The EU Clinical Trials Regulation permits clinical trials involving pregnant participants. Art. 33 of Regulation (EU) No 536/2014 sets out the conditions. On paper, the path from permission to enrollment appears structured and navigable. In practice, the conditions interact with ethics committee standards that vary by Member State, benefit-risk assessment methodologies the regulation frames but does not operationalize, and an insurance market across twenty-seven jurisdictions that lacks the data to price the risk. The result is a landscape where exclusion remains the default precisely because the regulation’s apparently clear framework does not resolve the questions that matter most to sponsors.
1. A Regulation That Permits What Practice Prevents
The CTR, which has applied since 31 January 2022, replaced Directive 2001/20/EC with a framework intended to harmonize clinical trial conduct across the EU.1Regulation (EU) No 536/2014 [2014] OJ L158/1 (CTR), applicable since 31 January 2022. For pregnant and breastfeeding participants, Art. 33 CTR establishes conditions that supplement the general requirements of Art. 28 CTR. The structure implies that inclusion is anticipated. A regulation does not define conditions for something it expects never to occur. The data tells a different story. According to figures cited by the EMA in its 4 June 2025 announcement of the ICH E21 draft guideline, fewer than 0.4% of all clinical trials submitted in the EU include pregnant people, and that figure falls to 0.1% for lactating individuals.2EMA, ‘New guideline on inclusion of pregnant and breastfeeding individuals in clinical trials’ (Press Release, 4 June 2025), citing CTIS data.
That aggregate figure conceals a sharp internal divide. Where a product is intended for use in pregnancy and the benefit to the woman or fetus is direct, trials that enrol pregnant participants do proceed: the maternal respiratory syncytial virus vaccine authorised in the EU in 2023 was licensed for administration during pregnancy on the strength of a phase III trial that enrolled pregnant participants directly.3Abrysvo (RSVpreF), EU marketing authorisation 23 August 2023 (maternal immunisation); the pivotal phase III MATISSE trial (Kampmann and others, 2023) enrolled pregnant participants. Multinational trial, cited as a therapeutic-class illustration. The deficit concentrates among therapeutics not designed for pregnancy, where the direct-benefit pathway is unavailable and only the harder no-benefit pathway of Art. 33 CTR remains.
The disparity between what the regulation permits and what sponsors do exposes a problem that is legal as much as it is practical. A sponsor who excludes pregnant participants avoids the complexity of Art. 33 CTR compliance, but that avoidance has its own consequences. Products reach the EU market without pregnancy-specific data. SmPCs lack the information prescribers need to make evidence-based decisions about use during pregnancy. The labeling gap transfers the risk from the clinical trial setting (where it would be monitored, documented, and subject to regulatory oversight) to the prescribing setting, where it is borne by the individual patient and the prescribing physician without the structured safety data that a trial would have generated. Whether that transfer of risk is itself a compliance problem (as labeling frameworks increasingly expect pregnancy-specific data rather than simply noting its absence) is a question the CTR does not address but that sponsors may need to consider.
2. Art. 33 CTR: The Conditions That Condition Everything
Art. 33 CTR sets out two alternative pathways for trials involving pregnant or breastfeeding women: a direct-benefit pathway, and a no-benefit pathway whose three cumulative conditions cover comparable effectiveness, wider population benefit, and minimal risk and burden to the woman and her embryo, fetus, or child after birth. The provision adds particular protection for the breastfeeding child and forbids financial inducements beyond expense compensation.4Art. 33 CTR, supplementing the general conditions of Art. 28; full enumeration in the footnote. Each condition is named; none is operationalized. The framework appears workable on its face. The difficulty lies in its application.
The “direct benefit” standard requires a benefit-risk assessment that the CTR does not operationalize. The regulation establishes the criterion but does not specify the methodology for evaluating whether a particular trial design satisfies it. That assessment falls to the ethics committee, which, under the CTR’s architecture, operates at the Member State level. Art. 10(3) CTR requires that when subjects are pregnant or breastfeeding women, “specific consideration shall be given to the assessment of the application for authorisation of a clinical trial on the basis of expertise in the relevant condition and the population represented by the subject concerned.”5Art. 10(3) CTR. The expertise requirement is implemented nationally, producing variation in what constitutes sufficient expertise, how that expertise is applied, and what threshold of evidence satisfies the benefit-risk assessment across different Member States.
