Table 3.
Mediation analysis* results
|
Total effect |
Direct effect |
Indirect effect |
Percentage mediated | ||||
|---|---|---|---|---|---|---|---|
| Effect size (95% CI) | p value | Effect size (95% CI) | p value | Effect size (95% CI) | p value | ||
| Paranoia (GPTS) outcome at week 10 | |||||||
| Insomnia at week 3 (SCI-8) | −2·27 (−3·03 to −1·51) | <0·0001 | −1·85 (−2·66 to 1·04) | <0·0001 | −0·67 (−0·86 to −0·48) | <0·0001 | 29·5% |
| Insomnia at week 10 (SCI-8) | −2·27 (−3·03 to −1·51) | <0·0001 | −0·97 (−1·80 to −0·14) | <0·0001 | −1·31 (−1·60 to −1·02) | <0·0001 | 57·8% |
| Hallucinations (SPEQ) outcome at week 10 | |||||||
| Insomnia at week 3 (SCI-8) | −1·60 (−2·00 to −1·20) | <0·0001 | −1·36 (−1·79 to −0·94) | <0·0001 | −0·33 (−0·43 to −0·23) | <0·0001 | 20·7% |
| Insomnia at week 10 (SCI-8) | −1·60 (−2·00 to −1·20) | <0·0001 | −0·90 (−1·34 to −0·46) | <0·0001 | −0·62 (−0·78 to −0·46) | <0·0001 | 38·6% |
Total n=1718. GPTS=Green et al Paranoid Thought Scales. SCI-8=Sleep Condition Indicator 8-item version. SPEQ=Specific Psychotic Experiences Questionnaire.
*
Outcome and mediators modelled by means of linear mixed effects models and the total, direct, and indirect effects determined using the Baron and Kenny31 approach. The effect size is the adjusted treatment difference (ie, non-standardised treatment difference).