To determine whether vertically integrated hospital and skilled nursing facility (SNF) care is associated with more efficient use of postdischarge care and better outcomes.
Medicare provider, beneficiary, and claims data from 2012 to 2014.
We compared facility characteristics, quality of care, and health care use for hospital‐based SNFs and “virtually integrated” SNFs (defined as freestanding SNFs with close referral relationships with a single hospital) relative to nonintegrated freestanding SNFs. Among patients admitted to integrated SNFs, we estimated differences in health care use and outcomes for patients originating from the parent hospital (ie, receiving vertically integrated care) versus other hospitals using linear regressions that included SNF fixed effects. We estimated bounds for our main estimates that incorporated potential omitted variables bias.
We identified hospital‐based SNFs based on provider data. We defined virtually integrated SNFs based on patient flows between hospitals and SNFs. We identified SNF episodes, preceding hospital stays, patient characteristics, health care use, and patient outcomes using Medicare data.
Consistent with prior research, integrated SNFs performed better on quality measures and health care use relative to nonintegrated SNFs (eg, hospital‐based SNFs had 11‐day shorter stays compared with nonintegrated SNFs adjusting for patient characteristics, < .001). Stroke patients admitted to hospital‐based SNFs from the parent hospital had shorter preceding hospital stays (adjusted difference: −1.2 days, = .001) and shorter initial SNF stays (adjusted difference: −2.7 days, = .049); estimates were attenuated but still robust accounting for potential omitted variables bias. For stroke patients, associations between vertically integrated care and other outcomes were either statistically insignificant or not robust to accounting for potential omitted variables bias.
Vertically integrated hospital and SNF care was associated with shorter hospital and SNF stays. However, there were few beneficial associations with other outcomes, suggesting limited coordination benefits from vertical integration.
