Missing Values

Abstract

Missing data are a common problem in medical research. We may encounter missing values of predictor values (X) and for the outcome (y) that we want to predict. Traditional complete case analysis suffers from inefficiency, selection bias of subjects, and other limitations when developing a prediction model. We briefly review the theoretical background on mechanisms of missingness of predictor values and how these may affect prediction models. We further concentrate on imputation methods as a solution, where a completed data set is created by filling in missing values for the statistical analysis. Special attention is given to the specification of an imputation model, which is the essential step in imputation. Multiple imputation is a method is to generate completed data sets multiple times, while single imputation is more straightforward and may be sufficient for some prognostic research questions.