Composite model for profiling physicians across domains of care

Area of Science:

• Health Services Research
• Health Informatics
• Biostatistics

Background:

• Physician profiling aims to enhance healthcare quality by recognizing individual physician contributions.
• Developing reliable physician performance metrics is challenging due to small sample sizes, incomplete data, and variations in physician panels.
• Existing methods often struggle with data limitations and variability, hindering accurate comparisons.

Purpose of the Study:

• To develop and demonstrate a hierarchical, weighted composite model for reliably comparing primary care physicians across different domains of care.
• To create a robust methodology for physician performance assessment within an integrated healthcare delivery system.
• To establish a reliable framework for quality and efficiency measurement in primary care.

Main Methods:

• Evaluated 199 primary care physicians using 19 quality and 2 efficiency measures from the Healthcare Effectiveness Data and Information Set (HEDIS) and pay-for-performance programs.
• Developed a hierarchical, weighted composite model to group individual measures into quality and efficiency composites.
• Assessed the reliability (signal-to-noise ratio) of each composite and weighted them to generate a single summary score for each physician.

Main Results:

• The quality-of-care composite demonstrated high reliability, with a median signal-to-noise ratio of 0.98 and 99.5% of physician estimates exceeding the reliability threshold.
• The efficiency composite also showed strong reliability, with a median signal-to-noise ratio of 0.97 and 94.9% of physician estimates exceeding the threshold.
• The hierarchical composite model proved effective in generating reliable physician performance scores across care domains.

Conclusions:

• Reliable physician profiling is achievable across diverse care domains using a hierarchical composite model integrating multiple data sources.
• The developed model successfully generated reliable quality and efficiency scores for primary care physicians.
• The model is adaptable for various healthcare settings and has been utilized for distributing incentive payouts.