There is a good deal of biomedical research that does not produce scientifically useful data because it fails to recruit a sufficient number of subjects. This fact is typically not disclosed to prospective subjects. In general, the guidance about consent concerns the information required to make intelligent self-interested decisions and ignores some of the information required for intelligent altruistic decisions. Bioethics has worried about the ‘therapeutic misconception’, but has ignored the ‘completion misconception’. This article argues that, other things being equal, prospective subjects should be informed about the possibility of non-completion as part of the standard consent process if (1) it is or should be anticipatable that there is a non-trivial possibility of non-completion and (2) that information is likely to be relevant to a prospective subject's decision to consent. The article then considers several objections to the argument, including the objection that disclosing non-completion information would make recruitment even more difficult.
- Informed Consent
- Research Ethics
- Clinical Ethics
Statistics from Altmetric.com
I begin with a personal story. In October 2008, a purely incidental finding from my annual physical—a high white blood cell count—led to further testing and my being diagnosed with chronic lymphocytic leukaemia (CLL). Many CLL patients remain asymptomatic for an extended period. Some develop symptoms more quickly. My haematologist at Fletcher Allen Health Care in Burlington, Vermont (the hospital is associated with the College of Medicine at the University of Vermont) invited me to participate in a multisite Phase III clinical trial that was sponsored by the National Cancer Institute and the Cancer and Leukaemia Group B (CALGB 10501).1
In the 1980s, researchers studied whether the drug chlorambucil (a standard CLL treatment) when given to asymptomatic CLL patients would help them live longer. The studies showed that those who received early treatment did not live longer than those who received treatment only after their CLL had worsened. Two things have changed since the 1980s. First, current treatments for CLL are more effective. Second, new blood tests can more accurately predict the subset of asymptomatic CLL patients who are at high risk for early symptoms. These changes raised the question: what is the best time to begin treatment for those CLL patients who are at high risk for early symptoms?
CALGB 10501 was a three-arm randomised trial. The first stage was a blood test for the IgVH unmutated gene. This—rather expensive—test was not generally available to those not participating in the trial, but was provided to trial participants at no cost. Those with the unmutated gene, who are at high risk for early symptoms, would be randomised to Arm A or B. Those in Arm A would receive the standard treatment, and those in Arm B would be monitored and would receive the same treatment when their condition worsened. The treatment was six cycles of weekly treatment, 4 weeks apart. Each cycle included one infusion of rituximab and five infusions of fludarabine. Those with the mutated gene—at low risk—were placed in Arm C. Their disease would be monitored and they would receive treatment when their condition worsened. The primary end-point was ‘time to second treatment’, but all participants would be monitored every 3 months for disease progression and for overall survival. The study was designed to enrol about 1750 people.
I consulted with several physician colleagues in the Department of Bioethics at the National Institutes of Health. They thought that the trial was sound and well designed. Although it was not designed to test a ‘breakthrough’ drug, the timing of treatment is an important question in oncology treatment. The consent forms were very detailed—about 18 pages. I decided to enrol.
I had the genetic test early in 2009. It revealed that I had the unmutated gene and was at high risk for early symptoms. I was randomised into the ‘early treatment’ arm (A) and was scheduled to begin treatment. Prior to receiving treatment, subsequent blood tests revealed that my disease had worsened, and I was no longer technically asymptomatic. I was therefore excluded from the trial.
Although, like everyone, I do not have full access to my own motivations, I believe that I consented to participate for four reasons. First, I was curious as to what it would be like to be a research subject, particularly given my professional interests. Second, I liked and respected my physician and—like many patients—I wanted to please him. Third, I wanted to contribute to medical research so that others might benefit from the study. Fourth, as someone who believed and had written that people have an obligation to participate in low-risk medical research, I did not want to be or appear to be hypocritical.2
I did not regard this as a high net risk trial. Given that it was unknown as to whether presymptomatic treatment for high-risk patients was more or less efficacious than postsymptomatic treatment, I had no reason to prefer Arm A to Arm B if I were at high risk. As I saw it, there were three risks or burdens associated with participation. First, because treatment would be administered according to the protocol, I was giving up the option of having the intensity of the chemotherapy regime adjusted in light of my response. Second, participation may have involved a few extra blood draws and perhaps a bone marrow aspiration and biopsy, although most of the procedures performed for study purposes would also be performed in the course of clinical care. Third, I was not sure that I wanted to know whether I was at high risk for early symptoms. Although I'm not sure that I regarded participation as obligatory, given the relatively small risks and burdens of participation, I certainly did not regard it as an act of great altruism.
