The United States has produced more policy on the inclusion of pregnant participants in clinical trials than any other jurisdiction. A dedicated congressionally mandated task force, multiple FDA draft guidance documents, a labeling rule that replaced letter categories with a demand for evidence, and the publication for comment of the ICH E21 draft guideline all point in the same direction. Sponsors rarely ask whether the binding rules have kept pace with that policy, or whether the gap between the two creates risks of its own for those who do what the policy encourages.
1. Policy Direction Without Operational Clarity
The policy infrastructure is substantial. The 21st Century Cures Act, signed into law on 13 December 2016, directed the Secretary of Health and Human Services under Section 2041 of the Act to establish the Task Force on Research Specific to Pregnant Women and Lactating Women (PRGLAC) to advise the Secretary on gaps in knowledge and research regarding safe and effective therapies for pregnant women and lactating women.121st Century Cures Act, Pub L No 114-255 (2016) s 2041. PRGLAC delivered its report in September 2018, containing fifteen recommendations spanning preclinical development through post-market surveillance.2PRGLAC, ‘Report to the Secretary of Health and Human Services’ (September 2018). The FDA has issued draft guidance on scientific and ethical considerations for inclusion (9 April 2018), on clinical lactation studies (9 May 2019), on postapproval pregnancy safety studies (9 May 2019), and final guidance on enhancing diversity of clinical trial populations (9 November 2020). ICH endorsed the E21 draft guideline at Step 2 of the ICH process on 14 May 2025; on 21 July 2025, the FDA published its notice of availability for that draft in the Federal Register (Docket FDA-2025-D-1797), with a comment period closing on 19 September 2025.3ICH, ‘E21 Guideline on Inclusion of Pregnant and Breastfeeding Individuals in Clinical Trials’ (Draft, Step 2, endorsed 14 May 2025); FDA Federal Register notice of availability, 21 July 2025, Docket FDA-2025-D-1797.
The direction is unambiguous; the operational rules are not. A sponsor designing a global trial for FDA submission works in a US system where policy encourages inclusion but the rules that govern trial design, ethics review, and participant protection have not been systematically updated to support it. Compliance decisions are made in that gap, and their consequences remain undefined.
2. The Dual Framework: FDA Regulations and HHS Subpart B
The United States splits regulatory authority across institutions in ways that create complexity absent in other jurisdictions. FDA clinical trial requirements under 21 CFR Parts 50 and 56 govern human subjects protection and IRB oversight for trials intended to support marketing applications.421 CFR Part 50 (Protection of Human Subjects); 21 CFR Part 56 (Institutional Review Boards). Separately, HHS regulations under 45 CFR 46, Subpart B provide additional protections for pregnant women, human fetuses, and neonates involved in research, imposing conditions that include prior nonclinical studies (extending to pregnant animal models where scientifically appropriate) and a prospect-of-direct-benefit analysis that determines the permissible risk level.545 CFR 46, Subpart B (Additional Protections for Pregnant Women, Human Fetuses and Neonates Involved in Research).
The interaction between these frameworks is not straightforward. FDA’s human subjects protection regulations do not contain provisions analogous to Subpart B. But the FDA’s April 2018 draft guidance recommends that sponsors comply with Subpart B’s requirements regardless of whether the research is conducted or supported by HHS, which is the jurisdictional trigger for Subpart B’s mandatory application.6FDA, ‘Pregnant Women: Scientific and Ethical Considerations for Inclusion in Clinical Trials’ (Draft Guidance, April 2018). A guidance document recommending compliance with a regulation that would not otherwise apply creates an intermediate category, neither mandatory nor irrelevant, in which the sponsor must decide without the clarity that either a binding rule or plain inapplicability would give.
US policy says include. The regulations say: where there is no prospect of direct benefit, keep the fetal risk within a definition of minimal risk that says nothing about the fetus.
The 2018 revision of the Common Rule removed pregnant women from the 45 CFR 46.111(b) list of populations ‘vulnerable to coercion or undue influence’ that triggers heightened IRB scrutiny, a change that signaled a conceptual shift away from categorizing pregnancy as inherently rendering subjects vulnerable.7Federal Policy for the Protection of Human Subjects (Final Rule, 82 FR 7149, 19 January 2017), general compliance date 21 January 2019 (83 FR 28497, 19 June 2018). Subpart B’s additional protections, however, remain in force. PRGLAC’s 2018 report recommended modifying Subpart B to remove provisions that function as unnecessary barriers to research participation, in particular by amending 45 CFR 46.204(e) to require maternal consent alone. That recommendation remains unimplemented. As a result, the overarching framework has moved in one direction while the protections specific to pregnant participants have not moved at all, leaving a gap between the Common Rule’s updated conceptual approach and Subpart B’s unreformed operational requirements.
