Gender disparity in authorship broadly persists in medical literature, little is known about female authorship within pulmonary medicine.
MethodsA bibliometric analysis of publications from 2012 to 2021 in 12 journals with the highest impact in pulmonary medicine was conducted. Only original research and review articles were included. Names of the first and last authors were extracted and their genders were identified using the Gender-API web. Female authorship was described by overall distribution and distribution by country/region/continent and journal. We compared the article citations by gender combinations, evaluated the trend in female authorship, and forecasted when parity for first and last authorship would be reached. We also conducted a systematic review of female authorship in clinical medicine.
Results14,875 articles were included, and the overall percentage of female first authors was higher than last authors (37.0% vs 22.2%, p<0.001). Asia had the lowest percentage of female first (27.6%) and last (15.2%) authors. The percentages of female first and last authors increased slightly over time, except for a rapid increase in the COVID-19 pandemic periods. Parity was predicted in 2046 for the first authors and 2059 for the last authors. Articles with male authors were cited more than articles with female authors. However, male-male collaborations significantly decreased, whereas female-female collaborations significantly increased.
ConclusionsDespite the slow improvement in female authorship over the past decade, there is still a substantial gender disparity in female first and last authorship in high-impact medical journals in pulmonary medicine.
In modern-day society, gender disparities broadly persist in many professions. Despite the increasing numbers of women in the medical field, they are under-represented in leadership positions,1,2 editorial boards,3-5 and conference speakers.6,7 As authorship plays a crucial role in faculty promotion, grant application, and academic position,8-10 understanding gender distribution of prominent author positions helps clarify the roles women hold in an academic environment.
In recent years, gender equity in academic medicine has attracted interest from different medical disciplines.11-17 Except for nursing,18 obstetrics and gynecology,19 and dermatology,20 most clinical specialties have reported that women have lower representation than men in published articles,12-17 especially in the last author position.12,14-17 In articles published between 2008 and 2018 in 40 critical care medicine journals, fewer than one-third of first authors and one-fifth of last authors were women.14
Gender disparities in authorship vary between geographic areas and journals. The reported proportions of first and last women authors in countries vary by medical specialties. For example, the percentage of women first authors in radiology was reported to rank high in Asian countries like China and South Korea,21 in contrast to their representation in surgical medicine,22 neurosurgical science,23 and critical care medicine.14,24,25 Female authorship in high-impact journals is low, especially in the last author position.26,27 Also, articles with male authors are cited more often than articles with female authors.28,29 so we sought to explore if there are similar trends in pulmonary medicine journals.
This study aimed to investigate the gender distribution of authorship among high-impact medical journals in pulmonary medicine over the past decade. Additionally, to compare authorship distribution in pulmonary medicine to other medical specialties, we reviewed the publications with the topic of gender distribution in the academic literature on clinical medicine. Because the first authorship typically belongs to the person responsible for most of the work, and the last author is often considered the senior author who supervises the project, we used those positions for this paper.
Material and methodsData sourceWe included the top eight journals that were ranked by journal citation reports (JCRs, 2020) in the category of the respiratory system (https://jcr2.clarivate.com), according to the Web of Science database core collection. These journals included Lancet Respiratory Medicine, American Journal of Respiratory and Critical Care Medicine (AJRCCM), European Respiratory Journal (ERJ), Journal of Thoracic Oncology, Journal of Heart and Lung Transplantation, Chest, Thorax, and European Respiratory Review. In addition, we included four general medical journals with the highest impact factors (British Medical Journal [BMJ], Journal of the American Medical Association [JAMA], Lancet, and New England Journal of Medicine [NEJM]). For the four general medical journals, only articles with study content related to pulmonary medicine were included, with article titles and abstracts screened independently by two researchers. Any disagreement regarding if it was pulmonary medicine related article was resolved by discussion with a third researcher.
Articles published in the 12 journals between January 1st, 2012, and December 31st, 2021 were searched on Web of Science core collection. Only original research studies and reviews were included. Articles with a single author were excluded.
