The “inclusion benefit” in clinical trials

Abstract

See related article, p. 151. Our current regulatory systems to protect the subjects of human research seem misguided in many ways. If clinicians try a new therapy with the idea of studying it carefully, evaluating outcomes, and publishing the results, they are doing research. Research is thought to be risky, and the subjects of the research are thought to be in need of special protection. Therefore an institutional review board will review the protocol, the informed consent form will be carefully scrutinized, and the research may be forbidden. If the study is permitted, every adverse event will be carefully documented and scrutinized. If, however, clinicians try the same new therapy without any intention of studying it, simply because they believe it will benefit their patients, it is not research and does not need institutional review board approval, consent may be obtained in a manner governed only by the risk of malpractice litigation, and adverse events may not necessarily be noticed or analyzed. It would seem that the patients in the second situation are at much higher risk than the patients in the first. After all, the physicians in the first situation are carefully evaluating the therapy, whereas the physicians in the second situation are using the therapy based on imperfect hunches. Schmidt et al1Schmidt B Gillie P Caco C Roberts J Roberts R. Do sick newborn infants benefit from participation in a randomized clinical trial?.J Pediatr. 1999; 134: 151-155Abstract Full Text Full Text PDF PubMed Scopus (63) Google Scholar report that infants who were enrolled in the placebo arm of a randomized controlled trial had better outcomes than infants who were eligible for the study but not enrolled. They are not the first to report such a finding. Other investigators have reported similar findings after trials for diseases ranging from cancer to cholera. A number of explanations have been offered for the apparent benefit of RCT participation, including selection bias, placebo effects, and adherence to well-defined protocols. Each of these explanations might suggest narrowly focused methodological responses. Taken together, however, the implication that patients actually benefit from participation in clinical trials creates a number of challenges for the conduct, interpretation, and regulation of clinical research. As a thought experiment, let us suppose that it is really true that participants in RCTS do have better outcomes than similar patients with similar diseases treated in the same institution at the same time. We might call this phenomenon an “inclusion benefit,” that is, a benefit that accrues to patients simply by being included in a research protocol. Let us further suppose that we do not know what the precise explanation for the inclusion benefit might be. What would the implications be? One implication would be that RCTs are not quite what they seem to be. RCTs, like other scientific experiments, seek to hold the environment constant in such a way that we can isolate the effect of the discrete, specific intervention that we are studying. The phenomenon of inclusion benefit suggests that this is not the case. Instead, RCTs change the very environment of care in ways that have an impact on the results that we can see and measure. One of the more obvious ways that this is likely to occur is through the use of strict protocols in study patients, whether they are in the intervention arm or the placebo arm. Clinical trials thus incorporate a style of treatment that looks quite similar to practice guidelines, and the inclusion benefit phenomenon might be seen as a validation of more strict controls on practice variation and physician autonomy. Schmidt et al1Schmidt B Gillie P Caco C Roberts J Roberts R. Do sick newborn infants benefit from participation in a randomized clinical trial?.J Pediatr. 1999; 134: 151-155Abstract Full Text Full Text PDF PubMed Scopus (63) Google Scholar suggest another implication. RCTs may not be as generalizable as they are sometimes thought to be. This might be secondary to problems in patient enrollment, but that might not be the whole story. If the difference persists between study patients and those not in the study, even with modified recruitment techniques, then clinical trials would routinely have less statistical power to detect differences between any 2 treatments than was previously thought. This would lead to widespread underestimates of the value of new drugs or treatments and possible rejection of certain drugs or interventions that are, in fact, beneficial. Correcting for this would first require an acknowledgment of the phenomenon of inclusion benefit and then an estimate of the magnitude of the phenomenon. We could then discount a certain amount of benefit in the control arm or slightly relax our standards for statistical significance, accepting, perhaps, P values of .1 rather than the standard .05, as the threshold of significance. This would be a practical response to the problem of incorporating inclusion benefit into our clinical epidemiology. The existence of an inclusion benefit also suggests that our ethical, legal, and regulatory framework for oversight of clinical research may be somewhat misguided. The framework