Therapeutic misconceptions and moral conflict in clinical research

Consider a physician whose patient meets inclusion criteria for a randomized clinical trial centered at a nearby institution, where the clinical trial has been undertaken to resolve a controversy regarding which of 2 treatment plans is more efficacious in this patient’s condition. The physician must then choose whether to treat the patient “empirically” by prescribing the treatment that she judges most likely to benefit the patient, or to recommend enrollment in the clinical trial where the patient will be randomized to 1 of 2 treatment arms. Intuitively, this physician should consider the interests of 2 different parties: her present patient, who could be randomized to a treatment that she believes to be less likely to benefit him; and future patients with similar conditions, who stand to benefit from medical knowledge gained by testing interventions in a controlled fashion. Clinical researchers and bioethicists have long recognized a potential ethical conflict in clinical research, between the clinical aim of benefiting the individual patient and the scientific aim of producing generalizable data to guide the care of future patients. (The equipoise requirement represents an established, but I think ultimately unsatisfactory, attempt at resolving this conflict.) However, in their article in this issue of Annals of Neurology, Patrick Lyden and colleagues acknowledge no conflict between these aims. They claim that clinical trial enrollment is superior to empiric treatment both from the perspective of the individual patient and from the perspective of future patients, and that empiric treatment would therefore be unethical. They go on to propose a blanket approach to clinical decision making in medical emergencies (although most of their arguments are applicable to clinical practice quite generally) in which clinical trial enrollment is prioritized over empiric treatment. Let us grant that enrolling this patient in a clinical trial would promote the interests of future patients. I wish to focus instead on Lyden and colleagues’ repeated assertion that trial enrollment is also best for the individual patient who is enrolled in the study. The grounds for this assertion are weak. First, they claim that “Most available data suggest that patients enrolled in trials may benefit, regardless of the treatment group into which they are randomized.” Much appears to depend on what these authors mean by “most,” “suggest,” and “may”; the 2 sources cited are a 2001 meta-analysis of 21 studies that concludes that there may be such an effect but the evidence is poor, and a 2008 meta-analysis of 85 studies (including many of the same studies, and excluding several others) that found no effect. Second, they allude to other features of clinical trials that may be beneficial to enrolled patients, such as more intensive monitoring and risk factor management, or earlier access to new interventions. This represents a rather selective view, as they give no consideration to other features of clinical trials that limit the quality of care available to research participants for the sake of producing generalizable data. These include not only randomization but also washout periods, placebo controls, restricted flexibility in the dosing of study drugs, restrictions on the use of concomitant treatments, invasive data-gathering interventions such as blood draws and lumbar punctures, sham surgeries and other interventions performed to preserve blinding, and imaging procedures performed not for the sake of guiding treatment but instead for evaluating efficacy. (It should also be noted that some of the beneficial features they cite are not intrinsic to the clinical research setting—if patients do in fact receive more effective monitoring and risk factor management in clinical trials than in routine care, this may not suggest that more patients should be enrolled in clinical trials, but instead that our models for the delivery of routine care must be improved.) Lyden and colleagues begin their article with the arresting anecdote of a young patient with a devastating right middle cerebral artery occlusion who failed to respond to standard treatment with recombinant tissue plasminogen activator, was enrolled in an investigational protocol, and was then randomly assigned to the active treatment arm, underwent embolectomy, and had a good outcome. According to their arguments, it would have been unethical, for instance, for a physician convinced of the utility of embolectomy to empirically proceed to embolectomy after the patient failed to respond to standard treatment, without EDITORIAL