The ethics of research on pregnant women: is maternal consent sufficient?
This is now a matter of debate, some medical ethicists argue that it should be the patient who determines whether equipoise exists, not the doctor.
In September 2003, two papers were published in this journal describing the renal biopsy findings characteristic of healthy pregnancy and pre-eclampsia and correlating these findings with serum cystatin C levels.1,2 The research subjects were adult volunteers and they were ‘thoroughly informed’. The research had been approved by the local research ethics committee (LREC) and was ‘in accordance with the Helsinki Declaration of 1975’. Despite this, the BJOG has been criticised for publishing unethical research.3–5 The criticisms of the papers are divided into three main areas:
- 1Should the studies have taken place at all, even though ‘informed consent’ was given by the volunteers?
- 2Were the participating women informed of the true frequency and potential severity of the complications that might follow renal biopsy?
- 3Should the BJOG have published the articles?
In order to establish answers to these questions, we give an account of the key ethical thinking that informs the principled conduct of medical research today and apply this reasoning to the controversial publications in this journal.1,2
In the western world, respect for individual autonomy (and the ‘right to information’) has become of the highest legal and ethical importance. As a result, the ‘informed consent’ of a volunteer has rapidly become the pre-eminent test for the ethics of a research project. There are two important problems with this. The first problem is that obtaining ‘informed consent’ is not an objective process. In scientific papers, it is usually described in the ‘methods’ section of the manuscript, which endows it with the appearance of scientific objectivity, but the communication of information is open both to subjective differences in emphasis and interpretation. Thus, one person might describe the risks of a research project in a completely different way to another, while both may believe they are giving sufficient information to the subject and therefore acting ethically. The statement ‘all participants were thoroughly informed’ can therefore be characterised as a subjective process, unwittingly presented as an objective one. The second problem is that the research subject's ‘informed consent’ alone can sometimes appear to confer ethical acceptability on a research study, even though it is generally agreed that while ‘obtaining consent is crucial [it is] not sufficient [in isolation] to justify research on human beings’.6 This second problem might also be framed as a question ‘If informed consent alone is insufficient to determine the ethics of research on human subjects, what are the other criteria that should be used?’ To answer this we turn to the most ‘authoritative set of rules for the protection of human subjects in medical research’7 which were formulated at the end of the second world war.
The trial of trials
In January 1947, during the Nuremberg trials, the lawyers charged with the defence of Nazi doctors who had conducted experiments on concentration camp prisoners, challenged the conduct of researchers in ‘civilised’ countries such as the UK, France and the USA.8 These countries, it was revealed, had also performed dangerous medical experiments on prisoners without their consent. This unexpected turn of events made it necessary for the Nuremberg court to define the conditions under which human experimentation was ethically permissible.8 The resulting ethical formulation became known as the Nuremberg Code and it is one of the foundation stones on which modern research ethics are based (e.g. ‘The Helsinki Declaration’, which modifies the code9). The primary requirements of the code were that: voluntary consent should always be given by the human subject; that the consent should be fully informed; that the experiment should be the only way in which the information could be gained; that the experiment should avoid all unnecessary physical and mental suffering; and that the degree of risk to be taken should never exceed that determined by the humanitarian importance of the problem.10 The overriding purpose of this ethical guidance is to protect the research subject's human rights.
Is all research the same?
In general, medical research can be divided into therapeutic and non-therapeutic subgroups. Therapeutic research is conducted with therapeutic intent and there is usually the potential for a therapeutic benefit to the research subject. It usually involves the comparison of one sort of treatment with another. Non-therapeutic medical research is conducted solely for scientific purposes, to answer a research question. For therapeutic research to be ethical there must be uncertainty as to the optimal treatment of a particular condition and the physician, who owes a ‘duty of care’ to the subject, must believe that there is (a kind of) clinical equipoise between the therapeutic options.a The informed subject must consent to their involvement in the study and be prepared to sacrifice their interests, at least slightly, in order to contribute to important scientific knowledge.11 This has been characterised as a harm–benefit assessment.12 Conversely, the research subject involved in non-therapeutic research receives no immediate or direct benefit. In this case, the altruistic subject agrees to take a risk in order to benefit science. It is harder to justify these studies. Such studies need to be the only way that the scientific information may be gained and the risks should be reasonable (those risks that a reasonable person might be expected to take) and proportionate to the knowledge gained.10 This sort of assessment has been described as a harm–knowledge assessment.12
Do ‘future people’ have rights?
