Sex Differences in Academic Rank and Publication Rate at Top-Ranked US Neurology Programs

Key Points

Questions  What is the proportion of male vs female neurologists at each academic faculty rank, and how does the number of publications differ between sexes at each rank?

Findings  In this cross-sectional study that included 1712 academic neurologists, 31% were women and 69% were men. Men outnumbered women at all academic faculty ranks, and the discrepancy increased with advancing rank; after adjusting for years since medical school graduation, men had a significantly larger number of publications than women at all academic ranks, but the disparity in publication number decreased with advancing rank.

Meaning  Future studies should explore the reasons for the underrepresentation of women among academic neurologists.

Importance  Women are underrepresented in academic neurology, and the reasons for the underrepresentation are unclear.

Objective  To explore potential sex differences in top-ranked academic neurology programs by comparing the number of men and women at each academic faculty rank and how many articles each group has published.

Design, Setting, and Participants  Twenty-nine top-ranked neurology programs were identified by combining the top 20 programs listed on either the 2016 or 2017 Doximity Residency Navigator tool with the top 20 programs listed in the US News and World Report ranking of Best Graduate Schools. An internet search of the departmental websites was performed between December 1, 2015, and April 30, 2016. For each faculty member on a program site, the following biographical information was obtained: first name, last name, academic institution, sex, academic faculty rank, educational leadership (clerkship, fellowship, or residency director/assistant director), and year of medical school graduation.

Main Outcomes and Measures  To compare the distribution of men vs women and the number of publications for men vs women at each academic faculty rank. Secondary analyses included Scopus h-index, book authorship, educational leadership (clerkship, residency, or fellowship director/assistant director), and clinical activity as inferred through Medicare claims data in men vs women after controlling for years since medical school graduation.

Results  Of 1712 academic neurologists in our sample, 528 (30.8%) were women and 1184 (69.2%) were men (P < .001). Men outnumbered women at all academic faculty ranks, and the difference increased with advancing rank (instructor/lecturer, 59.4% vs 40.5%; assistant professor, 56.7% vs 43.3%; associate professor, 69.8% vs 30.2%; and professor, 86.2% vs 13.8%). After controlling for clustering and years since medical school graduation, men were twice as likely as women to be full professors (odds ratio [OR], 2.06; 95% CI, 1.40-3.01), whereas men and women had the same odds of being associate professors (OR, 1.04; 95% CI, 0.82-1.32). Men had more publications than women at all academic ranks, but the disparity in publication number decreased with advancing rank (men vs women after adjusting for years since medical school graduation: assistant professor [exponentiated coefficient, 1.85; 95% CI, 1.57-2.12]; associate professor [1.53; 95% CI, 1.22-1.91]; and full professor [1.36; 95% CI, 1.09-1.69]). Men had a higher log Scopus h-index than women after adjustment (linear coefficient, 0.44; 95% CI, 0.34-0.55). There was no significant association between sex and clinical activity (linear coefficient, 0.02; 95% CI, −0.10 to 0.13), educational leadership (OR, 1.09; 95% CI, 0.85-1.40), or book authorship (OR, 2.75; 95% CI, 0.82-9.29) after adjusting for years since medical school graduation.

Conclusions and Relevance  Men outnumber women at all faculty ranks in top-ranked academic neurology programs, and the discrepancy increases with advancing rank. Men have more publications than women at all ranks, but the gap narrows with advancing rank. Other measures of academic productivity do not appear to differ between men and women.

Quiz Ref IDWomen constituted 49.8% of 2016 medical school matriculants¹ and 46.1% of residents and fellows in Accreditation Council for Graduate Medical Education–accredited programs in 2013.² In 2013, 44.9% of neurology residents were women. A sex gap persists among all practicing physicians, however. In 2013, women constituted 32.6% of the overall physician workforce, and only 26.9% of the 13 142 active neurologists. A nearly identical sex gap persists in academic medicine, including neurology, and the disparity increases with advancing academic rank.^2-4

In medical schools and across academia, women have fewer publications than men. From 1990 to 2012, 27.2% of authors of all academic publications in the natural sciences, social sciences, and humanities were women.⁵ Among all medical school faculty members in 2014, women had fewer total and first- or last- author publications (mean total, 11.6 for women vs 24.8 for men; difference, −13.2; 95% CI, −13.6 to −12.7).⁴Quiz Ref ID Among 980 articles from the journals Nature, Neuroscience, and Neuron, only 29.0% of the first authors and 17.8% of the senior authors were women.⁶

In this study, we performed a systematic analysis of top-ranked US academic neurology departments to determine the distribution of men and women at each academic faculty rank. Quiz Ref IDWe further evaluated the number of PubMed⁷ publications of male vs female neurologists after controlling for years since medical school graduation. Because publication rate is not the only measure of academic productivity, we also compared Scopus h-index, book authorship, educational leadership (clerkship, residency, or fellowship director/assistant director), and clinical activity as inferred through Medicare claims data⁸ in men vs women after controlling for years since medical school graduation.

