Outcomes of COVID-19 manuscripts submitted to the Canadian Journal of Anesthesia: a retrospective audit of author gender and person of colour status

Abstract

Purpose

We aimed to evaluate the representation of women and persons of colour (POC) authors of COVID-19 manuscripts submitted to, accepted in, and rejected from the Journal and to evaluate trends in their representation during the pandemic.

Methods

All COVID-19 manuscripts submitted to the Journal between 1 February 2020 and 30 April 2021 were included. Manuscript data were retrieved from Editorial Manager, and gender and POC status were obtained through: 1) e-mail communication with corresponding authors; 2) e-mail queries to other coauthors; 3) NamSor software, and 4) Internet searches. The data were described using percentages and summary statistics. A two-sample test of proportions was used for comparisons and trends were analyzed with linear regression.

Results

We identified 314 manuscripts (1,555 authors), 95 (461 authors) of which were accepted for publication. Of all authors, 515 (33%) were women, and women were the lead and senior authors of 101 (32%) and 69 (23%) manuscripts, respectively. There were no differences in women’s representation as authors between accepted and rejected manuscripts. Overall, 923/1,555 (59%) authors were identified as POC, with a significantly lower proportion of POC authors among accepted vs rejected manuscripts (41%, 188/461 vs 67%, 735/1,094; difference, -26%; 95% CI, -32 to -21; P < 0.001). We did not observe significant trends in the proportion of women and POC authors over the study period.

Conclusion

The proportion of women authors of COVID-19 manuscripts was lower than men’s representation. Further research is required to determine the factors that account for the higher proportion of POC authors across rejected manuscripts.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12630-023-02455-w.

Women are under-represented compared with men in academic scholarly activities, including as authors,¹ conference speakers,² members of clinical practice guideline panels,³ and peer reviewers and editorial board members.⁴ The COVID-19 pandemic has had a disproportionate impact on women researchers and has exacerbated gender inequities in formal and informal work, as well as in the distribution of home responsibilities and dependent care obligations.^5,6 Consequently, women scientists have reported larger declines in the time they could devote to research compared with men; and these declines were magnified for women with young children and single mothers.⁵ In a comparative quantitative content analysis, the proportion of 1,893 COVID-19 publications with a woman first author was 19% lower than that for publications in the same journals in 2019.⁷ In another study of more than 1,200 COVID-19 publications, women accounted for 34% of all authors, with slightly lower representation as lead and senior authors (29% and 26%, respectively).⁸ While they noted that representation of women authors was greatest in Oceania and lowest in Africa, these differences were not statistically significant.⁸

The COVID-19 pandemic also disproportionately affected socially marginalized individuals, and may have had an uneven impact on opportunities for scholarly activities. Publications arising from these sources provide an important and valuable perspective. In contrast to gender data, there are no Canadian or international studies reporting the racial and ethnic representation of authors of COVID-19 manuscripts to date. The Royal Society of Chemistry Joint Commitment Statement on Inclusion and Diversity for Scholarly Publishing⁹ includes a recommendation to understand our research community and enable collection of diversity data to understand where action is needed.

We sought to investigate the gender and person of colour (POC) status of authors of COVID-19 manuscripts submitted to the Journal and determine potential trends in their representation during 15 months of the pandemic.^4,10

We previously reported the diversity of authors of publications by the Canadian Critical Care Trials Group (CCCTG): 40.1% and 33.6% of 354 publications had a woman first and last author respectively, and 18.0% of authors identified as people of colour.¹⁰ Goldstone et al. reported that 36.2% of 1,300 manuscripts submitted to the Journal in 2016 and 2017 had a woman as first and/or corresponding author.⁴ Based on these recent data, we hypothesized that: 1) the percentage of lead authors identified as women would be less than 35%; 2) the percentage of senior authors identified as women would be less than 25%; and 3) the proportion of authors identified as POC would be less than 18%. Secondarily, we compared the gender and racial status of authors of accepted vs rejected manuscripts and explored trends in proportion of women and POC authors over the study period.

Methods

The study was approved by the Sinai Health Research Ethics Board (#21-0034-E), who waived the need for informed consent. Access to Editorial Manager was provided by the Journal. We analyzed all manuscripts that met the following criteria: 1) submitted between 1 February 2020 and 30 April 2021; 2) historically classified by the Journal as COVID-19-related at the time of submission; and 3) received a final decision. Information regarding each manuscript was extracted from Editorial Manager (Aries Systems Corp., North Andover, MA, USA), the software platform used for manuscript submission.

