‘The randomized controlled trial: an excellent design, but can it address the big questions in neurodisability?’
Article first published online: 27 MAY 2010
© The Authors. Journal compilation © Mac Keith Press 2010
Developmental Medicine & Child Neurology
Volume 52, Issue 11, pages 1066–1067, November 2010
How to Cite
SUSSMAN, M. D. (2010), ‘The randomized controlled trial: an excellent design, but can it address the big questions in neurodisability?’. Developmental Medicine & Child Neurology, 52: 1066–1067. doi: 10.1111/j.1469-8749.2010.03708.x
- Issue published online: 27 MAY 2010
- Article first published online: 27 MAY 2010
SIR–I was as pleased as I was astounded to read the editorial by Peter Rosenbaum in DMCN wherein he questioned the applicability of the randomized controlled trial (RCT) design for clinical research in cerebral palsy (CP).1 As the author is the North American Editor of the journal and one of the preeminent clinicians/scientists in the field of CP (who has made huge contributions over the years), his notion is particularly shocking and carries with it a great deal of weight. As Professor Rosenbaum has so eloquently stated, the RCT is the very best way to answer a clinical question; however, this study design is unrealistic for most studies of interventions in children with CP.
The RCT is ideal for the study of issues where there is a large sample size, the population is relatively homogeneous, and the end points are objective and easily measurable. For example, in order to assess a medication to prevent myocardial infarction in males with high cholesterol, large numbers of patients can be gathered, the only intervention is either the presence or absence of the medication, and the end point would either be a cardiac event and/or death. Unfortunately, in CP our population is not homogeneous, the interventions and their follow-up therapies are very difficult to standardize, and it is almost impossible to obtain a large enough sample to have enough power to answer a question. For this reason investigators have resorted to multicenter studies which make studies even more complex and costly. Although there are other research models, one significant deterrent to using these has been journal editors and grant review panels who will not consider any design other than the RCT.
In the early 1990s my colleagues and I initiated a study comparing patients who underwent selective dorsal rhizotomy (SDR) along with physical therapy management, with those undergoing standard orthopaedic surgical treatment and physical therapy management. We anticipated that it would be almost impossible to randomize patients since families were unlikely to leave such an irreversible decision to the chance of randomization. Therefore, we chose a different model, where we selected patients who we felt would be appropriate for either intervention, and then let each family choose the intervention which the child would undergo. When the study was submitted for publication many journal editors would not even consider it since it did not use an RCT design. Although several randomized trials in this area were published, these compared SDR-treated patients with those undergoing no surgical treatment and followed them for a short period, after which the untreated group would have the option to proceed with the SDR. This model did not address what we felt was the critical question.
If one reviews the efforts of the clinical outcomes committee of the American Academy for Cerebral Palsy and Developmental Medicine, who have diligently reviewed the literature on a variety of different treatment issues, essentially all of these meta-analyses have come to the conclusion that the data is just not available to make a statement regarding the efficacy of the interventions they studied.
Therefore, it is clear that we need to reassess how clinical research is to be done in the future, so that we can begin to acquire information which will allow us to practice evidence-based medicine.
When we perform prospective studies, we should not be exclusively locked into an RCT design, but be allowed to utilize other research designs which will still be robust enough to provide useful and statistically powerful data, This approach may not be as rigorous, but it should be offset by increasing the numbers of participants and, therefore, the power of the study. We have been encouraged or mandated to use the International Classification of Functioning, Disability and Health model of the World Health Organization and consider multiple levels of function and participation, in addition to just structural changes induced by our intervention. I think this is an excellent goal, but perhaps at this point much too lofty. Children’s lives are influenced by many factors and a single intervention, no matter how powerful, may not influence function in the multiple other domains of their life. Outcome measures need to be focused on the area wherein the intervention lies, i.e. if it is a structural intervention we should be able to focus on structural outcomes and not be forced to look at the influence of our intervention on the child’s overall lifestyle.
How do we deal with the inadequate research which is available to us now in making clinical decisions? I think that we have to accept imperfect research at this point, and use what is available as best we can, recognizing the limitations of the information at hand.
Professor Rosenbaum has initiated what hopefully will become a significant dialogue among editorial boards and grant review panels regarding research in CP. Rather than being hypnotized by the ‘siren call of the RCT’, I think that investigators feel shackled by the grant review panels and editorial boards insisting that the RCT is the only valid approach to use. Unfortunately, the RCT is, as Professor Rosenbaum clearly states, not realistic in many circumstances and if we are to determine best treatments, we need to explore other models for our clinical research.