Objective: To simultaneously evaluate actuarial and dynamic predictors of severe in-patient violence among women involuntarily admitted for severe drug abuse.
Method: All patients admitted to special facilities for involuntary treatment of absconding-prone, previously violent, drug abusing women in Sweden were assessed with the Staff Observation Aggression Scale, revised. Actuarial data on risk factors for violence were collected and considered in an extended Cox proportional hazards model with multiple events and daily assessments of the Broset Violence Checklist as time-dependent covariates.
Results: Low-grade violence and being influenced by illicit drugs were the best predictors of severe violence within 24 h. Significant differences in risk for violence between different institutions were also found.
Conclusion: In-patient violence risk is rapidly varying over time with being influenced by illicit drugs and exhibiting low-grade violence being significant dynamic predictors. Differences in violence between patients could not be explained by patient characteristics.