To the Editor The analysis by Dr Bardach and colleagues1 of pay-for-performance incentives rests on data recorded by clinicians. But only the incentivized group had reason to put a positive spin on their numbers. Subtle manipulation could easily bias their results.
For instance, colleagues of ours speak of clinic managers who have coached them to change their digit preference when recording blood pressure levels from 140 mm Hg to 139 mm Hg. In other cases, administrators have discouraged further monitoring (until the next pay-for-performance incentive year) once a goal blood pressure level or hemoglobin A1C has been recorded. Either strategy would distort Bardach et al’s blood pressure results. Their process of care measures are even easier to undermine. Notably, their outcome measure (cholesterol level), which is harder to game, showed no improvement.
David U. Himmelstein, Steffie Woolhandler. Physician Payment Incentives to Improve Care Quality. JAMA. 2014;311(3):304. doi:10.1001/jama.2013.284475