Correcting for non-compliance in randomized trials: An application to the ATBC Study

Different methods for estimating the effect of treatment actually received in a longitudinal placebo-controlled trial with non-compliance are discusse d. Total mortality from the ATBC Study is used as an illustrative example. In the ATBC Study some 25 per cent of the participants dropped out from act ive follow-up prior to the scheduled end of the study. The 'intention-to-tr eat' analysis showed an increased death risk in the beta-carotene arm when compared with the no beta-carotene arm. Owing to considerable non-complianc e it is also of interest to estimate the effect of beta-carotene actually r eceived. We use a simple model for the treatment action and discuss three m ethods for estimation of the treatment effect under the model - the 'intent ion-to-treat' approach, the 'as-treated' approach and the g-estimation appr oach. These approaches are compared in a simulation study under different s ettings for non-compliance. Finally the data from the ATBC Study are analys ed using the proposed methods, Copyright (C) 1999 John Wiley & Sons, Ltd.