Medicare Spending and Outcomes After Post-Acute Care
Posted: 22 Jun 2007
Elderly patients who leave an acute care hospital after a cerebrovascular accident (CVA) or a hip fracture (HIP) may be discharged to home or undergo rehabilitation care in an inpatient rehabilitation facility (IRF) or skilled nursing facility (SNF). Knowing the costs and health outcomes of these alternatives is important to maximize the efficiency of post-acute care (PAC), but naive estimates are likely to be biased by unobserved patient selection. This study assesses Medicare payments for and outcomes of patients discharged to IRFs, SNFs, and home. In addition to adjusting for observable differences in patient severity across PAC sites, we use instrumental variables (IV) to account for unobserved patient selection.
We analyzed data on elderly Medicare CVA and HIP patients discharged from acute care between January 2002 and June 2003. We used acute hospital, home health, skilled nursing, and inpatient rehabilitation data to obtain total payments over 120 days after acute care discharge. Health outcomes were death and the joint outcome of death or long-term institutionalization in a custodial nursing home at 120 days after acute discharge. We estimated a pair of multiple-equation selection models, one for payments and one for health outcomes. In each model, there was a set of multinomial logit equations for the choice of PAC site and another equation was a linear model (for log-transformed payments) or a binary logit (for the health outcome). Independent variables included age; sex; clinical variables such as severity of the main condition, comorbidities, and complications; and hospital characteristics. The PAC site choice equations included measures of each patient's proximity to and local availability of IRFs and SNFs as IVs. In addition, we specified the joint distribution of latent factors that were incorporated into both equations to allow for unobserved influences on care choice to affect outcomes.
Our results indicate that selection has a sizable influence on estimates of health outcomes, but a much smaller effect on payments. Both IRF and SNF are more costly than going home. In addition, IRF improves health outcomes for both CVA and HIP. By contrast, SNF reduces mortality for HIP, but increases institutionalization for both CVA and HIP.
Of note, our IV estimates show the effects of IRF and SNF use for marginal patients whose decision to use IRF or SNF is swayed by the proximity and availability of these PAC sites and for whom, in a sense, the clinical decision is gray. Nonetheless, our prior work has shown that this is a sizeable group and thus policies designed to direct them to sites of care with paid less and/or with better outcomes could be warranted.
Keywords: health care financing, post-acute care, Medicare, selection models
JEL Classification: I11, I12
Suggested Citation: Suggested Citation