Objective:
The 30-day readmission rate in the Medicare population is near 18%, with an associated cost of $16 billion dollars annually. Policy makers have become focused on trying to identify successful strategies to reduce both the clinical and economic burden of rehospitalizations, and to this end, the Affordable Care Act sets up penalties for hospitals with high readmission rates. Despite the national attention to readmissions, there are many hospitals that have failed to improve their readmission rates. Understanding who these persistently poor-performing hospitals are is key to helping them improve.
Methods:
We used national Medicare data from 2007 through 2009 to calculate mean readmission rates across six common conditions (heart failure, acute myocardial infarction, chronic obstructive pulmonary disease, pneumonia, stroke and gastrointestinal bleeding) for all acute-care hospitals in the U.S. We identified poor baseline performers as those hospitals with performance in the worst quartile of readmission rates in 2007. We then categorized these hospitals into two groups: those who improved by 2009 and those who did not. We compared the characteristics of hospitals and markets in each of these groups.
Results:
Our sample was comprised of 869 poor-performing acute-care hospitals. Baseline median composite readmission rates were 27.8% (IQR 25.8%-32.5%). Of these, 214 (24.6%) hospitals failed to improve their readmission rates by 2009; the median 2009 readmission rate for persistently poor performers was 32.0% (IQR 27.1%-38.0%) while the rate fo hospitals that improved was 20.9% (IQR 18.5%-23.6%). Persistently poor performers were more likely to be small hospitals (71% versus 32%, p<0.001), publicly owned (32% versus 21%, p=0.003), rural (43% versus 18%, p=0.02), non-teaching hospitals (86% versus 74%, p<0.001). They were less likely to have a medical intensive care unit (20% versus 52%, p=0.001). Persistently poor performers were located in areas with fewer specialist physicians (7 per 100,000 population versus 9 per 100,000 population, p<0.001) and lower median income ($33,299 versus $34,523, p<0.001). In multivariate logistic regression analyses, the strongest predictors of being a persistently poor performer were being a small hospital (odds ratio 13.3, p<0.001) and lacking a medical intensive care unit (odds ratio 2.9, p<0.001).
Conclusions and Implications:
Between 2007 and 2009, small, public, non-teaching hospitals were far less likely to improve their readmission rates than others; in general, persistently poor performers were hospitals with lower resource levels and more socioeconomically disadvantaged populations. As hospitals face looming penalties for high readmission rates, our findings raise concern about the ability of the small, worst-performing hospitals and those with poor resource bases to improve their outcomes, and thus raise concerns about the potential of readmissions penalties to widen disparities in care. Policymakers may need to consider coupling readmission penalties with programs and resources to help these vulnerable hospitals improve in order to avoid unintended consequences of this policy initiative.
Does Implementation of a Previously Validated Prediction Tool Reduce Readmission Rates Into a Medical Intensive Care Unit?
