3.1 Causal effect of household income status on knee osteoarthritis
Table 1 shows the results of Mendelian randomization analysis, in which we identified 48 sensitive SNPs to examine the causal effects between household income status and knee osteoarthritis. The results of Mendelian randomization showed that IVW, MR-Egger, WM, Simple Mode, and Weighted Mode had the same direction. The Odds ratio (95% CI) of IVW was 0.71(0.60~0.85), and the P value was 0.0001, suggesting a causal association between household income status and knee osteoarthritis. Simple mode analysis all supported the evidence of a causal relationship. OR values for MR-Egger, Weighted median, and Weighted mode all suggested that an increase in household income status was associated with a reduced risk of knee osteoarthritis, but their p values suggested no statistical significance. To further validate these findings, we examined the effect size and F-statistics of the 48 SNPs, as listed in Table 2. Notably, each F-statistics of the SNPs was greater than 10, indicating the absence of weak instrumental variable bias.
3.2 Results of sensitivity analyses in the MR study
To test the robustness of the results, we performed a sensitivity analysis. Cochran's Q test did not reveal heterogeneity (p= 0.106). The funnel plot shows that when a single SNP is used as the instrumental variable, the causal effects of household income status and knee osteoarthritis are symmetrically distributed (Figure 2). This indicates that the results obtained by using 48 SNPs as the instrumental variables are less likely to be affected by potential bias, and the results are stable and reliable. The MR-PRESSO global test indicated no horizontal pleiotropy(p=0.119). The pleiotropy test showed that egger intercept was not statistically significant (egger intercept =0.006, p= 0.404) ( Figure 3). The "leave-one-out" method showed that no single SNP significantly affected the overall results ( Figure 4). Finally, the forest plot calculated by the IVW method reflects that increased household income status significantly reduces the risk of knee osteoarthritis as the interval estimates are all on the right side of the 0 point during the IVW analysis (Figure 5).
3.3 Results of reverse MR study and sensitivity analyses
The number of SNPs identified as IVs in knee osteoarthritis is 8 in the reverse MR study. The number of SNPs that were ultimately identified as IVs in knee osteoarthritis is 8 in the reverse MR study. We calculated the F-statistic for these 8 SNPs, all>10 (minimum 31.517, maximum 96.867, Table 3). According to IVW model results, there is no causal relationship between knee osteoarthritis and household income status (OR: 0.965,95%CI = 0.927- 1.005, P = 0.09). Weighted median (OR:0.962,95% CI =0.921-1.005, P =0.08), Simple mode(OR: 1.014,95% CI =0.923-1.115, P =0.778), Weighted mode(OR: 0.922, 95% CI =0.849-1.002, P =0.096) and MR Egger (OR:0.791,95% CI =0.645-0.97, P =0.065) all supported the conclusions of the IVW model (Table 4). Cochran 's Q test did not reveal heterogeneity (p= 0.053). The MR-PRESSO global test indicated no horizontal pleiotropy(p=0.055). This shows that our reverse MR study results are reliable.