Data collection

We extracted the following data for each manuscript: category (i.e., reports of original investigation, case reports/series, review articles, reflections articles, special articles, letters to the Editor presenting original material, letters to the Editor commenting on published material, images), specialty focus (anesthesiology, critical care, or combination), total number of authors, publication date, and the geographic location of each author’s primary affiliation. Manuscripts that did not include any authors with a primary affiliation in Canada were designated as “authored outside of Canada.” We noted the final disposition of each manuscript: submitted but not accepted (“rejected”), or accepted for publication (“accepted”). In addition to gender and POC status, we categorized authors as being in the lead (first), senior (last), or middle position. The lead author was defined as the first named author on the manuscript. The last-named author was referred to as the senior author. An author who was not the lead or senior author was termed as an author in the middle position. If a manuscript had a single author, we considered it to have only a lead author and no senior author.

Determination of gender and person of colour status

We identified the gender of each author and POC status using a series of methods. The definition of POC as per the Canadian Employment Equity Act is “persons, other than Aboriginal peoples, who are non-Caucasian in race or non-White in colour;”¹¹ and a single question on perception or treatment as a POC has been shown to identify visible minorities.¹² Gender and POC data were obtained through the following hierarchical sequence: 1) e-mail communication with lead/senior/corresponding authors; 2) e-mail queries to other coauthors; 3) NamSor software (NamSor SAS, Versailles, France), which identifies with 97% accuracy a person’s gender based on their first name;¹³ and 4) Internet search of individual authors for pronouns, photos, or reference to gender or race/ethnicity. The e-mail asked recipients about the gender (man, woman, or non-binary) of coauthors; and “whether she/he/they would self-identify as a person of colour (a term which applies to all people who are not seen as White by the dominant group).”

Statistical analysis

The data are described using numbers or total sample sizes, and percentages. Missing data were omitted from the analysis. We used a two-sample test of proportions to explore the proportion of women and POC authors between accepted and rejected manuscripts. We evaluated changes in the proportion of women and POC authors over the 15-month period using linear regression. Data analysis was performed using STATA 12.1 (StataCorp LLC, College Station, TX, USA), and P < 0.05 was considered statistically significant. For comparisons between accepted and rejected manuscripts, P < 0.0025 was considered significant after Bonferroni adjustment for 20 comparisons. For our analysis of trends over time, P < 0.004 was considered significant after adjusting for 12 comparisons.

Results

We identified 314 COVID-19-related manuscripts: 95 manuscripts with 461 authors were accepted for publication and 219 manuscripts from 1,094 authors were submitted but rejected for publication. Forty-one (43%) accepted and 176 (80%) rejected manuscripts were authored outside of Canada, with the top countries including the USA, China, and India. The authors of accepted and rejected manuscripts represented 18 and 26 countries, respectively (see Electronic Supplementary Material [ESM] Figs. 1–2). Using World Bank country income categories,¹⁴ 16%, 19%, and 65% submission were from lower-middle, upper-middle, and high-income countries, respectively; there were no submissions from low-income countries (ESM Table 1). The specialty focus of included articles was 65% anesthesiology (205/314), 22% critical care (68/314), and 13% (41/314) combination.

Table 1.

Manuscript and author demographic variables

Demographic variable
	Accepted manuscripts N = 95	Rejected manuscripts N = 219	Difference (95% CI)	P value
Gender
Women authors	154/461 (33%)	361/1094 (33%)	0% (-5 to 6)	0.92
Women lead authors	32/95 (34%)	69/219 (32%)	2% (-10 to 14)	0.80
Women coauthors if woman lead authora	43/119 (36%)	116/288 (40%)	-4% (-15 to 7)	0.50
Women coauthors if man lead author	79/247 (32%)	176/587 (30%)	2% (-5 to 9)	0.62
Women senior authorsb	17/85 (20%)	52/210 (25%)	-5% (-16 to 6)	0.47
Person of colour status
POC authors	188/461 (41%)	735/1094 (67%)	-26% (-32 to -21)	< 0.001*
POC lead authors	50/95 (53%)	160/219 (73%)	-20% (-33 to -8.)	< 0.001*
POC coauthors with POC lead authorc	89/158 (56%)	525/622 (84%)	-28% (-37 to -19)	< 0.001*
POC coauthors with White lead author	49/208 (24%)	50/253 (20%)	-4% (-4 to 12)	0.38
POC senior authorsb	43/85 (51%)	144/210 (68%)	-18% (-31 to -5)	0.006
Intersectionality
Women POC authors	67/461 (15%)	263/1094 (24%)	-10% (-14 to -5)	< 0.001*
Women POC lead authors	10/95 (11%)	53/219 (24%)	-14% (-23 to -5)	0.009
Women POC senior authorsb	17/85 (20%)	44/210 (21%)	-0.9% (-12 to 21)	0.98

Data are presented as n/total N (%).

