A simple and powerful method for the estimation of intervention effects on serological endpoints using paired interval-censored data.

ABSTRACT

Clinical trials often use a binary "fold increase" endpoint defined according to the ratio of interval-censored measurement at end-of-study to that at baseline. We propose a simple yet principled analytic approach based on the linear mixed-effects model for interval-censored data for the analysis of such paired measurements. Having estimated the model parameters, the risk ratio can be estimated by explicit composite estimand and the variance is estimated using the delta method. The estimation can be implemented using the existing procedures in popular statistical software. We use antibody data from the Chloroquine for Influenza Prevention Trial for illustration.