A regulation that permits inclusion on conditions it does not operationalize has created a framework where exclusion is the path of least regulatory resistance.
The “minimal risk” threshold (applicable to trials without prospect of direct benefit) compounds the difficulty. The CTR does not define what “minimal risk” means in the context of fetal exposure, and the term carries different connotations across Member State regulatory traditions. For a multinational trial, the sponsor must satisfy the conditions in Art. 33 CTR as interpreted by every participating Member State. The Part I/Part II assessment structure in CTIS separates scientific assessment (Part I, coordinated by the reporting Member State) from ethical and national-level assessment (Part II, conducted independently by each concerned Member State). Pregnancy-specific risk assessment straddles both dimensions. A protocol that clears Part I’s scientific assessment may encounter Part II objections grounded in a particular Member State’s interpretation of what constitutes acceptable risk to the fetus, an interpretation the reporting Member State’s Part I assessment did not and could not control.
The practical consequence for sponsors is that satisfying Art. 33 CTR in one Member State does not guarantee satisfaction in another. A protocol approved by the ethics committee in Germany may face objections in France or Italy, not because the science or the risk profile differs, but because the institutional expectations applied to the same regulatory text differ. That variation is not a failure of the CTR; Art. 33 CTR deliberately leaves ethical assessment to national systems. But it creates a compliance landscape where the sponsor cannot design a single protocol with confidence that it will satisfy the Art. 33 CTR assessment in every intended jurisdiction without engaging each jurisdiction’s decision-makers in advance.
3. CTIS, Transparency, and the Visibility of Exclusion
The Clinical Trials Information System centralizes submission but not assessment standards. Sponsors submit through CTIS; Member States assess through their own institutional structures. The system’s transparency provisions (which require publication of trial data and documents, subject to deferral settings for commercially confidential information) have produced a consequence that may not have been anticipated: inclusion rates for pregnant participants are measurable, attributable, and increasingly scrutinized. The EMA’s citation of CTIS data in its 4 June 2025 announcement was not a neutral observation. It was a regulatory signal that the disparity between permission and practice has been noticed.6EMA (n 2); CTIS transparency rules revised, applicable since 18 June 2024. The under-representation of pregnant participants is, moreover, an express theme of the Accelerating Clinical Trials in the EU (ACT EU) initiative run jointly by the EMA, the Heads of Medicines Agencies, and the European Commission, which gives that signal an institutional home.
The structured data fields in CTIS create a record that extends beyond individual trial assessment. Over time, that record may enable regulators (or the Commission, or interested stakeholders) to identify patterns of unjustified exclusion across sponsors, therapeutic areas, or Member States. Whether such identification creates regulatory consequences is uncertain, as the CTR does not establish a duty to include pregnant participants, only conditions under which inclusion is permitted. But the distinction between “no duty to include” and “no consequence of excluding” may not hold indefinitely. The deferral settings for commercially confidential information protect proprietary data about the product and the trial design. The decision to exclude pregnant participants is not commercially confidential; it is a design choice that the transparency framework makes visible.
Sponsors based outside the EU (whether in Switzerland, the US, the UK, or elsewhere) submitting trials through CTIS for EU conduct face this transparency regime regardless of home-jurisdiction expectations, a dimension of the cross-border landscape examined in Insight 27 in the Swiss context. Whether a sponsor’s consistent pattern of excluding pregnant participants across its CTIS submissions creates a reputational or regulatory exposure distinct from the analysis of any individual trial is a question without a clear answer under the CTR framework as it stands.
