Does linear equating improve prediction in mapping? Crosswalking MacNew onto EQ-5D-5L value sets

Area of Science:

• Health Economics
• Psychometrics
• Biostatistics

Background:

• Preference-based measures are crucial for calculating Quality-Adjusted Life Years (QALYs) in economic evaluations.
• Mapping algorithms can estimate utilities from disease-specific measures when preference-based measures are unavailable.

Purpose of the Study:

• To develop and validate mapping algorithms connecting the MacNew Heart Disease Quality of Life Questionnaire (MacNew) to the English and US EQ-5D-5L value sets.
• To enable the estimation of utilities for MacNew data, facilitating its use in economic evaluations.

Main Methods:

• Recruited individuals with heart disease from six countries.
• Applied parametric and non-parametric statistical techniques to develop mapping algorithms.
• Selected optimal algorithms based on RMSE, MAE, CCC, and r-squared, with cross-validation for generalizability.

Main Results:

• A one-inflated beta regression model demonstrated superior performance for both English and US EQ-5D-5L value sets.
• The preferred model explained 59% (English) and 60% (US) of the variance.
• Cross-validation confirmed model generalizability, with linear equating yielding equivalent predicted and observed values.

Conclusions:

• The developed mapping function accurately predicts utilities for MacNew data using recent English and US EQ-5D-5L value sets.
• This enables studies using the MacNew to be incorporated into cost-utility analyses.
• Facilitates consistent comparison of health interventions across the healthcare system.