The Changing Composition of the Pediatric Medical Subspecialty Workforce
http://www.100md.com
《小儿科》
Cecil G. Sheps Center for Health Services Research
Department of Health Policy and Administration
Department of Biostatistics, School of Public Health, University of North Carolina, Chapel Hill, North Carolina
ABSTRACT
Objectives. To characterize the composition of the pediatric subspecialty workforce in terms of the distribution of women and international medical graduates (IMGs) across pediatric medical subspecialties and to determine whether the proportions of board-certified pediatric subspecialists who are women or IMGs differ between graduation cohorts.
Study Design. We used board certification data from the American Board of Pediatrics. Within each pediatric subspecialty, we classified physicians into 2 groups, ie, recent graduates, defined as those who completed medical school after January 1, 1987, and nonrecent graduates, who completed medical school before that date. We calculated the percentage of female physicians for each subspecialty and computed 95% confidence intervals around those estimates to identify male-dominated subspecialties. Using Pearson 2 tests, we compared the percentages of women between the 2 graduation cohorts for each subspecialty. Similar calculations were performed for the percentage of IMGs in each subspecialty. Sensitivity analyses were performed with data from the 2002 American Medical Association Physician Masterfile.
Results. For 9 of 16 pediatric medical subspecialties studied, the percentages of board-certified women were significantly greater in the recent cohort than in the nonrecent cohort. Subspecialties that remain predominantly male in the recent graduation cohort include cardiology, critical care medicine, gastroenterology, pulmonology, and sports medicine. In contrast, the percentages of board-certified IMGs were significantly lower for 6 of the 16 specialties studied; endocrinology and gastroenterology remain relatively reliant on IMGs.
Conclusions. For the majority of pediatric medical subspecialties, concerns that the predominance of women in pediatric training may negatively affect the supply of subspecialists are likely unfounded; however, a small number of procedure-based specialties, as well as sports medicine, continue to rely disproportionately on men. There do not seem to be consistent differences in the role of IMGs across the pediatric medical subspecialties between recent and nonrecent graduates, which may reflect differing tendencies to become certified.
Key Words: women international medical graduates workforce
Abbreviations: IMG, international medical graduate USMG, US medical graduate ABP, American Board of Pediatrics AMA, American Medical Association
Currently there is debate about the adequacy of the pediatric subspecialty workforce.1,2 Although recently Cull et al3 noted an increase in the percentage of third-year pediatric residents intending to subspecialize, from 13% in 1997 to 27% in 2002, they also reported a decrease in the percentage with an intention to provide both primary care and subspecialty care, from 19% to 6%. The total percentage of pediatric trainees intending to subspecialize either full-time or part-time did not change dramatically; however, the absolute number of pediatric subspecialty trainees increased 46%, from 728 in 1996 to 1061 in 2003.4
According to the American Board of Pediatrics (ABP), women represented 63% of the first-time candidates taking the general pediatrics certifying examination in 2003. Given their predominance in the pediatric workforce, the extent to which female pediatricians subspecialize has important implications for the pediatric subspecialty workforce.5 In past studies, female pediatricians were significantly more likely than male pediatricians to be general pediatricians6,7 and less likely to work in selected pediatric subspecialties.7 A survey-based study found that female pediatric residents expressed significantly less intention to subspecialize than did their male counterparts.8 Notably, Brotherton et al7 did not find a significant relationship between female gender and the likelihood of choosing a general pediatrics career among pediatricians <46 years of age but found a strong relationship between gender and choosing a general pediatrics career among pediatricians >45 years of age, which suggests a cohort effect. To date, no studies have explored how the gender composition of the pediatric subspecialty workforce has changed over time.
In the early 1990s, international medical graduates (IMGs) represented one third of third-year categorical pediatric residents.9 One study found that pediatric residents who trained abroad were significantly more likely than US medical graduates (USMGs) to indicate an intention to subspecialize, were significantly less likely to report primary care as a practice goal, and were significantly more likely to rate research opportunities as an essential or very important factor in job selection.8 In addition, past studies showed that IMGs represent disproportionate percentages of pediatric subspecialty trainees overall2 and in child neurology.10 However, the specific role of IMGs in the pediatric subspecialty workforce (ie, after training) has not been documented.
Given the predominance of women in pediatric residency and the historical reliance of pediatric subspecialty training programs on IMGs, the gender and training characteristics of the pediatric workforce may have important implications for the long-term supply of pediatric subspecialists in the United States. Continued trends toward male predominance suggest that the supply of pediatric subspecialists may be threatened by the predominance of women in general pediatrics training programs. Similarly, reliance on IMGs may be problematic because IMG entry into the United States physician workforce is a constant policy issue. Our objectives were to characterize the composition of the pediatric subspecialty workforce in terms of the distribution of women and IMGs across pediatric medical subspecialties, to identify male-dominated and IMG-reliant pediatric medical subspecialties, and to determine whether the proportions of certified subspecialists who were female or IMGs differed between graduation cohorts. Our findings inform current discussions of the pediatric subspecialty workforce and demonstrate how gender and training characteristics may influence pediatric subspecialty supply in coming years.
METHODS
We used 2003 diplomate data from the ABP and therefore studied the pediatric medical subspecialties that are certified by the ABP. Pediatricians who are board certified in allergy/immunology are also included, because the ABP maintains data on these individuals. The file, which includes data for 14780 board-certified pediatric subspecialists, excludes physicians known to be deceased or retired. The file includes data on gender, date of medical school graduation, all pediatric subspecialty boards for which the individual has been certified, and certification and expiration dates for each certification. The ABP data file also includes an indicator of IMG status that identifies diplomats who graduated from a medical school outside the United States or Canada. The file does not contain any information about diplomates’ current clinical involvement in the specialties in which they are certified. Therefore, board-certified providers who no longer practice in the subspecialty cannot be excluded from analyses unless their certification has expired. The ABP data file does include the entire population of board-certified pediatric subspecialists but, because not all trained or practicing subspecialists are certified, we consider this group to function as a sample, rather than a population, for statistical purposes (ie, we performed statistical tests).
