A threshold causal model for clinical trials with departures from intendedtreatment

Randomized clinical trials often are planned to study a specific interventi on. However, the collection of data on treatment actually received often re veals variable levels of treatment exposure (or 'dose') across subjects, du e to non-compliance or other reasons, This paper presents a new method, usi ng such 'dose' data as well as control group responses, to assess a causal dose-response relationship. The specific model utilizes a threshold functio n and incorporates a random effect term to allow for heterogeneous treatmen t responses among subjects. Further modelling of the random effects allows for reduction of error variance and control for potential confounders. The threshold dose is estimated using a residual variance criterion based on a transformed model. Estimates of standard errors and confidence intervals ar e obtained using a bootstrap procedure. The method is applied to data from an AIDS clinical trial. A simulation study demonstrates the adequacy of the threshold estimates for particular sample sizes and error variances. The l imitations of this essentially exploratory method, as well as some possible extensions, are discussed. Published in 1999 by John Wiley & Sons, Ltd. Th is article is a US Government Work and is in the public domain in the Unite d States.