Following the third straight year in which the Centers for Medicare and Medicaid Services estimated the growth in national health expenditures to be a record-low 3.9 percent, considerable speculation on the causes of slower spending growth has come from a variety of sources. There seems to be a consensus among actuaries, academics, and other analysts that the recession and the associated increase in unemployment and decline in insurance coverage led individuals to cut back on their use of health care services. (See here, here, here, here, and here.) But, while the recession is clearly associated with the dramatic slowdown in spending growth from 2007-2009, there is also evidence that the slowdown in spending preceded the recent recession and seems to be continuing during the modest economic recovery.
Observers of this more general trend have begun to suggest that fundamental structural changes in the health system are playing a role in recent spending trends. The ability of some high profile providers and health systems to achieve high quality outcomes with greater efficiency has garnered a lot of attention and some suggest that more salaried employment of physicians could be altering the practice patterns that developed under a fee-for-service system. Others have pointed to patient-centered medical homes, accountable care organizations, and other payment and delivery system reforms as potential contributors to the slowdown in spending growth. The Obama administration has also argued that the Affordable Care Act has started to have a moderating effect on spending growth.
The extent to which the economy versus broader systemic changes has been driving slower spending growth has enormous implications for forecasting future spending trends. If the economy has been the primary driver of recent trends, we should expect spending growth to return to historically high levels as the economy recovers. The Congressional Budget Office (CBO) and the CMS actuaries have revised their Medicare and Medicaid forecasts downward to reflect the latest trends, but both entities seem to suggest that spending growth over the long term will return to historical levels. If, however, more structural changes are at work, then perhaps there is reason to be hopeful that health care spending growth will continue at a rate much closer to the rate of growth in the economy.
Until recently, despite the implications for future spending forecasts, there have been few attempts to systematically estimate the factors that contributed to the slowdown in spending growth. In a recent report by the Kaiser Family Foundation and the Altarum Institute, analysts modeled the effects of macroeconomic factors on growth in national health spending. They estimate that 77 percent of the recent decline in health spending growth can be attributed to economic forces, as measured by gross domestic product (GDP) and inflation. While this study provides some useful insights, additional analysis is needed to tease out the drivers of spending trends in more detail.
Other Perspectives On The Spending Slowdown
Two studies in this month’s issue of Health Affairs extend this line of inquiry and further attempt to disentangle the role of cyclical versus structural factors in recent spending trends. Michael Chernew, Alexander Ryu, and colleagues argue that if the spending slowdown occurred primarily as a result of job loss and deterioration in benefit generosity (i.e., increased cost sharing), spending growth could revert to higher levels as the job market recovers and benefits improve. Their analysis shows, however, that for a population not affected by job loss, spending growth slowed considerably. Further analysis holding benefit generosity constant also showed slower spending growth. They conclude that these results are suggestive of a more permanent slowdown and cite “health reform, payment changes, and transformation of the delivery system” as likely contributors to longer lasting effects.
In their paper, David Cutler and Nikhil Sahni argue that the medical spending slowdown can only be partially explained by the recession, Medicare policies, and coverage changes. They rightfully argue that the slowdown preceded the recession and further demonstrate that medical spending has slowed more than the drop in income in the most recent recession would predict. They also find that changes in the coverage distribution explain a rather small proportion of the slowdown.
Since the factors that are quantifiable account for only 45 percent of the slowdown in spending growth, the residual 55 percent is as yet unexplained. Some of the proposed explanations for the residual include a sharp slowdown in prescription drug expenditures and new developments in imaging technology. Substantial increases in cost sharing for private payers are also cited as a likely driver of slower cost growth. Finally, a number of interesting examples of cost savings achieved by major hospital systems are considered as evidence that improved provider efficiency has contributed to the slowdown.
The Chernew and Cutler analyses make a compelling case that the recession, and its associated job losses and coverage changes, are not the only factors contributing to the slowdown in health spending growth. Both suggest that there is reason to believe that more fundamental changes have occurred and that this implies that future spending growth will remain modest. If they are correct, this has enormous implications for business and household spending, as well as federal and state budgets. However, if a collection of changes did occur to slow the growth in health spending, the question remains as to what caused them to materialize with enough strength and sustainability over this period to actually bend the curve.
Our Take On The Spending Slowdown
In a recent paper, we joined the analysts and researchers questioning the causes and sustainability of the slowdown in spending growth. Our study reviews the trends in health spending growth over the last decade and shows that growth began to slow well before the most recent recession. Private insurance spending peaked in 2002 and declined through 2009, with slight increases in 2010 and 2011, but some of the decline reflects the drop in private coverage.
