This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Exploring Patients' Perception of Osteoporosis Following a Fragility Fracture: Results of a Literature Review and
Analysis of a Provincial Database
by
Rebeka Sujic
A thesis submitted in conformity with the requirements for the degree of Master of Science
Department of Health Policy, Management and Evaluation University of Toronto
Chapter 2 A Review of Patient-centered Post-fracture Osteoporosis Interventions in the Context of Health Behaviour Change Theories .......................................................................
Chapter 3 Critical Appraisal of the Ontario Osteoporosis Strategy’s Fracture Clinic Screening Program Database as a Source of Data for the Thesis ............................................
3.2 Background and description of the database ....................................................................
3.3 Critique of the database: establishing its suitability to the research question...................
3.3.1 Purpose and content of the database, sampling, comparison of the sample to the target population and data collection methods............................................
3.3.2 Database structure, content of each observation, duplicates and coding.................
3.3. 3 Manual inspection of data for data flaws................................................................
Chapter 4 Factors Associated with Patients Making the Link Between Their Fragility Fracture and Osteoporosis........................................................................................................
Papadimitropoulos, 2001). Mortality associated with hip fractures is particularly high:
approximately 20% of women and 40% of men die within one year of fracturing their hip
(Papaioannou et al., 2000; Wiktorowicz et al., 2001; Bierman AS, 2010).
Health care costs of OP and related fractures are high both in Canada and internationally. In
Canada, the estimated health care costs of treating OP and fractures are expected to rise from
estimated $1.3 billion in 1993 to $10-$20 billion annually (Lorrain et al., 2003; Goeree &
O'Brien B, 1996). Despite literature suggesting a decline in the incidence of hip fractures (Jaglal
et al., 2005), the burden will still be considerable. In Canada, the annual economic burden of hip
fractures alone is projected to increase to $2.4 billion by 2041 (Wiktorowicz et al., 2001).
European estimates from 2000 show that healthcare costs for 3.8 million osteoporotic fractures
totalled 31.7 billion Euros (Kanis & Johnell, 2005). Based on fracture probability and changing
demographic factors in Europe, this figure is expected to increase to 76.7 billion Euros in 2050
(Kanis & Johnell, 2005).
1.3 The importance of osteoporosis management after a fragility fracture
The importance of assessing fragility fracture patients for OP has been emphasized in both
previous research (Astrand, Thorngren, & Tagil, 2006; Lindsay et al., 2001; Center, Bliuc,
Nguyen, & Eisman, 2007; Astrand et al., 2006) and guidelines for OP management (Brown et
al., 2006; Brown & Josse, 2002; Kanis et al., 2008; DH/SC, 2009; National Osteoporosis
Foundation, 2010; van Staa et al., 2001). Previous research found that screening fragility fracture
patients is an effective way of identifying those with a low bone mineral density (BMD) who
should be assessed for OP: one study found that 87% of fragility fracture patients in orthopaedic
environment had BMD that was at least one SD below the mean (Astrand et al., 2006). A large
3
OP epidemiological study of community-dwelling men and women in Australia found that about
50% of fragility fracture patients experienced another fracture in the 10 years following the
initial fracture (Center et al., 2007). Because of the large number of patients who re-fractured,
the study concluded that even one fragility fracture is a clear indication for therapy aimed at
prevention of future fractures. An international study analysing data from 373 study centers in
North America, Europe, Australia, and New Zealand found that 20% of post-menopausal women
with vertebral fractures sustained another fracture within a year of the first fracture (Lindsay et
al., 2001). These findings suggest that urgent action needs to be taken to prevent future fractures
particularly because they may occur a lot sooner than expected.
In addition to previous research, both Canadian and international guidelines emphasize the need
for OP management and prevention of future fractures in fragility fracture patients (Brown &
Josse, 2002; Brown et al., 2006; 1998; DH/SC, 2009; Guideline Development Group for the
Royal College of Physicians, 2010; National Osteoporosis Foundation, 2010; NIH Consensus
Development Conference on Osteoporosis Prevention, 2000; U.S.Department of Health and
Human Services, 2004; National Institute for Health and Clinical Excellence., 2008). According
to the Canadian guidelines, a previous fragility fracture is one of the four most significant risk
factors for future fractures and a major risk factor for identifying who should be assessed for OP
(Brown & Josse, 2002; Brown et al., 2006). 2008 US guidelines recommend testing in all adults
with fragility fractures and considering pharmacotherapy for those over the age of 50 who have
sustained a hip or vertebral fracture and those with a low BMD who have a high probability of
future fractures (National Institute for Health and Clinical Excellence., 2008). 2002 UK Royal
College of Physicians’ guidelines recommend the that bone protective therapy be considered for
all fragility fracture patients over the age of 64 while the National Institute for Health and
Clinical Excellence recommends that female fragility fracture patients over the age of 75 should
receive treatment even without undergoing a diagnostic test (National Institute for Health and
Clinical Excellence, 2008).
1.4 Current post-fracture osteoporosis management: gap in care Effective treatment options for OP are available. Medication can reduce fracture risk by 30- 50%
among those who have OP (or a previous vertebral fracture) (MacLean et al., 2008; Cranney et
4
al., 2002). Bisphosphonates, a class of drugs which inhibits bone resorption and therefore
prevents the loss of bone mass, have been shown to decrease the risk of fractures: the relative
risk of fracture for women using bisphosphonates (vs. placebo) is 0.47 for hip, 0.52 for vertebral
and 0.70 for other fractures (Black & Cooper, 2000). Both Canadian and international guidelines
support the use of such pharmacological agents in reducing the risk of future fractures (Kanis et
al., 2008; Brown & Josse, 2002; Brown et al., 2006; Lewiecki & Watts, 2009). Canadian Clinical
Practice Guidelines, for example, advise that bisphosphonates be used as treatment to decrease
vertebral, non-vertebral and hip fractures and recommend Calcium and vitamin D as essential for
prevention of OP and as an adjunct to the main pharmacotherapy for OP treatment (Brown &
Josse, 2002; Brown et al., 2006).
Despite the existence of effective treatment options and clinical practice guidelines
recommending that fragility fracture patients be assessed (and treated if appropriate) for OP, the
majority of fragility fracture patients do not receive appropriate OP management. A systematic
review examining OP management of fragility fracture patients in the US, Canada, Europe,
Australia and New Zealand found that these patients were not commonly tested and treated for
OP: 1- 32% of patients had diagnostic testing (BMD test) done while pharmacotherapy was
reported in 1% to 65% of patients (Giangregorio, Papaioannou, Cranney, Zytaruk, & Adachi,
2006). Another systematic review of diagnostic and treatment patterns of OP after a fragility
fracture similarly found that less than 32% of patients underwent BMD testing as part of OP
management, even though 42% to 100% of tested patients ended up diagnosed with OP (Elliot-
Gibson, Bogoch, Jamal, & Beaton, 2004). This review also showed that only 0.5–38% of patients
who had a BMD test were prescribed bisphosphonates, while Calcium and Vitamin D were used
in only 8–62% of cases (median 18%). A large UK audit in 172 trusts confirmed the low rate of
post-fracture testing and treatment: despite a number of government-led initiatives that aim at
improving the care of fragility fracture patients, post-fracture assessment and treatment rates
remained low with only 20-35% of older female fragility fracture patients assessed for OP and
only 19% of those with non-hip fractures prescribed pharmacotherapy (Clinical Effectiveness
and Evaluation Unit.London, 2007).
5
1.5 Barriers to more effective osteoporosis management Barriers to more effective OP management are associated with both the healthcare environment
and the patient. Barriers associated with the healthcare environment include a lack of OP
knowledge among physicians (such as belief that OP is a normal part of aging with a low
mortality rate), existence of competing conditions that take priority over discussing OP during a
medical visit / an insufficient amount of time that physicians may have to address OP, a lack of a
consensus on which health professional is responsible for OP management, a lack of access to
diagnostic testing as well as the complexity in coordinating patient identification, testing and
Levasseur et al., 2007), as well as creation of complex programs where a dedicated coordinator
facilitates diagnostic testing and coordinates post-fracture OP care (Bogoch et al., 2006;
Majumdar et al., 2007; Levasseur et al., 2007).
Even though addressing and removing healthcare-level barriers is an important step in improving
OP management, it is ultimately the patient who needs to implement the recommendations and
adhere to them. Without the commitment of the patient, interventions are not likely to be
successful: even in cases where specific interventions to improve OP management were
implemented, up to 79% of patients with OP or osteopenia did not initiate treatment (Bliuc et al.,
2006; Chevalley et al., 2002; Majumdar et al., 2004). Of those who initiated treatment, 74%
ceased to take their medication within one year (Edwards et al., 2005). A study involving
patients with low trauma distal forearm fractures revealed that patients may remain resistant to
OP diagnosis and treatment despite the increased OP interventions involving both orthopaedists
and primary care physicians (Cuddihy et al., 2004). The authors of this study suggested that
additional efforts from health care professionals are just one part of the puzzle and that patient-
level barriers to OP treatment need to be examined more closely.
6
1.6 Patient-level barriers to improving osteoporosis management Patient-level barriers to more effective OP management that have been identified include a lack
of perceived susceptibility to OP and future fractures (Meadows, Mrkonjic, & Lagendyk, 2005;
Giangregorio et al., 2008; Burgener et al., 2005), not making the link between OP and fractures
(Meadows 2005, Giangregorio 2008), a lack of readiness to accept OP therapy (Mauck et al.,
2002) and perception that treatment is ineffective or that no treatment is necessary following a
healed fracture (Kaufman et al., 2003). Considering these barriers, it is not surprising that many
patients are reluctant to undergo further testing and initiate treatment: if a patient believes that
OP management is not applicable to them or that the existing treatments are ineffective, they are
not likely to initiate and adhere to the prescribed therapy and recommendations. Meadows
(2003), for example, found that female fragility fracture patients who perceived their fractures as
accidents, rather than markers of poor bone health, took little or no steps towards OP
management.
Luckily, these patient-level barriers are modifiable. They could be modified through
interventions that focus on changing patients’ perception of OP-related issues with the goal of
ultimately changing patients’ health behaviour in favour of initiation of OP management. One
does not have to search far to find information on health behaviour change. The process by
which individuals change their behaviours has been the focus of a large area of research:
individual theories of health behaviour change. Different theories propose different models of
behaviour change but they all concentrate on an individual’s perception of his or her
health/condition (Glanz, Rimer, & Lewis, 2002). Given that many of the barriers to more
effective OP management stem from patients’ perception of their condition, it would be fitting to
apply these theories to the context of post-fracture OP management. Whether these theories have
been applied to the context of post-fracture OP management (and what the resulting outcomes
are) has not been examined so far.
An important barrier to more effective OP management and prevention of future fractures
involves patients’ lack of recognition that their fragility fracture may be associated with an
underlying bone health disease and therefore with the risk of future fractures. Even though a
fragility fracture is the best predictor of future fractures (Klotzbuecher et al., 2000) and the main
risk factor that identifies who should be assessed for OP (Brown et al., 2006), many fragility
7
fracture patients do not make the link between their fracture and OP (Giangregorio et al., 2008).
Making this link is important because a fragility fracture is usually the first and, with the
exception of vertebral fractures, the only “symptom” of OP. The importance of making this link
has been recognized and incorporated into interventions by two domestic, province-wide post-
fracture OP screening initiatives: Recognizing Osteoporosis and its Consequences in Quebec
(ROCQ) (Bessette et al., 2008), a health-management program aimed at evaluation of OP care
following a fragility fracture in Quebec, and the Ontario Osteoporosis Strategy’s Fracture Clinic
Screening Program, a province-wide initiative of the Ministry of Health and Long Term Care of
Ontario (Jaglal et al., 2010). Promoting the awareness of the OP-fracture link is a part of the
intervention implemented by ROCQ, while increasing patient’s awareness of the OP-fracture
link is one of the goals of the Ontario’s Fracture Clinic Screening Program. The two initiatives
emphasize the OP-fracture link for a good reason: if patients perceive that their fracture is due to
an isolated accident, rather than poor bone health, they are not likely to take action following the
fracture as they are not likely to perceive OP management as relevant to them. Perception of
one’s fracture as osteoporotic has been identified as a predictor of follow up with a physician for
investigation of OP (Bliuc et al., 2006) and initiation of OP management (Beaton, Jiang, Sujic,
Sale, & Bogoch, 2009; Beaton, Bogoch, Sujic, & Elliot-Gibson, 2007; Meadows et al., 2005).
