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

Area of Science:

• Cardiovascular Nursing
• Allied Health Research
• Biostatistics

Background:

• Missing data are prevalent in cardiovascular nursing and allied health research.
• Conventional methods for handling missing data have limitations and can introduce bias.
• There is a need for robust methods to address missing data in health research.

Purpose of the Study:

• To present principled methods for handling missing data in cardiovascular nursing and allied health research.
• To illustrate the application of multiple imputation and full information maximum likelihood estimation.
• To provide worked examples for applying these advanced statistical techniques.

Main Methods:

• Multiple Imputation (MI) with chained equations
• Full Information Maximum Likelihood Estimation (FIML)
• Application to missing independent and dependent variables

Main Results:

• Principled methods mitigate bias more effectively than conventional approaches.
• Multiple imputation and FIML utilize available data more fully.
• Worked examples demonstrate practical implementation of these techniques.

Conclusions:

• Advanced statistical methods like MI and FIML are superior for handling missing data in health research.
• These methods reduce bias and improve the validity of research findings.
• Researchers should adopt these principled approaches for more accurate results.