The discordant MZ and DZ twin pairs were then utilized in conditional logistic regression analyses to examine the link between EAU and self-harm while accounting for shared predispositions. Importantly, to test whether associations were significantly different in discordant DZ and MZ pairs (the latter being critical for demonstration of causal/individual-specific environmental effects), an interaction term between zygosity (MZ or not) and EAU status was entered into the adjusted models. If the interaction was significant, then the unadjusted models were re-run in MZ and DZ twins separately to examine differences in ORs.