Knowing which patients are less likely to make the link (even after a baseline intervention)
would help target interventions to that group of patients in order to help them link OP and their
fracture. The predictors of which undiagnosed and untreated patients are more or less likely to
make the OP-link after a fragility fracture have not yet been examined.
1.7 Objectives The overall purpose of this thesis is to examine the role of patients’ perception of OP in the
context of post-fracture care in order to contribute to the body of knowledge necessary for
designing more effective OP interventions. There are two specific objectives:
1. To determine whether current OP interventions (that aim to improve post-fracture care)
account for patient perceptions of their health and OP. Patients’ perceptions are an important
factor in behaviour change and are a main building block of individual-level theories of
health behaviour change that aim to explain how and why behaviour change occurs. It is
8
unknown whether these theories have been used in the context of post-fracture OP
interventions. My first objective will therefore consist of performing a literature review to
examine whether individual-level theories of health behaviour change have been applied to
post-fracture OP interventions and if so, what were the outcomes. This objective will be
addressed in Chapter 2 (manuscript “A Review of Patient-centered Post-fracture
Osteoporosis Interventions in the Context of Health Behaviour Change Theories”).
2. To identify factors predictive of patients’ making the link between their fracture and OP.
Making the OP-fracture link is important in initiation of post-fracture OP management as
fractures are usually the first signs of an underlying bone health issue and are strongly
associated with future fractures. Factors that predict which patients (who were not previously
diagnosed or treated for OP) are more or less likely to make this link are unknown.
Identification of patients less likely to make the link (after a baseline intervention) and their
characteristics would help create interventions targeted to that group of patients. The second
objective of this thesis is to identify and examine predictors of making the OP-fracture link in
previously undiagnosed and untreated patients. Chapter 4 (manuscript “Factors Associated
with Patients Making the Link Between Their Fragility Fracture and Osteoporosis”) will
focus on this objective.
1.8 Chapter summary OP and related fractures are a global health problem associated with high human and health care
costs. Previous fragility fracture is a major risk factor for future fractures and a major indication
for OP assessment. Despite effective treatment options, guidelines and previous research
advocating assessment and treatment of fragility fracture patients, the majority of patients do not
receive appropriate OP management. Barriers to more effective OP management are associated
with both the healthcare environment and the patient. Removing healthcare-level barriers is
important, however, interventions are not likely to be successful if patient-level barriers are not
addressed. Patient level-barriers are modifiable through interventions focusing on changing
patients’ perception of their condition with a goal of changing their behaviour in favour of
initiating OP management. Theories of health behaviour change would be applicable to this
9
context as they focus on patient perception and aim to explain how and why behaviour change
occurs. The first objective of the thesis is to examine whether theories of health behaviour
change have been applied to post-fracture interventions. The second objective will focus on
patient perception of the OP-fracture link as many patients do not make this link. Making the
OP-fracture link is important because a fragility fracture is usually the first and only sign of
deteriorating bone health. The second objective will examine the predictors of making and not
making this link in undiagnosed and untreated patients in order to identify the patients who are
not likely to make the link and who may require a more intensive intervention.
10
Chapter 2 A Review of Patient-centered Post-fracture Osteoporosis
Interventions in the Context of Health Behaviour Change Theories
2.1 Introduction OP is a chronic disease characterized by a decrease in bone density, making bones fragile and
susceptible to fractures. This disease usually remains asymptomatic until a fragility fracture
occurs; patients who sustain one fracture are at a much higher risk of additional fractures
(Klotzbuecher et al., 2000). Osteoporotic fractures are associated with high human and health
care costs including loss of mobility, decreased independence, high financial cost and, in the case
of hip fractures, death (Wiktorowicz et al., 2001; Goeree & O'Brien B, 1996).
Effective treatment options and clinical practice guidelines for OP care and prevention of future
fractures exist (Brown & Josse, 2002; Brown et al., 2006) however, less than 32% of patients
who sustain a fragility fracture receive appropriate OP management (Elliot-Gibson, Bogoch,
Jamal, & Beaton, 2004). Barriers to more effective OP management are well documented and
associated with both the patient and the healthcare environment. Patient-related barriers include a
lack of knowledge of OP and consequences of fractures, as well as the perception that treatment
is ineffective and that fracture healing is the end of the health concern (Kaufman et al., 2003;
Meadows et al., 2005). It is therefore not surprising that many patients are reluctant to undergo
further testing and treatment and that medication initiation and adherence remain suboptimal
initiation and adherence are ongoing health behaviours that require patients to make and sustain
changes to their existing behaviour. Whether patients will change their behaviour and
initiate/adhere to the prescribed therapy is, in part, dependent on the patients’ perception of the
state of their health (Gold et al., 2006; Wichowski & Kubsch, 1997). Patients’ perception of their
health plays a central role in individual-level theories of health behaviour change. Major theories
of health behaviour change such as the Health Belief Model (Rosenstock, Strecher, & Becker,
1988), Teachable Moment (McBride, Emmons, & Lipkus, 2003), Transtheoretical Model
(Prochaska & Velicer, 1997), Theory of Planned Behaviour (Fishbein M & Ajzen I, 1975) and
11
Social Cognitive Theory (Bandura, 2004) have been developed to explain how and why
individuals change their behaviour. Each of these theories is characterized by a defining
component. Patients’ perception of their susceptibility to a condition is a defining component of
the Health Belief Model; self-efficacy (one’s sense of control and confidence in ability to take
action) is a core feature of the Social Cognitive Theory; the Teachable Moment theory
emphasizes the importance of the affective reaction to a cueing event (event that triggers action);
behavioural intention is a key factor of the Theory of Planned Behaviour; and the
Transtheoretical Model is characterized by the idea that change is staged and not immediate.
Although these theories propose different models of behaviour change, they also share many of
the same features: in order for health behaviour change to occur, most of them would postulate
that patients need to first perceive a health threat, see themselves as susceptible to it, believe that
benefits of the behaviour change outweigh the barriers, perceive themselves as having enough
ability to perform this change and perceive the behaviour as socially desirable. Key concepts of
major theories of behaviour change and their definitions are presented in Table 2.1 (on the next
page).
Theories of health behaviour change have been applied to a variety of behaviours such as
smoking cessation, dietary change, weight control, exercise acquisition, HIV prevention and
chronic disease prevention and management (Becker et al., 1978; Glanz et al., 2002; Munro,
Lewin, Swart, & Volmink, 2007; Prochaska et al., 1994). Theories of health behaviour change
have also been used in the context of OP, mainly in the area of primary prevention (Hsieh,
Wang, McCubbin, Zhang, & Inouye, 2008; Popa, 2005; Wallace, 2002). Some theories have also
been applied to the context of OP management and medication initiation: the Health Belief
Model was used to examine the role of OP beliefs and attitudes in antiresorptive medication use
in community-dwelling women (Cline, Farley, Hansen, & Schommer, 2005) while Gibson’s
patient-centered, self-empowerment model was used to investigate the potential of an
Osteoporotic Pain Program (Jensen & Harder, 2004). Mauck and colleagues utilized a stage of
change model (Weinstein Precaution Adoption Process Model) to examine patients’ decision to
accept OP treatment after hip fracture (Mauck et al., 2002). The same theoretical framework was
used to explore barriers to OP care from the perspective of female fragility fracture patients
(Edwards, Iris, Ferkel, & Feinglass, 2006). Gold and colleagues implemented a self-management
12
OP program for post-menopausal women with OP (or OP risk) that incorporated elements of the
Social Cognitive Theory (Gold & Silverman, 2004).
Table 2.1 Main concepts of five major individual-level theories of health behaviour change Moderating factor = + Main factor = + + Defining factor = defining
Concept
Definition
Health Belief Model
Social cognitive theory
Teachable moment
Theory of planned behaviour
TTM Stages of change
Predisposing factors
Pre-existing factors that may influence the behaviour
+ + +
Cueing event Event that triggers ‘readiness’ to change or provides a cue to action
+ (cues to action)
defining
Affective response
Emotional reaction to a cueing event + +
Perceived susceptibility
to a condition
One’s perception of threat risk or belief about the likelihood of developing a condition
defining + +
Perceived severity or seriousness
One's belief about seriousness of a condition/threat and its consequences
+ +
Perceived benefits
One’s beliefs about the positive outcomes associated with a behaviour
+ + + +
+ + + + +
Perceived Barriers (to
taking action)
One’s perception of the obstacles to specific behaviour/action
+ + + + + + +
Self-Efficacy One’s sense of control and confidence in ability to take action
+ defining + + + +
Social approval
/norm
One’s perception about the social context and the extent to which the behaviour is socially desirable
+ + + + + +
Proximal goals
(intention)
Behavioural plans predictive of performing behaviour
+ + defining
Staged change
Behaviour change involves movement through a sequence of discrete stages
defining
13
A wide range of post-fracture interventions have been created and implemented within the
orthopaedic environment as this is a setting where the concentration of fragility fracture patients
(and cost-effectiveness of interventions) is the highest. These interventions range from sending
reminders and education to GP’s, to complex programs where a dedicated screening coordinator
identifies, educates and coordinates testing and treatment of possible OP (Elliot-Gibson, Bogoch,
Jamal, & Beaton, 2004). It is not known whether behaviour change theories have been used to
create any post-fracture OP interventions within the orthopaedic environment.
Given that post-fracture OP management is a multi-stage process requiring patients to make and
sustain changes to their existing behaviour, it is important to evaluate whether models of
behaviour change have been used in this context. The purpose of this literature review is to
determine whether, and to what extent, current post-fracture interventions utilize theories of
behaviour change and whether those that are theory-based are more successful in changing
health behaviours.
2.2 Methods This literature review was based on studies identified as part of a broader systematic review of
non-surgical OP interventions in the orthopaedic environment (Sale J, under review). The initial
search using keywords “osteoporosis” and “intervention” was conducted in CINAHL, EMBASE
and OVID Medline databases in May 2008; an additional search was conducted in September
2008. Articles were limited to those published in English in the last 20 years that contained the
above keywords in either the title or the abstract (n=608). These articles were further limited to
333 articles that focused on post-fracture OP care in the orthopaedic environment. Letters,
commentaries, literature reviews and articles one page in length or less were discarded. The 333
articles were then restricted to patient-targeted, primary studies that aimed at improving post-
fracture OP care. There were 35 articles that fulfilled these criteria based on a full article review
(Figure 1.1). The 35 studies included in the final review were described in terms of study design,
target population, interventions and outcomes and examined for explicit mention of an
underlying theory.
14
608 articles identified through title
and abstract screen Keywords: OP and non-
surgical intervention
333 articles identified through title
and abstract (full article if needed)
35 articles identified
through full article review
Restricted to articles on post-fracture care. Excluded letters, commentaries, literature
reviews, articles that were 1 page in length.
Restricted to: - primary studies (no literature reviews) - post-fracture population - patient-targeted interventions - studies aimed at improving OP care
Figure 2.1 Search strategy for the literature
review of post-fracture osteoporosis interventions
If the authors did not explicitly state that they used a theory, we looked for a description of a
theoretical framework consistent with one of the major behaviour change theories. To do this, we
searched for presence of constructs from theories of behaviour change and examined whether
they were integrated in a way suggested by one of the theories of behaviour change. Appendix I
table provides a description of all the studies in terms of study design, target population,
theoretical framework, interventions and outcomes. Eight studies contained descriptions of three
programs; those that described the same program were grouped together.
15
2.3 Results Thirty-five studies were included in the final review (Ashe et al., 2004; Astrand, Thorngren, &
Tagil, 2006; Astrand, Thorngren, Tagil, & Akesson, 2008; Bessette et al., 2008; Bliuc et al.,
2006; Blonk, Erdtsieck, Wernekinck, & Schoon, 2007; Bogoch et al., 2006; Che, Ettinger, Liang,
Pressman, & Johnston, 2006; Chevalley et al., 2002; Cuddihy et al., 2004; Davis, Guy, Ashe,
Liu-Ambrose, & Khan, 2007; Edwards et al., 2005; Feldstein et al., 2006; Feldstein et al., 2007;
Chapter 3 Critical Appraisal of the Ontario Osteoporosis Strategy’s Fracture Clinic Screening Program Database as a Source of Data for the
Thesis
3.1 Introduction The first objective of this thesis, a review of current OP post-fracture interventions in the context
of theories of health behaviour change, has been described in Chapter 2. The review showed that
theories of behaviour change have not been integrated into post-fracture interventions and that
more research into the role of patient perceptions is necessary in this area.