So what's the problem? As of 3 November 2011, only 84 persons had enrolled in CALGB 10501.1 Although the study has not been officially terminated, it is no longer recruiting subjects. I do not know whether I would have agreed to participate if I had known that there was a reasonably high probability that the trial would never complete or generate useful scientific data, but it is distinctly possible that I would not have agreed. Although I was not in the grips of a ‘therapeutic misconception’, I was in the grips of a ‘completion misconception’. It did not occur to me that the study would never be completed, and that possibility was not mentioned by my colleagues.
The purpose of this article is to consider whether prospective subjects should be informed about the possibility of non-completion as part of the standard consent process. As an argumentative strategy, I will defend the principle of non-completion disclosure: other things being equal, investigators should unambiguously disclose the possibility of non-completion if (1) it is, or should be, anticipatable that there is a substantial possibility of non-completion and (2) that information is likely to be relevant to a prospective subject's decision to consent.
Although it is possible that it would be best not to disclose non-completion information, all things considered, I believe that this principle—as stated—is not only true, but is obviously true. I bother to make the argument in some detail and wrapped in my personal narrative because (to the best of my knowledge) it has not been previously made, and no such principle is incorporated in any standard guidance about informed consent.
It is difficult to prove a negative—that non-completion has not been discussed in the bioethics literature about informed consent—but it is not specifically mentioned in any of the major documents that provide ethical guidance for clinical research. In general, the guidance about consent concerns the information required to make intelligent self-interested decisions and ignores some of the information required for intelligent altruistic decisions. The Nuremberg Code, for example, states that subjects should have the information required to ‘make an understanding and enlightened decision [including] the nature, duration and purpose of the experiment; the method and means by which it is to be conducted; all inconveniences and hazards reasonably to be expected; and the effects upon his health or person.’3 (Emphasis added) All subject-oriented.
The Common Rule opens up the space for information about non-completion when it states that consent forms should include ‘A description of any benefits to the subject or to others which may reasonably be expected from the research’.4 (Emphasis Added) It might be argued that the probability of completion is implied by the italicised phrase because the expectation of benefits ‘to others’ turns on whether the research will yield scientifically valid data. But IRBs (or Research Ethics Committees (RECs)) rarely if ever require that consent forms contain information about the possibility of non-completion.
Many institutions have a volunteer or participant's ‘Bill of Rights’ which is fashioned after the document provided by the Center for Information and Study of Clinical Research Participation.5 The relevant provisions state that subjects should be provided with information about the purpose of the trial, the benefits, risks, side effects or discomforts that can be expected, whether study interventions or procedures differ from those used in standard medical care, and so on. Again, this is all subject-oriented information.
I believe that the well-accepted principle that prospective subjects should be informed about the risks and benefits of participation to them is best understood as derived from the more general principle that they should be given whatever information can reasonably be expected to be relevant to an intelligent decision about participation. And that implies the principle of non-completion disclosure. There is no reason to privilege the information required for an intelligent self-interested decision over the information required for an intelligent altruistic decision. Indeed, it could be argued that the obligation to disclose information about the possibility of non-completion is greater than the obligation to disclose information about personal risks and benefits, since most prospective subjects will seek out information about the latter, but many, like myself, will not think about the former.
Trials may fail to complete for various reasons. A trial may be stopped because there are too many adverse events, or because a superior intervention has been introduced, or because of inadequate recruitment. The recruitment rate may vary with the type of disease, the type of sponsor and the incentives that are offered. For example, one study suggests that research sponsored by pharmaceutical companies is more likely to complete than government funded or academic research. A recent study of research approved by one REC in The Netherlands reported that almost half (46%) the academic or non-profit sponsored studies were terminated before the requisite number of subjects was reached. This percentage was notably lower in studies that were initiated by a pharmaceutical company (12.5%).6
Although I lack solid comparative data, it appears that oncology research is less likely to complete than research on many other diseases, and within that category, adult oncology research is much less likely to complete than paediatric oncology research where there is a greater history of integration of paediatric care and research.7 A recent article maintains that one out of five studies sponsored by the National Cancer Institute failed to enrol a single subject, and only half reached the minimum number for meaningful results.8 One study maintains that only 3% of adult cancer patients are enrolled in clinical trials.
There are several reasons as to why many studies fail to complete for lack of enrolment. Some patients recoil at the notion that their treatment will be chosen randomly even when the arms of the trial are in equipoise. Some patients may not understand why research-related procedures are necessary, and may overestimate the burden they pose. But while it is rarely discussed in these terms, participation in net risk research—that is, research in which the risks and burdens exceed the benefits to subjects—constitutes a classic collective action problem. It is in the ex ante interest of most people that socially valuable net risk research be conducted, but participation in such research is contrary to the interests of each individual. And to the extent that people are self-interested, they will seek to reap the benefits of research without paying the cost of participation. In many other collective action contexts (such as paying taxes) we use coercion to solve the problem. In medical research, however, we insist that subjects give their voluntary consent with the predictable result that participation rates are often quite low.
