Hospitalist Workload and Its Impact on Quality of …...Hospitalist Workload and Its Impact on Quality of Care and Patient Safety Henry Michtalik, MD, MPH An Intensive Intro to Clinical

Hospitalist Workload and Its Impact on Quality of Care and Patient Safety

Henry Michtalik, MD, MPHAn Intensive Intro to Clinical Research July 22, 2011

Outline

1. Background/Context2. Objective and Hypotheses 3. Who to study?4. What to study?5. How to study?6. Why should you care?

Background Physician workload is an integral part of many

compensation systems ◦ Has implications for workstaff planning, comparing physicians, and

public policy.

Patient safety is an integral part of quality healthcare ◦ Has implications for the delivery, benchmarking, and reimbursement

of medical care.

These two areas of healthcare research are often explored independently, with limited focus on the interaction of one area upon the other.

Historically, nursing-patient ratios and resident-physician workhours have explored this interaction.

The Attending Perspective?

http://nathanbond.wordpress.com/2008/12/28/the-elephant-in-the-room/

Context Ratios and staffing plans have been studied and established for

nurses; ACGME workhour rules are in effect. Hospitalists now account for nearly 40%, and in some regions

up to 70%, of inpatient claims for general internist services. Hospitalists provide a unique venue to study the effect of

physician workload on patient safety and quality of care measures. ◦ Breadth and integration of their services◦ Focus on quality of care training◦ Specialization in inpatient medicine

Johns Hopkins Clinical Research Network◦ “The JHCRN is a unique research resource that increases patient access

to innovative therapies and outcomes research in their own local communities. It also empowers physicians to design and conduct a broad array of research projects relevant to their communities.”

Johns Hopkins Clinical Research Network

Objectives To study Hospitalists from the seven hospitals within the Johns Hopkins

Clinical Research Network to assess the typical patient:physician ratio and examine and describe its variability. ◦ There will be minimal variation in the patient:physician ratios within each

hospital. The ratio between hospitals will vary based on the Hospitalist system structure.

To adjust this ratio for the significant patient, physician, and hospital level factors which affect the number of patients that a single attending physician may be responsible for and assess its impact on quality of care measures, including readmissions, healthcare acquired conditions and mortality. ◦ Higher patient:physician ratios will be associated with poorer quality of care

and safety measures, even after adjustment for patient, physician, and hospital level factors.

Ultimately, we seek to identify factors in attending physician workload which can be improved upon to make inpatient care a safer, more efficient, and higher quality care experience.

/ Hypotheses

PredictorsTable 1: Summary of Factors Affecting Patient:Physician Ratio

Level Variable Name Type

Patient

Average Age (Years) Continuous Race Categorical Gender Dichotomous Typical Insurance Status Categorical Average Acuity of Care Categorical Frequency of Readmissions Continuous

Physician

Age (Years) Continuous Race Categorical Gender Dichotomous Average Workday (Hrs) Continuous Assistance by Midlevels or Housestaff (%) Continuous Clinical Experience (Years) Continuous Annual Salary ($) Continuous Bonus ($) Continuous Physician Group Size Continuous Non-Direct Patient Care Responsibilities (%) Continuous

Hospital

Practice Area Categorical Practice Location Categorical Magnet Status Dichotomous System to Deal With Increased Patient Volumes Categorical

Outcome: Percent Compliance with JCCore Measures

Table 2A: Selected Joint Commission Quality of Care Measures Applicable to Hospitalists Category Descriptiona

Acute MI

AMI-2 Aspirin prescribed at discharge AMI-3 ACEI for LVSD AMI-4 Adult smoking cessation advice/counseling AMI-5 Beta blocker prescribed at discharge AMI-9 Inpatient mortality

Heart Failure

HF-1 Discharge instructions HF-2 LVF assessment HF-3 ACEI for LVSD HF-4 Adult smoking cessation advice/counseling

Community Acquired

Pneumonia

CAP-1 Oxygenation assessment CAP-2 Pneumococcal screening and/or vaccination CAP-3 Blood cultures CAP-4a Adult smoking cessation advice/counseling CAP-5 Antibiotic timing

a: Labels correspond to Joint Commission core measure

Outcome: Absolute Number of Events Table 2B: Selected Maryland Health Services Cost Review Commission Quality of Care Measures Applicable to Hospitalists

Category Descriptionb

Neurologic

1 Stroke & Intracranial Hemorrhage 2 Extreme CNS Complications 36 Acute Mental Health Changes 47 Encephalopathy

Pulmonary

3, 4 Acute Pulmonary Edema and Respiratory Failure 5 Pneumonia & Other Lung Infections 6 Aspiration Pneumonia 7 Pulmonary Embolism 8 Other Pulmonary Complications 49 Iatrogenic Pneumothorax

Cardiac

9 Shock 10 Congestive Heart Failure 11 Acute Myocardial Infarction 12 Cardiac Arrhythmias & Conduction Disturbances 13 Other Cardiac Complications 14 Ventricular Fibrillation/Cardiac Arrest

Gastrointestinal 17, 18 Major Gastrointestinal Complications 19 Major Liver Complications 20 Other Gastrointestinal Complications

