The ethics of randomised controlled trials from the perspectives of patients, the public, and healthcare professionals

Since the introduction of randomised controlled trials, professionals and lay people alike have worried over whether doing this sort of experiment in humans is ethical. It has been argued that participants may be called to sacrifice their own best interests for the benefit of future patients.¹ The scientific rationale for conducting a trial rests in collective equipoise, which means that the medical community as a whole is genuinely uncertain over which treatment is best. The key point, however, is that future patients benefit at no cost to participants, provided that participants are in personal equipoise and give informed consent on this basis. In these circumstances, the trial arms are an equally good bet prospectively.²

Summary points

• Doctors do not seem to take informed consent from competent patients as seriously as they should; ways in which practice might be improved need to be tested empirically

• Most doctors expressed willingness to enter their patients in trials even when the treatments offered were widely available but were not an equal bet prospectively; the fact that members of the public suspect this might undermine their confidence in trials

• Willingness to undergo randomisation drops as prospective participants are given more preliminary data and as they are made aware of any accumulating evidence of effectiveness

• A large number of participants, even in phase III trials, emerge from consultations expecting to benefit personally; self interest, rather than altruism, seems to be their motive for participating

Methods

To find out what patients, the general public, and healthcare professionals thought about trials, we undertook a review of the ethics of randomised controlled trials from these perspectives as part of a broader review relating to the ethics of designing and conducting clinical trials.¹ We searched BIDS, Medline, and Psychlit (for strategy see the BMJ website). There were 61 studies on attitudes to trials, 54 based on quantitative methods and eight based on qualitative ones (one study used both). Twenty studies made use of hypothetical trial scenarios. Seven studies focused on early trials (phases I and II). Our search also identified three reviews that included views on clinical trials,^3–5 but only one was a systematic review and even this study gave only limited information about the quality of the composite studies.⁴ In all three reviews, the number of studies cited was much lower than the number we found.

Abstraction of results and quality assessment for all studies were carried out independently by SE and JH. We have chosen to focus here on the issues of informed consent, doctor-patient relationship, what motivates participants, equipoise, and restricting new treatments to trials. Other issues are summarised elsewhere.¹ Quality assessment was difficult in many cases because insufficient information was given by the authors—perhaps itself an indicator of poor quality. The dimensions of methodological rigour used are listed in table 1. Our detailed findings are available in the form of a Health Technology Assessment monograph.¹ Here, we discuss the salient findings in narrative style. The results were often surprising and sometimes even shocking.

Table 1.

Quality of evidence checklist

Code	Dimension of methodology
A	External validity of study—that is, how clearly defined was the target population which the sample results may in theory be generalised? (+ acceptable, − flawed)
B	Sampling: how representative is the sample of the target population?
	1 Entire population (consenters and refusers) approached or random selection of eligible patients
	2 “Quota” sampling (deliberate selection from specific groups)
	3 Grab sampling
	4 Not reported
C	Response rate must be given and acceptable at 70% or above (+ acceptable, − flawed)
D	Interobserver reliability should be assessed where appropriate for questionnaires or interviews (particularly where open ended questionnaires are used) (+ acceptable, − flawed)
E	Questionnaire should be supplied in study or available from author (+ acceptable, − flawed)
F	The statistical method used to analyse data must be appropriate (+ acceptable, − flawed)

Informed consent

The concept that informed consent should always be obtained from competent patients was widely, though not universally, accepted.^6–13 Two of these studies (one as recent as 1994) found that up to one in five doctors regularly entered competent patients in trials without even obtaining informed consent.¹²,¹³

In three studies, trialists were asked how confident they were that participants had grasped the key issues. An astonishing 47% of responding doctors in a multinational study thought that few patients knew they were taking part in a controlled experiment, even though they had given written consent.¹³ Even more extreme results were obtained in the two remaining studies, in which more than three quarters of responding doctors thought their patients rarely understood all the information given to them.¹⁴,¹⁵ These three studies were all carried out in the 1980s and involved clinicians from North America and Europe.

Despite this, patients generally seemed content with the information they received. The experiences of patients who had actually taken part in clinical trials were audited in four studies, which reported that at least 80% of responding participants were satisfied that they had made an autonomous decision to take part.^16–19 Of course, there is no way of knowing whether they really had received or understood all that they would have wished.