Article information extractionThe article information was obtained from Web of Science database and extracted by Python programming software, including publication date, journal, article type, title, abstract, citation, and author information. Detailed information about data extraction is available in the Supplement (page 2). Gender was identified for authors listed in the first and last positions. For the authors who had multiple affiliations, the affiliation listed first was used.
Author's gender identificationThe extracted authors’ names were entered in Gender-API (https://gender-api.com/), which has been used to identify gender based on name with a reported accuracy of over 98%.30,31 Gender-API reports gender (male, female, or null) and the predicted accuracy of the determination. If the author's gender was not identified (null) or the predicted accuracy was < 60%, a manual search on Google and/or social media was conducted using the author's name and affiliations listed in the publication. The author's photo and/or biographies by pronouns were used to identify gender. When either or both the first and last authors’ genders could not be identified, the articles were excluded.
Evaluation on the publications of female authorship in clinical medicineWe also reviewed the publications on female authorship in clinical medicine. Two investigators independently searched the PubMed database for articles with a focus on female authorship in clinical medicine published in English from January 1st, 2012 to December 31st, 2021. The literature search strategy and eligibility criteria for including studies are available in the Supplement page 3.
To explore the association between the percentages of female faculty/physicians and female authorship in different disciplines of clinical medicine, we extracted the percentages of female physicians and female faculty reported by AAMC.32 The percentages of female first and last authorship from articles published from January 1st, 2018, to December 31st, 2019, in relevant disciplines were calculated as an average.
Statistical analysisThe distribution of female first and last authors was reported as percentages by overall, article type, countries/regions/continents, and journals. The differences were analyzed using chi-square test. To evaluate female first and last authorship changes over time, linear regression was performed to forecast the year in which parity would be reached between women and men for first and last authorship. Similarly, linear regression was used to analyze the changes over time in female first and last authors by country/region, journal, and gender pairs from 2012 to 2021. To understand the pattern of the pairs of first and last authors, we paired the first and last authors by gender to create four categories, including: (1) female-female, (2) female-male, (3) male-female, and (4) male-male.
The Kolmogorov-Smirnov test was used to test the normality of distribution for article citations, and Mann-Whitney test was used to compare the article citations with different pairs of genders. Multi-factor logistic regression was performed to assess the influential factors with the first and last author's gender. Factors in this model included year (reference: 2012), article type (reference: original research), region of the author (reference: North America), and journal (reference: European Respiratory Review). References were chosen as they were closest to the average overall distribution data.
The first and last authors with the highest number of articles in pulmonary medicine during the study period were included in the top 100 most prolific authors.
Statistical analysis was conducted using SPSS (version 22.0; IBM Corp, Armonk, New York), and visualization of statistics was performed using GraphPad Prism version 8 (GraphPad Software, La Jolla, CA). A two-sided P< 0.05 was considered significant.
Results14,875 articles were included in our study, including 13,297 original research studies and 1,578 reviews (Fig. 1A). Of those included articles, 14,411 had the gender of the first author identified by gender-API, with a probability of 99% (98-100). Likewise, the gender of last authors was identified by gender-API in 14,328 articles, with a probability of 99% (99-100). In the remaining articles, a manual search was performed to determine the gender.
Flow diagram for selecting articles for (A) selection of respiratory system articles published between 2012 and 2021 and (B) evaluation on clinical medicine. Abbreviations: BMJ, British medical journal; JAMA, Journal of the American Medical Association; NEJM, New England Journal of Medicine.
For the evaluation, 1,331 articles were screened and 117 articles were included (Fig. 1B).
Manuscript characteristics and authorship position by genderOf the 14,875 articles included, 37.0% were first-authored by a woman, which was higher than the percentage of female last authors (37.0% vs 22.2%, P< 0.001) (Table 1). For original research and review articles, female first authors were similar (37.1% vs 35.9% respectively), which was also noted for last authors (22.2% vs 22.1% respectively).
