Ethical issues in therapeutic use and research in pregnant and breastfeeding women

Pregnant or potentially pregnant women have historically been excluded from clinical trials of new medications. However, it is increasingly recognised that it is imperative to generate evidence from the population in whom the drugs are likely to be used to inform safe, evidence‐based shared clinical decision making. Reluctance by researchers and regulators to perform such studies often relates to concerns about risk, particularly to the foetus. However, this must be offset against the risk of untreated disease or using a drug in pregnancy where safety, efficacy and dosing information are not known. This review summarises the historical perspective, and the ethical and legal frameworks that inform the conduct of such research, then highlights examples of innovative practice that have enabled high quality, ethical research to proceed to inform the evidence‐based use of medication in pregnancy.


| INTRODUCTION
The risks to pregnant women of having no information upon which to make choices about their own treatment with medication have begun to outstrip concerns about the risks to them of participating in research. In this paper, we review the historical context and the ethical and legal frameworks for decisions relating to the inclusion of pregnant or potentially pregnant women in clinical trials, then illustrate examples of best practice drawn from a range of disciplines.
The thalidomide catastrophe of the 1950s clearly demonstrated that potential harms of medications in pregnancy are genuine. 1 However, this was a lesson badly learned, as it arose precisely in the wake of enormous harms arising from ill-advised research exclusion: widespread rollout and clinical use of thalidomide in pregnancy without ever having included pregnant women in the initial clinical trials. 2 The incidence of thalidomide-associated phocomelia is high (20-30%, and probably much higher if the exposure occurs during the critical window of 20-36 days postconception), which means by the Rule of Threes that an early-phase study would only need to have included 10-15 pregnant individuals to have detected this harm. 3 Contrasting this to the 10 000 infants born with thalidomide syndrome between 1957 and 1962, the possibility of staggering harms averted by research becomes clear. At the policy level, the global response to the thalidomide crisis, in addition to withdrawing the drug from the market, was to exclude all women of reproductive potential from pharmaceutical research; this has cast a long shadow and resulted in a lack of information about the safety, pharmacokinetics and effects of drugs in women more broadly, including inadequate information about instances where sex-specific differences exist (a recent effort to inventory sex-specific safety, efficacy or pharmacokinetic differences Tweetable Summary: Principles of justice, autonomy, beneficence and nonmaleficence underpin the ethical design of clinical trials in pregnant and lactating women. among drugs available in Sweden found that 15% of drugs have sexspecific differences in at least some populations, and 30% of drugs have inadequate data to determine whether sex-specific differences exist 4 ). Thoughtful advocacy over the past 3 decades has shifted the global focus somewhat from a paternalistic framework of protecting women from research to a perspective of protecting women through research. This approach maps to the established bioethical domains of justice, autonomy, beneficence and nonmaleficence ( Figure 1).

| EXTENT OF KNOWLEDGE DEFICIT ON DRUGS IN PREGNANCY AND LACTATION
A recent study including almost 10 000 pregnant women demonstrated that the majority of women in Europe, North America, South America and Australia used at least 1 medication during pregnancy. 5 In the USA, despite the fact that about 7 in 10 women take at least 1 prescription drug in pregnancy, 6  and 2019, Byrne and colleagues noted that animal pregnancy data were available in 90% of cases, but human pregnancy data in only 10%. 7 Notwithstanding the widespread use of prescribed or over the counter medication in pregnancy and breastfeeding, there is rarely evidence-based information to fully describe the dosing and safety of the medication to either mother or infant. The FDA guidance calls for such studies to be done around the time of licensing, when it is anticipated that the drug will be used by women of childbearing age, [8][9][10] but in reality this is rarely done. It is important to physicians and to patients to know whether the dose, efficacy and safety profile of a drug required in pregnancy are the same as in nonpregnant adults. Substantial physiological changes occur in pregnancy, impacting the absorption, distribution, metabolism and elimination of the drug. These processes, summarised as pharmacokinetics, may result in a lower exposure of drug in pregnancy and risk of therapeutic failure. 11

| HIV PHARMACOLOGY HAS LED THE WAY