We grouped diplomates according to pediatric subspecialty (eg, neonatal/perinatal medicine or nephrology). Physicians with current (ie, not expired) certification in a subspecialty were included and classified accordingly. Only 3% of providers had expired certification for their subspecialty certifications. Physicians with 2 current certifications were included in analyses for each of the subspecialties for which they had certification, because our data did not allow us to discern the discipline in which each physician was active currently.
In addition to pediatric subspecialty, we classified physicians into 1 of 2 cohorts on the basis of the date of medical school graduation. We elected to use graduation dates to construct these cohorts because many pediatric medical subspecialties did not offer certification before the early 1990s,2 despite the existence of training programs before that time, and the date of residency completion was not available in the file. Therefore, the date of medical school graduation serves as a proxy for the period in which the individual entered the subspecialty. We classified physicians who graduated from medical school after January 1, 1987, as recent graduates. Although this date is somewhat arbitrary, it allows up to 16 years to complete residency and fellowship training and to receive certification. Physicians who completed medical school before 1987 were considered to be nonrecent graduates. Less than 1% (n = 120) of diplomates were excluded from analyses because of missing data regarding the date of graduation.
For each subspecialty, we calculated the percentage of physicians who were female and the large-sample 95% confidence intervals. Specialties for which the large-sample 95% confidence interval did not contain 0.5 were considered to be dominated by 1 gender. This threshold represents equal proportions of male and female physicians and reflects a 2000 estimate that women represent 47.8% of practicing pediatricians in the United States.11 The analysis allows easy application of other thresholds. In addition, we calculated the percentages of female physicians for subspecialties according to graduation cohort and applied Pearson 2 tests to determine whether the percentage of female physicians in each subspecialty differed significantly between recent and nonrecent graduates. We used a Bonferroni adjustment to account for multiple comparisons. With this adjustment, P values of .0031 were considered significant. Similar calculations were performed for the percentages of physicians who were IMGs, overall and according to graduation status. Because IMGs currently represent 25% of US physicians,12 subspecialties were considered to be relatively IMG reliant if the lower limit of the 95% confidence interval was >0.25.
For 14 of the 16 subspecialties presented here, additional analyses were performed with data on clinically active physicians from the 2002 American Medical Association (AMA) Physician Masterfile. The AMA Physician Masterfile underestimates the number of board-certified pediatric medical subspecialists. Therefore, we developed an algorithm to identify providers in each of these disciplines. For these analyses, we identified physicians for each pediatric medical subspecialty in one of the following ways: (1) certified in the relevant pediatric subspecialty, (2) not certified but selected the relevant pediatric subspecialty as his or her primary or secondary specialty, (3) certified in general pediatrics and reported his or her specialty as the nonpediatric version of the subspecialty, or (4) certified in general pediatrics and certified in the internal medicine version of the subspecialty.
We provide the following example to clarify. To identify all relevant providers in pediatric rheumatology, we classified all physicians who were board certified in pediatric rheumatology as pediatric rheumatologists. In addition, we considered any physician whose self-reported primary or secondary specialty was pediatric rheumatology to be a pediatric rheumatologist, regardless of certification. This ensured inclusion of physicians who work full-time or part-time in the specialty but have not been certified. In addition, physicians who were certified as pediatricians but reported their specialty as rheumatology were included to capture reporting errors. Finally, physicians who were certified in pediatrics and rheumatology were included as pediatric rheumatologists to identify providers who trained before the existence of pediatric fellowship training programs, as well as internal medicine/pediatrics providers who pursued an internal medicine fellowship.
Using the algorithm described above, we identified 25% more pediatric subspecialists with the AMA data than with the ABP data. For 2 subspecialties, namely, allergy/immunology and adolescent medicine, many more providers were identified with the AMA data, because internists can also subspecialize in these fields. Therefore, the data for these 2 specialties are not comparable between the AMA and ABP data sets. Because the AMA data allowed physicians to self-designate their specialty regardless of certification status, analyses with these data can assess biases introduced by restricting the sample to certified physicians; however, the AMA data may overestimate the number of providers. Because the AMA file had <10 observations for developmental and behavioral pediatrics and neurodevelopmental disabilities, we excluded these specialties from the statistical analyses using AMA data.
RESULTS
Currently, women represent 37.2% of certified pediatric subspecialists (Table 1). Women represented only one third of nonrecent graduates but nearly one half of recent graduates. For 9 of 16 pediatric medical subspecialties, the percentages of women were significantly higher among board-certified recent graduates than among board-certified nonrecent graduates. The percentages of women certified in allergy/immunology, adolescent medicine, developmental and behavioral pediatrics, neurodevelopment, pulmonology, rheumatology, and sports medicine did not differ significantly between the 2 graduation cohorts.
Pediatric medical subspecialties in which male physicians predominate (ie, specialties for which the upper bound of the 95% confidence interval for the percentage of female physicians is <0.5) include allergy/immunology, cardiology, critical care, gastroenterology, neonatology, nephrology, pulmonology, and sports medicine (results not shown). Pediatric medical subspecialties in which female physicians predominate (ie, specialties for which the lower bound of the 95% confidence interval for the percentage of female physicians is >0.5) include adolescent medicine, developmental and behavioral pediatrics, emergency medicine, and neurodevelopment. Among board-certified recent graduates, those in cardiology, critical care, gastroenterology, pulmonology, and sports medicine continue to be disproportionately male (Fig 1). In contrast, women predominate among board-certified recent graduates in adolescent medicine, developmental and behavioral pediatrics, emergency medicine, and neurodevelopment.
