The Exclusion Habit and What It Costs

Part of Consent and Vulnerable Populations: Pregnancy, Breastfeeding and the Evidence Gap.

About 6 minutes. Free, in your language, no account needed.

Why this matters

Pregnant and breastfeeding women are excluded from most clinical trials by default. The exclusion is usually written into the eligibility criteria without discussion, and it is presented as protective.

It protects the trial. Whether it protects women is a different question, and the answer is largely no, because the consequence is that women become pregnant while taking medicines nobody has studied in pregnancy, and their clinicians must advise them with no evidence.

Understanding the cost of the exclusion habit is what allows a person to recognize when it is being applied reflexively rather than reasoned.

The evidence gap it produces

Pregnancy changes pharmacokinetics substantially. Blood volume, renal clearance, hepatic metabolism and protein binding all shift, and a dose appropriate outside pregnancy may be too low or too high during it.

Because pregnant women are excluded from the trials that establish dosing, the dose used in pregnancy is frequently an extrapolation. For many widely used medicines, the evidence about use in pregnancy comes from registries, case series and post-marketing reports rather than from trials, which means it arrives late, is observational, and is weakest exactly where the decisions are hardest.

The same applies to breastfeeding, where the question is transfer into milk and effect on the infant, and where the evidence base is thinner still.

The result is a population that takes medicines and about whom there is no good evidence, which is a worse outcome than the one the exclusion was intended to prevent.

Why the habit persists

Genuine risk to the fetus. This is real and it is the legitimate core of the concern.

Liability. A harm occurring in a fetus is among the most serious outcomes imaginable and sponsors reasonably fear it.

Regulatory caution, both actual and anticipated, where teams exclude in the expectation that an authority would require it.

Convenience. Excluding removes a category of complexity: pregnancy testing, contraception requirements, follow-up of pregnancies, and the design questions that follow.

Inertia. The criterion is in the template, and removing it requires someone to raise it.

The last two are the ones worth attacking, because they are not about risk at all.

What is actually required

Regulation (EU) No 536/2014 addresses clinical trials on pregnant or breastfeeding women and permits them subject to conditions: that the trial has the potential to produce a direct benefit to the woman or the embryo, fetus or child after birth outweighing the risks and burdens involved, or, where there is no direct benefit, that the trial aims to contribute to results capable of benefiting pregnant or breastfeeding women or others in relation to reproduction, and that it entails minimal risk and minimal burden; and that research on women who are not pregnant or breastfeeding cannot provide equivalent results.

Read that carefully and the structure is the same as for other populations in whom research requires particular justification. It is not a prohibition. It is a set of conditions, and the framework contemplates that such trials happen.

The middle position

The choice is not between blanket exclusion and unrestricted inclusion.

Excluding from early phase and including later, once human safety data exists.

Including women who become pregnant during a trial, rather than withdrawing them automatically. Automatic withdrawal is common and loses exactly the follow-up data that would be most informative, so the better design offers continued follow-up even where treatment stops.

Dedicated pregnancy studies once a product is in wider use, designed for the question rather than as a subgroup.

Pregnancy registries with real follow-up, which are weaker than a trial and much better than nothing.

Including breastfeeding women where the question is milk transfer and the study can be designed to answer it with limited exposure.

In practice

A team drafting eligibility criteria for a trial in a chronic condition affecting many women of childbearing age includes the standard exclusion.

The medical lead asks what the evidence is for the product's use in pregnancy, given that the condition does not pause for pregnancy and women with it will become pregnant while taking something.

The answer is that there is none, and that if this product is approved, clinicians will face that question with no data.

The trial is not redesigned to include pregnant women, which would be a substantial change with its own justification requirements. What changes is smaller and real: women who become pregnant during the trial are no longer withdrawn from follow-up, a pregnancy outcome follow-up is added with its own consent, and the development plan gains a commitment to a dedicated study post-approval.

The trial's primary question is unaffected. The evidence gap for the next generation of women narrows slightly, because somebody asked why the criterion was there.

What this looks like when it goes wrong

A trial excludes pregnant women and requires contraception. A participant becomes pregnant in month seven.

The protocol requires immediate withdrawal. She is withdrawn, study treatment stops, and she exits the trial.

No follow-up is collected on the pregnancy outcome, because the protocol has no provision for following a withdrawn participant and no consent covering it.

She was exposed to the investigational product for seven weeks of an early pregnancy. That exposure happened and cannot be undone, and it is precisely the circumstance in which outcome information has value, for her and for every woman who is exposed in the same way later.

The protocol's design ensured that the exposure occurred and the information did not.

Key takeaways