RESUMO
Sequentially randomized designs are commonly used in biomedical research, particularly in clinical trials, to assess and compare the effects of different treatment regimes. In such designs, eligible patients are first randomized to one of the initial therapies, then patients with some intermediate response (e.g. without progressive diseases) are randomized to one of the maintenance therapies. The goal is to evaluate dynamic treatment regimes consisting of an initial therapy, the intermediate response, and a maintenance therapy. In this article, we demonstrate the use of pattern-mixture model (commonly used for analyzing missing data) for estimating the effects of treatment regimes based on familiar survival analysis techniques such as Nelson-Aalen and parametric models. Moreover, we demonstrate how to use estimates from pattern-mixture models to test for the differences across treatment regimes in a weighted log-rank setting. We investigate the properties of the proposed estimators and test in a Monte Carlo simulation study. Finally we demonstrate the methods using the long-term survival data from the high risk neuroblastoma study.
RESUMO
In this companion article to "Dynamic Regime Marginal Structural Mean Models for Estimation of Optimal Dynamic Treatment Regimes, Part I: Main Content" [Orellana, Rotnitzky and Robins (2010), IJB, Vol. 6, Iss. 2, Art. 7] we present (i) proofs of the claims in that paper, (ii) a proposal for the computation of a confidence set for the optimal index when this lies in a finite set, and (iii) an example to aid the interpretation of the positivity assumption.