Gender Segregation by Specialty during Medical School

Abstract

Over the past 30 years, the representation of women in the active physician workforce climbed from 6.8% in 1970 to 23.0% in 1997.1 Moreover, women's presence in medicine will continue to grow, because 45.8% of all U.S. first-year medical students in 2000 were women.2 By 2010, women physicians are projected to constitute 29.4% of U.S. physicians.1 In spite of this dramatic progress, women physicians remain concentrated in relatively few, lower-income medical specialties3 and are less likely than are men to hold tenured faculty positions.4 In addition to their effects on women physicians, gender disparities in medicine may have implications for patient care. Patients who desire providers of a specific sex but cannot access them may be more reluctant to seek medical care initially and less compliant when they ultimately end up with providers whom they do not trust. Having providers of a specific sex has been shown to influence the satisfaction of women seeking prenatal care,5 and satisfaction has been associated with patients' compliance.6 It follows that a similar preference for women physicians may exist for women seeking other types of sensitive care, such as treatment for breast cancer or urologic disorders. Although the supply of women obstetricians has grown substantially, the availability of women breast surgeons and urologists remains very low. Further, data suggesting that substantial numbers of women prefer women primary care physicians may imply that some women prefer to see women providers for all of their health care needs. In addition, a number of studies have suggested that women physicians differ from their men counterparts with respect to time spent with patients7 and attention devoted to preventive health.8 Women physicians may also treat certain conditions more aggressively.9 Thus, the presence of women across a broad range of specialties may affect patient care in a number of different ways. There is extensive literature investigating why men and women medical students pursue different specialties. These studies suggest that demographic characteristics, personal values, and/or unique family responsibilities are responsible for gender differences in specialty choice.10–12 Surprisingly, little effort has been made to assess the role of academic experience in promoting gender segregation across specialties. Coffin and Babbot found that the gender disparity in interest in pediatrics grew during medical school.13 Several studies have documented an association between medical school experiences and specialty choices, and at least one study suggests that women and men have different experiences in medical school.13,14 In this study, we used the well-established sociologic concept of segregation to describe differences in distributions. We specifically examined how gender segregation across specialties has changed during the past 20 years and how gender segregation changes during medical school. In so doing, we began to assess whether the entry of women into medicine: (1) causes the profession to become woman-dominated; (2) exacerbates the concentration of women in a few, lower-income, specialties; or (3) encourages the equal representation of men and women in the profession. This analysis sheds light on whether specialties are continuing to integrate, as Jacobs found for the 1970s and 1980s, or whether a process of resegregation by gender is occurring in the medical profession, as Reskin and Roos suggest.15,16 The analysis also indicates whether the process of gender segregation is fixed at entrance to medical school, or reproduced during medical school. Methods The primary data for this study were from the Jefferson Longitudinal Study of Medical Education.17 The study represents an annual panel of students entering Jefferson Medical College since 1964. It contains information about students prior to, during, and after medical school. We limited our analysis to the cohorts who began their medical education in 1975 or later and who graduated between 1979 and 2000. These cohorts have complete data on specialty plans as freshmen and seniors. They included 1,275 women and 3,037 men, for a total of 4,312 analytic subjects. The representation of women increased over the course of the study. Women constituted 26.7% of students in the entire sample. They were 18.9% of students matriculating between 1976 and 1980 and 37.1% of students matriculating between 1991 and 1995. Recently, the characteristics of the Jefferson student body have closely resembled those of the national population of medical students in important respects, such as average Medical College Admission Test (MCAT) science score and scores on the U.S. Medical Licensing Examination.2,18 Between 1995 and 1999, 43.0% of matriculants to medical schools were women and 18.5% were Asian. The corresponding figures at Jefferson were 36.7% and 19.5%. Finally, 7.6% of matriculants between 1995 and 1999 were Black. At Jefferson, the figure was lower, at 2.9%. We used nationally representative data from medical students during the 1990s to assess the generalizability of trends at Jefferson. These data come from the Association of American Medical Colleges' (AAMC's) graduating student