Figure. US Census regions and divisions. Map provided courtesy of the US Census Bureau.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Resneck JS, Kostecki J. An Analysis of Dermatologist Migration Patterns After Residency Training. Arch Dermatol. 2011;147(9):1065–1070. doi:10.1001/archdermatol.2011.228
Author Affiliations: Department of Dermatology and Philip R. Lee Institute for Health Policy Studies, University of California, San Francisco School of Medicine (Dr Resneck); and Department of Survey Research, American Academy of Dermatology, Schaumburg, Illinois (Mr Kostecki).
Objective To evaluate the migration patterns of dermatologists from residency training to eventual clinical practice to assess whether certain programs or regions were more likely to populate their own areas with graduates.
Design Analysis of existing data from the American Academy of Dermatology's membership database.
Setting The United States and Puerto Rico.
Participants Graduates of US dermatology residency programs completing training before 2005 and actively practicing in 2009. Data from 7067 practicing dermatologists were analyzed.
Main Outcome Measures Distance from training to practice site and state locations of training and practice sites.
Results Almost half (43%) of dermatologists practice within 100 miles of their residency training site, although substantial variation was observed in distance (mean, 538 miles; median, 189 miles). More than 70% of graduates from all but 1 New York City program remain within a 100-mile radius, and several California and Florida programs were most likely to retain trainees in state. The Midwest was a net exporter of residents to the West and South.
Conclusions The relationships between residency and eventual practice locations are complex, but certain regions and training programs have distinct graduate migration patterns. To the extent that further expansions in residency positions are undertaken with local supply and demand of dermatology services in mind, these patterns are among many factors that should be considered.
The dermatology workforce has been challenged by shortages, which appear to be worse in selected geographic areas as measured by variation in patient access to care, difficulty in hiring associates, and the subjective impressions of dermatologists and referring physicians.1-3 The ratio of dermatologists to population size does not always predict workforce adequacy, suggesting that the demand for care may vary regionally.4,5 This finding may be related to regional variability in the prevalence of disease (eg, skin cancer), the age of the patient population, health insurance coverage, patterns of specialty use, and the popularity of cosmetic services. Also, the effective supply of dermatologists in a region may be affected by differences in work intensity, types of practice venues, and the scope of practice of local dermatologists.
In many specialties including dermatology, market forces never have completely ameliorated workforce maldistribution problems.6-11 Dermatologists emerging from training report that geographic location is the top factor in evaluating potential job opportunities and that they are usually able to find plentiful opportunities in preferred locations.12 If an underserved area also is seen as undesirable, it may struggle to attract dermatologists, even if it boasts strong patient demand and outstanding job opportunities with relatively high compensation.
As the field of dermatology has sought opportunities to increase the number of dermatologists in areas with relative shortages, one approach has been to target growth in residency programs located in underserved areas.13 This practice has been tied to an assumption that programs located in areas with shortages may be most likely to place graduates in these same areas (based on evidence from other specialties).14,15 Little is known, however, about the patterns of migration of dermatology residency graduates. One study surveyed a group of recent graduates attending a board review course and found that 53% stayed in their state of training for their first job, but that substantial regional variations occurred, with a large migration of trainees out of the Midwest and to the West coast.16 This finding suggests that expanding training programs in a particular location may not ensure a proportional increase in the number of practicing dermatologists in the same region.
This study seeks to determine in greater detail how far dermatologists typically travel from their residency location to their eventual practice site. In addition to examining regional migration patterns, we evaluate the hypothesis that these larger patterns may obscure variability at the individual program level that affects how far residents move after training.
We used the American Academy of Dermatology's membership database (which includes >90% of practicing US dermatologists) to identify cohort individuals between September 1 and 30, 2009. All board-certified or board-eligible dermatologists who graduated from US allopathic dermatology residency programs before January 1, 2005, and were actively practicing in the US were included, yielding 7997 physicians. Those who graduated in 2005 or later from residency were excluded in order to minimize the effect of recent trainees who might be less likely to have settled in a long-term practice location. Location data for residency program and current primary practice site were available for 7067 (88%) of these cohort individuals, and the remainder were excluded from analysis. Zip codes for training and practice sites were converted to latitudinal and longitudinal coordinates using the 5-Digit Zip Code Data, Commercial Edition software (ZIP Code Download, Provo, Utah), and the distance between coordinates was calculated using the spherical law of cosines. This study was reviewed and approved as exempt by the Institutional Review Board at the University of California, San Francisco.
Most dermatologists practice at a distance from their training site (mean, 538 miles; median, 189 miles), but a substantial minority remain in the same state (43%) or within 100 miles (43%) (Table 1). Female dermatologists and those who had completed training more than 20 years ago were somewhat more likely to remain closer to their training site. Practicing in a multispecialty group was associated with relocating further from one's training site.
Most graduates of training programs in the Middle Atlantic and Pacific census divisions relocated within 100 miles of their residency site (see the Figure for a map of the divisions). However, those completing training in the Mountain census division and the Midwestern and Southern regions were substantially less likely to remain within 100 miles of their training site. Those training in the Pacific division were most likely to practice in the same state as their training site. Further analysis of regional patterns (Table 2) indicates that the Midwest retained the lowest proportion of its trainees (51%) and that the South and West were the biggest recipients of trainees from the Midwest and the Northeast.