Infectious

21 Clostridium Difficile Colitis 22 Urinary Tract Infection 33 Cellulitis 34 Moderate Infectious 35 Septicemia & Severe Infections 54 Infections due to Central Venous Catheters

Genitourinary 23 GU Complications Except UTI 24, 25 Renal Failure

Other

15 Peripheral Vascular Complications Except Venous Thrombosis 16 Venous Thrombosis 26 Diabetic Ketoacidosis & Coma 28 In-Hospital Trauma and Fractures 48 Other Complications of Medical Care 50 Mechanical Complication of Device, Implant & Graft 53 Infection, Inflammation & Clotting Complications of Peripheral Vascular Catheters & Infusions

b: Numbers correspond to MHSCRC Potentially Preventable Complication (PPC) indexing number

The Study Who? Hospitalists from the JHCRN

What to study? The patient:physician ratio (as assessed by billing encounters) Patient, physician, and hospital level factors Percent Compliance with Joint Commission Core measures Number of Highly Preventable Hospital Acquired Conditions Percent In-Hospital mortality Percent 30 day readmission

Predictors

Outcomes

The Study: The How?

Determine the averagepatient:physician ratio per weekday shift based on administrative billing data in each of the sites.

We will also describe the variability in the ratio over time (weekly) within and between sites.

Using linear regression (within sites) and ANOVA (between sites), we will calculate the adjustedpatient:physician ratio adjusted for patient, physician, and hospital factors.

Next, we will examine how quality of care varies with the patient:physician ratio.

Quality of care will be defined using two main outcome measures. One outcome will be percent compliance with JC core measures; the second will be absolute number of HPHAC.

We will examine the associationbetween the adjusted patient:physician ratio and our quality of care measure using Analysis of Variance for JCcompliance and poisson regression for HPHAC events.

We will perform a similar analysis between the adjusted ratio and percent 30 day readmission and in-hospital mortality

Why should you care? Improve our knowledge in the relationship between attending

physician workload and quality of care. Define the association and identify important patient-, physician-,

and hospital-level factors that affect physician workload. Control for differences seen between patient populations,

physicians, and hospitals, allowing for standardization of workload. Lead to a greater focus on patient:physician ratios, distribution of

responsibilities, and staffing plans, similar to those implemented and required for nursing.

Hopefully translate into safer and higher quality care. Necessary, especially in the setting of increased patient access, a

greater focus on patient safety and hospital acquired conditions, and rising healthcare costs.

It affects you, your family, and your institution.

Acknowledgements Daniel Brotman, MD, FACP; Director, Hospitalist Program, Johns

Hopkins Hospital Daniel E. Ford, MD, MPH; Professor of Medicine, Vice Dean for

Clinical Research Peter Pronovost, MD, PhD; Professor of Medicine, Medical

Director for the Center for Innovation in Quality Patient Care The Rockin’ Small Group Number 8◦ Steve Sozio, MD, MHS, FASN; Assistant Professor, Division of Nephrology ◦ Sherley Abraham, MD◦ Monica Giles, MD◦ Eleni Liapi, MD◦ Purva Sharma, MD

The Instructors of this Course!

Questions?

Conceptual Model

Conclusion: Little Picture Physicians reported that their patient load often (≥4/5)

led to:◦ incomplete patient/family discussions (24.6%)◦ ordering potentially unnecessary tests or procedures (22%),◦ delaying admitting or discharging patients until the next shift or

day (21.5%)◦ cross-covering (20.3%) or caring (16.5%) for too many patients,◦ worsened patient satisfaction (19.3%)◦ poorer handoffs (17.9%)◦ increased 30 day readmission (14%)◦ worsened overall quality of care (12.4%)◦ failure to promptly act on critical findings (9.8%)◦ treatment errors (6.5%)

With respect to adverse events, physicians reported that workload has likely (≥4/5) caused◦ transfers to higher levels of care (9.8%)◦ morbidity/complications (6.9%)◦ mortality (4.9%)◦ incident reports (5.7%)

Conclusion: Little Picture

Conclusion: Big Picture Summary ◦ Forty percent of Hospitalists

reported an unsafe workload at least monthly.

◦ Over 20% of Hospitalists reported workload has often caused incomplete patient discussions, unnecessary tests and procedures, admission/discharge delays, and excessive cross-coverage.

◦ Hospitalist workload may be adversely affecting patient safety and quality of care and should be further explored.

Limitations◦ Extreme values for number of

encounters, suggesting misinterpretation

◦ Extreme values for work day hours, suggesting misinterpretation

◦ Relying on self-report for all parameters

◦ External generalizability

The Future: Examining Predictors for Unsafe Census

202 Respondents 291 Respondents

890 Invitations to Participate

506 Respondents

5 Eliminated because did not complete any of the safety

questions

501 Respondents

At Least Monthly Less Than MonthlyFrequency of Unsafe Census

Just a taste… Associated WITH frequent UNsafe◦ Increasing inpatient percent

clinical work◦ No assistance

NOT Associated with frequent UNsafe◦ Practice area◦ Practice organization◦ Salary, bonus or total

Associated WITH frequent safe◦ Internal Medicine◦ Outpatient provider◦ Increasing years of practice◦ Increasing housestaff usage◦ Any system for census control◦ Fixed cap◦ Larger group size◦ Private insurance◦ More consult time