IMGs represent 29.8% of all board-certified pediatric subspecialists and 20.6% of board-certified recent graduates (Table 2). For 6 of 16 specialties, the proportions of board-certified IMGs within each specialty were significantly lower among recent graduates than among nonrecent graduates. For 5 subspecialties, the proportions of IMGs in the recent cohort were greater than those in the nonrecent cohort; however, these differences were generally small and not significant. Overall, IMG-reliant pediatric medical subspecialties (ie, specialties for which the lower bound of the 95% confidence interval for the percentage of IMGs is >0.25) include allergy/immunology, endocrinology, gastroenterology, hematology/oncology, neonatology, and nephrology (results not shown). Among board-certified recent graduates, only gastroenterology and endocrinology continue to rely disproportionately on IMGs (Fig 2).
Sensitivity analyses performed with the AMA Physician Masterfile generated comparable estimates of the proportions of women in the pediatric medical subspecialties among the nonrecent and recent cohorts (Tables 3 and 4). Like bivariate analyses performed with the ABP data, those performed with the AMA data found that significantly greater proportions of women were certified in the recent cohort, compared with the nonrecent cohort. The only difference in findings between the ABP data and the AMA data was that the proportion of women certified in allergy/immunology was significantly greater for the recent cohort, compared with the nonrecent cohort, in the AMA file only.
There were discrepancies between the IMG results from the ABP data and those from the AMA file, however. Overall, the AMA and ABP data estimated comparable percentages of IMGs for the nonrecent cohort (33.6% and 33.3%, respectively); however, the estimates of the percentages of IMGs in the pediatric medical subspecialties based on the AMA data were greater than those based on the ABP data for the recent cohort (29.1% and 20.6%, respectively). A greater proportion of IMGs in pediatric infectious diseases in the recent cohort, relative to the nonrecent cohort, which was not significant in the ABP analyses, was significant in the analyses with the AMA data. Allergy/immunology and pediatric critical care medicine, which had significantly smaller proportions of IMGs in the recent cohort, relative to the nonrecent cohort, in the ABP-based analyses did not have significantly different proportions in the AMA analyses. Significantly smaller proportions of IMGs in the recent cohort, relative to the nonrecent cohort, were found for neonatology, cardiology, emergency medicine, and hematology/oncology in both the AMA and ABP analyses.
DISCUSSION
Currently, women represent the majority of general pediatrics residents in the United States. Past studies raised concerns that the predominance of women in pediatrics could lead to supply constraints for pediatric medical subspecialties,5,6 with the reasoning that a previously noted lesser tendency to subspecialize among female pediatricians would lead to a smaller pool of pediatricians interested in subspecialty training. Our results demonstrate that women represent significantly larger proportions of board-certified pediatric medical subspecialists among recent graduates than among their nonrecent peers. Analyses performed with data from the AMA demonstrate that this finding can be generalized to a broader definition of pediatric medical subspecialists (ie, both certified and uncertified). For most pediatric medical subspecialties, the lower involvement of woman, compared with men, reflects a cohort effect rather than an ongoing phenomenon.
In a few pediatric medical subspecialties, male physicians continue to predominate among the cohort of recent graduates. Four of the 5 specialties characterized as male dominated are procedure-based specialties. It is notable that for 3 of these, namely, cardiology, critical care medicine, and gastroenterology, the proportions of certified women are significantly greater in the recent cohort than in the nonrecent cohort. The extent to which these trends continue could influence the gender balance in these specialties. Training data from the ABP showed that, when data were averaged over the past 6 years, women represented approximately two fifths of pediatric cardiology, critical care medicine, and gastroenterology trainees in their final fellowship year (Table 5), which suggests that these fields may remain male dominated for the foreseeable future.
Given a constant number of trainees, the absolute number of women in general pediatrics training will lead to an increase in the percentage of women in the pediatric medical subspecialties unless there is a decrease in the percentage of women who choose to subspecialize. Nonetheless, our findings provide reassuring evidence of increased involvement of women in the pediatric medical subspecialties among recent graduates and a general trend toward parity in the majority of these fields. Concerns about the impact of female predominance in general pediatrics training programs on the future supply of pediatric subspecialists seem unwarranted for the majority of pediatric medical subspecialties. Continued male predominance in procedure-intensive specialties suggests that these disciplines may need to increase efforts to attract women.
Another frequently discussed issue with regard to women in the pediatric workforce relates to differences in the relevant full-time equivalents.6,13 Brotherton et al6 showed that female and male pediatric subspecialists were equally likely to work full-time and treat the same number of patients per week, on average. Previous studies reported differences in academic productivity between men and women in academic pediatrics.14,15 Interestingly, female pediatric subspecialists surveyed through the Future of Pediatric Education II study were significantly less likely than their male counterparts to report facing competition or to perceive no need for additional pediatric subspecialists in their practice areas.1 The extent to which these differential perceptions reflect different practice styles is not known. At this time, it remains unclear whether the increasing presence of women in pediatric medical subspecialties will result in any changes in productivity attributable to variations in practice preferences and/or personal responsibilities.
In contrast to our findings on the increasing role of women as board-certified pediatric subspecialists, our findings suggest that the proportions of IMGs among board-certified pediatric medical subspecialists are declining for several disciplines. ABP data on pediatric subspecialty trainees from 1998 through 2003 clearly show substantial involvement of IMGs in pediatric subspecialty fellowship programs (Table 6). Past data from the ABP indicated that IMGs represented >40% of pediatric fellowship trainees during the 1990s.2 Therefore, there is a discrepancy between the percentage of IMGs among pediatric trainees and the percentage of IMGs among board-certified pediatric medical subspecialists among recent graduates. A likely explanation is that IMGs are less likely to be certified, an observation that has been noted by the ABP (Bob Guerin, PhD, written communication, April 9, 2004). Similarly, a previous study found that IMGs represented 23.4% of surveyed pediatricians and 21.5% of those certified in a pediatric subspecialty but represented 41% of those who had received training in a certifiable specialty but were not certified.16 Replication of our analyses with AMA data found that the estimates of the proportions of IMGs were greater with the AMA data than with the ABP data for most pediatric medical subspecialties. These results suggest that differences in the timing and successful completion of certification between USMGs and IMGs may lead to underestimates of the proportions of IMGs in the pediatric subspecialty workforce, especially among recent graduates. Despite this underestimate, the results of bivariate analyses with AMA data yielded comparable conclusions regarding the differences in IMG involvement in pediatric medical subspecialties between graduation cohorts for 9 of 14 specialties studied.
