Racial and Ethnic Disparities in Faculty Promotion in Academic Medicine

Context Previous studies have suggested that minority medical school faculty are at a disadvantage in promotion opportunities compared with white faculty.

Objective To compare promotion rates of minority and white medical school faculty in the United States.

Design and Setting Analysis of data from the Association of American Medical Colleges' Faculty Roster System, the official data system for tracking US medical school faculty.

Participants A total of 50,145 full-time US medical school faculty who became assistant professors or associate professors between 1980 and 1989. Faculty of historically black and Puerto Rican medical schools were excluded.

Main Outcome Measures Attainment of associate or full professorship among assistant professors and full professorship among associate professors by 1997, among white, Asian or Pacific Islander (API), underrepresented minority (URM; including black, Mexican American, Puerto Rican, Native American, and Native Alaskan), and other Hispanic faculty.

Results By 1997, 46% of white assistant professors (13,479/28,953) had been promoted, whereas 37% of API (1123/2997; P<.001), 30% of URM (311/1053, P<.001), and 43% of other Hispanic assistant professors (256/598; P = .07) had been promoted. Similarly, by 1997, 50% of white associate professors (7234/14,559) had been promoted, whereas 44% of API (629/1419; P<.001), 36% of URM (101/280; P<.001), and 43% of other Hispanic (122/286; P = .02) associate professors had been promoted. Racial/ethnic disparities in promotion were evident among tenure and nontenure faculty and among faculty who received and did not receive National Institutes of Health research awards. After adjusting for cohort, sex, tenure status, degree, department, medical school type, and receipt of NIH awards, URM faculty remained less likely to be promoted compared with white faculty (relative risk [RR], 0.68 [99% confidence interval {CI}, 0.59-0.77] for assistant professors and 0.81 [99% CI, 0.65-0.99] for associate professors). API assistant professors also were less likely to be promoted (RR, 0.91 [99% CI, 0.84-0.98]), whereas API associate professors and other Hispanic assistant and associate professors were promoted at comparable rates.

Conclusion Our data indicate that minority faculty are promoted at lower rates compared with white faculty.

While several studies have demonstrated that female faculty are less likely than men to be promoted to senior rank,^1-9 less is known about the promotion of racial/ethnic minority faculty members. In recent decades, the numbers of minority faculty have increased, and these increases have encompassed minority groups traditionally underrepresented in medicine, including blacks, Mexican Americans, Puerto Ricans, American Indians, and Alaska Natives, as well as minority groups that have not been considered underrepresented, including Asian or Pacific Islanders (APIs) and other Hispanic Americans.¹⁰

In 1980-1981, 2.6% of newly appointed assistant professors were underrepresented minorities (URMs), 10.3% were APIs, and 1.9% were other Hispanics (Figure 1); by 1996-1997, 4.6% of new assistant professors were URMs, 13.8% were APIs, and 2.1% were other Hispanics. Much smaller increases over time are noted among newly appointed associate professors (Figure 1).¹¹ Although these increases indicate steady improvement in the representation of minority faculty, they suggest that medical schools have been more successful at recruiting minority junior faculty and less successful at helping minority junior faculty achieve senior rank.

To our knowledge, only 2 studies have explicitly addressed racial/ethnic disparities in faculty promotion in medical schools. Petersdorf et al¹² examined faculty with a doctor of medicine (MD) degree in 1989 and found that minority faculty typically are promoted to the associate professor level 3 to 7 years later than white faculty. However, this seminal study excluded faculty without MD degrees and did not adjust for potential confounders such as professional age or faculty productivity. More recently, Palepu et al¹³ conducted a national survey of medical school faculty and found that minority faculty were less likely to be promoted to senior rank than white faculty. While both faculty with and without MD degrees were included and adjustment was made for age and faculty productivity, this study was limited by a sample that included only 344 minority faculty and the self-reported nature of the survey data.

This study builds on these previous studies of racial/ethnic disparities in faculty promotion and examines promotions using data from the official roster of all US medical school faculty maintained by the Association of American Medical Colleges (AAMC), while controlling for faculty productivity using a more objective criterion, receipt of an award from the National Institutes of Health (NIH), as well as other potential confounders.

