As above, there was high heterogeneity of PAE (Q=164, df=3, p<0.0001; I2=98%) and COA studies (Q=202, df=9, p<0.0001; I2=96%), with lower between-study variability for PAE (ζ2 =0.16). However, contrary to ADHD and CD, effect sizes for ODD yielded higher between-study variability for COA (ζ2 =0.29) studies. The LOR from the random effects model across PAE studies was 0.38 (red diamond, Figure 2E) and the 95% CI included 0 [−0.05, 0.80], (OR=1.46; 95% CI=[0.95, 2.23]), indicating approximately equal odds. The LOR from COA studies was 0.93 and the CI did not include 0, (95% CI=[0.51, 1.36], Figure 2F), resulting in an odds ratio of 2.53 (95% CI=[1.63, 3.94]). Due to the limited number of studies, only a comparison of PAE vs: COA, father AD and unknown parent AD could be tested. As seen in Figure 3, the mean LOR for PAE studies was smaller than the mean LOR for COA studies, but did not reach significance (p ≥ 0.08). All other comparisons were non-significant as well (p ≥ 0.15; Table 3C).