The second objective of this thesis is to further examine patient perceptions by analysing whether
patients perceive the link between their fracture and OP. The analysis of the OP-fracture link will
be described in Chapter 4. This chapter will provide background information to set the stage for
the OP-fracture link analysis by describing and critiquing the source of data.
3.2 Background and description of the database Data used in the OP-fracture link analysis were derived from the Fracture Clinic Screening
Program database. The Fracture Clinic Screening Program, one of five components that comprise
the Ontario Osteoporosis Strategy, was implemented by Osteoporosis Canada, under the
directive of the Ministry of Health and Long Term Care (MOHLTC) of Ontario (Jaglal et al.,
2010). As part of this program, Osteoporosis Canada placed screening coordinators in designated
fracture clinics across Ontario with the long term goal of increasing secondary prevention among
fragility fracture patients and decreasing the rates and associated costs of subsequent fractures.
The clinics were identified by estimating yearly volumes of fragility fracture patients using
administrative data. Sixty clinics with high or moderate volumes were identified as sites where it
could be cost effective to place a screening coordinator to perform screening and facilitate follow
up for OP management. At the time data used in this thesis were collected, there were 19
coordinators working in 36 (of 60) clinics across Ontario.
23
As part of the Fracture Clinic Screening Program, the role of the screening coordinators was to
identify, assess, refer and provide education to fragility fracture patients who presented
themselves at the clinics where the coordinators were located. The process started by
coordinators identifying patients 50 years of age and older who had sustained a possible fragility
fracture (i.e., who fractured due to low trauma, defined as a fall from standing height or less) of
the wrist, clavicle, elbow, spine, pelvis, femur, tibia/fibula, ankle or other selected sites.
Screening coordinators first approached the identified patients in order to confirm that they had
sustained a low trauma fracture. Patients who sustained a low trauma fracture (or were referred
by a surgeon) were interviewed by screening coordinators in order to complete the baseline
questionnaire. If the patient or his/her proxy was unable to answer the questions and was
therefore unable to complete the baseline questionnaire, the coordinators would complete only
the first section of the questionnaire which included information on patient’s age, primary
language, first three digits of the postal code (FSA), fracture (site, etiology, date) and reasons for
not completing the questionnaire (language/mental/physical barriers or refusal). Other sections of
the questionnaire covered OP risk factors, OP diagnosis and treatment history, bone mineral
density (BMD) testing, demographics, documentation of intervention and patient’s OP
knowledge and OP beliefs. With the exception of patient’s OP knowledge and beliefs, the
questionnaire could be filled out by an accompanying person (proxy), if the patient was unable to
do so.
The intervention that patients received as part of the Fracture Clinic Screening Program
consisted of: verbal recommendations for further assessment of bone health and BMD testing,
recommendations for Calcium and Vitamin D intake and provision of (written) general OP
educational materials, Program Information Letter and other Osteoporosis Canada resources
(details are available in Appendix II). Screening coordinators would also send a letter to the
patient’s family physician with a request for further assessment of bone health (unless the patient
requested that his or her family physician not be contacted).
All patients who were not diagnosed and/or treated for OP at the time of screening were asked if
they would agree to a telephone follow up at three and/or six months. Those who agreed were
called at the appropriate time and asked about post-fracture follow up with their family physician
and new testing, diagnosis and treatment information. The same OP knowledge/beliefs questions
24
asked at baseline were also asked at follow up. Patient flow through the Fracture Clinic
Screening Program is depicted in Figure 3.1.
Patients ≥ 50 years of age, potential low trauma fracture
approached by coordinators to establish fracture etiology
Low trauma patients (or those referred by a surgeon)
completed baseline questionnaire
Undiagnosed and/or untreated patients were asked to agree to follow
up
Screening stopped for moderate or high trauma, missed patients or those who refused to participate in the program
Interaction stopped for patients who were previously diagnosed or treated for OP
Interaction stopped for patients who were lost to follow up
Figure 3.1 Patient flow through the
Fracture Clinic Screening Program
Patients followed up at three or six months to complete follow up
questionnaire
Data collected at baseline and follow up were used for ongoing quality improvement within the
program and for reporting to the MOHLTC. A subset of data was also used for research: all
patients (or their proxies) capable1 of providing their consent were asked whether they would be
1 As per coordinator’s assessment
25
willing to: (a) allow the information gathered as part of the program to be used for research
purposes; and (b) provide necessary personal health information to allow linkage to the
MOHTLC databases for research into future fracture rates and OP-related care. Patients’ consent
was documented in the database and only those patients who consented to research use of their
baseline and follow up information were included in the research dataset. Approval of the site’s
research ethics board was also obtained prior to consenting patients for inclusion in the database.
3.3 Critique of the database: establishing its suitability to the research question
In this section, I will assess the strengths and weaknesses of the Fracture Clinic Screening
Program as a source of data for my research question. To make this assessment, I will use
elements from Connell’s (1987) framework for identifying and dealing with issues inherent in
working with large health care research databases. To understand possible data flaws, Connell
suggested examining the following categories: data collection methods (population, sampling
strategy, time frame, how were data collected, inclusion criteria, comparison of subgroups),
database structure (unit of observation, content of each observation, duplicates), quality of data
elements (coding, missing observations, out of range values) and linkage of datasets. As part of
the same framework, Connell also listed specific strategies such as manual examination of
records for data flaws, frequency analysis of all variables and inspection for unusual
distributions, out-of-range and missing values (Connell, Diehr, & Hart, 1987).
My description and critique of the Fracture Clinic Screening Program database will be based on
Connell’s framework and organized as follows:
1. Purpose and content of the database, sampling strategy, comparison of the sample to the target
population and data collection methods
In addition to discussing the original purpose and content of the database, I will examine the
sampling strategy (including time-frame and the inclusion/exclusion criteria), comparison of
the sample to the target population, data collection methods and possible biases associated
with data collection.
26
2. Database structure, content of each observation, duplicates and coding
In this section, I will review the overall structure of the database, the content of each
observation and the data quality issues (possible duplicates and coding of variables).
3. Manual inspection of data for data flaws and an assessment of the overall data quality
To assess features that may indicate a pattern of errors, I will perform a manual inspection of
three samples of raw data using the first, last and random 10% observations. I will critique
any patterns of miscoding, missing observations or out of range values that may be found.
Overall quality of the data will be assessed as well.
3.3.1 Purpose and content of the database, sampling, comparison of the
sample to the target population and data collection methods
Purpose: One of the first steps in establishing appropriateness of the database to the research
question is to consider the original purpose of the database and its content (Connell et al., 1987).
The Fracture Clinic Screening Program database was created as part of a province-wide
screening program whose short-term objectives were to enhance patient OP knowledge, the
awareness of the OP-fracture link and the need for testing, as well as to increase OP awareness of
healthcare providers and increase referrals of fragility fracture patients to family physicians for
assessment and treatment. The database was designed to collect data for ongoing evaluation of
the screening program (quality assurance) and future research into post-fracture OP care and care
gaps in Ontario, with the final goal of improving OP care in the province. Variable selection was
done by the research team and Osteoporosis Canada, based on the available literature and input
from experts in the field. Specific variables were selected in order to address a number of
possible research questions and to inform quality assurance initiatives. Variables were taken
from both previously validated instruments and previous studies to allow comparability with the
literature. Even though the research question asked in this thesis was not established prior to data
collection, the database contained variables (predictors and the outcome of the OP-fracture link)
that could be applied to the research question being addressed in this thesis. In addition, the
database contained the data from fragility fracture patients, the target population of this thesis.
27
Given that the database variables and the population met the needs of this research (to inform
efforts to optimizing post-fracture OP care in an orthopaedic setting), it can be concluded that the
Fracture Clinic Screening Program database was a good source of information for the research
question to be answered.
Content: At the time data used in this research were collected, the Fracture Clinic Screening
Program database consisted of self-reported baseline and follow-up data from fragility fracture
patients who were screened by 19 screening coordinators at 36 sites across the province.
Fragility fracture patients who were over the age of 50 at the time of the fracture were the target
population for the screening program and its research component. Baseline information was
collected on all patients who fit the screening criteria (over the age of 50 with a low trauma
fracture or surgeon-referred) while the follow up was done only with those patients who were not
diagnosed and/or treated for OP at baseline. Only these undiagnosed and untreated patients were
followed up because (a) the resources were not sufficient to follow up with everyone and (b) the
undiagnosed and untreated patients were a priority for the intervention (assessment, education,
referral) as those previously diagnosed and/or treated for OP had already taken steps towards OP
management. Patient data collected at baseline included questions on OP risk factors, OP
diagnosis and treatment history, BMD testing, demographics, documentation of
intervention/patient consent and patient’s OP knowledge and beliefs. The follow up
questionnaire also included questions about OP knowledge and beliefs in addition to follow-up
questions about post-fracture visits with a family physician and testing, diagnosis and treatment
information. Questions asking about OP knowledge and beliefs were derived from previous
research. OP knowledge was measured by four items from the “OP and You” knowledge
questionnaire (Brenneman, Blau, Chen, & Abbott, 2002) previously selected and validated by
Cadarette (Cadarette, Gignac, Beaton, Jaglal, & Hawker, 2007). Questions on OP beliefs
included questions on patients’ perception of OP treatment, the link between their fracture and
OP, the quality of patients’ bones as well as patients’ perception of benefits and barriers to OP
pharmacotherapy. Questions on both perceived benefits and barriers to pharmacotherapy were
derived from previous research (Cadarette, Beaton, & Hawker, 2004) and based on the
“Osteoporosis Health Belief Scale” (Kim, Horan, Gendler, & Patel, 1991). Perception of benefits
of OP pharmacotherapy was measured by a 5-item OP benefits scale while barriers were
28
assessed by two individual items. The measures are described in greater detail in Chapter 4 with
additional information presented in Appendix III (construct list).
Even though the database contained the variables of interest to this thesis, there are other
concepts which could have also been explored. For example, Tugwell’s concept of the patient as
an “effective consumer” (Tugwell, Santesso, O'Connor, & Wilson, 2007) or a related concept of
patient’s health literacy (McCabe, 2006), would have also been of interest but were not explored
because they were not captured by the database. The concept of health literacy, defined as ”the
degree to which individuals have the capacity to obtain, process and understand basic health
information and services needed to make appropriate health decisions" (National Library of
Medicine, 2010) could be a potential predictor of the OP-fracture link. Patients who have high
health literacy may be more likely to make the OP-fracture link because they would also be more
likely to obtain relevant health information and apply it to themselves appropriately. Whether
this ability to obtain, interpret and apply relevant health information is predictive of making the
OP-fracture link at follow up was left unexplored. Other concepts that could be useful predictors
of the OP-fracture link included cultural and social issues, education level and co-morbidities.
Despite the database not capturing every variable that would be of interest for this thesis’
research question, the database contained most variables of interest making it a rich and fitting
source of data.
Sampling strategy: The sample used to answer the research question on the OP-fracture link
was derived from the Fracture Clinic Screening Program research database. At the time of
analysis, this research database contained data from consenting patients who were screened
between December 2007 (when the consenting process started) and September 2009 (as per the
REB approval at the time2). For purposes of the OP-fracture link analysis, the sample was
further defined to include patients who:
a) Were eligible for follow up (not diagnosed or treated for OP at baseline) and agreed to
follow up
2This REB approval allowed for research use of baseline data collected up until September 30, 2009 and the follow up data of the same patients obtained thereafter.
29
b) Were followed up by December 31, 20093 and had complete data for the OP-fracture link
question at both baseline and follow up
c) Did not make the link between OP and their fragility fracture at baseline
In using the above inclusion criteria, the research dataset (containing only data from consenting
patients) was reduced from 4,657 fragility fracture patients over the age of 50 who were eligible
for follow-up to 853 patients who fulfilled the rest of the inclusion criteria (Figure 3.2 Sampling
strategy for the OP-fracture link analysis ).
