APPLICATION OF MAPPING ALGORITHMS FOR THE PEDIATRIC QUALITY OF LIFE INVENTORY GENERIC CORE SCALES (PEDSQL GCS) TO THE EUROQOL 5 DIMENSIONS (EQ-5D) UTILITIES IN CHILDHOOD CANCER SURVIVORS IN JAPAN
Author(s)
Masafumi Kato, MSc.1, Hiroko Fukushima, M.D., Ph.D.2, Reiko Okubo, M.D., Ph.D.3, Masahide Kondo, M.D., Ph.D.3, Kyoko Hidaka, Ph.D.4, Hidetoshi Takada, M.D., Ph.D.2;
1University of Tsukuba, Doctoral Program in Human Care Science, Graduate School of Comprehensive Human Sciences, Ibaraki, Japan, 2University of Tsukuba Hospital, Department of Pediatrics, Ibaraki, Japan, 3University of Tsukuba, Department of Health Care Policy and Health Economics, Institute of Medicine, Ibaraki, Japan, 4Mejiro University, Department of Psychology, Tokyo, Japan
1University of Tsukuba, Doctoral Program in Human Care Science, Graduate School of Comprehensive Human Sciences, Ibaraki, Japan, 2University of Tsukuba Hospital, Department of Pediatrics, Ibaraki, Japan, 3University of Tsukuba, Department of Health Care Policy and Health Economics, Institute of Medicine, Ibaraki, Japan, 4Mejiro University, Department of Psychology, Tokyo, Japan
OBJECTIVES: To assess the applicability of the mapping algorithms by Khan et al. (2014) to generate EQ-5D utilities based on the EQ-5D Three-Level, Youth Version (EQ-5D-Y-3L) from the PedsQL GCS total score in childhood cancer survivors in Japan, and explore its extension to the EQ-5D Five-Level (EQ-5D-5L), which is administered to younger adults than EQ-5D-Y-3L.
METHODS: We used existing mapping algorithms in which the PedsQL GCS total score was mapped to the EQ-5D-Y-3L utility. To assess prediction errors, the Mean Squared Error (MSE) and Mean Absolute Error (MAE) between the predicted and original utilities were calculated for EQ-5D-Y-3L and the EQ-5D-5L, respectively. All analyses were conducted separately for both EQ-5D-Y-3L and PedsQL GCS complete respondents and both EQ-5D-5L and PedsQL GCS complete respondents.
RESULTS: In the EQ-5D-Y-3L and PedsQL GCS complete respondents (N=36), the means (standard deviations, SD) of the original and predicted utilities across existing algorithms were 0.92 (0.09) and 0.91-0.93 (0.09 for all algorithms), respectively. The MSE and MAE were approximately 0.01 and 0.07, respectively, with no significant differences among the algorithms, and they were similar to the original study results. In addition, in the EQ-5D-5L and PedsQL GCS complete respondents (N=75), the means (SD) of the original and predicted utilities across existing algorithms were 0.92 (0.12), and 0.87-0.90 (0.11 for all algorithms), respectively. The MSE and MAE were approximately 0.01 and 0.08-0.09, respectively. These results were similar to those of the EQ-5D-Y-3L and PedsQL GCS complete respondents.
CONCLUSIONS: The prediction errors for both the EQ-5D-Y-3L and EQ-5D-5L were as small as those in the original study that developed the algorithms and considered acceptable. Our study is the first to show that the mapping algorithms proposed by Khan et al. (2014) are applicable to childhood cancer survivors in Japan.
METHODS: We used existing mapping algorithms in which the PedsQL GCS total score was mapped to the EQ-5D-Y-3L utility. To assess prediction errors, the Mean Squared Error (MSE) and Mean Absolute Error (MAE) between the predicted and original utilities were calculated for EQ-5D-Y-3L and the EQ-5D-5L, respectively. All analyses were conducted separately for both EQ-5D-Y-3L and PedsQL GCS complete respondents and both EQ-5D-5L and PedsQL GCS complete respondents.
RESULTS: In the EQ-5D-Y-3L and PedsQL GCS complete respondents (N=36), the means (standard deviations, SD) of the original and predicted utilities across existing algorithms were 0.92 (0.09) and 0.91-0.93 (0.09 for all algorithms), respectively. The MSE and MAE were approximately 0.01 and 0.07, respectively, with no significant differences among the algorithms, and they were similar to the original study results. In addition, in the EQ-5D-5L and PedsQL GCS complete respondents (N=75), the means (SD) of the original and predicted utilities across existing algorithms were 0.92 (0.12), and 0.87-0.90 (0.11 for all algorithms), respectively. The MSE and MAE were approximately 0.01 and 0.08-0.09, respectively. These results were similar to those of the EQ-5D-Y-3L and PedsQL GCS complete respondents.
CONCLUSIONS: The prediction errors for both the EQ-5D-Y-3L and EQ-5D-5L were as small as those in the original study that developed the algorithms and considered acceptable. Our study is the first to show that the mapping algorithms proposed by Khan et al. (2014) are applicable to childhood cancer survivors in Japan.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
P27
Topic
Patient-Centered Research
Topic Subcategory
Patient-reported Outcomes & Quality of Life Outcomes
Disease
SDC: Oncology