Weighing the participant against the society a study serves
A trial asks one person to carry risk so that many others can have evidence. That tension between the individual and society is weighed in what a participant accepts in signing, in how a cancer trial is designed without withholding treatment, in what a schedule of assessments asks of participants for data that may never be used, and in how many participants a trial enrols.
What a participant accepts for society's benefit
Ethics in health care asks when the individual should be given more emphasis than society, and when society should be given more emphasis than the individual. One view holds that this dynamic tension between the individual and society is the overarching theme of ethics in health care.
One view holds that where the aim of a patient's participation in a clinical study is meaningful information that allows a regulatory body to determine whether a medicine, device, diagnostic or vaccine has the potential to benefit the population at large, the individual patient in some ways has to give way to the benefit of society. That does not mean abusing the patient, or subjecting them to unethical or harmful treatment where that can be anticipated.
Researchers must always protect participants from harm and must be fair to them, and the needs of research participants should always come before the objectives of the study.
A clinical trial carries a potential for harm. Cohorts are often blinded, so neither the participant nor the investigator knows what is being given, and the products are unapproved: they are tested because too little is known about them to determine whether they are safe and effective enough to be marketed. A participant may receive placebo rather than an active treatment, and one with a progressive disease then forgoes any other therapeutic intervention for the duration of the study. The potential harm is that the disease is allowed to progress uninhibited, without the benefit of an intervention that might deter, slow or cure it.
On entering a trial, the participant signs an informed consent form, which is essentially a contract. In it they agree to take part and accept as their risk that they may or may not receive the investigational product, and that they may receive a placebo or a comparator over which the investigational product is believed to be superior. Taking part also reduces a patient's options, because they must follow the rules. Outside a trial a patient has agency, the right to self-determination. Inside one that right may be constrained: the patient may not have the right to access another drug, because the protocol forbids taking any drug during the study other than the investigational product or the comparator.
Trial design in cancer
Cancer is life-threatening and demands urgent treatment, so it poses ethical and technical problems for study design that would not usually arise in other indications. Placebo-controlled trials of cancer drugs are therefore unusual. The test drug is typically given in combination with an established chemotherapy or radiotherapy regimen, so that in a controlled study every group receives at least an established standard of treatment. Similarly, best supportive care may be offered to all patients in an effectively placebo-controlled study. A new product tested in isolation is generally tested in patients who have failed to respond to several established treatments and are not eligible for other standard regimens.
Cancer drugs are rarely if ever piloted in the usual healthy male subjects for first-in-man and other phase I trials, because most are so toxic that the risk to people who cannot benefit personally is unacceptably high. Early development is typically performed instead in patients with advanced, usually incurable disease.
An overly strict definition of the concomitant treatments cancer patients may receive during a trial is ethically questionable. Compared with other indications, a cancer trial demands greater tolerance of reduced compliance caused by missed treatments and permits a wider range of concomitant therapies. Cancer specialists have considerable freedom to determine the best treatment for each patient, and when a new product is given alongside a particular established regimen, recruitment should be restricted to patients who would otherwise qualify for that regimen. Trial treatment may be for a defined duration or number of cycles, and patients still alive or showing a response typically may continue treatment beyond the planned trial duration.
The burden of data that will never be used
An analysis of almost 3,500 protocols sorted their procedures into core procedures, related to the study endpoints; required procedures, demanded by the Declaration of Helsinki or by legislation; standard procedures, such as weight, height and body mass index; and non-core procedures, related to exploratory endpoints. Non-core procedures accounted for just under 20% of procedure costs in phase 2 and phase 3 studies, and that figure counts only what the hospital charges the sponsor, not the cost of monitoring and entering the data. Between 2002 and 2012 a typical participant's commitment grew from nine visits and 187 days on trial to eleven visits and 230 days.
Procedures whose results reach neither the clinical study report, nor the publication, nor the drug application waste resources and time at clinical sites, and this raises the question of whether asking participants to undergo them is also unethical.
The benefits of research should exceed the risks to the individual participants, and there should be a benefit to the community. Where historical data could give the same insight, there is little reason to expose research participants to the risk of a study. In that sense the possibilities AI offers for reusing data are tremendous.
How many participants a trial enrols
A larger trial is not always better, and at times a larger sample can border on the unethical. Since evidence-based medicine arrived in the early 1990s, practice has increasingly been shaped only by large randomised trials, of 200 or more patients, whose results were statistically significant but whose effect sizes were rather small. Small studies with dramatic effects raise a different question: whether it is even ethical to follow one with a randomised trial.
Such cases run from the extreme to the contested. Rabies had long been recognised as universally fatal when, in 1885, a 9-year-old mauled by a rabid dog received a new vaccine and became the first person to survive the illness. The vaccine was accepted worldwide after one further patient survived, without a randomised trial of its efficacy or safety.
Extracorporeal membrane oxygenation is a highly invasive technique used in neonates with severe, reversible cardiac and respiratory illness unresponsive to maximal medical therapy. In the 1980s it raised survival among very ill neonates, whose mortality risk was 80% or more, to 80%. The researchers felt strongly that they were condemning some children to death, but were concerned that colleagues would not accept so risky a procedure without a randomised trial. The trial was stopped after the fourth death among 10 controls, when all 9 treated neonates had survived.
In a study published in 1995, Esselstyn and colleagues coached 22 patients with severe, angiographically documented coronary disease on a plant-based diet with 10% fat. Five dropped out within two years, and 11 of the remaining 17 kept to the diet through a mean of five years: they lowered their cholesterol, saw their lesions regress or stabilise and had no new cardiovascular event. The 5 who returned to their earlier diet had 10 further cardiovascular events, and the 6 still on the diet at ten years had no further coronary events. The study showed a dramatic response to the diet but was roundly criticised for its small sample and for not being blinded. Larger randomised trials that are rigorously blinded, which is not possible in most outpatient nutritional trials, often show only a modest effect compared with standard of care, yet they often change practice patterns. The diet study described a 100% response rate with no safety issues. Had the diet been a pill and the control the average American diet, scientists would argue whether a control group in future studies was ethical at all.
The same question arises between phase 2 and phase 3. Phase 3 trials have traditionally committed sponsors to a large sample with no possibility of stopping early for benefit, harm or futility. If phase 2 suggests a marked benefit, is it ethical to randomise half or a third of phase 3 patients to placebo before the outcome is more certain? And if early studies suggest no difference from placebo, or even harm, is it ethical to randomise patients to the intervention to confirm that impression?
An adaptive design allows an interim analysis in which accumulating data are used to decide how aspects of the trial are modified as it continues, without undermining its validity and integrity. Done right, interim analyses may bridge the gap between small studies with large effects and large studies with small effects. An ethical trial should include only as many patients as are needed to answer a specific clinical question. Statistical calculations should be used to identify a plan that will answer that question effectively, not the other way round.