Yesterday we noted research to the effect that pay-for-performance programs may not be achieving their intended goals. Today we comment on another piece of recent research, published in the Journal of the American Medical Association, that undermines the supposed benefits of public reporting of outcomes. (JAMA Article) This was actually a forerunner strategy to pay-for-performance, with the notion being that if you collected and reported data on providers’ outcomes, patients and payers would use the data to select better-performers, which would put pressure on all providers to improve their outcomes, at least those that were being reported publicly. The current research compared three states (New York, Massachusetts and Pennsylvania) which require public reporting of percutaneous coronary intervention rates and mortality with several states in the same region which do not (Maine, Vermont, New hampshire, Connecticut, Rhode Island, Maryland and Delaware), using Medicare fee-for-service claims data.
The researchers reviewed whether patients with a heart attack diagnosis received PCI during their hospitalization and also looked at 30 day mortality rates. About 100,000 patients were included in the study. The authors found that the three states with public reporting of PCI rates ranked in the bottom quintile of all states and that the odds of a patient in one of those states have PCI was significantly lower than the odds for a patient in one of the non-reporting comparison states. In Massachusetts, which began public reporting during the study period, the odds of getting PCI were higher before public reporting was initiated. In regard to 30-day mortality rates, there was generally no difference between the reporting and non-reporting states, with the exception that in one sub-group there was a slightly higher mortality rate in the public reporting states. The authors give two possible explanations, one is that providers stopped doing unnecessary procedures and the other is that they avoided doing them on riskier patients so that they would not have to report potential bad outcomes on those patients. The mortality statistics don’t shed much more light, since they could be consistent with either not doing needed PCIs or with the avoidance of unnecessary ones. Whatever the explanation, the results can hal