Cautionary study on the effects of pay for performance on quality of care: a pilot randomised controlled trial using standardised patients

Abstract

BackgroundDue to the difficulty of studying incentives in practice, there is limited empirical evidence of the full-impact pay-for-performance (P4P) incentive systems.ObjectiveTo evaluate the impact of P4P in a controlled, simulated environment.DesignWe employed a simulation-based randomised controlled trial with three standardised patients to assess advanced practice providers’ performance. Each patient reflected one of the following: (A) indicated for P4P screenings, (B) too young for P4P screenings, or (C) indicated for P4P screenings, but screenings are unrelated to the reason for the visit. Indication was determined by the 2016 Centers for Medicare and Medicaid Services quality measures.InterventionThe P4P group was paid $150 and received a bonus of $10 for meeting each of five outcome measures (breast cancer, colorectal cancer, pneumococcal, tobacco use and depression screenings) for each of the three cases (max $300). The control group received $200.SettingLearning resource centre.Participants35 advanced practice primary care providers (physician assistants and nurse practitioners) and 105 standardised patient encounters.MeasurementsAdherence to incentivised outcome measures, interpersonal communication skills, standards of care, and misuse.ResultsThe Type a patient was more likely to receive indicated P4P screenings in the P4P group (3.82 out of 5 P4P vs 2.94 control, p=0.02), however, received lower overall standards of care under P4P (31.88 P4P vs 37.06 control, p=0.027). The Type b patient was more likely to be prescribed screenings not indicated, but highlighted by P4P: breast cancer screening (47% P4P vs 0% control, p<0.01) and colorectal cancer screening (24% P4P vs 0% control, p=0.03). The P4P group over-reported completion of incentivised measures resulting in overpayment (average of $9.02 per patient).LimitationsA small sample size and limited variability in patient panel limit the generalisability of findings.ConclusionsOur findings caution the adoption of P4P by highlighting the unintended consequences of the incentive system.