Data used in this study were obtained from the AAMC's Faculty Roster System, the official data system for tracking US medical school faculty.¹¹ This database contains updated information about faculty members including demographics, education, employment, and promotions. Based on a 1997 AAMC survey of department chairs of US medical schools, it is estimated that the database contains records for approximately 90% of all active full-time US medical school faculty (Charles A. Elliott, director of the Faculty Roster System, oral communication, March 1999).

Each medical school has a designated Faculty Roster System representative who is responsible for gathering faculty data, including information about race and ethnicity, and submitting this information to the AAMC. Typically, new medical school faculty complete the Faculty Roster form when they are first appointed, while Faculty Roster System representatives make updates to the database about changes in employment status. However, as the data are reported voluntarily by medical schools, there are variations in consistency across data elements.

The study population consists of full-time US medical school faculty who became assistant or associate professors between 1980 and 1989. These years were chosen to allow all faculty to be tracked for a minimum of 8 years during which a promotion could be identified. Faculty who remain in academia may not reflect all faculty who aspire for and are eligible for promotion. Because some evidence suggests that minority faculty may be more likely to leave academia,¹⁴ all faculty, including those who left academic medicine during the observation period, were included in the study population. Because predominantly minority medical schools may be atypical in providing minority faculty with more mentorship and professional support than other schools, faculty from historically black medical schools and Puerto Rican medical schools were excluded (387 assistant and 253 associate professors). In addition, faculty in departments other than clinical and basic science departments, such as social science or allied health departments, were excluded (674 assistant and 360 associate professors).

Because faculty composition and the probability of promotion may change over time, the study population was divided into 5 cohorts representing faculty who attained their rank in 1980-1981, 1982-1983, 1984-1985, 1986-1987, and 1988-1989. The construction of these study cohorts relied on data elements in the Faculty Roster System about the year in which each faculty member attained different ranks. These variables were missing for about a quarter of medical school faculty, and they were consequently excluded from the study population. About half of these faculty with missing data were listed as faculty members prior to 1980; hence, they are not part of the later study cohorts and their exclusion is appropriate. However, some of the remaining faculty likely belonged in the 1980-1989 cohorts used for our study. To address possible bias introduced by these excluded faculty, we compared racial/ethnic disparities in promotion among these faculty and among the faculty included in our study. We found the disparities to be quite similar between the 2 groups of faculty and believe that data selection bias is not a major factor in our study.

To define our key dependent variable, promotion, each faculty member in our study population was followed up through 1997. Faculty whose reported academic rank changed from assistant professor to associate or full professor or from associate professor to full professor during the years of observation were considered to have been promoted. Faculty for whom no rank change was reported, including faculty who left academia, were considered not to have been promoted.

Faculty race and ethnicity was our key independent variable of interest. Information about race and ethnicity is based on self-reported designations. The Faculty Roster System classifies faculty as American Indian or Alaska Native; API; black, not of Hispanic origin; Mexican American or Chicano; Puerto Rican; other Hispanic; white, not of Hispanic origin; or "do not wish to respond." For this study, faculty classified as black, Mexican American, Puerto Rican, American Indian, or Alaska Native were considered URMs. These classifications do not conform exactly to official AAMC designations, which include Native Hawaiians and exclude Commonwealth Puerto Ricans from URM designation. For most analyses, URM faculty are included as a single category because the counts of faculty of specific racial/ethnic groups are small. The URM faculty were compared with white, API, and other Hispanic faculty. Faculty who did not indicate their race/ethnicity were excluded (2230 assistant and 840 associate professors).

In addition to rank, cohort, and race and ethnicity, other faculty characteristics included in the analyses were sex, type of degree, tenure status, department, medical school type, and research productivity. Degrees differentiated include faculty with MD degrees from US medical schools, MD degrees from foreign medical school, doctor of philosophy (PhD) degrees, and other degrees. Faculty with both MD and PhD degrees were classified by their medical degree. Tenure status differentiated faculty with tenure or on a tenure track from faculty on nontenure tracks. Medical school departments differentiated basic science, primary care (family medicine, internal medicine, and pediatrics), surgery, and other clinical departments. Medical school type differentiated public from private institutions.

