Trends in diversity of scientific authorship by ethnicity, sex and age of authors in a National Institute for Health and Care Research Biomedical Research Centre in England: A retrospective longitudinal analysis

ObjectiveScientific publications are important markers of achievement in academic medicine. Whilst gender disparities in scientific authorship are well documented, there is little understanding of the influence of authors protected characteristics on authorship positions, publication outputs and citations in biomedical and translational research. This study examined the diversity of scientific authorship by three protected characteristics, i.e., ethnicity, sex and age of authors of biomedical research publications. DesignA retrospective longitudinal descriptive analysis of 476 unique authors protected characteristics, biomedical scientific publications and citations from 2012 to 2022. SettingNational Institute for Health and Care Research (NIHR) Oxford Biomedical Research Centre (BRC) in England. DataPeer-reviewed publications funded by the NIHR Oxford BRC during its funding round 2 (2012-2017) (henceforth BRC2) and round 3 (2017-2022) (henceforth BRC3). BRC author-level data were extracted on three self-reported protected characteristics, i.e., ethnicity, sex (male and female) and age of authors. Data were analysed by descriptive statistics, Fishers exact test and Pearsons chi-squared test of independence to explore the proportions of first and last (senior) authorship positions and number of publications and citations by ethnicity, legal sex and age of authors. Primary outcome measuresAuthorship positions (first and last authors), number of publications and citations of publications by author ethnicity, sex and age. ResultsThere were statistically significant differences in first and last authorship positions and number of publications and citations of publications by author ethnicity, sex and age. White researchers were 79% and 85% of first authors and 85% and 86% of last authors in BRC2 and BRC3, respectively. In BRC2, females were more (53%) in the first author category, while males were more (58%) in the last author category (p<0.014). In BRC3, male first (70%) and last authors (78%) were statistically significantly more than female first (30%) and last (22%) authors (p<0.001). For the total study period, 44% of first and 24% of last authors were female, which was statistically significantly less than male first (56%) and last (76%) authors (p<0.001). Authors aged 40-59 were statistically significantly more (79% and 71% of first authors and 73% and 75% of last authors in BRC2 and BRC3, respectively) than authors of other age groups (p<0.001). White authors published more publications (87%) than BAME authors (13%) in both BRC periods, and there were no differences in the overall proportions of publications by authors of either ethnicity in both BRC periods (p=0.91). In BRC2 and BRC3, male authors produced statistically significantly more publications (59% and 73%) than female authors (41% and 27%, respectively, p<0.001). Authors aged 40-59 years published more papers (64% and 69% of total papers in BRC2 and BRC3, respectively) than authors of other age groups (p<0.001). Citations of publications by white authors were statistically significantly higher (89% and 87% of total citations of BRC2 and BRC3 publications, respectively) than publications by BAME authors (p<0.001). Publications by male authors had higher citations (65% and 68% of total citations of BRC2 and BRC3 publications, respectively) than female authors publications (p<0.001). Citations of publications by authors aged 40-59 years were higher (56% and 72% of total citations of BRC2 and BRC3 publications, respectively) than citations of publications by authors aged less than 40 years and those aged 60 and above (p<0.001). Authors aged 60-64 years had the highest average citations per publication in both BRCs and the total period of the study. In the BRC3 period, some positive trends were observed, such as a substantial increase in the absolute number of BAME and female first and last authors and young first authors (aged up to 39 years); an increase in the number of publications by BAME, female and young authors (aged up to 39 years); and an increase in total citations of publications by BAME, female and young authors, and higher average citations of female authors publications (42 citations per publication) compared to male authors publications (33 citations per publication) in BRC3. ConclusionThere are statistically significant differences in diversity of authorship positions, publication output and citations by ethnicity, sex and age of authors. Positive trends include an increase in BAME, female and young authors and their publications and citations; however, under-representation of authors of BAME background, female and younger age is identified. Gaps in scientific authorship, publications and citations by author ethnicity, sex, and age could be addressed by systematic collection of data on protected characteristics of authors (researchers). In addition, tackling disparities in recruitment, career progression and research funding is imperative. STRENGTHS AND LIMITATIONS OF THIS STUDYO_LIThis is the first study to examine diversity at author level by ethnicity, sex and age to our knowledge in three publication metrics: authorship, publications output and citations count within the context of an NIHR BRC. C_LIO_LIMost studies have examined gender individually, here we explore the influence of ethnicity, sex and age of authors on authorship type (first and last authors), publications productivity and citations count. C_LIO_LIThis study provides benchmarking data on scientific authorship diversity of relevance to other NIHR funded BRCs. C_LIO_LIThis study identified the challenges of benchmarking protected characteristics at individual author level because such data is not routinely available or collected. C_LIO_LICompletion of protected characteristics data is self-reported, not mandatory and therefore incomplete. C_LIO_LIHowever, the data does provide important benchmarking data and the feasibility of gathering such information at author level. Further work could be completed to improve the reliability and comprehensiveness of such data. C_LI