SAGER guidelines: how to report sex and gender in a research paper
By Mubarak MutesasiraUpdated 7 min read
Short answer:
The SAGER guidelines (Sex and Gender Equity in Research) tell authors how to report sex and gender in a research paper: use the two terms correctly, state the sex and gender of participants or source material in the title and abstract where relevant, explain how sex and gender were considered in the design, report results separately by sex and gender where appropriate, and discuss what that means, including why if no such analysis was done. They were published by Heidari and colleagues in 2016 and apply to studies of people, animals and cells.
This guide is for anyone writing up an empirical study in health, life sciences, psychology or the social sciences. It sits in our series on clinical studies and reporting; for the trial set-up side, start with the site initiation visit checklist.
Why SAGER exists
Many studies have enrolled mostly men, used only male animals or did not record the sex of cell lines, and then reported results as if they applied to everyone. When a drug's dose, a device's fit or a risk factor works differently by sex, that gap can mislead readers and clinicians. SAGER was developed by the European Association of Science Editors (EASE) Gender Policy Committee to make the sex and gender dimension of a study visible in what is published, so readers can judge whom the findings apply to.
SAGER does not require every study to be powered for sex differences. It asks you to be clear about what you did and did not do.
Sex vs gender
The first SAGER principle is to use the terms carefully and not interchangeably.
| Sex | Gender | |
|---|---|---|
| What it is | Biological attributes: chromosomes, gene expression, hormones, reproductive anatomy | Socially shaped roles, identities, norms, behaviors and relations |
| Applies to | Humans, animals, cells, tissues | People |
| Usual categories | Female, male, intersex | Woman, man, non-binary and other identities, as the study defines them |
| Example question | Do female and male mice differ in drug clearance? | Do women and men differ in how often they seek care for chest pain? |
Three practical consequences:
- Say how you determined it. Self-report, medical records, genotyping, observation of the animals. "Gender" collected on a form that offered only "male" and "female" is usually sex, or a mix; describe what was asked.
- Do not swap the words for style. "Gender differences in bone density" is wrong if the measure is biological; write "sex differences".
- Some questions involve both. Smoking-related lung disease can differ by sex (biology) and by gender (who smokes and how much). Say which you think explains the result and how sure you are.
What SAGER asks for, section by section
SAGER gives its recommendations by part of the paper. The checklist below follows its published numbering; items 2 and 4 to 6 are split into parts so each can be checked on its own, as in the reporting checklists template. This is our paraphrase; read the original before submission.
| # | Section | What to report |
|---|---|---|
| 1 | General | Use the terms sex (biological) and gender (social) carefully and not interchangeably |
| 2a | Title and abstract | If only one sex was studied, or the findings apply to one sex or gender, say so in the title |
| 2b | Title and abstract | Name the sex of the participants or source material in the abstract, and give results by sex or gender where relevant |
| 3 | Introduction | Whether sex or gender differences may be expected, based on earlier work |
| 4a | Methods | How sex and gender were taken into account in the design |
| 4b | Methods | Whether females and males (or gender groups) were adequately represented, and how sex or gender was determined |
| 4c | Methods | Why either sex or a gender group was excluded, if it was |
| 5a | Results | Data presented separately by sex and gender where appropriate |
| 5b | Results | Sex- and gender-based analyses reported whatever their result, positive or negative |
| 5c | Results | For trials, withdrawals and dropouts reported by sex |
| 6a | Discussion | The possible effect of sex and gender on the results and their interpretation |
| 6b | Discussion | If no sex or gender analysis was done, the reason, and what it means for generalizing the findings |
SAGER complements the design checklist for your study. A trial reports against CONSORT and SAGER; an observational study against STROBE and SAGER.
Examples of good reporting
These are invented sentences that show the kind of reporting each item asks for.
Title (item 2a). Example (invented): "Effect of a high-fat diet on hepatic insulin resistance in male C57BL/6 mice." The reader knows at once that females were not studied.