The primary purpose of the Nuremberg Code was to protect the individual research subject's human rights and those subjects whose rights need most protection are the vulnerable. Historically, this group has included children, incapable adults, prisoners and pregnant women. While children, incapable adults and prisoners are vulnerable because they lack freedom or autonomy, what distinguishes pregnant women from the rest of the population is the prospect of causing harm to vulnerable ‘future people’ (their unborn children). Women do not lose capacity when they become pregnant and are not vulnerable per se. It is the incapacity of the vulnerable future person, developing within them, that makes the maternal–fetal unit a vulnerable entity. This appears at odds with current UK law, which allows women to smoke (causing growth restriction) or to request the killing of their fetus (through abortion). The reason for this difference is that while the ethical arguments surrounding the ‘rights’ of the fetus are complex, emotive and unresolved, the legal position is simple and pragmatic. In English law, the rights of a fetus cannot be realised until the fetus is born alive. Thus, the fetus is, in legal terms, a person-in-waiting13 and if it is aborted it never materialises as a legal entity. Also as a matter of pragmatic social policy the law14 disallows any claim by the child against its mother for antenatal injuries. While the mother may not be held liable in law, for damage caused by her conduct, the researcher is generally believed to have an ongoing moral responsibility for any harm done to a future person as a result of their experiments.
The maternal–fetal conflict
When judging the ethics of research on a vulnerable group, it is critical to determine whether the research is therapeutic or non-therapeutic. This distinction is particularly important when considering research in pregnancy, as there is always the potential for conflict between the interests of the mother and the person-in-waiting (the fetus has interests even though it cannot take an interest). Therapeutic research in pregnancy may be divided into that which might benefit the woman alone, the fetus alone or both the woman and the fetus. If the research potentially benefits both the woman and the fetus, it may be justified on the grounds that the potential therapeutic benefits of the study outweigh the possible risks. If the research might benefit the fetus but not the mother (e.g. the insertion of a therapeutic device, through the mother's body, by a fetal medicine expert), then, provided the competent mother freely accepts any potential risk, the study might be justified on the grounds that the potential therapeutic benefits outweigh the risks to the vulnerable fetus. If the therapeutic research might benefit the mother but carries a risk to the fetus, there is a potential for a conflict of interest. In some cases, it might be in the direct personal interest of a mother to be included in a therapeutic trial (if she has a rare and incurable disease for instance) and to deny her the right to participate in the experimental treatment because of the fetus within her, would not only infringe her autonomy but may limit her chances of survival.15 The greater the potential benefit to the pregnant woman, the greater the ethical consensus as to the acceptable risk to the fetus. However, when the benefits to a pregnant woman of a new treatment are potentially less substantial, how should the researcher proceed?
Primum non nocere
The effect of Thalidomide on the pharmaceutical industry cannot be under-estimated.16 The fear of successful litigation by future people, damaged by drugs taken by their mother, is understandable given the law and public sentiment. However, the self-imposed stringency of the limitations on research on pregnant women is not without its consequences. Information deficits are brought about by the exclusion of these women from research trials and this may have serious consequences.17 Pregnant women are often treated with drugs that have been superseded in every other branch of medicine (e.g. methyl dopa), because newer drugs have not been fully investigated in the pregnant population. Clinicians managing pregnant women are often forced to extrapolate from studies on men, which have specifically excluded pregnant women, before they prescribe drugs for an unlicensed use in pregnancy. This protects the pharmaceutical companies, but exposes the clinician to medico-legal risk and the pregnant woman to receive a drug that has not been tested in a relevant clinical trial. Furthermore, while participants in clinical studies should be protected by informed consent procedures and monitoring safeguards, these safeguards are not available to pregnant women who currently receive many drugs untested and not licensed for use in pregnancy. We think there needs to be a serious and ongoing debate about therapeutic research in the pregnant population and a consensus needs to be reached as to what levels of risk might be considered reasonable. We do not think the same debate is necessary for non-therapeutic research.
