Do Surgeons Strategically Select Patients to Improve Scores on Health Care Quality Report Cards?
Posted Aug 07 2012 10:22am
Do report cards improve quality or simply incentivize hospitals and physicians to stratetically choose which patients to treat?
To answer this question, a paper by Chen and Meinecke (2012) examine coronary artery bypass graft ( CABG ) report cards introduced in Pennsylvania. Report cards can improve quality but they can also induce selection behavior not only by providers but also by patients (which is a major reason for their existence).
Previous research by Dranove et al. (2003) “…use the publication date of CABG report cards for New York and Pennsylvania as the cutoff point in a difference-in-differences analysis. Using a comprehensive longitudinal Medicare claims data set, combined with data from the American Hospital Association, they show that the average illness severity of bypass patient decreases by 3.47%–5.30% because of the introduction of CABG report cards, concluding that providers shift treatment from sicker patients to healthier ones. However, it is not clear to what extent the decrease in illness severity is owing to the selection of patients by providers or selection of providers by patients.”
In this study, the authors use data from the Nationwide Inpatient Sample (NIS)–collected by U.S. Agency for Health Care Research and Quality ( AHRQ ). The NIS file contains approximately 8 million hospital stays each year. In 2010, the NIS contained all discharge data from 1,051 hospitals located in 45 States, approximating a 20-percent stratified sample of U.S. community hospitals.
The Pennsylvania report cards “summarize, for each provider, whether the actual mortality rate falls below or above the predicted mortality rate (adjusted for standard errors). Therefore, a provider can either be lower than, higher than, or in line with the expected mortality rate. This quality evaluation is also performed for individual surgeons.”
The paper finds that providers do not select patients in the period when report cards are distributed to providers, but not released to the public. Mortality rates decrease slightly over this time period (0.05 percentage points) which of very small magnitude. This is in contrast to previous empirical studies. The reason could be that hospitals and physicians have less of an incentive to select patients when the report cards are publicly reported. When the report cards are not publicly reported, then the incentive is lower. Although identifying the provider selection when report cards are publicly reported is confounded by the fact that patient selection of hospitals/surgeons occurs simultaneously, this may be the only time when hospital/surgeons are motivated to select patients. In fact, work by both Dranove et al. and Chen and Meinecke find that provider selection and quality improvement occurs when the report cards are publicly reported.
Thus, it appears that report cards only improve quality (and affect patient selection) when they are distributed to the public.