All main effects for PLE are significant (P < 0.001). For SelfRepDep, with each additional PLE the odds of being one SelfRepDep category higher are ~1.6 (from ordinal regressions) or 0.36 SD units on a normal underlying depression scale (from IRT analyses). Strikingly, for ClinDep and suicidality, these odds are ~1.3 and ~1.5, respectively or ~0.15 and ~0.23 SD units, respectively, even though ClinDep/suicidality represents clinical diagnoses measured 1–10 years after PLE, whereas SelfRepDep is measured in the period immediately after PLE. The association with NLE is in the same direction but only evident for ClinDep and SelfRepDep (not suicidality) where the effect is weaker (odds ratio of ~1.06 or ~0.04 SD units) and the P-value is inconsistent (ns to P < 0.01).