The HIV research community has led the way in many aspects of pharmacological research in pregnant and breastfeeding women, from trial design, advocating for the earlier and routine inclusion of pregnant or potentially pregnant women in such studies, 12 providing guiding principles for the design and analysis of high quality pharmacokinetic studies 13,14 and providing an ethical framework on which to underpin this. 15 A 2016 systematic review of all pharmacokinetic studies in pregnancy identified 198 studies exploring 121 different medications. These studies encompassed a broad range of interventions for which pregnant women were successfully recruited, speaking to the potentially broad acceptability of trial participation among pregnant women. Of note, more than a quarter of these studies were on antiretrovirals. 16 Initially, antiretroviral drugs were given to mothers with HIV to protect the infant from up to a 30% risk of acquiring HIV in utero or peripartum, which carries a 50% mortality by age 2 years without treatment, 17,18 rather than to improve the mothers' health. Considerations of risk and benefit reflected the F I G U R E 1 Bioethical principles relating to research in pregnancy substantial risk to the infant of not treating the mother. Furthermore, there has been a strong voice from affected communities and strong partnerships and stakeholder relationships prioritising the generation of evidence to inform the public health approach to management of this significant public health challenge. The innovations from this field can be drawn upon to shape best practice for pharmacological research among pregnant and breastfeeding women with other medical conditions, who require drug therapy and face related evidence gaps, but may lack representation by advocacy or political mobilisation on their behalf.

| JUSTICE AND RISK-URGENCY CALIBRATION IN THE INCLUSION OF PREGNANT WOMEN IN DRUG DEVELOPMENT TRIALS
Justice is fairness in the distribution of benefits and burdens. 19 It is a fundamental normative standard of bioethics and as such it applies not only to health care and to public health as domains of action directly affecting individual and community health, but also and fully to the large supporting institutional structures and activities of biomedical research including drug development research. 20 Justice applied to drug development research requires comprehensively identifying and evaluating disparities in the ways that different groups of people, such as pregnant women, are systematically treated and impacted by particular policies, programmes and practices of drug development (such as particular research agendas and trial designs). Justice is a comparative assessment of these differences as they arise among alternative approaches or actions, 22 such as between the inclusion vs. exclusion of pregnant women in drug development trial designs. An assessment of the justice of a particular drug development agenda or trial design must, for instance, take stock of the differential risks for study-related harm that pregnant women or their foetuses stand to face from participation as opposed to the risks they face from nonparticipation. Equal inclusion of pregnant women in drug development trials may pose unequal risks of studyrelated harm to them or their foetuses compared to other groups or to the general population. This well-recognised point of assessment, important though it is, is not in itself dispositive of the justice of including or excluding pregnant women in drug development trials as there are other countervailing domains of trial design-attributable benefits and burdens relevant to the comprehensive assessment of justice in drug development research.
Another justice-relevant domain of disparities in benefits and burdens from drug development trial design is the differential time to evidence for given levels of safety and efficacy data for a drug in the general adult population as compared to in pregnant women. Exclusion of pregnant women from drug development trials-and not simply sporadic exclusion from a particular drug development trial but systemically consistent exclusion from a wide array of drug development trials in multiple areas of biomedical research-risks causing significant delays in achieving particular levels of safety and efficacy data on which to base optimal use of drugs in the medical care of pregnant women as compared to general adult population. 23 The burden of these delays for pregnant women is a function of the magnitude of the delays as well as the urgency of the medical problems addressed by the particular drug development trials. This urgency in turn is a function of the severity of the medical problems, their frequency in pregnant women, the availability, safety and efficacy of existing alternatives for managing those problems during pregnancy, and the level of evidence supporting the safety and efficacy of those putative alternatives in pregnancy. Significant trial design attributable delays in time to evidence for high urgency medical problems is a burden of substantial ethical concern as a matter of justice. They represent high impact shortfalls in achieving equality of evidence and comparably evidencebased medical care for pregnant women as compared to the general adult population.
Whilst some delays in time to evidence may be inevitable irrespective of trial design, exclusionary trial designs can significantly prolong that delay. Trial design attributable delay in evidence is, in and of itself, a disparity of great ethical relevance under the bioethical principle of justice. Evidence-based medicine is rightly predicated on the benefit that a greater quantity and quality of evidence has on medical decision making, rather than lesser; such benefits should be equitable as a matter of justice.
