Spatiotemporal Trends and Age-Period-Cohort Modeling of the Incidence of Type 1 Diabetes Among Children Aged <15 Years in Norway 1973–1982 and 1989–2003

Abstract

OBJECTIVE—We have investigated age-period-cohort effects and spatial and temporal trends for the incidence of type 1 diabetes among 0- to 14-year-old children in Norway. RESEARCH DESIGN AND METHODS—We included children with the diagnosis of type 1 diabetes in Norway during 1973–1982 and 1989–2003. We studied age, calendar period, and birth cohort effects using Poisson regression, including Holford's method of parameterization, to model the dependencies between age, period, and cohort effects. To study spatiotemporal clustering of cases, we used spatial scan statistics. RESULTS—The overall incidence rate for the study population &lt;15 years of age was 22.7 cases per 100,000 (95% CI 22.1–23.4), showing an average annual increase of 1.2% (95% CI 0.7–1.5%) during the study period. One specific area with 30% increased incidence rates was identified in the southern part of Norway during 1976–1980 (P = 0.001). Also, children born during 1964–1966 in a specific region in the southern part of Norway as well as children born during 1987–1989 in a region in northern Norway showed 2.0 and 2.6 times, respectively, higher incidence rates compared with the rest of the country (both P = 0.001). CONCLUSIONS—The incidence of type 1 diabetes among children increased during the study period. Birth cohort effects were identified using the spatiotemporal scan statistic but not using age, period, and birth cohort modeling. Such effects, within the relatively homogenous Norwegian population, suggest the influence of nongenetic etiological factors.