Statistical issues in clinical trials that involve the double-blind, placebo-controlled food challenge

Abstract

The double-blind, placebo-controlled food challenge is a rigorous tool that has become popular for evaluating adverse reactions to foods. The standard use of the double-blind, placebo-controlled food challenge has been to document food allergies for individual patients, but it recently has been gaining acceptance as a procedure for investigating the effectiveness of therapies to prevent/minimize food-induced anaphylaxis. The purpose of this study was to describe the statistical design and analysis issues for clinical trials that use the double-blind, placebo-controlled food challenge in measuring sensitivity to food allergens. Nonparametric tests for within-group and between-group comparisons are described, as well as a discrete-time survival analysis. The statistical methods are applied to simulated data from a clinical trial that compares control therapy and experimental therapy groups. The results indicate that the experimental therapy is significantly better than control in improving the tolerance to peanut flour in patients with peanut allergy. Although simple nonparametric tests for within-group and between-group comparisons are easy to apply, a discrete-time survival analysis provides the best approach because of its flexibility in accounting for important independent variables (regressors) and longitudinal data. Statistical software packages can be adapted to perform such analyses.