No measure of research productivity is available from the Faculty Roster System. As a proxy for research productivity, data from the 17th Update of the Consolidated Grant Applicant File of the NIH¹⁵ were used to identify receipt of NIH awards. This file contains information on all individuals who have applied for NIH grants or contracts from fiscal year 1938 to 1998 and includes any NIH awards received. Included are awards that specifically target minority investigators, such as Research Supplements for Minority Investigators and Minority Opportunities in Research Faculty Development awards. For each faculty member, we determined whether he/she received any NIH awards during the observation period prior to promotion. Traditional research awards (RO1s) were differentiated from all other NIH awards including individual training grants.

Data for assistant and associate professors were analyzed separately. Bivariate analyses compared the characteristics and rates of promotion of white, API, URM, and other Hispanic faculty members. While the data used in this study approach the population of all US medical school faculty, significance testing with χ² tests is reported to facilitate interpretation of findings. Multivariate analyses used logistic regression models to examine the differential likelihood of promotion of minority and white faculty. These models adjust for cohort, sex, degree, tenure status, receipt of NIH research awards, department, and medical school type. To allow comparison with results from previous studies, results are presented as adjusted odds ratios (ORs) with 99% confidence intervals. However, because promotion is not an uncommon event, ORs tend to overstate relative risk.¹⁶ Hence, adjusted ORs were used to estimate risk ratios¹⁷ and are also presented. All analyses were performed using SAS statistical software (Version 6.12, SAS Institute Inc, Cary, NC).

The study population included 28,953 white, 2997 API, 1053 URMs, and 598 other Hispanic assistant professors and 14,559 white, 1419 API, 280 URMs, and 286 other Hispanic associate professors. The characteristics of faculty who became assistant professors between 1980 and 1989 differed by race and ethnicity (Table 1). All minority assistant professors shared a number of differences compared with white assistant professors. API, URMs, and other Hispanic assistant professors were more likely to be graduates of foreign medical schools or affiliated with other clinical science departments. These faculty members were less likely to be tenured or on tenure tracks, were less likely to be recipients of RO1 and other NIH awards, and were more likely to have appointments in private medical schools. In addition, API and URM assistant professors were more likely to be women. The URM assistant professors were more likely to be graduates of US medical schools, while API assistant professors were more likely to have PhD degrees or to be affiliated with basic science departments. Rates of promotion among faculty who became assistant professors between 1980 and 1989 also differed by race and ethnicity. By 1997, 46% of white assistant professors from these cohorts had been promoted. In comparison, 37% of API, 30% of URM, and 43% of other Hispanic assistant professors had been promoted.

Faculty who achieved associate professor rank between 1980 and 1989 demonstrated similar differences by race and ethnicity. Compared with white associate professors, API, URM, and other Hispanic associate professors were more likely to be graduates of foreign medical schools or affiliated with other clinical science departments and to have appointments in private medical schools. These faculty were less likely to be tenured or on tenure tracks and were less likely to be recipients of RO1 and other NIH awards. API and URM associate professors also were more likely to be women. White associate professors had the highest rate of promotion (50%), while API (44%), URM (36%), and other Hispanic (43%) associate professors had lower rates of promotion.

The numbers of new faculty of specific URM racial/ethnic groups are small (742 black, 150 Mexican American, 123 Puerto Rican, and 38 American Indian or Alaska Native assistant professors and 191 black, 37 Mexican American, 31 Puerto Rican, and 21 American Indian or Alaska Native associate professors). Hence, we limited analyses of these groups to rates of promotion only, and these results need to be interpreted with caution. Among URM faculty, different racial/ethnic groups tended to have comparable rates of promotion with the exception of American Indian or Alaska Natives who had higher rates of promotion. Twenty-nine percent of black, 29% of Mexican American, 30% Puerto Rican, and 45% of American Indian or Alaska Native assistant professors were promoted. Thirty-four percent of black, 24% of Mexican American, 39% of Puerto Rican, and 71% of American Indian or Alaska Native associate professors were promoted.

Among assistant professors, those on tenure tracks were more likely to be promoted than those on nontenure tracks (Table 2). However, within each track, rates of promotion differed by race and ethnicity. Among assistant professors on tenure tracks, API, URM, and other Hispanic faculty were less likely to be promoted than white faculty. Among assistant professors on nontenure tracks, API and URM faculty were also less likely to be promoted than white faculty.

Similarly, among assistant professors, those who had served as principal investigators on NIH awards were more likely to be promoted compared with assistant professors without NIH grant funding. However, among both assistant professors who received NIH awards and those who did not receive such support, API and URM faculty were less likely to be promoted than white faculty. In contrast, other Hispanic faculty were not promoted at lower rates.

