Relationship Between BMI and Age at Diagnosis of Type 1 Diabetes in a Mediterranean Area in the Period of 1990–2004

We analyzed data on 3,203 subjects (1,836 male, 2–24 years of age at diagnosis) with newly diagnosed type 1 diabetes included in the Catalan Registry database for new cases of type 1 diabetes between 1990 and 2004 (10). Data on BMI corresponded to information obtained during the 1st week after the diagnosis. The patients were classified into five groups according to age at diagnosis (2–4.9, 5–9.9, 10–14.9, 15–19.9, and 20–25 years). BMI was compared with standards from a cross-sectional epidemiological study of a representative sample of Spanish population (n = 3,534, 1998–2000, 2–24 years of age) (11). Sex- and age-adjusted BMI SD scores were calculated. A subgroup of 579 patients diagnosed during the period 1998–2000 was also selected from the total cohort exactly matching the study period of the reference population.

Comparisons between groups of age at diagnosis and reference values were performed using the Wilcoxon test. Simple correlations (Spearman's) were calculated between age at diagnosis and BMI SD score, as well as a linear regression analysis. A P value <0.05 was considered statistically significant.

From 1990 to 2004, 222 subjects (6.9%) were diagnosed between 2 and 4.9 years of age, 638 (19.9%) between 5 and 9.9 years of age, 877 (27.4%) between 10 and 14.9 years of age, 713 (22.3%) between 15 and 19.9 years of age, and 753 patients (23.5%) between 20 and 25 years of age. The proportion of subjects in each group by age at diagnosis during the period 1998–2000 (n = 579) did not differ greatly from the whole cohort: 5.9, 19.2, 28.1, 21.4, and 25.4% in each different age-group, respectively. In terms of BMI at diagnosis and age at diagnosis, there were no differences between the two cohorts.

The patients diagnosed with type 1 diabetes had a lower BMI than the reference population in all groups of age at diagnosis. These results remained unchanged when data were analyzed by sex or in the 1998–2000 cohort.

For the period of 1990–2004, we did not observe any significant change in the BMI SD score of either sex regardless of the age at diagnosis. In the cohort 1990–2004, there was a significant positive correlation between patient BMI SD score and the age at diagnosis of type 1 diabetes (Fig. 1.). Actually, each year increase in age at diagnosis increased BMI SD score by 2%. In regard to the relationship between these two variables in each age-group, a negative nonsignificant correlation was only observed in the youngest group (r = −0.101). In the remaining groups, the correlation between patient BMI SD score and age at diagnosis was also positive (r = 0.251, r = 0.228, and r = 0.141, respectively, P < 0.001; and r = 0.048, P = 0.219) in each different age-group at diagnosis. The relationship between BMI SD score and the age at diagnosis did not differ when the analysis was performed by sex.

When the subgroup of subjects diagnosed between 1998 and 2000 were considered as a whole, a positive correlation was again found between BMI SD score and age at diagnosis (r = 0.147, P < 0.001). A negative nonsignificant correlation was only observed in the youngest group (r = −0.174). In this cohort, when weight loss before diagnosis (reported for each individual subject) was considered in the analysis, we did not observe any change in the results.

Our data, collected from a National registry of newly diagnosed type 1 diabetes from a Mediterranean area over a period of 15 years (1990–2004) indicates that BMI at diagnosis of the disease has not substantially increased. In addition, we found a positive association between BMI SD score and the age at diagnosis of type 1 diabetes.

Compared with other geographical areas, Spain has an intermediate prevalence of overweight and obesity in children and young adults (11). However, over the past decades an increased trend toward higher rates has been observed (12,13). We did not observe an increase in BMI or BMI SD score at diagnosis in any of the age-groups evaluated. In the present study, a higher BMI SD score at diagnosis was not associated with a younger age at onset of type 1 diabetes. To the contrary, in the whole cohort we found a positive correlation between higher BMI SD score and older age at diagnosis. However, it should be pointed out that in both cohorts (1990–2004 and 1998–2000) in the 2–4.9 years of age group there was a nonsignificant trend toward an association between a higher BMI SD score and a younger age at diagnosis. It is well recognized that in type 1 diabetes the younger the age at diagnosis the lower the β-cell function. In the study by Dabelea et al. (8), 70% of the participants had fasting C-peptide <0.5 ng/ml with BMI and a younger age at diagnosis being associated only in these subjects. It is possible that the additional metabolic demands associated with higher BMI cannot be compensated only in subjects with a highly diminished β-cell function. Unfortunately, data concerning β-cell function was not available in our study.

Excluding recent data by Knerr et al., previous studies carried out in the U.K. showing the relationship between younger age at diagnosis and higher BMI involved a small number of subjects (4,5,8). To our knowledge, this is the first study in a large cohort of a Mediterranean National Registry for new cases of type 1 diabetes in which the accelerator hypothesis is tested. Our results did not fit with the hypothesis and point to the heterogeneity of the disease in different populations (14,15). Accordingly, in a recent study including children who were either from the U.K. or originally from the south of Asia, Porter et al. did not confirm the accelerator hypothesis (9). It is plausible that this hypothesis does or does not become manifest depending on the genetic background and environmental factors, including the prevalence of overweight and obesity.

Our study has several limitations. First, it is cross-sectional in nature, it has a long recruitment period, and influence by secular trends in BMI cannot be ruled out. Second, we lack of biological evidence to support diabetes classification in terms of pancreatic autoantibodies availability. However, recently, as a part of the European multicenter study Insulin Dependent Diabetes in Young Adults in Europe (IDA), we investigated the consistency in clinical diagnosis of type 1 diabetes. The percentage agreement between original classification and reclassification was 92%. Finally, we are aware that in order to truly test the accelerator hypothesis, longitudinal studies in risk groups (i.e., first-degree relatives of type 1 diabetic subjects) well before clinical onset would be needed.

In summary, in our large cohort of Mediterranean subjects with recently diagnosed type 1 diabetes, increasing BMI is not uniformly associated with younger age at diagnosis. Although our data did not agree with the accelerator hypothesis, the postulate is worthy of interest, mainly in childhood-onset type 1 diabetes in Western societies in which obesity has risen to epidemic proportions in this age-group.

We are indebted to all of those involved in reporting the data of newly diagnosed type 1 diabetic subjects to the National Catalan Registry. The National Catalan Registry for type 1 diabetes participates in the EURODIAB study group.