Manuscript characteristics and authorship position by gender.
Abbreviations: CI, confidence interval; ERJ, European Respiratory Journal; AJRCCM, American Journal of Respiratory and Critical Care Medicine; BMJ, British medical journal; JAMA, Journal of the American Medical Association; NEJM, New England Journal of Medicine.
The percentage of female first authors increased significantly from 2012 to 2021, with an average annual increase of 0.44% (95%CI, 0.13-0.75%, P= 0.011) (Fig. 2A), with parity predicted in 2046 (Fig. 2B). Likewise, the percentage of women in the last author position also increased significantly, with an average annual increase of 0.66% (95%CI 0.40–0.91%, P< 0.001). At this rate, parity would be reached in 2059.
During the COVID-19 pandemic 2020 and 2021), of 3372 articles published in the 12 journals, 39.0% were first-authored by a female and 25.5% had a female last author (Fig. 2A). While in the two years before COVID-19 (2018 and 2019), of 2581 articles, 36.3% and 23.8% had women first and last authors respectively. More females were first (P= 0.032) and last (P= 0.016) authors in the COVID-19 pandemic than in non-pandemic periods.
Geographic distribution of female authorsOver 85% of the first and last authors had North American (44.9% and 46.1% respectively) and European affiliations (40.3% and 40.1%, respectively) (eFigure 1A in the Supplement). Regarding authorship on different continents, Oceania had the highest female first (46.2%) and last (25.5%) authors, whereas Asia had the lowest female first (27.6%) and last (15.2%) authors. Over the past decade, female first authors increased significantly in North America (P= 0.002) and Asia (P= 0.028); and female last authors increased significantly in North America (P= 0.026), Europe (P= 0.004), and Oceania (P= 0.045) (eFigure 2).
When all countries were analyzed together, the United States had the most first and last female author positions (38.1% and 39.5%) in terms of the number of publications, respectively (eFigure 3A). The Netherlands had the highest proportion of female first (52.5%) and last (27.6%) authors by country; and Japan had the lowest proportions (10.1% and 2.6% respectively) (eFigure 3B). Female first authors increased significantly over time in the United States (β=0.66, 95% CI 0.31-1.02%, P= 0.003), whereas female last authors increased significantly over time in several countries, including the United States, the Netherlands, Australia, Switzerland, and Denmark (eFigure 4).
Author gender distribution in different journalsIn terms of journals, Thorax had the highest female first authors (44.9%), and NEJM had the lowest female first authors (27.3%). In contrast, NEJM had the highest female last authors (27.6%) and Journal of Heart and Lung Transplantation had the lowest female last authors (18.1%) (eFigure 5). Other than NEJM, all journals had a higher percentage of female first authors than female last authors.
Over time, female first authors increased significantly in the European Respiratory Review, Chest, and BMJ; and decreased significantly in Lancet Respiratory Medicine (β=-1.78, 95%CI -3.18 to -0.38%, P= 0.019). Female last authors increased significantly over time only in the Journal of Thoracic Oncology (β=1.12, 95%CI 0.71-1.54%, P= 0.001) (eFigure 6).
First and last author gender pairsOf the included articles, 50.9% had males in the first and last author positions (male-male), followed by female-male (26.9%), male-female (12.2%), and female-female (10.0%). Over the past decade, female-female author pairs increased (P= 0.001), and between male-male author pairs decreased (P< 0.001) (eFigure 7).
High productivity authorsAmong the 100 most prolific first and last authors in pulmonary medicine research during the study period, only 10 and 14 were women, respectively (eFigure 8).
Gender disparity in citationsArticles with women as first and last authors were cited less than those with men as first (25 [9-54] vs 29 [11-64] citations, P< 0.001) and last authors (24 [9-56] vs 28 [10-61] citations, P< 0.001) (eTable 1). In year-by-year data, this pattern was consistent each year from 2018 to 2021 for the articles with female first authors and in 2020 for articles with female last authors. Articles with females as both first and last authors were cited the fewest times (22 [7-49]) (eTable 2), whereas articles with males as first and last authors were cited most (29 [11-65]). Comparisons across the four pairs of first/last author genders were statistically significant (P< 0.001).
