Estimating healthy life years without activity limitations using medical claims data in Japan

Healthy life years are estimated from national population surveys that assess limitations in daily activities. However, the infrequent timing and limited sample sizes hinder the timely and detailed assessment of regional health conditions. We aimed to develop a novel method to estimate municipal-level healthy life years without activity limitations by using medical claims data.

We analysed medical claims data in Kyoto Prefecture, Japan. Outpatient data from May to July in 2016 and 2019 (n = 1 489 920) were used for development and data from the same period in 2022 (n = 739 236) were used for evaluation. A total of 5743 diagnostic codes were aggregated into 40 disease categories. A machine-learning model was employed to produce the probability of activity limitation, which was calibrated by using data from the Comprehensive Survey of Living Conditions and applied to derive age-specific prevalence rates. Healthy life years at the municipal level as of June 2022 were then estimated by incorporating a life table.

We observed variation in healthy life years among municipalities: 72.1 years across all regions for males, ranging from 67.3 to 75.2 years, and 75.8 years for females, ranging from 71.3 to 77.6 years. Regional disparities were also noted in the prevalence of diseases associated with activity limitations.

This study provides the first robust, scalable method to estimate healthy life years without activity limitations at the municipal level by using real-world administrative data. Timely monitoring of regional healthy life years will support targeted health-promotion policies and contribute to reducing health disparities.