Health Care Reform Center

Peter S. Hussey, Ph.D., Christine Eibner, Ph.D., M. Susan Ridgely, J.D., and Elizabeth A. McGlynn, Ph.D.

High U.S. health care spending has been characterized not only as a barrier to affordable insurance but also as the preeminent long-term threat to the economy and the competitiveness of American business. The current policy discussion in Congress does not address this problem. The search for government savings with which to fund coverage expansions makes public programs the main targets for spending reductions; opportunities for private-sector savings are left out of the equation. We think it is useful to consider the cost-control options available to both the public and the private sectors.

Although there is no consensus on what should be the target for reducing spending, constraining increases in health care spending to the rate of growth in the gross domestic product (GDP) — and so devoting a fixed share of national income to health care — offers a reasonable goal. To achieve this goal, spending on health care over the next decade would have to be reduced by 6.2% from the amount the Centers for Medicare and Medicaid Services estimates the country would otherwise spend.¹ This proposed reduction provides a framework for evaluating the options now under consideration.

We recently produced quantitative estimates of the likely impact of 12 policy options for reducing health care spending in Massachusetts,² and we have extrapolated from that work to produce estimates for the country as a whole. We identified 8 options that evidence suggests have the potential to reduce spending and are broadly applicable to the United States. For these options, we developed high and low estimates of cumulative cost savings over 10 years. The graph lists the options, ranked according to their savings potential, and shows the percentage change in spending that we estimate could be achieved if that policy alone were implemented. If our most optimistic assumptions are reasonable, health care spending can be reduced, but the lower-bound estimates are far more pessimistic, and the substantial spread between the two indicates that the effects are highly uncertain.

Estimated Cumulative Percentage Changes in National Health Care Expenditures, 2010 through 2019, Given Implementation of Possible Approaches to Spending Reform.

HIT denotes health information technology, NP nurse practitioner, and PA physician assistant.

Among the most promising options are those related to changing the payment methods for health care services. Fee-for-service payments, today’s dominant mode, encourage higher volume rather than better value. Most payment reforms under consideration are designed to provide incentives for both efficiency and quality in care delivery.

A “bundled-payment” approach would provide a single payment for all services related to a given treatment or condition, causing providers to assume risk for preventable costs; this approach has proved effective in limited demonstration projects. Bundled payment provides a mechanism for reducing both the volume of services and the prices charged for them. We estimate that under optimistic scenarios and with broad use of the Prometheus model³ of bundled payment for six chronic conditions and four acute conditions or procedures requiring hospitalization, national health care spending could be reduced by 5.4% between 2010 and 2019. This estimate assumes that providers can achieve a reduction of 25 to 50% in the costs associated with avoidable complications by providing higher-quality, more collaborative care. However, bundling payments only for hospital-based services would reduce spending by 0.1%. The Congressional Budget Office (CBO) has estimated that bundled payment would save only $19 billion (0.05%) between 2010 and 2019,⁴ but it considered only hospital and post-acute care services paid for by Medicare, and since Medicare already bundles hospital payment through the diagnosis-related-group system, the savings opportunities are limited. We find that there would be greater opportunities in bundling payments for the treatment of chronic diseases and in applying the model to all payers. The difference between the CBO’s estimates and ours illustrates the limits of focusing solely on savings in the federal budget.

A second possible option for payment reform, all-payer hospital rate setting, has had a lower profile. This approach involves the creation of a regulatory authority that would set the prices for hospital services to be paid by all private and public payers. Hospital rate setting was widely implemented in the 1970s and 1980s but was later abandoned by all states except Maryland and West Virginia. The most likely result, given the previous experience, is that hospital rate setting would not reduce spending; our most optimistic estimate is a 2.0% reduction in national health care spending between 2010 and 2019.

Options for improving the quality and efficiency of health care require increased short-term spending on infrastructure that could eventually lead to more efficient care. Two important infrastructure-related options are accelerating the adoption of health information technology (HIT), which was funded in the American Recovery and Reinvestment Act of 2009, and expanding primary care capacity. We examined several options for improving primary care delivery, including creating medical homes, increasing the use of physician assistants and nurse practitioners, and expanding the number of retail clinics. Although these approaches seem unlikely to produce substantial savings by themselves, they provide a foundation for other cost-control efforts.⁴ For example, HIT adoption could facilitate both improvements in health services delivery and innovations in payment methods.

Our analysis of the literature indicates that the effect of disease-management programs on spending ranges from an increase of 1.0% to a decrease of 1.3%. Such programs typically require up-front payments for services for a broad population, and there is little evidence of substantial cost offsets.

We also examined “value-based” insurance designs, in which drug copayments for patients with certain chronic diseases are reduced to give them an incentive for taking their medications regularly. Substantial evidence suggests that lower copayments lead to better adherence to drug regimens among patients with chronic diseases, but the effect on total health care spending would probably be small, since the resulting reduction in the use of hospital and other services would be relatively modest.

Overall, our analysis indicates that it is possible to reduce spending on health care services, although numerous political and implementation barriers stand between these policies and actual savings. All savings represent lost income for somebody, and affected stakeholders have successfully blocked, weakened, or circumvented past attempts at cost control. Widespread implementation of policies is also hampered by the heterogeneity of practice settings, with their varying data systems, organizational forms, and degrees of readiness to change. As reflected in the uncertainty of our estimates, none of these options have a strong history of reducing spending in full-scale national implementation. Skeptics note that these potential savings targets have been identified in the past, but solutions have eluded us. To achieve substantial savings, we will need to formulate and implement policies more effectively this time around. And a combination of effective approaches will probably be needed; no single option will hold growth to the increase in the GDP.

Although many of the options being considered are likely to improve the value of our health care system, only some have the potential to reduce spending. The reform legislation moving through Congress includes both promising and unpromising approaches. Furthermore, formulations of the promising options in the current bills are unlikely to be as strong as those which our highest-savings scenarios would require and so are also unlikely to achieve those levels of savings.

Because of the considerable uncertainty surrounding all options, better evaluation methods will be a critical adjunct to the experimentation that will be required. Today we have few nimble mechanisms for rapidly assessing the effects of a policy innovation. We need to develop strategies for effectively designing such interventions, evaluating them, and then deciding within an appropriate time frame whether to abandon them or systematically deploy them nationally. Many “reforms” have worked in one place, but we have almost no examples of their successful replication. If we can develop a common set of tools for design, evaluation, and assessment, we will be able to move more quickly and effectively to reject or embrace policy solutions on the basis of the evidence.

Financial and other disclosures provided by the authors are available with the full text of this article at NEJM.org.

This article (10.1056/NEJMp0910315) was published on November 11, 2009, at NEJM.org.

Source Information

From RAND Health, Santa Monica, CA.

References

1. Sisko A, Truffer C, Smith S, et al. Health spending projections through 2018: recession effects add uncertainty to the outlook. Health Aff (Millwood) 2009;28:w346-w357.
2. Eibner C, Hussey P, Ridgely MS, McGlynn EA. Controlling health care spending in Massachusetts: an analysis of options. Santa Monica, CA: RAND, 2009.
3. de Brantes F, Rosenthal MB, Painter M. Building a bridge from fragmentation to accountability — the Prometheus Payment model. N Engl J Med 2009;361:1033-1036.
4. Budget options, volume 1: health care. Washington, DC: Congressional Budget Office, 2008. (Accessed November 5, 2009, at http://www.cbo.gov/doc.cfm?index=9925.)