All Times EDT
Keywords: Robust causal estimate of treatment effect, clinical trials
Causal estimands of interest are treatment effects from comparing the endpoint for the same patients, or for similar groups of patients, on different treatments. Motivated by a trial to compare 3-year relapse-free survival (RFS) between locally treated high-risk ocular melanoma patients on adjuvant combination immunotherapy vs a matched contemporaneous external control population, we develop causal estimates for the appropriate estimand via improving the doubly robust estimators (DREs) which have been developed for various designs in the last two decades. The approach combines propensity score and outcome models of the confounding variables. It yields unbiased estimator of the target parameter if at least one of the two models is correctly specified, a desirable property and an improvement on the inverse propensity score weighted estimate. However, it is difficult to know what the correct model could be and both propensity score and outcome models may be incorrectly specified. Importantly, it is known that DRE may give estimates with large bias and variance, even when the propensity and/or outcome models are mildly misspecified. To resolve this issue, we propose an enhanced DRE (eDRE) method utilizing semiparametric models with nonparametric monotone link functions for both the propensity score and the outcome models. We estimate the model with an iterative algorithm and then derive the asymptotic properties of the eDREs. We show by simulation that eDRE has reduced bias and increased efficiency even when both models are misspecified. The method is then applied to analyzing two real studies. Furthermore, fundamentally the causal inference is to estimate the mean of a partially observed variable. The enhanced DRE approach thus yields an efficient alternative to imputation. We will explore the benefit of this approach for several causal estimands of interest, e.g. those where endpoint data is missing after rescue medication. The work is a collaboration with Ao Yuan.