Figure 3.2
Sampling strategy for the OP-fracture link analysis
n=4657 Eligible for follow up
3 December 31, 2009 was chosen as a cut-off date because the OP-fracture link analysis started in January 2010.
n=4011 Did not refuse follow up
n=2207 Followed up with
n=916 Had complete data on OP-fracture
link at baseline and follow up
n=853 did not make the OP-fracture link
at baseline
30
The inclusion criteria reduced the sample size but also helped meet the requirements of the OP-
fracture link research question. These inclusion criteria were set for the following reasons:
a) The research question examined the change in the OP-fracture link over time; therefore
those ineligible to be followed up (treated and/or diagnosed with OP) were excluded as
they would not have the necessary follow up data. Second, including patients who were
previously treated and/or diagnosed with OP would also make the direction of the OP-
fracture association unclear as previously diagnosed and/or treated patients have already
taken steps towards OP management and could make the OP-fracture link at follow up as
a result of being diagnosed and/or treated at baseline. The OP-fracture link analysis
focused on identifying clinically-useful predictors of making the link which was
hypothesized to lead to taking steps towards OP management: a finding that patients who
were already diagnosed and/or treated at baseline make the OP-fracture link at follow up
would therefore have not been clinically useful.
b) Since the research question examined predictors of the OP-fracture link at follow up, it
was necessary to know which patients did or did not make the link at baseline and follow
up. Only patients who had data on this variable (OP-fracture link) at both baseline and
follow up were included in the analysis.
c) Because the research question aimed to identify predictors of the OP-link, including
patients who had already made the link would not contribute to the findings. The analysis
was therefore further restricted to those who did not make the OP-fracture link.
Comparison of the sample to the target population: Whenever conclusions are drawn from
sample-specific observations, there is a possibility that these conclusions may not be
generalizable because the sample may be systematically different from the population from
which it was drawn (Fletcher RH, 2005). Such sampling bias can occur anytime a sample is
drawn from a population of interest. To examine whether the sampling bias was present in this
study, I compared my study sample to the target population. Since I was only interested in
generalizing the findings to patients who did not make the OP-fracture link at baseline, I
compared my sample (n=853) to patients who were eligible for follow up, who did not make the
OP-fracture link at baseline (n=3090) and who were not in my study sample (3090 – 853 =
2237). Figure 3.3 shows the comparison that took place.
31
Figure 3.3 Comparison of the sample (n=853) to the
target population (n=2237)
n=4011 Did not refuse follow up
n=2207 Followed up with
n=916 Had complete data on OP-
fracture link
n=853 did not make the OP-fracture
link at baseline
n=4657 Eligible for follow up
n=3090 did not make the OP-
fracture link at baseline
Compare n=2237 (3090-853=2237)
to the sample n=853
The results of the comparison are shown in Table 3.1. The sample and the rest of the population
were compared on possible confounders, predictors and the outcome of interest. They differed on
four variables: work status, perception of OP treatment benefits, steroid use and one OP
knowledge item “If a person has OP, something as simple as lifting a bag of groceries can break
a broke”. Specifically, the sample used in the OP-fracture link analysis consisted of fewer
patients who were employed (30.1% vs. 35.4%), more oral steroid users (6.3% vs. 4.3%), more
patients who perceived higher benefits of OP drug treatment (4.2/5 and 4/5) and more patients
who correctly answered the above-mentioned knowledge item (84% vs. 78%). Fewer employed
patients in the sample could have been a result of the follow up procedure that coordinators used
where most follow up calls were made during working hours (9am-5pm), resulting in more
complete data among those who were not working and were available during these hours. The
second variable that my sample and the comparison group differed on was the perception of OP
drug treatment benefits. Even though the difference was statistically different in two groups, both
groups perceived four out of five benefits to OP pharmacotherapy.
32
Table 3.1 Description and comparison of the sample n=853 to the target population n=2237 Variables n=853 n=2237 p value
Mean age (SD) range 66.6 (10.5) 50-94
66.6 (10.8) 50-95 0.97
Gender – Female
79.7%
76.8%
0.08
Marital status: Married/common law
59.7%
62.5%
0.34
Living situation: Live alone
28.7%
26.8%
0.31
Work full or part time
30.1%
35.4%
0.0054
Hip or shoulder fracture (may include others)
25.6%
24.5%
0.56
Previous fracture >age 40
25.2%
25%
0.94
Maternal fracture >age 40
28%
25%
0.09
Previous fall in the last year
26.6%
27.2%
0.73
Feeling unsteady
27%
30.2%
0.08
Current or past steroid use
6.33%
4.31%
0.02
Perceived benefits of OP drug treatment (1-5)
4.2
4
0.0032
% Who perceived bones as thin
10.6%
10%
0.11
Knowledge item: There is a way to prevent OP (reversed)
64.2%
67.7%
0.07
Knowledge item: Bones can be rebuilt once they are thin from OP (reversed)
57.1%
55.5%
0.41
Knowledge item: If a person has OP, something as simple as lifting a bag of groceries can break a bone
84.3%
78.7%
0.0006
Knowledge item: Health problems caused by OP can be life-threatening
69.2%
71.1%
0.33
Perceived barriers item: Taking too many medications
22.2%
20.8%
0.38
Perceived barriers item: Stomach problem limits ability to take drug treatment
17.5%
20.3%
0.1
The difference in use of oral steroids and correctly answering the knowledge item “If a person
has OP, something as simple as lifting a bag of groceries can break a bone” was statistically
33
significant and could have biased the sample; however, the differences were not large (2% oral
steroids and 14% knowledge question). The sample and the population did not differ on other
variables of interest: there were no statistically or clinically important differences in age, gender,
marital status, living situation, type of fracture, existence of previous or maternal history of
fractures, previous falls, feeling unsteady, perception of the bone quality, nor remaining three
knowledge items and perceived barriers to OP drug treatment.
The results of the comparison of the sample to the rest of the population revealed that the sample
did not differ from the population on most variables but was slightly different when it came to
the employment rate, steroid use and perceived benefits of OP treatment. Since the differences
were not large, the results of the analysis should be generalizable to the target population as long
as these differences are taken into account.
Data collection methods: Data collection process used to create the Fracture Clinic Screening
Program database was described in the first section of this chapter. Data collection was
electronic: coordinators entered data directly into a centralized database hosted and maintained
by an external vendor. Other modes of data collection were also available in case direct
connection to the internet and server was unavailable: coordinators could collect data on tablets
and upload them at a later time. In case of electronic data collection failure, paper versions of the
questionnaire were available. The main advantage of electronic data collection was that it saved
time by skipping the data entry step (which can also carry a risk of data entry errors).
In comparison to other modes of data collection, such as mailed surveys, the disadvantage of
data entry done by coordinators was the dependence of data capture on the time that coordinators
had available to perform data collection. Time allocated to certain tasks could also be an issue: a
number of patients were not followed up because of the limited number of follow up phone calls
that coordinators were supposed to make4. However, in contrast to mailed surveys where the
completeness of responses and return of the survey are at the sole discretion of the participant,
data collection done by coordinators allowed for more control over data capture.
4 According to the protocol, coordinators were to make three follow up calls at three months and three calls at six months.
34
A number of systematic errors can occur during the process of data collection. Measurement
bias, a systematic error that poses a threat to the validity of results, occurs when the methods of
measurement are different among different groups of patients (Fletcher 2005). Since data
collection for the purposes of the Fracture Clinic Screening Program involved different screening
coordinators collecting data at different sites, measurement bias could have occurred. To
minimize the possibility of such bias occurring, each coordinator was trained in data collection
by the same consultant and the same protocol for data collection and documentation had been
followed at each of the 36 sites. Quality checks were also built into the system to avoid out of
range values or incorrect skip patterns. As part of ongoing quality assurance efforts, coordinators
were regularly given feedback about data they collected. A private website was created in order
to regularly communicate data quality issues with coordinators.
Another potential bias associated with measurement could have occurred in situations when
coordinators needed to use paper copies of the questionnaire because the electronic data capture
was unavailable. In such cases, coordinators were able to omit entire sections of the
questionnaire and a deviation from skip patterns, which are automatic in the electronic version,
was possible. To examine the possibility of such deviation occurring, a manual examination of
raw data (described in the next section) was undertaken.
Systematic errors are a possibility during any data collection procedure, particularly in multi-site
studies. Fracture Clinic Screening Program placed a number of procedures into place in order to
decrease such errors during data collection and ensure that potential measurement bias was
minimized.
3.3.2 Database structure, content of each observation, duplicates and coding
Structure: The Fracture Clinic Screening Program database was designed with fixed variables
with pull-down menus (multiple choice) and “free text” fields when “other” value was chosen.
Appendix III contains the construct list with baseline and follow-up variables contained in the
database. As part of the quality assurance, a number of smaller datasets, corresponding to
different sections of the baseline and follow up questionnaire, were extracted from the
35
centralized database and sent to the evaluation and research team at St. Michael’s Hospital where
the smaller datasets were merged into a single database. Data cleaning was done as part of the
quality assurance. Screening coordinators were regularly given feedback on the quality of the
data they were collecting.
Content of each observation: In the study sample, each observation referred to one patient and
contained data collected at both baseline and follow up (this was not the case in the Fracture
Clinic Screening Program database where every patient did not necessarily have follow up
information). Each observation/record in the database was marked by a unique study ID referring
to one individual patient. The unique study ID was a seven digit number assigned by the
screening coordinator who performed data collection: first two digits corresponded to the pre-
assigned site ID (indicating which hospital the patient was seen at) and the last five were patient-
specific (in chronological order). Another unique study ID was also assigned by the electronic
system, when the observation was entered into the database. Having two separately-assigned
ID’s (assigned by the system and the person performing data collection) was useful in
identification of errors and possible duplicate data entries.
Duplicates: Checking for duplicates in the unique ID was performed as part of the data quality
assurance. The potential for coordinators-assigned ID errors was high given that this ID field was
in “free-text” format where a number of variations of the same ID were possible. For example, a
patient number 36 from site 20, who would have been identified as “20-00036”, could have also
been identified as “20 036”, “20-00000036” or “00036”). A number of variations of the required
seven-digit format were indeed found during the data cleaning procedure (part of quality
assurance). This issue was resolved by creating a new subject ID that consisted of the last three
digits of the coordinator-assigned subject ID (the part that was correctly formatted) combined
with the ID number specific to the site where the patient was screened. Duplicates were checked
for by examining frequency distributions of both the newly-created subject ID and the subject ID
assigned by the system during data entry. In case of doubt, records were manually checked for
possible duplicate content. All duplicates found were excluded from the overall dataset. All of
this was done as part of quality assurance efforts, prior to extracting the study sample from the
database.
36
Coding: Initial coding structure was agreed upon at the time of variable selection for the
Fracture Clinic Screening Program database. Possible inconsistencies in coding in the study
sample were checked by examining the frequency distribution for all variables. All data elements
were appropriately coded: no changes in coding were found and there were no variables that
were not accounted for. Some categorical variables were re-coded after examination. For
example, the codes “thin” and “definitely thin” were re-coded into one category “thin”; “agree”
and “strongly agree” were re-coded as “agree”; and “disagree” and “strongly disagree” were re-
coded as “disagree”. Distributions of continuous variables were also examined and no out-of-
range values were found. During frequency distribution checks, variables that were marked by a
large number of missing observations were flagged for further sensitivity analysis.
3.3. 3 Manual inspection of data for data flaws
In this section, I made comparisons across three data samples to determine any time-dependent
shifts in data quality. Namely, I evaluated data flaws (undocumented codes, unexpected
distributions and missing values) by manually examining three samples of raw data: the first
10% (observations 1-85), last 10% (observations 768-853) and random 10% (observations
numbered 86-767). I also examined the frequency distribution of the key variables (predictors,
moderators, outcomes) for potential data flaws.
Raw data examination showed that there were no undocumented codes. The only unexpected
distribution was found in the last 10% of the data where 16% of answers for the knowledge
question “The health problems caused by OP can be life-threatening” were marked as “refused”
(27/170). The same question had 12% (10/85) of missing values in the random 10% sample.
The only difference in the structure across the three sections was that there were no missing
values in the last 10% of observations, which could be due to OP knowledge/beliefs questions
becoming mandatory later on in the data collection process (which would also explain the high
number of “refused” in the same section). Table 3.2 shows the distribution of missing values for
the three sections.
