questions before we begin?”

Introduction

1. Name

2. Education

3. Current job title and responsibilities

4. Years of experience (total + specialty as an immunologist)

5. Involved with research (clinical or otherwise)

6. Practice setting/site of care (multiple?)

7. How many PID patients have you treated in your career/how many now?

199

8. What is the average length of time with the patient?

9. Most have co-morbidities?

10. Are all receiving IG: IV or ScIG?

Focus in on factors that determine how you make decisions

1. Describe a typical patient

a. How do they get to you?

b. Diagnosis to treatment to maintenance

2. Describe the types of decisions you make?

a. SC vs IV

3. Do you use decision aids (describe/evidence-based?)

4. Do you find that your patients are educated or well educated on self-

management?

a. High levels of self-efficacy (I am confident I can manage my situation)

b. High levers of self-activation

c. Do you think more informed patients result in fewer health resources

and better outcomes?

d. Are you patients actively involved in patient networks (IDF, JMF,

internet networks such as patients-like-me)?

e. Do you have a patient portal where patients can review their health

history?

5. What seeing a patient: look for patterns that match past experience?

a. Tend to quickly assess symptoms and diagnosis or it it’s a slow

painstaking process that has lots of complexities (particularly with co-

morbidities)

b. Tend to spend more time on protocol or treatment

c. Would you describe yourself as patient-centric or evidence centric

(EBM)?

6. If you were to guess: do you as the physician make the final treatment

decision or leave it up to the patient

200

a. Protocols

b. Drugs

7. Does your DM style vary based on the patients’ level of understanding and

interest?

a. Does it vary based on complexity or uncertainty?

8. Do you think your status as a physician or authority figure influences how

patients respond to you in a clinical setting?

a. Is Intimidated or encouraged to share information or

b. More likely to tell you their treatment preferences or express an option

on treatment options?

9. Do you routinely ask about patient preferences: lifestyle and how treatment

will affect patient goals and values?

10. When discussing pros and cons of a potential treatment (protocol or drugs) do

you tend to lead with the pros or cons

11. How much and in what way do cost or reimbursement influence your decision

making

a. Protocol

b. Drug

12. Do patients want to play an active role in their decisions

a. Function of health literacy

b. Function of health numeracy

13. Do you see yourself a continuation of care or coordinator of care?

a. Is coordination of care an issue for your PID patients

b. Are patients actively looking for you to coordinate their care?

c. Do your PID patients have problems accessing healthcare services or

getting adequate treatment?

14. Have you ever made a mistake/misdiagnosis?

a. Proper follow-up

b. Diagnostic test

c. Adequate history

201

15. Prefer face-to-face or is phone or internet possible and how often

16. Trustimpact on patient participation

17. Does DM change over time as uncertainty changes

a. Less shared decision making

18. Provider number

19. Are you rated? Do you recall your score?

20. How would you assess or describe the quality of your communication style?

a. Outgoing

b. seeking

21. Do patients tend to speak up or does it depend of the context and your

relationship

22. Impact of the organization on your decision making

a. IOM: patient-centered care: patient perspectives are now being

factored into Medicare value-based payments to hospitals

202

Appendix F: Mixed Methods Tutorial

Research and Analysis Methods

This section summarizes the methods used in this study to gather and analyze

data. The topics are as follows: mixed-methods, qualitative methods, and then

quantitative methods.

Mixed Methods

This research follows an exploratory and developmental mixed-methods10

approach that “combines quantitative and qualitative research methods in the same

research inquiry [to] help develop rich insights that cannot be fully understood using only

[one] method” (Venkatesh et al., 2013: 21). It consists of two sequential parts:

qualitative, then quantitative. Taking a developmental approach means we adapted and

follow questions that emerge from the different parts of the study (Venkatesh et al.,

2013). A sequential approach means that each study is separate and without overlap. An

exploratory approach means that the qualitative findings were tested by the quantitative

study (Creswell & Plano Clark, 2007). The triangulation, or merging, of both the

qualitative and quantitative approaches helps to identify factors of the physician decision-

making process that facilitate or impede SDM (Creswell & Plano Clark, 2007).

Mixed methods offers a more robust foundation to this research and the results

through the triangulation of the different methods (Caracelli & Greene, 1993; Gibson,

2017). The advantage of mixed methods over single-method studies is the ability to

10

Mixed methods should not be confused with a multi-method approach; multi-method involves two or

more of the same type of research approach for an individual research inquiry (i.e., two qualitative

methods) Venkatesh, V., Brown, S. A., & Bala, H. 2013. Bridging the qualitative-quantitative divide:

Guidelines for conducting mixed methods research in information systems. Mis Quarterly, 37(1): 21-54..

203

address both exploratory and confirmatory research questions, provide stronger

inferences, reduce common method bias, and provide more divergent or complementary

views (Teddlie & Tashakkori, 2003, 2009). Additionally, mixed methods can improve the

validity and reliability of research findings and are beneficial in studies where the

research area is not well documented or previous studies yielded contradictory research

findings (Greene, Caracelli, & Graham, 1989; Silverman, 2011; Tashakkori & Teddlie,

2008). “Quantitative studies typically rely on quality criteria, such as internal validity,

generalizability, and reliability, whereas qualitative studies have roughly comparable

quality criteria of credibility, transferability, and dependability” (Wisdom, Cavaleri,

Onwuegbuzie, & Green, 2012). The triangulated results from both studies capitalizes on

the strengths of each method, likely providing a more complete understanding of

influencing factors involved in a physician’s decision process for the treatment of PID

(Wisdom et al., 2012). The following sections discuss the strengths and weaknesses of

qualitative and quantitative methods.

Qualitative Methods

Qualitative methods assume the constructivist and interpretivist paradigms, with

an emphasis on process, constant change and meaning (Tashakkori & Teddlie, 2008).

This method is very effective in identifying patterns and commonalities in participant

responses and themes of a phenomenon; interviews can be “one of the most important

data gathering techniques for qualitative researchers in business and management …

interviews allow us to gather rich data from people in various roles and situations”

(Myers, 2013: 119). Furthermore, a qualitative approach may offer a more nuanced

understanding of the treatment process of complex chronic diseases.

204

The qualitative methods used in this research followed grounded theory. Meaning

study was conducted in stages: data collection, open-coding, axial coding and focused

coding (Charmaz, 2014; Corbin & Strauss, 2008). Open-coding is the preliminary

analysis and labeling of the data, axial coding is the grouping of open-coded labels, and

focused coding is constructing a formal framework with a variable (Charmaz, 2014;

Weick, Sutcliffe, & Obstfeld, 2008). Simultaneous data collection and analysis facilitated

the identification and pursuit of themes that shaped data collection and frame the

emerging analysis (O’Reilly, Paper, & Marx, 2012). Despite the advantages of this

method, there are also weaknesses of qualitative methods (Error! Reference source not

found.).

Table 36. Strengths and Weaknesses of Qualitative Methods

(Johnson & Onwuegbuzie, 2004)

Suggested samples sizes of greater than 20 participants are usually suggested to

achieve saturation (Creswell, 2013a; Glaser, 2017; Morse, 1994). Adequate physician

205

response rates to surveys is a well-documented problem (Flanigan et al., 2008). The

findings of the qualitative phase will guide the instrument development and hypothesis

testing for the subsequent quantitative study, in part by validating the scales used

(Creswell, 2013b; Venkatesh et al., 2013). Completion of the qualitative interviews

occurs when the information has reached a saturation point (i.e., when additional

interviews are not providing more themes) (Charmaz, 2014; Golden-Biddle & Locke,

2007).

Quantitative Methods

Quantitative methods assume a paradigm of positivism in testing and validating

qualitative research findings in a larger context by estimating the prevalence factors and

causal links between them (Johnson & Onwuegbuzie, 2004). Despite the advantages of

this method, there are also weaknesses of quantitative methods (Error! Reference

source not found.).

206

Table 37. Strengths and Weaknesses of Quantitative Methods

(Johnson & Onwuegbuzie, 2004)

Constructs and Validity

Construct development and their validity is argued to be the most important and

significant part of any study (Podsakoff, Podsakoff, MacKenzie, & Klinger, 2013). A

scale, or set of items, can only measure a construct if (a) the [construct] exists, and (b)

variations in the [construct] causally produce variations in the outcomes of the

measurement process” (Borsboom, Cramer, Kievit, Zand Scholten, & Franić, 2009: 150).

Furthermore, the observed constructs must avoid redundancy (Shaffer, DeGeest, & Li,

2016). Strategies to increase the validity of the constructs include testing them before

207

collecting data and conducting tests during the analyses post-collection (Worthington &

Whittaker, 2006).

To increase the construct validity of the instrument, the survey items should be

pre-tested for using the Q-sort technique. Q-sort is a powerful and well-established

technique for quantitatively evaluating opinions and attitudes (Thomas & Watson, 2002).

Like other pretesting techniques, the method can be used to establish content, and

discriminant and convergent validity before the post-data collection analysis (Bolton,

1993; Krosnick, 1999). Using software such as Qualtrics, mock-participants can be asked

to drag randomized questions into categories they believe it relates to potentially unveil

cross-loading items.

After data collection, validity is measured during the analysis. The exploratory

factor analysis (EFA) was performed to conduct tests to confirm or increase the validity

of the item set; the tests include the Kaiser-Meyer-Olkin test (KMO) for adequacy,

rotation type to estimate a simple structure for the data, cross-loading analysis, and item

deletion (Worthington & Whittaker, 2006). An EFA is suitable for this study because

formal hypotheses are tested and the number of factors to be analyzed are data-driven

(Fabrigar, Wegener, MacCallum, & Strahan, 1999). Multiple EFAs were conducted if

necessary as depending on results from subsequent steps in the analysis (MacKenzie,

Podsakoff, & Podsakoff, 2011).

SEM Methods

One method of analysis for quantitative data is called structural equation

modeling (SEM) and involves the use of SPSS AMOS software version 23. SEM is a

multivariate statistical analysis technique used to analyze structural relationships by

208

combining factor analyses and linear regression models (Williams, Vandenberg, &

Edwards, 2009). This analysis is appropriate when the goal of the research is to assess

direct, mediating, and moderating relationships between variables within a specific

hypothesized model (Anderson & Gerbing, 1988). Mediation was tested using a

bootstrapping method running 2,000 iterations and 95% bias-corrected confidence

intervals “to gauge the extent and significance of indirect effects” (Preacher, Rucker, &

Hayes, 2007: 189). Interactions (moderation) were analyzed using a chi-square difference

test to determine the conditions under which an effect varies in size (Aguinis, Edwards, &

Bradley, 2016). Model fit is observed throughout all analyses (Error! Reference source

not found.).