It is extremely difficult to justify non-therapeutic research on a pregnant woman. She might be willing to risk harm to herself in order to increase scientific knowledge, but should she impose a risk of harm on her fetus when there is no potential for therapeutic gain? Some might argue that to interfere with her choice represents a violation of a competent woman's autonomy and others that the minor infringement of maternal autonomy is justified, in order to protect the vulnerable person-in-waiting from a risk of future harm. We would argue that non-therapeutic research on a pregnant volunteer is justifiable when the competent woman gives her consent and there is no risk to the vulnerable person-in-waiting. In the present legal environment, this would include research on women planning to have an abortion. In this case the person-in-waiting does not materialise (in legal terms), thus the ‘risk of future harm’ cannot eventuate. It would also include research in which the scientific evidence suggests that there is only the most minimal risk of harm to the fetus. In this instance, minimal means comparable to the risks of daily living.18 When there is more than a minimal risk of harm to the fetus, we do not think non-therapeutic research is justifiable. Preventing women from participating in non-therapeutic research does not involve an invasion of their bodily integrity nor can it be described as a major infraction of their autonomy as nothing is being done to them and they are not being denied any potential therapeutic benefit.15 Risking damage to a vulnerable person-in-waiting however, for the benefit of other people, is, in our view, in direct contravention of the spirit of the Nuremberg Code.
What do we think are the risks of renal biopsy?
The renal biopsy papers in the BJOG1,2 involved non-therapeutic research on a vulnerable population. The arguments in favour of such research must be very strong and should answer, at the very least, the primary requirements of the Nuremberg Code. As originally published, it is difficult to comment on whether the research subjects were properly informed. For example, in the paper on serum cystatin C, a careful scientific case is made, over three pages, for the use of cystatin C as a marker of endotheliosis while the description of the process of consent and the ethics of the study is consigned to a single short paragraph. It is noted that women were ‘thoroughly informed of the risks of complication’ but what risks and what complications?
In the authors' reply to criticism,19 it is stated that the volunteers were informed that renal biopsy is a morbid procedure with a rate of renal haematoma or haematuria of 1/59–111.20,21 They also state that ‘All participants…agreed to receive blood transfusion if this should be necessary’, but they do not comment on the rate of major complications, including haemorrhage-requiring blood transfusion, nephrectomy or death; a maternal death following renal biopsy during pregnancy has been reported.22,23 The risks quoted by Strevens et al. are derived from series published in 1960 and 1987. In the series by Packham and Fairley21 there was one major renal haemorrhage (1/111) requiring 7 units of blood and resulting four days later in the spontaneous delivery of a 25-week-old baby that died after 48 hours. In an accompanying editorial, however, Lindheimer and Davison23 state that pregnancy adds little or no risk to the complications of renal biopsy.
