Long-term mortality in nationwide cohorts of childhood-onset type 1 diabetes in Japan and Finland

Abstract

Objective: This study compares mortality from type 1 diabetes in Japan and Finland and examines the effects of sex, age at diagnosis, and calendar time period of diagnosis on mortality.

Research design and methods: Patients with type 1 diabetes from Japan (n = 1,408) and Finland (n = 5,126), diagnosed from 1965 through 1979, at age <18 years, were followed until 1994. Mortality was estimated with and without adjustment for that of the general population to assess absolute and relative mortality using Cox proportional hazard models.

Results: Overall mortality rates in Japan and Finland were 607 (95% CI 510-718) and 352 (315-393), respectively, per 100,000 person-years; standardized mortality ratios were 12.9 (10.8-15.3) and 3.7 (3.3-4.1), respectively. Absolute mortality was higher for men than for women in Finland, but relative mortality was higher for women than for men in both cohorts. Absolute mortality was higher in both cohorts among those whose diabetes was diagnosed during puberty, but relative mortality did not show any significant difference by age at diagnosis in either cohort. In Japan, both absolute and relative mortality were higher among those whose diagnosis was in the 1960s rather than the 1970s.

Conclusions: Mortality from type 1 diabetes was higher in Japan compared with Finland. The increased risk of death from type 1 diabetes seems to vary by sex, age at diagnosis, and calendar time period of diagnosis. Further investigation, especially on cause-specific mortality, is warranted in the two countries.

Figure 1

A– C: Stratified cumulative-predicted survival probabilities in the Japanese and Finnish cohorts. The predicted survival probabilities were calculated with the Cox hazard models and stratified by sex, age at diagnosis, and calendar time period of diagnosis in each cohort. A: Stratified cumulative-predicted survival probabilities by sex. B: Stratified cumulative-predicted survival probabilities by age at diagnosis (prepubertal age versus pubertal age: ≥ 11 years old for women and ≥12 years old for men). C: Stratified cumulative-predicted survival probabilities by calendar time period of diagnosis. The survival curves for those with diabetes diagnosed in 1970–1974 are not shown in C because they had a longer duration of diabetes before starting follow-up than did those with diabetes diagnosed in 1965–1969 and 1975–1979. To draw the predicted survival curves, the proportions of sex, age at diagnosis (prepubertal age versus pubertal age), and calendar time period of diagnosis (1965–1969, 1970–1974, and 1975–1979), as well as the mean years before entering the follow-up in the combined sample of the two cohorts, were applied as covariates of the models.