Methods (items 4a to 4c). Example (invented): "Sex was recorded from the hospital record (female or male). Gender identity was asked in the baseline questionnaire with the options woman, man, non-binary and prefer to self-describe. Randomization was stratified by sex. We planned to enroll at least 40% of each sex, based on the sex distribution in the clinic population."
Exclusion (item 4c). Example (invented): "Only male rats were used because the outcome, a model of testicular injury, does not apply to females." Compare a weak version: "Male rats were used to avoid variability from the estrous cycle", which is a common reason that evidence has not supported as a general rule; if you use it, cite your evidence.
Results (items 5a to 5c). Example (invented): "Of 312 participants, 171 were women and 141 men. Withdrawals were 9 women and 11 men (Table 2). The difference in systolic blood pressure at 12 weeks was −6.1 mmHg (95% CI −9.0 to −3.2) in women and −5.4 mmHg (95% CI −8.6 to −2.2) in men; the treatment-by-sex interaction was not significant (P = 0.74)."
Discussion (item 6b). Example (invented): "We did not analyze results by gender because gender identity was not collected; the findings may not apply to transgender or non-binary people, and future studies should collect it."
How to analyze and report by sex when the study was not powered for it
Most studies are powered for the overall effect, not for a difference between women and men. SAGER does not ask you to pretend otherwise. It asks you to report what you have, honestly.
- Report descriptive results by sex. Numbers enrolled, baseline characteristics and outcomes for each sex, in a table or supplement. This costs nothing and lets future meta-analyses pool your data.
- Give effect estimates with confidence intervals for each sex. Wide intervals are informative: they show how little you know.
- Test the interaction, not two separate P values. "Significant in men (P = 0.03), not in women (P = 0.12)" does not show a difference between them. The right comparison is a treatment-by-sex interaction term, or a difference in effects with its confidence interval. See which statistical test to use for the models.
- Label it as exploratory. Unless sex was a prespecified hypothesis with power for it, say "exploratory" and avoid causal language.
- Prespecify when you can. Put the sex analysis in the protocol or statistical analysis plan, so it is not seen as a data-driven subgroup.
- Consider pooling. If your study is one of several, plan to contribute sex-disaggregated data to a meta-analysis.
- Say what it means. If the estimates in women and men overlap widely, write that the study cannot tell whether the effect differs by sex, not that it does not.
Common mistakes
- Using "gender" for a biological measure, or "sex" for a social one.
- Reporting "participants were 58% female" and nothing else by sex.
- Studying only male animals or cells with no reason given.
- Claiming a sex difference from separate subgroup P values.
- Leaving out a negative sex analysis because it was "not interesting" (item 5b asks for it).
- Not reporting dropouts by sex in a trial.
- Saying nothing in the discussion when no sex analysis was done.
Checking your paper against SAGER in AI for Research
The Reviewer guideline check in AI for Research now includes SAGER (2016) and SAMPL (2013), next to CONSORT 2010, STROBE 2007, PRISMA 2020, CARE 2013, STARD 2015, COREQ 2007, SRQR 2014 and SQUIRE 2.0. Upload or paste your manuscript, choose SAGER, and the check goes item by item: for each item it shows the passage that appears to report it, or marks "check by hand" when it finds none. The items are paraphrased and numbered as published. It costs 5 credits per 10,000 words; the Free plan has 45 credits a month (see pricing).
The check drafts the checklist; you confirm it. It can miss reporting phrased in an unusual way, and it cannot tell whether your sex analysis was the right one. It does not guarantee that a journal will accept the paper. More: reporting guideline checker.
For a blank version to fill by hand, use the reporting checklists template.
Learn more
- SAMPL guidelines: reporting the statistics
- Site initiation visit checklist, the pillar of this series
- ClinicalTrials.gov results reporting, where baseline sex or gender is a required measure
- How to review your own paper
- Reporting guideline checker and AI for medicine