Variations In The Costs And Quality Of Medical Care: Is More Always Better?

Elliott S. Fisher, M.D. M.P.H., 

Professor of Medicine and of Community and Family Medicine,
Dartmouth Medical School for the Dartmouth Atlas Project

America’s Health Rankings have historically focused on the decisions made by individuals, community leaders and state and national policy makers -- and the impact of these decisions on health.  This year, for the first time, the rankings have sought to look at  health care services -- their quality and cost-effective delivery -- as additional parts of the mosaic that contribute to our health.      For many years, researchers at Dartmouth have been studying how health care is delivered across the nation and have come to believe that Americans have much to gain from a better understanding of how the quality and costs of care vary across regions -- and how they can best contribute to creating a healthy population.

The care of patients with chronic illness presents a major challenge to health care systems throughout the world.  According to the U.S. Centers for Disease Control, more than 90 million Americans live with chronic disease such as diabetes, cancer and heart disease.  Of the 1.7 million Americans who die each year, seven out of ten deaths are caused by chronic diseases.  Additionally, the medical care costs for people with chronic disease account for more than 75% of all U.S. health care expenditures. (CDC, http://www.cdc.gov/nccdphp/overview.htm, cited April 25, 2006)

The care of patients with chronic illness also presents an important opportunity.  As the most recent edition of the Dartmouth Atlas of Health Care reveals, dramatic variations in treatment of Medicare beneficiaries with severe chronic illness exist across U.S. states, regions and hospitals.   (We focus here on differences at the state level, but similar differences are observed across both regions and hospitals; see www.dartmouthatlas.org for details).  This table and graph provide state-specific information on utilization, costs and quality of care for Medicare beneficiaries with serious chronic illness. To ensure that differences in utilization are not due to differences in underlying illness levels, the analyses focus on patients with at least one serious chronic illness who are in their last two years of life: for these patients we can be confident that on at least one measure -- their risk of death -- they are identically ill.  Differences in utilization, therefore, reflect variations in how similar patients are treated in different health systems.

A few key findings have emerged from our work:

Patients with chronic illness are treated very differently in different states.  The average number of days spent in the hospital by seriously ill Medicare beneficiaries during their last six months of life varies more than twofold.   States in New England, the Midwest, the Mountain states and the Pacific Northwest had low rates compared to residents of Hawaii (16.4 days), New York (16.3 days) and New Jersey (15.2 days).  Residents of Utah (7.3 days) and Oregon (7.8 days) had rates less than half the average among residents of New York and Hawaii.   Utilization rates for other services, such as physician visits and the number of different physicians seen during the last six months of life, are highly correlated with hospital stays. 

Differences in utilization drive important differences in spending.  As a consequence, the amount of money the Medicare program spent per patient varied nearly twofold.  Some of the differences in Medicare spending are a consequence of differences in the prices Medicare pays providers.  The most important factor, however, is the greater volume and intensity of care delivered in high cost states and regions.  In other words, the variations in costs are largely due to differences in the volume of discretionary services provided to similarly ill patients -- variations in how much time similarly ill patients spend in the hospital, in how often they see physicians, in how many specialists are involved in their care, and how frequently patients have tests and minor procedures. 

Variations in spending have several causes, including limited evidence, optimistic assumptions and unmanaged supply. Evidence-based medicine focuses primarily on the “what” of treatment (what drug, which surgical procedure) rather than the “how” (by whom, where delivered, over how many visits).   Current research provides no guidance on whether a patient with well-controlled high blood pressure should be seen once per month (as is the case for patients cared for by some physicians in high spending regions) or once every six to twelve months (as in other regions).  In the absence of strong evidence, other factors drive clinical decisions – including the widely held assumption that more medical care means better care.  Although this assumption is reinforced by fee-for-service payment and physician fears of malpractice --- these factors do not vary across regions.  What varies across U.S. regions and health care organizations is the supply of medical resources relative to the size of the population served.  High spending states have many more physicians and acute care hospital beds on a per-capita basis than low spending states -- and the current payment system ensures that they stay busy. 

More services don’t necessarily mean better outcomes.   In the end, the critical question is whether greater use of these discretionary “supply-sensitive” services (hospital stays, visits, specialist referrals) results in better health outcomes.   Extensive research – both across U.S. regions and among leading academic medical centers – has now documented that greater use of these services across the range of practice observed in the U.S. is, if anything, associated with slightly worse outcomes, poorer quality and lower satisfaction with care (Fisher, Annals 2003, Parts 1 and 2, Fisher Health Affairs 2004) . Physicians report that quality is worse in higher spending regions (Sirovich, Annals 2006), and, likewise, the most recent measures of the quality of hospital care (Figure 1) show no evidence that higher spending is associated with better hospital quality.  On the contrary, as in earlier work that focused on both inpatient and outpatient quality measures (Baicker Health Affairs 2005), there is a weak negative association between spending and state-level average performance on Medicare’s current measures of hospital quality (r = - 0.27, p = 0.06).  (If New Jersey, which is clearly what would be termed a “statistical outlier”, is dropped, the correlation is much stronger:
r = - 0.42, p = 0.002).   The reasons why higher spending would be associated with worse quality on these measures remains a topic of research, although a likely possibility is that the higher spending regions have more complex delivery systems (e.g. more physicians involved in a given patient’s care), and greater complexity increases the chance of errors.  

The remarkable variations in the costs and quality of care for patients with chronic disease -- and the evidence that states and regions that provide lower cost care can do so with equal or better quality and outcomes -- represent an important opportunity.  States have increasing responsibility for the financing and regulation of care.  All states play an active role managing the Medicaid program which, because of its role in financing much of long term care, has a direct impact on the costs and quality of the care delivered to the chronically ill.  States are also increasingly concerned about rising health care costs because of their responsibilities as payers for their employees and retirees.  And states have taken an active role in expanding or limiting resources for health care, through decisions to increase the number of medical schools or to constrain resource growth through certificate of need programs.   This is potentially an important role because as the Dartmouth Atlas Project (http://www.dartmouthatlas.org/) has demonstrated, capacity strongly influences both the quantity and per-capita costs of care provided to patients with chronic illnesses. (Wennberg, Health Affairs 2002)

The U.S. has made tremendous gains in understanding the underlying biological causes of disease and disability.  However, as the variations across the United States and America’s Health Rankings reveal, we have much to learn about how best to translate our knowledge into policies and clinical practices that achieve the best possible health outcomes for all at an affordable price.   Research on variations in practice can help guide these efforts. 

Baicker, K. and A. Chandra (2004). "Medicare spending, the physician workforce, and beneficiaries' quality of care." Health Aff (Millwood) Suppl Web Exclusive: W184-97. 

Fisher, E. S., D. E. Wennberg, et al. (2003). "The implications of regional variations in Medicare spending. Part 1: the content, quality, and accessibility of care." Ann Intern Med 138(4): 273-87.

Fisher, E. S. and J. E. Wennberg (2003). "Health care quality, geographic variations, and the challenge of supply-sensitive care." Perspect Biol Med 46(1): 69-79. 

Fisher, E. S., D. E. Wennberg, et al. (2004). "Variations in the longitudinal efficiency of academic medical centers." Health Aff (Millwood) Suppl Web Exclusive: VAR19-32. 

Sirovich, B. E., D. J. Gottlieb, et al. (2006). "Regional variations in health care intensity and physician perceptions of quality of care." Ann Intern Med 144(9): 641-9. 

Wennberg, J. E., E. S. Fisher, et al. (2002). "Geography and the debate over Medicare reform." Health Aff (Millwood) Supp Web Exclusives: W96-114.