Similar patterns were observed among URM associate professors. Compared with white associate professors, URM faculty were less likely to be promoted regardless of tenure status or receipt of NIH grant funding. API and other Hispanic associate professors demonstrated a slightly different pattern of promotion. Compared with white associate professors, API and other Hispanic associate professors on nontenure tracks or who did not receive NIH awards were less likely to be promoted, but those with tenure or on tenure tracks or who did receive NIH grant funding were promoted at rates comparable with white associate professors.

Compared with white assistant professors, minority assistant professors experienced lower rates of promotion in every cohort from 1980 to 1989 (Table 3). These gaps in promotion rates were largest for URM faculty, smaller for API faculty, and not statistically significant for other Hispanic faculty. These gaps appeared constant across the cohorts, and there is no evidence that these gaps had narrowed over time. Similarly, minority associate professors tended to experience lower rates of promotion than white associate professors, although this difference was not statistically significant in most cohorts. There was no evidence that these gaps in promotion rates had narrowed over time.

Because later cohorts had shorter follow-up periods during which promotion could occur and be reported, their promotion rates were expected to be lower. To account for this effect, analyses were repeated counting only those promotions that were reported during a fixed 10-year follow-up period of observation for each cohort (excluding the 1988-1989 cohort which could not be followed up for 10 years). The results of these analyses also demonstrated lower rates of promotion among minority faculty.

Logistic regression analyses that adjusted for cohort, faculty sex, degree, tenure status, receipt of NIH awards, department, medical school type, and tenure status were the strongest predictors of promotion among assistant professors (Table 4). Men, physicians, faculty not affiliated with other clinical departments, and faculty at public medical schools also were more likely to be promoted. After controlling for these covariates, API and URM assistant professors remained less likely to be promoted compared with white assistant professors.

Similar findings, although of a smaller magnitude, were observed among associate professors. Again, men, physicians, faculty with tenure, recipients of NIH awards, faculty in surgical departments, and faculty at public schools were more likely to be promoted. After controlling for these covariates, URM faculty remained less likely to be promoted compared with white associate professors, whereas differences in rates of promotion among API and other Hispanic faculty were not statistically significant.

The major finding of this study is that racial/ethnic minority faculty, at both the assistant and associate professor rank, are lagging in rates of promotion compared with white faculty, even though their representation in academic medicine has steadily increased over time. These findings are consistent with the findings by Pertersdorf et al¹¹ and Palepu et al.¹² However, our study goes beyond previous work by demonstrating an association using a database that includes all medical school faculty and after controlling for cohort effects and multiple potential confounding factors.

One difference with the study by Palepu et al¹² merits comment. In our study, estimates of the influence of race/ethnicity on promotion tend to be smaller than comparable influence reported by Palepu et al. For example, our study found an OR of promotion for URM assistant professors relative to white assistant professors of 0.54 (risk ratio, 0.68) whereas Palepu et al reported an OR of 0.29. Similarly, our study found an OR of promotion for API assistant professors of 0.85 (risk ratio, 0.91) whereas Palepu et al reported an OR of 0.58. We believe that these differences can be attributed to differences in study design and population. Palepu et al used a case-control design, and consequently, the study population included faculty whose first faculty appointment spanned many decades. A much higher proportion of the oldest faculty were white men. Since we have observed that promotion rates were higher in the 1960s and 1970s, inadequate adjustment for cohort effects may overestimate disparities between minority and white faculty. In contrast, our study used a retrospective cohort design that allowed focus on cohorts of faculty who attained their rank between 1980 and 1989.

Our study has several limitations. First, while our study used the official roster of medical school faculty and tracked the progress of more than 33,000 assistant professors and more than 16,000 associate professors, some faculty were excluded because the year they first attained particular ranks is unknown. While about half of these faculty are known to belong to cohorts from the 1970s and earlier, some faculty undoubtedly belonged in the study cohorts. To address possible bias due to the exclusion of these faculty, racial/ethnic disparities in promotion among these faculty and among the faculty included in our study were compared and found to be similar. In addition, to allow focus on the experiences of minority faculty in the typical US medical school, faculty of historically black and Puerto Rican medical schools were excluded from the study population. The promotion experiences of the 10% of black faculty with appointments in historically black medical schools and the 50% of Puerto Rican faculty with appointments in Puerto Rican schools may differ from the experiences of minority faculty presented in this study.