P values represent comparison of manuscripts that were accepted and those that were rejected for publication

*Statistically significant if P < 0.0025 following Bonferroni correction for 20 comparisons

^aNumber of women coauthors if a manuscript had a woman lead author (excludes lead author)

^bStudies with one author only (accepted n = 10, rejected n = 9) were removed from the senior author data, hence the denominators for “accepted” and “rejected manuscripts” are 85 and 210, respectively

^cNumber of POC coauthors if a manuscript had a POC lead author (excludes lead author)

CI = confidence intervals; lead author = first named author; POC = person of colour; senior author = last named author

We were able to establish the gender, POC status, and geographical location of all authors using the approach outlined in the methods. For most authors, gender and POC status data were reported by the lead or corresponding author. Gender and POC status were assigned (using NamSor software and internet searches, as listed in the Methods) to 4% (18/461) and 12% (127/1,094) authors of accepted and rejected manuscripts, respectively.

Author demographics

The characteristics of included manuscripts are provided in Table 1. Of 1,555 total authors, 515 (33%) were women, with a similar proportion in accepted and rejected manuscripts (n = 154 [33%] and n = 361 [33%], respectively; P = 0.92). Women were the lead and senior authors of 101 (32%) and 69 (23%) of all manuscripts, respectively, with no differences between accepted and rejected manuscripts (Table 1). Across all authors, 1,040 (67%) identified as men, and no authors identified with a non-binary gender.

Overall, 923 (59%) authors were POC; with 188 (41%) and 735 (67%) POC authors among the accepted and rejected manuscripts, respectively. Persons of colour authors were the lead and senior authors of 50 (53%) and 43 (51%) accepted manuscripts, respectively. The proportion of total POC authors was lower across accepted compared with rejected manuscripts (41%, 188/461 vs 67%, 735/1094; difference, -26%; 95% confidence interval [CI], -32 to -21; P < 0.001). Similarly, the proportion of POC lead authors was lower in accepted manuscripts than in rejected manuscripts (53%, 50/95 vs 73%, 160/219; difference, -20%; 95% CI, -33 to -8; P < 0.001). The proportion of POC senior authors was similar across accepted vs rejected manuscripts after correcting for multiple comparisons (51%, 43/85 vs 69%,144/210; difference, -18%; 95% CI, -31 to -5; P = 0.007). In the 50 accepted manuscripts with POC lead authors, a greater proportion of coauthors identified as POC compared with manuscripts with white lead authors (56%, 89/158 vs 23%, 49/208; Table 1).

Among all authors, 21% (330/1,555) identified as both women and POC; this proportion was lower among accepted vs rejected manuscripts (15%, 67/461 vs 24%, 263/1,094; difference, -10%; 95% CI, -14 to -5; P < 0.001). After correcting for multiple comparisons, the proportion of women POC lead authors was similar across accepted vs rejected manuscripts (11%, 10/95 vs 24%, 53/219; difference, -14%; 95% CI, -23 to -5; P = 0.009). There were no significant differences in the proportion of women POC senior authors (Table 1).

Across all manuscripts (accepted and rejected), there was an increase in women lead authors over the study period from 28% (55/200) during the first five months to 34% (13/38) in the last five months (2% increase per month; 95% CI, 0 to 4; P = 0.041); however, this result was not significant after adjusting for multiple comparisons (P < 0.004) and our study included only a small number of manuscripts in the final five months of the study period. The overall percentage of total women (1% per month; 95% CI, -0 to 2; P = 0.16) and women senior authors (0.8% per month; 95% CI, -2 to 3; P = 0.52) was similar across the study period (Figure). We found similar proportions of overall women authors of accepted and rejected manuscripts across the study period (ESM Fig. 3), with a similar finding in total, lead, and senior POC authors over the 15-month period (ESM Fig. 4).

Figure.

Percentage of women authors overall, in the lead author position, and senior author position across all (accepted and rejected) COVID-19 manuscripts from February 2020 to April 2021. Across all manuscripts (accepted and rejected), the percentage of women lead authors (orange line) increased by approximately 2% per month (95% CI, 0.04% to 4%; P = 0.041); however, this result was not significant after adjustment for multiple comparisons (P < 0.004). The changes in percentages of total women authors (0.96% per month; 95% CI, -0.42% to 2%; P = 0.16) and women senior authors (0.77% per month; 95% CI, -2% to 3%; P = 0.52) were not statistically significant.

Manuscript type

Of 314 total manuscripts, the most common manuscript categories were letters to the Editor presenting original material (41%, n = 129) followed by reports of original investigation (19%, n = 59), and editorials (9%, n = 29) (ESM Table 2).

