key: cord-1023756-b7dlqdbf
authors: Cevik, Muge; Haque, Syed Arefinul; Manne-Goehler, Jennifer; Kuppalli, Krutika; Sax, Paul E.; Majumder, Maimuna S.; Orkin, Chloe
title: Gender disparities in COVID-19 clinical trial leadership
date: 2021-01-05
journal: Clin Microbiol Infect
DOI: 10.1016/j.cmi.2020.12.025
sha: f61d803e50a391bb76bbfa262ecb36fefb71fda8
doc_id: 1023756
cord_uid: b7dlqdbf

OBJECTIVES: We aimed to compare the gender distribution of clinical trial leadership in COVID-19 clinical trials. METHODS: We searched https://clinicaltrials.gov/and retrieved all clinical trials on COVID-19 from January 1, 2020 to June 26, 2020. As a comparator group, we have chosen two fields that are not related to emerging infections and infectious diseases: and considered not directly affected by the pandemic: breast cancer and type 2 diabetes mellitus (T2DM) and included studies within the aforementioned study period as well as those registered in the preceding year (pre-study period: January 1, 2019 and December 31, 2019). Gender of the investigator was predicted using the genderize.io API (application programming interface). The repository of the datasets used to collect and analyse the data available at https://osf.io/k2r57/. RESULTS: Only 27.8% (430/1548) of principal investigators (PIs) among COVID-19-related studies were women, which is significantly different compared to 54.9% (156/284) and 42.1% (56/133) for breast cancer (p<0.005) and T2DM (p<0.005) trials over the same period, respectively. During this “pre-study” period, the proportion of PIs who were predicted to be women were 49.7% (245/493) and 44.4% (148/333) for breast cancer and T2DM trials, respectively and the difference was not statistically significant when compared to results from the study period (p>0.05). CONCLUSION: We demonstrate that less than one-third of COVID-19-related clinical trials are led by women PIs, half the proportion observed in non-COVID-19 trials over the same period which remained similar to the pre-study period. These gender disparities during the pandemic may indicate not only a lack of women's leadership in international clinical trials and involvement in new projects but also may reveal imbalances in women's access to research activities and funding during health emergencies.

women occupied fewer leadership positions, led a fewer funded studies, and applied for and received 61 less grant funding than men when they did apply [4] [5] [6] [7] . The employment gap that occurs when women 62 take parental leave impacts the rate of academic advancement and in turn the receipt of institutional 63 support to apply for and secure funding [6, 7] . These imbalances contribute to systemic inequalities that 64 hamper women's access to and progress in science [2, 7, 8] . A review of the gender distribution of 24 65 COVID-19 national task forces suggests that many committees are comprised of less than a quarter 66 women, indicating that women's voices and expertise have been excluded from decision making during 67 this unprecedented public health emergency [9].

For example, emerging data suggest that across all disciplines, despite an increased number of peer- The COVID-19 pandemic offers numerous opportunities in clinical research. These include trials to assess 78 the safety and efficacy of medical interventions, with protocols in various stages of implementation.

Here, we compare the gender distribution of clinical trial leadership in COVID-19 clinical trials.

We systematically searched https://clinicaltrials.gov/ and retrieved all clinical trials on COVID-19 83 registered from January 1, 2020 to June 26, 2020 using "COVID" as a keyword. As a comparator group,

we have chosen two fields that are not related to emerging infections and infectious diseases, and 85 considered not directly affected by the pandemic: breast cancer and type 2 diabetes mellitus (T2DM).

We retrieved all clinical trials related to these comparator conditions registered at 87 https://clinicaltrials.gov/ within the aforementioned study period as well as those registered in the 88 preceding year (pre-study period: January 1, 2019 and December 31, 2019). We retrieved the names of 89 investigators listed; study director, principal investigator (PI) (the person who is responsible for the 90 scientific and technical direction of the entire clinical study) and study chair (whose role involve toxicity 91 and accrual monitoring). Gender of the investigator was predicted using the genderize.io API 92 (application programming interface). This tool has been used to predict the gender of first names in 93 studies regarding gender bias [12, 13] and achieves a minimum accuracy of 82%, with an F1 score 94 (weighted average of precision and recall) of 90% for women and 86% for men [14] . Clinical trials were 95 excluded if i) investigator information was not provided; ii) the genderize.io API could not predict any of 96 the investigators' gender from their first name; or iii) organization or company names were provided as 97 the investigator. The number of studies that were excluded for the above reasons are reported in the 98 supplementary flow diagram. An exploratory temporal analysis was conducted with the available data.

Categorical variables were summarized by frequencies and percentages. We compared groups using Chi-100 square testing for equality of proportions with continuity correction [15] . The analysis was performed using R (Version 4.0.2). The repository of the datasets used to collect and analyse the data available at 102 https://osf.io/k2r57/.

We identified 2 345 COVID-19-related clinical trials. Of those, 1 448 had at least one investigator listed 106 (i.e., principal investigator, study director, or study chair) whose gender could be predicted. In the 107 comparator group, we identified 449 trials on breast cancer and 272 on T2DM that were registered. Of 108 those, 274 breast cancer studies and 139 T2DM studies had at least one investigator whose gender 109 could be predicted.

