The menstrual cycle is still treated as noise in clinical research

Controlling for cycle phase is easier than studying it. That choice has quietly removed a variable that changes drug response, pain and symptom reporting.

8 September 2026 · last reviewed 8 September 2026 · 3 min read

A laboratory researcher examining a sample under a microscope — Photo by Edward Jenner on Pexels

There is a defensible reason clinical research controls for the menstrual cycle rather than studying it: variance is the enemy of statistical power, and hormonal fluctuation adds variance. There is also an indefensible consequence, which is that we now know very little about how a normal physiological cycle interacts with the treatments half the population is given.

Key points

Cycle phase can affect symptoms, measurements and treatment response.

Excluding variation can make findings less useful in real life.

Teams can collect practical cycle data without overburdening participants.

How the variable disappears

Three routine practices remove cycle phase from the data.

Timing. Assessments are scheduled in the early follicular phase, when hormone levels are lowest and most stable, precisely because it looks most like a male baseline.

Recruitment. Studies preferentially enrol postmenopausal women, or women on hormonal contraception, or men, when the outcome is thought to be hormone-sensitive.

Silence. Cycle phase is not recorded at all, so it cannot be analysed later and its absence never appears in the limitations section.

The third is the most common and the most damaging, because it leaves no trace.

What gets lost

Pharmacokinetics. A review of pharmacokinetic research found that menstrual-cycle effects vary by medicine and that the evidence is limited and inconsistent, which is why compound-specific study is needed. [1]

Symptom measurement. Cycle-related changes have been reported across pain, mood and sleep outcomes, but their size and consistency differ across studies. Poorly balanced assessment timing can therefore confound a modest treatment effect. [2] [3]

Condition-specific effects. Migraine and epilepsy are among the conditions with documented menstrual-cycle patterns. Evidence for other conditions is less consistent and should be assessed separately. [4] [5] Treating those patients on evidence that averages across the cycle is treating them on a blurred picture.

The cycle itself as an outcome. If cycle disruption is never measured, a treatment that disrupts it will not be flagged, and patients discover the effect after approval.

The circular problem

Cycle phase is excluded because it adds variance. Because it is excluded, we have no good estimate of how much variance it actually adds, or for which outcomes. The absence of evidence that it matters is then used as evidence that it does not.

Breaking the circle requires studies that deliberately measure across phases, which cost more and need larger samples. That is a funding and recruitment problem before it is a science problem.

What a research team can do now

Always record the date of the start of the last period, plus cycle regularity, contraception and menopausal status. Four questions. Even if you do not analyse by phase, the next person can.
Report the distribution. Say what phases your sample was measured in. It costs a sentence and it makes your study reusable.
Decide deliberately, not by default. Either constrain to a phase window and say so, or power for phase analysis. Pooling without acknowledgement is the option that hides the problem.
Ask for the date, not the label. Participants can tell you when their last period started. Most cannot reliably tell you which phase they are in.

FeFe screens on cycle stage at the panel level, so studies that need a specific phase window can be matched rather than searched for. The practical guidance for that is in recruiting for women's health research.

For the wider context, see the women's health data gap.

References

Common questions

Does the menstrual cycle affect drug response?

For some drugs, yes. Hormonal fluctuation can influence gastric emptying, plasma protein binding, drug metabolising enzyme activity and receptor sensitivity. The size of the effect varies by compound, and for many drugs it has simply never been tested, which is the point.

How should a study account for cycle phase?

At minimum, record the date of the start of the last period and cycle regularity, and report the distribution. If the outcome is plausibly hormone-sensitive, either schedule assessments within a defined phase window or power the study to analyse by phase rather than pooling.

Recruiting for a women's health study?

Filter a screened panel by condition, treatment and cycle stage. Free to set up, and you only pay for completed responses.

Start recruiting free