January 1, 1992Biometrika
Estimation of the time-dependent accelerated failure time model in the presence of confounding factors
Authors
Discussion
Loading...
Member takes
Implication
Methodological study demonstrates consistent estimation of time-dependent accelerated failure time models with time-varying confounding, highlighting robust causal inference in survival analysis.
Key Points
- To develop valid semiparametric estimation methods for time-dependent accelerated failure time models in survival studies subject to time-varying confounding.
- Formulated semiparametric accelerated failure time models to evaluate exposure effects on survival under dynamic, time-dependent treatment regimens.
- Constructed G-estimation methodology within structural nested failure time models to account for time-varying confounders affected by prior exposure.
- Demonstrated that standard survival analysis techniques yield biased causal estimates when intermediate variables act simultaneously as time-varying confounders.
- Derived consistent and asymptotically normal estimators that accurately recover treatment effects under conditional exchangeability assumptions.
Cite This Study
James M. Robins (1992) studied this question.