Part II: Adaptive designs in pediatric clinical trials: specific examples, comparison with adult trials and a discussion for the child health community

Abstract

Abstract Background In Part I of this review, we outlined the study characteristics and methodologies utilized in adaptive clinical trials reported in the literature from 2010–2020. Herein, the second part of this analysis presents a secondary analysis of the trials captured within this timeframe that enrolled children. Methods This analysis seeks to generate an evidence base that can inform practical recommendations that can shape the design, ethical considerations, and training on methods and reporting for pediatric adaptive design (AD) trials. We performed a secondary analysis of 43 AD trials involving children and compared the study characteristics with those of adult AD trials. Results There were one to five arms in these pediatric AD trials, with the most commonly reported adaptive methods being dose modifications (20/43, 46.5%) in dose-finding trials, followed by continual reassessment method (CRM), a model-based Bayesian design, reported in 20 studies (46.5%), and adaptive randomization (9/43, 20.9%). The frequentist framework (68.8%) was most commonly used for statistical analysis. Reporting indicated a lack of patient and parent engagement with clinicians and scientists during the clinical trial (CT) planning process and was only reported in 1 of the reviewed studies (1/43, 2.32%). Conclusion We reviewed examples of the most common types of adaptive designs used in pediatric trials and compared the methods used with adults’ trials. Against this background, we provide an overview of the different statistical approaches used and highlight the ethical considerations. The results of this review could serve as a reference for the development of guidelines and training materials to guide clinical researchers and trialists in the use of pediatric adaptive clinical trials. Study protocol registration: DOI:10.1186/s13063-018-2934-7