Fig. 4

Hazard ratios based on number of prior relapses (primary proxy). Hazard ratios represent the hazard of relapse for those having the specified number of prior relapses relative to those with 0 relapses. Covariates in the Cox model included prior number of relapses, the calendar year, gender and age class at the start of follow-up. Primary proxy: identified a relapse episode based on a psychiatric hospitalisation ≥7 days (n=2994). Error bars represent 95% CIs. Statistical significance was estimated by the 95% CI. If the 95% CI included 1, then the time to relapse was not statistically significantly different from the reference group. If the 95% CI excluded 1, then time to relapse was significantly different from the reference group. CI: confidence interval