[Access article in PDF]
Health Care Quality, Geographic Variations, and the Challenge of Supply-Sensitive Care
THE UNITED STATES SPENDS MORE per capita on health care services than any other country, yet there are mounting concerns about the uneven and unpredictable quality of care provided at even the "best" American institutions. In 1998, the National Roundtable on Health Care Quality reported serious and widespread quality problems throughout the American health care system, regardless of organizational setting or payment mechanism (Chassin and Galvin 1998). A review of published studies indicated that, on average, 50 percent of patients did not receive recommended preventive care; 30 percent did not receive needed care for acute medical conditions, and 40 percent went without necessary care for chronic conditions (Schuster, McGlynn, and Brook 1998). An Institute of Medicine (2001) panel concluded that safe and error-free delivery of effective care will require a fundamental redesign of the health care system.
Studies of geographic variations in medical care underscore the scope of the challenge, documenting twofold differences in Medicare spending among regions (Dartmouth Medical School 1999). There are dramatic variations in the [End Page 69] quality of care provided (Dartmouth Medical School 1999; Jencks et al. 2000; O'Connor et al. 1999), and the quality of care is unrelated to the level of spending (Wennberg, Fisher, and Skinner 2001). Recent studies have highlighted the largely unappreciated role of local health system capacity as a determinant of both the costs and the quality of care (Fisher et al. 2000; Fisher et al., parts 1 and 2, in press; Skinner, Wennberg, and Fisher 2002). Reform that ignores the influence of local capacity misses a major opportunity to improve the quality of care for all Americans.
In this article we discuss some of the conclusions of a population-based approach to measuring the quality of health care in the United States, drawing on data presented in the Dartmouth Atlas of Health Care and summarizing the findings of other recent studies (Fisher et al. 2000; Wennberg et al. 2001). We first provide a brief overview of the methods used in the Atlas project. We then propose an analytic framework for assessing quality in the American health care system, incorporating our understanding of variations in practice. We summarize the evidence about how capacity influences both the cost and the quality of care. Finally, we discuss the implications of our analysis, and the necessity of a population-based approach to monitoring and improving the quality of health care.
A Population-Based Approach to Performance Monitoring
The methods of small area analysis were first developed more than 30 years ago; they are the analytic foundation for the Dartmouth Atlas of Health Care (Wennberg and Gittelsohn 1973). The first step in small area analysis is the definition of natural health care markets. Using analyses of patients' travel patterns, we identified 306 hospital referral regions in the United States that reflect prevailing patterns of where defined populations seek hospital-based care (Dartmouth Medical School 1996). Each region contains both local community hospitals and at least one referral hospital. Regions are as diverse as Rochester, Minnesota, and Minneapolis; Salt Lake City; East Long Island, New York; Reno; and Miami. The resident populations in these regions receive almost all of their medical care from providers located within the region. The second step of small area analysis is to characterize resource levels, patterns of practice, and spending levels (chiefly, in the instance of the Dartmouth Atlas, Medicare reimbursements) for the residents of each region. Comparing the practice patterns, resources, and spending in these regional health systems makes it possible to draw inferences about the relative importance of the several determinants of the quality and cost of care.
An Analytic Framework
We define three major categories of medical services (Table 1), distinguished by the relative importance of four factors in decision making: medical evidence, clinical theory, patient preferences, and the local supply of health...