Skip to content

The equal air-time solution for controversial research

When are placebos ethical in medical research? One common answer is that it is only appropriate to use placebos in research when there is no proven effective treatment for the condition (1). On this view, if there is a proven treatment placebos would be unethical, and any trial should compare new drugs or treatments with the existing proven one. But what if the question of ‘proof’ is in dispute? For new medical treatments there often comes a point where some researchers and doctors are convinced that the new treatment is effective and safe while others remain unconvinced. When placebo-controlled trials take place in this setting they are often controversial.

A scientific review paper published this week in the medical journal Pediatrics provides an example of this sort of controversy. The paper suggests that a new treatment could halve the risk of death for very premature babies. It looks at evidence from 11 previous trials involving more than 2000 babies. Infants who received probiotics (so called ‘friendly bacteria’) were less likely to die or to develop nasty infections of the bowel. In the review, and in an accompanying editorial (which I co-wrote) it is argued that further placebo-controlled trials are not needed, and indeed that they may be unethical.

However, others within the medical community are not yet sure about probiotics for premature babies. The idea of giving fragile premature infants (whose immune systems are not fully functioning) a deliberate dose of bacteria seems counterintuitive and potentially dangerous. There are several large placebo controlled trials planned of probiotics in premature babies – in the UK and in other parts of the world. A number of very experienced and well-respected doctors and scientists believe that these trials should go ahead.

There are different ways in which we could go wrong. If we adopt new treatments too soon we risk potentially harming a large number of babies. There are past examples of medical treatments in premature babies that looked promising and were enthusiastically embraced, only to later turn out to be ineffective or even harmful. On the other hand, if we are too cautious about the evidence and insist on more and bigger trials there is a real danger than infants will be denied an effective treatment and harmed as a result.

The usual response to these controversies is to turn to the scientific evidence. We just need to know whether or not science has reached sufficient certainty in favour of the new treatment. Doctors who are in favour of probiotics (the ‘No trial’ camp) will point to the strength of the existing evidence, while those who are unsure (the ‘Yes trial’ camp) will highlight the gaps in knowledge and potential weaknesses. But it does not seem likely that the two camps will reach agreement about the evidence anytime soon. Arguments will rage in the letters page of Pediatrics for some time.

However, an alternative answer to the problem is to involve patients (or in this case parents) in the debate. It is paternalistic for doctors alone to be debating about whether or not the risks to the patient are or are not too great for the research to take place. After all, it is the patient who is going to shoulder those risks (or be denied them). If we explain the evidence and the uncertainties to patients, and they agree to be randomised to placebo or to the new treatment, then it would, arguably, be ethical for a trial to take place.

One problem with this approach, relying on patient autonomy to answer the problem, is that the ‘No trial’ camp may worry about whether prospective participants in a trial are actually given a fair summary of the evidence. Any trials are likely to be conducted by the ‘Yes trial’ camp, those who are unsure about existing evidence. These scientists are likely to emphasise the negatives in previous trials and the unknowns. They may downplay the significance of trials that had shown benefit. Participants may not even be aware that doctors are divided about the treatment and about the trial.

But we could get around this problem. One solution (particularly topical in view of current events in the UK) would be to involve both camps in the preparation of information that is given to trial participants. Similar to policies relating to public broadcasting or to television debates during political election campaigns we could give ‘equal airtime’ in information sheets to the ‘Yes’ camp, and to the ‘No’ camp. Research ethics committees would need to make sure that information presented was accurate and appropriate.

What arguments might be given against the ‘equal airtime’ solution? One objection might be that information sheets for trials are already reviewed by ethics committees, and already contain information about the potential benefits and harms of trial participation. However, if that were genuinely the case, then researchers would have nothing to fear by allowing those opposing trials (for genuine reasons) to be involved in the development of trial information sheets. A second objection might be that this would impose an excessive burden on those designing trials. However, the equal-airtime solution would not apply to all clinical trials. It might be invoked only where there is genuine controversy within the medical community about the wisdom of further trials. A third possible objection is that providing potential trial participants with the point of view of the ‘No trial’ group would lead to a collapse in enrolment; few, if any, patients or parents would agree to the trial if given that information.