Multivariable analyses.
Abbreviations: CI, confidence interval; ERJ, European Respiratory Journal; AJRCCM, American Journal of Respiratory and Critical Care Medicine; BMJ, British medical journal; JAMA, Journal of the American Medical Association; NEJM, New England Journal of Medicine.
When using the percentages of female authorship in 2012 as a reference, the odds of female first and last authorship increased overtime (Table 2). Moreover, compared to the female authorship in North America, the odds of being the first author were higher in Europe, Oceania, and South America, but lower for the last author in Europe. In Asia, the odds were lower for the first and last authorship. Furthermore, compared to female authorship in the European Respiratory Review, which had a percentage of female authors equal to the average of the 12 included journals, the odds of female first authorship were higher in Thorax, while Lancet Respiratory Medicine, JAMA, and NEJM had lower odds. For Journal of Heart and Lung Transplantation, the odds were lower for the first and last authorship.
Evaluation of female authorship in biomedical journalsThe proportions of female authorship in different disciplines of clinical medicine varied (Table 3). Using the aggregated data published between 2012 and 2021 to generate the overall proportion of female authorship in different disciplines, the top three disciplines ranked by the percentages of female first authorship were nursing (83.0%), obstetrics and gynecology (60.3%), rheumatology (51.5%). Furthermore, the top three disciplines ranked by the female last authorship were nursing (72.0%), obstetrics and gynecology (41.7%), dermatology (35.6%). There were increasing proportions of female first and last authors in most disciplines, such as oncology, pediatric surgery, and anesthesiology (eFigs. 9 and 10).
Percentages of female first and last authors in publications from 2012 to 2021 and percentages of female physicians and faculty reported by AAMC.
Abbreviations: AAMC, Association of American Medical Colleges; NR, Not Reported; a, Data from Our Research Results.
In this study, we found that the percentages of female first and last authors increased slightly between 2012 and 2021, especially a significant increase during the COVID-19 pandemic periods. Parity was foreseen in 2046 for the first authors and 2059 for the last authors. Articles with male-male pairs were cited more than articles with female-female pairs. Male-male pairs decreased significantly from 2012 to 2021, whereas female-female pairs increased significantly.
Our findings are consistent with several recent studies illustrating gender disparities in academic publications.13,15-17 Our findings were not surprising considering the low number of female physicians in pulmonary medicine (12.3%) and critical care medicine (26.8%) in the United States, according to AAMC.32 That said, the percentage of female first authors (37.0%) in our study is relatively high among all the medical specialties (Table 3).
In many specialties, including emergency medicine, internal medicine, and surgery, the COVID-19 pandemic reduced female researchers’ productivity,33,34 due to family responsibility, childcare needs,33,35 and the lack of patients or research materials.36 In contrast, we found a significant increase in publications with females as the first and last authors. This might be explained by the fact that pulmonary medicine was directly impacted by the COVID-19 pandemic. In general, the lack of childcare support and flexible working hours might explain the reduced output of female researchers.
In our study, the percentage of female first authors was nearly double that of female last authors and might be explained by the lack of female senior research mentors.37 According to the 2021 AAMC report, the percentages of female faculty in the ranks of assistant professors and instructors is similar to that of male faculty (29% vs 29%),32 while the percentages of female faculty in higher ranks (Associate and full Professors) is lower than that of male faculty (13% vs 26%).32 Our study suggests that this disparity might eventually be resolved, but this will take 20-30 years at the current rate of change. However, efforts to promote female authorship, grant awards, and academic leadership positions might accelerate the process.29 We also found a growth in female-female pairs and a decline in male-male pairs, which might be explained by the feminization of the workforce38 and the increasing numbers of female senior researchers in the field.39
Regarding citations of articles authored by females and males, we found articles with female first authors were cited fewer times than articles with male first authors within four years of publication. These findings agree with results from research by Chatterjee and Werner, who found that articles by women as primary authors had fewer citations within four years of publication.29 Interestingly, we found that the difference in citations between female and male first authors was not significantly different after four years of publication. Compared to females, males tend to have more opportunities to present at conferences, participate in clinical practice guideline development, serve as content experts, and share their research findings,40-43 which may result in more citations shortly after study publication.27 In the long term, citations might be more dependent on the content and relevance of the publication, thus the citations become similar.