governing clinical medical research begins from the assumption that research is a risky endeavor. In particular, the guiding view is that it is more risky for a patient to be in a research protocol, such as a randomized clinical trial, than it would be for that patient to receive standard medical therapy. Persons with this view reason from famous examples of research abuse, such as the Tuskegee syphilis study or the Willowbrook hepatitis experiments, to conclude that the conduct of clinical trials requires special safeguards to protect participants.2Rothman DJ. Were Tuskegee & Willowbrook ‘studies in nature’?.Hastings Cent Rep. 1982; 12: 5-7Crossref PubMed Scopus (92) Google Scholar At the same time, there seems to be another view of research that is gaining momentum, namely, that participation in clinical research is neither risky nor dangerous but is, instead, beneficial and desirable. For example, AIDS activists demand the “right” to be included in clinical trials, implying that it is not a risk or a burden to be in a trial but instead is a benefit that allows access to the latest experimental drugs.3Merigan TC. You can teach an old dog new tricks. How AIDS trials are pioneering new strategies.N Engl J Med. 1990; 323: 1341-1343Crossref PubMed Scopus (36) Google Scholar Women protest when their insurance does not pay for experimental protocols for the treatment of breast cancer, again implicitly suggesting that participation in clinical research is desirable.4Daniels N Sabin JE. Last chance therapies and managed care. Pluralism, fair procedures, and legitimacy.Hastings Cent Rep. 1998; 28: 27-41Crossref PubMed Scopus (54) Google Scholar And, in some areas of clinical care, such as pediatric oncology, the standard of care is to enroll virtually all patients in a research protocol with the assumption that there is little conflict between the patient’s interest, the researcher’s interest, and the interest of society to do so.5Kodish ED Pentz RD Noll RB et al.Informed consent in the Children’s Cancer Group: results of preliminary research.Cancer. 1998; 82: 2467-2481Crossref PubMed Scopus (80) Google Scholar The phenomenon of inclusion benefit, if real, raises a profound challenge to the current framework. What if, instead of creating increased risk, clinical research creates increased benefit? Perhaps we should include, as part of the informed consent process for clinical research, a statement to the effect that participation in a research protocol has been shown to lead to better outcomes than nonparticipation. And as we try to protect patients, perhaps we should shift the focus of our oversight efforts from carefully designed and defined research protocols to the more murky area of innovative therapies. Perhaps we would all be better off, in terms of measurable health outcomes if, for every health problem, we were enrolled in an RCT. The moral intuitions that are incorporated in guidelines for research ethics appear to be at odds with the vision of many clinical investigators who perceive themselves as acting primarily to benefit their patients. They believe that evaluation of outcomes is part of their commitment to excellent patient care and that participation in clinical trials is the best way for patients to receive state-of-the-art care. Clinical trials, by this view, seek new knowledge in order to benefit patients. The research component is subsumed under and is secondary to the patient care component. Why then, one might ask, do ethical codes regulate the responsible investigator rather than the irresponsible adventurer? I think the codes are not as irrational as they may seem. The basis of the ethical codes must be understood in terms of traditional expectations of the doctor-patient relationship. Research is thought to present a risk to patients not because it is new, poorly understood, or hazardous to the physical well-being of the patients. Much medical therapy is new, poorly understood, and potentially risky. Instead, research is considered hazardous because the physicians who conduct research are using the patients as a means to an end. Their goal is no longer solely the well-being of the patients. They are also, perhaps primarily, committed to creation of generalizable knowledge. Research thus transforms the doctor-patient relationship into something other than a fiduciary relationship in which the physicians’ sole concern is the well-being of the individual patient. The real goal of ethical guidelines is to acknowledge that conflict and provide psychic safeguards against it. These psychic safeguards are more symbolic than real because they rely on a dichotomy between research and therapy that is fundamentally faulty but symbolically useful. By labeling certain activities “research,” we reassure our patients and ourselves that we have a high degree of certainty about the remainder of clinical activities that we call “standard therapies.” The study by Schmidt et al1Schmidt B Gillie P Caco C Roberts J Roberts R. Do sick newborn infants benefit from participation in a randomized clinical trial?.J Pediatr. 1999; 134: 151-155Abstract Full Text Full Text PDF PubMed Scopus (63) Google Scholar is disconcerting because it illustrates how some of the presumptions about the risks of research may be just wrong.