Farnia et al Dovepress Instruments Greenhouse-Geisser epsilon value(E).For ANOVAs,effect Self-assessment of depressive symptoms by using the BDI sizes are indicated with the partial eta squared (n2),with Patients completed the BDI,64 which a self-report of symp- 0.059≥n≥0.01 indicating small(s),0.139≥n2≥0.06 toms of depression.The questionnaire consists of 21 items indicating medium(M),and n2=0.14 indicating large (L) and covers such areas as depressive mood,loss of appetite, effect sizes.All computations were performed as intention- sleep disorders,and suicidal thoughts.Answers are given to-treat and the last observation carried forward. on 4-point Likert scales with the anchor points 0(for"as The nominal alpha-level was set at 0.05;post hoc analy- always”or“no change')and3(for“not able anymore” ses were performed with Bonferroni-Holm corrections of or "dramatic change")and with higher scores reflecting P-values for multiple testing.Statistical analyses were per- greater severity of depressive symptoms(Cronbach's formed with SPSS*20.0(IBM Corporation,Armonk,NY, alpha =0.89). USA)for Apple MacIntosh". Self-assessment of sexual dysfunction by using the BSFI Results The BSFI63 contains eleven questions that cover five domains Sexual dysfunction over time and of sexual function:1)sexual drive (two items);2)erectile between verum and placebo groups function (three items);3)ejaculatory function (two items): Tables I and 2 show the descriptive and statistical overview 4)sexual problem assessment (three items);and 5)sexual of sexual function levels separately by assessment time(base- satisfaction(one item).Answers are given on a 5-point Lik- line,week 4,and 8 weeks later [at the end of the study])and ert scale with scores ranging from 0(none,big problem,or group (verum versus placebo). no activity)to 4(always,no problem,or high activity)and Sexual function improved significantly over time.Post with lower mean scores reflect greater sexual dysfunction hoc analyses with Bonferroni-Holm corrections for P-values (Cronbach's alpha =0.91). showed that sexual dysfunction reduced from week 4 to week 8.All effect sizes were large.No statistically significant Statistical analysis differences between the groups were observed. Preliminary calculations Significant time x group interactions were observed for To detect possible confounders,Spearman's correlations all sexual function variables.Effect sizes were large.Post were computed between sociodemographic data (age,edu- hoc analyses with Bonferroni-Holm corrections for P-values cation,civil status,medication intake,number of children) showed that sexual dysfunction was lower in the verum than and indices of sexual dysfunction and depressive symptoms. in the placebo group at week 8.Figure 2 shows the mean All correlation coefficients were between-0.05 and 0.15 values for the two groups over the three time points. (Ps >0.56);accordingly,sociodemographic data were not introduced as possible confounders. Depressive symptoms Next,a series of Pearson's correlations was performed Patients rated their depressive symptoms via the BDI at between dimensions of sexual dysfunction and depres- baseline and at the end of the study (Tables 1 and 2 and sive symptoms.Further,a series of analyses of variances Figure 3).Depressive symptoms declined over time;the (ANOVAs)for repeated measures was performed with the significant time x group interaction showed that depressive factors time (baseline,4 weeks,and 8 weeks)and group symptoms declined more in the verum group than in the (verum versus placebo),and with sexual dysfunction areas placebo group. (sexual drive,erections,ejaculations,problem assessment, overall satisfaction,and overall mean score)as dependent Correlations between sexual dysfunction variables.Additionally,another ANOVA was computed and symptoms of depression with the factors time(baseline,8 weeks)and group(verum Correlating symptoms of depression at baseline and after versus placebo),with the dependent variable BDI score,and 8 weeks (the end of the study)with sexual dysfunction with Bonferroni-Holm corrections for P-values.Because of showed that symptoms of depression and sexual dysfunction deviations from sphericity,ANOVAs for repeated measures were unrelated (allrs<0.111,ps>0.41).When correlations (for the factor time with three values)were performed using were calculated separately for the verum and placebo group, Greenhouse-Geisser corrected degrees of freedom,although the following pattern of results was observed:for the verum the original degrees of freedom are reported with the relevant group,again,symptoms of depression and sexual function 630 submit your manus Neuropsychiatric Disease and Treatment 2015:II DovepressNeuropsychiatric Disease and Treatment 2015:11 submit your