|Home | About | Journals | Submit | Contact Us | Français|
U.S. medical schools are in the midst of a significant expansion in their capacity to train new physicians (Cooper, Stoflet, and Wartman 2003), with recent reports from the American Association of Medical Colleges indicating plans that would increase the aggregate U.S. medical school class size 17 percent by 2012 (AAMC 2007). These efforts have been driven by, among other things, official calls from the Council on Graduate Medical Education for a 15 percent expansion in medical school enrollment (Council on Graduate Medical Education 2005), and the AAMC, which has reportedly called for 30 percent more U.S. medical school graduates annually (Goodman and Grumbach 2008). Coupled with activities aimed at expanding funding for residency programs, expansion of medical school capacity would tend to expand physician supply.
Efforts to expand physician supply reflect a variety of factors including concerns about potential doctor shortages as the population grows and ages, the hope that more physicians will help address concerns about quality of care, and the view that adding to the pool of physicians could help address the fact that physicians are currently not well distributed around the country. While these efforts may be well intentioned, they deserve considerable scrutiny, as there are many reasons to be concerned that expansions of supply that are not carefully considered may not meet their stated goals.
In this issue of HSR, the research by Gravelle, Morris, and Sutton highlights one key aspect of discussions about physician supply: the specialty distribution. They studied the relationship between the supply of family practice physicians in the “local areas” that are part of the health care system in England, and the self-reported health of individuals living in those areas. Using three different measures of health status, they report evidence that increases in the supply of family practice physicians are associated with improvements in health. One of the notable aspects of this study is its recognition of the potential for reverse causality—it may be that aspects of the health of the local population drive changes in family practice physician supply, not the other way around—and for unobserved characteristics of areas to bias results. After taking steps to deal with these challenges, the authors find that a 10 percent increase in the supply of family practice physicians in an area is associated with a 6 percent increase in the probability that area individuals will report being in very good health. While this research and other related studies point out the importance of generalist physicians for improving health care delivery, relatively little attention has been paid to issues of specialty mix amid current efforts to expand physician supply. Beyond the specialty mix, other issues also deserve attention, including whether the number of physicians should be expanded at all and the importance of the geographic distribution of physicians.
The main arguments for expanding physician supply draw on existing evidence about the current number of physicians and forecasts about future population and economic growth (e.g., Cooper et al. 2002). By extrapolating from past experience, projections for future physician demand assume that existing approaches to utilizing physician care are the right models to rely on for the future—if we have kept the physicians we have busy, then it must be that we will need more in the future.
However, it is not difficult to conclude from available studies that existing patterns of physician supply and utilization are not models for efficient and effective health care that should be perpetuated into the future. There are currently substantial variations in physician availability from one part of the country to another—among the 200 largest Metropolitan Statistical Areas, the number of physicians per 100,000 persons varies by a factor of more than three. And research from the Dartmouth Atlas, among other places, suggests that these increases in physician supply are not regularly associated with improvements in quality measures or health outcomes (e.g., Fisher et al. 2003a, b).
We should expect that expanding physician supply will drive up utilization of physician services. Studies over many decades report links between expanding health care capacity and increases in utilization. In one early piece on surgeons, John Bunker, an anesthesiologist from Stanford University, published an article in the New England Journal of Medicine in which he studied the performance of surgery in the United States and in England and Wales (Bunker 1970). His specific interest was in the then-current debate about whether there was a shortage of anesthesiologists in the United States, but his study also presented information about the supply of surgeons and the use of surgery. In 1967, he reported, there were 39 surgeons per 100,000 persons in the United States, and less than half as many—18 per 100,000—in England and Wales. America, he also found, had a much higher rate of surgery, about 7,400 surgeries performed per 100,000 people in 1965, about twice the 3,770 reported for 1966 in England and Wales.
Bunker cast around for explanations from several areas, considering the extent of group practice, public health practices, financial incentives for performing services, and the “disproportionate number of physicians [in] the practice of surgery.” Though Bunker's data were not perfect, and not everyone accepted his conclusions (e.g., Rutkow 1987), his findings provide an early example of a link between physician supply and utilization. Since then, others have furthered the point. In 1973, Wennberg and Gittlesohn reported that across 13 hospital service areas in Vermont, the supply of surgeons was correlated positively, and relatively strongly, with the number of surgeries performed in the area (Wennberg and Gittlesohn 1973). In a 1978 study, Fuchs estimated that 10 percent increases in the surgeon/population ratio resulted in about a 3 percent increase in per capita utilization of surgeries (Fuchs 1978). A few years later, Sloan and Schwartz (1983) concluded that a 10 percent increase in the supply of physicians would be associated with a 4 percent increase in spending for physician services. In the last few years, Fisher and colleagues argued that in the Medicare program, having more specialists per capita in an area is associated with higher surgery rates and higher procedure rates (Fisher et al. 2003a, b).