Table 38. Model Fit Thresholds

Measure Threshold Citation

Chi-SQ N/A (Hair et al., 2010)

Df N/A (Hair et al., 2010)

Chi-square / df

(cmin / df)

< 3.0 (Bagozzi & Heatherton, 1994; Baumgartner, 2010; Hair et

al., 2010; Matsunaga, 2010)

p-value for model > .05 (Hair et al., 2010)

CFI > .90 (Baumgartner, 2010; Hair et al., 2010; Hu & Bentler, 1999;

Matsunaga, 2010)

GFI > .95 (Bollen, 1989; Hair et al., 2010)

AGFI > .80 (Hair et al., 2010)

PGFI (Hair et al., 2010)

NFI > .9 (Bentler, 1990; Bentler & Bonett, 1980; Hair et al., 2010)

PNFI (Hair et al., 2010)

RMSEA < .05 (Kenny, 2016)

PCLOSE > .05 (Costello & Osborne, 2005)

RMR < .09 (Hair et al., 2010)

SRMR < .08 (Hair et al., 2010; Hu & Bentler, 1999)

209

In summary, “quantitative studies typically rely on quality criteria, such as

internal validity, generalizability, and reliability, whereas qualitative studies have roughly

comparable quality criteria of credibility, transferability, and dependability” (Wisdom et

al., 2012: 724). It is the combination of these methods that provides the best picture of the

factors that affect SDM implementation, since, through triangulation, a better

understanding of the phenomenon can be achieved. This study integrated the two

methods at the data analysis and interpretation stage after the data was collected. The

analysis for each of the three studies was done separately for each study and is provided

in Chapters 4, 5 and 6. The information was compared at the interpretation stage of the

research and is provided in Chapter 7 (Creswell & Plano Clark, 2007: 175; Hossler &

Vesper, 1993)

210

REFERENCES

Abolhassani, H., Aghamohammadi, A., Pourjabbar, S., Sadaghiani, M. S., Nikayin, S.,

Rabiee, A., Imanzadeh, A., Gharaei, J. M., Arbabi, M., & Rezaei, N. 2013.

Psychiatric aspects of primary immunodeficiency diseases: the parental study.

Iranian Journal Of Allergy, Asthma And Immunology, 12(2): 176-181.

Agency for Healthcare Research and Quality. 2014a. Mean Expenses per Person with

Care for Selected Conditions by Type of Service: United States, 2014. .

Agency for Healthcare Research and Quality. 2014b. MEPS HC-171: 2014 Full Year

Consolidated Data File. In Agency for Healthcare Research and Quality (Ed.).

Agency for Healthcare Research and Quality. 2014c. Table 4.1: Among adults age 18 and

over who reported going to a doctor's office or clinic in the last 12 months,

percent distribution of how often their health providers listened carefully to them,

United States, 2014., Medical Expenditure Panel Survey Household Component

Data, Vol. 2017.

Agency for Healthcare Research and Quality. 2014d. Table 4.5: Among adults age 18 and

over who reported going to a doctor's office or clinic in the last 12 months,

percent distribution of how often their health providers showed respect for what

they had to say, United States, 2014., Vol. 2017.

Agency for Healthcare Research and Quality. 2014e. Table 4.7: Among adults age 18 and

over who reported going to a doctor's office or clinic in the last 12 months,

percent distribution of how often their health providers spent enough time with

them, United States, 2014., Vol. 2017.

Agency for Healthcare Research and Quality. 2014f. Total Expenses and Percent

Distribution for Selected Conditions by Type of Service: United States, 2014.,

Vol. 2017.

Agency for Healthcare Research and Quality. 2018. MEPS: Survey Background, Vol.

2018.

Aggarwal, A., Davies, J., & Sullivan, R. 2014. “Nudge” in the clinical consultation–an

acceptable form of medical paternalism? BMC Medical Ethics, 15(31): 1-6.

Aguinis, H., Edwards, J. R., & Bradley, K. J. 2016. Improving our understanding of

moderation and mediation in strategic management research. Organizational

Research Methods, 20(2): 665-685.

Al-Herz, W., Bousfiha, A., Casanova, J.-L., Chatila, T., Conley, M. E., Cunningham-

Rundles, C., Etzioni, A., Franco, J. L., Gaspar, H. B., Holland, S. M., Klein, C.,

Nonoyama, S., Ochs, H. D., Oksenhendler, E., Picard, C., Puck, J. M., Sullivan,

K., & Tang, M. L. K. 2014. Primary immunodeficiency diseases: An update on

211

the classification from the international union of immunological societies expert

committee for primary immunodeficiency. Frontiers in Immunology, 5(Article

162): 1-33.

Allen, P., Maguire, S., & McKelvey, B. 2011a. The sage handbook of complexity and

management: Sage Publications.

Allen, P. M., Maguire, S., & McKelvey, B. 2011b. Complexity and management. In C.

Hooker (Ed.), Philosophy of complex systems handbook of the philosophy of

science, Vol. 10: 783-808: Elsevier.

Alston, C., Paget, L., Halvorson, G., Novelli, B., Guest, J., McCabe, P., Hoffman, K.,

Koepke, C., Simon, M., & Sutton, S. 2012. Communicating with patients on

health care evidence. Washington DC: Institute of Medicine.

Anderson, G., & Horvath, J. 2013. Chronic conditions: Making the case for ongoing

care. 2002. Princeton, NJ: Robert Wood Johnson Foundation’s Partnership for

Solutions.

Anderson, J. C., & Gerbing, D. W. 1988. Structural equation modeling in practice: A

review and recommended two-step approach. Psychological Bulletin, 103(3):

411.

Ansell, D. A. 2017. The Deah Gap: How Inequality Kills: University of Chicago Press.

Arterburn, D., Wellman, R., Westbrook, E., Rutter, C., Ross, T., McCulloch, D.,

Handley, M., & Jung, C. 2012. Introducing decision aids at Group Health was

linked to sharply lower hip and knee surgery rates and costs. Health Affairs,

31(9): 2094-2104.

Ashforth, B. E., & Mael, F. 1989. Social identity theory and the organization. Academy

of management review, 14(1): 20-39.

Babbie, E. R., Halley, F., & Zaino, J. 2007. Adventures in social research: data analysis

using SPSS 14.0 and 15.0 for Windows: Pine Forge Press.

Bagozzi, R. P., & Heatherton, T. F. 1994. A general approach to representing

multifaceted personality constructs: Application to state self‐esteem. Structural

Equation Modeling: A Multidisciplinary Journal, 1(1): 35-67.

Baker, D. P., Day, R., & Salas, E. 2006. Teamwork as an essential component of high‐

reliability organizations. Health Services Research, 41(4p2): 1576-1598.

Baker, D. W. 2017. History of The Joint Commission’s Pain Standards: Lessons for

Today’s Prescription Opioid Epidemic. JAMA, 317(11): 1117-1118.

212

Bansal, G., Zahedi, F. M., & Gefen, D. 2016. Do context and personality matter? Trust

and privacy concerns in disclosing private information online. Information &

Management, 53(1): 1-21.

Barnett, K., Mercer, S. W., Norbury, M., Watt, G., Wyke, S., & Guthrie, B. 2012.

Epidemiology of multimorbidity and implications for health care, research, and

medical education: a cross-sectional study. The Lancet, 380(9836): 37-43.

Barrett, L. F., Tugade, M. M., & Engle, R. W. 2004. Individual differences in working

memory capacity and dual-process theories of the mind. Psychological bulletin,

130(4): 553.

Barry, M. J., & Edgman-Levitan, S. 2012. Shared decision making—the pinnacle of

patient-centered care. New England Journal Of Medicine, 366(9): 780-781.

Barua, S., Greenwald, R., Grebely, J., Dore, G. J., Swan, T., & Taylor, L. E. 2015.

Restrictions for Medicaid reimbursement of Sofosbuvir for the Treatment of

Hepatitis C virus infection in the United States. Annals of Internal Medicine,

163(3): 215-223.

Bauer, A. M., Parker, M. M., Schillinger, D., Katon, W., Adler, N., Adams, A. S.,

Moffet, H. H., & Karter, A. J. 2014. Associations between antidepressant

adherence and shared decision-making, patient–provider trust, and

communication among adults with diabetes: diabetes study of northern California

(DISTANCE). Journal of general internal medicine, 29(8): 1139-1147.

Baumgartner, H. 2010. Structural Equation Modeling, Wiley International Encyclopedia

of Marketing. The Pennsylvania State University, University Park, PA, USA.

Beach, L. R., & Lipshitz, R. 2017. Why classical decision theory is an inappropriate

standard for evaluating and aiding most human decision making. In D. Harris, &

W.-C. Li (Eds.), Decision making in aviation: 85-102. London and New York:

Routledge.

Beevers, C. G. 2005. Cognitive vulnerability to depression: A dual process model.

Clinical psychology review, 25(7): 975-1002.

Belanger, E., Rodríguez, C., & Groleau, D. 2011. Shared decision-making in palliative

care: A systematic mixed studies review using narrative synthesis. Palliative

Medicine, 25(3): 242-261.

Bensing, J. 2000. Bridging the gap.: The separate worlds of evidence-based medicine and

patient-centered medicine. Patient Education And Counseling, 39(1): 17-25.

Bentler, P. M. 1990. Comparative fit indexes in structural models. Psychological

bulletin, 107(2): 238.

213

Bentler, P. M., & Bonett, D. G. 1980. Significance tests and goodness of fit in the

analysis of covariance structures. Psychological bulletin, 88(3): 588.

Bernardo, R. 2017. 2017’s Best & Worst States for Health Care, Vol. 2018: WalletHub.

Berner, E. S., & Graber, M. L. 2008. Overconfidence as a cause of diagnostic error in

medicine. The American journal of medicine, 121(5): S2-S23.

Bertakis, K. D., Franks, P., & Azari, R. 2002. Effects of physician gender on patient

satisfaction. Journal Of The American Medical Women's Association (1972),

58(2): 69-75.

Berwick, D. M., & Hackbarth, A. D. 2012. Eliminating waste in US health care. JAMA,

307(14): 1513-1516.

Beshears, J., & Gino, F. 2015. Leaders as decision architects. Harvard Business Review,

93(5): 52-62.

Blumenthal-Barby, J. 2017. ‘That’s the doctor’s job’: Overcoming patient reluctance to

be involved in medical decision making. Patient education and counseling,

100(1): 14-17.

Boland Jr, R. J., & Collopy, F. 2004. Managing as designing: Stanford University Press.

Bollen, K. A. 1989. A new incremental fit index for general structural equation models.

Sociological Methods & Research, 17(3): 303-316.

Bolton, R. N. 1993. Pretesting questionnaires: content analyses of respondents'

concurrent verbal protocols. Marketing science, 12(3): 280-303.

Borrell-Carrió, F., Estany, A., Platt, F. W., & MoralesHidalgo, V. 2014. Doctors as a

knowledge and intelligence building group: Pragmatic principles underlying

decision-making processes. Journal of Epidemiology and Community Health,

69(4): 303-305.

Borsboom, D., Cramer, A. O., Kievit, R. A., Zand Scholten, A., & Franić, S. 2009. The

end of construct validity. In R. W. Lissitz (Ed.), The concept of validity:

Revisions, new directions, and applications: 135-170. Charlotte, NC: IAP

Information Age Publishing.

Bousfiha, A. A., Jeddane, L., Ailal, F., Benhsaien, I., Mahlaoui, N., Casanova, J.-L., &

Abel, L. 2013. Primary immunodeficiency diseases worldwide: more common

than generally thought. Journal Of Clinical Immunology, 33(1): 1-7.

Boyle, J., & Buckley, R. 2007. Population prevalence of diagnosed primary

immunodeficiency diseases in the United States. Journal Of Clinical

Immunology, 27(5): 497-502.