An easily overlooked point must first be made about the precise relevance to justice of comparatively differential study-related risks between pregnant women and their foetuses and the general population. This requires logically distinguishing between 2 contrasting ideas ( Figure 2). Idea 1: the idea of including pregnant women in a drug development trial to study a question that would be otherwise well answered by simply including members of the general population-the general safety and efficacy of the drug in the general human population. Idea 2: the very different idea of including pregnant women in a drug development trial to study something that cannot justifiably be assumed to be otherwise well studied simply by the inclusion of members of the general population-the precise safety, efficacy and optimal dosing in pregnant women themselves. In fact, the latter idea can and should be stated more strongly in many instances: it is not simply that we cannot justifiably assume that the safety, efficacy or optimal dosing of a drug are all equally well studied with or without the inclusion of pregnant women and their foetuses in a drug trial, but rather that we often have compelling scientific reasons to believe they will not be equally well studied without the inclusion of pregnant women and their foetuses. The 2 ideas are points along a continuum of thought between viewing pregnant women as commensurable with the general population and as incommensurable with the general population.
If the relevance to justice of differential study-related risk is framed within Idea 1, then the inclusion of pregnant women in a drug development trial would appear to be a gratuitously higher-risk approach to the very same research objective that could just as well and more safely be achieved by a trial that excludes pregnant women.
Trial designs excluding pregnant women would then appear to offer an ethically superior approach to studying drug safety and efficacy in the general human population, as they avoid the distribution of what is made to appear as needlessly excessive risk. That appearance is deceiving, however, because it arises from a misleading conceptual frame. Idea 1 is archaic and harkens to an earlier period in human subject research in which there existed an implicit idea that clinical trials were able to produce evidence and data about human beings as such,

| Equity of access to research
Moving from a commensurability paradigm to an incommensurability paradigm, similarity or difference between nonpregnant and pregnant individuals is a matter to be proven, on a drug-by-drug basis, and not assumed. Therefore, it is imperative that drugs are studied in the populations in whom they are to be used. Examples abound of harms resulting from inadequate study of drugs in certain populations, including inappropriate withholding of drugs. At a time when efavirenz was the most virologically effective available HIV drug, it was systematically withheld from women in the first trimester of pregnancy because of concerns about a preclinical toxicity signal in monkeys. Clinical research that followed in nonpregnant people found no toxicity but it was assumed that they were incommensurable with pregnant women; therefore the drug was withheld in pregnancy. Only a decade later, when registry data had accumulated and the studies were eventually done in pregnant women, did it become clear that there was a species difference in the observed toxicity; no embryotoxicity was observed among pregnant humans. 12,24 The potential benefits of research participation include later access to the fruits of a particular research endeavour: a safe, tolerable and effective drug. Systematic exclusion from research is unjust partly because it results in systematic exclusion from these benefits.