questionnaire (GQ), which is administered to all fourth-year medical students in the country. The key outcome of interest in this analysis was students' responses to questions about their intended specialties. In this analysis, we collapsed choices into the following ten broad categories: (1) anesthesiology, pathology, and radiology, referred to as hospital-based specialties; (2) emergency medicine; (3) family practice; (4) internal medicine; (5) obstetrics—gynecology; (6) ophthalmology; (7) pediatrics; (8) psychiatry; (9) surgery; and (10) other. There were a total of 11 categories in our analysis because undecided students formed their own group. We used collapsed categories because of sample-size constraints. The sample did not allow an assessment of segregation among 25 specialties unless all the data were aggregated and time trends were ignored. Collapsing detailed specialty characteristics inevitably results in some loss of precision. However, that loss was limited in this case. We found that our collapsed specialties captured over 90% of the segregation revealed by the 25 fields for Jefferson first-year students, Jefferson fourth-year students and AAMC—GQ fourth-year students. Our analysis had two phases. First, we documented changes over time, and between the first and the fourth years, in the levels of gender segregation across specialties. We then examined the relative odds of women's representation in specific fields. We used the standard measure of segregation, Duncan's index of dissimilarity (D), for the first phase of our analysis.19 D is interpreted as measuring the proportion of men or women who would have to change specialties in order to produce an even distribution across all specialties. D measures the difference between the allocation of women across fields and the allocation of men across fields. For example, if 1% of all women intend to pursue surgery and 10% of all men intend to pursue surgery, then the absolute value of the difference is 9%. D equals half of the sum of these individual differences. We produced two sets of D for our analysis, one for all students and one for students who elect particular specialties as first-year students. We calculated the second set of Ds to determine whether the inclusion of first-year “undecideds” affected our results. Since the group of “undecideds” was large (36.5% first-year students overall and 40% of matriculants between 1991 and 1995) and generally integrated, they could have distorted the implications of D. (Women appeared among the “undecideds” roughly in proportion to their presence among students in general, especially since 1981.) Statistical analysis used the chi-square and jackknife tests to assess changes in the D statistic. Results Table 1 compares the indices of segregation for first- and fourth-year students between 1976 and 1995. Gender segregation across specialties increased during this time period for the first-year students. Between 1976–1980 and 1991–1995, the index of dissimilarity increased from 15.5 to 24.2 or 56.1% for first-year students. The upward trend was also evident among fourth-year students. D rose 67.1% for graduating students. Both chi-square tests and jack-knife tests indicated that these changes in segregation were significant (p < .01).TABLE 1: Gender Distribution at Jefferson Medical College as Depicted by the Index of Dissimilarity, Duncan's D*The increase in segregation among fourth-year students at Jefferson mirrored national trends. Between 1981 and 1985, the segregation of fourth-year students at Jefferson was 24.8. The comparable figure for the 1991–1995 period was 27.4. Based on our analysis of the AAMC's graduating student questionnaire, between 1990 and 1999, segregation among all fourth-year students increased from 22.3 to 27.3. We also found that the level of segregation was generally higher among fourth-year than among first-year students. This difference suggested that medical school experience is associated with an increase in the gender-based segregation of students by specialty. However, the absolute increase in D changed over time. The role of the medical school in determining D appeared to peak between 1986 and 1990 and then began declining. The inclusion of undecided first-year students lowered the level of segregation significantly for these students. The second panel of Table 1 indicated that first-year students who reported intended fields were about as segregated by field in the first year as they were in the fourth year. Overall, when undecided students were excluded, segregation declined slightly between the first and fourth years, but when the data were divided into four periods, there was no consistent pattern. A first-year—fourth-year increase in 1976–1980 gave way to almost no change in 1981–1985, followed by a sharp increase in 1986–1990 and an even steeper decline in 1991–1995. Thus, when undecided students were excluded, we found that students left medical school about as segregated as when they started their training. Nevertheless, there remained an important sense in which medical school contributed to the gender differentiation of specialties. Only about one fifth of the fourth-year students left medical school in the same specialties they had intended to pursue