214

Braun, V., & Clarke, V. 2006. Using thematic analysis in psychology. Qualitative

research in psychology, 3(2): 77-101.

Bravo, P., Edwards, A., Barr, P. J., Scholl, I., Elwyn, G., & McAllister, M. 2015.

Conceptualising patient empowerment: A mixed methods study. BMC Health

Services Research, 15(252): 1-14.

Brooks, M. 2017. Rare Disease Treatments Make Up Top 10 Most Costly Drugs, Vol.

2017: Medscape.

Buck, J. N., & Daniels, M. H. 1985. Assessment of Career Decision Making (ACDM):

Manual: Western Psychological Services.

Burgess, D. J., Dovidio, J., Phelan, S., & van Ryn, M. 2011. The Effect of Medical

Authoritarianism on Physicians' Treatment Decisions and Attitudes Regarding

Chronic Pain. Journal Of Applied Social Psychology, 41(6): 1399-1420.

Burgess, D. J., Van Ryn, M., Crowley-Matoka, M., & Malat, J. 2006. Understanding the

provider contribution to race/ethnicity disparities in pain treatment: insights from

dual process models of stereotyping. Pain Medicine, 7(2): 119-134.

Butler, M., Ratner, E., McCreedy, E., Shippee, N., & Kane, R. L. 2014. Decision aids for

advance care planning: an overview of the state of the ScienceDecision aids for

advance care planning. Annals of internal medicine, 161(6): 408-418.

Byrne, B. M. 2004. Testing for multigroup invariance using AMOS graphics: A road less

traveled. Structural Equation Modeling, 11(2): 272-300.

Byrne, B. M. 2008. Testing for multigroup equivalence of a measuring instrument: A

walk through the process. Psicothema, 20(4): 872-882.

CAHPS. 2012. CG CAHPS for ACOs – Field Test Survey Content by Survey Domain

Overview, Vol. 2017.

Calder, L. A., Forster, A. J., Stiell, I. G., Carr, L. K., Brehaut, J. C., Perry, J. J.,

Vaillancourt, C., & Croskerry, P. 2011. Experiential and rational decision making:

A survey to determine how emergency physicians make clinical decisions.

Emergency Medicine Journal, 29(10): 811-816.

Campbell, R. 2013. Step inside the physician's head. EHRs can enhance a physician's

cognitive processing and eliminate diagnostic errors, with some HIM-led changes.

Journal of American Health Information Management Association, 84(11): 44.

Caracelli, V. J., & Greene, J. C. 1993. Data analysis strategies for mixed-method

evaluation designs. Educational evaluation and policy analysis, 15(2): 195-207.

215

Carman, K. L., & Workman, T. A. 2017. Engaging patients and consumers in research

evidence: applying the conceptual model of patient and family engagement.

Patient education and counseling, 100(1): 25-29.

Casto, A. B., Layman, E., & Association, A. H. I. M. 2006. Principles of healthcare

reimbursement: American Health Information Management Association Chicago.

Castro, F. G., Kellison, J. G., Boyd, S. J., & Kopak, A. 2010. A methodology for

conducting integrative mixed methods research and data analyses. Journal of

mixed methods research, 4(4): 342-360.

CDC. 2017. Center for disease control and prevention, Vol. 2017: CDC.

Certo, S. T., Busenbark, J. R., Woo, H. s., & Semadeni, M. 2016. Sample selection bias

and Heckman models in strategic management research. Strategic Management

Journal, 37(13): 2639-2657.

Chambers, D., Booth, A., Baxter, S. K., Johnson, M., Dickinson, K. C., & Goyder, E. C.

2016. Evidence for models of diagnostic service provision in the community:

literature mapping exercise and focused rapid reviews. Health Services and

Delivery Research, 4(35): 1-362.

Chang, S.-J., Van Witteloostuijn, A., & Eden, L. 2010. From the editors: Common

method variance in international business research. Journal Of International

Business Studies, 41(2): 178-184.

Chapel, H., Prevot, J., Gaspar, H. B., Español, T., Bonilla, F. A., Solis, L., & Drabwell, J.

2014. Primary immune deficiencies–principles of care. Frontiers In

Immunology, 5.

Chapman, E. N., Kaatz, A., & Carnes, M. 2013. Physicians and implicit bias: how

doctors may unwittingly perpetuate health care disparities. Journal Of General

Internal Medicine, 28(11): 1504-1510.

Charles, C., Gafni, A., & Whelan, T. 1999. Decision-making in the physician–patient

encounter: revisiting the shared treatment decision-making model. Social Science

& Medicine, 49(5): 651-661.

Charmaz, K. 2006. Constructing grounded theory: A practical guide through

qualitative research. London: Sage Publications.

Charmaz, K. 2014. Constructing grounded theory: Sage.

Chen, S.-L. 2016. Economic Costs of Hemophilia and the Impact of Prophylactic

Treatment on Patient Management. The American Journal Of Managed Care,

22(5 Suppl): s126-133.

216

Christozov, D., Chukova, S., & Mateev, P. 2009. Informing processes, risks, evaluation

of the risk of misinforming, in foundations of informing science. In T. G. Gill, &

E. Cohen (Eds.), Foundations of informing science: 323-356. Santa Rosa, CA:

Informing Science Press.

Chu, L. F., Utengen, A., Kadry, B., Kucharski, S. E., Campos, H., Crockett, J., Dawson,

N., & Clauson, K. A. 2016. “Nothing about us without us”—patient partnership in

medical conferences. BMJ, 354(i3883).

CMS. 2015. National Health Expenditures 2015 Highlights: Centers for Medicare &

Medicaid Services.

CMS. 2016. Medicare Shared Savings Program quality measure benchmarks for the 2016

and 2017 reporting years, Vol. 2018.

CMS. 2017. NHE Fact Sheet, Vol. 2017: CMS.

Coget, J.-F., & Keller, E. 2010. The critical decision vortex: lessons from the emergency

room. Journal Of Management Inquiry, 19(1): 56-67.

Coghlan, A. 2017. US ranked worst healthcare system, while the NHS is the best, New

Scientist.

Cole, J., Kiriaev, O., Malpas, P., & Cheung, G. 2017. ‘Trust me, I’m a doctor’: A

qualitative study of the role of paternalism and older people in decision-making

when they have lost their capacity. Australasian Psychiatry, 25(6): 549-553.

Condino-Neto, A., Franco, J., Espinosa-Rosales, F., Leiva, L., King, A., Porras, O.,

Oleastro, M., Bezrodnik, L., Grumach, A. S., & Costa-Carvalho, B. 2012.

Advancing the management of primary immunodeficiency diseases in Latin

America: Latin American Society for Immunodeficiencies (LASID) Initiatives.

Allergologia Et Immunopathologia, 40(3): 187-193.

Cook, K., Kramer, R., Thom, D., Stepanikova, I., Bailey, S., & Cooper, R. 2004. Trust

and distrust in patient-physician relationships: Perceived determinants of high and

low trust relationships in managed care settings. In R. M. Kramer, & K. S. Cook

(Eds.), Trust and distrust in organizations: Dilemmas and approaches: 65-98.

Thousand Oaks, CA: Russell Sage Foundation.

Cooper-Patrick, L., Gallo, J. J., Gonzales, J. J., Vu, H. T., Powe, N. R., Nelson, C., &

Ford, D. E. 1999. Race, gender, and partnership in the patient-physician

relationship. JAMA, 282(6): 583-589.

Cooper, L. A., & Roter, D. 2003. Patient-provider communication: The effect of race and

ethnicity on process and outcomes of healthcare. In B. D. Smedley, A. Y. Stith, &

A. R. Nelson (Eds.), Unequal treatment: Confronting racial and ethnic

disparities in health care. Washington, DC: National Academies Press.

217

Cooper, L. A., Roter, D. L., Carson, K. A., Beach, M. C., Sabin, J. A., Greenwald, A. G.,

& Inui, T. S. 2012. The associations of clinicians’ implicit attitudes about race

with medical visit communication and patient ratings of interpersonal care.

American journal of public health, 102(5): 979-987.

Corbin, J., & Strauss, A. 2008. Basics of qualitative research (3rd ed.). London: Sage.

Corbin, J., & Strauss, A. 2014. Basics of qualitative research: Techniques and

procedures for developing grounded theory: Sage publications.

Costa-Carvalho, B. T., Grumach, A. S., Franco, J. L., Espinosa-Rosales, F. J., Leiva, L.

E., King, A., Porras, O., Bezrodnik, L., Oleastro, M., & Sorensen, R. U. 2014.

Attending to warning signs of primary immunodeficiency diseases across the

range of clinical practice. Journal Of Clinical Immunology, 34(1): 10-22.

Costello, A. B., & Osborne, J. 2005. Best practices in exploratory factor analysis: Four

recommendations for getting the most from your analysis. Practical Assessment

Research & Evaluation, 10(7): 1-9.

Cottone, R. R. 2001. A social constructivism model of ethical decision making in

counseling. Journal of Counseling & Development, 79(1): 39-45.

Cottone, R. R. 2004. Displacing the psychology of the individual in ethical decision-

making: The social constructivism model. Canadian Journal of Counselling,

38(1): 5.

Couët, N., Desroches, S., Robitaille, H., Vaillancourt, H., Leblanc, A., Turcotte, S.,

Elwyn, G., & Légaré, F. 2015. Assessments of the extent to which health‐care

providers involve patients in decision making: a systematic review of studies

using the OPTION instrument. Health Expectations, 18(4): 542-561.

Creswell, J. W. 2013a. Qualitative inquiry and research design: Choosing among five

approaches: Sage.

Creswell, J. W. 2013b. Steps in conducting a scholarly mixed methods study: DBER

Speaker Series. 48. http://digitalcommons.unl.edu/dberspeakers/48.

Creswell, J. W., & Plano Clark, V. L. 2007. Designing and conducting mixed methods

research: Sage Publications.

Croskerry, P. 2009a. Clinical cognition and diagnostic error: applications of a dual

process model of reasoning. Advances In Health Sciences Education, 14(1): 27-

35.

Croskerry, P. 2009b. A universal model of diagnostic reasoning. Academic Medicine,

84(8): 1022-1028.

218

Croskerry, P. 2014. Bias: a normal operating characteristic of the diagnosing brain.

Diagnosis, 1(1): 23-27.

Curreri, A., & Lyytinen, K. 2017. Mindfulness, Information Technology Use, and

Physicians’ Performance in Emergency Rooms. Paper presented at the Academy

of Management Proceedings.

D'Amour, D., Ferrada-Videla, M., San Martin Rodriguez, L., & Beaulieu, M.-D. 2005.

The conceptual basis for interprofessional collaboration: core concepts and

theoretical frameworks. Journal of interprofessional care, 19(sup1): 116-131.

Dalton, D. R. 2015. Hemophilia in the managed care setting. The American journal of

managed care, 21(6 Suppl): S123-130.

Davis, M. S. 1971. That's interesting! Towards a phenomenology of sociology and a

sociology of phenomenology. Philosophy of the social sciences, 1(2): 309-344.

