Analysis of 20 years of prospective registration of childhood onset diabetes - time trends and birth cohort effects

The Swedish Childhood Diabetes Registry has been recording all cases of childhood onset diabetes nationwide, with a high level of ascertainment, since 1 July 1977. The present report describes and analyses the 8358 childhood onset cases occurring between 1 January 1978 and 31 December 1997. The mean annual incidence was 26.4/100000 children per year (1978. 21.1 and 1997: 31.9). There was a significant log-linear increase over time, with a mean annual increase of 1.7%. The steepest mean increase was seen among the young onset cases (2.5%) and the steepest yearly increase (6.3%) was seen in this age group during the last 10-y period. A shift towards a younger age at onset was clearly indicated, as the age at onset was less during the last compared with the first 10-y period of observation. The increase over time was similar between the sexes and during winter and summer. When analysing the six full birth cohorts covered, we found no clear-cut shift in the trend. Birth cohorts (1978-82) up to 5 y of onset showed a time variability but no clear trend over time. Ecological analyses associating cumulative incidence by birth cohort to breastfeeding frequency showed no significant association. A statistically significant log-linear association was found to the official estimate of gross domestic product adjusted for similar price levels (p = 0.002). Conclusion: The incidence of childhood onset diabetes is rapidly increasing in Sweden with a shift towards the younger age groups but with no trend in birth cohorts. Precipitating rather than initiating environmental risk factors are suggested, and the correlation to gross domestic product may suggest risk factors associated with wealth-such as a high growth rate, a known risk factor for childhood diabetes.