Misclassification Risk of Tier-Based Physician Quality Performance Systems.

Area of Science:

• Healthcare Quality Measurement
• Physician Performance Evaluation
• Health Services Research

Background:

• Increasing interest exists in identifying high-quality physicians based on performance thresholds.
• Current methods for distinguishing above- versus below-threshold physicians require evaluation for accuracy.

Purpose of the Study:

• To evaluate the accuracy of current methods in distinguishing above- versus below-threshold physicians.
• To estimate misclassification rates for two-category physician identification systems.

Main Methods:

• Utilized claims data for Medicare fee-for-service beneficiaries in Florida and New York (2010).
• Estimated quality scores for colorectal cancer, glaucoma, and diabetes for 23,085 physicians.
• Employed a beta-binomial model to assess physician score reliabilities and computed misclassification proportions under three scoring systems.

Main Results:

• Misclassification rates ranged from 4.5% to 25.7% across three scoring systems.
• True positive rates varied from 34.7% to 100.0%, and true negative rates from 10.5% to 99.9%.
• Positive predictive values ranged from 55.2% to 99.1%.

Conclusions:

• Current physician profiling methods may yield misleading results due to small numbers of eligible events.
• Misclassification rates are policy-relevant, impacting providers, payers, and patients.
• Quantifying misclassification should inform high-performance network construction and quality improvement efforts.