Delaney, R., Strough, J., Parker, A. M., & de Bruin, W. B. 2015. Variations in decision-

making profiles by age and gender: A cluster-analytic approach. Personality and

individual differences, 85: 19-24.

DeMeester, R. H., Lopez, F. Y., Moore, J. E., Cook, S. C., & Chin, M. H. 2016. A model

of organizational context and shared decision making: application to LGBT racial

and ethnic minority patients. Journal of general internal medicine, 31(6): 651-

662.

DeSantis, C. E., Ma, J., Goding Sauer, A., Newman, L. A., & Jemal, A. 2017. Breast

cancer statistics, 2017, racial disparity in mortality by state. CA: A Cancer

Journal for Clinicians, 67(6): 439-448.

Desroches, S. 2010. Shared decision making and chronic diseases. Allergy, Asthma &

Clinical Immunology, 6(4): A8.

Dewey, J. 1925. Experience and nature. New York, NY: Dover Publications.

Dieleman, J. L., Squires, E., Bui, A. L., Campbell, M., Chapin, A., Hamavid, H., Horst,

C., Li, Z., Matyasz, T., & Reynolds, A. 2017. Factors Associated With Increases

in US Health Care Spending, 1996-2013. Jama, 318(17): 1668-1678.

DiMichele, D. M. 2008. Inhibitors in hemophilia: A primer (4th ed.). Montreal, Quebec:

World Federation of Hemophilia.

Djulbegovic, B., & Guyatt, G. H. 2017. Progress in evidence-based medicine: A quarter

century on. The Lancet, 390(10092): 415-423.

219

Djulbegovic, B., Hozo, I., Beckstead, J., Tsalatsanis, A., & Pauker, S. G. 2012. Dual

processing model of medical decision-making. BMC Medical Informatics And

Decision Making, 12(1): 94.

Du, L., & Lu, W. 2016. U.S. Health-Care System Ranks as One of the Least-Efficient,

Vol. 2017: Bllomberg.

Durning, S. J., Dong, T., Artino, A. R., van der Vleuten, C., Holmboe, E., & Schuwirth,

L. 2015. Dual processing theory and expertsʼ reasoning: exploring thinking on

national multiple-choice questions. Perspectives On Medical Education, 4(4):

168-175.

Eapen, Z. J., & Jain, S. H. 2017. Redesigning care for high-cost, high-risk patients,

Harvard Business Review.

Ehlayel, M. S., Bener, A., & Laban, M. A. 2013. Primary immunodeficiency diseases in

children: 15 year experience in a tertiary care medical center in Qatar. Journal Of

Clinical Immunology, 33(2): 317-324.

Eisenberg, L. 1977. Disease and illness Distinctions between professional and popular

ideas of sickness. Culture, medicine and psychiatry, 1(1): 9-23.

Eisenhardt, K. M. 1989. Agency theory: An assessment and review. Academy of

management review, 14(1): 57-74.

Eldar-Lissai, A., Hou, Q., & Krishnan, S. 2015. The changing costs of caring for

Hemophilia Patients in the US: insurers’ and patients’ perspectives. Value in

Health, 18(3): A304.

Elwyn, G., Frosch, D., Thomson, R., Joseph-Williams, N., Lloyd, A., Kinnersley, P.,

Cording, E., Tomson, D., Dodd, C., & Rollnick, S. 2012. Shared decision making:

a model for clinical practice. Journal Of General Internal Medicine, 27(10):

1361-1367.

Elwyn, G., Laitner, S., Coulter, A., Walker, E., Watson, P., & Thomson, R. 2010.

Implementing shared decision making in the NHS. BMJ, 357: j1744.

Ely, J. W., Kaldjian, L. C., & D'Alessandro, D. M. 2012. Diagnostic errors in primary

care: lessons learned. The Journal Of The American Board Of Family

Medicine, 25(1): 87-97.

Eraker, S. A., & Politser, P. 1982. How decisions are reached: physician and patient.

Annals of Internal Medicine, 97(2): 262-268.

Erikson, C., Jones, K., & Tilton, C. 2012. Physician specialty data book. Washington,

DC: Association of American Medical Colleges.

220

Espinosa-Rosales, F. J., Condino-Neto, A., Franco, J. L., & Sorensen, R. U. 2016. Into

action: Improving access to optimum care for all primary immunodeficiency

patients. Journal of Clinical Immunology, 36(5): 415-417.

Evans, J. S. B., & Stanovich, K. E. 2013. Dual-process theories of higher cognition

advancing the debate. Perspectives On Psychological Science, 8(3): 223-241.

Eysenck, H. J. 1963. Biological basis of personality. Nature, 199(4898): 1031-1034.

Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. 1999. Evaluating the

use of exploratory factor analysis in psychological research. Psychological

methods, 4(3): 272.

Fiske, S. T., Dupree, C. H., Nicolas, G., & Swencionis, J. K. 2016. Status, power, and

intergroup relations: The personal is the societal. Current opinion in psychology,

11: 44-48.

Flanigan, T. S., McFarlane, E., & Cook, S. 2008. Conducting survey research among

physicians and other medical professionals: A review of current literature.

Paper presented at the Proceedings of the Survey Research Methods Section,

American Statistical Association.

Flynn, D. 2003. Non-medical influences upon medical decision-making and referral

behavior: an annotated bibliography: Greenwood Publishing Group.

Fram, A., & Freking, K. 2017. Sanders would make government health care role even

bigger, U.S. News & World Report.

Friesen-Storms, J. H., Bours, G. J., van der Weijden, T., & Beurskens, A. J. 2015. Shared

decision making in chronic care in the context of evidence based practice in

nursing. International Journal Of Nursing Studies, 52(1): 393-402.

Frosch, D. L., & Kaplan, R. M. 1999. Shared decision making in clinical medicine: past

research and future directions. American Journal Of Preventive Medicine, 17(4):

285-294.

Frosch, D. L., Moulton, B. W., Wexler, R. M., Holmes-Rovner, M., Volk, R. J., & Levin,

C. A. 2011. Shared decision making in the United States: policy and

implementation activity on multiple fronts. Zeitschrift Für Evidenz, Fortbildung

Und Qualität Im Gesundheitswesen, 105(4): 305-312.

Frost, J., & Massagli, M. 2009. PatientsLikeMe the case for a data-centered patient

community and how ALS patients use the community to inform treatment

decisions and manage pulmonary health. Chronic respiratory disease, 6(4): 225-

229.

221

Gabel, S. 2012. Perspective: physician leaders and their bases of power: common and

disparate elements. Academic Medicine, 87(2): 221-225.

Gallan, A. S., Jarvis, C. B., Brown, S. W., & Bitner, M. J. 2013. Customer positivity and

participation in services: an empirical test in a health care context. Journal Of

The Academy Of Marketing Science, 41(3): 338-356.

Gardulf, A., & Nicolay, U. 2006. Replacement IgG therapy and self-therapy at home

improve the health-related quality of life in patients with primary antibody

deficiencies. Current Opinion In Allergy And Clinical Immunology, 6(6): 434-

442.

Gardulf, A., Winiarski, J., Thorin, M., Arnlind, M. H., von Döbeln, U., & Hammarström,

L. 2017. Costs associated with treatment of severe combined

immunodeficiency—rationale for newborn screening in Sweden. Journal of

Allergy and Clinical Immunology, 139(5): 1713-1716. e1716.

Gerteis, J., Izrael, D., Deitz, D., LeRoy, L., Ricciardi, R., Miller, T., & Basu, J. 2014.

Multiple chronic conditions chartbook. Rockville, MD: Agency for Healthcare

Research and Quality (AHRQ) Publications.

Gibson, C. B. 2017. Elaboration, Generalization, Triangulation, and Interpretation: On

enhancing the value of mixed method research. Organizational Research

Methods, 20(2): 193-223.

Gigerenzer, G. 2015. Simply rational: Decision making in the real world: Oxford

University Press, USA.

Gigerenzer, G., & Goldstein, D. G. 1996. Reasoning the fast and frugal way: models of

bounded rationality. Psychological Review, 103(4): 650.

Giguère, A., Labrecque, M., Njoya, M., Thivierge, R., & Légaré, F. 2012. Development

of PRIDe: A tool to assess physicians’ preference of role in clinical decision

making. Patient Education And Counseling, 88(2): 277-283.

Glaser, B. 2017. Discovery of grounded theory: Strategies for qualitative research:

Routledge.

Glaser, B., & Strauss, A. 2017. Discovery of grounded theory: Strategies for qualitative

research: Routledge.

Godager, G., Henning-Schmidt, H., & Iversen, T. 2016. Does performance disclosure

influence physicians’ medical decisions? An experimental analysis. Journal of

Economic Behavior and Organization, 131: 36-46.

Godolphin, W. 2009. Shared decision-making. Healthcare Quarterly, 12(Sp): e186-e190.

222

Goffman, E. 1959. The moral career of the mental patient. Psychiatry, 22(2): 123-142.

Golden-Biddle, K., & Locke, K. 2007. Composing qualitative research: Sage.

Goodnough, L. T., Shander, A., & Brecher, M. E. 2003. Transfusion medicine: looking to

the future. The Lancet, 361(9352): 161-169.

Graber, M. L., Franklin, N., & Gordon, R. 2005. Diagnostic error in internal medicine.

Archives Of Internal Medicine, 165(13): 1493-1499.

Gravel, K., Légaré, F., & Graham, I. D. 2006. Barriers and facilitators to implementing

shared decision-making in clinical practice: A systematic review of health

professionals' perceptions. Implementation Science, 1(1): 16.

Green, A. R., Carrillo, J. E., & Betancourt, J. R. 2002. Why the disease-based model of

medicine fails our patients. Western journal of Medicine, 176(2): 141.

Greene, J., & McClintock, C. 1985. Triangulation in evaluation: Design and analysis

issues. Evaluation Review, 9(5): 523-545.

Greene, J. C., Caracelli, V. J., & Graham, W. F. 1989. Toward a conceptual framework

for mixed-method evaluation designs. Educational Evaluation And Policy

Analysis, 11(3): 255-274.

Grembowski, D., Schaefer, J., Johnson, K. E., Fischer, H., Moore, S. L., Tai-Seale, M.,

Ricciardi, R., Fraser, J. R., Miller, D., & LeRoy, L. 2014. A conceptual model of

the role of complexity in the care of patients with multiple chronic conditions.

Medical care, 52: S7-S14.

Grifols. 2017. Investors’ & Analysts’ Meeting 2017: Grifols.

Gringeri, A. 2011. Factor VIII safety: plasma-derived versus recombinant products.

Blood Transfusion, 9(4): 366.

Gringeri, A., Mantovani, L. G., Scalone, L., & Mannucci, P. M. 2003. Cost of care and

quality of life for patients with hemophilia complicated by inhibitors: the COCIS

Study Group. Blood, 102(7): 2358-2363.

Groopman, J. 2008. How doctors think: Houghton Mifflin Harcourt.

Groopman, J., & Hartzband, P. 2012. Your medical mind: How to decide what is right

for you: Penguin Books.

Groot, G., Waldron, T., Carr, T., McMullen, L., Bandura, L.-A., Neufeld, S.-M., &

Duncan, V. 2017. Development of a program theory for shared decision-making:

a realist review protocol. Systematic Reviews, 6(1): 114.