Finally, we found lower percentages of female first and last authors in Asian countries, in contrast to high percentages of female authors in European countries. In particular, we found that among the developed countries, the Netherlands has the highest percentages of female authorship in contrast to the lowest proportions in Japan, this finding agrees with research in critical care medicine44 and transplantation research.45 The discrepancy between Asian and European countries might be explained by differences in cultural expectations and roles/responsibilities of women in society and family,45 the overall lack of research training/opportunities, and the lack of female mentors and academic leaders in educational institutions.14,44,46 Likewise, the only two journals that exceeded the average percentage of overall female first authorship among the 12 journals were Thorax and ERJ, which both belonged to European scientific/medical societies, and both societies made rules or policies on female representations and participation,47,48 with an endeavor to achieve gender equality. To what extent these rules or policies contribute to female authorship remains unclear. Recently, Chest journal starts to collect authors’ gender during manuscript submission and whether these data help improve gender equality requires more investigation.
Our findings suggest that female researchers should have the same opportunities to disseminate the findings of their research as their male counterparts. Solutions may include increasing mentorship and sponsorship opportunities, encouraging and inviting women to pursue formal research training, and for senior investigators to invite early-career women onto their teams. Further studies are needed to identify factors that influenced our findings.
Strengths of our study include the large sample size and comprehensive analyses of gender distribution in different geographic areas, journals, trends over time, collaborations between first and last author genders, and article citations in high-impact journals of pulmonary medicine. Notably, the journal impact factor is just one measure of scientific output as highlighted and advocated by the Declaration on Research Assessment (DORA), thus high-impact journals included in this study might not represent their true academic reputation and scientific influence. Additionally, we conducted a thorough literature search and performed an evaluation of female authorship. This study also has limitations. First, we only included original research and review articles published by pulmonary medicine journals with the highest impact factors, thus our findings may not be generalizable to all journals in academic medicine. Second, gender was determined by an online web tool, which may have resulted in misclassification, even though genders that were assigned with a low probability of accuracy were manually evaluated. Third, there was a risk of misclassifying publication types. Fourth, single-author articles were excluded. Lastly, the exact number of female researchers (the denominator) in academic pulmonary medicine worldwide is unknown.
ConclusionIn conclusion, this study demonstrates a significant difference between female and male authorship in medical journals pertaining to pulmonary medicine. These findings appear to differ between countries and continents, suggesting that some regions and academic environments might be more supportive of female researchers. We suggest that efforts be made by academic institutions globally to ensure equal opportunities for any individual pursuing research.
Author contributionsJL and BD designed this project. YR and FG conducted the screening and data extraction for the study of academic literature in pulmonary medicine, while YR and HJH performed literature search, screening, and data extraction for the evaluation. FG performed the data analysis. BD participated in the resolution of discrepancies on data extraction. FG and YR had full access to the data and verified the data. JBS, SLS, and SM interpreted the data and provided critical edits and comments on the manuscript. JL, BD, YR, FG, and HJH attended bi-monthly web meetings. JL drafted the manuscript, all authors reviewed the manuscript for important intellectual content, and approved the final manuscript. FG, YR, HJH, and BD equally contributed to the overall project described in this article. JL was responsible for the decision to submit the manuscript.
Funding/supportThis study did not receive any funding support.
Data Sharing StatementData will be available immediately after article publication to researchers who provide a proposal for any purpose of analysis.
