Sex Disparities in Ophthalmic Research: A Descriptive Bibliometric Study on Scientific Authorships

Key Points

Question  Are there sex disparities in ophthalmic research?

Findings  In this bibliometric analysis of authorships from ophthalmic articles, women held 34.9% of authorships, were underrepresented in terms of prestigious last authorships, published fewer articles, and were less commonly cited in a key role compared with men. Their representation showed cross-journal uniformity but transnational differences; in recent years, percentages of women with first or last authorship have increased.

Meaning  These data suggest the integration of women into ophthalmic research and that male researchers, on average, still dominate senior ranks; however, with increasingly more women in ophthalmology, sex inequalities may diminish in the future.

Importance  Previous studies examined sex distributions in different medical faculties, especially because increasingly more women entered the medical field in recent decades. Little is known at present about the female representation in ophthalmic research.

Objective  To clarify sex equalities in ophthalmic research by evaluating the representation of female authorships.

Design and Setting  This bibliometric analysis included original English-language articles published in ophthalmologic journals indexed in the Web of Science from January 2008 to August 2018. Authorships were assigned by sex according to first name.

Main Outcomes and Measures  Outcomes included the proportion of female authorships, odds ratios of women holding prestigious first and last authorships compared with men (measured by the prestige index), citation rates, a productivity analysis, and cross-journal and transnational female representation within ophthalmic research. The hypothesis was formulated during data collection.

Results  Bibliometric data were abstracted from 87 640 original articles published in 248 ophthalmologic journals. Of 344 433 total authorships, female scholars held 34.9% (120 305 of 344 433) of all authorships, 37.1% (24 924 of 67 226) of first authorships, 36.7% (77 295 of 210 372) of coauthorships, and 27.1% (18 086 of 66 835) of last authorships. The female-to-male odds ratio was 1.12 (95% CI, 1.10-1.14) for first authorships, 1.20 (95% CI, 1.18-1.22) for coauthorships, and 0.63 (95% CI, 0.62-0.64) for last authorships, with annual growth rates of 1.6% overall, 1.6% for first authorships, 1.3% for coauthorships, and 2.5% for last authorships. Women were underrepresented in prestigious authorships (prestige index = −0.22). The underrepresentation remains almost stable for articles with many authors (prestige index = −0.17 for articles with >9 authors per article). Articles with female key authors were cited slightly less frequently (95% CI for female vs male authors, 10.8-11.0 vs 11.5-11.7 citations/articles). Women published fewer papers than men (42.5% [n = 41 383]; women held 34.9% [n = 120 207] of the authorships), show cross-journal uniformity and differences among single countries (change in prestige index = 0.66 vs 1.96). Overall, 44.1% of female authorships and a sex-neutral distribution of prestigious authorships are prognosticated for 2028.

Conclusions and Relevance  This algorithm analysis suggests the integration of women into ophthalmic research is average compared with other disciplines. A sex-specific gap exists for last authorship, suggesting career inequalities. With a growing number of female researchers in ophthalmology, the number of women in senior ranks may increase in the future.

In recent decades, academic medicine has undergone a huge transformation in terms of sex distribution. In 1966, only 6.9% of US graduates from medical school were women.¹ In 2016, the proportion increased enormously to 46.3%.¹ Improved educational opportunities for women, desire for financial independence, and high academic goals may provide an explanation for this development.² Data from 2018 show that women account for 21% of professors, 37% of associate professors, 46% of assistant professors, and 54% of instructors of ophthalmology at all US medical schools,³ emphasizing a female underrepresentation in senior ranks.

In this study, we focused on the distribution of authorships between the sexes in ophthalmic publications. In medical research, the first author usually “indicates the person whose work underlies the paper as a whole,”⁴ whereas the last author “indicates a person whose work or role made the study possible without necessarily doing the actual work”^4,5 and is held by a person in a senior position.^6,7 Consequently, the prestige of authorships follows a ranked order with a higher reputation of first and last authorships and a lower reputation of coauthorships.^5,6,8,9 Regarding original articles, young scientists usually publish as first authors or coauthors and senior researchers as last authors.^6,7 Therefore, the analysis of authorships allows conclusions to be drawn on the academic position of women. This axiomatic view is only valid for original research articles; by contrast, the order of authorships is often reversed in review articles.⁵

In this article, we identify female proportions of prestigious authorships, international differences, female representation in collaborative articles,¹⁰ and sex-specific citation rates. The study also provides an outlook for the coming decade.