223

Guaní-Guerra, E., García-Ramírez, U. N., Jiménez-Romero, A. I., Velázquez-Ávalos, J.

M., Gallardo-Martínez, G., & Mendoza-Espinoza, F.-J. 2013. Primary

immunodeficiency diseases at reference and high-specialty hospitals in the state

of Guanajuato, Mexico. Biomed Research International, 2013(Article ID

187254).

Guyatt, G. H. 1991. Evidence-based medicine. ACP Journal Club, 114(2): A16-A16.

Hair, J. F. J., Black, W. C., Babin, B. J., & Anderson, R. E. 2010. Multivariate Data

Analysis (7th Edition ed.): Pearson Prentice Hall.

Hajjaj, F., Salek, M., Basra, M., & Finlay, A. 2010. Non-clinical influences on clinical

decision-making: a major challenge to evidence-based practice. Journal Of The

Royal Society Of Medicine, 103(5): 178-187.

Harden, A. 2010. Mixed-methods systematic reviews: Integrating quantitative and

qualitative findings. Focus, 25: 1-8.

Hardin, G. 1985. Filters against folly: Viking Books.

Hathi, S., & Kocher, R. 2017. The right way to reform health care: To cut costs, empower

patients. Foreign Affairs, 96(4): 17.

Hawley, S. T., & Morris, A. M. 2017. Cultural challenges to engaging patients in shared

decision making. Patient education and counseling, 100(1): 18-24.

Hayes, A. F. 2009. Beyond Baron and Kenny: Statistical mediation analysis in the new

millennium. Communication Monographs, 76(4): 408-420.

Hayes, A. F. 2013. Introduction to mediation, moderation, and conditional process

analysis: A regression-based approach: Guilford Press.

Heifetz, R. A. 1994. Leadership without easy answers: Harvard University Press.

Helman, C. G. 1981. Disease versus illness in general practice. Journal of the Royal

College of General Practitioners, 31(230): 548-552.

Henseler, J., Ringle, C. M., & Sarstedt, M. 2015. A new criterion for assessing

discriminant validity in variance-based structural equation modeling. Journal of

the academy of marketing science, 43(1): 115-135.

Hernandez-Trujillo, V. P., Scalchunes, C., Hernandez-Trujillo, H. S., Boyle, J., Williams,

P., Boyle, M., & Orange, J. S. 2015. Primary immunodeficiency diseases: An

opportunity in pediatrics for improving patient outcomes. Clinical Pediatrics,

54(13): 1265-1275.

HFA. 2014. Inhibitors, Vol. 2017.

224

Hibbard, J. H. 2017. Patient activation and the use of information to support informed

health decisions. Patient education and counseling, 100(1): 5-7.

Hickner, J., Thompson, P. J., Wilkinson, T., Epner, P., Shaheen, M., Pollock, A. M., Lee,

J., Duke, C. C., Jackson, B. R., & Taylor, J. R. 2014. Primary care physicians'

challenges in ordering clinical laboratory tests and interpreting results. The

Journal of the American Board of Family Medicine, 27(2): 268-274.

Hiltzik, M. 2018. Reducing healthcare costs doesn't require Bezos/Buffett/Dimon magic:

Every other country already knows how, Vol. 2018: Las Angelas Times.

Hock, D. 1999. Birth ofthe chaordic age. Los Angeles, CA: Pub Group West.

Holroyd-Leduc, J., Resin, J., Ashley, L., Barwich, D., Elliott, J., Huras, P., Légaré, F.,

Mahoney, M., Maybee, A., & McNeil, H. 2016. Giving voice to older adults

living with frailty and their family caregivers: engagement of older adults living

with frailty in research, health care decision making, and in health policy.

Research Involvement and Engagement, 2(1): 23.

Hossler, D., & Vesper, N. 1993. An exploratory study of the factors associated with

parental saving for postsecondary education. The Journal of Higher Education,

64(2): 140-165.

Hostetter, M., Klein, S., & McCarthy, D. 2017. CareMore: Improving outcomes and

controlling health care spending for high-needs patients: The Commonwealth

Fund.

Hu, L. t., & Bentler, P. M. 1999. Cutoff criteria for fit indexes in covariance structure

analysis: Conventional criteria versus new alternatives. Structural Equation

Modeling: A Multidisciplinary Journal, 6(1): 1-55.

Hunink, M. M., Weinstein, M. C., Wittenberg, E., Drummond, M. F., Pliskin, J. S.,

Wong, J. B., & Glasziou, P. P. 2014. Decision making in health and medicine:

integrating evidence and values: Cambridge University Press.

Iacobucci, D. 2010. Structural equations modeling: Fit indices, sample size, and advanced

topics. Journal of Consumer Psychology, 20(1): 90-98.

Isaacson, W. 2011. Steve Jobs: Die autorisierte Biografie des Apple-Gründers:

Verlagsgruppe Random House.

Isen, A. M. 2001. An influence of positive affect on decision making in complex

situations: Theoretical issues with practical implications. Journal of consumer

psychology, 11(2): 75-85.

225

Janssen, S. M., & Lagro-Janssen, A. L. 2012. Physician's gender, communication style,

patient preferences and patient satisfaction in gynecology and obstetrics: a

systematic review. Patient Education And Counseling, 89(2): 221-226.

Johnson, C. Y. 2017. After single payer failed, Vermont embarks on a big health care

experiment, The Washington Post.

Johnson, R. B., & Onwuegbuzie, A. J. 2004. Mixed methods research: A research

paradigm whose time has come. Educational researcher, 33(7): 14-26.

Jolles, S., Orange, J., Gardulf, A., Stein, M., Shapiro, R., Borte, M., & Berger, M. 2015.

Current treatment options with immunoglobulin G for the individualization of

care in patients with primary immunodeficiency disease. Clinical &

Experimental Immunology, 179(2): 146-160.

Jones, R. 1992. Decision-making and hospital referrals. Oxford General Practice Series,

22: 92-92.

Joseph-Williams, N., Elwyn, G., & Edwards, A. 2014. Knowledge is not power for

patients: a systematic review and thematic synthesis of patient-reported barriers

and facilitators to shared decision making. Patient education and counseling,

94(3): 291-309.

Joshi, A. Y., Iyer, V. N., Hagan, J. B., Sauver, J. L. S., & Boyce, T. G. 2009. Incidence

and temporal trends of primary immunodeficiency: A population-based cohort

study. Mayo Clinic Proceedings, 84(1): 16-22.

Kaba, R., & Sooriakumaran, P. 2007. The evolution of the doctor-patient relationship.

International Journal Of Surgery, 5(1): 57-65.

Kahneman, D. 2011. Thinking, fast and slow: Macmillan.

Kahneman, D. 2017. Less-than-Rational-Actors. In J. Christian (Ed.), Cheung Kong

Graduate School of Business. CKGSB Knowledge Magazine.

Kahneman, D., Knetsch, J. L., & Thaler, R. H. 1991. Anomalies: The endowment effect,

loss aversion, and status quo bias. The journal of economic perspectives, 5(1):

193-206.

Kaplan, S. H., Greenfield, S., Gandek, B., Rogers, W. H., & Ware, J. E. 1996.

Characteristics of physicians with participatory decision-making styles. Annals

Of Internal Medicine, 124(5): 497-504.

Keehan, S. P., Sisko, A. M., Truffer, C. J., Poisal, J. A., Cuckler, G. A., Madison, A. J.,

Lizonitz, J. M., & Smith, S. D. 2011. National health spending projections

through 2020: Economic recovery and reform drive faster spending growth.

Health Affairs, 30(8): 1594-1605.

226

Kenny, D. A. 2016. Miscellaneous Variables: Formative Variables and Second-Order

Factors.

Kiesler, D. J., & Auerbach, S. M. 2006. Optimal matches of patient preferences for

information, decision-making and interpersonal behavior: evidence, models and

interventions. Patient education and counseling, 61(3): 319-341.

Klein, J. G. 2005. Five pitfalls in decisions about diagnosis and prescribing. BMJ:

British Medical Journal, 330(7494): 781.

Kobrynski, L., Powell, R. W., & Bowen, S. 2014. Prevalence and morbidity of primary

immunodeficiency diseases, United States 2001–2007. Journal of clinical

immunology, 34(8): 954-961.

Konrad, T. R., Link, C. L., Shackelton, R. J., Marceau, L. D., von Dem Knesebeck, O.,

Siegrist, J., Arber, S., Adams, A., & McKinlay, J. B. 2010. It’s about time:

physicians’ perceptions of time constraints in primary care medical practice in

three national healthcare systems. Medical care, 48(2): 95.

Kourakos, M., Fradelos, E. C., Papathanasiou, I. V., Saridi, M., & Kafkia, T. 2017.

Communication as the Basis of Care for Patients with Chronic Diseases.

American Journal of Nursing, 7(3-1): 7-12.

Kramer, R. M., & Cook, K. S. 2004. Trust and distrust in organizations: Dilemmas and

approaches: Russell Sage Foundation.

Krosnick, J. A. 1999. Survey research. Annual review of psychology, 50(1): 537-567.

Krupat, E., Rosenkranz, S. L., Yeager, C. M., Barnard, K., Putnam, S. M., & Inui, T. S.

2000. The practice orientations of physicians and patients: the effect of doctor–

patient congruence on satisfaction. Patient Education And Counseling, 39(1):

49-59.

Lamb, C., Boland Jr, R. J., Lyytinen, K., & Wolfberg, A. 2015. Patient-driven vs.

evidence-centric decision models in the treatment of hemophilia. In W. S. o. M.

C. W. R. University (Ed.).

Lamb, C., Lyytinen, K., & Wang, Y. 2016. Does a physician’s decision making process

effect patient participation in the treatment choices of primary

immunodeficiency? In W. S. o. M. C. W. R. University (Ed.).

Larcher, V., Craig, F., Bhogal, K., Wilkinson, D., & Brierley, J. 2015. Making decisions

to limit treatment in life-limiting and life-threatening conditions in children: a

framework for practice. Archives of disease in childhood, 100(Suppl 2): s1-s23.

227

Lawrence, R. E., Rasinski, K. A., Yoon, J. D., & Curlin, F. A. 2015. Physician race and

treatment preferences for depression, anxiety, and medically unexplained

symptoms. Ethnicity & Health, 20(4): 354-364.

Lee, E., & Emanuel, E. J. 2013. Shared decision making to improve care and reduce

costs. New England Journal Of Medicine, 368(1): 6-8.

Lee, W., Joshi, A., Woolford, S., Sumner, M., Brown, M., Hadker, N., & Pashos, C.

2008. Physicians’ preferences towards coagulation factor concentrates in the

treatment of Haemophilia with inhibitors: a discrete choice experiment.

Haemophilia, 14(3): 454-465.

Légaré, F., Labrecque, M., Cauchon, M., Castel, J., Turcotte, S., & Grimshaw, J. 2012.

Training family physicians in shared decision-making to reduce the overuse of

antibiotics in acute respiratory infections: a cluster randomized trial. Canadian

Medical Association Journal, 184(13): E726-E734.

Légaré, F., Ratté, S., Gravel, K., & Graham, I. D. 2008. Barriers and facilitators to

implementing shared decision-making in clinical practice: update of a systematic

review of health professionals’ perceptions. Patient Education And Counseling,

73(3): 526-535.