More research is needed to elucidate the role of these IMGs in the pediatric subspecialty workforce in the United States. It is clear that IMGs represent large percentages of fellowship trainees in many pediatric medical subspecialties, and studies showed that 80% of IMGs who train in the United States remain here.17 We need a better understanding of how these physicians are involved in the pediatric subspecialty workforce after they complete their training, the reasons for their lesser propensity to be certified, and the association between certification status and the quality of care provided. Moreover, we need more research to identify the data sources that provide the most accurate estimates of pediatric subspecialists and their characteristics; sources contingent on certification status may underestimate the number of clinically active pediatric subspecialists, whereas those that rely on self-designation may overestimate the number of providers.
Several limitations are noteworthy. As discussed, the ABP data contain only physicians who are certified in a pediatric subspecialty. Additional analyses with AMA data yielded similar findings for women; however, these analyses were less consistent with regard to IMGs. Another potential limitation is the use of only 2 cohort periods. We choose 2 cohorts for ease of presentation; however, a reasonable concern is that the same pattern might not be observed with different cohort definitions or smaller and more frequent cohort intervals. To assess the validity of this concern, we performed analyses with a cutoff year of 1990 and again with 3 age cohorts, ie, <41 years of age, 41 to 55 years, and 56 years. Our results were similar in both the direction of the trends and their significance levels; therefore, we feel confident that the results are robust for other reasonable cohort constructions. Finally, there is the potential for an interaction effect between female gender and IMG status. USMGs were slightly more likely than IMGs to be female (37.9% vs 35.3%, P = .006). It is unclear how this interaction might develop, however, because the trends for these 2 groups are in opposite directions. Gender analyses restricted to USMGs yielded comparable findings; however, cell sizes were too small to allow sufficient power for detecting significant differences between female physicians in the recent and nonrecent cohorts among IMGs only. Finally, the use of the Bonferroni adjustment requires a very small P value to indicate significance; therefore, our results are somewhat conservative.
CONCLUSIONS
Women are increasingly represented among pediatric medical subspecialties. The few remaining male-dominated pediatric medical subspecialties may need to factor the changes in the composition of the overall pediatric workforce into their recruitment strategies and manpower projections. In contrast, several pediatric subspecialties may experience a decrease in IMG presence. The extent of IMG involvement in the US pediatric subspecialty workforce may be underestimated, however, because IMGs may be less likely to be certified. More research is needed to elucidate the role of IMGs in the pediatric subspecialty workforce after completion of fellowship training.
ACKNOWLEDGMENTS
This research was supported through grant 1-K02-HS013309-01A1 from the Agency for Healthcare Research and Quality.
Special thanks go to William J. Steinbach, MD, Morris Weinberger, PhD, participants in the Works in Progress Lunch at the Cecil G. Sheps Center for Health Services Research, and anonymous reviewers for helpful comments on earlier versions of the manuscript.
FOOTNOTES
Accepted Jan 5, 2005.
No conflict of interest declared.
REFERENCES
Stoddard JJ, Cull WL, Jewett EA, Brotherton SE, Mulvey HJ, Alden ER. Providing pediatric subspecialty care: a workforce analysis. Pediatrics. 2000;106 :1325 –1333
Gruskin A, Williams RG, McCabe ER, et al. Final report of the FOPE II Pediatric Subspecialists of the Future Workgroup. Pediatrics. 2000;106 :1224 –1244
Cull WL, Yudkowsky BK, Shipman SA, Pan RJ. Pediatric training and job market trends: results from the American Academy of Pediatrics third-year resident survey, 1997–2002. Pediatrics. 2003;112 :787 –792
Jones MD Jr, Stanton BF. Women in pediatric subspecialties. J Pediatr. 2004;144 :143 –144
Brotherton SE, Mulvey HJ, O’Connor KG. Women in pediatric practice: trends and implications. Pediatr Ann. 1999;28 :177 –183
Brotherton SE, Tang SfS, O’Connor KG. Trends in practice characteristics: analyses of 19 periodic surveys (1987–1992) of Fellows of the American Academy of Pediatrics. Pediatrics. 1997;100 :8 –18
Pan RJ, Cull WL, Brotherton SE. Pediatric residents’ career intentions: data from the leading edge of the pediatrician workforce. Pediatrics. 2002;109 :182 –188
Oliver TK Jr, Tunnessen WW Jr, Butzin D, Guerin R, Stockman JA III. Pediatric work force: data from the American Board of Pediatrics. Pediatrics. 1997;99 :241 –244
Laureta E, Moshe SL. State of training in child neurology 1997–2002. Neurology. 2004;62 :864 –869
American Medical Association. Physician Characteristics and Distribution in the US: 2004. Chicago, IL: AMA Press; 2003
Shipman SA, Lurie JD, Goodman DC. The general pediatrician: projecting future workforce supply and requirements. Pediatrics. 2004;113 :435 –442
Schaller JG. Women in academic pediatrics. Pediatr Ann. 1999;28 :184 –189
Kaplan SH, Sullivan LM, Dukes KA, Phillips CF, Kelch RP, Schaller JG. Sex differences in academic advancement: results of a national study of pediatricians. N Engl J Med. 1996;335 :1282 –1289
Brotherton SE. Pediatric subspecialty training, certification, and practice: who’s doing what. Pediatrics. 1994;94 :83 –89
Mullan F, Politzer RM, Davis CH. Medical migration and the physician workforce: international medical graduates and American medicine. JAMA. 1995;273 :1521 –1527(Michelle L. Mayer, PhD, M)
Department of Health Policy and Administration
Department of Biostatistics, School of Public Health, University of North Carolina, Chapel Hill, North Carolina
ABSTRACT
Objectives. To characterize the composition of the pediatric subspecialty workforce in terms of the distribution of women and international medical graduates (IMGs) across pediatric medical subspecialties and to determine whether the proportions of board-certified pediatric subspecialists who are women or IMGs differ between graduation cohorts.
