The Swedish Childhood Diabetes Study--a multivariate analysis of risk determinants for diabetes in different age groups.

In a nationwide incident case-referent study stepwise univariate analysis has revealed several risk determinants for childhood diabetes mellitus. In a multivariate analysis we have determined the set of risk determinants that would independently predict childhood Type 1 (insulin-dependent) diabetes. Possible interactions between the risk determinants and differences in risk profiles with different ages at onset were also examined. Reported familial insulin-treated and non-insulin-treated diabetes were significant risk factors in all age groups, as was also a low frequency of milk intake. The frequency of infections and a high intake of foods rich in nitrosamine tended to interact (OR 11.8, p = 0.053) indicating a synergistic effect. A Cox regression analysis revealed that stressful life events during the last year was the only variable that tended to affect the age at onset (p = 0.055). This indicated that psychological stress may rather precipitate than induce Type 1 diabetes. A short breast-feeding duration (OR = 3.81), and an increased body height (OR = 3.82) contributed significantly to the predictive model in only the youngest age group (0-4 years). An increased frequency of infections in the year preceding onset (OR = 2.15) and no vaccination against measles (OR = 3.33) contributed significantly to the model only in the age group 5-9 years. Various nutrients had different impacts on the risk of developing Type 1 diabetes in different age groups. It is concluded that in the genetically susceptible child, risk factors which are associated with eating habits, frequency of infections, vaccination status, growth pattern and severe psychological stress affect the risk of developing diabetes independently of each other.(ABSTRACT TRUNCATED AT 250 WORDS)