Temporal and geogrpahic trends in female representation within European cardiovascular imaging among 13,436 individuals associated with EACVI

Abstract Introduction Women are under-represented in cardiology trial enrolment, specialty training, faculty, scientific authorship and board membership in cardiology journals and societies. An EACVI Task Force on Women in Cardiovascular Imaging aims to promote women’s representation and leadership, but data is lacking in this area. Purpose To characterise the representation of women in cardiovascular imaging within Europe and examine temporal and geographic trends. Methods We scraped publicly available name data from over 100 online sources including EACVI taskforces, committees, research project members, consensus and position paper authors, board membership and past presidents alongside members of affiliated cardiovascular imaging national societies and working groups. We also included names from EACVI or national accreditation records for cardiac imaging. We recorded data on country location and year of affiliation. We defined gender in a binary manner as either male or female. The previously validated artificial intelligence (AI) system Namsor assigned gender based on name morphology and geolocation data. We pre-set a confidence of gender-name matching of 95%. Results The AI was able to assign gender at 95% confidence to 12,643/13,436 (94.1%) names. The overall number of names assigned female gender was 5903/12643 (46.7%), rejecting the null hypothesis that gender was equally distributed (i.e. 50% female) [p<0.0001]. Among the 9712 individuals with year of affiliation we observed female representation to increase over time: 1989-2000 4/18 (22.2%), 2001-2004 107/287 (37.8%) 2005-2009 77/170 (45.3%), 2010-2014 109/203 (53.7%), 2015-2019 1063/2022 (51.3%) and 2020-2024 3352/3064 (52.2%) (Cochran-Armitage trend test 24.196 [p<0.0001]) (Figure 1). Female representation was lower in countries outside of Europe, compared to within Europe; 740/2654 (27.9%) vs. 5163/9989 (51.7%) [p<0.0001]. The lowest rate of female representation was in Saudia Arabia (15.1%), the highest in Lithuania (83.3%). Figure 2 displays a choropleth of gender representation (%) between European countries with n>20. Conclusion(s) There are significant limitations in this form of analysis, including that we have arbritarily defined gender in a binary format, thereby excluding individuals who do not identify with such a social construct. It is also likely that the excluded names based on the 95% confidence of the AI software disproportionately represents individuals from ethnic minorities. However, our data provides a basic understanding of the representation of women in European cardiovascular imaging over time and between countries. We observed a trend of increasing female participation but with substantial variation between countries both inside and outside of Europe. This data may be useful to refine and direct efforts to improve female representation in areas in which it is needed most.