Study Design. We used board certification data from the American Board of Pediatrics. Within each pediatric subspecialty, we classified physicians into 2 groups, ie, recent graduates, defined as those who completed medical school after January 1, 1987, and nonrecent graduates, who completed medical school before that date. We calculated the percentage of female physicians for each subspecialty and computed 95% confidence intervals around those estimates to identify male-dominated subspecialties. Using Pearson 2 tests, we compared the percentages of women between the 2 graduation cohorts for each subspecialty. Similar calculations were performed for the percentage of IMGs in each subspecialty. Sensitivity analyses were performed with data from the 2002 American Medical Association Physician Masterfile.
Results. For 9 of 16 pediatric medical subspecialties studied, the percentages of board-certified women were significantly greater in the recent cohort than in the nonrecent cohort. Subspecialties that remain predominantly male in the recent graduation cohort include cardiology, critical care medicine, gastroenterology, pulmonology, and sports medicine. In contrast, the percentages of board-certified IMGs were significantly lower for 6 of the 16 specialties studied; endocrinology and gastroenterology remain relatively reliant on IMGs.
Conclusions. For the majority of pediatric medical subspecialties, concerns that the predominance of women in pediatric training may negatively affect the supply of subspecialists are likely unfounded; however, a small number of procedure-based specialties, as well as sports medicine, continue to rely disproportionately on men. There do not seem to be consistent differences in the role of IMGs across the pediatric medical subspecialties between recent and nonrecent graduates, which may reflect differing tendencies to become certified.
Key Words: women international medical graduates workforce
Abbreviations: IMG, international medical graduate USMG, US medical graduate ABP, American Board of Pediatrics AMA, American Medical Association
Currently there is debate about the adequacy of the pediatric subspecialty workforce.1,2 Although recently Cull et al3 noted an increase in the percentage of third-year pediatric residents intending to subspecialize, from 13% in 1997 to 27% in 2002, they also reported a decrease in the percentage with an intention to provide both primary care and subspecialty care, from 19% to 6%. The total percentage of pediatric trainees intending to subspecialize either full-time or part-time did not change dramatically; however, the absolute number of pediatric subspecialty trainees increased 46%, from 728 in 1996 to 1061 in 2003.4
According to the American Board of Pediatrics (ABP), women represented 63% of the first-time candidates taking the general pediatrics certifying examination in 2003. Given their predominance in the pediatric workforce, the extent to which female pediatricians subspecialize has important implications for the pediatric subspecialty workforce.5 In past studies, female pediatricians were significantly more likely than male pediatricians to be general pediatricians6,7 and less likely to work in selected pediatric subspecialties.7 A survey-based study found that female pediatric residents expressed significantly less intention to subspecialize than did their male counterparts.8 Notably, Brotherton et al7 did not find a significant relationship between female gender and the likelihood of choosing a general pediatrics career among pediatricians <46 years of age but found a strong relationship between gender and choosing a general pediatrics career among pediatricians >45 years of age, which suggests a cohort effect. To date, no studies have explored how the gender composition of the pediatric subspecialty workforce has changed over time.
In the early 1990s, international medical graduates (IMGs) represented one third of third-year categorical pediatric residents.9 One study found that pediatric residents who trained abroad were significantly more likely than US medical graduates (USMGs) to indicate an intention to subspecialize, were significantly less likely to report primary care as a practice goal, and were significantly more likely to rate research opportunities as an essential or very important factor in job selection.8 In addition, past studies showed that IMGs represent disproportionate percentages of pediatric subspecialty trainees overall2 and in child neurology.10 However, the specific role of IMGs in the pediatric subspecialty workforce (ie, after training) has not been documented.
Given the predominance of women in pediatric residency and the historical reliance of pediatric subspecialty training programs on IMGs, the gender and training characteristics of the pediatric workforce may have important implications for the long-term supply of pediatric subspecialists in the United States. Continued trends toward male predominance suggest that the supply of pediatric subspecialists may be threatened by the predominance of women in general pediatrics training programs. Similarly, reliance on IMGs may be problematic because IMG entry into the United States physician workforce is a constant policy issue. Our objectives were to characterize the composition of the pediatric subspecialty workforce in terms of the distribution of women and IMGs across pediatric medical subspecialties, to identify male-dominated and IMG-reliant pediatric medical subspecialties, and to determine whether the proportions of certified subspecialists who were female or IMGs differed between graduation cohorts. Our findings inform current discussions of the pediatric subspecialty workforce and demonstrate how gender and training characteristics may influence pediatric subspecialty supply in coming years.
METHODS
We used 2003 diplomate data from the ABP and therefore studied the pediatric medical subspecialties that are certified by the ABP. Pediatricians who are board certified in allergy/immunology are also included, because the ABP maintains data on these individuals. The file, which includes data for 14780 board-certified pediatric subspecialists, excludes physicians known to be deceased or retired. The file includes data on gender, date of medical school graduation, all pediatric subspecialty boards for which the individual has been certified, and certification and expiration dates for each certification. The ABP data file also includes an indicator of IMG status that identifies diplomats who graduated from a medical school outside the United States or Canada. The file does not contain any information about diplomates’ current clinical involvement in the specialties in which they are certified. Therefore, board-certified providers who no longer practice in the subspecialty cannot be excluded from analyses unless their certification has expired. The ABP data file does include the entire population of board-certified pediatric subspecialists but, because not all trained or practicing subspecialists are certified, we consider this group to function as a sample, rather than a population, for statistical purposes (ie, we performed statistical tests).