37
Table 3.2 Manual examination of raw data: missing values in three data samples
Variable First 10%
(n=85)
Random 10%
(n=85)
Last 10%
(n=85) Age 0 0 0
Gender 0 0 0
Marital status 0 0 0
Living situation 0 1 0
Work status 0 0 0
Fracture type 0 0 0
Previous fracture 0 0 0
Maternal fracture 0 1 0
Fall in the last year 0 3 0
Feeling unsteady 0 2 0
Current or past oral steroid use 0 0 0
OP knowledge items • There is a way to prevent OP (rev.) • Bones cannot be rebuilt once they are thin from OP • If a person has OP, something as simple as lifting a bag of
groceries can break a bone • The health problems caused by OP can be life-threatening
0 1 1 0
1 1 0
10
0 0 0 0
Perceived benefits of OP pharmacotherapy • Drug treatments can help build strong bones • You would feel good about taking drug treatments to treat OP • Drug treatments can cut the chances of broken bones • You would consider taking drug treatments to prevent broken
bones • If your doctor advised you to, you would take drug treatments
to prevent broken bones
3 3 3 2 6
1 0 2 0 1
0 0 0 0 0
Perception of bones as thin 0 0 0
Perceived barriers to OP pharmacotherapy • You are taking too many medications • You have a stomach problem that limit your ability to take
drug treatment
1 1
0 0
0 0
38
Overall, the manual inspection of data for data flaws did not reveal any patterns of errors that
would be a cause of concern: there were no undocumented codes, out of range values and
missing observations were few.
Assessment of the overall data quality: The first part of the critique consisted of evaluating the
database purpose, content, sampling, comparison of the sample to the target population and data
collection methods. The purpose and content of the database were concluded to be a good fit
with the research question for two reasons. First, the purpose of the database was to collect data
in order to help improve the current OP care in the province and the purpose of the research
question was to identify factors which could also be used to improve current OP interventions.
Second, even though the secondary use of already collected data prevented exploration of
additional predictors, most variables of interest were present. The sampling strategy and
comparison of the sample to the target population revealed that the two were similar on all,
except for three variables where slight differences were found. Data collection procedures and
ongoing quality assurance were put in place to monitor data quality across sites and rectify any
issues. The last two parts of the critique showed that the data were of high quality with no
structural inconsistencies and a few missing observations. The quality of the data was evaluated
on a regular basis, as part of quality assurance: if any issues were found, steps were immediately
taken to rectify them. Overall, the provincial database used in this thesis was a unique, valuable
source of information that provided rich, high-quality data that I was able to use to answer the
research question appropriately.
3.4. Chapter summary Data used to address the second objective of this thesis, the OP-fracture link, were derived from
the database created as part of the Fracture Clinic Screening Program, MOHLTC-commissioned
province-wide initiative implemented by Osteoporosis Canada. In order to establish the
suitability of the database to answer the research question, the database purpose, content,
structure and quality were examined and a manual examination of raw samples of data was
performed. The results showed that the database was a fitting source of information for the
research question to be answered. Although some limitations were noted (inability to explore
additional predictors, missing data for those not followed up with, possibility of systematic
39
errors), the strengths and opportunities afforded by this database exceed its limitations. The
database remained the most suitable source of data for my research question.
40
Chapter 4 Factors Associated with Patients Making the Link Between Their
Fragility Fracture and Osteoporosis
4.1 Introduction OP is an asymptomatic disease which often only presents clinically after a fragility fracture
occurs (Brown et al., 2006). In addition to indicating a possibility of compromised bone health,
fragility fractures are strong predictors of future fractures (Klotzbuecher et al., 2000). Many
fragility fracture patients, however, do not perceive themselves to be at an increased risk for
future fractures or that their fracture may be linked to OP (Giangregorio et al., 2008; Meadows et
Gerend, Erchull, Aiken, & Maner, 2006; Fiandt, Pullen, & Walker, 1999). This disconnect is
concerning because fragility fracture patients who do not believe that they are at risk for OP or
future fractures are not likely to perceive OP management as applicable or important to them and
are therefore not likely to engage in it (Giangregorio et al., 2009; Meadows et al., 2005). In
interviews with female fragility fracture patients, Meadows and colleagues (2003, 2004, 2005)
found that patients who did not make the OP-fracture link took little or no action to prevent
another fracture. These patients were not motivated to change their behaviour and reduce risk
because they perceived their fracture as an isolated incident, attributable to external factors
(Meadows et al., 2005; Meadows et al., 2004; Meadows & Mrkonjic, 2003). In our previous
research that used structural equation modeling, we found that perceived need (perception of the
OP-fracture link and perception of bones as “thin”) was the best predictor of initiating first-line
OP treatment (Beaton et al., 2009). Our qualitative research also pointed to the importance of
patients making the OP-fracture link (Beaton et al., 2007): focus group participants described the
link as the pivotal step towards initiating OP management (testing and treatment as appropriate).
Similar to making the OP-fracture link, the concept of perceived susceptibility to OP has been
associated with the acceptance of OP therapy (Cline et al., 2005; Cadarette, Gignac, Jaglal,
Beaton, & Hawker, 2009) and has also been shown to be a predictor of follow up with a
physician for investigation of OP (Bliuc et al., 2006).
41
Previous research examining factors that were associated with patients’ perception of
susceptibility to OP and the OP-fracture link was mostly cross-sectional or qualitative in nature
(Meadows et al., 2005; Meadows et al., 2004; Meadows & Mrkonjic, 2003; Giangregorio et al.,
2008; Giangregorio et al., 2009), making the direction of the association unclear. One such cross-
sectional study involving fracture patients found that OP diagnosis was associated with patients
making the OP-fracture link (Giangregorio et al., 2008), while another study involving
community-dwelling women, found that the use of antiresorptive medication was associated with
high perceived susceptibility to OP (Cline et al., 2005). Other factors found to be associated with
higher perceived susceptibility to OP include younger age (Giangregorio et al., 2008), female
gender (Doheny, Sedlak, Estok, & Zeller, 2007), family history of OP (Gerend et al., 2006) and
more OP knowledge (Doheny et al., 2007). Apart from being cross-sectional, previous studies
included patients who were already diagnosed and/or treated for OP (Cline et al., 2005;
Giangregorio et al., 2008), thus making it more difficult to determine which came first: OP
diagnosis/treatment or the perception of the link.
Although previous research has identified some factors associated with fragility fracture patients’
perception of the OP-fracture link and their susceptibility to OP, its cross-sectional nature makes
it impossible to predict which undiagnosed, untreated patients are more likely to make this link
after sustaining a fragility fracture. The objective of this study is to fill this gap in knowledge by
identifying baseline predictors that characterize undiagnosed and untreated patients who make
the OP-link at follow up. Identification of factors predictive of (rather than associated with)
making the OP-fracture link can help identify specific groups of patients at the time of initial OP
screening, when intervention can be delivered. Patients identified as less likely to make the OP-
fracture link might require a different, more intense intervention to make this link. The ability to
predict which patients will (and will not) associate their fracture with OP would therefore: (a)
help identify the patients who need a modified intervention and (b) help create an appropriate
intervention tailoring it to the patients less likely to make the link.
Since only two studies examined the OP-fracture link (Giangregorio et al., 2008; Meadows et
al., 2005), most predictors were based on research examining a related concept, patients’
perceived susceptibility to OP and future fractures (Doheny et al., 2007; Gerend et al., 2006).
Other predictors were drawn from our previous research on the pathway to OP care (Beaton et
al., 2007). Based on the previous research, we hypothesized that patients more likely to make the
42
link at follow up would be: younger, female, those who sustained a more complex fracture (hip,
shoulder or multiple fractures), those with a previous fragility fracture after the age of 40, as well
as maternal fracture (also after the age of 40). We also hypothesized that the following
modifiable factors would be associated with making the OP-fracture link: perception of bones as
“thin”, having more knowledge of OP and perceiving more benefits and fewer barriers to OP
treatment.
4.2 Methods Sample: The sample for this study was drawn from the database that was created as part of a
provincial screening program for OP. At the time data were collected, there were 19 screening
coordinators placed in 36 designated fracture clinics across the province in order to identify,
assess, refer and educate fragility fracture5 patients over the age of 50 (or younger if referred by
a surgeon). Screening coordinators collected self-reported patient data at baseline and follow up.
The following data were collected at baseline: OP risk factors, OP diagnosis, bone mineral
density (BMD) testing and OP treatment history, demographics, documentation of
intervention/patient consent and patient’s OP knowledge and beliefs. Patients who were not
diagnosed or treated for OP at baseline were contacted at three or, if necessary six months6 and
asked about post-fracture follow up with their family physician and new testing, diagnosis and
treatment information in addition to the same OP knowledge/beliefs questions that were asked at
baseline.
The study sample included consenting patients who were not diagnosed or treated for OP, whose
baseline data were completed by September 30, 2009 and who were followed up by December
31, 2009. For the purpose of answering the research question, the sample was further restricted to
those patients who, at baseline, did not think that their fracture was related to OP. This study is a
part of a thesis project which received ethics approval from St. Michael’s Hospital and the
University of Toronto research ethics boards (Appendix IV). The research database from which
5 Fragility fracture was defined as a fracture that resulted from a fall from standing height or less 6 Patients who have not followed up with their GP by three months post-fracture were contacted again at six months
43
the study sample was drawn was approved by the research ethics boards of each participating
site.
Measures: The study’s outcome variable was the OP-fracture link as measured by a “yes”
response to the following question “Do you think your broken bone could have been caused by
having OP (thin or brittle bones)?” Contrasting responses (“no” and “do not know”) were
collapsed into one category “did not make the link”. This was done because the purpose of the
research question was to distinguish only between those aware of the OP-fracture link (“yes”)
and those unaware of this link (“not yes”). Also, the sample only included those patients who
answered the OP-fracture link question at both baseline and follow up.
The variables chosen as potential predictors are shown in figure 4.1 and listed below:
1. Patient’s age at the time of screening: <65 vs. ≥65. Patients’ age was dichotomized into
these categories because patients over the age of 65 are at a higher risk for OP (Brown &
Josse, 2002; Brown et al., 2006) and because the division into <65 vs. ≥65 provided us
with two groups with similar number of observations: n=415 and n=438 (65 was the
median).
2. Gender: male vs. female. Gender was chosen as a predictor because previous research has
not addressed the question of whether there was a gender difference in who is more likely
to make the link.
3. Fracture site: shoulder, hip and multiple vs. single fractures. Our previous qualitative
research showed that patients who made the OP-fracture link also reported having a more
complicated fracture, such as a hip or a shoulder fracture (Beaton et al., 2007). We
therefore divided all patients into two groups: those who had a shoulder, hip or a multiple
fracture and those who sustained a single fracture.
4. Previous fragility fracture after the age of 40: yes vs. no (as measured by a yes/no answer
to the question “Has the patient ever broken any other bones since the age of 40 from a
simple trip and fall?”). Previous fracture was associated with awareness of bone health in
previous research (Giangregorio et al., 2008).
44
5. Maternal history of fractures: yes vs. no (as measured by a yes/no answer to the question
“Did this patient’s biological mother have any fractures after the age of 40?”). A 2006
study found that women who rated their risk of developing OP as high attributed this risk
primarily to family history (Gerend et al., 2006).
6. Perception of bone quality: thin vs. do not know vs. normal. Perception of one’s bone
health was measured by patient’s agreement with the statement “My bones may be
thin/my bones are definitely thin”. This variable was chosen because previous research
pointed to the lack of awareness of poor bone health in most fragility fracture patients
(Meadows et al., 2005; Meadows et al., 2004; Meadows & Mrkonjic, 2003).