A neurology program was considered a top-ranked program if it was among the top 20 programs listed on either the 2016 or 2017 Doximity Residency Navigator tool,⁹ sorting by reputation, or the US News and World Report ranking of Best Graduate Schools.¹⁰ Twenty-nine unique programs were identified. Faculty members at these programs were included in our analysis if they had completed a residency in adult neurology and held an active clinical appointment. This study was approved and participants were deemed exempt from providing informed consent by the institutional review board of the University of Michigan, Ann Arbor.

An internet-based search of departmental websites was performed between December 1, 2015, and April 30, 2016. For each of 1712 eligible faculty, the following biographical information was obtained: first name, last name, academic institution, sex, academic rank, educational leadership (clerkship, fellowship, or residency director/assistant director), fellowship specialty, medical school attended, year of medical school graduation, residency program attended, and year of residency graduation. When this information was not readily apparent on a departmental website, study team members contacted the neurology residency program coordinator to provide or confirm biographical information of the faculty member. Sex was determined by review of faculty photographs on department websites. Photographs were available for all faculty. Because of small numbers, we combined neurologists described as lecturer, clinical instructor, or instructor into 1 group that we labeled “instructor/lecturer.”

Individualized measures of faculty publication number and citation impact were obtained from Scopus¹¹ and PubMed.⁷ On Scopus, an author search was performed by cross-referencing the name and institution of the faculty member. The number of published documents and h-index were recorded. On PubMed,⁷ an author search was conducted using the last name, first name, and middle initial of the faculty member. The search was refined by selecting for a given institution, using the author’s full name, or by adding the criterion “and neurology” after the author’s name. The total number of publications, number of first-author publications, and number of last-author publications were recorded.

On August 1, 2017, we searched for books available from Amazon¹² using the search term neurology. On the first 10 search pages that resulted, there were 113 unique books. Multiple printings or editions of a book were only counted once. Amazon lists up to 2 authors or editors for each book on the search page. Each of these names was considered an authorship.

We used summary Medicare provider utilization and payment data⁸ from 2012 to 2015 to explore whether “clinical time” varied by sex. These data summarize which procedures were performed by each clinician in the United States and how many of each type of procedure they performed. To estimate clinical time, we linked each procedure code to an estimated total procedure time and then summed the total procedure time across all the procedures and years for each clinician.¹³ Clinicians with a specialty of neurology were linked by first name and last name to the publication database of the 1712 academic neurologists in our sample, which matched data for 79% of the respondents in the publication database.

We used descriptive statistics to investigate the proportion of men vs women neurologists, as well as the distribution of men vs women by academic faculty rank, across all of our selected institutions. We used a χ² test to compare the proportion of men vs women overall and within each academic faculty rank.

We compared men and women with respect to years since medical school graduation. Because our data were not normally distributed, we performed a Wilcoxon rank sum test to compare the years since medical school graduation by sex overall as well as within each academic rank. We used a logistic regression model to estimate full professorship for men and a model to estimate associate professorship for men while controlling for clustering at the facility level. We repeated the analysis while also controlling for years since medical school graduation.

Because we found that the publication characteristics we analyzed (PubMed publications, first-author publications, last-author publications, and Scopus h-index) were correlated, we elected to focus on a single characteristic in our primary analyses. We chose the number of PubMed⁷ publications because we thought this would be the most readily understood of the publication variables investigated.

We used a Wilcoxon rank sum test to investigate the difference in publication number by sex within each academic rank. Next, we fit a negative binomial model for each rank while adjusting for years since medical school graduation. Our full regression model included male sex, academic faculty rank, and years since medical school graduation. We used female sex and assistant professor as the referent in this model.

The association between secondary outcomes and sex was analyzed using regression models (logistic regression for binary outcomes and linear regression for continuous outcomes) adjusting for clustering at the facility level and including years since medical school graduation and sex as independent variables. Given the large right-skews, h-index and clinical time were log transformed.

