Women’s representation among academic first authors has risen steadily over the past two decades. Their representation among senior (last) authors and corresponding authors — the positions most closely tied to research leadership, grant credit, and career advancement — has not kept pace. A cluster of bibliometric studies published or updated in 2025 quantifies that gap in detail, across fields ranging from critical care medicine to the natural sciences broadly, and the pattern is consistent even where the exact numbers differ by discipline.
What the 2025 data shows
The most granular recent dataset comes from a 2025 study in Critical Care by Bruns and colleagues, which analyzed 42,970 articles published between 2005 and 2024 in five of the highest-impact critical care and respiratory medicine journals (including The Lancet Respiratory Medicine, Intensive Care Medicine, and the American Journal of Respiratory and Critical Care Medicine). Among the 34,743 multi-author papers, the authors found:
- 56.7% had a male senior (last) author and a male first author.
- 23.1% had a female first author paired with a male senior author.
- 12.5% had a male first author paired with a female senior author.
- Only 7.8% had both a female first author and a female senior author.
The trend line over the study’s 20-year window shows real but slow movement: papers with a female first author and male senior author rose from 21.0% of the sample in 2005 to 24.5% in 2024, and papers with both a female first and female senior author roughly doubled, from 4.8% to 12.1%, over the same period. At the annual rate of change the authors calculated, they concluded that parity in senior authorship in this field is decades away, not years.
A broader picture comes from Nature Index’s own tracking of authorship across the journals it indexes, reported in Nature‘s news coverage of the data. In the natural sciences overall, women held 17% of last-author positions in 2025, up only marginally from 15% in 2015 — despite first-authorship share rising from 28% to 29% over the same decade. Health sciences show a similar split at a higher overall base: women accounted for 41% of all authorships and 44% of first authorships in 2025, but only 31% of last authorships. Within clinical sciences specifically, women held 44% of first-author positions but just 27% of last-author positions. Across every breakdown, the gap between first-authorship share and senior-authorship share for women is the consistent finding — not the underrepresentation of women in research output generally, which has been closing for years, but a narrower, more persistent gap in who leads it.
Why corresponding and senior authorship specifically
Senior (last) authorship is a positional convention, strongest in biomedical and life-science fields, that signals the researcher who supervised or led a project — typically a lab head or principal investigator — rather than the person who did the most hands-on work (usually the first author). Corresponding authorship is a related but formally distinct role: the person named as the point of contact for editorial correspondence, and often, informally, the person treated as accountable for the paper. The two roles frequently, but not always, overlap — see CASRAI’s comparison of senior author vs. corresponding author for how journals and disciplines draw the line differently, and the co-corresponding author convention some journals now formally support.
Both positions matter disproportionately for career advancement: tenure and promotion committees, grant reviewers, and hiring committees widely read senior/corresponding authorship as evidence of independent research leadership in a way they do not read middle authorship or even first authorship in the same field. A persistent gap at this specific position in the authorship line — even alongside real progress in overall representation and first authorship — is therefore a gap in who gets credited as a research leader, not simply a gap in who publishes.
Methodology caveats worth reading past the headline numbers
A few limitations recur across this body of research and are worth carrying into any policy use of these figures:
- Gender is usually inferred, not self-reported. The Bruns et al. study, like most large-scale bibliometric gender research, used an algorithmic name-to-gender classifier (NamSor), validated against a smaller human-coded sample (94.3% agreement in that study). Name-based inference is a reasonable large-N proxy but is known to be less reliable for names common across multiple genders, for authors from naming conventions the classifier was trained less well on, and it cannot represent non-binary authors at all — a limitation acknowledged across this literature, not specific to any one study.
- “Senior author” is not a standardized field. Author-position conventions vary by discipline and even by sub-field; some studies equate “senior author” with “last-listed author” as a proxy, which is a reasonable approximation in fields where that convention is strong (much of biomedicine) and a weaker one where it isn’t (e.g., large collaborations, alphabetical-order fields, or humanities-adjacent disciplines).
- Findings are field-specific and shouldn’t be over-generalized. A 2025 methodological review in the Journal of the Association for Information Science and Technology surveying bibliometric approaches to studying gender in science notes that results vary meaningfully by discipline, country mix, and time window studied — a critical care finding and a physics finding are not interchangeable evidence for a single “the gender gap in science” claim, even when the direction of the finding is the same.
Why this matters for research administration and DEI policy
For research offices, funders, and institutions tracking equity commitments, this data has practical implications beyond general awareness:
- Authorship-position data is auditable institutional data. Unlike self-reported climate surveys, authorship position by gender can, in principle, be tracked from an institution’s own publication output — making it a concrete metric research offices can monitor over time, alongside grant PI rates and promotion-committee outcomes, rather than relying solely on perception surveys.
- It connects to existing equity frameworks institutions may already be signed onto. Programs like the Athena SWAN Gender Equality Charter already ask departments to examine gender patterns in research roles and career progression; authorship-position data is a natural, low-cost input to that kind of self-assessment.
- It intersects with the shift toward contribution-based crediting. Frameworks like CRediT (Contributor Roles Taxonomy) that record specific contributions rather than relying solely on author-list position and order give institutions a second, complementary data source — one less dependent on a single positional convention — for understanding who actually led a given piece of work. See CASRAI’s CRediT and authorship hub for more on how structured contribution statements work alongside traditional author order.
- Responsible research assessment initiatives are a relevant policy lever. Signatories of DORA and the Coalition for Advancing Research Assessment (CoARA) commit to evaluating researchers on criteria broader than journal-level metrics and simple authorship position — a commitment that, taken seriously, should also reduce the structural premium currently placed on senior/corresponding authorship specifically, since that premium is part of what these bibliometric studies are measuring the effects of.
None of this changes with a single study or a single year of data — the consistent message across the 2025 literature is a slow rate of change, not a sudden shift. For research administrators, the practical takeaway is that first-authorship parity, where it exists, is not a proxy for parity in research leadership credit; the two need to be tracked separately, and the second is moving much more slowly than the first.
