Elizabeth J Conroy1, Anna Rosala-Hallas2, Jane M Blazeby3, Girvan Burnside4, Jonathan A Cook4, Carrol Gamble5. 1. Department of Biostatistics, University of Liverpool, A member of Liverpool Health Partners, Liverpool, UK. Electronic address: ejconroy@liverpool.ac.uk. 2. Department of Biostatistics, University of Liverpool, A member of Liverpool Health Partners, Liverpool, UK. 3. Centre for Surgical Research, Population Health Sciences, University of Bristol, Bristol, UK. 4. Centre for Statistics in Medicine, University of Oxford, Oxford, UK. 5. Department of Biostatistics, University of Liverpool, A member of Liverpool Health Partners, Liverpool, UK; North West Hub for Trials Methodology Research, University of Liverpool, Liverpool, UK.
Abstract
OBJECTIVE: The objective of this study was to provide insight into current practice in planning for, and acknowledging, the presence of learning and clustering effects, by treating center and surgeon, when developing randomized surgical trials. STUDY DESIGN AND SETTING: Complexities associated with delivering surgical interventions, such as clustering effects, by center or surgeon, and surgical learning should be considered at trial design. Main trial publications, within the wider literature, under-report these considerations. Funded applications, within a 4-year period, from a leading UK funding body were searched. Data were extracted on considerations for learning and clustering effects and the driver, funder, or applicant, behind these. RESULTS: Fifty trials were eligible. Managing learning through establishing predefined center and surgeon credentials was common. One planned exploratory analysis of learning within center, and two within surgeon. Clustering, by site and surgeon, was often managed through stratifying randomization, with 81% and 60%, respectively, also planning to subsequently adjust analysis. One-third of responses to referees contained funder led changes accounting for learning and/or clustering. CONCLUSION: This review indicates that researchers do consider impact of learning and clustering, by center and surgeon, during trial development. Furthermore, the funder is identified as a potential driver of considerations.
OBJECTIVE: The objective of this study was to provide insight into current practice in planning for, and acknowledging, the presence of learning and clustering effects, by treating center and surgeon, when developing randomized surgical trials. STUDY DESIGN AND SETTING: Complexities associated with delivering surgical interventions, such as clustering effects, by center or surgeon, and surgical learning should be considered at trial design. Main trial publications, within the wider literature, under-report these considerations. Funded applications, within a 4-year period, from a leading UK funding body were searched. Data were extracted on considerations for learning and clustering effects and the driver, funder, or applicant, behind these. RESULTS: Fifty trials were eligible. Managing learning through establishing predefined center and surgeon credentials was common. One planned exploratory analysis of learning within center, and two within surgeon. Clustering, by site and surgeon, was often managed through stratifying randomization, with 81% and 60%, respectively, also planning to subsequently adjust analysis. One-third of responses to referees contained funder led changes accounting for learning and/or clustering. CONCLUSION: This review indicates that researchers do consider impact of learning and clustering, by center and surgeon, during trial development. Furthermore, the funder is identified as a potential driver of considerations.