Overall 27.8% (430/1548) of PIs among COVID-19-related studies were predicted to be women, which is 112 significantly different compared to 54.9% (156/284) and 42.1% (56/133) for breast cancer (p<0.005) and 113 T2DM (p<0.005) trials over the same period, respectively (Table 1) . While there has been a small 114 increase in the proportion of PIs who were predicted to be women in May 2020, clinical research 115 leadership for COVID-19 among this group was below 25% for the remainder of the study period 116 (Supplementary Material). While 31.4% (76/242) of study chairs were predicted to be women in COVID-117 19-related studies, 32.1% (9/28) (p=0.7) and 63.6% (7/11) (p<0.01) were predicted to be women in 118 breast cancer and T2DM trials, respectively. Proportion of study chairs were not significantly different 119 across the three fields.

We also reviewed comparator group studies registered before January 1, 2020 to determine whether 122 the pandemic might have affected gender distribution of trial leadership. We identified 839 clinical trials 123 related to breast cancer and 533 on T2DM over a 12-month period prior to January 1, 2020. Of those, 124 573 breast cancer studies and 359 T2DM studies yielded at least one investigator whose gender could be predicted. During this "pre-study" period, the proportion of PIs who were predicted to be women 126 were 49.7% (245/493) and 44.4% (148/333) for breast cancer and T2DM trials, respectively and the 127 difference was not statistically significant when compared to results from the study period (p>0.05).

In this study, we demonstrate that less than one-third of COVID-19-related clinical trials are led by As a community, we must recognise that there is a tendency to "turn to men" in times of crisis both for 151 leadership and scientific expertise [2, 3, 16, 17] , highlighting the need to challenge this culture. Research 

The evidence while sparse indicates that teams that are diverse in terms of gender, ethnicity, and social 162 background produce better health science, are more highly cited, generate a broader range of ideas and 163 innovations, and better represent society [2, 16, 18, 19] . Not only can these women drive discovery and 164 innovation, but they can act to address health disparities and provide role models for the next 165 generation of women scientists [2, 16, 18, 19] . Ensuring gender representation would also reflect the 166 commitment of the global community to promoting gender equality in academic medicine and research: 167 inclusion, diversity, representation, progression, and success for all. Therefore, the disadvantage not 168 only affect women themselves and their research career but has much more profound implications for 169 the wider society especially given the disproportionate burden of such outbreaks for communities who 170 are marginalized due to their gender, sexuality, class, ethnicity, and ability [20] [21] [22] . Our analysis has some limitations. We could include only ~50-75% of trials for which an investigator's 173 gender could be algorithmically predicted because the majority of studies had no investigator 174 information, or the investigator names were not distinguishable (supplementary material). Furthermore, 175 while such algorithms allow for the rapid analysis of gender disparities such as those conducted here, 176 they can also be exclusionary to gender non-conforming, non-binary, and trans individuals. Beyond 177 these limitations, although there were several observational studies in our dataset, clinicaltrials.gov may 178 be biased towards randomised control trial registration and women may be more likely to be involved in 179 observational studies, which still demonstrates gender disparities in types of trials women lead. Also, we 180 did not consider studies that received private funding, which may not have been registered on 181 clinicaltrials.gov; however, it is worth noting that clinicaltrials.gov is an international database with 182 widespread international representation. Finally, while we attempted to provide a comparison with two 183 other fields, a potential for bias could arise from the difference of gender distributions of researchers 184 working in the fields of infectious diseases, breast cancer and diabetes. 

MC: conceptualisation, methodology, investigation, literature review, data curation, writing -original 197 draft. SH and MM: investigation, data curation, formal analysis, writing -review and editing; JS, KK, PS: 198 methodology, writing -review and editing, supervision. CO: conceptualisation, methodology, 199 investigation, literature review

She has also received research grants to her institution from the 207 above-mentioned companies. PES has received honoraria, fees for lectures

KK has received personal fees from GSK, outside the submitted work

Parental involvement in home schooling and 244 developmental play during lockdown -Initial findings from the COVID-19 Survey in Five 245 National Longitudinal Studies

Women Physicians and the COVID-19 Pandemic

Sex discrepancies in infectious 252 disease research funding 1997&#x2013;2010: a systematic analysis

Increasing 255 women's leadership in science in Ho Chi Minh City. The Lancet

Who Applies for Research Funding

Sex Differences in Achievement 261 and Faculty Rank in Academic Infectious Diseases

Covid-19 response: a call for more inclusive and transparent decision-making

COVID-19 medical papers 267 have fewer women first authors than expected

Achieving women's equity in 269 academic medicine: challenging the standards. The Lancet

The gender gap in science: How long until women 272 are equally represented?

Historical comparison of gender inequality in 276 scientific careers across countries and disciplines

Inferring gender from names on 280 the web: a comparative evaluation of gender detection meth-ods

Interval estimation for the difference between independent proportions: 283 comparison of eleven methods

19. A call for more inclusive Covid-19 research: Tackling disparities during a pandemic

low-income-and-communities-of-color-at-higher-risk-of-serious-illness-if-infected-with-300 coronavirus/. 301 21. Finch WH, Hernández Finch ME. Poverty and Covid-19: Rates of Incidence and Deaths 302 in the United States During the First 10 Weeks of the Pandemic

Revealing the unequal burden of COVID-19 by income, 305 race/ethnicity, and household crowding: US county vs. ZIP code analyses: Harvard Center for 306 Population and Development Studies