We grouped diplomates according to pediatric subspecialty (eg, neonatal/perinatal medicine or nephrology). Physicians with current (ie, not expired) certification in a subspecialty were included and classified accordingly. Only 3% of providers had expired certification for their subspecialty certifications. Physicians with 2 current certifications were included in analyses for each of the subspecialties for which they had certification, because our data did not allow us to discern the discipline in which each physician was active currently.
In addition to pediatric subspecialty, we classified physicians into 1 of 2 cohorts on the basis of the date of medical school graduation. We elected to use graduation dates to construct these cohorts because many pediatric medical subspecialties did not offer certification before the early 1990s,2 despite the existence of training programs before that time, and the date of residency completion was not available in the file. Therefore, the date of medical school graduation serves as a proxy for the period in which the individual entered the subspecialty. We classified physicians who graduated from medical school after January 1, 1987, as recent graduates. Although this date is somewhat arbitrary, it allows up to 16 years to complete residency and fellowship training and to receive certification. Physicians who completed medical school before 1987 were considered to be nonrecent graduates. Less than 1% (n = 120) of diplomates were excluded from analyses because of missing data regarding the date of graduation.
For each subspecialty, we calculated the percentage of physicians who were female and the large-sample 95% confidence intervals. Specialties for which the large-sample 95% confidence interval did not contain 0.5 were considered to be dominated by 1 gender. This threshold represents equal proportions of male and female physicians and reflects a 2000 estimate that women represent 47.8% of practicing pediatricians in the United States.11 The analysis allows easy application of other thresholds. In addition, we calculated the percentages of female physicians for subspecialties according to graduation cohort and applied Pearson 2 tests to determine whether the percentage of female physicians in each subspecialty differed significantly between recent and nonrecent graduates. We used a Bonferroni adjustment to account for multiple comparisons. With this adjustment, P values of .0031 were considered significant. Similar calculations were performed for the percentages of physicians who were IMGs, overall and according to graduation status. Because IMGs currently represent 25% of US physicians,12 subspecialties were considered to be relatively IMG reliant if the lower limit of the 95% confidence interval was >0.25.
For 14 of the 16 subspecialties presented here, additional analyses were performed with data on clinically active physicians from the 2002 American Medical Association (AMA) Physician Masterfile. The AMA Physician Masterfile underestimates the number of board-certified pediatric medical subspecialists. Therefore, we developed an algorithm to identify providers in each of these disciplines. For these analyses, we identified physicians for each pediatric medical subspecialty in one of the following ways: (1) certified in the relevant pediatric subspecialty, (2) not certified but selected the relevant pediatric subspecialty as his or her primary or secondary specialty, (3) certified in general pediatrics and reported his or her specialty as the nonpediatric version of the subspecialty, or (4) certified in general pediatrics and certified in the internal medicine version of the subspecialty.
We provide the following example to clarify. To identify all relevant providers in pediatric rheumatology, we classified all physicians who were board certified in pediatric rheumatology as pediatric rheumatologists. In addition, we considered any physician whose self-reported primary or secondary specialty was pediatric rheumatology to be a pediatric rheumatologist, regardless of certification. This ensured inclusion of physicians who work full-time or part-time in the specialty but have not been certified. In addition, physicians who were certified as pediatricians but reported their specialty as rheumatology were included to capture reporting errors. Finally, physicians who were certified in pediatrics and rheumatology were included as pediatric rheumatologists to identify providers who trained before the existence of pediatric fellowship training programs, as well as internal medicine/pediatrics providers who pursued an internal medicine fellowship.
Using the algorithm described above, we identified 25% more pediatric subspecialists with the AMA data than with the ABP data. For 2 subspecialties, namely, allergy/immunology and adolescent medicine, many more providers were identified with the AMA data, because internists can also subspecialize in these fields. Therefore, the data for these 2 specialties are not comparable between the AMA and ABP data sets. Because the AMA data allowed physicians to self-designate their specialty regardless of certification status, analyses with these data can assess biases introduced by restricting the sample to certified physicians; however, the AMA data may overestimate the number of providers. Because the AMA file had <10 observations for developmental and behavioral pediatrics and neurodevelopmental disabilities, we excluded these specialties from the statistical analyses using AMA data.
RESULTS
Currently, women represent 37.2% of certified pediatric subspecialists (Table 1). Women represented only one third of nonrecent graduates but nearly one half of recent graduates. For 9 of 16 pediatric medical subspecialties, the percentages of women were significantly higher among board-certified recent graduates than among board-certified nonrecent graduates. The percentages of women certified in allergy/immunology, adolescent medicine, developmental and behavioral pediatrics, neurodevelopment, pulmonology, rheumatology, and sports medicine did not differ significantly between the 2 graduation cohorts.
Pediatric medical subspecialties in which male physicians predominate (ie, specialties for which the upper bound of the 95% confidence interval for the percentage of female physicians is <0.5) include allergy/immunology, cardiology, critical care, gastroenterology, neonatology, nephrology, pulmonology, and sports medicine (results not shown). Pediatric medical subspecialties in which female physicians predominate (ie, specialties for which the lower bound of the 95% confidence interval for the percentage of female physicians is >0.5) include adolescent medicine, developmental and behavioral pediatrics, emergency medicine, and neurodevelopment. Among board-certified recent graduates, those in cardiology, critical care, gastroenterology, pulmonology, and sports medicine continue to be disproportionately male (Fig 1). In contrast, women predominate among board-certified recent graduates in adolescent medicine, developmental and behavioral pediatrics, emergency medicine, and neurodevelopment.