The slowdown in health spending growth was not limited to private payers, however. Medicare spending generally declined between 2004 and 2010, after excluding the effect of adding the Medicare Part D drug benefit, with part of the decline in Medicare spending growth due to the influx of a younger and less costly population. Medicaid spending growth also declined during much of the period. Even the 8.8 percent increase at the peak of the recession in 2009 largely reflected increases in enrollment. Per enrollee Medicaid spending growth was relatively slow throughout the decade because of faster growth among lower cost enrollees and aggressive state actions to control costs.
A decade of slow economic growth. We discussed many of the same drivers of the trends described above as those cited by Chernew, Cutler and others, including the role of high-deductible health plans, Medicare payment policies, Medicaid cost containment efforts, prescription drug trends, and provider efficiencies. While these and other structural changes could have been a spontaneous response to the consistent growth in health care costs that exceeded growth in incomes, it is likely that there were other factors at play. Namely, there has been a decade-long period of slow economic growth over which a decline in real incomes and a shift towards less generous insurance arrangements has occurred. These economic pressures seem to have slowed the growth in provider revenues and forced efficiencies. Some of the more recent payment and delivery system reforms could well build upon these efforts and help sustain the slow growth rate.
From 2000 to 2011, the U.S. economy had a period of remarkably slow economic growth. The decade began with a recession, had 3 to 4 years of modest recovery, and then in 2007 the nation entered the Great Recession and unemployment surged. Exhibit 1 shows that the declines in real incomes began in 2000, never fully recovered at the peak of the economic recovery in the middle of the decade, and declined thereafter. Real median household incomes in 2011 are about 10 percent below their 2000 levels and real per-capita incomes about 5 percent below 2000 levels. Our analysis also finds that all of the net population growth between 2000 and 2011 was among those below 200 percent of poverty. The net change in the population includes the net effect of movements up and down in the income distribution, but also births and deaths, and in and out migration.
Changing patterns of coverage. We also argue that coverage changes had a major influence on the slowdown in health care spending growth. The proportion of the population with employer-sponsored insurance (ESI) declined from 69.0 percent in 2000 to 58.0 percent in 2011 (exhibit 2). Overall, 14.2 million fewer people had ESI in 2011 than in 2000, despite an increase of 21.3 million in the U.S. population. The decline in ESI led to an increase in Medicaid enrollment of 19.3 million individuals. In 2000, 8.4 percent of the nonelderly population had Medicaid coverage; by 2011, 15.0 percent were in the program. Similarly, the percentage of the population that was uninsured increased from 14.8 percent to 18.0 percent between 2000 and 2011.
At the same time, there has been an increase in the elderly population because of the aging of the baby boom generation. By 2010, the Medicare population was growing at about 2.4 percent per year, much faster than the overall population growth of about 1 percent.
This movement from private to public or no coverage is significant. Private insurance plans pay hospitals about 30 percent and physicians about 20 percent above Medicare. Medicaid payments are lower and payments by the uninsured are lower still. Simply stated, about 15 percent of the nonelderly population has moved into insurance arrangements that pay approximately 25 to 40 percent below that of private coverage. Furthermore, utilization by the uninsured is considerably lower than that for individuals with private coverage. Thus, we calculate that if the coverage distribution had remained constant, the average annual growth rate from 2000 to 2010 would have been higher by half a percentage point.
Ultimately, we agree with both the Chernew and Cutler papers that the recession alone was not responsible for the slowdown in health spending growth, but further argue that economic change over a much longer period of time caused the structural changes that we and others have observed. Specifically, there was a decade-long decline in real incomes and a shift from private to public coverage that reduced provider revenues. Provider responses to the slow revenue growth then resulted in behavioral changes that further reduced cost growth.
The Crucial Sustainability Question
The question remains whether the slower spending growth is sustainable. History suggests that high growth rates are likely to return. If the economy improves and coverage is expanded because of the Affordable Care Act (ACA), the forces that have resulted in slow cost growth may diminish. But there are reasons to believe otherwise. First, the projections for the economic recovery are relatively modest. Relatively low growth rates – 3.4 to 3.6 percent – are forecast for the next five years. After 2018, even slower economic growth is expected – 2.2 percent per year between 2019 and 2023. While insurance coverage is expected to expand substantially under the ACA, those gaining coverage will largely be enrolled in Medicaid or subsidized exchange plans. Medicaid programs can be expected to continue to be aggressive in their cost-containment efforts. Competition within exchanges is likely to be intense, meaning that these will be relatively low-cost plans as well. A modest economic recovery and increasingly cost-conscious payers are therefore likely to place continued pressure on providers to become even more efficient.
Further, the ACA is tightening Medicare payments to hospitals and other providers. A large number of Medicare demonstration programs, including bundled payment mechanisms, value-based purchasing, and shared savings programs are being implemented. All of these factors taken together suggest that a return to a high historic growth rates in health care spending may not materialize. To sum up, we, like Michael Chernew and David Cutler and their coauthors, are cautiously optimistic.