For our secondary analyses, we analyzed the association of log-transformed Scopus h-index and sex by using linear regression adjusted for years since medical school graduation. We used logistic regression to investigate the association of sex with being an educational leader after controlling for years since medical school graduation. In addition, we used logistic regression to examine the association between sex and book authorship. We used linear regression to examine the association between sex and clinical time after controlling for years since medical school graduation. All analyses were performed using R, version 3.3.2,¹⁴ and Stata, version 14.0.¹⁵

In our sample of 1712 academic neurologists, 528 (30.8%) were women and 1184 (69.2%) were men (P < .001). Table 1 lists the 29 academic institutions included in our sample, the number of neurologists on faculty at each academic center, and the percentage of neurologists who were men. Table 2 shows the distribution of men vs women within each academic faculty rank and shows the mean years since medical school graduation for men vs women within each academic rank. Men at the rank of instructor/lecturer had graduated from medical school earlier than their female counterparts (mean [SD] years since medical school graduation, 25.7 [16.2] vs 13.8 [8.7] years; P = .003), associate professor (25.8 [9.3] vs 22.0 [7.0] years; P = .001), and professor (34.9 [10.5] vs 30.3 [7.5] years; P < .001). There was no significant difference in years since medical school graduation for men vs women at the assistant professor rank.

When clustering at the facility level was accounted for, men were 4 times as likely as women to be full professors (odds ratio [OR], 3.98; 95% CI, 3.00-5.29). After additional adjustment for years after medical school graduation, men were twice as likely as women to be full professors (OR, 2.06; 95% CI, 1.40-3.01). When clustering was accounted for, men had the same odds of being associate professor as women (OR, 1.04; 95% CI, 0.82-1.32). After additional adjustment for years since medical school graduation, the odds remained the same (OR, 1.00; 95% CI, 0.71-1.20).

Men had significantly more publications than women in unadjusted analysis at the ranks of assistant professor (mean number [SD] of PubMed⁷ publications, 20.2 [23.5] vs 10.5 [11.8]; P < .001), associate professor (48.6 [58.8] vs 32.8 [27.5]; P = .03), and full professor (144.0 [131.9] vs 95.2 [90.8]; P < .001). After adjustment for years since medical school graduation, men had more publications than women at the ranks of assistant professor (exponentiated coefficient, 1.85; 95% CI, 1.57-2.17), associate professor (1.53; 95% CI, 1.22-1.91), and full professor (1.36; 95% CI, 1.09-1.69) (Table 3).

Men also had a higher log Scopus h-index than women after adjustment for years since medical school graduation (linear coefficient, 0.44; 95% CI, 0.34-0.55). There was no significant association between sex and clinical activity, educational leadership, or book authorship after adjustment for years since medical school graduation (Table 3).

Men outnumber women in top-ranked academic neurology programs; these findings are consistent with previous studies, both for neurology and for other medical specialties.^3,4 Even at entry-level instructor/lecturer positions, the ratio of male to female neurology faculty is higher than the ratio of male to female neurology residents. This finding suggests that women are less likely to accept junior faculty positions in top-ranked academic neurology departments.

The disproportionate number of men among academic neurologists increases with advancing rank. This is to be expected because individuals with more senior academic rank are more likely to have graduated when the ratio of male to female medical students was higher. However, even after controlling for years since medical school graduation, fewer women than men have achieved the rank of full professor.

Our results may indicate ongoing discrepancies in the promotion of men and women in top academic neurology programs, but other interpretations are possible. Academic department administrators have increasingly recognized that scholarship encompasses a wide variety of activities. Departments have developed a choice of academic career paths that differs by criteria for achieving promotion and also by the degree of which promotion is considered important. Our methods did not distinguish among different academic career paths with distinct criteria for promotion. Women may tend to select appointments in academic career paths that are less focused on promotion to senior faculty rank. Quiz Ref IDSimilarly, whereas the traditional measures of academic productivity have been publication rate, publication impact, and grant support, there is a growing recognition of the importance of other factors, such as the quality and quantity of teaching, the development of educational resources, and administrative effectiveness. Ideally, studies of sex disparity would incorporate these nontraditional criteria for promotion; unfortunately, these criteria vary widely across institutions. Although number of publications represents only 1 component of productivity in most academic medical schools, it continues to be an important component for promotion and is the one factor that can be evaluated most consistently across institutions. It remains relevant to examine publication factors when studying the subject of sex disparity. Quiz Ref IDWe found that male neurologists in top-ranked academic neurology programs had more publications than women at every rank above instructor/lecturer, even after controlling for years since medical school graduation. The sex disparity in publication volume was highest among assistant professors, lower among associate professors, and lowest among professors, and the difference was significant at all ranks above the instructor/lecturer level.

