Common pitfalls of perioperative clinical trials in response to ‘Barriers and enablers to recruiting participants within paediatric perioperative and anaesthetic settings: lessons learned from a trial of melatonin versus midazolam in the premedication of anxious children (the MAGIC trial)’

Summary

In this issue, Hyslop and colleagues discuss challenges leading to inadequate enrolment and premature termination of the MAGIC trial, a multicentre randomised controlled trial of melatonin vs midazolam premedication for anxiety among children. The authors should be commended for sharing their experiences, as they underscore common pitfalls of well-intended trials in the perioperative setting, including issues with study design and implementation.

Randomised controlled trials (RCTs) are considered the gold standard to generate data to guide clinical management. In paediatrics, trials are often more challenging to conduct, and enrolment goals are challenging to reach. Subsequently, clinical care, with emphasis on medication selection and dosing, is often guided by provider preference and clinical site norms. As historic context, in the USA, the 2003 Pediatric Research Equity Act (PREA) and 2012 Food and Drug Administration (FDA) Safety and Innovation Act permitted the FDA to mandate paediatric studies for medications likely to be used in children.¹ Despite regulatory support, an estimated the majority FDA-approved medications are still not approved for use in paediatric populations.² Significant barriers persist and impair the ability to complete paediatric trials and disseminate results.

Although the MAGIC trial's enrolment efforts were hampered by pandemic-era restrictions and medication shortages, several of the authors' key findings underscore two fundamental challenges when conducting any trial.³ Firstly, the study question itself must be of sufficiently high interest to those at the enrolling sites that they consider engagement with the trial a high priority. Secondly, site infrastructure and workflows vary, highlighting the importance of thorough assessment of each site before commencing the trial and ideally during design of the trial.

Preprocedural anxiety is common among children. Effective pharmacological and non-pharmacological interventions exist with varying degrees of efficacy, tolerability, and unwanted effects. Midazolam, a benzodiazepine, has historically been the mainstay for paediatric premedication despite being associated with paradoxical reactions and increased postoperative sedation.⁴ Data support that non-pharmacological interventions, including watching cartoons on a tablet or engaging in virtual reality, may be as effective as midazolam without side-effects for many children.^5,6 The authors chose to study midazolam as the standard of care vs melatonin, a synthetic hormone involved in regulating circadian rhythm with associated sedation. Melatonin use for preprocedural anxiety has been extensively studied among adults. A Cochrane systematic review concluded that it was effective for preoperative anxiety vs placebo and found no evidence of difference compared with midazolam.⁷ A small Italian RCT comparing midazolam vs melatonin among paediatric patients did not identify a difference in preoperative anxiety or acceptance of induction but conceded that the low numbers significantly reduced the power of the trial and concluded that larger trials were needed.⁸ This trial may have been the impetus for the MAGIC trial.

Trials are large enterprises with numerous stakeholders. Success requires all stakeholders to be engaged champions for the trial. In particular, stakeholders must agree that the research question is a question that must be answered. Those planning a trial must be sure they are asking the right question: convincing a funder to fund a trial, or an ethics committee to approve a trial, is not the same as convincing your peers. Some aspects of the failure of the MAGIC trial suggest that not all sites involved were sufficiently engaged to overcome the inevitable hurdles that appear when doing trials. Did the investigators spend enough time ‘socialising’ the question during the design phase? Were they to have done this, might the problems later encountered have been identified, and the study abandoned?

Inadequate stakeholder interest in the study question is directly related to another common pitfall: incorporation of clinical setting practices and preferences during the study design phase. As one example, unanticipated variation in anaesthetist equipoise was cited as a barrier to enrolment. In hindsight, authors recognised that some sites had local policies that guided premedication selection and that other sites had increased preference for dexmedetomidine for preprocedural sedation. Authors note that they could have considered undertaking a survey to understand clinical practices across sites, although ideally, this would have been a consideration when assessing study feasibility during the pilot. Another example relates to time sensitivity within pressured environments. During planning, investigators could have documented the preoperative workflow at each of the 20 study hospitals, including verification that adequate space existed for patient intake and for time needed to complete study procedures.

It is possible that a more comprehensive pilot study may have identified some of these barriers. Many consider that pilot studies are essential before large trials, a sentiment shared by many funding bodies. Pilot studies where a smaller sample are enrolled, randomised, treated, and followed-up can help clarify and quantify suspected issues and can also uncover unexpected issues. Ideally, the pilot study is completed before applying for funding for the definitive trial, which may allow for more comprehensive modifications. In the case of the MAGIC trial, one might also wonder about the role that escalation of commitment may have played, as investigators pressed on despite a pilot that failed to recruit the target number of pilot participants. In the context of this failed pilot, and focus on a single medication supply issue, was it any surprise that the larger trial fell short of recruitment targets?

We commend the authors for candidly sharing their experiences, and it is our hope that these ‘lessons learned’ will allow other investigators to avoid pitfalls that often affect trials. Key stakeholders should be identified early in the planning process to verify interest in the study question and to understand logistical challenges at each potential site to inform study design. Pilot studies continue to be recognised as an essential step to identify potential study design issues before launching a large trial, and completion before applying for definitive funding should be strongly encouraged.

Declarations of interest

The authors declare that they have no conflicts of interest.

Handling Editor: Dr Susan M. Goobie