Why healthy life expectancy estimates diverge: insights from survey-based, administrative, and model-based metrics across Japanese prefectures, 2001–2019: a comparative ecological study

Healthy life expectancy (HALE) can be measured using different approaches, yet estimates may diverge. Understanding divergence sources could help identify which health dimensions current systems capture or miss. However, systematic subnational comparisons remain scarce.

We conducted a comparative ecological study across Japan’s 47 prefectures from 2001 to 2019. Three Ministry of Health, Labour and Welfare (MHLW) metrics—disability-free life expectancy based on activity limitation (DFLE-AL) and life expectancy in subjective health (LE-SH) derived from national surveys, and disability-free life expectancy based on activities of daily living (DFLE-ADL) derived from long-term care insurance records—were obtained from MHLW. Model-based HALE from the Global Burden of Disease (GBD) 2023 Study was extracted for corresponding years. We assessed cross-sectional concordance and examined associations with disease burden, risk factors, and socioeconomic indicators using fixed-effects panel regression.

Over 2001–2019, DFLE-AL and LE-SH correlated with GBD HALE among males (r = 0.33 to 0.65) but weakly among females (r =  − 0.11 to 0.32); DFLE-ADL showed concordance for both sexes (r = 0.66 to 0.96). GBD HALE exceeded MHLW estimates in females (0.53 years for DFLE-AL, 0.23 for LE-SH) but not males (− 0.18, − 0.14). In fixed-effects analysis of MHLW-minus-GBD differences, mental disorders showed negative associations with DFLE-AL and LE-SH differences among males; high BMI and musculoskeletal disorders showed positive associations with DFLE-AL differences. Among females, population ageing showed positive associations with DFLE-AL, LE-SH, and DFLE-ADL differences.

MHLW metrics and GBD HALE capture complementary health dimensions and should not be treated as interchangeable. The consistent sex difference in concordance indicates that divergence between HALE metrics reflects differences in which health dimensions are emphasised—self-reported health states, objectively assessed functional status, or disease-based disability modelling—rather than a single, uniform construct of population health.