Table 3.
Results of the Mendelian randomization sensitivity analyses using summary-level data from genetic liability to ADHD (9 SNPs) to a standard deviation change in attained educational level, household income, Townsend deprivation index, and intelligence
| Outcome | Method | OR (95% CI) | p-value |
|---|---|---|---|
| Education | Weighted median | −0.07 (−0.1, −0.03) | <0.001 |
| MR-Egger | 0.03 (−0.15, 0.21) | 0.760 | |
| MR-PRESSO (4 outliers) | −0.06 (−0.10, −0.02) | 0.046 | |
| Contamination mixture | −0.12 (−0.19, 0.00) | <0.001 | |
| MR-Mode | −0.08 (−0.15, 0.00) | 0.078 | |
| MR-Mix | −0.15 (−0.21, −0.09) | <0.001 | |
| GSMR (<10 SNPs) | NA | ||
| Household income | IVW-random effects | −0.09 (−0.11, −0.06) | <0.001 |
| Weighted median | −0.09 (−0.12, −0.05) | <0.001 | |
| MR-Egger | −0.03 (−0.14, 0.07) | 0.564 | |
| MR-PRESSO (0 outlier) | NA | NA | |
| Contamination mixture | −0.09 (−0.11, −0.07) | <0.001 | |
| MR-Mode | −0.08 (−0.14, −0.02) | 0.026 | |
| MR-Mix | −0.04 (−0.06, −0.02) | 0.001 | |
| GSMR (<10 SNPs) | NA | ||
| Townsend deprivation index | Weighted median | 0.06 (0.04, 0.09) | <0.001 |
| MR-Egger | 0.01 (−0.07, 0.09) | 0.762 | |
| MR-PRESSO (0 outliers) | NA | NA | |
| Contamination mixture | 0.08 (0.05, 0.09) | <0.001 | |
| MR-Mode | 0.07 (0.03, 0.10) | 0.003 | |
| MR-Mix | 0.07 (−0.02, 0.16) | 0.132 | |
| GSMR (<10 SNPs) | NA | ||
| Intelligence | Weighted median | −0.05 (−0.10, −0.01) | 0.030 |
| MR-Egger | −0.24 (−0.49, 0.01) | 0.108 | |
| MR-PRESSO (4 outliers) | −0.06 (−0.13, 0.01) | 0.149 | |
| Contamination mixture | −0.11 (−0.15, −0.08) | 0.020 | |
| MR-Mode | −0.10 (−0.21, 0.02) | 0.130 | |
| MR-Mix | −0.14 (−0.22, −0.06) | 0.001 | |
| GSMR (<10 SNPs) | NA |
NA not applicable given that the genetic instrument for the ADHD exposure included only 9 SNPs, GSMR is not a suitable method