統計学輪講（第23回）

日時      2012年01月10日(火)    15時00分～15時50分
場所      経済学部新棟３階第３教室
講演者    篠崎　智大 (医D1)
演題      Estimating controlled direct effects for time-varying treatments
using structural nested mean models

概要
Estimating direct effects without bias requires that two fundamental
assumptions hold, that is, the absence of unmeasured confounding for
treatment and outcome, and the intermediate and outcome. Even if the above
two assumptions hold, one cannot estimate direct effects via standard
methods such as stratification or regression modeling if treatment affects
confounding factors, namely, time-dependent confounding problem arises.
Sequential g-estimation method for the structural nested mean models has
been developed for estimating controlled direct effects (i.e., direct
effects controlling intermediates to be fixed at a specified level) in the
presence of time-dependent confounders. In this article, we extend this
method for longitudinal data with time-varying treatments and repeatedly
measured intermediates. Simulation studies showed that usual regression
approaches were heavily biased in the presence of time-dependent
confounders, but our sequential g-estimator remained unbiased. The proposed
method was applied to data from a large primary prevention trial for
coronary events in which pravastatin was used for lowering cholesterol. We
demonstrated in the analyses what extent of the benefit for coronary
prevention experienced by pravastatin could be attributed to the effect of
treatment on the cholesterol levels.