Many published reports have noted that there are too many specialists and probably too few primary care physicians both nationally and regionally.1–4 Studies measuring physician density have been used repeatedly with little apparent prognostic value. Most recently, there have been documented situations where physicians in certain specialties have not found desired employment after completing residency,5–8 which has concerned many medical educators and policymakers.
Anecdotal evidence indicates that graduates of residency programs at the University of Washington School of Medicine (UWSOM) appear to locate desired positions in their communities of choice without unusual difficulty. The UWSOM is the main regional medical education institution, producing 89.7% of residency graduates in the five-state WWAMI region (Washington, Wyoming, Alaska, Montana, and Idaho). The difference between the results of previous national studies and the apparent demand for physicians in our area raises questions about the appropriate public policy and educational response.
Using data that are based on actual demands for physicians can greatly improve the appropriate responsiveness of medical educators and can focus the efforts of policymakers on plans and programs that reflect the actual market for physicians. If there are regional variations, national and local policies need to account for these differences.
Ascertaining demand is difficult given a paucity of instruments that have been tested and validated. Using documentation from program directors, studies have assessed the ability of graduates to get their desired employment.6,7 Unfortunately, follow-up surveys with the graduates themselves show a discrepancy between the directors' and the graduates' assessments of potential placement in desired employment.8 There is also no documentation of whether this measure validly predicts the actual placement of graduates, not to mention whether the measure can be considered a reasonable proxy for demand.
Other studies have used the self-reported recruiting by physicians in practice as an indicator of demand.9 On face value, this appears to be a good measure of demand, but it is unclear how serious the recruiting physician is in the recruitment process and how much the recruiter is willing to pay. A recruitment offering that is less than the market rate for a physician may result in an unfilled recruitment and may not be a good measure of actual demand. In this context, demand may be best measured by the availability of positions at the market salary. However, this recruiting indicator may or may not be reflective of actual market-rate recruiting, and no study have been done validating this measure as a proxy for demand.
The only other indicator of demand available in the literature is the use of the volume of advertisements in journals read by physicians.10 This measure is a more general indicator and does not give information about regional demand or the specific demand for graduates of a particular institution.
It appears from the available studies that the demand for the graduates of one region have not been evaluated. No study has been done to compare the demand for graduates of all residency programs in a region by following up once the graduates have entered practice. This survey of graduates of the UWSOM residency programs was done to measure the apparent demand for physicians trained at the UWSOM and its affiliated institutions, and to assess whether the national assumptions regarding excess in the physician supply have decreased the demand for graduates.
The primary data-collection effort was a written questionnaire sent to graduates of all residency programs affiliated with the UWSOM. The questionnaire was developed by the authors and was based on other follow-up studies done by the UWSOM and its Family Medicine Network.
The questionnaire asked about demographics, practice location and type, and managed care activity, and included two questions about demand. The first question about demand gauged the ability of recently graduated residents to find employment of the type and in the location of their choice. It asked whether and why the respondent had been unable to find employment in his or her specialty and location of choice within two years of completing residency. (The two-year time frame was used to be consistent with previous surveys and because studies have shown that many graduates take more than one year to settle into their practices of choice, with a 20% rate of practice-changing noted in the first two years of practice.11) The second question about demand asked whether the practices of graduates of the UWSOM residency programs residing in the Northwest were recruiting any physicians. If they were recruiting, the question asked what type of physician they were recruiting. This proxy has been used in surveys of family physicians within Washington State as an indicator of employment opportunities.9
A single mailing of the questionnaire was sent to all residents (not fellows) in all departments completing their residency programs between 1975 and 1995 (n = 3,824). The participants were informed that, following data aggregation and analysis, the results would be returned to the departments and to the school of medicine for program planning. All surveys that were returned as undeliverable were recorded in a separate database of non-respondents, which produced a list of non-responders divided by apparently valid and invalid addresses. A stratified sample of 200 non-responders with apparently valid addresses, drawn in the same proportion by department as the original mailing, was sent a follow-up questionnaire. Only the key questions about demand and demographics were asked in the follow-up survey. Analysis included frequencies and descriptive statistics compiled by department and for the entire sample. We used chi-square statistics to test certain interrelationships.
A total of 1,923 (50.29%) graduates responded to the initial mailing. The respondents were sorted as generalists (family medicine, internal medicine, and pediatrics; n = 1,839, 46.6%), and specialists (all other programs; n = 1,985, 53.7%) for the evaluation of ability to locate a position. In assessing recruitment, all respondents were assessed by department, but are reported here in groupings of generalists, support specialists (anesthesiology, radiology, and pathology) and all other specialists (neurology, neurosurgery, general surgery, obstetrics and gynecology, ophthalmology, orthopedics, otolaryngology, psychiatry, rehabilitation, and urology) to maintain confidentiality. The years of residency completion among the respondents were evenly arrayed over the target years of 1975 to 1995. Of the non-responders, 56 of the 200 (28%) returned follow-up questionnaires.
