Double Sampling and Semiparametric Methods for Informatively Missing Not at Random Data

Missing data arise in almost all applied settings and are ubiquitous in electronic health records (EHR). When data are missing not at random (MNAR) with respect to measured covariates, sensitivity analyses are often considered. These solutions, however, are unsatisfying in that they are not guaranteed to yield actionable conclusions. Motivated by an EHR-based study of long-term outcomes following bariatric surgery, we consider the use of double sampling as a means to mitigate MNAR outcome data when the statistical goals are estimation and inference regarding causal contrasts based on mean counterfactuals. We describe identification assumptions and derive efficient and robust estimators of the average causal treatment effect under a nonparametric model as well as under a model assuming the missing outcomes were, in fact, initially missing at random (MAR). We compare these in simulations to an approach that adaptively estimates based on evidence of violation of the MAR assumption. Finally, we show how the methods can be extended to: (i) estimation/inference regarding causal quantile treatment effects; and (ii) hypothesis testing regarding MNAR.

Host: Bo Li