Second, faculty research productivity was measured using receipt of NIH awards. We focused on measures of research productivity because objective and uniform measures of teaching and administrative productivity are generally not available. We chose receipt of NIH awards because this information is available and because the competitive merit review process used by the NIH ensures that these awards uniformly represent high-quality research. We believe that medical schools routinely use this information in making promotion decisions and found that receipt of NIH awards is one of the strongest predictors of promotion. However, other measures, particularly publication of articles in peer-reviewed journals may reflect research productivity more comprehensively and also are used to make promotion decisions. Hence, we cannot exclude the possibility that minority faculty are less likely to be promoted because they publish less frequently.

Third, faculty tenure status captured in the database may reflect status at initial appointment rather than at subsequent points in time. Hence, adjustment for tenure status in our models may not correctly classify faculty who transfer from tenure tracks to nontenure tracks. However, analyses that focus on faculty who were never on a tenure track demonstrate that minority faculty in this group were also less likely to be promoted than comparable white faculty. Promotion was defined as a dichotomous variable, and hence, faculty who were not promoted represent a heterogeneous group. This group includes faculty who sought but were denied promotion, faculty who did not seek promotion, faculty who left academic medicine to pursue other career opportunities, faculty who left academic medicine because they perceived that they would never be promoted, and faculty who were terminated. Analyses of associate professors, who presumably have a significant investment in academic medicine, as well as of assistant professors who remain in academic medicine for at least 3 years demonstrate lower promotion rates among minority faculty. Hence, we do not believe that differences in desire for promotion or commitment to an academic career can explain these differences in promotion rates. However, additional work is needed to examine other reasons faculty are not promoted and to differentiate lack of promotion from attrition.

The findings of our study have implications for faculty members, medical schools, and health policymakers. All faculty members may be discomforted by the low rates of promotion. Only half of the faculty members who became assistant or associate professors in 1980-1981 had been promoted after 17 years of follow-up. Minority faculty members, in particular, may be concerned by the knowledge that they face many barriers to advancement. In general, minority faculty are more likely to be affiliated with departments and medical schools with lower promotion rates, and are less likely to be on tenure tracks or to receive NIH awards, the 2 strongest positive predictors of promotion. API and URM faculty are more likely to be women, for whom lower promotion rates have been well documented. Moreover, after these factors are taken into consideration, minority faculty are still less likely to be promoted.

Medical schools may be equally concerned about these findings. Academic medicine has long been committed to increasing the diversity of the physician workforce. It has championed efforts to ensure a diverse applicant pool to medical school, defended equal opportunity in admissions to medical schools, and has led the opposition to activities to curb affirmative action in medical education. Hence, in the interest of equity, medical schools may perceive the need to examine the reasons racial/ethnic disparities in promotion exist in their institutions. Specifically, they may be encouraged to review promotion criteria that may place too much emphasis on basic research and undervalue contributions in education, administration, and community service often made by minority faculty.

From a practical perspective, medical schools also may be concerned that they have inadequate numbers of minority faculty to properly mentor minority students entering medical school and residency programs. Moreover, as many medical schools expand their faculty practices and compete with managed care organizations for patients, they may wonder if they have adequate numbers of minority faculty to provide culturally competent care and to meet the medical care needs of an increasingly diverse patient population in the United States.

Policymakers also may be interested in these findings. They may appreciate that efforts to train minority researchers and health professional school faculty will be jeopardized if these individuals are unable to find faculty positions with reasonable opportunities for professional growth after completion of training. Policymakers may consider expanding initiatives to support minority investigators and educators during later stages of their careers as medical school faculty.

Additional research is needed to address issues and questions raised by this study. For instance, why are minority faculty less likely to be promoted? Are they isolated or burdened with service duties that limit their academic pursuits? Are these faculty members subjected to unconscious discrimination as suggested by some?¹⁸ Or are less subtle factors at play? Do culture, language, or skin color factor into the promotion process? Can faculty development programs be devised to help minority faculty overcome barriers to promotion or will medical educators continue to mirror our tiered system of health care delivery? Answers are needed to enable faculty and medical schools to better understand the reasons for racial/ethnic disparities in faculty promotion and to ensure an equitable system of professional advancement for all faculty members.