The Effect of Integration Between Hospital and Post-acute Care Providers on Patient Outcomes

Emerging payment models in the U.S., including Accountable Care Organizations (ACOs) and bundled payments, focus on coordination of care. Improving integration across providers is expected to be a key component to success with these payment models and is particularly pressing for acute care hospitals and post-acute care (PAC) providers, given the high rate of hospital readmissions and the frequency, expense, and variability in the use of PAC. However, the effects of vertical integration on patient outcomes are ambiguous in theory and limited empirical work has examined this issue.  In this paper we investigate changes in hospital-PAC integration over time and examine empirically the effect of integration on patient outcomes.

We use 2005-2013 data on all PAC providers and all fee-for-service Medicare beneficiaries receiving PAC in the U.S. to examine integration between hospitals and three common post-acute care settings: skilled nursing facilities (SNFs), inpatient rehabilitation facilities (IRFs), and home health agencies (HHA). We use provider-level data to measure formal vertical integration and we use patient-level data to construct a second measure of patient-flow-based integration using a Herfidahl Index for each PAC provider (the concentration of hospital discharges at each PAC provider within health care market). We link these measures to hospital claims to examine the effect of hospital-PAC integration on patient readmission within 30 days of hospital discharge. Our identification strategy is twofold. First, we use longitudinal models with a fixed effect for each hospital-PAC pair in the same market to test how changes in integration impact patient outcomes. Second, we use multiple instrumental variables approaches, including the differential distance between the nearest integrated PAC provider and the chosen PAC provider and market-level measures of integration as our instruments.

Our preliminary results indicate that rates of formal vertical integration decreased over our study period for all three types of PAC providers, dropping from 6.5% in 2005 to 3.5% in 2013 for SNFs; from 82.2% to 78.8% for IRFs; and from 13.9% to 3.5% for HHAs. PAC concentration measures of integration declined slightly over this same period but, compared to formal vertical integration, remained relatively stable. Using naïve OLS model with fixed effects for hospital-SNF pairs, we find that SNFs that are vertically integrated with hospitals have a 30-day readmission rate that is close to 10 percentage points lower than non-vertically integrated SNFs. Using an instrumental variable for patient selection to SNF and a two-stage residual inclusion model with hospital-SNF fixed effects, this effect changes direction, with vertically integrated SNFs having higher rates of 30-day readmission. On the other hand, SNFs that have hospital discharges that are more concentrated among hospitals have lower 30-day readmission rates in instrumental variables specifications.  Results for IRF and HHA are in progress.

These preliminary results suggest that functional hospital-PAC integration improves patient outcomes, though hospital ownership of a PAC does not.