Subgroup analyses and sensitivity analyses
The point estimates for ADHD with true SSRI/SNRI-exposure were slightly larger (about 10%) than the main results; for ADHD symptoms the results were almost identical (Table 3).
Further adjustment for pre-existing clinical depression/anxiety attenuated the observed associations in the age band 7-9 years (Table S7). Maternal pre-existing clinical depression/anxiety was independently associated with child ADHD in these analyses (weighted, adjusted HR: 1.34, 95% CI: (1.05-1.72)). Results of other sensitivity analyses are given in the Supplement, Tables S8-10 and Figures S6-15, and these did not deviate from the main analysis.