ABSTRACT
Missing data can significantly affect the performance of predictive risk modeling, an important technique for developing medical guidelines. The two most commonly used strategies for managing missing data are to impute or delete values, and the former can cause bias, while the later can cause both bias and loss of statistical power. Several techniques designed to deal with missing data are described and applied to an illustrative example. These methods include complete-case analysis, available-case analysis, as well as single and multiple imputation. The causes of missing data should always be investigated and more data should be collected if possible. There are three types of missing data: missing completely at random, missing at random and missing not at random. A simple imputation includes mean, median and mode substitution is a relatively easy approach. In this approach, the missing values is replaced by a mean or median for numerical variable, and mode for categorical variable.
