Similarly, the link from maternal AD liability (ukb-b-14699) to a lower risk of insomnia (ukb-b-3957) remained highly significant with the Weighted Median method ( p = 7.67 × 10 −7 ), However, in line with our multi-method approach to identify potential bias, a significant MR-Egger intercept ( p = 0.008) suggests that substantial some caution is warranted due to potential horizontal pleiotropy.
This screen required putative causal relationship to concurrently demonstrate: (1) a nominally significant IVW effect ( p < 0.05), (2) a concordant direction of effect from the weighted median estimator to ensure the signal was not driven by a minority of potentially invalid instruments, (3) no evidence of directional pleiotropy as indicated by a non-significant MR-Egger intercept ( p > 0.05), and (4) no significant heterogeneity between instruments as assessed by Cochran’s Q statistic ( p > 0.05).