Background: Childhood diabetes has been shown to have increased over time, to have a north-south gradient within Sweden and to exhibit seasonal variation. In the present study we have analysed the variation in incidence of childhood diabetes with time, and climatological factors such as mean temperature and sunshine hours.
Methods: The Swedish childhood diabetes register prospectively records all incident cases of diabetes with onset at 0-14 years since 1 July 1977. By 1 January 1993, 5831 cases had been registered. The data set was analysed using Poisson regression modelling and Cox' proportional hazards model.
Results: The mean annual incidence rate was 25/100,000. Over the 15 years there was an overall increase with time. When analysing the first 7 years and the last 8 years separately the basal linear increase was a mean 6% per year during the first period but no significant increase was noted during the second period. The increase during the first period was mainly due to an increase in incidence before the age of 10. Significant inverse and independent relationships were found between incidence rate and mean sunshine hours and mean temperature, respectively, when measured monthly in each county. We did not find any association between the climatological factors and age at onset.
Conclusions: The increasing trend in childhood diabetes in Sweden during the past 15 years was mainly due to an increase in incidence in children younger than 10 years of age with onset during the first half of the observation period. During the last 8-year period incidence levelled off. A low mean temperature and a low number of sunshine hours are inversely and independently correlated to the incidence of childhood diabetes which may partly explain the north-south gradient and seasonal variation in the incidence of the disease.