| AUTONOMY AND CONSIDERATIONS OF INFORMED CHOICE
Autonomy requires both the removal of barriers to an individual's decision making and providing the necessary support for that individual to make informed decisions. Movement away from paternalistic F I G U R E 2 Assessment of commensurability top-down models has led to an increasing emphasis on clinical shared decision-making with patients after providing them with relevant information that they need in order to decide. This type of inclusive shared decision-making is not possible in the absence of relevant evidence-based information on which to base shared decisions. Per recent guidance from the General Medical Council in the UK, needed information includes "the potential benefits, risks of harm, uncertainties about, and likelihood of success for each option." 25 When deciding about drugs for pregnant and lactating women, however, generally the most guidance physicians can provide their patients is to highlight the many unknowns at hand, caused by a dearth of research involving pregnant women, and a resulting information vacuum. Even if high-quality evidence does exist, some women may decline medication which has proven safe and effective, and it is within their autonomy to do so. This is not unique to pregnancy, as topically evidenced by widespread COVID vaccine hesitancy across populations. It is of high importance to seek out and understand the beliefs, perspectives and knowledge of pregnant and breastfeeding women around participation in clinical trials. Full understanding of concerns and barriers among patient groups requires a qualitative approach followed by targeted public engagement endeavours to provide relevant information in an accessible manner. These approaches are beyond the scope of this review. Regarding trial hesitancy, however, it is well worth noting that a recent trial of antiretroviral therapy in the third trimester of pregnancy was initially expected to take 18 months to enrol, but due to demand for participation, had completed enrolment by 6 months. 26 Autonomy is the opportunity and capability of individuals to make whereas its opponents argue that it fails to respect the autonomy of the pregnant woman, fails to recognise the diversity of family structures, is difficult to apply and may interfere with access to research which may be beneficial to the child. The US regulation on research with pregnant women requires paternal consent should the research be of potential benefit to the foetus alone and not to the mother. It carries the caveat that paternal consent is not required if there is prospect of benefit to the mother or if the father "is unable to consent because of unavailability, incompetency, or temporary incapacity or the pregnancy resulted from rape or incest", 27 and is similar to guidance from other countries such as Uganda. 28 An additional complexity in the context of HIV is that a woman may not have disclosed her HIV status to her partner and that to do so during pregnancy may carry a real threat of gender-based violence and abandonment; some studies have proceeded with paternal objection as an exclusion criterion rather than insisting on a requirement for paternal consent (NCT02245022 and NCT03249181). 26,29 These considerations are complex and nuanced and there will be variation between families and partnerships even within a community with a strong prevailing view on decision making and gender dynam- PHASES also considered the perspectives of researchers. Sixtytwo HIV researchers described the challenges including ethical concerns such as how to weigh risks and benefits in pregnancy, and legal concerns relating to interpretation of current regulations, concluding that advancing research in pregnancy will require clearer guidance regarding ethical and legal uncertainties. A requirement for ethically responsible, action-guiding recommendations presented in a userfriendly format was noted. 30 Such guidance relating to HIV-related research in pregnancy, was produced in late 2020. 15  increases. Particular biases include the over-interpretation of the risk of making a decision, without recognising that to avoid such a decision or to take no action may itself carry greater risk, a concept known as risk distortion. A good example of this is excluding pregnant women from COVID treatment and vaccine trials because of the perceived risks of their participation, even given the disproportionately higher risk they face of severe disease from COVID. 32 Furthermore, concerns of any perceived theoretical risk to the foetus may be given greater weight than actual threats to the health of the mother. 33 In lactation, this risk perception may be even more distorted: in high-income countries, a default option has been to advise a lactating woman that there are no data to support the safety of a drug in pregnancy and lactation, and that therefore switching to artificial feeding might be preferable.
However, this assumes that prevention or premature discontinuation of breastfeeding carries no risks, whereas the body of evidence supporting the myriad of benefits of breastfeeding to both mother and infant continues to grow. 34 A balanced consideration of risks and benefits for each proposed clinical trial must be conducted. Figure 5 considers 2 examples at opposite extremes of both risk and benefit. In Ebola virus disease, without treatment there has been a maternal case fatality rate of over 80% accompanied by pregnancy outcomes, which were universally grim with spontaneous abortion, stillbirth or neonatal death in all documented cases. 38 These risks, and potential lifesaving benefits, would justify use of a drug with an unknown safety profile in either the short or long term, or indeed a drug which was known to carry substantial risk. Despite this, women were initially excluded from the trials-a stance that has been described as being protected to death. 39   follow-up is common and adherence to daily oral pills is challenging. 43,44 Long-acting strategies have never been studied in pregnancy or lactation, so their pharmacokinetics, safety and efficacy in this period are unknown. A strong call to study these agents among pregnant and postpartum women has been supported in part by the high acceptance among women of long-acting injectable depot forms of contraceptives. 45

| WOMEN OF CHILDBEARING POTENTIAL