3. FDA Guidance: Four Documents Without a Common Framework
Across the four guidance documents it issued between 2018 and 2020, the FDA articulates a recommended approach to trial design, lactation pharmacokinetics, post-approval safety surveillance, and inclusion across demographic groups (pregnancy being one part of a broader diversity agenda). Each document is coherent on its own, but the four have not been consolidated. The force of these recommendations depends on how one understands the FDA’s guidance framework as a whole.
FDA guidance represents the agency’s thinking at the time of issuance and creates expectations without establishing legally enforceable requirements. The ICH E21 draft guideline, when finalized and adopted, is likely to occupy the same space: a representation of how the FDA believes sponsors should approach the question, operating alongside but not amending the underlying regulatory text. A sponsor that follows all applicable guidance may still face IRB objections grounded in provisions the guidance does not override; one that departs from guidance would need to justify the departure without clear criteria for what constitutes sufficient justification, and without certainty about whether that justification might be assessed at the IRB level, the FDA review level, or both.
Because the guidance documents overlap, non-US sponsors also have to determine which of them apply to which decisions, and how FDA reviewers might weigh compliance with one document against non-compliance with another. The 2018 draft guidance and the ICH E21 draft guideline address overlapping questions about trial design, benefit-risk assessment, and preclinical data requirements. Where those recommendations diverge (even in emphasis rather than substance), the sponsor may face a judgment call about which representation of FDA thinking controls. The absence of a single, consolidated framework means that the sponsor’s regulatory strategy for a pregnancy trial would typically need to synthesize recommendations from documents issued years apart, in different regulatory contexts, addressing slightly different versions of the same underlying questions.
The Pregnancy and Lactation Labeling Rule (PLLR), finalized on 4 December 2014 and effective for new applications from 30 June 2015, replaced the prior A/B/C/D/X pregnancy category system with a narrative labeling framework.8Content and Format of Labeling for Human Prescription Drug and Biological Products; Requirements for Pregnancy and Lactation Labeling (Final Rule, 79 FR 72064, 4 December 2014). The PLLR requires data where data exists and descriptive information about the absence of data where it does not. The rule’s practical effect is to make the evidence gap visible. A labeling section that states there are no adequate and well-controlled studies in pregnant women (the formulaic language that previously sufficed) sits alongside a narrative framework designed to prompt the question of why that gap exists and whether it can be closed. Whether the PLLR’s transparency function creates a regulatory pressure toward conducting pregnancy trials, or merely documents the absence of data without consequence, depends on how regulators, prescribers, and litigators interpret the resulting label over time.
4. IRB Review and Informed Consent Requirements
On IRB composition and expertise for trials involving pregnant participants, the ICH E21 draft guideline recommends that consideration be given to using review bodies experienced in working with pregnant and breastfeeding participants. That recommendation, however, has no binding force under US regulations. The requirements of 21 CFR 56.107 address IRB composition in general terms: diversity of members, inclusion of at least one scientist and one non-scientist, and sensitivity to community attitudes. Where an IRB regularly reviews research involving a vulnerable category of subjects, the regulation directs that consideration shall be given to including one or more individuals knowledgeable about and experienced in working with those subjects; it stops short of mandating maternal-fetal medicine expertise on the panel for any particular trial population.921 CFR 56.107 (IRB Membership). An IRB may or may not have access to maternal-fetal medicine expertise when reviewing a pregnancy trial protocol. Whether its approval is adequate without that expertise is a question the regulation does not address and that ICH E21 will recommend but cannot require.
The practical consequence is that a protocol may receive IRB approval at one institution and face objections at another, with the inconsistency traceable not to regulatory disagreement but to differences in institutional expertise and risk appetite. For a multi-site trial, this may produce a coordination problem that the centralized IRB model (increasingly common for multi-site trials since the 2018 Common Rule revisions) might mitigate but is unlikely to fully eliminate. A central IRB’s approval binds all participating sites for cooperative research subject to the Common Rule, but the relationship between centralized review and site-level implementation for pregnancy-specific protocols raises questions about whether a central IRB’s risk assessment adequately accounts for the clinical capabilities and monitoring infrastructure at each participating site.