manuscript | www.dovepress.com Dovepress Dovepress 630 Farnia et al Instruments Self-assessment of depressive symptoms by using the BDI Patients completed the BDI,64 which a self-report of symp￾toms of depression. The questionnaire consists of 21 items and covers such areas as depressive mood, loss of appetite, sleep disorders, and suicidal thoughts. Answers are given on 4-point Likert scales with the anchor points 0 (for “as always” or “no change”) and 3 (for “not able anymore” or “dramatic change”) and with higher scores reflecting greater severity of depressive symptoms (Cronbach’s alpha =0.89). Self-assessment of sexual dysfunction by using the BSFI The BSFI63 contains eleven questions that cover five domains of sexual function: 1) sexual drive (two items); 2) erectile function (three items); 3) ejaculatory function (two items); 4) sexual problem assessment (three items); and 5) sexual satisfaction (one item). Answers are given on a 5-point Lik￾ert scale with scores ranging from 0 (none, big problem, or no activity) to 4 (always, no problem, or high activity) and with lower mean scores reflect greater sexual dysfunction (Cronbach’s alpha =0.91). Statistical analysis Preliminary calculations To detect possible confounders, Spearman’s correlations were computed between sociodemographic data (age, edu￾cation, civil status, medication intake, number of children) and indices of sexual dysfunction and depressive symptoms. All correlation coefficients were between -0.05 and 0.15 (Ps .0.56); accordingly, sociodemographic data were not introduced as possible confounders. Next, a series of Pearson’s correlations was performed between dimensions of sexual dysfunction and depres￾sive symptoms. Further, a series of analyses of variances (ANOVAs) for repeated measures was performed with the factors time (baseline, 4 weeks, and 8 weeks) and group (verum versus placebo), and with sexual dysfunction areas (sexual drive, erections, ejaculations, problem assessment, overall satisfaction, and overall mean score) as dependent variables. Additionally, another ANOVA was computed with the factors time (baseline, 8 weeks) and group (verum versus placebo), with the dependent variable BDI score, and with Bonferroni–Holm corrections for P-values. Because of deviations from sphericity, ANOVAs for repeated measures (for the factor time with three values) were performed using Greenhouse–Geisser corrected degrees of freedom, although the original degrees of freedom are reported with the relevant Greenhouse–Geisser epsilon value (ε). For ANOVAs, effect sizes are indicated with the partial eta squared (η2 ), with 0.059 $ η2 $0.01 indicating small (S), 0.139 $ η2 $0.06 indicating medium (M), and η2 $0.14 indicating large (L) effect sizes. All computations were performed as intention￾to-treat and the last observation carried forward. The nominal alpha-level was set at 0.05; post hoc analy￾ses were performed with Bonferroni–Holm corrections of P-values for multiple testing. Statistical analyses were per￾formed with SPSS® 20.0 (IBM Corporation, Armonk, NY, USA) for Apple MacIntosh®. Results Sexual dysfunction over time and between verum and placebo groups Tables 1 and 2 show the descriptive and statistical overview of sexual function levels separately by assessment time (base￾line, week 4, and 8 weeks later [at the end of the study]) and group (verum versus placebo). Sexual function improved significantly over time. Post hoc analyses with Bonferroni–Holm corrections for P-values showed that sexual dysfunction reduced from week 4 to week 8. All effect sizes were large. No statistically significant differences between the groups were observed. Significant time × group interactions were observed for all sexual function variables. Effect sizes were large. Post hoc analyses with Bonferroni–Holm corrections for P-values showed that sexual dysfunction was lower in the verum than in the placebo group at week 8. Figure 2 shows the mean values for the two groups over the three time points. Depressive symptoms Patients rated their depressive symptoms via the BDI at baseline and at the end of the study (Tables 1 and 2 and Figure 3). Depressive symptoms declined over time; the significant time × group interaction showed that depressive symptoms declined more in the verum group than in the placebo group. Correlations between sexual dysfunction and symptoms of depression Correlating symptoms of depression at baseline and after 8 weeks (the end of the study) with sexual dysfunction showed that symptoms of depression and sexual dysfunction were unrelated (all rs,0.11, ps.0.41). When correlations were calculated separately for the verum and placebo group, the following pattern of results was observed: for the verum group, again, symptoms of depression and sexual function