Concerns about potential risk mean that many studies, particularly those involving novel agents, require women to be using a form of contraception that the investigators consider adequate. This requirement may result in reluctance of women to participate. However, in the clinical realm, there are few drugs where contraception is insisted upon, primarily the known teratogens such as isotretinoin. 46 Once more, risk shifting occurs, due to not undertaking research as to the actual risks and benefits in pregnancy. 33 If long-acting antiretroviral therapy is rolled out among women of reproductive age, some will become pregnant while on the drugs (as 50% of pregnancies are unplanned 47 ). Accepting this as a potential risk from the outset, a clinical trial enables evaluation in a controlled context: women who become pregnant in the context of a study should be allowed to continue on the trial and on study drug, with all evaluations and appropriate monitoring of both the foetus and the mother-otherwise nothing will be learned to benefit others who follow. 12,13 As discussed by Fairlie and colleagues, rather than immediate withdrawal from a clinical trial, follow-up for safety endpoints should be standard irrespective of whether the participant remains on the investigational drug or is switched to standard of care. If she is in the second or third trimester, she could be invited for pharmacokinetic analysis to provide data on the drug disposition in pregnancy. 12 F I G U R E 5 Balance of risks and benefits to pregnant women of participation in research 8 | SHOULD PREGNANT WOMEN BE CONSIDERED SPECIAL OR VULNERABLE?: NONMALEFICENCE Different terms are used to described pregnant and lactating women in the context of research, including special and vulnerable. Special populations is a broad term referring to individuals who are often excluded from clinical trials; but is it right to consider 50% of a population who are undergoing a normal and natural physiological process to be special? Vulnerable is a term that is less frequently used in recent years. It could be argued that it is rather the lack of evidence to inform safe treatment that renders a pregnant or lactating woman to be vulnerable! Increasingly, populations that have been deemed vulnerable have advocated for themselves, calling for quite different policy of inclusion in research and access to research products, rather than the status quo: discriminatory systematic exclusion supposedly on their behalf. [48][49][50][51] This situation was seen with the antiretroviral dolutegravir.
Licensed in 2013, it was shown to reduce viral load far more rapidly than comparator regimens, 52 to have a more favourable drug-drug interaction profile, 53,54 to be suitable for co-formulation with other antiretrovirals and potentially to incur fewer adverse effects. 55 Botswana made an early decision to transition national policy to use dolutegravir-based regimens as first-line, knowing that pregnancies would occur. During this period, the Tsepamo birth surveillance study was active in the country, exploring the association between the current standard of care efavirenz-based regimen and birth defects; the protocol adapted to explore birth outcomes with dolutegravir. Other countries began to introduce dolutegravir among specific populations with a view to scale up. To inform international guidance, in 2018 the World Health Organization (WHO) requested an interim analysis of the Tsepamo data, which revealed the surprising finding of a significantly higher incidence of neural tube defects among the babies of mothers who had conceived whilst taking dolutegravir (4/426 exposed pregnancies, 0.9%). 56 This led to immediate international alerts, and many countries withdrew dolutegravir as an option for women considered to be of childbearing potential (often taken to be women with a uterus who had not been surgically sterilised, aged between 15 and 49 years). However, by this point, an appreciable proportion of women who had tolerated disabling adverse effects from the efavirenz-based regimens, 57 knowing that there was no alternative first-line option, had been switched to dolutegravir and experienced a substantially improved quality of life.
Anger was expressed by community advocacy groups that women were not engaged in the dialogue surrounding dolutegravir and the potential risks. A well-publicised protest took place at the International AIDS Society conference in Amsterdam in July 2018, and the Kigali Stakeholder Meeting, convened in April 2019 concluded that "it is critical to not just view a pregnant mother, or any woman of childbearing potential, as a vessel for a baby, but as an individual in her own right, who deserves access to the very best, evidence-based treatment available and the right to be adequately informed to make a choice that she feels is best for her." 58 Women wished acknowledgement of their ability to manage their reproductive health, including the choice to delay or not to undergo a pregnancy and stated that "blanket exclusions that deny women equitable access to this optimal HIV treatment are not warranted or justified". 58 Qualitative research revealed the tension between the desires of women for information and autonomous choice 59 with the significant concerns of the stakeholders to protect their populations from potential harm. 60 Detailed modelling of potential risks and benefits of widespread dolutegravir use in South Africa, should the increased risk be genuine, suggested that even despite such an increased risk, there would be significantly improved population health and the intervention would be cost-effective. 61 As further Botswanan women who had conceived on dolutegravir reached delivery, the signal for Vulnerable may have meant originally to refer more to the foetus than the pregnant woman. The vulnerability, however defined, should be understood to apply to the same demographic both inside and outside the clinical trial setting; only some of a vulnerable population may stand to participate in a given study but the other members of that vulnerable population may nevertheless stand to be affected in ethically relevant ways by the design and conduct of that study. Accounting for the impact of trial design on a vulnerable population is not just this looking beyond the borders of the studies themselves, although this is a first step towards their full ethical assessment, but also accounting for the cumulative magnitude of impacts within and beyond the trials-this means not just comparing the harms and benefits to representative member(s) of the vulnerable population inside and outside the trial, but also accounting for the numbers of individuals in the vulnerable population who stand to be affected by the study design, for how long they stand to be affected, and to what degree. If the vulnerable group beyond the confines of a trial is large or faces a significantly prolonged delay in therapeutics or preventives for a high morbidity or mortality illness from trial design that excludes that group then, the adverse impact of the trial design on that vulnerable group may be great indeed due to a design-attributable period of de facto clinical or public health neglect of the vulnerable population compared to other populations.