The practice locations of the graduates were mainly in the WWAMI region. Of the respondents, 936 (50.3%) were in private practice and 393 (18.8%) were in full-time academic positions. Single-specialty groups (30.9%) and multispecialty groups (20.6%) were the most common practice settings. A small but significant proportion of respondents (9.5%) were employed in underserved locations, defined as health professional shortage areas, community and migrant health centers, or Indian Health Service sites.
About 50% of the volume of practices of graduates working in Washington, California, Oregon, and other non-WWAMI sites were paid through managed care contracts, defined as capitated or discounted fee schedules. Alaska, Montana, and Wyoming have much less penetration of managed care, with about 13% to 18% of volume in managed care.
Almost all graduates had found employment in their specialties of choice. Table 1 lists the physicians who had not been able to find employment in their specialties and places of choice. The few personal reasons that graduates cited in the comment sections for not attaining such positions related to family obligations, such as spouse's employment or child rearing. Those graduating in the 1990s were not significantly different from graduates of earlier years, and the responders and non-responders were not significantly different in their abilities to find positions.
Table 2 summarizes the findings for the likelihood that physician graduates of UWSOM programs are recruiting more physician(s) into their practices. Overall, 30% of the respondents working in the WWAMI region were recruiting physicians. Generalist practices were recruiting more than were specialist practices. Recruitment was not different between urban and rural locations. The breakdown by specialty type appears in Figure 1 (departments are masked for confidentiality).
While the generalists were more active in recruiting physicians, one fourth of all specialty practices that had graduates of the UWSOM were recruiting. Almost all of these specialty practices were in urban areas. Also, the rural and urban generalist providers were recruiting at similar rates, but the largest numerical demand for generalist physicians was in urban communities. Four times as many urban-based as rural-based generalist practices were recruiting. The follow-up survey of non-responders demonstrated similar (no significant difference) recruiting activity.
Measurements of physician workforce demand provide information for medical educators as they contend with a changing physician marketplace, but previous findings and their applicability to the Northwest region of the country are confusing. The studies done by Miller et al.6–8 queried program directors about the ability of graduates in 1995 and 1996 to find employment, and surveyed pre-graduates in 1997. Their findings in the latter survey raised questions about the methods of the two earlier surveys. Our study method differed by questioning graduates later in their careers, which allowed comparisons of graduates over 20 years.
Our findings suggest there is a demand for the physician graduates of the regionally based graduate medical education programs in the Northwest. The demand is proportionally similar in rural and urban communities, but the numerical demand is four times greater in urban areas than it is in rural areas, for both specialists and generalists. Physicians trained at the UWSOM and affiliated programs are able to locate employment in their types of practice and communities of choice, and one third of the graduates of the residencies already in practice in the Northwest are actively recruiting.
Compared with rough measures of oversupply or limited demand used in other studies, the demand found in this study is not consistent with the expected finding. Other studies using benchmarks, need-based standards developed from current health maintenance organizations' staffing, and the demand for physicians would predict that physician graduates in the Northwest should be having more difficulty finding employment and fewer practices should be recruiting.4,6–8,12,13 The discrepancy between our findings and these predictions could be due to the limitations of this survey, real differences in demand, or differences in the physician market in the Northwest.
However, closer scrutiny does not support an outlier health care market in the WWAMI region. The health care market in the WWAMI region is divided between Washington State, which is very similar to the national norms, and three rural states. (For the purposes of this part of the discussion, only the states of Washington, Alaska, Montana, and Idaho are referred to, because these were the states in the network at the time of the residents' graduation.) The population-to-physician ratios in Washington parallel the national averages (Washington, 1990 = 1:415, 1998 = 1:373; U.S., 1990 = 1:422, 1998 = 1:357). Other WWAMI states have higher numbers of people per physician (1990 = 1:552-646; 1998 = 1:451-564).14 Washington State had a health maintenance organization market share of 27% (31% nationally) in 1997.15 Alaska, Montana, and Idaho had health maintenance organization market shares of 0-5%. The percentage of women physicians in the WWAMI states was similar to the national average (24.7%), with 28% women in Washington, 31% in Alaska, 16% in Montana, and 21% in Idaho.16 The health care market in Washington, the predominant segment of the physician market in this region, is relatively similar to the national norms in these important factors that affect productivity. The other states in the region differed from the norm and also influenced the findings in this study, but they constituted only 18% of the respondents.
