To the Editor: Several different viruses are associated with Type 1 diabetes, but no studies have observed a gender difference in the impact of infections on diabetes. Childhood Type 1 diabetes is increasing worldwide; this has prompted several study groups to search for risk factors explaining this worldwide increase. All studies investigating risk factors for Type 1 diabetes have been matched for gender, because there is a known difference in incidence between boys and girls [1].

Several viral infections (e.g. rubella, enterovirus) in early life are associated with the risk of Type 1 diabetes [2]. Age is the most important determinant of enterovirus infections but data also supports a male predominance particularly for the more severe disease [3]. Gender differences are found in immunological response to some vaccinations given in early childhood [4].

These observations prompted us to look at gender differences in the effect of neonatal infections in a population based case-control study based on the Danish National Diabetes Register.

In 1996 a Danish national prospective register of childhood diabetes was opened. The ascertainment in the register during the first 4 years was more than 99%. All children below the age of 15 years with Type 1 diabetes diagnosed from 1996 to 1999 were invited to participate in a case-control study. Two control subjects per case were randomly selected from the National Population Register matched by date of birth and gender. Information on hospitalisation and perinatal diagnosis such as preterm delivery (<37 weeks), small or large for date (>2 standard deviation from the expected mean), and neonatal jaundice and infections were obtained from the hospital discharge register. Data on maternal and paternal age and ethnicity were obtained from the Danish population register. Additionally all cases and controls received a questionnaire including questions on smoking during pregnancy, early life nutrition, growth, siblings and family history of diabetes. Immigrants and offspring of immigrants were excluded from the study. For the study 602 diabetic children were eligible and 1459 control subjects were matched by gender and date of birth. Hospital information was available for 598 cases and 1445 control subjects. Questionnaire data were available for 490 (81%) cases and 696 (48%) of control subjects. We estimated the effect of neonatal infections among boys and girls with multiple logistic regression using all cases and control with hospital information. The model included age, gender and interaction between gender and infections. Of the children 49 had an infectious diagnosis in the neonatal period and 28 of those had either meningitis or sepsis, the remaining diagnoses were upper respiratory tract infections. There was no information on treatment with antibiotics.

Interestingly we found an increased risk of diabetes in boys with any neonatal infections [OR=2.42 (CI 1.14−5.15) p=0.02] and a decreased risk in girls with neonatal infections: [OR=0.40 (CI 0.12–1.38) p=0.15]. The estimated odds ratio for boys was six times the estimated odds ratio for the girls [CI 1.41–25.5) p=0.015] which corresponds to the significant interaction between gender and neonatal infections). There was no association between diabetes and neonatal infections [OR=1.23 (CI 0.67–2.28) p=0.51] when the interaction term was omitted.

The difference between boys and girls remained after adjustment for confounders such as introduction of cow's milk (<2 month), maternal age at delivery, family history of diabetes, smoking during pregnancy and birth order with an odds ratio in boys of 4.36 (CI 1.29–14.8), and in girls of 0.49 (CI 0.10–2.36).

The number of hospitalised infections in first year of life were also analysed for interaction with gender. This showed no difference between boys and girls (p=0.94).

To our knowledge this is the first study indicating a gender difference in the impact of neonatal infections on the risk of Type 1 diabetes in children. The results are not likely to be caused by recall bias since all information is based on register data collected prior to diabetes onset and less than 1% of the patients have missing information. The analysis is based only on hospital records and since boys are more often hospitalised in the first year of life there could be a selection bias towards higher registration rate of infections in boys. However, if we assume that this higher registration rate is the same in diabetic and non-diabetic cases, this should not influence the results.

The results might be explained by a gender difference in susceptibility to T cell-mediated autoimmune diabetes and gender difference in response to infections [5]. The observation of an interaction between gender and risk factors affecting the development of the immune regulatory system is of crucial relevance. Accordingly the effects of those risk factors should be taken into account in analyses where boys and girls are analysed together because of opposite effects in boys and girls. This result suggests that all studies on risk factors of autoimmune diabetes which possibly occur through an effect on the immune regulatory system have to be analysed separately for each gender or with interaction. This has implications for the evaluation of risk factors such as infections, vitamin D supplement and breastfeeding, especially when considering the impact of these risk factors in early childhood.

J. Svensson, B. Carstensen, H. B. Mortensen, K. Borch-Johnsen and the Danish Study Group of Diabetes in Childhood (DSBD)