Comparing EQ-5D scores for comorbid health conditions estimated using 5 different methods

Area of Science:

• Health Economics
• Biostatistics
• Psychometrics

Background:

• Estimating health state utility values (HSUVs) for patients with multiple health conditions lacks a consensus method.
• Accurate HSUVs are crucial for cost-effectiveness analyses in healthcare.

Purpose of the Study:

• To evaluate the accuracy of five distinct methods for estimating HSUVs in a large dataset.
• To compare the performance of additive, multiplicative, minimum methods, adjusted decrement estimator, and linear regression models.

Main Methods:

• Utilized EQ-5D data from 41,174 participants in the Health Survey for England.
• Compared HSUVs generated by five different estimation techniques.
• Analyzed errors in HSUVs based on the estimation method and actual HSUV values.

Main Results:

• Additive and multiplicative methods tended to underestimate HSUVs, with errors increasing for higher values.
• Minimum and adjusted decrement estimator methods overestimated HSUVs, with errors increasing for lower values.
• The simple linear regression model yielded the most accurate HSUVs but showed a tendency to underpredict high scores and overpredict low scores.

Conclusions:

• Additive and minimum methods demonstrated poor performance in this dataset.
• The simple linear model requires external validation using EQ-5D and other preference-based measures.
• The multiplicative method, combined with univariate sensitivity analyses, is recommended for estimating HSUVs in comorbid conditions.