Technology Growth and Expenditure Growth in Health Care
New health care technologies offer the promise of improved health and longevity, but also are widely viewed as the biggest contributor to rising health care costs in the U.S. This duality raises the question of whether new technologies are worth the cost and how the rate of health care innovation can be slowed if the costs of new technology exceed the benefits.
In Technology Growth and Expenditure Growth in Health Care (NBER Working Paper 16953), researchers Amitabh Chandra and Jonathan Skinner explore technological growth in health care and its impact on cost growth and productivity improvements.
The researchers develop a model of patient demand and supplier behavior to explain the parallel trends of technology and expenditure growth. The model is one where health spending can affect individuals' longevity and quality of life and providers care about both their patients' health and their own income. The model's key finding is that the productivity of a health care innovation depends on the shape of the health production function (which translates health spending into health outcomes), the heterogeneity of treatment effects across patients, and the cost structure (many procedures have high fixed costs and low marginal costs).
The authors use this finding to develop a typology of medical technology productivity. The first category consists of "home run" treatments that are highly cost effective and useful for nearly everyone. One example is the development of antibiotics, which were highly effective in reducing mortality from pneumonia, tuberculosis, and other diseases starting in the 1930s. Category I treatments can be expensive, so long as they are cost effective and unlikely to be used on patients who will not benefit from the treatment; the use of antiretroviral drugs to treat HIV is an example in this vein.
The second category includes those technologies that are highly cost-effective in some patients but less useful for others. Despite their value to some patients, Category II treatments may have modest or poor average cost-effectiveness due to their use by many patients who experience few health gains. A leading example is angioplasty, which dramatically improves survival following a heart attack if administered within 24 hours, but yields no survival benefit and only modest functioning improvements for those with stable coronary disease.
The third category consists of treatments for which benefits are small or as yet unproven. Category III includes treatments like arthroscopic surgery for osteoarthritis of the knee, which was famously found to have no medical value in a randomized control trial where some patients received "placebo surgery," despite the fact that some 650,000 such surgeries were being performed annually at a cost of more than $5,000 each. Category III also includes treatments for which there is little scientific evidence of their value. Ethical and logistical considerations can make it difficult to conduct double-blinded trials, the gold standard for establishing the efficacy of medical treatments, and even when such trials are possible, it can take years for studies to be done.
Next, the authors ask how much of the gains in survival and cost increases over the past several decades have been driven by diffusion of each type of treatment. Using cardiovascular disease as an example, they note that 44 percent of the reduction in mortality from 1980 to 2000 was due to improved health behaviors. Another 22 percent of the decline was due to inexpensive Category I treatments such as aspirin and beta blockers, 12 percent was due to Category II treatments like angioplasty, and perhaps 10 percent was due to Category III treatments. On the cost side, the spread of Category I and II treatments appears to have contributed only modestly to cost growth, suggesting a larger role for Category III spending. Despite the rapid diffusion of "home run" technologies like beta blockers during this period, the average cost of saving an additional life-year tripled, to nearly $250,000.
Taking an international perspective, the authors note that the US is often a leader in the use of expensive technologies with unproven benefits, such as robotic surgery and proton-beam therapy for prostate cancer. Yet improvements in life expectancy in the U.S. have if anything lagged behind those in other OECD countries. This suggests the more rapid diffusion of the less productive Category II and Category III treatments in the U.S. may help to explain why it has experienced higher growth in health care spending relative to GDP without commensurate gains in life expectancy.
Lastly, the authors turn to the question of how to control health care costs. In other countries, regulatory boards use cost factors in setting standards for the use of Category II and III treatments. While discouraging the building of costly facilities such as MRIs and ICU beds is theoretically possible, this approach "would require a tectonic shift in the U.S. regulatory and policy environment."
Making consumers responsible for a larger share of costs is another approach. As consumers are often unaware of the costs and benefits of different treatments, charging higher prices for Category III and some Category II treatments and lower prices for Category I treatments could help consumers to make more appropriate treatment decisions, though few insurance plans are currently structured this way.
Doing more comparative effectiveness research could also improve the productivity of health care spending, though the fact that treatment effects can vary by patient type complicates such work. Finally, reimbursing providers based on the value rather than the volume of services provided may help to ensure that innovations are focused on cost-effective treatments. Some analysts believe that the fragmentation in the health care delivery system leads to higher costs and suggest that integrated delivery systems (like the "accountable care organizations" cited in the 2010 health care reform law) could be part of the solution, though their ability to promote cost-effective treatments has not yet been established.
The authors conclude "U.S. growth in health care costs is neither inevitable nor necessarily beneficial for overall productivity gains. Instead, cost growth is the aggregated outcome of a large number of fragmented decisions regarding the use and spread of both old and new health technologies." They warn "there does not appear to be a single magic bullet to solve the health care problem. The extent of waste in the U.S. could, ironically, prove to be a boon if a fundamental restructuring of health care unleashed some of this lost productivity. The alternative to not making such changes is far more worrisome: rising political and economic resistance against tax hikes, insurance premium increases, or coverage expansion could serve as particularly inefficient brakes on both health care costs and health care innovation."
The researchers acknowledge funding from the National Institute on Aging (P01 AG19783) and the Robert Wood Johnson Foundation.
