The EURODIAB Group has recently published that the incidence rate of childhood type 1 diabetes rises across Europe by an average of approximately 3–4%.1 On the other hand, several publications2 agree that the age of the children at diagnosis is decreasing, so whether there is an actual increase in the incidence of type 1 diabetes or if there is only a displacement to earlier ages is still under discussion.3 To the best of our knowledge, there are some papers showing the incidence of type 1 diabetes in adults, over 30 years,4 but no studies have been carried out regarding the variation of age at onset in these patients.

In this study, we analysed if the age at onset has changed during the last two decades in patients with type 1 diabetes in Navarre (Spain). The subjects of the study are the 716 patients (413 men and 303 women, 280 <15 years and 436 ≥15 years) with type 1 diabetes diagnosed between 1990 and 2008, that have been treated in the Navarra Hospital Complex. Patients with LADA (positive antibodies but insulin dependency ≥6 months after diagnosis) were excluded. There have been no changes in the admission policy of the hospital or in the uptake area during the study period. To assess the evolution of overweight at onset, we calculated the yearly proportion of patients over 85th percentiles from Carrascosa's normal body mass index (BMI) distributions for Spanish population.5

To assess the relationship between age at onset and calendar year, we used General Additive Models (GAM). As GAM models showed a linear trend (1 degree of freedom), we additionally adjusted linear regression models with age at onset as dependent variable, and calendar year and BMI as explanatory variables. In the case of children, as normal BMI is different for sexes and changes with age, BMI values were translated into their corresponding percentiles form Carrascosa's distributions.

This study was approved by the regional Ethical Review Board of Navarra.

GAM models show a linear relationship between age at onset and calendar year, both in children and in adults. In children we did not find a statistically significant trend, though the findings may suggest certain decreasing tendency (p=0.114). In contrast, in patients ≥15 years, the mean age at onset increases as time goes by (Fig. 1). The average increase was of 0.48 years per calendar year (95%CI: 0.28–0.67), and it is observed all through the study period. The mean (SD) ages at onset were 26.8 (10.3) years in 1990–1994, 27.7 (9.9) years in 1995–1999, 30.6 (10.4) years in 2000–2004, and 33.3 (12.4) years in 2005–2008. This increase is reflected in an increased relative proportion of adults whose age at onset fell into the age group of 30–45 years, and in a reduction in that of the group of 15–29 years (p<0.001). The proportion of adults with an age at onset above 45 years remains of around 10% in the study period. No patients had the onset above 60 years.

Fig. 1.

Age at diabetes onset by calendar year: means and 95%CI (error bars), and linear regression models (lines).

(0.16MB).

The proportion of patients with overweight at onset is 3% in those under 15 and 15% in those ≥15 years. In patients with onset under 15 years of age, we did not observe any association between BMI and the age at onset, nor a trend along the study period in the proportion of patients with overweight. The relationship of calendar year with age at onset remains non-significant after adjusting for BMI. In patients ≥15 years, we found a positive association of the age at onset with BMI (p=0.001), as well as a positive trend in the proportion of patients above 85th BMI percentile (p=0.002). However, including the BMI in the linear regression model does not meaningfully affect the estimate of the increase in the age at onset per calendar year (0.44 vs 0.48 years).

We did not find significant differences between males and females in the average age at onset or in its evolution along the study period.

Although high-risk HLA genotypes is a predictor for diabetes progression, we have not made statistical adjustments for other covariables as family history of diabetes, because previous studies dealing with this item, have found that BMI, body size and insulin resistance are the most important factors influencing the age at onset.2

The delay in the age at onset seen in patients ≥15 years of age could be reflecting an actual epidemiological change in type 1 diabetes onset, but also may be the result of higher accuracy in the diagnosis of type 1 diabetes in adults. Further research is needed in order to clarify this issue.