In addition, some of the ‘blame’ for inadequate enrolment lies with doctors who do not refer their patients to trials or tell their patients about them, either because they do not know what trials are available or, as seems common, because referral is time consuming and costly if only because they may lose a patient. But whatever the aetiology of inadequate accrual, many researchers seem to follow Lasagna's law (after Louis Lasagna, MD), namely, that they systematically overestimate the pool of available patients who meet the inclusion criteria and would be willing to enrol.9
Non-completion and participant motivation
Broadly stated, we can distinguish between self-interested reasons and altruistic motivations for participation, although both types of reasons can play a role in the psychic economy of a given person's decision. A person participates in research for self-interested reasons when she believes that the expected benefits to them of participation are greater than the risks and burdens of participation. As I noted above, there are several ways in which participation may be risky or burdensome. The personal benefits may include access to medical interventions that are otherwise unavailable, access to free care that would otherwise be unaffordable, access to superior care, payment and so on. With respect to such self-interested reasons, a subject needs to have the standard information about the risks, burdens and personal benefits; whether or not research leads to generalisable knowledge that is of benefit to other persons is irrelevant.
By contrast, if a person is considering participating (at least in part) in order to advance medical knowledge, it remains important that she be informed about the various risks and burdens mentioned above. After all, a person may be prepared to make a small sacrifice for the benefit of others, but not a large one. But to know whether the benefit to others warrants one's sacrifice, one also needs to know whether one's participation is likely to contribute to one's intended aim, and it will not be even a small contribution to that aim if that aim is never achieved. There may be some marginal scientific value to participation in medical research that does not meet the target enrolment, but I suspect that much non-completed research yields no generalisable knowledge. Although it is an empirical question, it seems likely that some prospective subjects would not consent to participate if they were told that their participation might be for naught.
Disclosing non-completion information
Damen and colleagues correctly state that non-completion is both a scientific and an ethical problem: ‘The value and validity of the principal study results would be questionable, and study subjects would have participated in these studies needlessly’.10 But how should this problem be addressed? They suggest that investigators ‘should thoroughly estimate the feasibility of attaining the required number of subjects’ and that ‘other stakeholders can contribute to resolving or reducing this problem’.11 They suggest, for instance, that journals should verify whether the submitted manuscripts are in conformity with the approved protocol, and that RECs should be ‘more critical of the feasibility of studies before granting institutional approval’. Perhaps funding agencies should also be more rigorous. Interestingly, however, the other stakeholders seem not to include the prospective subjects. Their position reflects the stance of the International Conference on Harmonisation Guidelines On Research (ICH): ‘The investigator should be able to demonstrate (eg, based on retrospective data) a potential for recruiting the required number of suitable subjects within the agreed recruitment period’.12 But the ICH does not recommend informing prospective subjects that the potential for recruiting the required number of subjects may not be realised.
IRBs surely should consider the possibility of non-completion as a variable in risk/benefit assessment. The Common Rule states that before research is approved, the IRB must determine that ‘the risks to the subjects are reasonable in relation to the anticipated benefits, if any, to subjects, and the importance of the knowledge that may reasonably be expected to result’. The knowledge that may reasonably be expected to result is a function of the probability that the study will yield valid scientific data as well as the importance of the knowledge. I do not know whether IRBs systematically consider the likelihood of completion, but colleagues with considerable experience suggest that they do not. But even if they do so, it surely would not obviate the need to provide non-completion information to prospective subjects who are entitled to make their own judgments as to whether the benefits to others warrants their sacrifice.
So what should be done? Other things being equal, the consent forms and process should specifically acknowledge the possibility that a study will not complete because of inadequate enrolment. At a minimum, the consent documents might include a blanket statement to the effect ‘This study is designed to benefit future patients, but there is some chance that it will not recruit sufficient subjects to provide valid scientific information’. In my view, however, it would be better if scholars, investigators and IRBs work to classify research protocols in terms of the likelihood of non-completion based on historical studies of the type of research in question. If adult oncology research has a low rate of completion, then that could be specified. In addition, and perhaps more importantly, once accrual has begun, investigators would have data as to how well a trial is meeting its milestones, and we could require that investigators disclose that information during the consent process. In any case, the underlying aim is to provide prospective subjects with information that allows them to make an intelligent decision about participation that is consistent with their interests, values and aims.