Debate about the exact interpretation of these relationships continues. The extent to which observing that increases in supply go with increases in utilization implies that expansions in physician supply cause changes in utilization, as opposed to other factors such as variations in demand across areas that increase both supply and utilization, may never be entirely resolved. Nonetheless, the persistence of these relationships over time, the fact that they are found in a wide variety of studies, and their concordance with literature on other types of medical care suggest a real possibility that expanding supply drives utilization. Expanding utilization of physician services may be worth it if we receive benefits that would offset the costs. There are many instances in which expanded utilization has the potential to improve outcomes, but we should also be cautioned by evidence that suggests that more physician utilization does not always bring better outcomes (e.g., Fisher et al. 2003a, b).
Efforts to expand physician supply are often not targeted at specific specialties. Perhaps, given evidence about the differential needs for physicians in different specialties, they should be. Expanding supply without carefully addressing the specialty mix runs the risk of expanding the pool of specialists more than the pool of generalists—for a variety of reasons, our medical education system has a strong tendency to produce specialized physicians. This would be appropriate if evidence supported the view that expanding the specialist supply would produce benefits akin to or greater than expanding the generalist supply, but the available literature often suggests the opposite.
The literature on physician supply suggests that specialty matters. Most studies, including the work by Gravelle and colleagues in this issue, suggests that expansions of primary care, an area in which the United States has historically lagged other countries, can be beneficial while expanding the supply of specialists is not always so. Shi compared capacity and outcomes across states, and found that higher primary care capacity was correlated with lower overall mortality rates, lower death rates due to diseases of the heart and cancer, longer life expectancy, lower neonatal death rate, and lower rates of low-birth-weight births (Shi 1994). Having more specialist physicians, by contrast, was statistically significantly related to higher total mortality, more deaths due to heart diseases and cancer, shorter life expectancy, higher neonatal mortality, and higher rates of low-birth-weight births (Shi 1994). Starfield et al. (2005) reported a similar set of findings looking across counties in the United States—higher primary care physician-to-population ratios were generally associated with lower all-cause and cause-specific mortality rates, while higher specialist-to-population ratios were generally associated with no change in outcomes or with higher mortality rates. Baicker and Chandra (2004) showed that states with the more specialists tend to rank lower in quality than states with fewer, and vice versa for generalists.
As described above, there are wide variations in physician supply across geographic areas in the United States. Even among larger Metropolitan Statistical Areas, there are a number of places with 300 or more physicians per 1,000 persons, along with areas with 100 or less. Including rural areas, the variations across areas are even wider. We also know that there are clusters of physicians in places where medical education programs are most numerous. Indeed, research suggests that historically most newly trained physicians have located in areas with the highest levels of physician supply, not the lowest (Goodman 2004).
One risk is that expanding the number of medical school slots to increase the number of physicians trained will lead to increases in the number of physicians in a relatively small number of geographic areas, which may be well supplied already. Expanding the supply of physicians should be done with an eye toward addressing the geographic imbalances that have crept into our system, but which could easily go unaddressed in generalized efforts to expand supply.
Because the United States now has more residency slots than can be filled by the graduates of U.S. medical schools, one possibility is that expanding the enrollment of U.S. medical schools will just displace some of the international medical graduates (IMGs) who now fill these slots, with no net change in the total number of physicians entering practice from U.S. residencies each year. Even without expansions in the supply of physicians, shifts in the share of physicians coming from United States as opposed to international medical schools could have implications for the geographic distribution of physicians; IMGs have, for example, been an important source of primary care in rural and other underserved areas (Hart et al. 2007). Shifts in the share of U.S. medical graduates might also affect quality of care, although available evidence does not suggest poor quality of care delivered by IMGs trained in strong residencies (Ko et al. 2005). Nonetheless, given efforts to by the AAMC to convince policy makers to raise caps on funding for graduate medical education, the potential for a net expansion in the supply of physicians should be taken seriously.
The arguments made here are not necessarily new—they all have bases in previous debates about physician supply (e.g., Goodman and Grumbach 2008), though they are easy to ignore amid the uncertainties about the needs of an expanding population and perhaps a desire to be conservative in estimates so as not to underprovide for the future. It is also true that moving toward more efficient systems for using physician care may be difficult and require non-trivial changes in the ways in which at least some doctors work. At the same time, there is real reason to be concerned that simply expanding the number of doctors in ways that could easily mainly lead to expansions of the number of specialists practicing in places with many physicians already would not make the population substantially better off. There may well be opportunities to expand physician supply in valuable ways, and existing expansion efforts should pay more attention to identifying and exploiting them. Laurence Baker, Ph.D.