Readmission penalties and racial disparities in readmissions

Medicare's Hospital Readmissions Reduction Program penalizes hospitals whose 30-day readmission rates exceed what would be expected given their patient mix. Because safety-net hospitals care for a disproportionate share of low-income and minority patients, and because those patients have historically experienced higher readmission rates, the penalties fall disproportionately on safety-net institutions.

So, a reasonable concern is that if a payment program leaves hospitals serving the most disadvantaged patients with fewer resources, it could lead to reduced quality at hospitals least able to absorb the loss. This could cause already substantial racial disparities in readmissions to widen as a result. This concern shaped much of the commentary about the program in the years following its implementation.

What we found

With Michael Thompson and Teresa Waters, I examined 30-day readmissions following index admissions for acute myocardial infarction, congestive heart failure, or pneumonia. The study covered 1,745,686 Medicare patients over 65 who were discharged alive from hospitals in five states between 2007 and 2014, and we stratified the analysis according to whether the hospital met the definition of a safety-net institution.

Two findings stood out, and neither corresponded to the anticipated pattern. The first is that the improvement preceded the policy, since before the program passed in 2010 readmission rates for both Black and White patients and the disparity between them were already declining, with the largest reductions occurring at safety-net hospitals. In 2007, Black patients treated at safety-net hospitals had 13 percent higher odds of readmission than White patients, and by 2010 that gap had narrowed to 5 percent higher odds.

The second is that the anticipated reversal did not occur, since those trends continued after the program took effect and safety-net hospitals did not see their progress on disparities undone despite bearing higher penalties. Disparities did, however, continue to persist at non-safety-net hospitals, which faced considerably lower penalties throughout the period we examined.

What this means

It is worth being careful about what these results do and do not establish. The analysis does not show that the readmissions program caused the improvements observed at safety-net hospitals, since the trends were already underway beforehand and separating a continuing trend from a policy effect is genuinely difficult with a design of this kind. What the results do suggest is that the specific mechanism that concerned many observers, in which penalties degrade care at the hospitals least able to absorb them and disparities widen as a consequence, is not visible in these data through 2014.

The results also direct attention somewhere less expected. Much of the equity discussion around value-based payment has focused on whether penalties harm safety-net providers, which remains a legitimate question and one worth continuing to ask. If disparities are nevertheless proving more durable at better-resourced hospitals facing minimal penalties, then the equity problem in readmissions may not be located primarily in the safety net, but rather at institutions that have faced comparatively little pressure to address it.

Evaluating payment programs

More broadly, value-based payment programs tend to be evaluated according to whether they move their headline metric. The distributional question is usually the more informative one, since it asks which organizations respond, which do not, and what happens to the patients treated in each. A program can produce an acceptable result on average while leaving the disparity it was implicitly expected to narrow largely untouched across a substantial part of the system.