7. Knowledge of OP. OP knowledge was associated with perceived susceptibility to OP in
previous research (Doheny et al., 2007). We measured the OP knowledge using four
items from the “OP and You” knowledge questionnaire (Brenneman et al., 2002): “There
is no way to prevent OP”, “Bones cannot be rebuilt once they are thin from OP”, “If a
person has OP, something as simple as lifting a bag of groceries can break a bone” and
Outcome: “My fracture
may have been caused by OP”
Initiation of OP
management
OP therapy benefits outweigh
barriers
Quality of bones -
thin
Previous fracture
Younger age
Fracture site: more painful
fracture
Maternal fracture
Female gender
Figure 4.1 Hypothesized predictors of making the osteoporosis-fracture link by follow up
More OP knowledge
45
8. Perceptions of benefits and barriers to OP pharmacotherapy. These variables were
included as predictors because our previous qualitative research found that patients’
health beliefs played a role in making the OP-fracture link. Perceived benefits of OP
pharmacotherapy were measured by a five-item OP benefits scale while the perception of
barriers to OP pharmacotherapy was assessed by two individual items. Both the
perceived OP benefits scale and the items measuring OP barriers were derived from
previous work by Cadarette (Cadarette et al., 2004) who based her scale and items on the
Osteoporosis Health Belief Scale (Kim et al., 1991). The number of perceived benefits
was measured on a scale 1-5 while barriers were assessed individually (yes/no to each of
the two questions).
Analysis: The sample was first described on a univariate level. Cronbach’s alpha coefficients of
the OP knowledge, OP pharmacotherapy benefits and OP pharmacotherapy barriers questions
were examined. The sample was compared to the population from which it was drawn to
examine any potential systematic sample biases. Distribution of the outcome variable (OP-
fracture link at follow up) was assessed and dichotomized into two responses: making the OP-
fracture link (“yes”) and remaining unaware of this link (“not yes”). Bivariate (unadjusted)
analysis was used to examine the relationship between each predictor and the outcome (chi
square tests for categorical and 2-tailed t tests for continuous variables). Multivariable logistic
regression was then used to determine which baseline factors were associated with the OP-
fracture link at follow up while controlling for other predictors. As per Hosmer and Lemeshow
(Hosmer & Lemeshow, 2000) only predictors with a p-value of <0.25 (based on unadjusted
analysis) were used in model building. Multicolinearity was assessed by examining the tolerance
and variance inflation factors of each variable to determine whether or not any of the predictor
variables were correlated. Tolerance levels below 0.4 and variance inflation values greater than
46
2.57 were to be taken as an indication that predictors were highly correlated8 (Allison, 1999).
Manual backwards elimination was then used to reduce the number of predictors in the final
model (Harrell, 2001): it was decided a-priori that a non-significant predictor would be kept in
the model if it changed the estimate of other predictors by 20% when taken out. Manual
backwards elimination allowed us to closely monitor the impact of the removal of each variable
on the model as a whole and on other predictors. The maximum number of predictors in the final
model was limited to one per ten events in the smallest category. Goodness of fit was assessed by
Hosmer and Lemeshow’s Goodness-of-Fit Test and accuracy of the model as a predictor of
outcomes by concordance statistic (Hosmer & Lemeshow, 2000). All analyses were performed
using SAS Version 9 (Statistical Analysis System, Cary, N. Carolina). The criterion for
interpreting statistical significance at the final stage was set at alpha ≤ 0.05.
4.3 Results Psychometric analysis: Cronbach’s alpha coefficient for the two OP knowledge domains that
the OP knowledge questions tapped into confirmed previous findings (Cadarette et al., 2007) of
low internal consistency of the two domains (Cronbach’s alpha coefficients of 0.43 and 0.24).
We therefore analysed the four OP knowledge questions as separate items, dichotomized into
“correct” and “incorrect”. Cronbach’s alpha coefficient for questions tapping into the patient
perception of OP benefits and barriers showed that the internal consistency for the OP benefits
scale was acceptable for group level analyses such as ours (Cronbach’s alpha coefficient of 0.76)
but was low for the OP barriers domain (Cronbach’s alpha coefficient of 0.19). Based on this
analysis and previous recommendations (Cadarette et al., 2004), the perception of OP
pharmacotherapy benefits was analysed as a scale, while the two barriers items were analysed
separately as a dichotomy of correct and incorrect answers.
7 High inflation value indicates that the variance of a regression coefficient is increased because of collinearity; low tolerance value also indicates that the variable under consideration is highly correlated with another variable already in the model. 8 In case of multicollinearity, a less important and less modifiable variable would be taken out of the model.
47
Analyses examining OP-fracture link: Of 916 patients who had data available on the OP-
fracture link question (at both baseline and follow up), 93% (n=853) did not make this link at
baseline. The distribution at follow up was similar: of those patients who did not make the link at
baseline, 91.2% (n=778) still did not make the link or remained unsure about it (Figure 4.2).
Figure 4.2 Distribution at follow up of those who did not make the OP-fx link at baseline
At baseline: 853 (93% of 916) patients did not make a link between OP
and their fracture (or were unsure about it)
At follow up:
75 (8.8%) patients made the link by follow up
At follow up:
778 (91.2%) patients still not making the link
In the sample used for the analysis (patients who did not make the link at baseline, n=853), the
majority of patients were female (80%) with the mean age of 67 (SD =10.5, range 50-94).
Twenty-five percent of the patients sustained a hip, shoulder or a multiple fracture while 28%
reported a history of maternal fractures (after the age of 40). The baseline fracture was not the
first fragility fracture after the age of 40 for 26% of the patients. This sample was compared to
the population it was drawn from on variables of interest: potential confounders, predictors and
the outcome. The results are shown in Table 3.1
As compared to the target population, the sample consisted of fewer patients who were employed
(30.1% vs. 35.4%), a greater number of patients who perceived more benefits of OP drug
treatment (4.2 of 5 and 4 of 5), more oral steroid users (6.3% vs. 4.3%) and more patients who
correctly answered the knowledge question stating “If a person has OP, something as simple as
lifting a bag of groceries can break a bone” (84% vs. 78%). Although statistically significant, the
differences in perceived treatment benefits or steroid use were not substantial: both groups
perceived four of five benefits on average and there was a 2% of difference in steroid use
between the two groups. The sample and the population did not differ on other variables of
interest. Specifically, there were no statistically important differences in age, gender, marital
status, living situation, type of fracture, existence of previous or maternal history of fractures,
48
previous falls, feeling unsteady, perception of the bone quality, the remaining knowledge items
(three of four) and perceived barriers to OP drug treatment.
Table 3.1 Description and comparison of the sample n=853 to the target population n=2237 Variables n=853 n=2237 p value
Mean age (SD) range 66.6 (10.5) 50-94
66.6 (10.8) 50-95 0.97
Gender – Female
79.7%
76.8%
0.08
Marital status: Married/common law
59.7%
62.5%
0.34
Living situation: Live alone
28.7%
26.8%
0.31
Work full or part time
30.1%
35.4%
0.0054
Hip or shoulder fracture (may include others)
25.6%
24.5%
0.56
Previous fracture >age 40
25.2%
25%
0.94
Maternal fracture >age 40
28%
25%
0.09
Previous fall in the last year
26.6%
27.2%
0.73
Feeling unsteady
27%
30.2%
0.08
Current or past steroid use
6.33%
4.31%
0.02
Perceived benefits of OP drug treatment (1-5)
4.2
4
0.0032
% Who perceived bones as thin
10.6%
10%
0.11
Knowledge item: There is a way to prevent OP (reversed)
64.2%
67.7%
0.07
Knowledge item: Bones can be rebuilt once they are thin from OP (reversed)
57.1%
55.5%
0.41
Knowledge item: If a person has OP, lifting a bag of groceries can break a bone
84.3%
78.7%
0.0006
Knowledge item: Health problems caused by OP can be life-threatening
69.2%
71.1%
0.33
Perceived barriers item: Taking too many medications
22.2%
20.8%
0.38
Perceived barriers item: Stomach problem limits ability to take drug treatment
17.5%
20.3%
0.1
49
Logistic regression was used to examine which of the hypothesized baseline factors were
associated with making the OP-fracture link at follow up. Of the 13 hypothesized factors
(including four knowledge items and two barriers), nine predictors had a p < 0.25 based on the
unadjusted analysis. These factors were entered into the adjusted (multivariable) analysis:
manual backwards elimination was used to remove the least significant variable sequentially to
narrow down the number of predictors, while examining the change in estimates in the remaining
variables. The final model contained six predictors (maximum number of predictors for the
number of events). P values associated with Type I error of 0.05 (p<0.05) were considered
statistically significant. The results of the unadjusted and adjusted analyses are presented in
Table 4.1.
Table 4.1 Regression of baseline factors most likely to be associated with a transition to making the link
Baseline characteristics: Unadjusted OR (95% CI) p value Adjusted OR
(95% CI) p value
Age < 65 vs. ≥ 65 1.5 (0.91-2.4) 0.12 dropped* -
Gender: female vs. male 1.96 (0.95-4.0) 0.06 1.8(0.8-4.1) 0.18
Previous fracture > age 40 yes vs. no 2.6 (1.6-4.2) <.0001 2.2 (1.3-3.9) 0.004
Maternal fracture >age 40 yes vs. no 1.4 (0.9-2.3) 0.18 dropped* -
OP knowledge item: Bones can be rebuilt once they are thin from OP (correct vs. incorrect)
1.43 (0.9-2.4)
0.16 1.5 (0.8-2.9) 0.2
Perceived OP pharmacotherapy benefits (yes vs. no)
1.3 (1.0-1.8) 0.03 1.2 (0.9-1.6) 0.13
Perceived barrier to OP pharmacotherapy Patient cannot take more drugs due to stomach problems (agree vs. disagree)
0.6 (0.3-1.3) 0.2 1.6 (0.8-3.5) 0.2
Perception of bone quality: • Do not know what the bone quality is • Bones thin or thinning • Normal or normal for age
3.4 (1.9-6.1) 9.8 (5.3-18.0)
ref
<.0001 <.0001
-
2.8(1.5-5.4) 7.1(3.7-13.8)
ref
0.0016 <.0001
-
*Dropped = dropped during the adjusted analysis
50
In the unadjusted analysis, a previous fragility fracture, high perceived benefits of OP
pharmacotherapy and perception of bones as thin or ambiguity about the bone quality were all
associated individually with making the OP-link at follow up. Perception of benefits was not
significant in the adjusted analysis which showed that, controlling for other factors, the odds of
making the link at follow up were: 2.2 times higher for those who sustained a fracture prior to
baseline fracture, 2.8 times higher for those who were unsure about the quality of their bones at
baseline (vs. those who thought bones were normal) and 7.1 times higher for those who believed
at baseline that their bones were thin (vs. those who thought the bones were normal). This model
showed good fit with the data: the Hosmer and Lemeshow Goodness-of-Fit Test resulted in a p
value of 0.63, while the model accuracy was confirmed by the concordance statistic of 0.76.
4.4 Discussion The results of this study indicate that most fragility fracture patients do not associate their
fracture with OP, even after a post-fracture intervention. This study identified the baseline
characteristics predictive of making the OP-fracture link at follow up: those who became aware
of the OP-fracture link by follow up were more likely to have had a previous fracture, perceive
their bones as thin or be unsure about their bone quality.
The findings from our study confirm the results of previous research on the OP-fracture link
which showed that most fragility fracture patients do not link their fracture with underlying OP
(Giangregorio et al., 2008). A novel finding from our study is that, for most patients, this
perception does not change over time: 91% of patients who did not make the link at baseline do
not make this link at follow up either. This is concerning because a majority of fragility fracture
patients in the orthopaedic environment have a low BMD (Astrand et al., 2008) and are in need
of OP management (Lindsay et al., 2001; Astrand et al., 2006a; Center et al., 2007) but they are
not likely to initiate if they do not find OP management to be applicable to them (Giangregorio et
al., 2008; Meadows et al., 2005).
Only two of the 13 hypothesized baseline factors were found to be predictive of the OP-fracture
link at follow up. Patient perception of their bones as “thin” and a previous fragility fracture
were mentioned in previous research as being correlated with perceived susceptibility to OP and
51
future fractures. Meadows (2005) found that the only group of fragility fracture patients that took
action towards managing their OP was the group that perceived their bone health as
compromised and themselves at risk for future fractures. In addition to the perception of thin
bones, previous research found that having more than one fragility fracture was related to the
perception of susceptibility to OP (Giangregorio et al., 2009): this study found that having one
fracture was not necessarily associated with the awareness of bone health but more than one was.
Having more than one fragility fracture was found to be associated with the OP-fracture link in
our previous focus group research; however, not everyone associated their fractures with thin
bones: one patient thought she was just “clumsy” and was repeatedly fracturing due to external
circumstances, not compromised bone health (Beaton et al., 2007).
