Background: Policymakers are increasingly using public report cards to improve health care quality. In July 2015, policymakers in the United States began issuing quality star ratings to home health agencies, which serves a disproportionately older, sicker, and low-income population. These quarterly ratings are intended to be simple to comprehend and increase demand for quality, incentivizing home health agencies to compete and improve. Despite the theoretical appeal of the star system, whether this effort has shifted patients from lower- to higher-rated home health agencies is unknown.
Objective: To determine whether providing quality information in the form of summary star ratings affect older patients' use of home health care at the margin.
Methods: This study uses a regression discontinuity design comparing agencies that were virtually identical but barely on the opposite sides of an arbitrary star rating threshold, to assess whether one more half star increased an agency's market share of new patients. This design enables me to disentangle the effects of reported quality from effects arising from other sources (e.g., changing quality beliefs), a common challenge in the information-disclosure literature. I use national Medicare data from 2014 through 2016 to examine fee-for-service beneficiaries new to home health, since they face no travel or out-of-pocket costs and no network restrictions that could affect patient choice. I estimate the overall effect of each additional half-star by pooling the star thresholds and six rating releases. I also test for heterogeneity across the star distribution and over time by examining each threshold and rating release separately. I focus on subsamples where effects should be more pronounced, such as community-entry patients who have less time constraints to seek information and patients in markets with greater variation in agency ratings where information is more useful. Finally, I assess how star ratings affect patient shares for the highest-rated agencies in a market, hypothesizing that the ratings may affect choice only for agencies at the top of the distribution.
Results: In first 18 months following the availability of the home health star ratings, the share of new patients increased by a statistically and economically negligible amount of 1.4 percent or 0.26 (95% CI: -0.62 to 1.14) percentage points for every half-star increase in an agency's ratings. There was no evidence that effects varied across the star distribution, over time, for community-entry patients, or in markets with more rating variation. Interestingly, highest-rated agencies within each market decreased their new patient shares by 3.12 (SE=0.66) percentage points, along with a 2.54 (SE=1.42) percentage point decrease in total caseloads; no such corresponding decrease in caseloads was observed in competitive markets (-0.029, SE=0.38), however. These results suggest that the null main findings may be partly attributable to practice pattern changes among home health agencies in response to market conditions.
Conclusion: For the millions of people relying on home health to remain in the community, the star ratings did not help them use higher-rated agencies. Even if consumers are not underusing star information, these findings suggest that supply-side practices may also limit patient choice.