Mendelian Randomization Analysis
In the IVW analysis, children with asthma were at higher risk for T1DM (OR=1.308, 95% CI 1.030-1.661, P =0.028) (Table 1). Higher risk for T1DM in children with asthma was also apparent in MR-Egger analysis (OR=2.082, 95%CI 1.115-3.888, P=0.029 ). Analysis using the weighted median method failed to establish an association, but the direction of the general finding was consistent (OR=1.033, 95%CI 0.849-1.256, P=0.745) . Figure S3 illustrates method comparison plot(a) and forest plot(b). MR-Egger regression failed to identify evidence to support the existence of unbalanced pleiotropy in the genetic instruments (intercept β = -0.071, P = 0.126). The MR-PRESSO outlier test detected 2 specific horizontal pleiotropic outlier variants (rs4795399 and rs28407950). However, these 2 SNPs were not genome-wide significant associated with T1DM. The association of childhood asthma with T1DM was not statistically significant in MR-PRESSO outlier-corrected IVW analysis (OR = 1.175, 95%CI 0.960-1.438, P = 0.118).
IVW analysis suggested a protective effect of T1DM on the risk of childhood asthma (OR = 0.937, 95%CI 0.881-0.996, P  = 0.037). Subsequent analysis using the MR-Egger (OR=0.918, 95%CI 0.850-0.991, P=0.041) and weighted median methods (OR=0.928, 95%CI 0.872-0.987, P=0.017) supported this finding. Figure S4 illustrates method comparison plot(a) and forest plot(b). No evidence of pleiotropy or outlier variants was found by either MR-Egger regression (intercept β = 0.016, p = 0.383) or MR-PRESSO outlier test.