Does Vertical Integration Improve Care Coordination? Evidence from Hospital-SNF Linkages

Research objective

With the implementation of new payment models such as bundled payments, hospitals increasingly face financial responsibility for patient outcomes and spending even after hospital discharge. Some have predicted that the expansion these new models would lead to greater vertical integration of hospitals with post-acute care providers. Increased vertical integration could improve care coordination reducing costs and improving outcomes. However, there is scant evidence on the benefits of vertical integration. Prior work has shown that patients admitted to hospital-owned skilled nursing facilities (SNFs) have shorter lengths of stay without higher hospital readmissions or mortality rates. It is unclear, however, whether the better outcomes among patients admitted to hospital-based SNFs reflects: (1) improved care coordination due to closer relationships with hospitals versus (2) hospital-based SNFs being higher quality than other SNFs on average, independent of hospital linkages. In this paper, we disentangle these effects to evaluate whether vertical integration lowers costs and improves patient outcomes.

Study design

In order to identify a distinctive “coordination” effect, we compared patients admitted to the following SNF-hospital combinations relative to non-integrated hospitals and non-integrated SNFs: (1) vertically integrated hospital and SNF, (2) vertically integrated hospital but a different, non-integrated SNF, and (3) non-vertically integrated hospital and a vertically integrated SNF. To the extent that combination (1) is associated with lower length of stay and better patient outcomes, this could either represent a care coordination effect or just that vertically integrated SNFs are higher performing facilities. In contrast, patients admitted to combination (3) will benefit from a “performance” effect without the care coordination effect. Thus, the differential effect of combination (1) versus (3) isolates the “care coordination” effect of being admitted to a vertically integrated SNF and hospital on length of stay and outcomes.

We accounted for selection into the different SNF-hospital combinations described above by using two-stage residual inclusion. The first stage was a multinomial logit model with the four categories of SNF-hospital integration as the dependent variable (described above). The differential distances of the closest integrated hospital and SNF relative to the closest hospital and SNF to a patient’s zip code were used as instruments. In the second stage, we examined the effect of SNF and hospital integration combinations on SNF days, readmissions, and post-discharge spending.

Population studied

Medicare fee-for-service enrollees that were hospitalized for a stroke and subsequently admitted to a SNF between January 2012 and June 2014.

Principal findings

We found shorter lengths of stay for patients admitted to vertically integrated SNFs relative to non-integrated SNFs, but patients admitted to vertically integrated SNFs from the parent hospital exhibited significantly shorter stays than patients admitted to vertically integrated SNFs from other hospitals. Patients admitted to vertically integrated SNFs from the parent hospital exhibited lower hospital readmission rates while patients admitted from other hospitals did not.

Conclusions

Our findings suggested the presence of a “care coordination” benefit for being admitted to a vertically integrated SNF from the parent hospital. More work is needed to identify the specific mechanisms for this coordination relationship.