83 The enhancing effects of testosterone in exposure treatment for social anxiety disorder In the post-hoc model in which the baseline testosterone data were included, the effects of Time were confirmed: Estimate(linear) = -82.23(16.96), F(1,48) = 23.45, p <.001, Estimate(quadratic) = −87.07(13.69), F(1,48) = 40.39, p <.001, Estimate(cubic) = -23.42(11.54), F(1,47) = 4.09, p = .049. No Time x Group interactions were found: Estimate(linear) = -1.00(16.96), F(1,48) = .00, p = .995, Estimate(quadratic) = 8.01(13.69), F(1,47) = .341, p = .619, Estimate(cubic) = -1.08(11.45), F(1,47) = .01, p = .926. We did find a significant Time(quadratic) x Group x Baseline-T effect: Estimate = 2.26(.94), F(1,48) = 5.72, p = .021. The fear patterns recorded per group depended on baseline testosterone levels. For the P group, fear scores showed the same quadratic pattern regardless of basal testosterone: Estimate = .82(1.14), F(1,24) = .52, p = .476, while in the T group fear levels showed a higher peak which then reduced more sharply for the participants with high endogenous testosterone levels, while for the participants with low endogenous testosterone the peak fear levels flattened and reduced less: Estimate = -3.73(1.48), F(1,24) = 6.32, p = .019 (Figure 4.2, main manuscript, panel A). Details analyses of transfer to unenhanced exposure (session 2) The model showed that fear levels reduced over time (linear, cubic, and quadratic), confirming that the unenhanced exposure resulted in the expected within-session reduction: Estimate (linear) = -62.94(15.42), F(1,50) = 16.66, p <.001, Estimate (quadratic) = −48.32(11.03), F(1,50) = 19.18, p < .001, Estimate (cubic) = -36.76(8.90), F(1,50) = 17.01, p <.001. Of main interest was the Group x Time effect; a significant interaction was found between Time (quadratic) and Group: Estimate = 23.68(11.03), F(1,50) = 4.61, p = .037, indicating that the fear patterns differed per group. Compared to the P group, the participants in the T group reported higher fear levels throughout the session, with a steeper decline towards the end of the session (see Figure chapter 4 – appendix 3, panel B). Details HR analyses session 1 and session 2 Session 1 Baseline HR was a significant predictor: Estimate = .65(.11), F(1,49) = 32.72, p <.001, i.e., the higher the baseline HR, the higher the HR during exposure. The linear and quadratic time terms were both significant predictors and each added to the model fit: Estimate (linear) = -26.51(6.44), F(1,49) = 16.89, p < .001; Estimate (quadratic) = 7.50(3.09), F(1,49) = 5.84, p = .019. Mean HR during exposure reduced over time, confirming that exposure resulted in the expected within-session reduction. We did not find a significant interaction between Time and Group: Estimate(linear) = 5.43(6.44), F(1,49) = .71, p = .404, 4
RkJQdWJsaXNoZXIy MTk4NDMw