Does Selecting Covariates Using Factor Analysis in Mapping Algorithms Improve Predictive Accuracy? A Case of...

Area of Science:

• Health Economics
• Biostatistics
• Psychometrics

Background:

• Selecting covariates for mapping non-utility measures to utilities is crucial for health economics research.
• Exploratory factor analysis (EFA) is one method, but its impact on predictive accuracy is unclear.

Purpose of the Study:

• To evaluate if exploratory factor analysis (EFA) improves the predictive ability of algorithms mapping non-utility measures to utilities.
• To compare EFA with other covariate selection methods in predicting health utilities.

Main Methods:

• The Women's Health Questionnaire (WHQ-23) was mapped onto the EQ-5D-5L and SF-6D utility instruments.
• Covariate selection methods included stepwise regression (SW), including all covariates, multivariable fractional polynomial (MFP), and EFA.
• Predictive accuracy was assessed using mean absolute error, root mean squared error, and correlation.

Main Results:

• SW, "Include all," and MFP methods yielded the best predictions for EQ-5D-5L and SF-6D.
• Root mean squared error ranged from 0.0762-0.1434, and mean absolute error ranged from 0.0590-0.0924.
• Exploratory factor analysis (EFA) was less effective than the other covariate selection methods.

Conclusions:

• Regression models using SW, "Include all," and MFP are effective for predicting health utilities from the WHQ-23.
• These methods provide a reliable way to derive utility values for health economic evaluations.