Heterogeneity

Subgroup analyses will examine variation by setting, demographics, tool type, administration mode, administrator type, and epilepsy subtype. If heterogeneity prevents meaningful pooled estimates, the analysis shifts to medians and interquartile ranges. Formal subgroup pooling will require at least five studies in a subgroup. Country-level incidence rates are calculated by dividing summed deaths by summed annual incarcerated populations and multiplying by 100,000. Standardised Mortality Ratios divide prison mortality incidence by sex-standardised general population mortality for the same country. Random-effects models are fitted by restricted maximum likelihood for heterogeneity estimation. The review will assess heterogeneity both graphically and statistically. Random-effects meta-regression is used to identify ecological factors associated with mortality. Heterogeneity is assessed using I², with a fixed-effect model applied when I² is below 50% and a random-effects model when I² is 50% or above with persistent heterogeneity. When Egger's test indicates significant publication bias, the trim-and-fill method will estimate missing studies and recalculate adjusted pooled effect sizes…