There are further limitations to our study. The response rate could have been much better if we had followed up more on the original survey. The small non-responder survey appeared to confirm the findings of the larger survey, suggesting that the non-responders were not different from the responders, but to be generalizable to other settings, a higher rate of return is preferable.
The types of instruments available to study physician demand are not well studied, nor are they validated. To improve this type of survey, more work needs to be done to improve the measures. The predictive value of the supply information, such as the Dartmouth Atlas and demand information in this study, are yet to be determined. Because physician employment appears to be driven by the health care market, improvement of market-based measures would assist educators and policymakers. Some of these measures are being used nationally, but their accuracy needs to be confirmed.
The fact that new providers accepted positions does not ensure that they were satisfied with the conditions of that employment. This survey did not query whether the physicians had obtained positions in their first-choice locations, or whether they received compensation that was satisfactory.
From our perspective at one regional medical school, the findings of this survey support the anecdotal information we are receiving from our graduates. The respondents are getting employment with little difficulty in the fields and communities of their choice. There are practices in the region (albeit we surveyed our own graduates only) that are recruiting in urban and rural areas and for specialists and primary care physicians. The demand found in our study is statistically higher for primary care physicians than it is for specialist physicians, and is similar for rural and urban areas. There appears to be substantial demand for physicians in all of these categories. The numericall demand in urban areas is four times higher than it is in rural areas. This level of demand is not similar to that described in national surveys, and for that reason, we feel the survey's results will be helpful for program planning for this one regional medical school. The findings are contrary to the expected outcome based on other attempts to predict supply and demand for physicians in the Northwest and nationally.
While this survey is helpful, a number of questions need to be answered before the supply-and-demand indicators currently found in the literature are used to predict local, regional, or national demand, or to guide policy decisions concerning physician production. Are there regional differences in demand that are not captured in national assessments? Are there local factors that make a graduate of a particular program more desirable than a graduate of another program, despite being in the same specialty? Should we use demand as a more responsive measure than physician-to-population ratios during policy development? Are market forces affecting the demand for physicians differently in different communities? Do we have accurate measures of demand?
1. Goodman DC, Fisher ES, Bubolz TA, Mohr JE, Poage JF, Wennberg JE. Benchmarking the US physician workforce: an alternative to needs-based or demand-based planning. JAMA. 1996;276:1811–7.
2. Pew Health Professions Commission. Critical Challenges: Revitalizing the Health Professions for the Twenty First Century. San Francisco, CA: University of California, San Francisco, 1996.
3. Council on Graduate Medical Education. Fourth Report; Recommendations to Improve Access to Health Care through Physician Workforce Reform. Rockville, MD: U.S. Department of Health and Human Services, 1994.
4. Weiner JP. Forecasting the effects of health reform on US physician workforce requirement, evidence from HMO staffing patterns. JAMA. 1994;272:222–30.
5. Meyer GS, Jacoby I, Krakauer H, Powell DW, Aurand J, McCardle P. Gastroenterology workforce modeling. JAMA. 1996;276:689–94.
6. Miller RS, Jonas HS, Whitcomb ME. The initial employment status of physicians completing training in 1994. JAMA. 1996;275:708–12.
7. Miller RS, Dunn MR, Whitcomb ME. Initial employment status of resident physicians completing training in 1995. JAMA. 1997;277:1699–704.
8. Miller RS, Dunn MR, Richter TH, Whitcomb ME. Employment-seeking experiences of resident physicians completing training during 1996. JAMA. 1998;280:777–83.
9. Hart LG. Family Physician Survey. Olympia, WA: Washington Academy of Family Physicians, 1996.
10. Seifer SD, Troupin B, Rubenfeld GD. Changes in marketplace demand for physicians: a study of medical journal recruitment advertisements. JAMA. 1996;276:695–9.
11. West PA, Norris TE, Gore EJ, Baldwin LM, Hart LG. The geographic and temporal patterns of residency-trained family physicians: University of Washington family practice residency network. J Am Board Fam Physicians. 1996;9:100–8.
12. Dartmouth Medical School. Center for the Evaluation of Clinical Sciences. The Dartmouth Atlas of Health Care. Chicago, IL: American Hospital Publishing, 1996.
13. Dartmouth Medical School. Center for the Evaluation of Clinical Sciences. The Dartmouth Atlas of Health Care in the Pacific States. Chicago, IL: American Hospital Publishing, 1996.
14. Physician Characteristics and Distribution in the U.S. 2000 Edition. Chicago, IL: American Medical Association, 2000.
15. Hoechst Marion Roussel, Inc. Managed Care Digest Series. Kansas City, MO: Hoechst Marion Roussel, 1997:18–21.
© 2001 Association of American Medical Colleges
16. Doescher MP, Ellesbury KE, Hart LG. The distribution of female generalist physicians in the United States. J Rural Health. In press.