Pay-for-Performance Results

Experts and policymakers have a lot of hope for pay-for-performance programs.  CMS has converted almost all Medicare reimbursement to some aspect of value-based purchasing.  But the research continues to show modest effects, at best, from these efforts.  A study published in Health Affairs continues this trend.  (HA Article)  The research was based on a pay-for-performance method at Fairview Health System in Minnesota that based up to 40% of compensation on quality metrics.  The measures included diabetes care, heart and vascular care, cancer screening, depression care and asthma care.  If clinical performance was at the median for all providers in the state, the clinician received the median compensation, for the portion of their compensation covered by the P 4 P program.  If clinical quality was above the median, the provider could get up to 50% more than median compensation and if it was below median, would get up to 50% less.  Other factors in total compensation included patient satisfaction, number of patients on the panel, number of patient interactions and being a team player.

During the study period between 2010 and 2012 Fairview improved by 9% in vascular care and 2% in cancer screening, but its diabetes care quality scores fell by 2%.  Fairview did not show better scores or improvement than other providers in Minnesota, most of whom were not using such an extensive pay-for-performance program.  For the primary care subset of physicians, there was significant improvement in cancer screening.  However, individual providers who started with the lowest scores showed the greatest improvement.  Another interesting result was that doctors with low-income panels tended to start with the lowest scores, but also showed the highest improvement, again demonstrating the strong effect that sociodemographic factors can have on quality measurement.  The failure of physicians in the top performing group to show much improvement indicates that there may be a point at which it is simply not reasonable to expect much further gain from quality incentive programs.  Overall, the study would suggest that pay-for-performance may not have the significant quality effects promoted by advocates, but that, properly targeted, they can raise quality among doctors who start out with lower scores and they can also be used to reward those who consistently maintain high scores.