| LEGISLATION AND LEGAL ASPECTS
In the USA, the 21st Century Cures Act of 2016 (Section 2041) established the PRGLAC Task Force on research specific to pregnant and lactating women, which has issued guidance to the Secretary of Health and Human Services on how to improve the inclusion of pregnant and lactating women in research. 63 Specifically, the PRGLAC mandate was to develop "a plan to identify and address gaps in knowledge and research regarding safe and effective therapies for pregnant women and lactating women, including the development of such therapies." 64 A key recommendation was a call for reducing liability around the study of therapeutics in pregnancy and lactation.
An important consideration in the design and conduct of clinical studies that include pregnant women are the various national regulatory requirements that such studies must satisfy in order to secure state sponsored funding, obtain institutional review board approvals, and move forward to approval for market entry in particular countries. It is essential to remember that all of these guidelines lack an enforcement mechanism, but do express legislative sense that increased inclusion of unrepresented populations is beneficial, rather than harmful, and to be supported. The goal of PRGLAC was to identify opportunities to promote such inclusion through future regulatory and legislative action (e.g. changing liability law or legislative incentives for the inclusion of pregnant people, or requirement to explain and justify their exclusion).  13,68 Population pharmacokinetic approaches are a powerful tool to explore variability in drug exposure between individuals through analysis of sparse datasets 69 but are unable to correct for poor study design or lack of accurate dose and time information. Therefore, such studies represent a missed opportunity and may yield confusing or contradictory results. 70,71 In any situation where a participant provides consent for their data and samples to be used in research, the investigators have a responsibility to undertake this to the highest possible standards to maximise the understanding of the clinical question. Recognising some of these challenges, the WHO and the IMPAACT study group hosted a 2-day consultation in 2019 to define best practice for pharmacokinetic studies in pregnant and breastfeeding women with HIV, and published guidance on these aspects of study design, sample collection and data analysis. 13

| Statins for preeclampsia
Use of innovative study design and multi-disciplinary working between basic and clinical pharmacologists can increase mechanistic understanding, challenge previously unexplored assumptions and pave the way for safe, ethical conduct of clinical trials that transform practice. 72 This can be illustrated by the repurposing of pravastatin to treat pre-eclampsia. 73 Introduced in 1987 for the management of hypercholesterolaemia, the earliest member of the 3-hydroxy-3 methyl-glutaryl coenzyme-A reductase inhibitors (statins) class, lovastatin, was given the 1979 FDA category X for use in pregnancy: "studies in animals or humans have demonstrated fetal abnormalities and/or there is positive evidence of human fetal risks … and the risks involved in use of the drug in pregnancy women clearly outweigh potential benefits", under the assumption that there would be no benefit from using this class of drug in pregnancy. Additionally, there was a suggestion of congenital malformations with lipophilic statins in rodent models, 74 which was never revised to reassess risk with hydrophilic statins and was not borne out through pharmacoepidemiological studies 75 or analysis of pharmacovigilance registries. 76 It was increasingly recognised that the actions of statins are multifaceted rather than restricted to lipid lowering, and the reversal of angiogenic imbalance, endothelial dysfunction and reduction in inflammatory and oxidative stress are all mechanisms implicated in the development of pre-eclampsia which affects 3-5% of pregnancies. 77,78 Pravastatin, the most polar hydrophilic member of the statin class, was shown in animal models and in vitro studies to reverse some pathophysiological pathways associated with pre-eclampsia, and accumulating evidence 75,79 refuted the concerns about teratogenicity. Prior to large scale clinical trials, ex vivo placental models were used to further define potential foetal exposure, 80 and longer term neurodevelopmental studies were conducted in murine models. 81 Finally, phase I and II safety and pharmacokinetic studies were undertaken in high-risk pregnant women, defined as those with a history of pre-eclampsia resulting in a previous delivery before 34 weeks, 78

