MatchThem:: Matching and Weighting after Multiple Imputation

Balancing the distributions of confounders across exposure levels in an observational study through matching or weighting is accepted method to control for confounding due these variables when estimating association between and outcome reduce degree dependence on certain modeling assumptions. Despite increasing popularity practice, procedures cannot be immediately applied datasets with missing values. Multiple imputation data a popular approach account values while preserving number units dataset accounting uncertainty However, best our knowledge, there no comprehensive software that can easily implemented multiply imputed datasets. In this paper, we review problem suggest framework map out 5 actions as well practices assess balance after weighting. We also illustrate approaches using companion package R, MatchThem.