Responses to readmissions penalties: What can we learn about hospital behavior?

点击翻译按钮获取中文摘要

Project Abstract There are approximately 40 million readmissions in the United States per annum, with a price tag of more than $40 billion. Starting in 2013, the Hospital Readmission Reduction Program (HRRP) penalizes hospitals with excess Medicare readmissions for three conditions: acute myocardial infarction, congestive heart failure, and pneumonia. Elective hip/knee replacements and chronic obstructive pulmonary disease were added in 2015, and coronary artery bypass grafts were added in 2016. CMS calculates hospitals’ risk-adjusted readmission rates using a three-year average, and if a hospital’s rate exceeds the national average the hospital receives a reduction in Medicare reimbursement rates in the following year. The reach of the policy has been profound; nearly 80% of hospitals receive penalties under the HRRP each year. In this study, we will first determine how hospitals with varying readmission rates responded to the HRRP’s financial incentives. Using data from 2009-2018 for HRRP-eligible conditions, we will study the effects of hospital distance from the HRRP performance threshold in a given year on future changes in readmission rates. Further, we include several years of pre-HRRP data to control for potential mean reversion in readmissions rates. Next, we will evaluate whether characteristics of hospitals or the communities they serve potentially moderate hospital responsiveness to the HRRP. We will stratify our analyses to determine whether these, as well as other institutional and geographic characteristics, moderate hospitals’ responses to the HRRP. We hypothesize that hospitals which are for-profit, larger in size, have a higher Medicare share of total payments, or have lower patient acuity will demonstrate a larger response to the HRRP’s incentives. Lastly, we will assess whether the observed reductions in hospital readmissions under the HRRP were associated with concomitant increases in hospital mortality rates. As a sub-aim, we will also determine whether hospitals’ participation in value-based care programs (e.g. accountable care organizations, bundled payments) moderates this potential substitution between readmissions and mortality. We hypothesize that hospitals with greater participation may lower readmissions without increasing mortality, but the converse will be true for hospitals with lower participation. The study’s findings may have important implications for several AHRQ priority populations such as the elderly, low-income, urban residents, and those with chronic condition who are disproportionately affected by hospital readmissions. First, our approach allows us to examine hospitals’ responses to different aspects of the HRRP’s financial incentives, which may suggest modifications to the HRRP’s program design to further reduce readmissions. Second, we will check for potential unintended consequences in terms of mortality, which may suggest that quality of care declined under the HRRP for certain hospitals.