Surprisingly, factors such as gender, age, OP knowledge and perceived benefits of OP
pharmacotherapy were not found to be predictive of the OP-fracture link at follow up. Women
were no more likely than men to make the OP-fracture link, a finding that was surprising in light
of previous cross-sectional research on perceived susceptibility to OP that identified women as
more likely to make the link (Doheny et al., 2007). This finding also does not reflect the fact that
women are at a higher risk of fragility fractures and OP (Kanis & Johnell, 2005; Tenenhouse et
al., 2000). In previous research examining the association between age and future fracture risk
(Giangregorio et al., 2008), decreasing age was found to be related to the perception of future
fracture risk which is particularly concerning because increasing age is actually an independent
predictor of subsequent fracture (Kanis et al., 2004). Our study demonstrated that age was not a
significant predictor of making the OP-fracture link between baseline and follow up, however,
we did not assess whether age was a predictor of perception of future fracture risk. The findings
of our study indicate that gender and age may only be correlated with concurrent perception of
OP-fracture link and/or susceptibility to OP but are not predictive of becoming aware of this link.
OP knowledge at baseline (knowledge of OP treatments and consequences of OP) was also not
related to making the OP-fracture link at follow up. This finding was unexpected given that
previous research found that those who knew more about OP also thought they were more
susceptible to it (Doheny et al., 2007). Previous research using structural equation modelling
(Beaton et al., 2009) also showed that baseline knowledge of OP was indirectly associated with
making the OP-fracture link at follow up. The analysis showed that previous OP knowledge,
along with the perception of OP treatment benefits, was strongly associated with OP knowledge
52
(and perception of treatment benefits) at follow up which, in turn, was associated with a
combined variable called “perceived need” consisting of the perception of thin bones and the
OP-fracture link at follow up (our OP-fracture link analysis, however, kept the two variables
separate). It is also possible that baseline knowledge alone may not predictive of the OP-fracture
link but that, in conjunction with OP benefits, it may indirectly impact making of the link at
follow up.
The current study is unique because it is a first cohort study that examined predictors of the OP-
fracture link in fragility fracture patients who were not diagnosed or treated for OP and who also
were unaware of the OP-fracture link at baseline. Previous research, due to its cross-sectional
nature, was only able to identify factors that were associated with the link and not predictive of
it. By determining which factors were predictive of OP-fracture link, our study also established
which factors were not predictive of the link, an equally important finding. An important strength
of the study rests with the provincial data that the study was based on: having data from multiple
sites across the province ensured that our sample was representative of the fragility fracture
population to which the results were to be generalized. Lastly, limiting our sample to
undiagnosed and untreated patients allowed us to eliminate the possibility of patients making the
OP-fracture link because of their diagnosis/treatment recommendations. Associating OP and
fractures is important because OP is an asymptomatic disease: patients who are unaware of their
risk factors are unlikely to take any action to prevent bone disease or to prevent already
compromised bone health from further deteriorating.
The main limitation of our study is the reliance on data which were not specifically collected for
the purpose of answering the research question on the OP-fracture link. We were therefore
unable to include and examine more potential predictors of interest or to base data collection on a
specific model of determinants of health behaviour change. Another limitation associated with
the database is a possibility of sampling bias as our sample and the target population were
different on three variables.
The findings of this study could be incorporated in post-fracture OP interventions to better
communicate to patients their risk of future fractures and OP. Patients who, at baseline, did not
have concerns about their bone quality and those who did not have a previous fracture are at a
particular risk for not making the OP-fracture connection. Future interventions could target these
53
patients (who did not make the link at baseline) emphasizing the link between a previous fracture
and OP and delivering a clear message about the quality of patient’s bones as “thin”. The same
message should be reinforced in patients who have experienced a previous fracture and who
already have a perception of their bones as “thin” in order to link the two with the need for
initiating OP management. Future post-fracture interventions should therefore support the
awareness in the patient group likely to make the OP-fracture link and use a different approach in
those not fitting this profile. Future research should explore the impact of such interventions on
acceptance and adherence to recommended therapies to reduce fracture risk.
54
Chapter 5 Discussion and Conclusion
5.1 Discussion The objectives of this thesis were twofold: to explore the extent to which fragility fracture
patients’ perception of their bone health has been considered in existing post-fracture OP
interventions and to examine fragility fracture patients’ awareness of their bone health in relation
to the fracture they sustained. The ultimate goal of the thesis was to inform the creation of more
effective, patient-centered OP interventions that are based on patients’ perception of their bone
health.
This thesis focused on the patient and patient-related barriers to more effective OP management
because it is ultimately the patient who needs to implement and adhere to the recommendations
and prescribed therapy. For OP management to take place, patients need to make changes to their
existing behaviour (e.g., increase their Calcium intake) or adopt a new behaviour altogether (e.g.,
initiate bone sparing medication). Whether such behaviour change will occur largely depends on
patients’ perception of the state of their health and perceived health risks: it is known that
patients’ perception of their health has a large influence on health-related behaviours including
treatment decisions (Gold et al., 2006; Deeks, Zoungas, & Teede, 2008). Patients’ perception of
their health and the process of behaviour change are concepts central to individual-level theories
of health behaviour change. This focus on patients’ perception distinguishes theories of health
behaviour change from educational interventions that provide generic information to all patients
without taking into consideration patients’ perception of the issue at hand. Application of
theories of health behaviour change takes into account the differences in patients’ perceptions
and strives to adapt the health information to each patient. These theories have been applied to a
variety of contexts that involve stopping, initiating or maintaining specific behaviours such as
prevention in fragility fracture patients: the impact of a simple clinical information procedure.
Journal of Applied Research.2004; 4, 570-575.
127. Simonelli, C., Killeen, K., Mehle, S., & Swanson, L. (2002). Barriers to
osteoporosis identification and treatment among primary care physicians and orthopedic
surgeons. Mayo Clin.Proc., 77, 334-338.
128. Skedros, J. G. (2004). The orthopaedic surgeon's role in diagnosing and treating
patients with osteoporotic fractures: standing discharge orders may be the solution for timely
medical care. Osteoporos.Int., 15, 405-410.
129. Tenenhouse, A., Joseph, L., Kreiger, N., Poliquin, S., Murray, T. M., Blondeau,
L. et al. (2000). Estimation of the prevalence of low bone density in Canadian women and men
using a population-specific DXA reference standard: the Canadian Multicentre Osteoporosis
Study (CaMos). Osteoporos.Int., 11, 897-904.
130. Tosi, L. L., Gliklich, R., Kannan, K., & Koval, K. J. (2008). The American
Orthopaedic Association's "own the bone" initiative to prevent secondary fractures. J Bone Joint
Surg.Am., 90, 163-173.
131. Tugwell, P. S., Santesso, N. A., O'Connor, A. M., & Wilson, A. J. (2007).
Knowledge translation for effective consumers. Phys.Ther., 87, 1728-1738.
132. U.S.Department of Health and Human Services (2004). The 2004 Surgeon
General's Report on Bone Health and Osteoporosis: What It Means To You U.S. Department of
Health and Human Services, Office of the Surgeon General.
133. Vaile, J., Sullivan, L., Bennett, C., & Bleasel, J. (2007). First Fracture Project:
addressing the osteoporosis care gap. Intern.Med.J., 37, 717-720.
134. van Staa, T. P., Dennison, E. M., Leufkens, H. G., & Cooper, C. (2001).
Epidemiology of fractures in England and Wales. Bone, 29, 517-522.
135. Wallace, L. S. (2002). Osteoporosis prevention in college women: application of
the expanded health belief model. Am.J.Health Behav., 26, 163-172.
79
136. Wichowski, H. C. & Kubsch, S. M. (1997). The relationship of self-perception of
illness and compliance with health care regimens. J.Adv.Nurs., 25, 548-553.
137. Wiktorowicz, M. E., Goeree, R., Papaioannou, A., Adachi, J. D., &
Papadimitropoulos, E. (2001). Economic implications of hip fracture: health service use,
institutional care and cost in Canada. Osteoporos.Int., 12, 271-278.
80
Appendix I
Description of the studies included in the literature review of patient-centered post-fracture osteoporosis interventions in the
context of theories of behaviour change
First author year
Study design or program type
Fracture type and patients’ age
Theoretical framework specified
Elements of intervention
Ashe 2004
Controlled trial
Low trauma wrist fx 50+
No Usual care (BMD offered, specialist referral, discussion of meds, exercise prescription and lifestyle modification) vs. intervention (in addition to usual care): letter to patient explaining that she or he has suffered a low trauma fx diagnostic of OP, letter for GP for patient to take, FU telephone call to patient to remind of visit with GP, faxed letter to GP
Astrand 2006 Astrand 2008
Screening program in orthopaedic department
Fragility fxs 50-75
No, only premise that patients can be trusted to seek appropriate medical advice.
Patients given: one letter for GP, another one for themselves, a copy of the BMD and questionnaire (on risk factors) that they completed. Diagnosed patients referred to GP.
Bessette 2008
Randomized controlled trial/ ongoing provincial
Women 50+ post-fx
No All patients told about the link between fragility fx and another fx. Documentation group received written OP information emphasizing the link between
Description of the studies included in the review Abbreviations: PCP=primary care practitioner GP=general practitioner OP=osteoporosis fx(s)=fracture(s) tx=treatment BMD=bone mineral density FU=follow up
81
OP health management program
OP and fragility fx, patients’ GPs received written information. Intervention group received documentation group material and detailed video about link between OP and fx.
Bliuc 2006
Randomized trial
Low or minimal trauma patients
No Information-based interventions. Both control and intervention groups received personalized letter with participant's risk factors and recommendation for FU with a PCP. Intervention group also included an offer for a free BMD test.
Blonk 2007
Screening program: Fracture and OP clinic
Low trauma 50+
No BMD arranged, results explained, questionnaire including risk factors and life habits explained, diagnosis and tx strategy explained by a nurse
Bogoch 2006
Screening program: OP exemplary care program
Fragility fxs Women 40+ Men 50+
No Individual counselling assessing risk factors, questionnaire, BMD arranged, visit to a specialist booked, summary sent to GP
Che 2006
Pilot study of an intervention (disease-management) program
Osteoporotic fxs 55+
No Lab tests and BMD done. Patients received written information and counselling over the phone: fx risk score, link between OP and fx reviewed. Meds offered. Summary sent to both PCP and patient.
Chevalley 2002
Screening program, OP Clinical pathway
Low trauma any age
No Interactive educational program for patients and families on the management of OP and lifestyle, physical therapy and nutrition (done by multidisciplinary team). BMD and tx suggestions sent to PCP and orthopaedic surgeon.
Cuddihy 2004
Clinical practice intervention
Moderate impact distal forearm fx 45+
No Patients provided with written information and filled out OP quality of life questionnaire. BMD test booked. Appointment
82
with GP facilitated. Davis 2007
Randomized controlled trial: Patient Empowerment and Physician Alert
Hip fxs Women 60+
No. The study tested an intervention involving patient empowerment
Intervention group: patients given written information, asked to see a GP and take a letter to GP.
Edwards 2005
Screening program: OP and fx intervention program
Minimal trauma fx 40+
No Patients given printed material on OP, diet, home safety and educational consult, lifestyle modification, BMD recommended, tx initiated if needed (recommendations mailed to PCP at first, then hospital resources mobilized).
Feldstein 2006
Randomized controlled trial
Women 50-89
No Usual care or PCP reminder or PCP reminder plus education (patient were mailed written letter with OP information about their risk, guidelines for management and general OP information. Physicians received the same material and an electronic record reminder).
Feldstein 2007
Outreach program evaluated – retrospective cohort
Women 67+
No Patients received counselling regarding their risk for OP, fx and lifestyle. BMD results explained and meds initiated. PCPs sent an electronic record reminder with guidelines and suggested tx.
No Patients received an educational consult with a nurse, BMD arranged, tx recommended if needed. Assessment letter sent to GP.
Gardner 2005
Randomized controlled
Low trauma hospitalized
No Control group: given a 2 page pamphlet on fall prevention, not
83
trial hip fx patients
going into detail about OP. Contacted after 6 months to see if OP was addressed by DXA or tx. Study group: coordinator discussed OP, future fx, available tx. Patients given 5 questions to ask their GP. Contacted after 6 weeks to remind about importance of FU and management of OP.
