When we performed gene-matched case-control logistic re-
gressions the effects were more pronounced. Regarding NOx ex- posure the ORs were above one for all three trimesters (1.36, 1.38 and 1.58, respectively, Table 2) when we compared the highest exposure category with the reference category. However, it was only exposure during the third trimester that reached statistical significance (OR 1.58 95% CI 1.06–2.35). When only the exact HLA- matched case-control sets were included in the analyses the cor- responding OR was 1.47 (95% CI 0.96–2.24). The estimates only changed marginally when we performed the sensitivity analyses (data in Table 3). In multivariate analyses the adjusted ORs for developing T1D when comparing highest with lowest category of exposure were 1.38 (95% CI 0.98–1.93) for NOx (data in Table 4).