The informed consent framework under Subpart B introduces requirements beyond the standard consent elements of 21 CFR 50.25. Where a trial involves pregnant participants and holds the prospect of direct benefit solely to the fetus, Subpart B requires the consent of both the pregnant woman and the father, except that the father’s consent need not be obtained if he is unable to consent because of unavailability, incompetence, or temporary incapacity, or where the pregnancy resulted from rape or incest.1045 CFR 46.204(e) (Subpart B). The scope and practical application of this requirement generate questions that the regulatory text does not resolve and that few clinical trial agreements address with the specificity the situation demands. Three in particular recur: how Subpart B interacts with evolving reproductive-rights jurisprudence across US states; what “unable to consent because of unavailability” means in operational terms; and what documentation obligations the consent rule creates for site staff.
5. The Compensation Gap: Liability Without a Federal Framework
Unlike the EU’s Art. 76 CTR requirement that Member States ensure compensation mechanisms, and unlike Switzerland’s HFG-mandated insurance regime, the United States has no federal compensation scheme for clinical trial-related injuries.11Art. 76 CTR; Art. 20 HFG (SR 810.30); Art. 13 KlinV (SR 810.305). PRGLAC’s 2018 report identified this gap and recommended reducing liability barriers to facilitate the evidence base for products used in pregnancy and lactation, expressly invoking the Vaccine Injury Compensation Program as a model whose liability protection might extend whether or not a product reaches market approval.12PRGLAC (n 2), Recommendation 7. The recommendation remains unimplemented. Sponsors rely on indemnification provisions in clinical trial agreements and commercial insurance, against a litigation backdrop shaped by the learned intermediary doctrine, state-level tort law, and the unpredictability of jury determinations on causation and damages in cases involving fetal or neonatal outcomes.
In a pregnancy trial, liability extends beyond the enrolled participant to fetal and neonatal outcomes, with claims that may have long latency periods and damages calculated on the claimant’s life expectancy rather than the participant’s. The contractual allocation between sponsor, CRO, and clinical site may distribute financial responsibility, but it might not allocate the regulatory consequence of a serious adverse event, and is unlikely to control the litigation strategy of a plaintiff’s attorney whose claims may be brought in a jurisdiction selected for its favorable jury pool and damages framework rather than for its connection to the clinical trial.
The insurance problem has the same structure as in Switzerland and across the EU, covered in the articles on Swiss trials with pregnant participants and EU trials with pregnant participants, but it arises in a litigation culture that raises the financial risk. Damages awards for neonatal injury may be orders of magnitude larger than the actuarial models, built on adult-participant trial data, were designed to accommodate.
The structural consequence is circular and familiar from the Swiss and EU contexts, but with a US-specific inflection. Without trials, there is no data. Without data, insurers cannot price coverage. Without affordable coverage, sponsors decline to conduct trials. The policy infrastructure (PRGLAC recommendations, FDA guidance, ICH E21 published for comment) addresses the first element of the circle by encouraging sponsors to conduct trials. Apart from PRGLAC’s unimplemented Recommendation 7, it does not address the insurance market failure or the liability framework that sustains the circle’s logic. A sponsor that responds to the policy encouragement by designing a pregnancy trial would typically need to solve the insurance and liability problems independently, in a jurisdiction where liability depends on state-level tort regimes, jury trial rights, and damages methodologies that can vary materially within a single state. The clinical trial agreement can allocate risk between parties. It is unlikely to determine the quantum of that risk, the forum in which it might be adjudicated, or the standards by which causation between investigational product exposure and neonatal outcome would be assessed.
6. Between Encouragement and Implementation
The ICH E21 draft that the FDA published for comment on 21 July 2025 will, once finalized at Step 4 of the ICH process (anticipated Q1 2028), represent the agency’s thinking on how sponsors should approach the inclusion of pregnant and breastfeeding participants.13ICH E21 (n 3); finalization as a Step 4 document anticipated Q1 2028. It will not modify Subpart B. It will not reform the IRB composition requirements of 21 CFR 56.107. It will not establish the federal compensation framework that PRGLAC recommended. It will not bind IRBs that reach different conclusions applying the same standards to the same protocol. The guideline provides a framework for what sponsors should do. It does not resolve the structural impediments to doing it.
The period between finalization and operational integration is likely to test the distance between the policy infrastructure and the regulatory machinery. A sponsor who designs a pregnancy trial to ICH E21 standards may find the protocol satisfies the guideline’s expectations but encounters IRB objections grounded in Subpart B’s unreformed provisions, or insurance obstacles rooted in a market that has not adjusted to the policy direction, or site-level resistance from investigators whose professional liability coverage does not extend to the protocols the guidance envisions.
For non-US sponsors, in particular those designing trials from Switzerland for global regulatory submission, the United States combines some of the strongest policy encouragement anywhere with the least harmonized operational rules. Set beside the Swiss and EU rules, the same protocol design question produces three different regulatory answers at certain points.