I cannot see any plausible objections to the proposed principle in terms of respecting the interests and aims of prospective subjects. At worst, the information will fall on deaf ears. If the reasonable subject would participate for self-interested reasons because the expected personal benefits outweigh the risks and burdens of participation, then it will not matter whether the protocol will yield valid scientific data. But if participation is likely to be a net risk or net burden to subjects and is justified by the expected benefits of the knowledge, then the disclosure of non-completion information (for lack of recruitment or for other reasons) will enable prospective subjects to more intelligently consider whether the expected benefits to others warrants their sacrifice.
But therein lies a potentially serious difficulty. For while non-completion information should be disclosed, other things being equal, it may be objected that other things are decidedly not equal. It is difficult to recruit prospective subjects into clinical trials that involve net risks or burdens when they are not told that there is a reasonable probability that the study will not complete. It may well prove more difficult to recruit them when they are told, and the effect will be greatest on those studies that are least likely to complete in the absence of such disclosure. And so we should consider whether there are good reasons to refrain from such disclosures.
One objection to the principle of non-completion disclosure rests on an analogy with charitable organisations, which may deliberately or unintentionally rely on donor ignorance or misunderstanding about the purpose or recipients of their aid or the extent to which their donation will further their aims. Donors may not understand that their donation may be used for purposes other than what they intended, or that a portion (perhaps a large portion) of their contribution goes to administrative costs and additional fund raising. If such organisations need not disclose such facts, then perhaps research subjects need not be informed about the possibility of non-completion.
There are two replies. First, it may be argued that the analogy should go the other way. Perhaps charities do have a positive obligation to disclose data about the actual beneficiaries of such aid, and the proportion of donations that reaches the intended beneficiaries. Second, it may be argued that there is a distinction between the biomedical context in which we have decided that there should be a formal consent process with disclosure of relevant information and charitable giving where we have always been satisfied with less rigorous criteria for what constitutes valid consent. I am not sure which position is correct, but on the assumption that investigators have an obligation to provide the sort of information that is likely to be relevant to a prospective participant's decision, then there is a presumption that non-completion information should be disclosed.
A second objection takes on the presumption just noted. It claims that while it is seriously wrong not to disclose information about the personal risks and burdens of participation, it is much less wrong not to disclose information about the probability of non-completion or, more generally, about the social value of the research. And given that the disclosure of non-completion information may impede research of potential social value, the beneficial consequences of non-disclosure are sufficient to justify non-disclosure.
I think this objection fails. If the point of insisting on informed consent is to allow people to make decisions about participation that is consistent with their interests, values and aims, then, and as I suggested above, I see no reason to privilege information that is relevant to considerations of self-interest over information that is relevant to altruistic aims, and this is particularly so if we think it desirable that people participate in research for altruistic reasons.
Finally, it may be objected that my proposal should be rejected mainly because it would seriously hamper valuable medical research. Now the potential adverse effect on recruitment is hardly a decisive objection to my proposal. After all, the negative effect on recruitment is a cost to the entire informed consent process. Baruch Brody describes a placebo-controlled study of the benefits of arthroscopic surgery in which subjects were required to write by hand that they understood that they might be receiving ‘sham surgery’.13 Although this procedure generated a ‘significant refusal rate’, Brody maintains that this is ‘the price you may have to pay if you increase potential subjects’ understanding’. No one said that ethics comes cheap.
Still, even if ethics does not come cheap, it is possible that the price of my proposal is too high. The Belmont Report, for instance, adopted a pluralistic approach with respect to the three core values—autonomy, beneficence and justice. It did not claim that respect for autonomy trumps or is even weightier than considerations of beneficence, which might support rejecting the proposed principle. As Beauchamp and Childress put it, ‘The principle of respect for autonomy does not by itself determine what, on balance, a person ought to be free to know or do, or what counts as a valid justification for constraining autonomy’.14 We might well decide on grounds of beneficence that, all things considered, it is morally preferable not to provide subjects with non-completion information full well knowing that we are thereby exploiting prospective subjects’ ignorance or misunderstanding about the probability that their participation will contribute to medical knowledge. If so, we should do so with our eyes open. We should face up to the fact that we have knowingly compromised the principle that prospective subjects should be provided with the information that is relevant to their decision to consent.
I wish to acknowledge the following persons who provided helpful comments upon a previous version of this paper: Franklin Miller, Seema Shah, David Wendler, Christine Grady, Nir Eyal, Paul Appelbaum, and Steven Joffe. I also thank two anonymous reviewers for JME for their helpful comments.
Contributors The views expressed are the author's own. They do not reflect any position or policy of the National Institutes of Health, US Public Health Service, or Department of Health and Human Services.
Competing interests None.
Provenance and peer review Commissioned; externally peer reviewed.
i Common Rule 45 CFR.111 Criteria for IRB approval of research.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.