Table 6


 G1Consider whether the research is concerned with sex (biological) or gender (behavioural) differences, and report the results accordingly*.
 G2Routinely provide sex-disaggregated results when reporting research on cardiovascular associations. This includes prespecifying subgroup analyses by sex. When there are no important sex differences, still include sex-specific results, most likely in the appendix of a manuscript for publication.
 G3Even when a study is concerned with associations for a single sex, where possible compare results for the other sex, as a control.
 G4Adjust at least for age when comparing sex-specific cardiovascular associations.
 G5Consider analyses on both the relative and absolute scales. When it is only appropriate to present relative risks, provide (at least) the number of events and the number at risk across the sex by risk factor exposure cross-classes, to give context to the reader.
 G6Quantify the sex difference (with accompanying measure of uncertainty, such as a 95% CI), rather than merely test for a significant difference.
 G7When analysing raw (ie, individual participant) data, use the full interaction model (with all main effects and two-way interactions) to obtain the sex-specific results, as well as the sex comparison(s).
 G8Unless there is statistical or clinical significance in the sex difference (ie, the sex interaction), avoid sex-specific conclusions.
Specific to meta-analyses
 M1Decide whether to use the fixed effect or random effects method before data are collected.
 M2Only include studies with results from both sexes.
 M3In the report, include a flow chart with reasons for exclusions. Clearly state the number of studies excluded for want of sex-disaggregated results.
 M4Use reliable, general, statistical software, such as R or Stata.
 M5Include forest plots by sex and to compare the sexes. Show age-adjusted and multiple-adjusted analyses separately, where appropriate. This will typically require placing some forest plots in the appendix of a manuscript for publication.
 M6Following the meta-analysis, use meta-regression and bubble plots to explore sources of heterogeneity, to include overall risk and the difference between the sex-specific risks.
 M7Take care when pooling ORs together with relative risks or HRs. Stratify pooling by the metric used where risk (or, in cross-sectional studies, prevalence) is typically high.
  • *In this manuscript no distinction is made, for simplicity of exposition.

  • †These have the advantage of offering a wide range of other tools, so that the extra work of learning the basics of such a package (if necessary) will be worthwhile.

  • ‡For example, through the ratio of relative risks—see figure 2.