Abstract

Objectives

The Children’s Dermatology Life Quality Index (CDLQI) assesses health-related quality of life in pediatric patients with atopic dermatitis. However, it cannot be used directly in economic evaluations. We developed algorithms to map CDLQI scores to EQ-5D-Y-3L health utility values, relevant for estimating quality-adjusted life-years.

Methods

Data were prospectively collected by recruiting pediatric patients with atopic dermatitis from a specialty children’s hospital in Singapore. Children aged 4 to 12 completed the EQ-5D-Y-3L and CDLQI questionnaires. For direct mapping approach, 3 model specifications were tested (CDLQI individual item scores, model 1; CDLQI domain scores, model 2; and CDLQI total score, model 3) using 4 regression techniques: ordinary least squares, general linear regression, multivariable fractional polynomials, and beta regression. A 4-fold cross-validation evaluated predictive performance using pooled root mean squared error and mean absolute error (MAE). For indirect response mapping, the ordered logit model predicted EQ-5D-Y-3L responses from CDLQI item scores.

Results

We extracted the data of 250 patients. Beta regression had the lowest MAE (0.0490) for item-based model 1, whereas ordinary least squares performed best for domain-based model 2 (MAE: 0.0499) and total score model 3 (MAE: 0.0532). However, the mapping algorithms overpredicted lower observed utility scores. The indirect mapping model achieved similar performance, with a MAE of 0.0452.

Conclusions

We mapped CDLQI to EQ-5D-Y-3L utilities using direct and indirect approaches with good predictive accuracy. These algorithms will allow studies with CDLQI data to calculate health utility estimates for use in atopic dermatitis-related economic evaluations.