Large sample results for frequentist multiple imputation for Cox regression with missing covariate data (Q778878)

The aim of the present paper is to study the properties of multiple imputation estimators from a compatible model. In particular, the authors following the idea by \textit{J. Bartlett} et al. (see [``Multiple imputation of covariates by fully conditional specification: accommodating the substantive model'', Stat. Methods Med. Res. 24, No. 4, 462--487 (2015; \url{doi:10.1177/0962280214521348})]) according to which such kind of imputations may be generated using rejection sampling and focusing on what \textit{A. A. Tsiatis} (see [Semiparametric theory and missing data. New York, NY: Springer (2006; Zbl 1105.62002)]) mean by frequentist multiple imputation managed to show how the combination of multiple imputations from a compatible model with suitably estimated parameters and the usual Cox regression estimators leads to consistent and asymptotically Gaussian estimators of both the finite-dimensional regression parameter and the infinite-dimensional cumulative baseline hazard parameter. Their results in connection with simulation studies on the finite sample performance of the proposed estimators as well as with an application to a study on survival after treatment for liver cirrhosis indicate the necessary justification for drawing correct statistical inference in the case of using multiple imputation in Cox regression with missing data leading to the increase of precision.