Abstract
Background
Patient outcomes can depend on the treating centre, or health professional, delivering the intervention. Skills of health professionals improve with experience in delivery, meaning that outcomes may also be associated with changes in skill, or learning. Considering any potential difference in intervention delivery at trial design will ensure that any adjustments can be made, as appropriate, at the analysis. The objective of this work was to establish current practice for the allowance of clustering and learning effects in the design and analysis of randomised multicentre trials.
Methods
A ten-question survey was developed comprising open and closed questions that drew upon quotes from existing guidelines, references to relevant publications, and example trial scenarios. It was piloted and checked for face validity within the research team. All registered UK Clinical Research Collaborative registered Clinical Trials Units were invited to participate.
Results
Completed surveys were obtained from 44 of 50 Units. Clustering was managed through design by stratification, commonly by centre and less commonly by treatment provider. Most Units had allowed for learning by design through defining a minimum level of expertise for treatment provider (89%). One third of Units reported experience of expertise-based designs. The majority of Units indicated experience in adjusting for clustering during analysis, by centre or treatment provider, although approaches to doing so varied. Analysis of learning was rarely performed for the main analysis (n=1), although many Units reported approaches to consider such effects, such as sensitivity analyses. Responders provided insight behind the approaches used within their Unit and reasons for, or against, alternative approaches.
Conclusion
This survey identifies widespread awareness of the potential methodological challenges associated with the design and analysis of multicentre trials, although approaches used and opinions on these vary. These results suggest that variation in approaches used exists both across Units and within, suggesting that this decision can depend on the type of trial being conducted. Reasons for approaches were provided and approaches justified by responders. These results highlight the need for more agreement between triallists about how to best design and analyse trials of different types and/or further research to establish optimal methods.