Trends in hospital-SNF relationships in the care of Medicare beneficiaries

Abstract

Improving the value of post-acute care at skilled nursing facilities (SNFs) has become a Medicare policy priority. Anecdotally, hospitals have responded by formally acquiring or pursuing tighter informal connections with SNFs. We evaluated the trend in connections between US acute care hospitals and Medicare-certified SNFs between 2000 and 2013 using vertical integration and two novel network-based measures (number of SNF partners, and discharge concentration). Among 4441 hospitals and 17,215 SNFs, hospitals with weaker connections with SNFs were more often non-profit, major teaching hospitals with a larger number of discharges and beds. We found an apparent weakening of hospital-SNF connections over time for all three measures. Over one-third (39%) of hospitals were vertically integrated in 2000 compared to 8.2% in 2013. The number of SNF partners increased between 2000 and 2013, while hospitals' discharge concentration declined steadily. Additional work is needed to understand the implications of these trends.

Keywords: Hospitals; Medicare; Post-acute care; Skilled nursing facilities.

Fig. 1.

Two measures of hospital-SNF connections based on network methods: Number of SNF Partners and Discharge Concentration Notes: This figure illustrates the concepts of a hospital's number of SNF partners and discharge concentration as network measures. The two fundamental components of networks are entities (termed “nodes”) and connections (termed “edges”). (A) Number of SNF partners describes the connectedness of a given node, and in this study quantifies the number of connections between hospitals and SNFs. For example, hospital A (the green node) is connected to 5 SNFs (blue nodes). (B) Discharge concentration is based on the Herfindahl-Hirschman Index, a common measure of market concentration. For a given hospital, its discharge concentration captures how concentrated a hospital's discharges are among the SNFs to which it sends any patients. For example, hospital A and hospital B both discharge 100 patients to 5 different SNFs. For hospital A, however, 96 of those discharges are to a single SNF, and 1 to each of the other 4 SNFs, while hospital B discharges 20 patients to each of the 5 SNFs. As described above, hospital A has a higher discharge concentration (discharge connection = 0.92) than hospital B (discharge concentration = 0.20).

Fig. 2.

Trends in Hospital-SNF Connections. Notes (for panels B & C): error bars represent IQR (25th–75th percentiles).