Views of healthcare professionals on the amount of information required to make consent valid were sought in a further six studies. In three Spanish studies, over 90% of trialists agreed on a minimum dataset for all patients,^7–9 and, in another study, a substantial proportion (58%) of doctors from the United Kingdom and eastern Europe said that they tried to give all information that might be pertinent.¹² However, 83% of respondents thought that patients may be overloaded with information,⁶ and about half the trialists doubted their capacity to judge what information to give patients.¹³ These results may reflect a widespread concern that fully informed consent may cause anxiety²⁰ and may even be needlessly cruel.²¹

Expectations with respect to informed consent varied according to the type of experiment in question. Some responding doctors believed that informed consent was obtained more rigorously in phase I trials than in phases II or III studies,²² and in trials of supportive care, than in trials testing curative or palliative therapies.¹²

Effect on doctor-patient relationship

Since the time of Hippocrates, the doctor-patient relationship has been the cornerstone of medical practice. Five studies reported the views of doctors on the effect of offering trial entry on the doctor-patient relationship.⁶,^13–15,²³ Of these, all but one small study¹⁵ showed that there was considerable concern among doctors about the effect of discussing trial entry on the trust and wellbeing of patients.

Motivation

Self interest was more commonly given than altruism as a reason for participating in trials (table 2). This was so whether the study was based on real (n=13) or hypothetical (n=9) scenarios. Fourteen studies explicitly involved phase III trials, and half of these recorded self interest as strongly influential. This is strange as patients have nothing to gain prospectively, save a putative non-specific trial effect, given personal equipoise and freely available treatments. Few of those asked were likely to have been aware of the extra medical attention that may go hand in hand with participating in a trial.¹ The way in which this information was elicited varied strikingly across the studies; some researchers recorded both motivations, others recorded only one. Of these, some had forced a choice, others had not, and, in some cases it was not even clear what exactly the researchers had done. As a result, the studies are not strictly comparable, but our main criticism is that they gave only a weak clue as to what patients had understood of the questions and hence what they meant by their responses. The majority of studies were (post-licensing) phase III trials comparing treatments which would normally be available in routine practice. Did the respondents realise this? They may not even have appreciated that they could be allocated to a control arm, and there is some literature to suggest that they might not.³³

Table 2.

Frequency with which self interested motivations and altruism were expressed by respondents

Motivation	Frequency
Self interest
Real scenarios:	Over 70% in four studies18 34 35 36
	Between 30% and 55% in eight studies17 19 37 38 39 40 41 42
	Under 20% in one study43
Hypothetical scenarios:	Over 50% in two studies26 44
	Between 25% and 50% in five studies10 26 38 45 46
Altruism
Real scenarios:	Over 60% in 3 studies35 36 38
	Between 40% and 60% in four studies17 18 39 40
	Under 40% in three studies41 42 43
Hypothetical scenarios:	Over 65% in four studies10 27 44 46
	Between 20% and 60% in three studies26 38 45
	Under 15% in one study47

Personal equipoise

In regard to personal equipoise, there are two issues: firstly, are decisions ever so finely balanced that a situation of personal equipoise may exist, and, secondly, what does and should happen when trial treatments are freely available but when personal equipoise does not apply? Regarding the first, Alderson and colleagues reported, on the basis of a standardised questionnaire, that only 25% of doctors could envisage themselves being in personal equipoise and still fewer (18%) thought their patients could ever reach this state.¹⁰ By contrast, it is possible to elicit Bayesian prior probabilities from clinicians²⁴ and these may interact, at least in theory, with a particular patient’s values to produce equal expected utilities—that is, equipoise.

Clinicians were frequently prepared to enter their patients in trials of freely available treatments even though they preferred one arm to another. This issue would not be important if patients could be “fully” informed. Over half (53%) of doctors who preferred tamoxifen for early breast cancer were prepared to enter their patients in a placebo controlled trial of this potentially lifesaving treatment.¹⁰ Nearly three quarters (73%) of the responding surgeons thought a trial of hormone replacement therapy in patients with treated breast cancer would be ethical, even though only 28% were “uncertain” whether this treatment could provoke a recurrence of this hormone sensitive tumour (J Marsden et al, unpublished data). It is possible, however, that they thought equipoise could exist nevertheless, as some patients might be pre- pared to trade off some survival advantage against the relief of symptoms. Taylor and Kelner explored this issue in more detail and found that 36% of clinicians thought that it was appropriate to enter patients in trials despite thinking that the treatments were an unequal bet prospectively. Eighty seven per cent went so far as to say that they would enter more patients if they could dispense with informed consent altogether, which could mean that they put greater store by scientific knowledge than the participants’ welfare in phase III trials.²⁵ But would these doctors subject one of their children to randomisation in such circumstances—and if not, would they at least disclose their prior opinion on the effects of the treatments in question? Otherwise doctors may have been allowing scientific affiliations to affect their behaviour.

