Pay for Performance Has Problems

People are really good at thinking up solutions to problems, convincing themselves that they understand the dynamics of the problem well enough to devise an effective solution.  Health care is full of these solutions.  Fortunately, we have research techniques to evaluate whether the solutions really do work.  A recent example is the CMS initiative to try to reduce hospital-acquired infections, a very worthy goal, by not paying for the care associated with such infections.  Researchers studied the effect of this non-payment policy and the results of the research are reported in the New England Journal of Medicine.   (NEJM Article)   CMS started this policy in 2008 and it will be extended to Medicaid under the ACA and a number of private payers have adopted similar policies.  The researchers avoided using claims data, which can be affected by changes in coding practices, and used surveillance data on 398 hospitals subject to the CMS initiative from the National Healthcare Safety Network.  The two primary infections targeted by CMS, central catheter bloodstream and catheter urinary tract ones, were generally already decreasing before the CMS program and the implementation of the program had no effect on the rates of infections or the trend.  In fact, there was a small, nonsignificant increase in the urinary tract infections.

When compared to rates of ventilator-associated pneumonia, another hospital-acquired condition but one not targeted by the CMS policy, the rates for the targeted infections and the trends of decline were similar, again suggesting that the CMS policy had no meaningful impact.  Similarly, comparing the effect of the policy on hospitals in states which did not have mandatory public reporting of such infections with hospitals that were in states that did have such reporting, there was also no significant impact of the CMS policy on the bloodstream infections, but it actually appears to be associated with a slowing of the decrease in urinary tract infections.  The study analysis also found that the relative number of Medicare patients treated by a hospital did not seem to make a difference in the effect of the CMS policy.  While it may be the right thing in any event for CMS to not pay for conditions that are acquired as a result of health care treatment when those conditions could have been avoided by the provider; it would be misguided to think that these non-payment policies are going to improve quality, at least at this level of penalty, and the result of this study is consistent with other research on other pay for performance programs.