| Autoimmune disorders
Some conditions, such as chronic inflammatory systemic diseases, disproportionately affect women, with peak incidence during reproductive years. There is growing recognition that uncontrolled disease activity prior to conception and disease flares during pregnancy present the greatest risks to maternal and infant outcomes. Biologics, particularly those targeting the proinflammatory cytokine tumour necrosis factor are increasingly used, but there has been uncertainty about the risk-benefit ratio of these drugs in pregnancy.
A particular challenge in determining risk results from the fact that women with more severe disease, and therefore a higher pre-existing risk of adverse pregnancy outcomes, are more likely to be prescribed biologics. Robust methodology is required to disentangle the potential confounding resulting from disease severity in assessment of risk of adverse birth outcomes resulting from drug therapy. Given the exclusion of pregnant women from trials of these drugs, Tsao and colleagues systematically reviewed 24 observational studies that included pregnant women with exposure to biologics from 3 months prior to conception or during pregnancy and reported on birth outcomes. Almost 60 000 mother-infant pairs including over 5000 where biologic exposure occurred, were included, and initial unadjusted analysis suggested higher risk among the exposed group. After stratification for severity of underlying disease, which can itself be associated with adverse pregnancy outcomes and which might skew the clinical decision towards the prescription of a biologic, there was no significant association with congenital anomalies or adverse birth outcomes. 83 This work supports the proactive study of these agents in women of reproductive potential, and underpins the ethical imperative to use the highest quality methods to correctly interpret available data.

| Collaboration, data sharing and reusability
The ethical imperative to maximise the benefits of data obtained through the study of drugs in pregnancy informs data management. It is increasingly recognised that datasets should be made findable, accessible, interoperable and reusable to others (FAIR), 84 but increasing awareness of these principles has not yet transformed practice. 71 It is recognised that a cultural shift is required to move towards widespread adoption of the recommendations. Whilst the principles apply to all trials and pharmacokinetic studies, there is particular importance in transparency of data arising from populations who are rare or difficult to recruit and where the research is believed to have involved some degree of risk. It would be unethical to need to repeat a study because of lack of availability of prior data. Reusability of data is the ultimate goal of FAIR, and this is a major factor separating traditional data management from FAIR data stewardship. Reusability allows data to be repurposed for new user communities, for new needs and for new applications. Data in this sense can become more valuable to more people across a range of organisations. 84 Furthermore, gaps in reporting can lead to misinterpretation of study findings and lack of generalisability. In recognition that clinical pharmacokinetic studies are not held to structured reporting guidelines such as CONSORT for clinical trials and PRISMA for systematic reviews, tools such as the ClinPK checklist have been developed to support the transparent and complete reporting of such studies. Involving 68 stakeholders from nine countries, 4 rounds of a modified Delphi survey and a series of small virtual meetings were required to generate consensus for a 24-item checklist considered to be essential to the reporting of clinical pharmacokinetic studies. 85 The use of such tools should be encouraged and promoted. Further information sharing, particularly on safety events, can be facilitated by the creation of national and international registries such as the antiretroviral drugs in pregnancy registry, 86 the FDA pregnancy registries, 87  The truly vulnerable group should properly be conceptualised as extending well beyond those individuals who participate in a trial and risk its harms. Finally, the longer pregnancy-and lactation-specific knowledge is delayed, the more urgent it becomes to fill these knowledge gaps.