Reducing Hospital Readmissions Through Preferred Networks Of Skilled Nursing Facilities

Abstract

Establishing preferred provider networks of skilled nursing facilities (SNFs) is one approach hospital administrators are using to reduce excess thirty-day readmissions and avoid Medicare penalties or to reduce beneficiaries' costs as part of value-based payment models. However, hospitals are also required to provide patients at discharge with a list of Medicare-eligible providers and cannot explicitly restrict patient choice. This requirement complicates the development of a SNF network. Furthermore, there is little evidence about the effectiveness of network development in reducing readmission rates. We used a concurrent mixed-methods approach, combining Medicare claims data for the period 2009-13 with qualitative data gathered from interviews during site visits to hospitals in eight US markets in March-October 2015, to examine changes in rehospitalization rates and differences in practices between hospitals that did and did not develop formal SNF networks. Four hospitals had developed formal SNF networks as part of their care management efforts. These hospitals saw a relative reduction from 2009 to 2013 in readmission rates for patients discharged to SNFs that was 4.5 percentage points greater than the reduction for hospitals without formal networks. Interviews revealed that those with networks expanded existing relationships with SNFs, effectively managed patient data, and exercised a looser interpretation of patient choice.

Keywords: Hospitals; Medicare; Organization and Delivery of Care; Rehospitalization Rates; Skilled Nursing Facilities.

EXHIBIT 2. Percent of rehospitalized patients discharged home or to a skilled nursing facility (SNF) from hospitals with and without a formal SNF network, 2009–13

SOURCE Authors’ analysis of data for 2009–13 from the Medicare Provider Analysis and Review files and the American Hospital Association Annual Survey. NOTES Hospitals with formal SNF networks are defined in the Notes to Exhibit 1. Only one difference in the trends shown was significant (p < 0:05): the difference between hospitals with and without a SNF network in SNF discharges.