Harrington 2005 Harrington 2007
Clinical improvement project: Osteoporosis Care Service
Fragility fxs 50+
No Explained the service to patients and family. Provided education and coordination of care, patient advised to contact PCP for OP management or referred to OP specialty clinic, patient complete a clinical questionnaire, assigned a nurse manager to answer questions and respond to problems.
Hawker 2003
Pre-post intervention design – pilot test of a simple intervention
Fragility fxs 40+
No Patients informed by orthopaedic that fx likely OP related and patients at risk for OP and fxs. Patients encouraged to FU with GP and given a standardized letter.
Ho 2006
Non- randomized prospective evaluation of a pilot intervention
Fragility fxs 45+
No Patients interviewed by pharmacy resident to identify barriers, patients also received an educational intervention (presentation on OP/falls/diet etc and written information), completed questionnaire on OP knowledge, encouraged to speak to physiotherapist and see a GP.
Johnson 2005
Prospective study - simple intervention
Fx patients Veterans (95% males)
No Patients were informed about the importance of diagnosing and treating OP, risk factors (also given written information). BMD test offered.. Report sent to PCP. Supplements recommended if needed.
84
Kuo 2007 Prospective cohort
Minimal trauma Adults
No Control group: information only. Intervention group: Face to face counselling re OP, patient-specific risk factors and lifestyle change. Results of BMD discussed, supplements and antiresorptive medications discussed and recommended (not prescribed). Information sent to GP and patients referred for BMD.
Laslett 2007
Clinical pathway
Admitted fragility fxs 45+
No Those patients recommended/started therapy also received individual counselling as well as written information about OP, letter to GP and resource information. No investigation of OP was undertaken – left to GP.
Levasseur 2007
Ortho department intervention
Low energy fx admitted to orthopaedic dept 50+
No During interview with rheumatologist, lifestyle changes and tx discussed. BMD performed. Letter to GP sent for those patients who were given a prescription at discharge.
Majumdar 2004
Non-randomized controlled trial
Wrist fx patients 50+
No Faxed physician reminders along with guidelines endorsed by local opinion leaders. Patient education: written message encouraging patients to seek further information from GP reinforced by a brief telephone counselling session.
Majumdar 2007
Randomized controlled trial
Hip fx patients 50+
No Intervention group consisted of usual care and counselling about the importance of BMD and tx in reducing the risk of another fx. BMD arranged and results discussed. Treatment facilitated. Summary sent to GP.
Majumdar 2008
Randomized controlled trial
Wrist fx patients 50+
No Control (pamphlet only) vs. Intervention: patients given the same pamphlet and counselling
85
reinforcing it. GP’s were mailed reminders and local opinion leader-endorsed OP tx guidelines.
Quintos-Macasa 2007
Retrospective cohort to examine rheumatology service
Low impact hip fxs
No Patients were prescribed medication after reviewing their records and scheduling a BMD if necessary. Rheumatology service communicated with PCPs.
Schmid 2004
Prospective intervention “OsteoCare project”
Low trauma 50-85 years old
No
Phase 1: Surgeons were asked to inform patients about OP possibly causing fx and give them written material including recommendation to see their GP for further evaluation of OP. Phase 2: GPs were informed about the project and asked to initiate diagnostic OP procedures.
Skedros 2004
Program aimed at surgeons (and indirectly at patients)
Low trauma 50+
No Ortho surgeons recruited to tell patients that they may have OP, were at risk of another fx and should see their GP. Ortho surgeons sent 2 letters to patients GP’s.
Tosi 2008 Evaluation of the “Own the Bone” tool /pilot intervention
Low energy fxs Any age
No Prompts to provide counselling to patients re OP, fall prevention, exercise and tx. Letter sent to GP.
Vaile 2007
“First Fracture Project” evaluation
Low trauma 50+
No, only that the intervention was aimed at patients with ‘fresh’ fx pain
Patients were educated about diet, exercise, risk factor reduction and falls prevention; BMD done, tx started if needed, letter sent to GP. Follow up at 1 month to encourage compliance with meds then a year after to repeat BMD and check for Vit D.
86
Appendix II
Fracture Clinic Screening Program’s patient educational materials and Program Information Letter
87
88
Appendix III
Construct list with description of all variables collected as part of the Fracture Clinic Screening Program
Data element (Number in
survey)
Field Description
Patient ID Numeric Provide link to de-identified information Facility ID Numeric Linked to patient ID DISCLAIMER 1 item Coordinator has read the disclaimer to the patient Health Card Number Numeric Only if the patient consents to his/her health card
number to be collected Agreed to health card number Yes/No Indicator that subject has agreed to link with health
data. Health card number is with identifiable data. Consent signed/stored Yes/No Indicator for central database that consent was
signed and is on site at clinic Gender Alpha Age Years Year of birth Numeric Last Name Alpha On contact sheet and consent form First Name Alpha On contact sheet and consent form Address On contact sheet Telephone Number(s) On contact sheet
Full date of birth Numeric Date completed Coordinator name
Marital Status 1 item 6 responses (choose 1)
Never married (single); Common Law; Married; Separated; Divorced; Widowed
Language 1 item 3 responses (choose 1)
English; French; Other (specify)
First 3 digits of postal code 1 item e.g. M5R
Date of fracture Numeric On baseline questionnaire Living Arrangement 1 question Where does this patient live?
Employment at time of fracture and 12 weeks post-fracture
1 item Yes/No
Baseline: At the time of this patient’s fracture, was he/she working full or part-time for pay? Follow up: Were you employed full time or part time at the time of your fracture?
89
Not employed at time of fracture: reason
1 item 7 responses (choose 1)
Homemaker; Unable to find work; Not working by choice; Volunteering; Retired; Student; Not working due to health conditions or disability.
Not employed at follow up (if employed at baseline)
1 item 2 responses
Not working due to fracture Other
Able to complete full or partial documentation? Yes/No If NO, why? Select one: Language; Mental/Physical barrier; Patient declined Other (specify).
Ability to complete Baseline Documentation
4 questions
Is there someone else able to complete partial documentation? Yes/No If YES, who: Patient; family member; friend; translator; chart
Fracture etiology (fragility fracture defined by WHO [51]
1 question 3 responses (choose 1)
Provided by patient, family member or surgeon Low trauma; Moderate/high trauma; Unknown
History previous fragility fracture
1 question Yes/No
Since you were 40, have you broken any other bones?
Trouble getting out of chair or problem with walking
1 question Yes/No
Do you have trouble getting out of a chair or feel unsteady when you walk?
Maternal fracture 1 question Yes/No
Did this patient’s biological mother ever have a hip fracture after the age of 40 years? Other bones?
History of falls 1 question Yes/No
Have you fallen in the past year besides the fall that may have led to your current fracture?
Weight Kilograms and pounds ______ lbs or _____ kgs
Steroid Use 1 question Have you ever taken steroids, such as prednisone, by mouth for more than 3 months at a time?
Alcohol Use 1 question On average, do you drink more than 2 alcoholic beverages a day?
Diagnosis of rheumatoid arthritis 1 question Has a physician specialist, such as rheumatologist,
ever told this patient that they have…? Did this physician prescribe medications…? Medications for
rheumatoid arthritis 2 questions What medications did their physician prescribe…?
90
Smoking 1 question: 3 answers
Does this patient currently smoke cigarettes? Yes, No, Refused Baseline: Prior to having this broken bone, was this patient ever told by a physician that they have OP or osteopenia? Diagnosis of
osteoporosis pre and post fracture by a physician
1 question Yes/No
Follow up: In the past 12 weeks has your family physician….? Predictive of treatment * not true diagnosis, but more perceived diagnosis
Physician management of osteoporosis
1 question 8 responses
Family physician (or general practitioner); Rheumatologist; Endocrinologist, Gynecologist; Internist; Orthopaedic Surgeon; No one, Other (specify)
Physician recommend supplements/ prescribe medication for osteoporosis/ osteopenia
1 question Yes/no
Did this patient’s physician recommend supplements or…?
Supplements recommended/ medications prescribed for osteoporosis
1 question Yes/no Patient currently taking medications prescribed?
Reasons for non-adherence with medications prescribed
1 question 10 responses
Medication adherence responses(follow up variable names in brackets): Side effects of the medications; Cost (too expensive); On too many medications already; Forgot to take them; I have decided not to take any medications; I have not made up my mind whether to take the medications or not; I have not started taking them yet I cannot take the medications because of: kidney disease, abnormalities of the esophagus, cannot swallow pills who; Other (specify).
Completion of bone mineral density (DXA) in past year
1 question Yes/No/ Don’t know
Knowledge of bone mineral density results (personal communication, J.
1 question 5 responses
BMD testing responses: (modified from Brown, 2006) Yes, the results show I have osteoporosis; Yes, the results show I have osteopenia;
91
Brown, 2006) Yes, the results show I have low bone density, but I do not know if I have osteopenia or osteoporosis; Yes, the results show I have normal bone density or high bone density; No, I do not know the results of my BMD test.
Osteoporosis Readiness for change (Mauck, 2002) [53]
1 question 8 responses
Responses: Never heard of OP; Aware, never thought seriously; Considered, but decided against; Currently considering treatment; Decided to use treatment but not yet started Recently started on treatment (modified response) <6 months; Been on treatment > 6 months; Refused
Patient opinion of bone quality (personal communication, G. Hawker, 2003)
1 question 5 responses (chose one)
OP Knowledge and Beliefs responses: My bone are normal; My bones are normal for someone my age; My bones may be thin; My bone are definitely thin (modified response); I do not know,
Did osteoporosis cause you to break your bone?
1 question Yes/ No/ Don’t know
(Personal communication, G. Hawker, 2003)
Osteoporosis Knowledge (Cadarette, 2006) [54]
4 questions Response: Strongly agree to Strongly disagree
Questions: There is no way to prevent osteoporosis Bones cannot be rebuilt once they are thin from osteoporosis If a person has osteoporosis, something as simple as lifting a bag of groceries can break a bone The health problems caused by osteoporosis can be life threatening
Osteoporosis drug treatment Caderette, 2006 [54]
7 questions Responses: Strongly agree to Strongly disagree
Questions: Drug treatments can help to build strong bones You would feel good about taking drug treatments to treat osteoporosis (modified from original) Drug treatments can cut down the chances of broken bones You would consider taking drug treatments to prevent broken bones If your doctor advised you to, you would take drug treatments to prevent broken bones You are taking too many medications You have a stomach problem that limit your ability to take drug treatment
Patient has a family physician?
2 questions Yes/No
Documentation of name, address, phone, fax of family physician Referral to College of Physicians and Surgeons if patient does not have a family physician
Recommendations from Osteoporosis Screening Coordinator
Check list
Standardized letter to family physician Discuss completing bone mineral density (DXA) test with family physician Discuss calcium and vitamin D supplements with pharmacist
92
Provide patient with information letter on program
Visit date Please specify the date of the visit Follow up with physician 1 question Did you see a physician for assessment of OP…?
Type of physician 1 question 9 responses What type of physician did you see?
Reasons for not following up with family physician for osteoporosis assessment
1 item 7 responses (choose 1)
Did not obtain a family physician My family physician has referred me to a specialist for assessment of bone health. Forgot/missed an appointment with my family physician Did not have time to see my family physician Appointment with family physician is coming up Did not want to follow up with family physician Other (please write):____________
OSC suggestion "I suggest you follow up with your family physician for further assessment…" Can I call you…"
DXA Testing Completed in past 12 or 24 weeks
Item Yes/No/Don’t Know
Bone mineral density (DXA) test completed (self-report)
Reasons for not completing a DXA test in 12 or 24 weeks
1 item 9 responses
Not recommended to have a bone mineral density test I do not have a family a family physician, therefore did not get a referral for this test I did not go to see my GP My physician said I did not need a bone mineral density test My appointment is coming up I missed/forgot my appointment I did not want to have a bone mineral density test Other (please write) Refused Did you miss work due to fracture? Time off work due
to fracture 2 questions How much time did you have to take off from your job (weeks)?
Quality of information and care
1 question What is your overall opinion of the quality of care…?
Information received on OP 2 questions In the past 12 weeks did you call…?
General comments General comments for the FU visit
93
Appendix IV
Research ethics approval from the University of Toronto and St. Michael’s hospital