3 Results
3.1 Bidirectional two-sample MR between the allergic disease and COVID-19
74 and 73 IVs were selected for the two-sample MR analysis of the allergic disease on SARS‐CoV‐2 infection and severe COVID‐19, respectively. In addition, 19 and 31 IVs were selected for MR analysis of SARS-CoV-2 infection and severe COVID-19 on the allergic disease, respectively. (Figure 2) The characteristics of IVs were presented in Supplementary Tables 1 to 4 in the order shown in Figure 2.
The allergic disease was not causally associated with SARS‐CoV‐2 infection (P=0.545; OR=0.99; 95% CI: 0.97-1.02) but have a causal and protective effect on severe COVID-19 (P=0.028; OR=0.91; 95% CI: 0.83-0.99). In the analysis of the effect of allergic disease on severe COVID-19, the four methods were in the same direction. The heterogeneity test indicated a minor problem (P=0.001) but no significant pleiotropy was observed (P=0.691). In contrast, no significant causal effect was found of SARS-CoV-2 infection/severe COVID-19 on the allergic disease. Incidentally, horizontal pleiotropy was not detected (all P >0.05) in any of the above analyses. (Figures 2 and 4)
3.2 Univariate MR between the various allergic diseases and severe COVID-19
The numbers of selected IVs were listed in Figure 3, with specific characteristics (including F values) described in Supplementary Tables 5 to 12 in the order shown in Figure 3. AD showed a clear causal relationship with severe COVID-19 (P=0.019; OR=0.94; 95% CI: 0.88-0.99). No significant pleiotropy or heterogeneity was observed (P=0.343 and 0.617, respectively). (Figures 3 and 4)
Although no causal relationship between asthma and severe COVID-19 was observed in univariate MR analysis (P=0.424; OR=0.97; 95% CI: 0.91-1.04), the analysis of each asthma subtype showed positive results: mixed asthma (P=0.007; OR=0.97; 95% CI: 0.94-0.99) and childhood asthma (age<16 years old) (P=0.017; OR=0.90; 95% CI: 0.83-0.98) were causal protective factors for severe COVID-19. Heterogeneity and horizontal pleiotropy were not observed in the analyses of both mixed asthma (P= 0.559 and 0.718, respectively) and childhood asthma (P= 0.111 and 0.896, respectively). (Figures 3 and 4)
Additionally, the univariate MR analysis of doctor-confirmed hay fever, AR, or eczema exhibited an ambiguous result: a positive protective result (P=0.035; OR=0.935; 95% CI: 0.878-0.995) was obtained using the primary method (IVW), and the directions of MR Egger and weighted median were consistent on the left, but the direction of weighted mode was not. There was small heterogeneity (P=0.008) but no horizontal pleiotropy (P=0.420) in this analysis.
The robustness of these univariate MR results was further proved by the insignificant MR-Egger intercepts (all P>0.05), implying the absence of horizontal pleiotropy. (Figure 3)
3.3 Multivariate MR of three allergic diseases and fourasthma subtypes
The results of the multivariate MR analysis between the three allergic diseases (AR, AD, and asthma) and severe COVID-19 were shown in Figure 5. In the multivariate MR analysis with AR, AD, and asthma adjusted for each other, asthma was surprisingly found to be a suggestive major causal protective factor against severe COVID-19 (P=0.024; OR=0.90; 95% CI: 0.83-0.98).
Figure 6 displayed the outcomes of multivariate MR analysis between four types of asthma (allergic, non-allergic, mixed, and childhood asthma) and severe COVID-19. Only childhood asthma remained suggestively causally associated with severe COVID-19 (P=0.048; OR=0.770; 95% CI: 0.594-0.998).