Data analysis was conducted using Gendermetrics.NET,¹¹ a SQL server-based platform for analyzing bibliometric data. Original English research articles from the Web of Science category of ophthalmology published from 2008 onwards were acquired on August 13, 2018, from the Web of Science Core Collection (search term: WC = “Ophthalmology” AND language: English AND document types: Article). These authors were engendered by grouping authorships by name and first name,⁵ which results in a nonempty collection of authorships for each author entity.¹⁰ A general bibliometric overview is given in eFigure 1 in the Supplement. Institutional board review approval was waived because the study does not involve human participants.

To ensure the validity of the analysis, only countries and journals with a relative and absolute detection rate of at least 60%¹⁰ and 750 male/female authorships were included in the corresponding subanalysis. China, South Korea, and Taiwan were excluded from the country-specific analysis owing to a high rate of unisex names¹⁰ (eFigure 4 in the Supplement).

This study considered first authorships, coauthorships, and last authorships, with coauthorship including authorships between first and last authorship.¹⁰ The proportion of female authors (FAP) is a percentage defined as the ratio of the number of female authors and all authors.⁸ The female authorship odds ratio (FAOR) indicates the odds of a woman obtaining a specific authorship type compared with male authorship odds ratio and was applied for first authorships, coauthorships, and last authorships with corresponding 95% CIs.¹⁰ Female authorship odds ratio triplets represent a female odds ratio surplus to hold a particular authorship.¹⁰ For example, the FAOR triplet (−, =, +) indicates that female researchers have a lower odds ratio for first authorships, equal odds ratio for coauthorships, and higher odds ratio for last authorships compared with men.⁶

The prestige index (PI) measures the female odds of holding prestigious authorships compared with male odds¹⁰ and is characterized as the prestige-weighted average of the FAOR excess ε_t calculated over all authorship types t: ε_t = w_t (FAOR_t – 1), if FAOR_t ≥ 1; otherwise ε_t = w_t (1 – 1/FAOR_t) with the weighting factor w_t.¹⁰ First and last authorships are presumed to be prestigious and provide higher reputation than coauthorships.⁶ Therefore, coauthorships were weighted negatively (w_co = –1), whereas first and last authorships were weighted positively (w_first = w_last = 1)¹⁰ (a potentially alphabetical order of the authors list was excluded; eFigure 5 in the Supplement). For example, the definition of the weighting factor indicates a lowered PI by both higher female odds for coauthorships and lower female odds for first and last authorships, as well as an equal distribution of prestigious authorships between the 2 sexes by a value of 0.¹⁰

Average annual growth rates characterize annual growth rates¹⁰ of FAPs and the number of authorships to make a linear 10-year forecast of temporal development of FAP, FAOR, and PI.¹⁰ We used the Pearson correlation to determine linear associations between FAP, PI, and the 5-year impact factor of a journal. A Kruskal-Wallis test and a post hoc multicomparison test with a significance threshold of .05 and .01 were applied to test the null hypothesis concerning whether the nonnormally distributed sex-specific citation rates (eFigure 6 in the Supplement) stem from the same distribution.¹⁰

Overall, 446 662 authorships from 87 640 original articles published in 248 ophthalmic journals from 2008 to 2018 were analyzed using the Gendermetrics Platform.¹¹ The algorithmic sex determination (see Bendels et al¹¹ and eFigures 2 and 3 in the Supplement for details) yielded 55 950 male authors (41.4%), 41 383 female authors (30.6%), 11 102 unisex authors (8.2%), and 26 788 undefined authors (19.8%). Unisex and undefined authors and their authorships (n = 102 229) were excluded from this study. By applying the threshold criteria, 58 of 248 journals, representing 95.3% of the authorships (328 415 of 344 433) and 33 of 159 countries, representing 92.2% (317 497 of 344 433) of the authorships were included in the journal- and country-specific analysis, respectively.¹⁰

The analysis reveals an underrepresentation of female authorships at a global level with an FAP of 34.9% (120 305 of 344 433), relatively more first authorships (24 924 of 67 226 [37.1%]) and coauthorships (77 295 of 210 372 [36.7%]), and an essentially smaller amount of last authorships (18 086 of 66 835 [27.1%]). The corresponding FAORs are 1.12 (95% CI, 1.10-1.14) for first authorships, 1.20 (95% CI, 1.18-1.22) for coauthorships, and 0.63 (95% CI, 0.62-0.64) for last authorships (Figure 1A). The global FAOR pattern is characterized by the FAOR triplet (+, +, −). Thus, female scholars have higher odds of first authorships and coauthorships and lower odds of last authorships. The PI is on average −0.22, indicating worse odds of women holding prestigious authorships (Table 1 and Figure 1B).

The FAP exhibits a moderate increase (32.1% [7956 of 24 795] in 2008 and 37.6% [8274 of 22 012] in 2018; Figure 1B and C) with an average annual growth rate of 1.6% overall, an equal annual growth for first authorships and a disproportionally high value for last authorships (2.5%) and low value for coauthorships (1.3%). As proven by the PI (−0.23 in 2008 and −0.12 in 2018), this development has led to a more sex-neutral distribution of authorships odds during recent years.

In the assessment of women’s representation at the country level, there is a large FAP span from 19.3% (1012 of 5232) in Iran to 52.8% (886 of 1677) in Belgium (Table 2). Different FAOR patterns range from unfavorable FAOR triplets (=, +, −) such as in Turkey, Portugal, Spain, and France to favorable FAOR triplets (+, −, =) in Austria and Norway. A sex-neutral distribution of chances for all authorships is given in Sweden and Egypt (FAOR triplet [=, =, =]). Although most countries offer higher female chances of a first authorship, no country reveals higher odds of women holding prestigious last authorships. The highest PIs are given in Norway (PI = 0.44) and Austria (PI = 0.29), whereas Greece (PI = −0.73), Denmark (PI = −0.76), Italy (PI = −0.83), and Ireland (PI = −1.52) exhibit the lowest PIs. An almost sex-neutral chance of prestigious authorships is given in Sweden (PI = 0.03) and Finland (PI = −0.06). We found no correlation between the FAP of a country and its PI (r₃₁ = 0.14, P > .05).

Between individual journals, there is a wide FAP range from 17.1% (140 of 817) in Iranian Journal of Ophthalmology to 43.1% (2392 of 5544) in Journal of AAPOS (Table 3). The predominant FAOR pattern is described by the unfavorable FAOR triplet (=, +, −) (22 of 58 [37.9%]), followed by the triplet (+, +, −) (12 of 58 [20.7%]). In almost every journal, female researchers have lower odds of last authorships. By contrast, all observed journals except Journal of Refractive Surgery and Iranian Journal of Ophthalmology provide higher or at least the same odds for women holding first authorships and coauthorships. Overall, the journals are characterized by a negative PI of −0.52 to −0.02.

We found no correlation between (1) the FAP of the journal and its mean impact factor (r₅₆ = 0.02, P = .87) and (2) the PI and the mean impact factor (r₅₆ = 0.15, P = .25). There is a slight positive correlation between the FAP and the PI (r₅₆ = 0.32, P = .02), ie, the higher the PI of a journal, the higher the FAP (Table 1 and eFigure 7 in the Supplement).

Considering female representation in multiauthor articles, the FAP varies in a small range and grows with an increasing number of authors from 31.5% (16 165 of 51 265) (1 to 3 authors per article) to 37.1% (18 171 of 48 971) (>9 authors per article). The FAOR pattern is described as (+, +, −) and is independent of the number of authors per article (Figure 2). The PI fluctuates between −0.14 (1 to 3 authors per article) and −0.21 (7 to 9 authors per article). To summarize, neither the FAORs for first authorships, coauthorships, and last authorships nor the PI show a clear correlation with the number of authors per article (Table 1).

Articles with male key authors are slightly more frequently cited than those with female key authors (eFigure 8A in the Supplement). Specifically, articles with a male last (11.7 citations per article) or first (11.5 citations per article) author are above the mean citation rate of 11.0 citations per article, whereas articles with a female first or last author exhibit citation rates of 11.0 and 10.8 citations per article, respectively.

Considering combined key authorships, the analysis reveals that male first/male last and male first/female last authorship articles have on average the highest citation rates with 12.0 and 11.8 citations per article, respectively, followed by female first/male last authorship articles (11.2 citations per article) and female first/female last authorship articles (10.4 citations per article). Single-authored articles have by far the lowest citation rates, with 7.7 citations per article for male single authors and 7.3 citations per article for female single articles (eFigure 8A in the Supplement).

Statistically, the citation rate of an article increases the more authors who are involved. Specifically, articles with 1 to 3 authors have a mean (SD) citation rate of 8.8 (18.9), while articles with more than 9 authors have a mean (SD) citation rate of 17.3 (32.8) (eFigure 8B in the Supplement). The sex-specific differences in citation rates impose at each article’s author count level (Table 1).

Regarding scientific productivity, we reveal marked differences between the 2 sexes: while women dominate the subgroups author of 1 article and author of 2 articles, all other subgroups reveal a relative underrepresentation of female authors. Especially the subgroup of most productive authors (author of more than 9 articles) is clearly dominated by male authors (8.8% [4922 of 55 950] of the male authors vs 5.2% [2137 of 41 383] of the female authors). This result underlines the higher productivity of male authors, as the 57.5% (55 950 of 97 333) male authors hold 65.1% (224 128 of 344 433) of all authorships (eFigure 8C in the Supplement).

In comparison with our previous studies,^8,10,12-15 ophthalmic research depicts an average quantitative integration of women with an FAP of 34.9% (Table 1; eFigures 10-19 in the Supplement). Analyzing the authorships, we found a discrepancy in terms of higher odds of women holding coauthorships and first authorships and a lower proportion of last authorships (FAOR triplet [+, +, −]), suggesting an imbalance of leadership positions and a sex-specific career dichotomy. Given that publishing in an academic leadership is a central point within the scientific career system,⁸ this disadvantage may be an obstacle to career advancement for female scholars.¹⁶ Franco-Cardenas et al¹⁷ discovered that despite the growing FAP for first and last authorships in 3 major ophthalmology journals between 2000 and 2010, editorial authorships continued to be overwhelmingly male dominated (87% in 2000 and 90% in 2010). Within that time, none of these journals ever had a female editor in chief.^17,18

Like other surgical subjects, ophthalmology has historically been male dominated.^19,20 The pipeline problem²¹ describes the phenomenon that despite the increasing number of women entering medicine during the past decade,²² there is no corresponding increase in women in leadership positions. The reasons are multifactorial.¹² According to a report from 2018, 47% of female and 67% of male US ophthalmologists aim for promotion.²³ It may last in sex-specific career affections, as women dedicate more of their professional time to teaching and patient care rather than research.²⁴ Moreover, a lack of mentors in upper positions,^22,24 raising children,¹² lack of support for childcare, unequal pay,^25,26 social constraints, inadequate promotions compared with their male colleagues,²¹ or prejudices against female executives^24,27 may all cause this sex-specific gap. A linear estimate based on our data forecasts a FAP of 44.1% and sex-neutral distribution of prestigious authorships (PI = 0) for 2028 (eFigure 9 in the Supplement).

We observed a slight increase of female representation (FAP) within collaboration articles when more authors were involved. Female representation in prestigious authorships also remains stable for articles with a high number of authors (eg, collaboration articles), which attract the highest citation rates (eFigure 8B in the Supplement). This may be a sign for an improving female representation in ophthalmic research.

In accordance with other scientific fields,^5,8,10,12,28 female authors as a group were on average less productive than the group of their male colleagues. Overall, 42.5% of female authors were responsible for 34.9% of all authorships. Furthermore, men show a higher productivity at the level of single authors (eFigure 8C in the Supplement). In a 2011 study, Reed et al²⁹ posited that women’s publication rates are initially lower, given that female researchers are more likely to work part time at the early stages of their career to assume a greater share of family responsibilities.^30-32 To evaluate the association of family responsibilities on academic productivity, Carr et al³⁰ discovered that women with children, on average, publish less than men with children. However, in the later part of their careers, the publication rates of female researchers increase and exceed those of men.^30-32 This result can also be transferred to ophthalmology.³³ The delayed career uplift of female researchers, on average, may account for both productivity and opportunities for leaderships.

When considering regional aspects, we focused on the differences between the FAP and the FAOR among single countries. Norway, Austria, and Sweden offer the best conditions for women, whereas Denmark, Italy, and Ireland are characterized by an advantage for men. This result partially correlates with the 2017 Global Gender Gap Report benchmarking 144 countries,³⁴ where Scandinavian countries are among the upper ranks (Norway ranked second and Sweden, fifth). However, it does not explain the poor results in Denmark, Italy, and Ireland, as they all lie above the global weighted average.

We did not find a correlation between the FAP and the countries’ PI, as also reported in our previous studies.^8,10,12,15 Accordingly, countries with a low FAP can still offer good chances for an advantageous authorship. For example, Norway provides the best PI (0.44) with only 37.7% of female authorships, whereas Poland has rather poor chances (PI = −0.57) with an FAP of 51.6%.

The PI does not depend on the number of female researchers but rather their academic position. Therefore, this result may be based on the differences in sociocultural and socioeconomic backgrounds between the countries.

We revealed a correlation between the journal’s FAP and its PI (r = 0.32), meaning that the more women who are included in ophthalmic research, the better their chances to hold prestigious authorships, suggesting an advanced change in ophthalmic research.

Moreover, it is in line with the theory of critical mass, postulating increasing female career opportunities and influence as soon as a threshold (30% to 35%) of female proportion is exceeded.³⁵ On the level of journal subject categories, the analysis confirms previous results regarding female participation^13,36 with highest percentages for the subfield pediatrics (42.1%) and lowest percentages for the subfield surgery (28.2%; eTable in the Supplement). We disclosed a homogeneity of sex-specific authorships odds, because 37 of 58 journals exhibit higher female odds for first and lower female odds for last authorships. The worldwide sex-specific hierarchy of research groups may be assigned to the corresponding journals, which may explain the small value range of the journal’s PI compared with that among the countries (change in PI: 0.66 vs 1.96).

Comparing different subjects, we noticed a wide FAP range from 29.8% in Nature Journal Index to 75.2% in Nursing (Table 1), while Otorhinolaryngology, as another small surgical subject, showed an FAP of 31.1%. In conclusion, the integration of women into ophthalmic research is average, with an FAP of 34.9% and stable female representation in prestigious authorships. However, all subjects suggest a considerable career dichotomy, supporting our assumption with many female novice authors in research and few women in senior positions (predominant FAOR triplet [+, +, −]).

As already discussed by Bendels et al,¹⁰ our analysis is limited to more recently published articles, owing to the use of first name initials before 2006 with a corresponding lack of sex identification¹¹ and the lack of information about academic positions of researchers, their academic degree, age, or employment status and their participation on editorial boards.^37,38 Limitations of the algorithm include that China, South Korea, and Taiwan were excluded from the country-specific analysis.¹⁰ Methodically, it should be mentioned that older articles (published from 2008 to 2010) have a stronger influence on citation rates compared with recent ones (cited half-life), reflecting the situation from the early phase of the study.^5,39 Moreover, all last authorships were presumed to be prestigious, even in articles with only a few authorships. The value of each article was not taken into account.

The results suggest an average integration of female scholars into ophthalmic research compared with other subdisciplines. Despite the average FAP and almost sex-neutral citation rates, women are unable to offset for the negative global and cross-journal PI. None of the observed countries or journals provide better chances for women to hold a last authorship, suggesting clear career inequalities. However, because there are plenty of female scholars at the beginning of their career,¹ female academic leadership may increase in the coming years, as confirmed by the high average annual growth rates for last authorships (2.5%). Nevertheless, a number of recently published studies report the persistence in female career dichotomy with a stagnant proportion of female executives.^38,40-42 Data in the coming years may clarify whether female career inequalities will cease and whether sex equitable opportunities in ophthalmic research can be achieved.

Corresponding Author: Michael H. K. Bendels, MD, PhD, Division of Computational Medicine, Institute of Occupational Medicine, Social Medicine and Environmental Medicine, Goethe-University, Theodor-Stern-Kai 7/Haus 9b, 60596 Frankfurt am Main, Germany (bendels@med.uni-frankfurt.de).

Accepted for Publication: June 16, 2019.

Published Online: August 15, 2019. doi:10.1001/jamaophthalmol.2019.3095

Correction: This article was corrected on November 27, 2019, to fix an error in Figure 2.

Author Contributions: Mr Kramer and Dr Bendels 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.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: Kramer, Groneberg, Bendels.

Drafting of the manuscript: Kramer, Groneberg, Bendels.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Kramer, Bendels.

Administrative, technical, or material support: Kohnen, Groneberg.

Supervision: Groneberg, Bendels.

Conflict of Interest Disclosures: Dr Kohnen reports consulting and research for Abbott/Johnson & Johnson, Alcon/Novartis, Avedro, Oculentis, Oculus, Presbia, Schwind, and Zeiss; consulting for Allergan, Bausch + Lomb, Dompé, Geuder, Merck, Rayner, Santen, Staar Surgical, Thea Pharmaceuticals, TearLab, Thieme Medical Publishers, Uni-Med Verlag, Med Update, and Ziemer; and research for Hoya Surgical Optics. No other disclosures were reported.