And patients suspected as much—the proportion of patients or members of the public who thought it likely that doctors put people in trials even if equipoise was not present ranged from 26% to 70% across two studies, both of which used hypothetical trial scenarios.²⁶,²⁷ Such variation might be explained in part by the populations sampled: an exclusively African-American population and the general public respectively.

Two studies examined the effect of preliminary information on willingness to participate in a hypothetical trial and found that fewer people were willing to enter a hypothetical trial when preliminary data indicated either an increasing difference in the effectiveness of the two treatments or an increasing statistical significance of that difference.²⁸ Consistent with these findings is the further observation that people given explicit data concerning the effects of hormone replacement therapy, along with the problems inherent in the interpretation of the data, were less willing to be randomised than those given less detail and told simply that the effects on the various outcomes were “uncertain” (JA Wragg et al, unpublished data). This is in keeping with the finding, based on comparative studies of different methods for obtaining consent to randomised controlled trials, that more information was associated with lower consent rates.¹

Restrictions on access

What did patients think about restricting potentially lifesaving treatments, where no “proved” treatment exists already, to trials for the sake of assessment? A large proportion (79%) of responding AIDS patients across two studies,²⁹,³⁰ and parents of very ill children in a further study,³¹ thought that such drugs should be available to patients outside the trial. Another study, brought people together by focus groups to discuss the scenario where the new treatment was available only within a trial. Even when the patients were not desperate, they preferred a three way choice—that is, treatment A, treatment B, or trial of A versus B (J Marsden et al, unpublished data). The preference for wide availability even before licensing has been taken up by Minogue, who argued that by offering potentially lifesaving treatments only within the confines of a trial, patient autonomy is infringed and it is desperation that compels them to “consent.”³² Logue, on the other hand, argued that although there are fewer options (either enter a trial or not), patient autonomy is not infringed, for limited choice is an inevitable part of everyday life.³³

Conclusion

The finding that so many people participate out of self interest needs exploring. Although the studies were of poor quality, especially regarding what exactly the respondents understood of the questions and hence what they meant by their answers, it is important to find out what well informed members of the public really think about trials and why they expect to benefit, if indeed they do. The finding that self interest features so strongly, even in phase III trials, is consistent with our other findings that patients do not always understand that they have been randomised to a control group,³¹ which was just what the doctors had anticipated.¹⁵

Given equipoise and freely available treatments, gain is not a realistic aim prospectively in late phase trials. Even if a trial effect is real,¹ this can hardly be used as an inducement to participate because to do so would violate the Declaration of Helsinki. Because patients are required to make informed choices, it is important that they both understand and accept that they stand neither to lose nor to gain in such cases. Rationally, the reason for such participation must be an altruistic one or else a way of resolving an otherwise difficult (or finely balanced) decision—this latter reason was picked up by respondents in Snowdon et al’s study.³¹ If patients expect to be personally worse off by being in a trial but nevertheless give their consent, they would be altruistic in a strong sense, meaning that they expect to sacrifice clinical benefit for the psychological satisfaction of helping others.

The other main theme is that a surprising number of doctors “owned up” to entering their patients when the treatments in the trial were probably otherwise freely available but when patients harboured a personal preference for one of the treatments. Worse, doctors seemed to have been aware that patients may not have fully understood what was going on. For many, informed consent seemed little more than a ritual. More research needs to be carried out to see precisely what factors motivate doctors to offer randomised controlled trials as an option in health care. It is interesting to speculate, though. Doctors may feel torn by their obligations to individual patients and to society at large, and they may even have got the impression, perhaps hyped up by advocates of evidence based medicine, that they are obliged to participate in research for the common good, even when they have a slight or moderate treatment preference. As a result, there may be a stark difference of opinion between ethicists and practitioners over what is ethically acceptable. This brings us back to the need for clear public discussion of ethics and the conduct of medical research.