Extended Data Table 6 Sensitivity analysis of the primary outcomes analyzed with linear mixed models was assessed at post-intervention, 6-month follow-up, and 12-month follow-up for trainer teachers (n = 223)

From: School-based suicide prevention using the gatekeeper programme: a cluster-randomized trial

  1. Statistical analysis was performed using linear mixed models to account for within-subject correlations and school-level clustering for the primary outcomes among trainer teachers. Controlled variables included age, gender, education level, family income, religion, and years of work. The model includes both main effects and interaction effects. Main effects include Group (representing the difference between the intervention and control groups), Time2, Time3, and Time4 (representing different time points compared to baseline), and interaction effects such as Group*Time2, Group*Time3, and Group*Time4, which examine how the intervention effect varies at different time points. For example, Group*Time2 represents the combined effect of being in the intervention group at the post-intervention timepoint. P-values were two-sided, and multiple comparisons were adjusted using the false discovery rate (FDR) correction.
  2. The results from the linear mixed models, after controlling for covariates and site effects, revealed significant main effects of group and time, as well as a significant interaction effect between group and time for all four primary outcomes among trainer teachers. Specifically, the interaction effects between group and time were of particular interest, as they highlighted the differential impact of the intervention across time points for the trainer teachers. The results indicate that LGTP intervention significantly reduced stigma against suicide at T2 (β = -1.27, 95% CI: [-1.50, -1.03], PFDR < 2.0×10−16), T3 (β = - 0.88, 95% CI: [-1.11, -0.64], PFDR = 2.8×10−12), and T4 (β = - 0.93, 95% CI: [-1.17, -0.70], PFDR = 1.3×10−13) compared to the control group. Additionally, the intervention significantly increased suicide literacy at T2 (β = 1.00, 95% CI: [0.75, 1.24], PFDR = 4.8×10−14), T3 (β = 0.90, 95% CI: [0.86, 1.14], PFDR = 1.5×10−11), and T4 (β = 0.77, 95% CI: [0.52, 1.02], PFDR = 6.9×10−9) relative to the control group. Furthermore, the intervention significantly increased the perceived competence at T2 (β = 1.34, 95% CI: [1.13, 1.54], PFDR < 2.0×10−16), T3 (β = 1.00, 95% CI: [0.80, 1.21], PFDR < 2.0×10−16), and T4 (β = 0.78, 95% CI: [0.58, 0.98], PFDR = 3.4×10−13) compared to the control group. Lastly, the intervention significantly increased the willingness to intervene at T2 (β = 0.92, 95% CI: [0.68, 1.16], PFDR = 9.2×10−13), T3 (β = 1.14, 95% CI: [0.90, 1.38], PFDR < 2.0×10−16), and T4 (β = 0.46, 95% CI: [0.22, 0.71], PFDR = 5.8×10−4) in comparison to the control group.
Back to article page