This last worry may be the strongest for those who would like trials to proceed. It may indeed be the case that few parents would agree to their infant being randomised to placebo if they were given a fair summary of the evidence both in favour of and against probiotics. But in that case, surely, we have our answer. The trials must stop.


(1) World Medical Association Declaration of Helsinki

The benefits, risks, burdens and effectiveness of a new intervention
must be tested against those of the best current proven intervention,
except in the following circumstances:

The use of placebo, or no treatment, is acceptable in studies
where no current proven intervention exists; or
Where for compelling and scientifically sound methodological
reasons the use of placebo is necessary to determine the efficacy or
safety of an intervention and the patients who receive placebo or no
treatment will not be subject to any risk of serious or irreversible

Girish Deshpande, Shripada Rao, Sanjay Patole, and Max Bulsara

Pediatrics published online
April 19, 2010


William Odita Tarnow-Mordi, Dominic Wilkinson, Amit Trivedi, and
Jesper Brok

Probiotics Reduce All-Cause Mortality and Necrotizing
Enterocolitis: It Is Time to Change Practice

Pediatrics published online
April 19, 2010


Dominic Wilkinson Therapeutic hypothermia and the equal air-time solution for controversial randomised trials Journal of Paediatrics and Child Health (forthcoming)

Share on

6 Comment on this post

  1. William Tarnow-Mordi, WINNER Centre for Newborn Research, University of Sydney

    Thanks, Dom, for this very clear summary of the background to the anticipated debate on probiotics.

    Can I add a point we raised in our commentary in Pediatrics? – We thought it appropriate to invite parents to enrol babies into new or ongoing placebo controlled trials if their babies are too ill or too young to meet the eligibility criteria of the earlier placebo controlled trials from which current evidence derives. This is because substantial uncertainty remains as to whether such babies will benefit or be harmed.

    As well, I wondered what you thought of the suggestion that, in new or ongoing randomised trials, parents of babies who do meet the eligibility criteria of previous trials be offered open label probiotic as an alternative to entering the trial?

    With these two modifications, new and ongoing placebo RCTs of probiotics may remain ethical and viable, but would almost certainly take longer to recruit their planned sample sizes.

  2. Thanks William,

    we should make a distinction between the claim that particular trials of a new treatment are unnecessary (or unethical) and a claim that all trials of a new treatment would be unnecessary or unethical. You are right of course to point out that it would be entirely appropriate to perform placebo comparisons where the new treatment (probiotics in this case) were being used for a different indication or in a different population – for example sicker or smaller infants excluded from previous trials.

    As for your second point, I am strongly inclined to think that patients *should* be able to choose an open label treatment outside a trial with a small number of exceptions
    a. the treatment is unlicensed/not registered outside a trial (this may be a particular problem for neonatal treatments where many are off-license)
    b. the treatment is only funded in the context of a trial (though patients should be able to choose to pay for it)
    c. the doctor is not willing to prescribe or provide it – doctors cannot be legally (or perhaps morally) required to provide a treatment that a patient requests that they do not personally feel is of benefit. (Though they should refer the patient to another physcian who may feel differently)
    d. where there is an overwhelming public interest in an RCT being performed, and there is good reason to think that allowing patients to choose the treatment outside a trial would preclude a trial

    There is more to say about conditions c and d. I think that doctors should take seriously a patient’s requests for treatment – and be prepared to provide a treatment that they would not personally endorse – if that treatment is one that a reasonably body of medical practitioners would support. So a doctor wouldn’t be required to provide a strange magnet-based alternative therapy for a patient with pneumonia. But they might be required to provide probiotics. On the other hand, I take seriously the view that doctors shouldn’t be required to act against their own sincerely held views about what is best for the patient.
    I also think that d is critical. So, for example, if we think that probiotics are highly likely to be of benefit to Australian or UK patients, but the only way that we can convince either regulators or our colleagues of this is to perform another trial, AND the only way to successfully perform a trial is not to allow patients open access outside a trial – then we may need to bite the bullet and restrict patient choice for the benefit of the wider community.

    We might link conditions b and d. So for new treatments where there is an overwhelming public interest in further trials it might be a condition of funding that it only be provided in the context of a trial. In fact this is often the case currently. This may (depending on the cost) have the desired effect of encouraging recruitment while still giving people the possibility to choose. However, of course, the freedom to choose the new treatment in this case would be linked to inequality in income. The burden of risk in trials would fall upon those who are economically disadvantaged, which hardly seems fair.