The cross-sectional nature of our study precludes inferences about the association between the number of publications and promotion. Differences regarding the number of publications expected for men and women may become less pronounced at higher academic ranks. Alternatively, the lower rate of publication among women overall may be one of the reasons fewer women achieve high academic ranks. The women who do achieve high academic ranks publish more frequently. A prospective study could help distinguish between these 2 explanations.

To investigate the possibility that promotions are influenced less by the quantity of publications than by quality, we analyzed the Scopus h-index as a function of sex. Our results paralleled the results of our analysis of the number of publications. To investigate the role of scholarly activities other than published articles, we also performed analyses of sex differences in book authorship, educational leadership, and clinical productivity. Women and men did not differ with respect to these variables.

A variety of explanations have been proposed for the sex gap in academic faculty rank and publications, including asymmetric home or childcare responsibilities, cultural stereotypes, professional isolation, and different career motivations.^16-19 For example, some literature suggests that women in academic medicine are less focused on salary, reputation, and leadership positions than their male colleagues, even though women do not differ in their patient care, teaching, or publication goals.²⁰ Our methodology was unable to account for variables, such as leave of absence and part-time appointment, that may differentially affect one sex more than another. Compared with men, women may be more likely to be recruited for employment positions that emphasize teaching and mentoring rather than research, or women may be more inclined to choose such positions. Previous studies have suggested a lack of same-sex mentorship for female junior faculty,^21-23 and bias favoring men in the National Institutes of Health award application review process has been suggested.²⁴ In addition, there may be increased attrition of women in academic medicine at higher ranks.²⁵ Overt and unconscious sex bias may also play a role in differences in publication rates between male and female neurologists.^26,27

Our study has several limitations. Our sample included faculty from top-ranked US academic neurology programs, potentially limiting the relevance of our findings more generally to other types of academic programs and to programs located outside the United States. Unlike other studies of sex disparity in academic medicine, we looked directly at departmental websites for information about faculty rank, sex, and year of medical school graduation. We assumed that these departmental websites would be more current than the general databases used in previous studies, but we made no attempt to verify this assumption. We are reassured that a core finding, that women constituted 30.8% of academic neurologists in our sample, is consistent with previous research using different methodologies. In a 2012 Association of American Medical Colleges benchmarking survey, 31% of 1073 full-time neurology faculty with medical degrees were women.³ In another recent study, 31.6% of 3880 neurology faculty in 2014 were women.⁴ As discussed above, some important variables, such as leave of absence, administrative and teaching responsibilities, and the specific promotion criteria of different academic career paths, were not surveyed. The strengths of our study include systematic data acquisition and a large sample size.

Our results in academic neurology can be viewed as either disappointing or encouraging, depending on whether they reflect persistent barriers to women trying to achieve similar goals as men, or whether they reflect a system that supports women with different goals altogether. If the former is true, it will be important to develop programs to heighten awareness of diversity in academic neurology. If the latter is true, academic neurology departments should be encouraged to foster a variety of career paths and expectations for all faculty. Future studies should explore reasons for the underrepresentation of female neurologists in academia. Prospective methods should examine the association of the number of publications and other aspects of productivity on promotion for men vs women in academic neurology.

Accepted for Publication: December 7, 2017.

Corresponding Author: Mollie McDermott, MD, MS, Department of Neurology, University of Michigan, 1500 E Medical Center Dr, Special Postal Code 5855, Third Floor, Reception C, Ann Arbor, MI 48109-5855 (mcdermom@med.umich.edu).

Published Online: April 2, 2018. doi:10.1001/jamaneurol.2018.0275

Author Contributions: Drs McDermott and Burke had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Gelb, Shelgikar, London.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: McDermott, Gelb, Pawloski, Shelgikar, London.

Critical revision of the manuscript for important intellectual content: McDermott, Gelb, Wilson, Burke, Shelgikar, London.

Statistical analysis: McDermott, Wilson, Burke.

Obtained funding: London.

Administrative, technical, or material support: Pawloski, Burke, London.

Study supervision: London.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was funded by the Jerry Isler Neuromuscular Fund.

Role of the Funder/Sponsor: The funding organization had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.