Légaré, F., Stacey, D., Brière, N., Fraser, K., Desroches, S., Dumont, S., Sales, A., Puma,

C., & Aubé, D. 2013. Healthcare providers' intentions to engage in an

interprofessional approach to shared decision-making in home care programs: a

mixed methods study. Journal Of Interprofessional Care, 27(3): 214-222.

Légaré, F., Stacey, D., Turcotte, S., Cossi, M. J., Kryworuchko, J., Graham, I. D.,

Lyddiatt, A., Politi, M. C., Thomson, R., Elwyn, G., & Donner-Banzhoff, N.

2014. Interventions for improving the adoption of shared decision making by

healthcare professionals. Cochrane Database of Systematic Reviews, 9(Art. No.:

CD006732).

Légaré, F., & Thompson-Leduc, P. 2014. Twelve myths about shared decision making.

Patient education and counseling, 96(3): 281-286.

Légaré, F., & Witteman, H. O. 2013. Shared decision making: examining key elements

and barriers to adoption into routine clinical practice. Health Affairs, 32(2): 276-

284.

Leimeister, J. M., & Krcmar, H. 2005. Evaluation of a systematic design for a virtual

patient community. Journal of Computer‐Mediated Communication, 10(4): 00-

00.

Letiche, H. 2008. Making healthcare care: Managing via simple guiding principles:

IAP.

228

Lin, M.-Y., & Kressin, N. R. 2015. Race/ethnicity and Americans’ experiences with

treatment decision making. Patient Education And Counseling, 98(12): 1636-

1642.

Liras, A., & García-Trenchard, R. 2013. Treatment for hemophilia: recombinant versus

plasma-derived coagulation factors-controversy and debate forever? An ethical

medical challenge? Expert Review Of Hematology, 6(5): 489-492.

Lissack, M., & Roos, J. 1999. Words count: Viewing organizations as emerging systems

of languaging.

Loewenstein, G., & Chater, N. 2017. Putting nudges in perspective. Behavioural Public

Policy, 1(1): 26-53.

Longtin, Y., Sax, H., Leape, L. L., Sheridan, S. E., Donaldson, L., & Pittet, D. 2010.

Patient participation: current knowledge and applicability to patient safety.

Paper presented at the Mayo Clinic Proceedings.

Lovallo, D., & Sibony, O. 2010. The case for behavioral strategy. Mckinsey Quarterly,

2(1): 30-43.

Lynch, S. M. 2007. Introduction to applied Bayesian statistics and estimation for social

scientists: Springer Science & Business Media.

MacKenzie, S. B., & Podsakoff, P. M. 2012. Common method bias in marketing: causes,

mechanisms, and procedural remedies. Journal Of Retailing, 88(4): 542-555.

MacKenzie, S. B., Podsakoff, P. M., & Podsakoff, N. P. 2011. Construct measurement

and validation procedures in MIS and behavioral research: Integrating new and

existing techniques. Mis Quarterly, 35(2): 293-334.

Maly, R. C., Umezawa, Y., Ratliff, C. T., & Leake, B. 2006. Racial/ethnic group

differences in treatment decision‐making and treatment received among older

breast carcinoma patients. Cancer, 106(4): 957-965.

Mamede, S., van Gog, T., van den Berge, K., Rikers, R. M., van Saase, J. L., van

Guldener, C., & Schmidt, H. G. 2010. Effect of availability bias and reflective

reasoning on diagnostic accuracy among internal medicine residents. Jama,

304(11): 1198-1203.

Manchikanti, L., Helm, S., & Janata, J. W. 2012. Opioid epidemic in the United States.

Pain Physician, 15: ES9-ES38.

Mannucci, P. M., Mancuso, M. E., & Santagostino, E. 2012. How we choose factor VIII

to treat hemophilia. Blood, 119(18): 4108-4114.

229

Marinova, D., Kozlenkova, I. V., Cuttler, L., & Silvers, J. 2016. To Prescribe or Not to

Prescribe? Consumer Access to Life-Enhancing Products. Journal of Consumer

Research, 43(5): 806-823.

Marshall, S., Haywood, K., & Fitzpatrick, R. 2005. Patient involvement and collaboration

in shared decision making: A structured review to inform chronic disease

management: Report from the Patient-Reported Health Instruments Group to the

Department of Health.

Mason, M. 2010. Sample size and saturation in PhD studies using qualitative

interviews. Paper presented at the Forum Qualitative Sozialforschung/Forum:

Qualitative Social Research.

Matsunaga, M. 2010. How to Factor-Analyze Your Data Right: Do’s, Don’ts, and How-

To’s. International Journal Of Psychological Research, 3(1): 97-110.

Mazzi, M. A., Rimondini, M., Deveugele, M., Zimmermann, C., Deledda, G., & Bensing,

J. 2016. Does gender matter in doctor–patient communication during standard

gynaecological consultations? An analysis using mixed methods. Communication

& Medicine, 11(3): 285-298.

McCormack, L., Thomas, V., Lewis, M. A., & Rudd, R. 2017. Improving low health

literacy and patient engagement: a social ecological approach. Patient education

and counseling, 100(1): 8-13.

McDonald, K., Sundaram, V., Bravata, D., Lewis, R., Lin, N., Kraft, S., McKinnon, M.,

Paguntalan, H., & Owens, D. 2007. Closing the quality gap: A critical analysis

of quality improvement strategies. Rockville, MD: Stanford-UCSF Evidence-

based Practice Center. Agency for Healthcare Research and Quality.

McGlynn, E. A., Asch, S. M., Adams, J., Keesey, J., Hicks, J., DeCristofaro, A., & Kerr,

E. A. 2003. The quality of health care delivered to adults in the United States.

New England journal of medicine, 348(26): 2635-2645.

McGuire, A. L., McCullough, L. B., Weller, S. C., & Whitney, S. N. 2005. Missed

expectations?: physicians’ views of patients’ participation in medical decision-

making. Medical Care, 43(5): 466-470.

McMurray, R., Pullen, A., & Rhodes, C. 2011. Ethical subjectivity and politics in

organizations: A case of health care tendering. Organization, 18(4): 541-561.

McNeil, B. J., Pauker, S. G., Sox Jr, H. C., & Tversky, A. 1982. On the elicitation of

preferences for alternative therapies. New England journal of medicine, 306(21):

1259-1262.

McWilliams, J. M. 2016. Cost containment and the tale of care coordination. New

England Journal of Medicine, 375(23): 2218-2220.

230

Mead, E. L., Doorenbos, A. Z., Javid, S. H., Haozous, E. A., Alvord, L. A., Flum, D. R.,

& Morris, A. M. 2013. Shared decision-making for cancer care among racial and

ethnic minorities: a systematic review. American journal of public health,

103(12): e15-e29.

Medicare Payment Advisory Commission. 2010. Aligning incentives in Medicare. Paper

presented at the Report to the Congress: MedPAC. June 2010, Washington, DC.

MedPAC. 2010. Report to the Congress: Aligning incentives in medicine.

Menzin, J., Sussman, M., Munsell, M., & Zbrozek, A. 2014. Economic impact of

infections among patients with primary immunodeficiency disease receiving IVIG

therapy. ClinicoEconomics and Outcomes Research: CEOR, 6: 297302.

Meunier, V., Laamen, S., Harison, M., Lewis, D. R., Muraoka, S., & Walker, A. L. 2015.

Hemophilia: reshuffling the cards. Global Pharmaceuticals.

Minno, M., Minno, G., Capua, M., Cerbone, A., & Coppola, A. 2010. Cost of care of

haemophilia with inhibitors. Haemophilia, 16(1): e190-e201.

Mirowsky, J. 2017. Education, social status, and health: Routledge.

Modell, V., Gee, B., Lewis, D. B., Orange, J. S., Roifman, C. M., Routes, J. M.,

Sorensen, R. U., Notarangelo, L. D., & Modell, F. 2011. Global study of primary

immunodeficiency diseases (PI)—diagnosis, treatment, and economic impact: an

updated report from the Jeffrey Modell Foundation. Immunologic Research,

51(1): 61-70.

Moens, L. N., Falk-Sörqvist, E., Asplund, A. C., Bernatowska, E., Smith, C. E., &

Nilsson, M. 2014. Diagnostics of primary immunodeficiency diseases: a

sequencing capture approach. PLoS One, 9(12): e114901.

Mondak, J. J., Hibbing, M. V., Canache, D., Seligson, M. A., & Anderson, M. R. 2010.

Personality and civic engagement: An integrative framework for the study of trait

effects on political behavior. American Political Science Review, 104(1): 85-110.

Montgomery, K. 2005. How doctors think: Clinical judgment and the practice of

medicine: Oxford University Press.

Montori, V. M., Gafni, A., & Charles, C. 2006. A shared treatment decision‐making

approach between patients with chronic conditions and their clinicians: the case of

diabetes. Health Expectations, 9(1): 25-36.

Morse, J. M. 1994. Designing funded qualitative research. In N. K. Denzin, & Y. S.

Lincoln (Eds.), Handbook of qualitative research 220-235. Thousand Oaks, CA:

Sage Publications.

231

Mulley, A. G., Trimble, C., & Elwyn, G. 2012. Stop the silent misdiagnosis: Patients’

preferences matter. BMJ, 345: 23.

Murray, E., Pollack, L., White, M., & Lo, B. 2007. Clinical decision-making: physicians'

preferences and experiences. BMC Family Practice, 8(1): 1.

Myers, M. D. 2013. Qualitative research in business and management: Sage.

Nannenga, M. R., Montori, V. M., Weymiller, A. J., Smith, S. A., Christianson, T. J.,

Bryant, S. C., Gafni, A., Charles, C., Mullan, R. J., & Jones, L. A. 2009. A

treatment decision aid may increase patient trust in the diabetes specialist. The

Statin Choice randomized trial. Health Expectations, 12(1): 38-44.

Nannestad, P. 2008. What have we learned about generalized trust, if anything? Annual

Review of Political Science, 11: 413-436.

Nguyen, H. 2011. The principal-agent problems in health care: evidence from prescribing

patterns of private providers in Vietnam. Health Policy and Planning,

26(suppl_1): i53-i62.

Nicolay, U., Haag, S., Eichmann, F., Herget, S., Spruck, D., & Gardulf, A. 2005.

Measuring treatment satisfaction in patients with primary immunodeficiency

diseases receiving lifelong immunoglobulin replacement therapy. Quality Of Life

Research, 14(7): 1683-1691.

Noonan, V., Lyddiatt, A., Ware, P., Jaglal, S., Riopelle, R., Bingham, C., Figueiredo, S.,

Sawatzky, R., Santana, M., & Bartlett, S. 2017. Montreal accord on patient-

reported outcomes use series–paper 3: Patient reported outcomes (PRO) can

facilitate shared decision-making and guide self-management. Journal of Clinical

Epidemiology, 89: 125-135.

NORC. 2014. Demonstrating the effectiveness of patient feedback in improving the

accuracy.

North, D. W. 1968. A tutorial introduction to decision theory. Systems Science And

Cybernetics, Ieee Transactions On, 4(3): 200-210.

