Does Regression to the Mean Explain the Decline in Readmissions Observed at Hospitals Penalized for Excess Readmissions?

Wednesday, June 26, 2019: 12:30 PM
Coolidge - Mezzanine Level (Marriott Wardman Park Hotel)

Presenter: Sushant Joshi

Co-Authors: Teryl Nuckols; Jose Escarce; Peter Huckfeldt; Ioana Popescu; Neeraj Sood

Discussant: Daniel Ludwinski

The Medicare Hospital Readmissions Reduction Program (HRRP) imposes penalties on hospitals with “excess” 30-day readmissions. Excess readmissions adjust for case-mix and are measured by a ratio – excess readmissions ratio (ERR) – calculated by taking the ratio of a hospital’s predicted 30-day readmissions by the number that would be expected based on an average hospital with similar patients. Only hospitals with ERRs greater than one for targeted conditions are penalized.

Case-mix adjusted readmissions started to decline after announcement of HRRP, and the reductions were larger at penalized than non-penalized hospitals. Prior studies interpreted this finding as suggesting that the HRRP’s financial penalties incentivized hospitals to improve quality of care resulting in a decline in excess readmissions. However, the changes in excess readmissions could also be explained, in part, by “regression to the mean” (RTM) -- a statistical phenomenon that occurs when a random outcome for the same entity is measured repeatedly and outcomes further away from the mean in one period are likely to be followed by outcomes closer to the mean in subsequent or preceding periods. Given that hospitals’ excess readmissions involve randomness, hospitals with high ERRs before the HRRP—i.e., penalized hospitals—would be more likely to exhibit subsequent reductions in ERRs due to chance alone.

Our analysis aimed to disentangle the potential role of RTM rather than true policy effects in explaining declines in readmissions at penalized hospitals. If RTM is the primary driver of decline in readmissions at penalized hospitals then we should expect three hypotheses to be true. First, we should expect to see an increase in ERRs after HRRP at hospitals with better than average performance (i.e. non-penalized hospitals). Second, we should expect that hospitals that had worse than average performance in a period that predates implementation of HRRP should exhibit subsequent improvement in performance even though excess readmissions were not penalized during this period. Third, hospitals penalized under HRRP would exhibit declines in ERRs going backward in time just as they do going forward in time.

We find strong evidence that the declines in excess readmissions experienced after the implementation of HRRP at hospitals penalized under were substantially affected by RTM. This evidence includes the fact that, while declines in ERRs occurred at the poor performing hospitals, increases occurred at the better performing hospitals. Furthermore, we observed similar changes in ERRs at worse and better performing hospitals even when we defined performance during an alternative measurement period which predated HRRP and when we examined changes in performance going backward rather than forward in time. We estimate that 74% to 86% of the improvement in performance of penalized hospitals can be explained by RTM. The analysis suggest that the benefits of penalizing hospitals might to be overstated and that both bad luck and bad quality of care determine whether a hospital is penalized.