Discussion

Across COVID-19-related manuscripts submitted to the Journal, women and POC authors were represented in similar proportions to historical prepandemic benchmarks. Overall, women authors comprised 32% of lead authors, and we identified similar proportions of total, lead, and senior women authors between accepted and rejected manuscripts. Persons of colour comprised 59% of authors and were well-represented, with a higher proportion of total POC and POC lead authors among rejected compared with accepted manuscripts. We did not identify clear trends in representation of women or POC authors over the study period, and our trend analysis was limited by a small number of manuscripts in the final months of the study. Finally, the majority of manuscripts originated from upper-middle or high-income countries, with no submissions from low-income countries; this was not the focus of our study but is deserving of further investigation. Our results capture a snapshot of the diversity of authors of COVID-19-related research in anesthesiology and critical care, which provides a benchmark for future comparison and several areas for further exploration.

Our results are consistent with some previous reports, although the literature is mixed on the impact of COVID-19 on author demographics. The proportion of women lead authors identified in our study (34%) was similar to a previous study by Goldstone et al., which found 36% of manuscripts published in 2016–2017 had a woman lead and/or corresponding author.⁴ Consistent with our study, other investigators have not found a significant decline in women’s representation during the pandemic.^8,15 The proportion of women authors across accepted COVID-19 manuscripts in our study aligns with the findings of Pinho-Gomes et al. (34%, 2,285/6,722).⁸ Misra et al. also found that, while women were underrepresented as authors relative to men, this was not significantly different for COVID-19 manuscripts.¹⁵ In contrast to our results, other studies have described a substantial reduction in authorship by women during the pandemic, and have hypothesized that this was related to disproportionate dependent care obligations.^7,8 Andersen et al. noted a 19% decline in women lead authors across several medical specialties, with the decline particularly evident early in the pandemic.⁷ This discordance may be due to the geographic origin of the manuscripts and differential domestic support available for women in these areas. We acknowledge the limitations of attempting to draw firm conclusions when comparing with other analyses due to inherent differences in sampling, data extraction, and overall trends that were independent of the pandemic.

Notably, we observed a greater proportion of lead or senior POC authors across manuscripts that were rejected for publication. While our study did not identify the reasons for this observation, possible explanations could include lower manuscript quality, geographical location, systemic bias, and other confounders. On the positive side, the proportion of lead authors across accepted manuscripts who identified as POC (53%) was higher than prepandemic publications by the CCCTG (15%).¹⁰ We also observed a higher proportion of women among POC senior authors than was observed in prepandemic publications.¹⁰ Currently, we lack robust comparator data on the representation of POC authors in multinational, prepandemic publications, which makes it difficult to draw firm conclusions. The present analysis provides a baseline assessment that can be used as a comparator going forward.

Manuscripts with POC lead authors had a larger proportion of POC coauthors while those with white lead authors had a greater proportion of white coauthors. This observation may be influenced by factors including geographical location, collaborations among national colleagues, or less likely homophily.⁴ This finding should be interpreted with caution given the limited sample size and confounders; and factors that contribute to research collaborations should be studied further.

The strengths of this study are the inclusion of both accepted and rejected manuscripts, and the minimal proportion of missing data on author gender and POC status. Our study is the first to report the representation of POC authors of COVID-19 manuscripts. Limitations of our study include the relatively small sample size, the decrease in the number of COVID-19 submissions over the study period, and the potential misassignment of author gender or POC status. Similarly, trends observed during the study period may reflect overall trends or variability unrelated to the pandemic. We cannot exclude that the representation of women authors reflects the proportion of women who are conducting research relative to other scholarly pursuits that may not involve publishing. While we have contrasted our results with recent publications, we did not conduct a formal comparison with the prepandemic representation of women and POC authors in the Journal. Finally, as we did not evaluate gender and race of editors and reviewers, we cannot comment on possible bias in peer review.

The “POC” term is imperfect because it distills diverse cultures, ethnicities, and narratives into a single term; however, we were unable to establish precise racial and ethnicity data without contacting every coauthor. We applaud initiatives by journals and publishers to prospectively collect self-reported granular demographic data from authors, reviewers, and editorial board members;¹⁶ this will hopefully obviate the need for retrospective data collection. Without information on the challenges faced by women and men during the COVID-19 pandemic, we can only hypothesize that disproportionate dependent care obligations impacted women’s scholarly work.

Our study provides novel data on the gender and racial representation of authors of COVID-19-related research publications in the Journal, and informs our understanding of equity issues in academic publishing throughout the pandemic and provides a baseline for future comparisons. While women and POC authors were represented at similar rates to prepandemic publications, women continue to be under-represented as authors relative to men. The impact of the ongoing pandemic on equity in scholarly productivity should continue to be investigated.

Supplementary Information

Below is the link to the electronic supplementary material.