O'Connor, A. M., Bruine de Bruin, W., Cassels, A., Driedger, S., Greenberg, J., Gully, P.,

Kreps, G., Lemyre, L., Lofstedt, R., & North, D. 2015. Health Product Risk

Communication: Is the message getting through?: Council of Canadian

Academies.

O'Connor, A. M., Llewellyn-Thomas, H. A., & Flood, A. B. 2004. Modifying

unwarranted variations in health care: shared decision making using patient

decision aids. Health Affairs: VAR63.

232

O'Hare, A. M., Rodriguez, R. A., & Bowling, C. B. 2016. Caring for patients with kidney

disease: shifting the paradigm from evidence-based medicine to patient-centered

care. Nephrology Dialysis Transplantation, 31(3): 368-375.

O’Connor, A. M., Wennberg, J. E., Legare, F., Llewellyn-Thomas, H. A., Moulton, B.

W., Sepucha, K. R., Sodano, A. G., & King, J. S. 2007. Toward the ‘tipping

point’: decision aids and informed patient choice. Health Affairs, 26(3): 716-725.

O’Hara, J., Hughes, D., Camp, C., Burke, T., Carroll, L., & Diego, D.-A. G. 2017. The

cost of severe haemophilia in Europe: the CHESS study. Orphanet journal of

rare diseases, 12(1): 106.

O’Reilly, K., Paper, D., & Marx, S. 2012. Demystifying grounded theory for business

research. Organizational Research Methods, 15(2): 247-262.

OECD. 2015. Health at a Glance 2015, OECD Indicators, Vol. 15. OECD Publishing,

Paris.

Ofri, D. 2017. What Patients Say, what Doctors Hear: What Doctors Say, what Patients

Hear: Beacon Press.

Oshima Lee, E., & Emanuel, E. J. 2013. Shared decision making to improve care and

reduce costs. New England Journal Of Medicine, 368(1): 6-8.

Paradies, Y., Truong, M., & Priest, N. 2014. A systematic review of the extent and

measurement of healthcare provider racism. Journal Of General Internal

Medicine, 29(2): 364-387.

Parekh, A. K., & Barton, M. B. 2010. The challenge of multiple comorbidity for the US

health care system. Jama, 303(13): 1303-1304.

Parekh, A. K., Goodman, R. A., Gordon, C., Koh, H. K., & Conditions, H. I. W. o. M. C.

2011. Managing multiple chronic conditions: a strategic framework for improving

health outcomes and quality of life. Public health reports, 126(4): 460-471.

Patel, V. L., Kaufman, D. R., & Arocha, J. F. 2002. Emerging paradigms of cognition in

medical decision-making. Journal Of Biomedical Informatics, 35(1): 52-75.

Patel, V. L., Yoskowitz, N. A., & Arocha, J. F. 2009. Towards effective evaluation and

reform in medical education: a cognitive and learning sciences perspective.

Advances in Health Sciences Education, 14(5): 791-812.

Paternotte, E., van Dulmen, S., van der Lee, N., Scherpbier, A. J., & Scheele, F. 2015.

Factors influencing intercultural doctor–patient communication: A realist review.

Patient education and counseling, 98(4): 420-445.

233

Pear, R. 2017. Minnesota finds a way to slow soaring health premiums, The New York

Times.

Peck, B. M. 2011. Age-related differences in doctor-patient interaction and patient

satisfaction. Current Gerontology and Geriatrics Research, 2011(Article ID

137492).

Peikes, D., Chen, A., Schore, J., & Brown, R. 2009. Effects of care coordination on

hospitalization, quality of care, and health care expenditures among Medicare

beneficiaries: 15 randomized trials. Jama, 301(6): 603-618.

Pelaccia, T., Tardif, J., Triby, E., & Charlin, B. 2011. An analysis of clinical reasoning

through a recent and comprehensive approach: The dual-process theory. Medical

Education Online, 16(1).

Pereira, C. M., Amaral, C. F., Ribeiro, M. M., Paro, H. B., Pinto, R. M., Reis, L. E.,

Silva, C. H., & Krupat, E. 2013. Cross-cultural validation of the Patient–

Practitioner Orientation Scale (PPOS). Patient Education And Counseling,

91(1): 37-43.

Petrič, G., Atanasova, S., & Kamin, T. 2017. Impact of social processes in online health

communities on patient empowerment in relationship with the physician:

Emergence of functional and dysfunctional empowerment. Journal of Medical

Internet Research, 19(3): e74.

Petroshius, S. M., Titus, P. A., & Hatch, K. J. 1995. Physician attitudes toward

pharmaceutical drug advertising. Journal Of Advertising Research, 35(6): 41-52.

Peyvandi, F., Rosendaal, F. R., O'Mahony, B., & Mannucci, P. M. 2014. Reply to: The

importance and challenge of pediatric trials of hemophilia drugs. Nature

Medicine, 20(5): 466-466.

Physicians Practice. 2017. 2017 Great American Physician Survey, Vol. 2017: Physicians

Practice.

Picard, C., Al-Herz, W., Bousfiha, A., Casanova, J.-L., Chatila, T., Conley, M. E.,

Cunningham-Rundles, C., Etzioni, A., Holland, S. M., & Klein, C. 2015. Primary

immunodeficiency diseases: an update on the classification from the International

Union of Immunological Societies Expert Committee for Primary

Immunodeficiency 2015. Journal of clinical immunology, 35(8): 696-726.

Pipe, S. W. 2012. The hope and reality of long‐acting hemophilia products. American

Journal Of Hematology, 87(S1): S33-S39.

Podsakoff, N. P., Podsakoff, P. M., MacKenzie, S. B., & Klinger, R. L. 2013. Are we

really measuring what we say we're measuring? Using video techniques to

234

supplement traditional construct validation procedures. Journal of Applied

Psychology, 98(1): 99.

Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. 2003. Common

Method Biases in Behavioral Research: A Critical Review of the Literature and

Recommended Remedies. Journal Of Applied Psychology, 88(5): 879-903.

Preacher, K. J., Rucker, D. D., & Hayes, A. F. 2007. Addressing moderated mediation

hypotheses: Theory, methods, and prescriptions. Multivariate Behavioral

Research, 42(1): 185-227.

Putnam, R. D. 1995. Bowling alone: America's declining social capital. Journal of

democracy, 6(1): 65-78.

Reio, T. G., & Shuck, B. 2014. Exploratory factor analysis implications for theory,

research, and practice. Advances In Developing Human Resources, 17(1): 12-25.

Resnick, E. S., Bhatt, P., Sidi, P., & Cunningham-Rundles, C. 2013. Examining the use

of ICD-9 diagnosis codes for primary immune deficiency diseases in New York

State. Journal of clinical immunology, 33(1): 40-48.

Richardson, W. C., Berwick, D. M., Bisgard, J., Bristow, L., Buck, C., & Cassel, C.

2001. Crossing the quality chasm: A new health system for the 21st century.

Washington, DC: Institute of Medicine, National Academy Press

Rosenthal, E. 2017. An American sickness: How healthcare became big business and

how you can take it back: New York, NY: Penguin Press.

Roter, D. 2017. Bridging the gap between patient-centered care and evidence-based

medicine: How and why communication matters: Kimmel Cancer Center,

Presentations and Grand Rounds. Paper 42.

http://jdc.jefferson.edu/kimmelgrandrounds/42.

Roter, D. L., Hall, J. A., & Aoki, Y. 2002. Physician gender effects in medical

communication: a meta-analytic review. Jama, 288(6): 756-764.

Rothenfluh, F., & Schulz, P. J. 2017. Physician rating websites: What aspects are

important to identify a good doctor, and are patients capable of assessing them? A

mixed-methods approach including physicians’ and health care consumers’

perspectives. Journal of Medical Internet Research, 19(5): e127.

Rothman, D. J. 2017. Strangers at the bedside: A history of how law and bioethics

transformed medical decision making: Routledge.

Rothstein, M. A. 2014. Autonomy and paternalism in health policy: currents in

contemporary bioethics. The Journal of Law, Medicine & Ethics, 42(4): 590-

594.

235

Rousseau, D. M. 2012. The Oxford handbook of evidence-based management: Oxford

University Press.

Sackett, D. L., Rosenberg, W. M., Gray, J. M., Haynes, R. B., & Richardson, W. S. 1996.

Evidence based medicine: what it is and what it isn't: British Medical Journal

Publishing Group.

Samaan, K., Levasseur, M.-C., Decaluwe, H., St-Cyr, C., Chapdelaine, H., Des Roches,

A., & Haddad, E. 2014. SCIg vs IVIg: Let's Give Patients the Choice! Journal Of

Clinical Immunology, 34(6): 611.

Sapolsky, R. M. 2017. Behave: The Biology of Humans at Our Best and Worst:

Penguin.

Say, R. E., & Thomson, R. 2003. The importance of patient preferences in treatment

decisions—challenges for doctors. BMJ: British Medical Journal, 327(7414):

542.

Scalone, L., Mantovani, L., Borghetti, F., Von Mackensen, S., & Gringeri, A. 2009.

Patients’, physicians’, and pharmacists’ preferences towards coagulation factor

concentrates to treat haemophilia with inhibitors: results from the COHIBA

Study. Haemophilia, 15(2): 473-486.

Schattner, A., & Simon, S. R. 2017. Diminishing patient face time in residencies and

patient-centered care: Elsevier.

Schiavone, G., De Anna, G., Mameli, M., Rebba, V., & Boniolo, G. 2014. Libertarian

paternalism and health care policy: a deliberative proposal. Medicine, Health

Care and Philosophy, 17(1): 103-113.

Schiff, G. D. 2008. Minimizing diagnostic error: The importance of follow-up and

feedback: Elsevier.

Schneider, E. C., Sarnak, D. O., Squires, D., Shah, A., & Doty, M. M. 2017. Mirror,

mirror 2017: The Commonwealth Fund, July 2017. Available from

http://www.commonwealthfund.org/publications/fund-reports/2017/jul/mirror-

mirror-international-comparisons-2017.

Schnell, M., & Currie, J. 2017. Addressing the opioid epidemic: Is there a role for

physician education?: National Bureau of Economic Research, NBER Working

Paper No. 23645.

Schoen, C., Guterman, S., Shih, A., Lau, J., Kasimow, S., Gauthier, A., & Davis, K.

2007. Bending the curve: options for achieving savings and improving value in

US health spending: New York: The Commonwealth Fund.

236

Schoenthaler, A., Montague, E., Baier Manwell, L., Brown, R., Schwartz, M. D., &

Linzer, M. 2014. Patient–physician racial/ethnic concordance and blood pressure

control: the role of trust and medication adherence. Ethnicity & health, 19(5):

565-578.

Schon, D. 1983. The reflective practitioner. New York, NY: Basic Books.

Schulman, K. A., Berlin, J. A., Harless, W., Kerner, J. F., Sistrunk, S., Gersh, B. J., Dube,

R., Taleghani, C. K., Burke, J. E., & Williams, S. 1999. The effect of race and sex

on physicians' recommendations for cardiac catheterization. New England

Journal of Medicine, 340(8): 618-626.

Schwartz, R. A. 2013. Factor VIII Treatment & Management.