These reports were from a time when renal biopsies were conducted with manually operated equipment and without ultrasound guidance and over the next 10–15 years the development of ultrasound-guided automated biopsy devices has reduced the rate of complications.24,25 The actual complication rate, however, depends on the experience of the operator, the size of the biopsy needle, inclusion of high risk patients and how thoroughly complications are sought. The lowest complication rate we could find was documented in a series that used an 18-gauge automated device with ultrasound guidance similar to that used by Strevens et al.26 In this series of 1090 renal biopsies there was only one major haemorrhage requiring interventional radiology and three cases requiring blood transfusion [i.e. four major complications/1090 (0.36%)]. It is not stated how many, if any, pregnant patients were included and all patients had a BP < 140/90 mmHg (raised BP was described as a contraindication to renal biopsy in this study), as hypertension significantly increases the risk of postbiopsy renal haemorrhage.27 In the studies by Strevens et al., all of the pre-eclampsia volunteers had a BP > 140/90 mmHg. Indeed, one of the pre-eclamptic volunteers (1/28) had a large perirenal haematoma requiring blood transfusion and then emergency caesarean section for a placental abruption.
Several series examining the complication rates of modern automated ultrasound-guided renal biopsy techniques have demonstrated a range of risk from 0.36% to 13% for major haemorrhage requiring blood transfusion, and lower risks of nephrectomy and death. The risk of minor complications, including macroscopic haematuria and perirenal haematoma, is between 15% and 40%.25,27–29 Renal biopsies are now rarely done during pregnancy, but a contemporary series of 15 antenatal renal biopsies documented seven postbiopsy haematomas, of which two required blood transfusion.30 As renal blood flow increases by 80% during healthy pregnancy, it is no surprise that the plethoric kidney should haemorrhage more following biopsy. From the data presented, we conservatively estimate that a renal biopsy during normotensive pregnancy will cause a minor complication (haematuria or haematoma) in approximately 20% of women and a major complication (haemorrhage requiring blood transfusion) in approximately 4% of women. These risks will be even higher for women with pre-eclampsia.
What should the research volunteer be told?
In the famous case of Rogers versus Whitaker (Rogers v. Whitaker 67 A.L.J.R. 47), the Australian High Court addressed the issue of informed consent. It set a very high standard for the level of information to be given to a patient prior to a therapeutic operation. The Judgement stated that ‘a risk is material if in the circumstances of the particular case, a reasonable person in the patient's position, if warned of the risk, would be likely to attach significance to it’. It is worth pointing out that Mrs Whitaker was suing Mr Rogers for his failure to mention a risk of 1 in 14,000 that she might become completely blind (the risk had unfortunately eventuated). We believe that the standard of information required to ‘thoroughly inform’ a voluntary participant in a non-therapeutic research project should be at least as high as this. Therefore, it is reasonable to expect researchers to mention the range of risks, in the published literature, associated with their particular project. It is for the volunteer, not the researcher, to determine what weight to give any particular risk. After all, a particular risk of death or haemorrhage may appear lower to a researcher than the volunteer!
Despite the fact that the research subjects ‘displayed great willingness to participate’, it is surprising (from a normative perspective) that they volunteered, if they had been informed of the published literature on renal biopsy complications and the particular risks associated with biopsies in the hypertensive and pregnant patient. Autonomous adults can however do surprising things and we accept that it may be argued that the research subjects were sufficiently informed. Setting aside our concerns about the standard of information given to the research volunteers, we have several further criticisms that do not turn on the issue of informed consent.
Were the renal biopsy papers ethical?
We make three critical points. The first is that it is far from clear that the ‘degree of risk taken was (proportionate to) the humanitarian importance of the problem to be solved by the experiment’ (Nuremberg Code). In other words, we do not believe that the risk of renal haematoma, nephrectomy or death, in a pregnant volunteer, was worth the potential discovery of a new test for the degree of renal involvement in pre-eclampsia. Furthermore, ‘in research on women, the interests of science and society should never take precedence over considerations related to the wellbeing of the subject’ (Helsinki declaration—1996 revision31) In their reply to the criticisms, Strevens et al. state that ‘the complication frequencies (of renal biopsy in the high risk pregnancy) are appalling and that renal biopsy is a morbid procedure. The whole point of (the) studies was to once and for all abolish this use of renal biopsy in (the high risk) pregnancy’.19 The authors appear to believe that the interests of science and society justified the complication frequencies of renal biopsy risked by their research subjects. This in our view runs counter to the Helsinki declaration. The third and final criticism is that the research was performed on a vulnerable (pregnant) population. We do not think that the minor and major risks to the mother of renal biopsy, described in this commentary, are comparable to the risks of daily living, to the fetus. We do not believe that the harm–knowledge test was passed and we do not think these studies should have taken place.