IMGs represent 29.8% of all board-certified pediatric subspecialists and 20.6% of board-certified recent graduates (Table 2). For 6 of 16 specialties, the proportions of board-certified IMGs within each specialty were significantly lower among recent graduates than among nonrecent graduates. For 5 subspecialties, the proportions of IMGs in the recent cohort were greater than those in the nonrecent cohort; however, these differences were generally small and not significant. Overall, IMG-reliant pediatric medical subspecialties (ie, specialties for which the lower bound of the 95% confidence interval for the percentage of IMGs is >0.25) include allergy/immunology, endocrinology, gastroenterology, hematology/oncology, neonatology, and nephrology (results not shown). Among board-certified recent graduates, only gastroenterology and endocrinology continue to rely disproportionately on IMGs (Fig 2).
Sensitivity analyses performed with the AMA Physician Masterfile generated comparable estimates of the proportions of women in the pediatric medical subspecialties among the nonrecent and recent cohorts (Tables 3 and 4). Like bivariate analyses performed with the ABP data, those performed with the AMA data found that significantly greater proportions of women were certified in the recent cohort, compared with the nonrecent cohort. The only difference in findings between the ABP data and the AMA data was that the proportion of women certified in allergy/immunology was significantly greater for the recent cohort, compared with the nonrecent cohort, in the AMA file only.
There were discrepancies between the IMG results from the ABP data and those from the AMA file, however. Overall, the AMA and ABP data estimated comparable percentages of IMGs for the nonrecent cohort (33.6% and 33.3%, respectively); however, the estimates of the percentages of IMGs in the pediatric medical subspecialties based on the AMA data were greater than those based on the ABP data for the recent cohort (29.1% and 20.6%, respectively). A greater proportion of IMGs in pediatric infectious diseases in the recent cohort, relative to the nonrecent cohort, which was not significant in the ABP analyses, was significant in the analyses with the AMA data. Allergy/immunology and pediatric critical care medicine, which had significantly smaller proportions of IMGs in the recent cohort, relative to the nonrecent cohort, in the ABP-based analyses did not have significantly different proportions in the AMA analyses. Significantly smaller proportions of IMGs in the recent cohort, relative to the nonrecent cohort, were found for neonatology, cardiology, emergency medicine, and hematology/oncology in both the AMA and ABP analyses.
DISCUSSION
Currently, women represent the majority of general pediatrics residents in the United States. Past studies raised concerns that the predominance of women in pediatrics could lead to supply constraints for pediatric medical subspecialties,5,6 with the reasoning that a previously noted lesser tendency to subspecialize among female pediatricians would lead to a smaller pool of pediatricians interested in subspecialty training. Our results demonstrate that women represent significantly larger proportions of board-certified pediatric medical subspecialists among recent graduates than among their nonrecent peers. Analyses performed with data from the AMA demonstrate that this finding can be generalized to a broader definition of pediatric medical subspecialists (ie, both certified and uncertified). For most pediatric medical subspecialties, the lower involvement of woman, compared with men, reflects a cohort effect rather than an ongoing phenomenon.
In a few pediatric medical subspecialties, male physicians continue to predominate among the cohort of recent graduates. Four of the 5 specialties characterized as male dominated are procedure-based specialties. It is notable that for 3 of these, namely, cardiology, critical care medicine, and gastroenterology, the proportions of certified women are significantly greater in the recent cohort than in the nonrecent cohort. The extent to which these trends continue could influence the gender balance in these specialties. Training data from the ABP showed that, when data were averaged over the past 6 years, women represented approximately two fifths of pediatric cardiology, critical care medicine, and gastroenterology trainees in their final fellowship year (Table 5), which suggests that these fields may remain male dominated for the foreseeable future.
Given a constant number of trainees, the absolute number of women in general pediatrics training will lead to an increase in the percentage of women in the pediatric medical subspecialties unless there is a decrease in the percentage of women who choose to subspecialize. Nonetheless, our findings provide reassuring evidence of increased involvement of women in the pediatric medical subspecialties among recent graduates and a general trend toward parity in the majority of these fields. Concerns about the impact of female predominance in general pediatrics training programs on the future supply of pediatric subspecialists seem unwarranted for the majority of pediatric medical subspecialties. Continued male predominance in procedure-intensive specialties suggests that these disciplines may need to increase efforts to attract women.
Another frequently discussed issue with regard to women in the pediatric workforce relates to differences in the relevant full-time equivalents.6,13 Brotherton et al6 showed that female and male pediatric subspecialists were equally likely to work full-time and treat the same number of patients per week, on average. Previous studies reported differences in academic productivity between men and women in academic pediatrics.14,15 Interestingly, female pediatric subspecialists surveyed through the Future of Pediatric Education II study were significantly less likely than their male counterparts to report facing competition or to perceive no need for additional pediatric subspecialists in their practice areas.1 The extent to which these differential perceptions reflect different practice styles is not known. At this time, it remains unclear whether the increasing presence of women in pediatric medical subspecialties will result in any changes in productivity attributable to variations in practice preferences and/or personal responsibilities.
In contrast to our findings on the increasing role of women as board-certified pediatric subspecialists, our findings suggest that the proportions of IMGs among board-certified pediatric medical subspecialists are declining for several disciplines. ABP data on pediatric subspecialty trainees from 1998 through 2003 clearly show substantial involvement of IMGs in pediatric subspecialty fellowship programs (Table 6). Past data from the ABP indicated that IMGs represented >40% of pediatric fellowship trainees during the 1990s.2 Therefore, there is a discrepancy between the percentage of IMGs among pediatric trainees and the percentage of IMGs among board-certified pediatric medical subspecialists among recent graduates. A likely explanation is that IMGs are less likely to be certified, an observation that has been noted by the ABP (Bob Guerin, PhD, written communication, April 9, 2004). Similarly, a previous study found that IMGs represented 23.4% of surveyed pediatricians and 21.5% of those certified in a pediatric subspecialty but represented 41% of those who had received training in a certifiable specialty but were not certified.16 Replication of our analyses with AMA data found that the estimates of the proportions of IMGs were greater with the AMA data than with the ABP data for most pediatric medical subspecialties. These results suggest that differences in the timing and successful completion of certification between USMGs and IMGs may lead to underestimates of the proportions of IMGs in the pediatric subspecialty workforce, especially among recent graduates. Despite this underestimate, the results of bivariate analyses with AMA data yielded comparable conclusions regarding the differences in IMG involvement in pediatric medical subspecialties between graduation cohorts for 9 of 14 specialties studied.