Scott, J. 2000. Rational choice theory. In G. Browning, A. Halcli, & F. Webster (Eds.),

Understanding contemporary society: Theories of the present: 126-138: Sage

Publications.

Seeborg, F. O., Seay, R., Boyle, M., Boyle, J., Scalchunes, C., & Orange, J. S. 2015.

Perceived Health in Patients with Primary Immune Deficiency. Journal Of

Clinical Immunology, 35(7): 638-650.

Shaffer, J. A., DeGeest, D., & Li, A. 2016. Tackling the problem of construct

proliferation: A guide to assessing the discriminant validity of conceptually

related constructs. Organizational Research Methods, 19(1): 80-110.

Shapiro, R. S. 2013. Supplement: Frontiers in immunoglobulin therapy of primary

immunodeficiency disease. Journal of Clinical Immunology, 37(2): 187.

Shapiro, R. S., Wasserman, R. L., Bonagura, V., & Gupta, S. 2014. Emerging paradigm

of primary immunodeficiency disease: Individualizing immunoglobulin dose and

delivery to enhance outcomes. Journal of Clinical Immunology, 37(2): 190-196.

Siegel, A. L., & Ruh, R. A. 1973. Job involvement, participation in decision making,

personal background, and job behavior. Organizational Behavior And Human

Performance, 9: 318-327.

Silverman, D. 2011. Interpreting qualitative data: A guide to the principles of

qualitative research: SAGE Publications Limited.

Silverman, D. 2015. Interpreting qualitative data: Sage.

Silvers, J., Marinova, D., Mercer, M. B., Connors, A., & Cuttler, L. 2010. A national

study of physician recommendations to initiate and discontinue growth hormone

for short stature. Pediatrics, 126(3): 468-476.

Simon, H. A. 1996. The sciences of the artificial: MIT press.

237

Singh, J., Cuttler, L., & Silvers, J. 2004. Toward understanding consumers' role in

medical decisions for emerging treatments: Issues, framework and hypotheses.

Journal Of Business Research, 57(9): 1054-1065.

Snipes, S. A., Sellers, S. L., Tafawa, A. O., Cooper, L. A., Fields, J. C., & Bonham, V. L.

2011. Is race medically relevant? A qualitative study of physicians' attitudes about

the role of race in treatment decision-making. BMC Health Services Research,

11(1): 183.

Sox, H. C., Higgins, M. C., & Owens, D. K. 2013. Medical Decision Making (2 ed.):

Wiley-Blackwell.

Sposito, V., Hand, M., & Skarpness, B. 1983. On the efficiency of using the sample

kurtosis in selecting optimal lpestimators. Communications In Statistics-

Simulation And Computation, 12(3): 265-272.

Squires, D., & Anderson, C. 2015. US health care from a global perspective: spending,

use of services, prices, and health in 13 countries. The Commonwealth Fund, 15:

1-16.

Srivastava, A., Brewer, A., Mauser‐Bunschoten, E., Key, N. S., Kitchen, S., Llinas, A.,

Ludlam, C., Mahlangu, J., Mulder, K., & Poon, M. 2013. Guidelines for the

management of hemophilia. Haemophilia, 19(1): e1-e47.

Stacey, D., Bennett, C. L., Barry, M. J., Col, N. F., Eden, K. B., Holmes-Rovner, M.,

Llewellyn-Thomas, H., Lyddiatt, A., Légaré, F., & Thomson, R. 2011. Decision

aids for people facing health treatment or screening decisions. Cochrane

Database Syst Rev, 10(10).

Stacey, D., Légaré, F., Col, N. F., Bennett, C. L., Barry, M. J., Eden, K. B., Holmes‐

Rovner, M., Llewellyn‐Thomas, H., Lyddiatt, A., & Thomson, R. 2014. Decision

aids for people facing health treatment or screening decisions. The Cochrane

Library.

Stanovich, K. E. 2011. Rationality and the reflective mind: Oxford University Press.

Stanovich, K. E., & West, R. F. 2000. Individual differences in reasoning: Implications

for the rationality debate? Behavioral and brain sciences, 23(5): 645-665.

Stanovich, K. E., & West, R. F. 2008. On the relative independence of thinking biases

and cognitive ability. Journal of personality and social psychology, 94(4): 672.

Stark, M., & Fins, J. J. 2014. The ethical imperative to think about thinking. Cambridge

Quarterly of Healthcare Ethics, 23(04): 386-396.

238

Stevenson, F. A., Barry, C. A., Britten, N., Barber, N., & Bradley, C. P. 2000. Doctor–

patient communication about drugs: the evidence for shared decision making.

Social Science & Medicine, 50(6): 829-840.

Strauss, A., & Corbin, J. 1998. Basics of qualitative research: Procedures and

techniques for developing grounded theory. Thousand Oaks, CA: Sage

Publications.

Street, A. 2012. Developing models of haemophilia care. Haemophilia, 18(s4): 89-93.

Street Jr, R. L., Gordon, H. S., Ward, M. M., Krupat, E., & Kravitz, R. L. 2005. Patient

participation in medical consultations: why some patients are more involved than

others. Medical Care, 43(10): 960-969.

Sturgis, P., & Smith, P. 2010. Assessing the validity of generalized trust questions: what

kind of trust are we measuring? International journal of public opinion

research, 22(1): 74-92.

Sturmberg, J. P., & Martin, C. 2013. Handbook of systems and complexity in health:

Springer Science & Business Media.

Sugawara, Y., Narimatsu, H., Hozawa, A., Shao, L., Otani, K., & Fukao, A. 2012. Cancer

patients on Twitter: a novel patient community on social media. BMC Research

Notes, 5(1): 699.

Sunstein, C., & Thaler, R. 2008. Why nudge?: The politics of libertarian paternalism:

Yale University Press.

Tanio, C. 2014. Developing physician culture in new risk models, The Health Care Blog.

Tapscott, D. 2010. Macrowikinomics: New solutions for a connected planet: Atlantic

Books Ltd.

Tashakkori, A., & Teddlie, C. 2008. Quality of inferences in mixed methods research:

Calling for an integrative framework. In M. M. Bergman (Ed.), Advances in

mixed methods research: 101-119: Sage Publications.

Teddlie, C., & Tashakkori, A. 2003. Major issues and controveries in the use of mixed

methods in the social and behvioral sciences. In A. Tashakkori, & C. Teddlie

(Eds.), Handbook of Mixed Methods in Social & Behavioral Research: 3-50:

Sage Publications.

Teddlie, C., & Tashakkori, A. 2009. Foundations of mixed methods research:

Integrating quantitative and qualitative approaches in the social and behavioral

sciences: Sage Publications Inc.

239

Thaler, R., & Sunstein, C. R. 2003. Libertarian paternalism. American Economic

Review, 93(2): 175-179.

The Joint Commission. 2007. “What did the doctor say?:” Improving health literacy to

protect patient safety Available from

https://www.jointcommission.org/assets/1/18/improving_health_literacy.pdf.

Thom, D. H., Wong, S. T., Guzman, D., Wu, A., Penko, J., Miaskowski, C., & Kushel,

M. 2011. Physician trust in the patient: development and validation of a new

measure. The Annals Of Family Medicine, 9(2): 148-154.

Thomas, D. M., & Watson, R. T. 2002. Q-sorting and MIS research: A primer.

Communications Of The Association For Information Systems, 8(1): 9.

Thygeson, M., Morrissey, L., & Ulstad, V. 2010. Adaptive leadership and the practice of

medicine: a complexity‐based approach to reframing the doctor–patient

relationship. Journal of Evaluation in Clinical Practice, 16(5): 1009-1015.

Tiffen, J., Corbridge, S. J., & Slimmer, L. 2014. Enhancing clinical decision making:

development of a contiguous definition and conceptual framework. Journal of

Professional Nursing, 30(5): 399-405.

Timmermans, S., & Berg, M. 2010. The gold standard: The challenge of evidence-based

medicine and standardization in health care: Temple University Press.

Topol, E. J. 2015. The patient will see you now: the future of medicine is in your hands:

Tantor Media.

Toro, R. 2012. Leading Causes of Death in the US: 1900 - Present (Infographic),

LiveScience, Vol. 2018: LiveScience.

Traylor, A. H., Schmittdiel, J. A., Uratsu, C. S., Mangione, C. M., & Subramanian, U.

2010. The Predictors of Patient–Physician Race and Ethnic Concordance: A

Medical Facility Fixed‐Effects Approach. Health Services Research, 45(3): 792-

805.

Truven Health MarketScan® Research Databases. 2010. Commercial claims and

encounters medicare supplemental.

Tversky, A., & Kahneman, D. 1973. Availability: A heuristic for judging frequency and

probability. Cognitive psychology, 5(2): 207-232.

Tversky, A., & Kahneman, D. 1986a. Judgment under uncertainty: Heuristics and biases.

Judgment And Decision Making: An Interdisciplinary Reader: 38-55.

Tversky, A., & Kahneman, D. 1986b. Rational choice and the framing of decisions.

Journal of Business, 59(4): S251-S278.

240

Tversky, A., & Kahneman, D. 1992. Advances in prospect theory: Cumulative

representation of uncertainty. Journal of Risk and uncertainty, 5(4): 297-323.

U.S. Department of Health Human Services. 2010. Multiple chronic conditions—a

strategic framework: Optimum health and quality of life for individuals with

multiple chronic conditions. Washington, DC: US Department of Health and

Human Services.

UKHCDO. 2012. UKHCDO annual report 2012 & bleeding disorder statistics for the

financial year 2011/2012.

UKHCDO. 2014. Bleeding Disorder Statistics for April 2013 to March 2014: UKHCDO.

Van de Ven, A. H. 2007. Engaged scholarship: A guide for organizational and social

research: Oxford University Press on Demand.

Venkatesh, V., Brown, S. A., & Bala, H. 2013. Bridging the qualitative-quantitative

divide: Guidelines for conducting mixed methods research in information

systems. Mis Quarterly, 37(1): 21-54.

Veroff, D., Marr, A., & Wennberg, D. E. 2013. Enhanced support for shared decision

making reduced costs of care for patients with preference-sensitive conditions.

Health Affairs, 32(2): 285-293.

Vohs, K. D. 2015. Money priming can change people’s thoughts, feelings, motivations,

and behaviors: An update on 10 years of experiments. Journal of Experimental

Psychology: General, 144(4): e86.

Wagner, E. H., Austin, B. T., Davis, C., Hindmarsh, M., Schaefer, J., & Bonomi, A.

2001. Improving chronic illness care: translating evidence into action. Health

affairs, 20(6): 64-78.

Wald, H. S. 2015. Professional identity (trans) formation in medical education: reflection,

relationship, resilience. Academic Medicine, 90(6): 701-706.

Walls, A. N. 2014. The “when” of libertarian paternalism, Ethical Issues in Health Care:

Emory.

Weick, K. E., Sutcliffe, K. M., & Obstfeld, D. 2008. Organizing for high reliability:

Processes of collective mindfulness. Crisis management, 3(1): 81-123.

Whitson, H. E., & Boyd, C. M. 2016. Managing multiple comorbidities, UpToDate:

UpToDate, Inc. .

WHO. 2000. The world health report 2000: health systems: improving performance: