The faces of endometriosis research (3/3): Gender

Endometriosis research: a fading male bastion

Historically, scientific research had been heavily male-dominated. Even today, gender inequalities in R&D remain (e.g., due to access to studies, care responsibilities, gender preferences…) and vary substantially across countries¹. However, the picture has also significantly improved in the past century, resulting in an ever-greater share of women among scientists. Endometriosis research is no exception. Indeed, the following graph summarizes the evolution of contribution to endometriosis R&D by gender (i.e., the number of author-publication pairs by gender).

Graphs may not be fully visible on cell phones. Using a computer is recommended for the best reading experience. The titles can be clicked to get a fullscreen version of the graphs.

Gradual feminization of endometriosis research

Note: Graph based on 30,565 publications involving 157,415 author-publication pairs, of which 108,500 can be attributed the male/female gender with a probability above 74%. Probabilities are derived from the Namespedia database. This graph shows the contribution or level of effort by gender, which is different from the number of authors of each gender. Indeed, on this graph, a single author can be counted multiple times if he or she contributes to more than one publication. The observed pattern holds with a sensitivity analysis taking a probability threshold of 90%.

Endometriosis research was thus almost exclusively male in the early 20^th century. First extensive female contributions can be identified in the mid-20^th century (about 5-10%), but the real take-off has occurred in the 1980s. From this period, the female contribution to endometriosis R&D has grown substantially. By 2020, it has reached almost half of the scientific output. The feminization of this particular stream of research has thus been gradual but also substantial and likely higher than in R&D in general. Parity is indeed at our doorsteps.

Many reasons can explain this pattern, including the strong feminization of science and medicine in many countries (notably the developed ones). Gender differences in interests/priorities could also explain this situation (e.g., the high share of women among gynecologists, which are likely to be strongly involved in endometriosis R&D).

However, this broad picture should be complemented by a more refined analysis of gender across institutional and national boundaries.

The various types of institutions involved in endometriosis R&D indeed exhibit different patterns of gender contribution, as shown in the following graph.

Gender contribution by type of institution

Note: Graph based on 29,532 publications with an attributable location/type of institution, involving 118,723 author-publication pairs, of which 88,889 can be attributed the male/female gender with a probability above 74%. Probabilities are derived from the Namespedia database. This graph shows the contribution or level of effort by gender, which is different from the number of authors of each gender. Indeed, on this graph, a single author can be counted multiple times if he or she contributes to more than one publication. These contributions are grouped by type of institution (excluding types with a low number of publications / that are difficult to interpret). The black line represents parity, i.e., an equal contribution for each gender to the produced research. Sensitivity analyses with a threshold of 90% have confirmed that the results are robust.

It is clear that the contribution of females to endometriosis R&D is higher in governmental institutions (45.86%) and nonprofits (47.99%) as compared to the average (36.9%). By contrast, the male contribution is much higher in companies (66.34%) and healthcare (64.79%). These patterns can notably be explained by the specialization differences between genders (e.g., surgeons are more likely to be male and thus involved in R&D performed in healthcare organizations) but also in priorities and forms of engagement (e.g., female overrepresentation in NGO-related publications).

However, differences between countries are much more important than between institutions. The range of gender differences (over the 1925-2020 period) is represented on the following map.

Gender contribution to endometriosis research by country (1925-2020)

Note: Map based on 29,531 publications with an attributable location at the national level, involving 118,720 author-publication pairs, of which 88,887 can be attributed the male/female gender with a probability above 74%. Probabilities are derived from the Namespedia database. This map shows the contribution or level of effort by gender, which is different from the number of authors of each gender. Indeed, on this map, a single author can be counted multiple times if he or she contributes to more than one publication. These contributions are grouped by country. The method for gender attribution by using first name hits some limits for Asia (e.g., China, Japan) and should thus be considered with caution for this part of the world. Similarly, some countries have a very low number of publications, impacting results (e.g., most African nations). The observed patterns hold with a sensitivity analysis taking a probability threshold of 90%.

Overall, most countries feature a male bias in contributing to endometriosis publications. For instance, male contribution reaches about 64% for the United States, which dominates the scientific landscape for the disease. However, this bias and its extent are subject to considerable variations. Some countries reach much higher levels of male contribution, but it is usually linked to a low number of publications, making interpretation more difficult. By contrast, some countries are almost gender-equal regarding endometriosis R&D, such as Italy, Sweden, Iran and India. There are even nations with a solid female-leaning contribution, including many countries from the former Socialist bloc. The important role of females in R&D is a notable feature of (former) Communist countries in general (i.e., beyond endometriosis). This finding thus tends to bear additional credence to the analysis of this article. Attribution of gender based on first names is sometimes considered to be less reliable for East Asian countries², so the strong male domination observed in, e.g., China or Japan should be nuanced. However, this is likely a reality, given that medicine is still a male bastion in these states (e.g., in 2015, 20.3% of medical doctors were male in Japan, 22.3% in Korea³).

The same patterns globally hold when restricting the analysis to the most recent endometriosis publications (i.e., those released in the 2010s). However, the shift towards a more significant feminization of research is clearly visible, as shown on the map below.

Gender contribution to recent endometriosis research by country (2010s)

Note: Map based on 13,606 publications from the 2010s with an attributable location at the national level, involving 65,625 author-publication pairs, of which 49,418 can be attributed the male/female gender with a probability above 74%. Probabilities are derived from the Namespedia database. This map shows the contribution or level of effort by gender, which is different from the number of authors of each gender. Indeed, on this map, a single author can be counted multiple times if he or she contributes to more than one publication. These contributions are grouped by country. The method for gender attribution by using first name hits some limits for Asia (e.g., China, Japan) and should thus be considered with caution for this part of the world. Similarly, some countries have a very low number of publications, impacting results (e.g., most African nations). The observed pattern holds with a sensitivity analysis taking a probability threshold of 90%.

For instance, the gender contribution to endometriosis R&D in the United States is much more balanced for the 2010s than for the entire period, with about 54% male / 46% female. Many developed countries exhibit a similar trend, as well as some developing ones, such as Brazil. Female overrepresentation is still present in some key states, such as those of the post-Soviet space.

In a nutshell, endometriosis R&D has historically been a male bastion, but over the last century, this situation has evolved. In particular, since the 1980s, the male share of contributions has eroded rapidly, resulting in a much more equal distribution of effort. Even if there are still significant inequalities across countries, parity is no longer a long shot in this field.

Endometriosis research and gender at the national level: the case of France and the USA

Zooming on a few countries can help us to understand the interactions between gender and endometriosis research. Indeed, as demonstrated by the previous maps, there are significant national differences in the gender balance among endometriosis researchers, e.g., with much more crucial female contribution in former Socialist countries. It is the case even between comparable, developed countries. Contextualizing these patterns by considering the broader national landscapes regarding gender can thus be relevant. Are these situations for endometriosis R&D the result of the feminization of medicine or research overall?

For this article, I chose to take a closer look at two countries that I know well and for which high-quality contextual data regarding gender is readily available. Namely, we will review the situation in France and the United States of America.

In France, only a minority of endometriosis research is carried out by women (34.38% overall, 37.37% when restricted to publications from the 2010s). This low share is surprising given the relatively high feminization of professions that feed the talent pool for endometriosis R&D.

A relatively low feminization of endometriosis R&D despite a favorable context in France (2010s)

Note: Graph based on multiple sources. Own elaboration using Dimensions and GRID data for endometriosis research contribution. Endometriosis research contribution is based on 1,558 publications with an attributable location matched to France, involving 6,576 author-publication pairs, of which 5,104 can be attributed the male/female gender with a probability above 74%. Probabilities are derived from the Namespedia database. This graph shows the contribution or level of effort by gender, which is different from the number of authors of each gender. Indeed, on this graph, a single author can be counted multiple times if he or she contributes to more than one publication. Recent endometriosis research contribution is similar but restricted to France-based publications from the 2010s (791 publications involving 3,872 author-publication pairs, of which 3,155 with an attributable gender). The observed pattern holds with a sensitivity analysis taking a probability threshold of 90%. French population is derived from INED data (average for all years of the 2010s)⁴. Scientists and Engineers information stems from Eurostat data (average for all years of the 2010s)⁵. Data for Medical Doctors (MDs) gynecologists (OB/GYNs), young Medical Doctors (below 35), young gynecologists (below 35) and midwives are all taken from DREES⁶ (average for years 2012-2019). The black vertical line shows gender parity.

Indeed, females are already a majority among French gynecologists, parity is almost reached when considering all medical doctors, and females also represent a significant share of all scientists and engineers. This feminization is even higher when considering only young medical doctors or young gynecologists. Similarly, French midwives are overwhelmingly women (even if they are traditionally less involved in R&D than, e.g., Medical Doctors).

So, French endometriosis R&D is much less feminized than we may have expected, given the gender breakdown of the relevant professions. Even in dynamic terms, recent French endometriosis research is not experiencing substantial growth in the share of female contribution, which contrasts with medicine’s fast feminization. The reasons are not precisely clear. Some time lag between access to the relevant professions and involvement in R&D may be present (but unlikely to explain the entirety of the situation). Other barriers are likely involved. Anyway, the French case suggests that even a (relatively) favorable context regarding gender in the professions of potential contributors does not guarantee a gender-balanced endometriosis R&D.

This French paradox can be contrasted with what is going on in the United States, as shown in the following graph.

A higher feminization of endometriosis R&D than what the context was likely to yield in the USA (2010s)

Note: Graph based on multiple sources. Own elaboration using Dimensions and GRID data for endometriosis research contribution. Endometriosis research contribution is based on 8,321 publications with an attributable location matched to the United States of America, involving 24,714 author-publication pairs, of which 20,340 can be attributed the male/female gender with a probability above 74%. Probabilities are derived from the Namespedia database. This graph shows the contribution or level of effort by gender, which is different from the number of authors of each gender. Indeed, on this graph, a single author can be counted multiple times if he or she contributes to more than one publication. Recent endometriosis research contribution is similar but restricted to US-based publications from the 2010s (3,208 publications involving 11,015 author-publication pairs, of which 8,885 with an attributable gender). The observed patterns hold with a sensitivity analysis taking a probability threshold of 90%. US population is derived from Census data (average for all years of the 2010s)⁷. Scientists and Engineers information stems from NSF data of 2015⁸. Data for Medical Doctors (MDs), gynecologists (OB/GYNs), Residents and Resident gynecologists are all taken from the AAMC⁹ (average for the 2010s with data from 2010, 2013, 2015, 2017 and 2019). Data for midwives is based on the report of the AMCB for 2019¹⁰. The black vertical line shows gender parity.

Indeed, medicine is much less feminized in the USA than in France. Males still account for a comfortable majority of physicians and are slightly more numerous than females, even among residents. Compared with France, science and engineering remain a male bastion. However, females are a majority among gynecologists, and women account for an even higher share of resident gynecologists. In this context that is likely to be less favorable to gender parity in R&D, the US reaches a higher percentage of female contribution in endometriosis R&D (36.04% overall, 45.83% for recent research only). Even better, the recent research shows a jump in the female contribution compared to the total historical output. Which factors may explain this relatively higher female contribution to endometriosis R&D in the US even when the pipeline of potential contributors seems less favorable than in France? Multiple factors may be at play (e.g., differences in access, interests, working conditions, etc.), and this comparison should serve as a cautionary tale against simple deductions. Anyway, even for these relatively advanced countries, there is still some important progress to be made regarding gender balance in endometriosis R&D, hoping that it would have positive repercussions on outcomes.

An impact of gender on endometriosis research?

Can the male overrepresentation in endometriosis scientific publications impact the produced knowledge? This possibility is far from being a preposterous idea. Indeed, gender can have various types of influence on R&D in general: prioritization of specific research interests to the detriment of others, gendered choices of hypotheses, of applied methods, of ways to categorize data, of assessment of the evidence (e.g., with sexist biases?)¹¹… To mention a notable example, according to Llyod, the biological analysis of the female orgasm has been plagued by unquestioned assumptions about the male understanding of orgasm, leading to a failure to review the evidence on this phenomenon rigorously¹².

Can we suspect something similar with endometriosis research? As gender may influence the research through various channels, it will not be possible to obtain a definitive answer. However, the available dataset can shed light on one dimension: are there differences in the topics covered by male or female scientists active in endometriosis research?

Indeed, we do have the gender data of authors contributing to the majority of publications (30,565 publications have at least one author with an attributable gender out of the total sample of 36,929 publications), as well as an identification of their topics (25,256 publications have both data on topics and the gender of at least one author). In particular, we can study how the gender balance within a team is related to the analyzed topics. This linkage allows us to see whether all-male, all-female or gender-balanced teams tend to have different research interests. Of course, only correlations will be produced, as it is impossible to control other factors interacting with both gender and topic choices (e.g., differences in scientific specializations, countries, periods…). But still, this would paint an interesting broad picture.

The following graphs highlight how the gender composition of the publications’ teams is connected to the focus on each broad topic.

Diversity of patterns linking the gender composition of teams and their focus on each broad topic

Note: Graphs based on 28,521 publications with attributable topics, of which 25,526 have data on the gender composition of their teams (gender of authors determined with a threshold of 74%). Probabilities for the first names’ gender are derived from the Namespedia database. Each publication is characterized by the gender composition of its team (excluding authors of undetermined gender) and by the shares of the different broad topics in its abstract. All the publications are put on a scatterplot. A loess regression is then applied to systematically study the linkages between the gender composition of the teams and the share of abstracts dedicated to each broad topic. As a result, gender-balanced teams are at the middle of each graph, all-male teams at the very right, and all-female ones at the very left. Sensitivity tests with a threshold for gender determination at 90% were carried out and showed robust results.

Different patterns between team composition and focus on broad topics can be observed at the publication level:

Monotonic relationship: the higher the share of one gender in the publications’ teams, the higher (or lower) is the focus on the relevant broad topics. It includes “anatomy and classification” and “treatment”, for which higher shares of male authors are associated with higher scientific prioritization. Similarly, higher percentages of female authors yield a higher focus on “symptoms and daily life” and “health and economic burden”.
U-shape relationship: all-male or all-female teams tend to focus the most on these broad topics compared to more gender-balanced teams. For instance, it is the case for the “connections to other diseases” and “miscellaneous” broad topics.
Inverted U-shape relationship: gender-balanced teams have the highest level of prioritization for these broad topics, as opposed to gender-biased teams, which typically contribute less to them. This is the case for “causes and mechanisms” and to a lower extent for “diagnosis and analyses” (the latter being also male-biased).

Differences in prioritization depending on the gender composition of the teams can thus be observed. However, the causal relationship is not demonstrated, and other factors may explain this situation. Moreover, the magnitude of the differences is typically small (at most a few percent, usually smaller). Overall, there is no major connection between the gender composition of teams and the prioritization of specific broad topics for endometriosis research. For detailed topics, the situation is largely similar. However, we can also see some overrepresentation of female-oriented teams in, e.g., “environmental, lifestyle and other risk factors”, “pain” and “daily life and experience”, and of male-oriented teams in, e.g., “detailed description of specific subtypes” or “specific surgical techniques”. But the same finding holds true: differences are very small in magnitude. This result does not mean that gender has no impact on endometriosis research activities. For instance, qualitative effects can be assumed regarding the formulated hypotheses, the type of evidence considered, or potential biases. For example, old publications written by white male gynecologists on endometriosis in black women have probably (read “definitely”) been plagued by sexist/racist assumptions, especially on the fact that the disease was less common in blacks¹³.

Moreover, this says nothing about the impact of gender on healthcare (rather than R&D). So much remains to be explored on that front as well.

Notes

1
See for instance https://www.pnas.org/content/117/9/4609
2
As documented in previous studies such as https://www.pnas.org/content/117/9/4609
3
https://www.oecd.org/gender/data/women-make-up-most-of-the-health-sector-workers-but-they-are-under-represented-in-high-skilled-jobs.htm
4
https://www.ined.fr/fr/tout-savoir-population/chiffres/france/structure-population/population-sexe-ages/
5
https://ec.europa.eu/eurostat/databrowser/view/HRST_ST_RSEX__custom_1424970/default/table?lang=en
6
https://drees.shinyapps.io/demographie-ps/
7
https://www.census.gov/topics/population/age-and-sex/data/tables.All.List_897222059.html
8
https://www.nsf.gov/statistics/2018/nsb20181/report/sections/science-and-engineering-labor-force/women-and-minorities-in-the-s-e-workforce
9
https://www.aamc.org/data-reports/workforce/data/2018-physician-specialty-data-report-executive-summary
10
https://www.amcbmidwife.org/docs/default-source/reports/demographic-report-201943ccaa1339164404a37c6993684dcd64.pdf?sfvrsn=9dd255e5_2
11
See https://philosophiedessciences.blogspot.com/2020/06/la-science-est-elle-objective-2.html for a short analysis of the social context (with a specific focus on feminism) and its linkages with the objectivity of science (in French)
12
https://plato.stanford.edu/entries/feminist-science/
13
See for instance https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1609884/

Endometriosis research: a fading male bastion

Endometriosis research and gender at the national level: the case of France and the USA

An impact of gender on endometriosis research?

Notes

Leave a Reply Cancel reply

Related Posts