Examination of individual residency training programs showed even greater variability (Table 3). Albert Einstein College of Medicine, Columbia University, Cornell University –New York-Presbyterian Hospital, The Mount Sinai Medical Center, New York Medical College, and SUNY Downstate Medical Center –Brooklyn each had more than 70% of their graduates currently practicing within a 100-mile radius. In contrast, the Mayo Clinic, Texas Tech University, the University of Iowa, and The University of Texas Medical Branch at Galveston each had fewer than 15% within a similar radius. Programs keeping the most trainees in state were located in California (Loma Linda University; Stanford University; University of California, Davis; University of California, Irvine; University of California, Los Angeles; University of California, San Diego; University of California, San Francisco; and the University of Southern California) or Florida (the University of South Florida and the University of Miami).
Not surprisingly, the relationships between dermatologists' residency training site and eventual practice location are complex. A substantial minority practice within 100 miles of their residency training site, but the rightward skew of the mean vs the median suggests that many of those who relocate travel a substantial distance. Some of those who relocate to distant places might be returning to areas where they or their significant other grew up. The finding that women were somewhat more likely than men to remain close to the location of their training programs was intriguing, but it cannot be explained by our data set.
This study confirmed the findings of a smaller study16 that had suggested the Midwest exports approximately half its graduates, with most relocating to the South and the West. Wide variability was observed at the individual program level, but no Midwestern programs were among the residency training sites with the highest proportions of graduates staying in state or within 100 miles. Other studies of physician migration suggest that the West and South are also the beneficiaries of the migration of primary care physicians.17
Because of the large geographic size and populations of their states, it was not surprising that California and Florida programs kept the largest proportion of trainees in state. Texas, however, did not retain as many trainees despite its large size. The programs with the most graduates locating within 100 miles were all in New York City, suggesting some self-selection of individuals who desire to live in that metropolitan area and thus seek training there. Nevertheless, these data suggest that expanding residency programs in certain regions may have results that can, partly, be predicted (eg, creating positions in New York City is most likely to increase the ratio of dermatologists per capita within that metropolitan area, but the same is not true for Mayo Clinic in Rochester, Minnesota).
This study has several limitations. We were able to analyze the migration patterns of graduates, but because no accepted list of dermatologically underserved areas exists, to our knowledge, we did not attempt to assess predictors of practicing in areas experiencing shortages. Although a major strength of the American Academy of Dermatology database is its inclusion of so many US dermatologists (>7000 met our inclusion criteria and were analyzed), which minimizes the risks of sampling or selection biases, the limited number of fields in that database precluded us from knowing more about each physician. For example, we were unable to assess the possibly important role that the location where dermatologists were born or raised played in their training site or practice locations. Also, we did not have any information regarding dermatologists' spouses or partners that might yield additional insight as to their migration patterns. Although our exclusion of dermatologists less than 5 years from residency graduation limited the effect of temporary practice locations immediately after training, the database did not permit analysis of serial relocations or relocations occurring later in careers. Also, we could not account for physicians who divide their time among practice locations in different areas.
The location of graduates is only a single outcome measure, which individual training programs likely view differently. Some residency programs might pride themselves on distributing their progeny widely across distant institutions and communities, but others might seek to produce graduates who stay at their institution or who populate nearby communities. Others may focus on other outcomes (graduates becoming outstanding physicians, leaders in the field, community leaders, successful academicians or researchers, etc) and may not care about eventual practice locations. These data must be viewed with that possibility in mind.
As the field of dermatology continues to struggle with a maldistribution of its workforce, the association between training location and practice location is highly variable. However, certain regions, and particularly certain training programs, have much more consistent graduate migration patterns, with some far more likely to produce a widespread diaspora and others more likely to populate nearby communities.
No common mechanism is in place for rational planning of allocation or prioritization of any funding for new residency positions, and individual training programs clearly have several other important goals and objectives in mind in addition to meeting patient demand in the most underserved areas. Nevertheless, to the extent that individual residency programs seek to produce physicians who will serve unmet patient needs, further research to identify predictive factors at the individual level may help them to select some trainees from the surplus of dermatology applicants who are more likely to locate in underserved areas.
Federal graduate medical education funding is currently allocated to individual medical centers where it is distributed to specific residency programs in a manner that is not typically focused on geographically based clinical need. The types of migration patterns we observed are among many factors that might be considered by future funders as the residency financing landscape changes. Those seeking to reform graduate medical education funding at the federal level, as well as emerging funders such as multispecialty groups making grants to residency programs, increasingly may seek to address the maldistribution of physicians by directing funding for positions to selected programs. This practice could have significant effects on the field of dermatology and on our training infrastructure.
Correspondence: Jack S. Resneck Jr, MD, Department of Dermatology, University of California, San Francisco School of Medicine, Campus Box 0316, San Francisco, CA 94143-0316 (firstname.lastname@example.org).
Accepted for Publication: April 14, 2011.
Author Contributions: Dr Resneck had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Resneck. Acquisition of data: Kostecki. Analysis and interpretation of data: Resneck. Drafting of the manuscript: Resneck. Critical revision of the manuscript for important intellectual content: Resneck and Kostecki. Statistical analysis: Resneck and Kostecki. Administrative, technical, and material support: Kostecki.
Financial Disclosure: None reported.
Disclaimer: The views expressed in this article are those of Dr Resneck and Mr Kostecki and do not necessarily represent the views of the American Academy of Dermatology.
Additional Information: Dr Resneck chairs the Council on Government Affairs, Health Policy, and Practice of the American Academy of Dermatology. Mr Kostecki is an employee of the American Academy of Dermatology.