The Effects of Vertical Integration on Health Care Costs and Patient Outcomes

In response to rising health care costs and poor coordination of health care across providers, Medicare and private insurers are increasingly instituting payment reforms to improve efficiency and quality of care. Prominent examples of payment reforms include accountable care organizations (ACOs), where hospitals, physicians, and health plans share the financial risk or savings for patients, creating incentives for better coordination and reducing costs. Another example is “bundled payment”, where a hospital or other entity receives a fixed payment for the hospital stay and all health care that occurs for a fixed time period following discharge from the hospital. 

The financial incentives in ACOs and under bundled payment may lead to greater vertical integration, as care coordination may be easier to accomplish within vertically integrated health care providers. Vertical integration may improve patient outcomes by facilitating communication across providers and ease transitions through physical proximity. It may also reduce health care costs by reducing duplication of tests and procedures, or by facilitating the allocation of tasks to the lowest cost provider.  At the same time, vertical integration may also be part of a strategy to increase market power, which may reduce quality of care and increase costs. In addition, the effects of vertical integration may vary depending on the types of providers involved and with patients’ health condition and severity. Despite trends towards greater integration, the effects of vertical integration in health care are not well understood.

In this paper, we examine the effects of vertical integration of hospitals with post-acute providers on treatment, costs of care, and patient health outcomes.  We focus on Medicare FFS patients admitted to hospitals in 2006 and 2007 with a primary diagnosis of hip fracture or stroke (as such patients are high users of post-acute care) using 100% claims data.  We define vertical integration as the presence of an inpatient rehabilitation facility (IRF) or skilled nursing facility (SNF) onsite at a hospital; we then compare patients seen in vertically integrated hospitals with patients in non-vertically integrated hospitals.  We make a number of contributions to the prior literature. First, we separately estimate the effects of different types of vertically integrated hospitals and post-acute providers. Second, we investigate heterogeneous effects by condition (and later by severity). Finally, we use a differential distance instrumental variables strategy that allows us to estimate the causal effects of being admitted to a vertically integrated hospital.  

We find evidence that hip fracture and stroke patients admitted to hospitals with onsite IRF units are more likely to be admitted to an IRF and less likely to be admitted to a SNF. For both conditions, this leads to higher episode costs (reflecting the higher intensity of IRF care relative to SNF care). Patients are also more likely to return to the community in the three months following hospital discharge, implying improvements in functional status.