An enhanced three-stage design with trend analysis for allergen immunotherapy trials

• Published in: PloS one 2023-09, Vol.18 (9), p.e0291533
• Main Authors: Tang, Xinyu, Rabin, Ronald L, Yan, Lihan K
• Format: Article
• Language: English
• Subjects: Allergens; Allergic reaction; Allergy; Alzheimer's disease; Analysis; Bias; Biology and Life Sciences; Clinical trials; Complications and side effects; Design; Design analysis; Engineering and Technology; Evaluation; Health aspects; Hypotheses; Immunotherapy; Medicine and Health Sciences; Patient outcomes; Physical Sciences; Progressions; Research and Analysis Methods; Statistical analysis; Statistics; Trend analysis; Trends
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0291533

We previously introduced a three-stage design and associated end-of-stage analyses for allergen immunotherapy (AIT) trials. End-of-stage differences alone may not provide a fuller picture of Stages 2 and 3 effects because they may depend upon stage-specific durations. Therefore, we introduce an additional trend analysis to evaluate the difference in progression curves of two groups over the entire stage. Results from such analysis are used to inform persistence of end-of-stage benefit and thus provide evidence for stagewise effects beyond the study periods. We jointly apply end-of-stage and trend analyses to support the enhanced three-stage design to determine treatment response over time and sustained response to AIT. A simulation study was performed to illustrate the statistical properties (bias and power) of trend analyses under varying statistical missing mechanisms and effect sizes. The extent of bias depended on the missing mechanism and magnitude. Powers were largely driven by effect and sample sizes as well as pre-specified success margins, particularly of relative trend. As an illustration, assuming relative treatment differences of 25–30%, stagewise dropout rate of 15%, and parallel outcome progressions, a sample size of 200 per group may achieve 97% power to demonstrate a treatment effect and 53% power to demonstrate a sustained effect post-treatment. Trend analysis supplements the end-of-stage analysis to enhance the statistical claims of stagewise effects. Inferential statistics support our proposed trend analysis for evaluating benefits of AIT over time and inform clinical understanding and decisions.