Aims/hypothesis: This study aimed to determine if the incidence of Type I (insulin-dependent) diabetes mellitus increased over time and if the time trend is attributable to linear trend, calendar period or birth cohort effects.
Methods: This study was based on a cohort of subjects aged 0-29 years from 1984 to 1996, who resided in the province of Turin, Italy. Poisson regression models were used to estimate the effect of sex, age, calendar time and cohorts on incidence rates.
Results: The mean incidence rate in the age group of 0 to 29 years was 7.78 of 100,000 person-years (95% confidence interval: 7.26-8.32), with a lower risk in women than in men [rate ratio (RR): 0.76 (0.67-0.88)]. We found a trend of incidence increasing over time (annual increase of risk 2.25 % a year, 95 % CI 0.44-4.10). In Poisson regression analysis we found that the best model was the one with sex, age and a linear term attributable to either calendar period or cohort effects. The linear term corresponds to a RR of 1.12 (1.02-1.22,p = 0.015) for each 5-year age span.
Conclusion/interpretation: We found that the time trend in this Mediterranean area was similar in magnitude to that reported in northern European countries. The increase was linear in pattern, each birth cohort and each calendar period showing a higher risk than the preceeding one with some evidence of two steep increases in the incidence of Type I diabetes centered on the years 1964 and 1984. We found that the incidence of diabetes increased in the 0 to 14 age group and also in the older age group of 15 to 29 years and that the age-period models were not statistically significantly better than the age-cohort models.