99 Social Avoidance and Testosterone Enhanced Exposure Efficacy in Women with Social Anxiety Disorder Predictive effects of automatic avoidance tendencies on exposure success Only effects relevant for the current research questions are reported. For a full overview of the results of the exposure and testosterone administration, we refer the reader to the report of the parent trial (Hutschemaekers et al., 2021). Fear levels Session 1 (Enhanced session) The three-way interaction of AAT effect score x Time (linear) x Group approached statistical significance, Estimate(linear) = -.44(.23), 95% CI [-.89, -.02]; F(1, 213) = 3.85, p = .051 (see step 1, table 5.2, also for the quadratic time term). To further test our hypothesis that those who show greater social avoidance tendencies at baseline and receive testosterone would profit better compared to those participants that receive placebo, we post-hoc decomposed this result. This follow-up analyses revealed a significant twoway interaction of AAT effect score X Time (linear) for the testosterone group: Estimate = .66(.33), 95% CI [-.01, 1.27]; F(1,106) = 4.04, p = .047, but not for the placebo group, Estimate = -.23(.31), 95% CI [-.82, .39]; F(1, 107) = .538, p = .465. Simple slope analyses further showed that, among those assigned to the testosterone condition, participants with lower AAT effect scores (- 1 SD, relative avoidance) reported greater reductions of fear (Estimate = -126.72(27.58), 95% CI [-187.13, -71.55]; t(106) = -4.60, p < .001) relative to those with higher AAT effect scores (mean + 1SD, relative approach; Estimate = -53.46(23.84), 95% CI [-103.61, -7.16], t(106) = -2.24, p = .03). These differential effects were not observed for the placebo group (see Figure 5.1). The inclusion of baseline testosterone as a control variable did not change the results (step 2, see Table 5.2). Session 2 (Non-enhanced) There was no significant three-way interaction effect between AAT effect score, time and group (see Table 5.2)2 or two-way interaction between AAT effect score and time. Inclusion of baseline testosterone in the model showed a significant interaction of AAT effect score with Group: Estimate = .09(.04), 95% CI [.001, .17]; F(1,45) = 4.44, p =.041. For the testosterone group, avoidance scores were not associated with fear levels while for the placebo group, stronger avoidance scores were associated with lower overall fear levels. All other main effects and interaction with AAT effect scores remained non-significant (see step 2, Table 5.2). 2 The residuals of the model did show some deviations from normality. However, log or square root transformations did not improve the distribution of the data. 5
RkJQdWJsaXNoZXIy MTk4NDMw