A number worth pausing on

Across published analyses of European cardiovascular trials, women have represented, on average, around 38% of participants, despite accounting for roughly half of the disease burden in several major cardiovascular conditions. This is not a single anomalous study. The pattern repeats across heart failure, acute coronary syndrome and arrhythmia trials, with the gap generally widening in trials of more invasive interventions and in older age groups, where the female share of the affected population is in fact higher.

A number like this is easy to state and easy to move past. It is worth pausing on, because its consequences are statistical, not only descriptive.

Why representation is a statistical issue, not only a fairness one

When a subgroup makes up a small share of a trial population, the trial has limited statistical power to detect whether the treatment effect in that subgroup differs from the overall result, even when the overall trial is adequately powered. This is a direct, mechanical consequence of smaller subgroup sample sizes, not a matter of study quality or good intentions. A trial can be well conducted, well reported, and still be structurally unable to tell us whether an intervention works the same way in women as in men, simply because the numbers were never sufficient to ask that question with any precision.

This matters because cardiovascular disease does not present identically across sexes. Differences in symptom presentation, in underlying pathophysiology for some conditions, and in pharmacokinetics have been documented in the literature. A trial that cannot statistically resolve sex-based treatment effects is not neutral on this question. It is simply unable to answer it either way.

Where the gap actually comes from

The causes are more varied than a single explanation about recruitment. They include eligibility criteria historically built around presentations more typical in men, exclusion of women of childbearing potential in earlier phases of drug development, referral patterns for cardiovascular investigation that differ by sex in routine clinical practice, and, in some trial designs, recruitment sites or timelines that do not account for these differences. None of this points to a single cause, and describing it fairly means recognising that most of these factors emerged from reasonable, if outdated, precautions, rather than from a deliberate choice to exclude.

What changes at the design stage

From a statistical design perspective, there are concrete steps that address this directly, and several are now expected rather than optional. The EU Clinical Trials Regulation includes a requirement that trial populations be representative of the population likely to use the medicine, which in practice means sex balance needs to be considered explicitly during recruitment planning, not left to whichever patients happen to be referred. Pre-specifying sex as a subgroup of interest in the statistical analysis plan, with an honest statement of the power available to detect an interaction, is a second step that costs little and avoids the more common alternative: an underpowered post-hoc subgroup analysis presented with more confidence than the numbers support.

What this means in practice

For a sponsor or a research team, the practical implication is straightforward, even if the underlying issue is not. Sex representation should be planned, monitored during recruitment, and reported, in the same way that any other important design element would be. Where representative recruitment genuinely is not feasible within a trial's scope, the honest response is to state the limitation clearly, together with the statistical power actually available for subgroup conclusions, rather than to let a low subgroup count speak for itself in a results table.

Key points