  3. There were trials of the beneficial use of probiotics on infants before this. How was the issue of notice and consent worked out in those cases. Of course, if the initial trial required notice and consent, there being no data from previous trials to indicate (and persuade the parents) that the child may benefit from probiotics, there might not have been any large-scale trials.

    This raises a serious question of social costs and benefits in a society in which we value autonomy, to the extent that it can exist in an individual human. How valuable is the research/ how costly is the consequence of not (adequately) getting consent (including the costs of lawsuits demanding compensation for the unconsented-to trial). Is this kind of calculation allowed in current hospital or university rules for medical research?

  4. Hi Dennis,

    Previous trials have, of course, required parents to be informed about the potential risks or benefits of being involved in the trial. The nature of the information that parents would have been given will have changed over time. In the initial trials parents would have been told that there was theoretical and animal-based research that indicated potential benefits from probiotics. Subsequent trials would (presumably) have included information that some previous trials in infants had suggested benefit, but that there was not enough information to be sure. The question now is what information should be provided to parents of infants potentially involved in ongoing or new controlled trials? In a supplement to the above editorial (available at William and I have provided a sample information sheet that might be provided to parents. As described above the information sheet provides on one side the case for further trials, and on the other side equal airtime to the opposing perspective. Although I personally think that there is a strong case against further trials, I think that there is an overwhelming case that if trials proceed parents are provided with the results of the most up to date meta-analysis.

    You also point to a deeper question. What if there were a strong case for benefit to society of research (for example to minimise the chance of errors in adopting a new treatment, and in detecting potentially rare side effects), but the existing evidence were such that no, or extremely few parents/patients if given a fair summary of the evidence would enroll in such a trial? That may be the case for probiotics. Then, you are right to point to the conflict between individual autonomy and collective benefit. If we thought that the benefit were significant enough there would be two theoretical options. One would be to perform trials without consent, or without full consent.
    Alternatively we could limits peoples choices in a way alluded to in the previous comment. If access to the new treatment were contingent on enrollment in the trial, or perhaps if free access to the treatment were contingent on enrolling in the trial, then a trial might be possible with the least infringement of the liberty of individuals. If we choose this option we should be open about it, and about the reasons for it, and we should acknowledge that it represents a compromise in patient choice (and potentially wellbeing) for the sake of the greater good.
    As for whether this sort of calculus is permitted in conventional rules governing medical research – I suspect not, though I think it occurs nevertheless.

  5. I agree with much of what has already been said. The key concern for me is not involving and informing parents. They deserve to know (and understand) the difficulties we have in answering this question ‘beyond reasonable doubt’ but also deserve to be treated with respect. Many parents are capable of weighing competing effects – some are very happy to trade future uncertainties with short term benefits. Doctors have rightly been accused of paternalism in the past [and present!] – surely we are making a major judgement on what is best if we don’t help inform parents. I would find it very difficult to persuade a parents who asked for their baby to receive probiotics that they were ‘wrong’. Currently most neonatal practitioners must be hoping parents don’t read the meta-analysis

    This is taken from from the publicly accessible Neonatal Formulary 5 on Anon. Probiotics ‘There is a strong case for informing parents that the strategy is new and will be used only with their consent. However, knowing what we now know, do we have the right to deny parents the option?’
    (Comment): Has the time come to start using probiotics more widely?
    In: Hey E. (ed) Neonatal Formulary 5

  6. Dominic Wilkinson

    Thanks Nick,

    the other interesting question that hasn’t yet entered into the debate, is whether neonatologists would be legally vulnerable if they failed to discuss probiotics with parents of a premature infant who subsequently died of NEC.

    While I don’t think it would be regarded as negligence using a Bolam standard in the UK, (given that the overwhelming majority of responsible members of the profession – rightly or wrongly, would not do so at present) other parts of the world have different tort standards.

    I don’t think fear of litigation is necessarily a good reason to change practice. But if people fail to be convinced by a meta-analysis of 2000 infants, trial sequential analysis and considerable supportive pre-clinical evidence, then perhaps they should pay attention to the possible legal consequences.

Comments are closed.