Aims: The aim of the study was to identify environmental risk factors for insulin-dependent diabetes mellitus (Type 1 DM) in childhood.
Methods: A matched case-control study of Type 1 DM conducted in Lancashire and Cumbria, UK, using a structured interview. Cases (n=196, participation rate 83%) were children under 16 years of age diagnosed prior to October 1998 and attending diabetic clinics. Controls (n=381) were healthy children from the community matched by gender and by age (within a few days of birth). The data were analysed by logistic regression using the technique of Breslow and Day for matched case control studies.
Results: The multivariate regression model showed that the following factors were significantly associated with the risk of developing Type 1 DM (odds ratio, 95% confidence intervals): sharing a room with a sibling (0.458, 0.290-0.721), social contact with other children when aged 6-11 months (0.439, 0.256-0.752), consumption of sugary food (0.080, 0.024-0.261), parental insulin dependent diabetes mellitus (10.651, 3.086-36.761), maternal thyroid disease (4.861, 1.681-14.058), consuming more than one pint of milk per day prior to school entry (0.498, 0.310-0.802), maternal smoking during pregnancy (0.373, 0.218-0.636), a father with no academic qualifications (0.504, 0.278-0.913), maternal age at time of birth (0.900, 0.854-0.948), maternal infections in pregnancy (2.453, 1.011-5.948), other maternal illnesses or conditions in pregnancy (2.007, 1.139-3.535), belonging to an Asian family (0.104, 0.028-0.394), and regular contact with pets and other animals (0.552, 0.309-0.987).
Conclusion: Many of the results are consistent with the hygiene hypothesis which links improved living standards with decreased exposure to microorganisms and increased risk of immune mediated disease in childhood. These findings challenge the idea that improved hygiene acts exclusively through a Th2 mechanism leading to atopic disease as Type 1 DM is mediated by a Th1 reaction. The association with maternal smoking could be due to recall bias but a causal link cannot be excluded with confidence.