Multivariate logistic regression of the association between proximal exposure to violencea and current mental health.

awitnessed, victimised and perpetrated categories are overlapping.cthe reference group is all participants who have not witnessed violence in the past 12 months.dthe reference group is all participants who have reported no victimisation in the past 12 months.ethe reference group is all participants who have not perpetrated violence in the past 12 months.fmodel adjusted for age and gender.gmodel adjusted for age, gender, ethnicity, employment, education, household income and migrant status.hmodel adjusted for all confounders and for the other two categories of violence.*p**p***p