Since 1973, when Jack Wennberg published his first paper describing geographic variations in health care, researchers have argued about both the magnitude and the causes of variation. The argument gained greater policy relevance as U.S. health care spending reached 18 percent of GDP and as evidence mounted, largely from researchers at Dartmouth, that higher spending regions were failing to achieve better outcomes. The possibility of substantial savings not only helped to motivate reform but also raised the stakes in what had been largely an academic argument. Some began to raise questions about the Dartmouth research.
Today, the prestigious Institute of Medicine released a committee report, led by Harvard’s Professor Joseph Newhouse and Provost Alan Garber, that weighs in on these issues. The report, called for by the Affordable Care Act and entitled “Variation in Health Care Spending: Target Decision Making, Not Geography,” deserves a careful read. The committee of 19 distinguished academics and policy experts spent several years documenting the causes and consequences of regional variations and developing solid policy recommendations on what to do about them. (Disclosure: We helped write a background study for the committee).
But for those trying to make health care better and more affordable, whether in Washington or in communities around the country, there are a few areas where the headlines are likely to gloss over important details in the report. And we believe that the Committee risks throwing out the baby with the bathwater by appearing, through its choice of title, to turn its back on regional initiatives to improve both health and health care.
What the committee found
The report confirmed three core findings of Dartmouth’s research. First, geographic variations in spending are substantial, pervasive and persistent over time — the variations are not just random noise. Second, adjusting for individuals’ age, sex, income, race, and health status attenuates these variations, but there’s still plenty that remain. Third, there is little or no correlation between spending and health care quality. The report also effectively identifies the puzzling empirical patterns that don’t fit conveniently into the Dartmouth framework, such as a lack of association between spending in commercial insurance and Medicare populations.
The committee also confirmed earlier work by Harvard investigators showing that, for the commercially insured population, variations in the prices paid by private health plans explain most of the variations in private insurance spending. The committee deserves considerable credit for deepening our understanding of this irrational world of pricing commercial health care services. Yet as the report finds, even in the commercially insured population, there are substantial differences in utilization rates across regions. We would therefore argue that for commercial populations both price and utilization deserve attention, especially because in many regions, avoidable utilization may be easier to address than price.
It is Medicare spending growth, however, that represents arguably the greatest risk to the financial health of the U.S Treasury, and in Medicare, variations are almost entirely the consequence of utilization of services, not prices. The report finds that the single largest component of the variation in Medicare spending across regions that remains after risk and price adjustment is due to post-acute care (including skilled nursing facility services, home health care, hospice, inpatient rehabilitation and long term acute care). These services have also been a major source of growth.
But this focus on post-acute rather than acute hospital and physician services misses the key point that dysfunctional regional health systems are characterized both by hospitals providing fragmented and expensive care and by a large and thriving post-acute care sector ready and eager to absorb the discharged patients. For example, Joan Teno and colleagues at Brown University have established the strong association of inpatient treatments with no medical benefit, such as feeding tubes for people with advanced dementia, with high rates of regional resource use.
Which brings us to…..
The IOM committee’s policy recommendations: Where they hit the mark …
The committee makes five policy recommendations — and we agree with all of them. First, they call for making more and better data available, on both Medicare and commercial populations. Second, they recommend that CMS continue to test new payment models that encourage clinical and financial integration. Third, they call for timely and iterative evaluation of current and new payment reforms so that improvements can be made to the models. Fourth, they call on Congress to grant CMS the flexibility to accelerate the transition to value-based payment models as successful approaches emerge.
The fifth recommendation focuses on whether Congress should adopt a geographically based payment adjustment. When the committee was first mandated by Congress in the midst of health care reform in 2010, congressional members from regions with lower costs espoused a “Value Index” in which Medicare would reward low-spending regions with higher reimbursements, at the expense of high-spending regions. The committee concluded that payment mechanisms should not be tied to region, but instead targeted to individual providers, rightly criticizing the Value Index approach as not providing institutions and systems with the right incentives to reduce costs and improve quality.
… and where they fall short: Geography does matter
We believe, however, that the committee, by subtitling the report “Target Decision-makers, Not Geography,” will confuse the media and casual readers (for example, those who don’t make it to page 3-3 in the full report) by appearing to cast doubt on the promise of geographic and regional efforts to improve the quality and efficiency of U.S. health care.
As the late Nobel-Prize winning economist Elinor Ostrom has emphasized, successful management of complex social problems can best be achieved through sustained collaboration among diverse stakeholders, often across traditional political boundaries. She demonstrated that cooperative agreements are often the most effective approach to solving the kinds of problems we face in health care. Among these are the natural instincts of physicians and hospitals within local health care systems to protect their financial health by expanding capacity and defending market share, whether by opening new cardiac centers when the one at the nearby hospital is perfectly adequate, or by buying proton accelerators that will be used to treat conditions where they offer no demonstrated benefit.
The rationale for a geographic focus on health care reform is strong: the factors that determine population health are largely local, rooted in the environmental, social, economic, and behavioral determinants of health. Many of the factors that influence health care quality and costs are also local, including local supply, pricing behavior, and the relative emphasis of providers on profit. For example, in the widely cited New Yorker article by Atul Gawande, Medicare utilization in McAllen was found to be nearly twice as high as that in another Texas border town, El Paso, despite the existence of multiple hospitals in both McAllen and El Paso, nearly identical Medicare prices, and common Texas malpractice laws.
Many regional multi-stakeholder initiatives have been established. Although most began with a focus on quality, many are beginning to act more broadly to both improve health and lower costs: Three examples include Pueblo Colorado (Regional Triple Aim), Akron, OH (Accountable Care Community), and the Atlanta Regional Collaborative for Health Improvement (focused on driving provider transitions to global payment, capturing savings, and reinvesting in strategic population health initiatives).
While the IOM Committee is exactly right to call for improved financial incentives for health care providers, we should also remember that both health and health care are local. Geography matters.