in the first year. As we have seen, over a third of the first-year students entered without firm plans regarding their specialties. Of those who did have intended specialties, majorities of both the men (65%) and the women (60%) switched their specialties by the fourth year. In all, 79% of the women and 75% of the men were either undecided initially or changed specialties during medical school. For the vast majority of students, the medical school experience had the potential to influence the choice of specialty. Thus, medical school was clearly implicated in the process of segregation even if D did not increase between the first and fourth years. We found that increases in the index of dissimilarity were due to changes in the characteristics of students choosing six fields: surgery, hospital specialties, internal medicine, obstetrics—gynecology, pediatrics, and family practice. Over time, we found declines in the percentage of fourth-year women intending to pursue surgery, hospital specialties, and internal medicine. The most pronounced change occurred in surgery. Between 1976 and 1980, fourth-year men students were twice as likely as were fourth-year women students to select surgery as their preferred or intended specialty. Between 1991 and 1995, men were almost three times as likely as women to choose surgery. We also found increases in the percentages of fourth-year women intending to pursue pediatrics, obstetrics—gynecology, and family practice. The most pronounced change occurred in obstetrics—gynecology. Between 1976 and 1980, senior women students were 1.1 times more likely than were senior men students to choose obstetrics—gynecology. Between 1991 and 95, senior women are 2.7 times more likely then senior men to choose obstetrics—gynecology. The specialties driving the increase in gender-based segregation at Jefferson were similar to the specialties responsible for segregation across the country. The increase in gender-based segregation among medical students nationally stems from increases in the proportion of men pursuing internal medicine and hospital specialties and increases in the proportion of women pursuing family practice and obstetrics—gynecology. Thus, it appears that the substantial increase in segregation over time, evident from Table 1, was due to the increased concentration of men in predominantly “male” fields such as surgery and the increased concentration of women in predominantly “female” fields such as obstetrics—gynecology. This conclusion holds for both first-and fourth-year students, and for the Jefferson data as well as the AAMC data. Discussion We found first, that there was a significant trend toward increased gender segregation among specialties between the 1980s and 1990s. This trend is explained by the increased concentration of men in surgery, hospital specialties, and internal medicine and the increased concentration of women in pediatrics, family practice, and obstetrics—gynecology. Furthermore, our analysis indicated that the trend toward increased gender segregation at Jefferson is representative of a larger national change in the physician workforce. This trend is consistent with the resegregation thesis advanced by Reskin and Roos16 and may imply that the entry of women into medicine has exacerbated the concentration of women in a subset of specialties. This trend may also stem from differences in how men and women medical students react to changes in medical school curricula, changes in the organization and financing of medical care, or changes in the status of women more generally. We also found an increase in gender-based segregation during medical school. This increase is attributable primarily to the large fraction of students who enter medical school without a specialty preference and ultimately distribute themselves across specialties in a segregated way. The extent of segregation among students who have preferred specialties in the first year did not change significantly between the first and fourth years because this population remains equally segregated throughout medical school. Although the extent of segregation in this group is constant, choices remain highly dynamic. Since women and men leave their preferred fields in a segregated fashion and enter new fields in an equally segregated fashion, the flows of students who select specialties in the first year do not contribute to changes in segregation between the first and fourth years. The fluidity of specialty choices among individuals is consistent with Jacobs' revolving-door perspective on gender segregation.15 The trends noted in this analysis are unexpected and may be inconsistent with goals pertaining to a highly effective physician workforce.20 Individual medical schools with workforce objectives may address these trends by offering counseling to undecided students early in medical school. We also note that Duncan's D provides a concise index for analysis of complex patterns of medical students' career decisions. We intend to expand these findings in future research by examining the correlates of Jefferson students' choice of medical specialty. We will also assess institutional effects more fully, by investigating trends in the gender distribution by specialty for multiple medical schools.