Indirect Effects of Introducing Performance Pay for Hospitals: Evidence from Post-Acute Care

The Affordable Care Act reformed Medicare payments by linking hospital reimbursements to performance. One of the initiatives, called the Hospital Readmissions Reduction Program (HRRP), financially penalizes hospitals with high unplanned readmissions. While existing literature shows that the HRRP has successfully lowered readmission rates, little robust evidence exists on mechanisms responsible for such improvements. We fill this gap by investigating a potential mechanism highlighting the role of post-acute care (PAC) providers in reducing readmissions. PAC represents 10% of Medicare spending and is the fastest growing category in Medicare. It plays an important role of stabilizing and improving patient health after hospital discharge.

We examine whether HHAs change their treatment patterns for patients referred by hospitals at greater risk of HRRP penalties. Imposing financial incentives on hospitals -- a dominant source of referrals for PAC providers -- could lead to changes in treatment patterns by PAC providers even if they are not acquired by hospitals. Competition for hospital business could force providers to adapt to new hospital priorities by placing greater emphasis on patients targeted by such penalties.

Our research design uses cross-sectional variation in penalties expected across hospitals. Since hospitals contribute differentially to HHAs in their markets, different HHA office locations will perceive differential penalty ‘pressure’ based on their reliance on penalized hospitals and exposure to conditions targeted by these programs. Hence we can exploit within and across-office location variation in penalty exposure to identify HHA treatment responses both on the extensive (number of visits, number of practitioners, types of experts) and intensive margins (length of visits, type of services) of effort.

To implement this strategy, we exploit access to novel operational data from a large multi-state HHA provider. Few studies have examined detailed treatment patterns within PACs, mainly due to data limitations. Existing studies mostly use national Medicare claims data, which have the advantage of being representative of general US population but do not record details on HHA effort like number and frequency of home visits, hours of service and so on. Our proprietary data contain detailed information at the level of home health visit, such as the length of visits, date and time of visits, and individual practitioner who provided the visit (and his service type). This information allows us to have a rich analysis of HHAs’ treatment intensity and patterns, using such measures as length and frequency of visits, frontloading of visits during a 60-day home health episode, share of visits by different nurse levels (i.e. RN vs LPN), and mix of different service disciplines (e.g. more nurse visits relative to home health aide visits). Moreover, the data provide rich patient health status assessment, which allows us to control for the underlying health risks of each patient.

We expect this project to generate several policy relevant insights on the role of performance pay and PAC in improving productivity in the US health care system.