Missing Data Essentials Part 2: Statistical Approaches to Missing Data in Cardiovascular Studies

Missing data are common in cardiovascular nursing and allied health research. Although conventional methods of handling missing data may appear straightforward, they have significant limitations and can introduce bias. In contrast, principled methods, including multiple imputation with and without chained equations and full information maximum likelihood estimation, offer more robust ways to mitigate bias and make full use of available information. These methods of handling missingness can be applied to missing independent and dependent variables, and this methods paper provides worked examples to illustrate each approach.