Does Public Quality Reporting Improve Quality? The Case of Nursing Home Prescribing of Antipsychotics
Poster, Association for Public Policy Analysts and Management (APPAM) Annual Conference, National Harbor, MD
The quality of services provided in nursing homes and other healthcare settings is a high societal concern. Public reporting of provider quality is an important part of the policy making toolkit, to incentivize providers to improve quality, to alleviate information problems in healthcare markets, and enable greater competition. Starting in 2012, nursing homes were required to publicly report the share of residents who received potentially inappropriate antipsychotic medications (APMs) on the popular CMS “Nursing Home Compare” website. This requirement was driven by concern regarding over-medication with APMs (“chemically restraining” to alleviate staff burdens) of NH residents. Policy makers recognized that the addition of this measure to the Compare website might instigate “cream skimming” if nursing homes avoid patients for whom APMs may be appropriate care, thus the policy exempted patients with certain diagnoses considered appropriate for treatment with antipsychotics (schizophrenia, Tourette Syndrome and Huntington’s disease) from the measure’s calculation. Reporting requirements may also prove unwieldy for small nursing homes, so only ones with 30 (or later 20) and more residents were required to report. Prior work has shown that new patient flows are influenced by reported quality, thus there may be strategic responses by nursing homes to public reporting requirements: to medically upcode residents at the margin of an exclusionary diagnosis. There may also be incentives to cream skim patients who are not at high likelihood of needing APMs, and, to prescribe medications that are a partial substitute for APMs (such as antidepressants) but are not included in public reporting. We test the hypothesis that nursing homes responses align with these incentives, using data from individual and administrative records from the nursing home sector. We use a DD design, pre- and post-policy, comparing responses among smaller facilities as a control group to larger facilities who were affected by the policy. More specifically, we use data from three Centers for Medicare and Medicaid Services (CMS) sources: 1) the Minimum Data Set (MDS) which contains an administrative record each year 2011-2017 for individual residents in all certified NHs in the U.S., recording their clinical diagnoses, as well as medications received; 2) 2011-2017 data from the Certification and Survey Provider Enhanced Reports (CASPER) which contains a facility record for each NH; 3) 2011-2017 data from Nursing Home Compare, a publicly available record on every NH containing quality measures relevant to this research. We verify prior findings that overall APM medication of long-term stay residents has been declining over time, especially in homes subject to public reporting. However, we find novel evidence from regression and event study analysis that there was also systematic upcoding: we find that diagnosis of schizophrenia, an exclusionary diagnosis, is increasing in nursing homes subject to public reporting, after the policy, by 3.2% (95%CI: 1.5 to 4.9) among residents with APM prescriptions. We find no evidence consistent with cream skimming. We do find evidence of increased prescribing of anti-anxiety medications, potential substitutes for APMs. We also find disparities in the increasing diagnosis of schizophrenia by marital status and in homes most reliant on Medicaid. Taken together, this evidence suggests that policy makers intending to improve quality in nursing homes through transparency of quality metrics should be aware of possible unintended consequences.