4. Safety Reporting and Fetal Outcomes Across Borders
SUSAR reporting under Art. 42 CTR and annual safety reporting under Art. 43 CTR apply to all clinical trials, including those enrolling pregnant participants. Member State cooperation in the safety assessment of those reports is structured by Implementing Regulation (EU) 2022/20.7Art. 42–43 CTR; Commission Implementing Regulation (EU) 2022/20 [2022] OJ L5/14 (Member State cooperation in safety assessment). The framework for capturing fetal and neonatal outcomes across multiple Member States, however, raises coordination questions the regulation addresses only partially. EudraVigilance integration requires reporting of suspected unexpected serious adverse reactions, but determining whether a fetal outcome constitutes an adverse reaction to the investigational product, or reflects the underlying condition for which the pregnant participant was being treated, or represents a background event unrelated to either, requires clinical judgment that may differ across investigators and across Member States.
Annual Safety Reports must aggregate safety data in a format that permits regulatory assessment, but the CTR does not specify how pregnancy-specific endpoints should be captured, categorized, or reported within that aggregation. The Maternal and Fetal Adverse Event Terminology (MFAET) developed by Spencer and colleagues offers one framework for consistent reporting across trial sites.8RN Spencer and others, ‘Development of standard definitions and grading for Maternal and Fetal Adverse Event Terminology’ (2022) 42 Prenatal Diagnosis 15. Its adoption, however, remains voluntary. The relationship between these definitions and the CTR’s own reporting requirements is unspecified. A sponsor that adopts the framework as its primary reporting standard for a pregnancy trial may find it insufficient for a particular competent authority’s expectations; a sponsor that develops its own approach may face challenges demonstrating consistency and comparability across participating Member States.
The post-authorization dimension amplifies the complexity. Pregnancy-specific data generated during a trial must ultimately be reflected in the product’s SmPC. The SmPC labeling framework expects data where data exists, but the pathway from trial-generated safety data to SmPC language involves regulatory assessments at the national and EMA level that may apply different standards to the same underlying evidence, depending on the therapeutic area, the product’s benefit-risk profile, and the agency’s own interpretive approach to pregnancy-related data.
A further layer sits beneath the safety data itself. A pregnancy trial processes special-category health data not only about the woman but, through fetal and neonatal monitoring and long-term follow-up, about a prospective second data subject who cannot consent and who, at enrollment, does not yet exist. The interplay between the CTR and the GDPR is settled in outline, with reliability and safety processing resting on the legal obligations in Art. 6 and Art. 9 GDPR and research processing subject to the Art. 89 safeguards.9Art. 9 GDPR and Recital 159; EDPB Opinion 3/2019 on the interplay between the CTR and the GDPR (23 January 2019). Art. 9(4) GDPR lets Member States add conditions for processing health data. But Art. 9(4) GDPR allows each Member State to maintain or introduce further conditions for the processing of health and genetic data, so the dyadic, multi-year data relationship a pregnancy trial creates is filtered through the same twenty-seven national overlays that fragment the ethics and liability analysis, rather than a single harmonized rule.
5. Liability Across Twenty-Seven Member States
Art. 76 CTR addresses damage to subjects in a clinical trial by requiring that “Member States shall ensure that systems for compensation for any damage suffered by a subject resulting from participation in a clinical trial conducted on their territory are in place in the form of insurance, a guarantee, or a similar arrangement that is equivalent as regards its purpose and which is appropriate to the nature and the extent of the risk.”10Art. 76(1) CTR. Art. 76(3) carves out low-intervention clinical trials where the applicable compensation system already in place covers the risk; that carve-out rarely engages in pregnancy trials of novel investigational products. The provision establishes the principle; implementation is national. The low-intervention carve-out in Art. 76(3) CTR rarely engages in pregnancy trials of novel investigational products. For a multinational pregnancy trial, this produces a patchwork of liability standards, insurance requirements, judicial approaches to causation, and damage calculation methodologies, each applied independently to the same clinical event depending on where it occurred.
The insurance challenge mirrors the structural problem encountered in Switzerland (examined in detail in Insight 27) but multiplied across jurisdictions. A sponsor may find adequate coverage in one Member State but not another, or coverage in both but on terms that are mutually inconsistent. The actuarial data that would allow insurers to price pregnancy trial risk with confidence does not exist in sufficient depth for any single jurisdiction, let alone for a portfolio of jurisdictions with different liability regimes. The Concept Foundation’s estimate of USD 5.7 million in additional development cost per pregnancy-specific medicine reflects the financial burden, but the legal complexity is not a cost that can be simply aggregated.11Concept Foundation, AIM market analysis (Geneva). Each Member State’s liability regime must be independently assessed, independently insured, and independently addressed in the clinical trial agreement, and the contractual allocation cannot override the mandatory national regime that applies when a claim is brought.