Should the BJOG have published these papers?
There is a form of ‘moral harm’ done to society when unethical research is knowingly published. Not publishing an unethical study may also be considered a form of ‘moral harm’ due to the potential loss of valuable scientific data. The Helsinki declaration is however clear, research known to be unethical should not be accepted for publication. In our view, the harm of publication lies not in publication per se but in the intention of the editor to knowingly publish and thereby condone unethical research. The ethics governing a controversial research project are unlikely to be black and white and generally fall along an ‘ethical spectrum’.32 The Researcher, LREC, Editor or Reader may differ in their opinion on where the research lies on this ‘ethical spectrum’. Thus, when, as in the work by Strevens et al., the editor is presented with research, which is ethically controversial and he is uncertain where along the ‘ethical spectrum’ this research lies, he should check the LREC approval and seek advice from an appropriate source (which was done in this case). If subsequent to publication good argument is mounted that in fact the study was unethical, then it may be argued that, in retrospect, the study should not have taken place and should not therefore have been published. This does not mean that the journal has acted improperly. In our view, the study should not have been conducted for the reasons given above, but if the editor believed prior to publication that the study, while controversial, was on balance ethically acceptable (following consultation), then we do not believe it was improper to publish it. The only proviso being that the ethical concerns about the paper are properly and publicly addressed.
Do ethics committees decide the ethics of a project?
Historically, the ‘local research ethics committee’ has usually determined the ethics of a research project and thus most scientific papers carry the declaration that ‘the study had ethics approval and the subjects were fully informed’. The problem with this is that ‘the evaluation of the conceptual and safety issues’ of research studies is technical and specialised and it has been argued that many LREC's are simply not set up to adequately protect the research subject.33 This is alarming as some medical researchers, who may only have the most rudimentary understanding of medical ethics, may use the LREC as the only ethical judge of their research project. Thus, the possibility exists that a researcher, in good faith, following scrutiny by their LREC (made up of unpaid volunteers, not necessarily experts in the field), may present an unethical paper for publication. At present, editors and referees of scientific medical journals tend to be scientists or doctors and they are well equipped to determine the scientific quality of a study. As a consequence, most scientific papers are arranged in order to defend the method of experiment and the statistics used to interpret results. Papers rarely have anything other than the most cursory discussion about the ethics of the project and so editors and referees generally (have little choice other than to) take the view that ‘if the REC has been properly constituted the paper should be published’.34 Thus, there is a potential for unethical research to be performed and published with the researcher believing they are acting ethically, the LREC acting with integrity, the research subjects acting altruistically and the editor publishing in good faith.
Can we do better?
We have two suggestions: the first is that research ethics committees should include specialist expertise and that members should be paid as specialist professionals whose job it is to assess the ethics of a study and protect research participants.33 The second is that journals should require authors to give a far more detailed account of the ethical issues pertinent to their research project32 and that the ruminations of the LREC should be made public. We would also suggest that in any discussion of risk, a numerical assessment is given of the range of risks in the published literature, which allows for an objective comparison of the risk and permits a much more open analysis of the subjective terms ‘minor’ or ‘reasonable’. These two recommendations would help to prevent unethical research from taking place, which must be better than suppressing the publication of data which has already been collected. They would also make the ethical deliberations of the researcher, the LREC and the editor more transparent and give them the column space they deserve. Given the present environment, a clear and transparent assessment of the ethical probity of controversial studies, in particular, and all research, in general, would offer protection to the reputation of the editor, the journal, the researcher and, most importantly, the human rights of research subjects.