Missing data imputation in quality-of-life assessment: imputation for WHOQOL-BREF

Area of Science:

• Psychometrics
• Health Outcomes Research
• Data Science

Background:

• Missing data in health-related quality of life questionnaires, such as the WHO Quality of Life Abbreviated Questionnaire (WHOQOL-BREF), can impact study results.
• Imputation methods are often used to address missing data, but their effectiveness requires careful evaluation.

Purpose of the Study:

• To investigate the impact of different data imputation strategies on the WHOQOL-BREF.
• To compare imputation accuracy at both item and domain levels.
• To examine how missing data rates and the number of imputed items influence accuracy.

Main Methods:

• Employed both empirical analysis and simulation studies to assess imputation accuracy.
• Performed item-level and domain-level imputations using varying amounts of data.
• Simulated missing data at 2%, 5%, and 10% levels across 20 datasets per condition.

Main Results:

• The number of items used for imputation had a minimal effect on accuracy.
• Increasing proportions of missing data did not significantly decrease imputation accuracy.
• Extreme responses demonstrated the lowest imputation accuracy and poorest computational performance.

Conclusions:

• Maximize imputation accuracy by including as many items as possible within the same WHOQOL-BREF domain.
• Imputing items across different domains offers limited benefit.
• Exercise caution when interpreting imputed values derived from extreme responses.