The New England Journal of Medicine has a study evaluating the English primary care physician pay-for-performance program. (NEJM article) The United Kingdom initiated its pay-for-performance system in 2004 with 136 measures for primary care physicians, with a focus on chronic disease management and patient access to care. The rewards are substantial, up to a third or more of the physician’s base income. The analysis looked at trends in quality improvement since the system was implemented.
Quality of care for the various diseases measured had been improving before the introduction of the program. After pay-for-performance was initiated, the rate of improvement on cardiac disease increased slightly but then fell back. The rate of improvement for asthma and diabetes improved at a faster clip, but then also fell back. Quality scores on indicators linked to pay-for-performance incentives were higher than for those not so linked, with an increasing gap during the term of the analysis. Access to care, measured by being able to see a physician within 48 hours, and physician communication showed no significant changes. Continuity of care, measured by seeing the same physician over time, declined.
The overall conclusion was that there was an improvement in quality scores but it plateaued. Possible explanations include near-maximum achievement, greater difficulty in getting the next level of improvement, phsycians reaching the maximum reward, and physicians having little motivation to seek further improvement given the income gains they already had achieved.
As with any of these programs, the unintended consequences are interesting. Quality scores for unincentivized measures declined, probably because the implicit message was that they didn’t matter that much. Hopefully the right measures were being used and those not used don’t have a negative impact on quality. The access to care and continuity measures may have been in conflict, because getting a patient in to see any physician may disrupt continuity with a particular physician. One takeaway may be that rather than designing and implementing a particular reform, it is best to comprehensively look at the whole system, with all its interactions and design reforms that work interactively to achieve reform’s objectives of more appropriate care, better health outcomes and lower costs.