Should newborn infants be excluded from multiple research studies?

Many infants admitted to neonatal intensive care, especially those born prematurely, are eligible to participate in several research studies. Uniquely, newborns may also have indirectly been involved in research if their mothers were recruited to studies while pregnant. The website of the British Association of Perinatal Medicine lists 20 ongoing clinical trials in which many units in Europe will be participating. In a typical subspecialty neonatal unit in the UK a pre term infant might be eligible to participate in six research studies, ranging from international, multi centre, randomised controlled trials to local, unit-based observational studies. The scientifi c and ethical implications of recruiting patients to more than one clinical trial or research study are unclear but are particularly important in perinatal medicine. We are aware of signifi cant diff erences in attitudes across Europe. The Association of British Pharmaceutical Industries gives guidance, based on expert opinion, that there should be a gap of at least 3 months between an individual’s (healthy volunteer) involvement in medicinal trials to allow for wash out of drug eff ects. The US Food and Drug Administration stipulates 28 days. There is some guidance from the UK National Research Ethics Service on multiple study involvement of patients, but none specifi cally relating to infants or children. The service’s recommendations emphasise the importance of distinguishing medicinal from non-medicinal studies, referring to the Association of British Pharmaceutical Industries and Food and Drugs Administration guidance for medicinal trials and suggesting that the involvement in multiple nonmedicinal studies should be within the remit of review by research ethics committees. There are currently no legislative restrictions under European Union directives for good clinical practice to limit recruitment of infants to only one trial. Our experience, and that of other clinicians coordinating clinical trials, is that there is no consistency in opinion on this matter from diff erent neonatal units, ethics committees, or other regulatory bodies. Although there may be ethical concerns about clinical trials in newborn infants because they are vulnerable and unable to consent for themselves, this has to be balanced by the concerns about the use of interventions in patients in whom they have not been tested. Most drugs prescribed in newborn infants have not been fully assessed in this age group. The new European Union Regulation on Paediatric Medicines, which came into force on Jan 26, 2007, aims to address this issue with a legislative framework to increase availability of medicines specifi cally adapted and licensed for use in paediatric patients. An increase in research involving children, including neonates, will clearly be needed. Research studies vary greatly in their design, and drug interventions are not the only actions taken with the aim of improving outcomes for patients. We must be clear to defi ne what constitutes an interventional study: the use of new drugs, old drugs used for new indications, new defi nitions of tolerable oxygen-saturation levels, new ventilation strategies, or the introduction of minimal handling are all interventions. Many observational studies do not involve specifi c treatment regimens; however, such studies may involve infants having additional investigations, such as imaging or blood samples, and therefore infants might be exposed to additional non-clinical procedures that are not for their own benefi t. However, such studies will not lead to the problems of interactions between separate interventions. If an infant has been randomly assigned to the placebo or control arm of a trial, and therefore has not received that trial intervention, then is it acceptable for them to take part in other interventional studies and trials? To allow multiple trial recruitment to only one arm of a study would immediately introduce bias into the analyses of the fi rst trial. Some interventions used in a study setting at one unit may be standard practice at another. Hence, patients receiving the same treatment might be eligible for study enrolment at one unit but not at another. This is a particular problem in neonatal intensive care in which the lack of an evidence base for many treatments means that use of clinical interventions might vary not only between units but also among diff erent doctors caring for the same infant. The most rigorous way forward would be to ensure that potential interactions between interventions are assessed through proactive involvement of infants in multiple studies. Recruitment of infants into more than one trial does pose potential problems for the scientifi c integrity of the trials. The fi rst problem is the potential eff ect on the power of each trial. If both interventions work (ie, they both decrease the outcome being measured) and there is substantial overlap in recruitment between the two trials, then the event rate on which each trial’s sample size was based will be lowered, which will mean, for most trials, that the eff ect size that can be detected will have to increase (ie, loss of power). This problem will not matter in many situations because the probability of both trials being eff ective is low, overlap in recruitment is usually not very high, and, perhaps most importantly, the event rates used to calculate sample sizes are usually imprecise Lancet 2008; 372: 503–05