Study design with staggered sampling times for evaluating sustained unresponsiveness to peanut sublingual immunotherapy.
Monica ChaudhariEdwin H KimPrabhashi W Withana GamageChristopher S McMahanMichael R KosorokPublished in: Statistics in medicine (2018)
In this work, we delineate an altered study design of a pre-existing clinical trial that is currently being implemented in the Department of Pediatrics at the University of North Carolina at Chapel Hill. The purpose of the ongoing investigation of the desensitized pediatric cohort is to address the effectiveness of sublingual immunotherapy in achieving sustained unresponsiveness (SU) as assessed by repeated double-blind placebo-controlled food challenges (DBPCFC). With scarce published literature characterizing SU, the length of time off-therapy that would represent clinically meaningful benefit remains undefined. We use the new design features to assess time to loss of SU, an important efficacy endpoint, that to our knowledge, no prior study has investigated. Our work has two-fold objectives: first is to propose and discuss aspects of the altered design that would allow us to study SU and second is to explore methodology to evaluate the time to loss of SU and its association with risk factors in the context of the data originating from the trial. The salient feature of the new design is the allocation scheme of study subjects to staggered sampling timepoints when a subsequent DBPCFC is administered. Due to this feature, the time to loss of SU is either left or right censored. Additionally, some participants at study entry fail the DBPCFC, leading to what can be construed as an instantaneous failure. Through in-depth numerical studies, we examine the performance and power of a recently proposed mixture proportional hazards model specifically designed for the analysis of interval-censored data subject to instantaneous failures.