More research is needed to elucidate the role of these IMGs in the pediatric subspecialty workforce in the United States. It is clear that IMGs represent large percentages of fellowship trainees in many pediatric medical subspecialties, and studies showed that 80% of IMGs who train in the United States remain here.17 We need a better understanding of how these physicians are involved in the pediatric subspecialty workforce after they complete their training, the reasons for their lesser propensity to be certified, and the association between certification status and the quality of care provided. Moreover, we need more research to identify the data sources that provide the most accurate estimates of pediatric subspecialists and their characteristics; sources contingent on certification status may underestimate the number of clinically active pediatric subspecialists, whereas those that rely on self-designation may overestimate the number of providers.
Several limitations are noteworthy. As discussed, the ABP data contain only physicians who are certified in a pediatric subspecialty. Additional analyses with AMA data yielded similar findings for women; however, these analyses were less consistent with regard to IMGs. Another potential limitation is the use of only 2 cohort periods. We choose 2 cohorts for ease of presentation; however, a reasonable concern is that the same pattern might not be observed with different cohort definitions or smaller and more frequent cohort intervals. To assess the validity of this concern, we performed analyses with a cutoff year of 1990 and again with 3 age cohorts, ie, <41 years of age, 41 to 55 years, and 56 years. Our results were similar in both the direction of the trends and their significance levels; therefore, we feel confident that the results are robust for other reasonable cohort constructions. Finally, there is the potential for an interaction effect between female gender and IMG status. USMGs were slightly more likely than IMGs to be female (37.9% vs 35.3%, P = .006). It is unclear how this interaction might develop, however, because the trends for these 2 groups are in opposite directions. Gender analyses restricted to USMGs yielded comparable findings; however, cell sizes were too small to allow sufficient power for detecting significant differences between female physicians in the recent and nonrecent cohorts among IMGs only. Finally, the use of the Bonferroni adjustment requires a very small P value to indicate significance; therefore, our results are somewhat conservative.
CONCLUSIONS
Women are increasingly represented among pediatric medical subspecialties. The few remaining male-dominated pediatric medical subspecialties may need to factor the changes in the composition of the overall pediatric workforce into their recruitment strategies and manpower projections. In contrast, several pediatric subspecialties may experience a decrease in IMG presence. The extent of IMG involvement in the US pediatric subspecialty workforce may be underestimated, however, because IMGs may be less likely to be certified. More research is needed to elucidate the role of IMGs in the pediatric subspecialty workforce after completion of fellowship training.
ACKNOWLEDGMENTS
This research was supported through grant 1-K02-HS013309-01A1 from the Agency for Healthcare Research and Quality.
Special thanks go to William J. Steinbach, MD, Morris Weinberger, PhD, participants in the Works in Progress Lunch at the Cecil G. Sheps Center for Health Services Research, and anonymous reviewers for helpful comments on earlier versions of the manuscript.
FOOTNOTES
Accepted Jan 5, 2005.
No conflict of interest declared.
REFERENCES
Stoddard JJ, Cull WL, Jewett EA, Brotherton SE, Mulvey HJ, Alden ER. Providing pediatric subspecialty care: a workforce analysis. Pediatrics. 2000;106 :1325 –1333
Gruskin A, Williams RG, McCabe ER, et al. Final report of the FOPE II Pediatric Subspecialists of the Future Workgroup. Pediatrics. 2000;106 :1224 –1244
Cull WL, Yudkowsky BK, Shipman SA, Pan RJ. Pediatric training and job market trends: results from the American Academy of Pediatrics third-year resident survey, 1997–2002. Pediatrics. 2003;112 :787 –792
Jones MD Jr, Stanton BF. Women in pediatric subspecialties. J Pediatr. 2004;144 :143 –144
Brotherton SE, Mulvey HJ, O’Connor KG. Women in pediatric practice: trends and implications. Pediatr Ann. 1999;28 :177 –183
Brotherton SE, Tang SfS, O’Connor KG. Trends in practice characteristics: analyses of 19 periodic surveys (1987–1992) of Fellows of the American Academy of Pediatrics. Pediatrics. 1997;100 :8 –18
Pan RJ, Cull WL, Brotherton SE. Pediatric residents’ career intentions: data from the leading edge of the pediatrician workforce. Pediatrics. 2002;109 :182 –188
Oliver TK Jr, Tunnessen WW Jr, Butzin D, Guerin R, Stockman JA III. Pediatric work force: data from the American Board of Pediatrics. Pediatrics. 1997;99 :241 –244
Laureta E, Moshe SL. State of training in child neurology 1997–2002. Neurology. 2004;62 :864 –869
American Medical Association. Physician Characteristics and Distribution in the US: 2004. Chicago, IL: AMA Press; 2003
Shipman SA, Lurie JD, Goodman DC. The general pediatrician: projecting future workforce supply and requirements. Pediatrics. 2004;113 :435 –442
Schaller JG. Women in academic pediatrics. Pediatr Ann. 1999;28 :184 –189
Kaplan SH, Sullivan LM, Dukes KA, Phillips CF, Kelch RP, Schaller JG. Sex differences in academic advancement: results of a national study of pediatricians. N Engl J Med. 1996;335 :1282 –1289
Brotherton SE. Pediatric subspecialty training, certification, and practice: who’s doing what. Pediatrics. 1994;94 :83 –89
Mullan F, Politzer RM, Davis CH. Medical migration and the physician workforce: international medical graduates and American medicine. JAMA. 1995;273 :1521 –1527(Michelle L. Mayer, PhD, M)