59 Endogenous testosterone levels are predictive of symptom reduction with exposure therapy in social anxiety disorder Analysis baseline testosterone and social anxiety symptoms with winsorized data The mixed model analysis for LSAS as dependent variable, with baseline testosterone (winsorized) and Assessment (pre/post) as predictors, showed a main effect of Assessment, again confirming efficacy of exposure, but now with respect to the social anxiety symptom levels, Estimate = -9.14(1.63), F(1,67) = 31.29, p <.001. In specific, LSAS scores reduced from pre to post-assessment (Mpre = 74.80, SD = 22.73; Mpost = 66.01, SD = 24.58). A main effect of Gender, Estimate = -9.40(2.75), F(1,68) = 11.68, p < 0.001 on the LSAS, also indicated again higher scores for females (M = 75.63, SD = 23.91) compared to males (M = 57.53, SD = 18.85). Most critically, an interaction of baseline testosterone and assessment Estimate = 3.87(1.86), F(1,67) = 4.33, p = .04, showed that stronger reductions in symptom levels were related to low baseline testosterone levels (see figure chapter 3 - Appendix 3, panel A for a scatter plot). Analysis reactive testosterone and social anxiety symptoms with winsorized data We repeated the same mixed models analysis but now with reactive testosterone level (winsorized) and Assessment (pre/post) as predictors. In line with our hypothesis, reactive testosterone levels were positively related to the pre-to post assessment decrease in social anxiety symptoms: apart from the main effects of Assessment (pre, post), Estimate = -8.63(1.60), F(1,66) = 29.26, p <.001 and Gender, Estimate = -9.73(2.76), F(1,68) = 12.41, p <.001, testosterone reactivity significantly modulated the effect of Assessment as indicated by a significant Time x Assessment interaction, Estimate = -1.63(0.64), F(1,66) = 6.42, p = .014. In specific, high reactive testosterone levels were associated with relatively stronger reductions in symptom levels pre to post assessment (see figure chapter 3 Appendix 3, panel B for a scatter plot). Analyses separated per gender Although we found no significant moderation by gender of the testosterone by assessment (pre- to post-intervention social anxiety symptom severity) interaction, we explored post-hoc whether similar patterns of results to the total sample held for females alone, with a disclaimer that we were underpowered to detect small effects. Fear levels females The mixed model analysis for the peak SUDs at the speech exposure session as dependent variable, with baseline testosterone and Time (speech 1, speech 2, speech 3, in the speech exposure session) as predictors, showed a main effect of Time, confirming that exposure resulted in the expected within-session reductions in fear levels, Estimate = -7.75(1.19), F(1,50) = 42.65, p <.001. More specifically, SUDs diminished over the 3
RkJQdWJsaXNoZXIy MTk4NDMw