Vertical integration and optimal reimbursement policy.

Abstract

Health care providers may vertically integrate not only to facilitate coordination of care, but also for strategic reasons that may not be in patients' best interests. Optimal Medicare reimbursement policy depends upon the extent to which each of these explanations is correct. To investigate, we compare the consequences of the 1997 adoption of prospective payment for skilled nursing facilities (SNF PPS) in geographic areas with high versus low levels of hospital/SNF integration. We find that SNF PPS decreased spending more in high integration areas, with no measurable consequences for patient health outcomes. Our findings suggest that integrated providers should face higher-powered reimbursement incentives, i.e., less cost-sharing. More generally, we conclude that purchasers of health services (and other services subject to agency problems) should consider the organizational form of their suppliers when choosing a reimbursement mechanism.

Full text

Page 1

NBER WORKING PAPER SERIES
VERTICAL INTEGRATION AND OPTIMAL REIMBURSEMENT POLICY
Christopher Afendulis
Daniel Kessler
Working Paper 17316
http://www.nber.org/papers/w17316
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
August 2011
We gratefully acknowledge funding from the National Institutes on Aging through the NBER. Afendulis
also gratefully acknowledges funding from the Marshall J. Seidman Program in Health Economics
in the Department of Health Care Policy at Harvard Medical School. The views expressed herein are
those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.
NBER working papers are circulated for discussion and comment purposes. They have not been peer-
reviewed or been subject to the review by the NBER Board of Directors that accompanies official
NBER publications.
© 2011 by Christopher Afendulis and Daniel Kessler. All rights reserved. Short sections of text, not
to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including
© notice, is given to the source.

Page 2

Vertical Integration and Optimal Reimbursement Policy
Christopher Afendulis and Daniel Kessler
NBER Working Paper No. 17316
August 2011
JEL No. I1,I11,I18,L2
ABSTRACT
Health care providers may vertically integrate not only to facilitate coordination of care, but also for
strategic reasons that may not be in patients' best interests. Optimal Medicare reimbursement policy
depends upon the extent to which each of these explanations is correct. To investigate, we compare
the consequences of the 1997 adoption of prospective payment for skilled nursing facilities (SNF PPS)
in geographic areas with high versus low levels of hospital/SNF integration. We find that SNF PPS
decreased spending more in high integration areas, with no measurable consequences for patient health
outcomes. Our findings suggest that subjecting integrated providers to higher-powered reimbursement
incentives, i.e., less cost-sharing, may enhance medical productivity. More generally, we conclude
that it may be efficient for purchasers of health services (and other services subject to agency problems)
to consider the organizational form of their suppliers when choosing a reimbursement mechanism.
Christopher Afendulis
Department of Health Care Policy
Harvard Medical School
180 Longwood Avenue
Boston, MA 02115
afendulis@hcp.med.harvard.edu
Daniel Kessler
Hoover Institution
Stanford University
434 Galvez Mall
Stanford, CA 94305
and NBER
fkessler@stanford.edu

Page 3

2
Introduction
Over the past 20 years, markets for health services have become significantly
more vertically integrated. Producers of complementary services that were once
independent are now increasingly commonly-owned. By the end of the 1990s, three-
quarters of acute-care hospitals had acquired a supplier of post-acute care such as a
rehabilitation hospital, home health service, or skilled nursing facility (SNF) (CBO
1997). The most important of these from the perspective of the Medicare program are
SNFs, which account for the largest share of post-acute spending (MedPAC 2010).
In theory, the optimal response of Medicare to the integration of hospitals into
post-acute services is indeterminate. Because providers have better information on the
severity of patients' illness, optimal reimbursement policy generally allows them to pass
on some but not all of the costs of treatment. This well-known result reflects the tradeoff
between incentives not to skimp on treatment and incentives to contain costs (Ellis 1998).
Because integration may change the terms of this tradeoff in either direction, or not at all,
whether integrated post-acute care providers should be offered more or less cost-sharing
than their stand-alone counterparts is an empirical question.
On one hand, integration may enhance these providers' incentives to supply
unnecessary treatment. This could occur, for example, if integration is used as a vehicle
to pay kickbacks for inefficient referrals (Pauly 1979).1 Inefficient referrals for post-
acute care are of particular interest in the Medicare program, because acute-care hospitals

1 Explicit payments for referrals are banned by Medicare (and Medicaid), effectively eliminating them, but
the legal prohibition can be circumvented by integration, because policing transfer payments among parties
that share fixed assets within an organization is extremely difficult.

Page 4

3
have traditionally faced higher-power reimbursement than post-acute care providers. 2
This anomaly could give integrated hospitals greater incentives to discharge patients to
post-acute facilities prematurely, thereby resulting in more avoidable post-acute stays
(e.g., Morrisey, Sloan, and Valvona 1988). In this case, giving integrated providers more
powerful incentives for cost containment could limit spending and improve efficiency.
On the other hand, integration may enhance post-acute providers' incentives to
skimp. This could occur, for example, if integration inhibits Medicare's ability to
monitor quality of care, or if integration enables providers to better align the incentives of
physician decision-makers with their own at patients' expense. In this case, giving
integrated providers less powerful incentives for cost containment could be optimal.
Integration may also have no effect on the principal/agent relationship between
Medicare and post-acute care providers, if (for example) it is simply a way to facilitate
coordination of care. In this case, integration would not change incentives for skimping
or cost containment, and should therefore not be a factor in determining the optimal cost-
sharing rule.
Despite the importance of this question, little research has examined the
relationship between hospital integration and Medicare reimbursement policy. In this
paper, we evaluate the effects of a change in such policy on health spending and the
quality of care. In 1997, Congress adopted prospective payment for skilled nursing
facilities (SNF PPS), under which SNFs are paid a largely fixed amount per patient per
day based on the patient's anticipated resource use.3 Prior to this change, Medicare

2 Before 1997, acute-care hospitals were reimbursed under a prospective payment system and post-acute
facilities under a cost-plus system, which creates an opportunity for profit from shifting care
inappropriately from the acute to the post-acute setting.
3 Congress later adopted prospective payment for other forms of post-acute care.

Page 5

4
reimbursed SNFs on the basis of their costs. This change therefore led to decreased cost-
sharing for both stand-alone SNFs and those that were integrated with an acute-care
hospital.4
We analyze the Medicare spending and health outcomes of all elderly patients
admitted to an acute-care hospital with stroke in 1997, 1998, and 1999. Rehabilitation
services provided by SNFs are frequently, but not always, indicated for patients who are
hospitalized with this disease (Gropen et al. 2009).
We compare the pre/post-SNF-PPS change in the spending on and the health
outcomes of an episode of illness in geographic areas with a high versus a low level of
hospital/SNF integration. We hold constant the characteristics of patients, hospitals,
geographic markets, and other aspects of Medicare reimbursement policy that were
changing over our study period, including area-level measures of the generosity of
Medicare for both hospital and SNF services. This enables us to assess whether
enhanced incentives for cost containment have different effects on spending and health
outcomes, depending on the extent of vertical integration.
The paper proceeds in five sections. Section I outlines the previous research on
the role of vertical integration in markets for health care, with a focus on strategic uses of
integration in a principal/agent context. Sections II and III present our data and models.
Section IV presents our results, and Section V concludes with an exploration of potential
implications for Medicare policy and contracting more generally.



4 We describe SNF PPS as leading to decreased cost-sharing, rather than no cost-sharing, because it still
allowed for cost-plus reimbursement on some margins such as the number of days in the SNF and certain
therapy services (White 2005/6).