The temporal dimension of pregnancy trial liability compounds the problem further. Fetal or neonatal adverse outcomes may not manifest until years after the trial concludes. The applicable liability regime is that of the Member State where the trial was conducted, but that regime may itself have changed in the intervening years. Whether a clinical trial agreement’s indemnification provisions, drafted to reflect the liability landscape at the time the trial commenced, remain adequate under a reformed regime is a question the agreement itself may not answer and that the CTR does not address.
6. ICH E21 and the CTR: Supplementation or Transformation?
The EMA’s CHMP adopted the ICH E21 draft at Step 2b on 22 May 2025; the EMA then released the draft for public consultation on 4 June 2025, characterizing it as a “change in paradigm”.12EMA (n 2); ICH E21 Step 2 endorsed 14 May 2025; CHMP Step 2b 22 May 2025; consultation closed 15 September 2025; finalization anticipated Q1 2028. The guideline’s scope is expansive: it recommends that inclusion of pregnant and breastfeeding people in clinical trials be considered for all medicines intended for populations that include people who can potentially give birth. The word “considered” carries weight. It does not mandate inclusion. It establishes an expectation that exclusion be a deliberate, justified decision rather than a default.
The relationship between ICH E21 and the CTR is one of supplementation, not amendment. The guideline, once finalized (anticipated Q1 2028), will represent harmonized scientific and regulatory expectations, but it cannot modify the conditions in Art. 33 CTR, the expertise requirements in Art. 10(3) CTR, or the national liability framework in Art. 76 CTR. This creates a specific category of complexity. Where ICH E21 expectations exceed what Art. 33 CTR requires, a sponsor designing a protocol to E21 standards may satisfy the guideline but still face Art. 33 CTR objections from a Member State whose ethics committee applies a more restrictive interpretation. Where E21 presupposes a harmonized approach to benefit-risk assessment, the CTR’s Member State-level assessment structure may produce divergent outcomes that the guideline’s harmonizing intent cannot prevent.
Supplementation also runs in the other direction. ICH E21 does not arrive into a vacuum: the EU has regulated pregnancy data for two decades, but on the back end rather than the front. Two CHMP guidelines, one on the need for post-authorisation data on exposure during pregnancy and the other on risk assessment from data to labelling, already govern how pregnancy information is collected after authorisation and translated into the SmPC.13EMA, ‘Guideline on the exposure to medicinal products during pregnancy: need for post-authorisation data’ (EMEA/CHMP/313666/2005, effective 2006); EMA, ‘Guideline on risk assessment of medicinal products on human reproduction and lactation: from data to labelling’ (EMEA/CHMP/203927/2005, effective 2009). What that regime presupposes, data that pre-authorisation trials systematically fail to generate, is precisely what E21 addresses. Read together, E21 is best understood not as a new departure but as the missing front-end complement to a long-standing back-end: rules for what to do with pregnancy data once it exists have been in place for years; rules for generating it during a trial have not.
The period between E21’s finalization and its integration into EU regulatory practice will itself generate uncertainty. The guideline will be adopted by the EMA and reflected in updated Q&A documents, assessment procedures, and potentially revised CTIS guidance. Whether Member State ethics committees will independently update their assessment approaches to reflect E21, and on what timeline, depends on national institutional processes that the EMA cannot direct. A sponsor submitting a trial through CTIS in 2028 or 2029 may encounter a reporting Member State that has fully integrated E21 expectations and concerned Member States that have not, or the reverse. The compliance question is not whether to follow Art. 33 CTR or ICH E21, but how to satisfy both simultaneously when each operates through different institutional actors and may produce divergent outcomes for the same protocol.
For the jurisdictional comparison with the US regulatory landscape (where a different institutional architecture produces its own variant of the tension between policy ambition and operational mechanism) see Insight 29 in this series.