Related Experiment Videos
Is risk-adjustor selection more important than statistical approach for provider profiling? Asthma as an example
I-Chan Huang1, Francesca Dominici, Constantine Frangakis
1Department of Health Policy and Management, Bloomberg School of Public Health, The Johns Hopkins University, Baltimore, Maryland 21205-1901, USA.
Summary
Selecting appropriate risk adjustors significantly impacts physician group performance profiles in patient satisfaction surveys. Careful consideration of both risk adjustors and statistical methods is crucial for accurate provider profiling.
Area of Science:
- Health Services Research
- Health Outcomes
- Medical Informatics
Background:
- Physician profiling using patient satisfaction data is common.
- Risk adjustment is essential for fair comparison of physician groups.
- The impact of different risk adjustment strategies on profiling outcomes is not fully understood.
Purpose of the Study:
- To evaluate how varying risk adjustors and statistical approaches influence physician group performance rankings based on patient satisfaction.
- To identify the most impactful components of risk adjustment models for provider profiling.
Main Methods:
- A cross-sectional study analyzing patient satisfaction surveys from 2515 asthma patients across 20 California physician groups.
- Compared risk adjustment models using sociodemographic, clinical, and health status variables.
- Evaluated statistical strategies including Observed-to-Expected (OE), Fixed Effects (FE), and Random Effects (RE) approaches.
- Assessed model performance using C-statistics and calibration, and ranking impact via Absolute Ranking (AR) and Quintile Ranking (QR).
Main Results:
- Sociodemographic, clinical, and health status variables significantly improved risk-adjustment model discrimination.
- Models incorporating these comprehensive variables (Model S-C-H) using FE or RE approaches showed better performance but fair discrimination (C-statistic=0.68).
- Risk adjustor selection had a greater impact on physician group ranking changes (AR: 50%-80%) than statistical strategy choice (AR: 20%-55%).
Conclusions:
- Comprehensive risk adjustment using sociodemographic, clinical, and health status variables optimizes performance profiling for patient satisfaction.
- The choice of risk adjustors substantially influences provider performance rankings, more so than the statistical approach.
- Stakeholders must carefully select both variables and statistical methods for accurate risk adjustment in provider profiling.