Predictors of Seclusion and Restraint Following Injurious Assaults on Psychiatric Units.

Document Type


Publication Date



doi: 10.1097/PTS.0000000000000657


OBJECTIVES: The aim of this study was to describe use of seclusion and restraint after injurious assaults by psychiatric inpatients in U.S. hospitals, including examination of hospital, unit, assaultive patient, and assault characteristics as predictors of seclusion/restraint use.

METHODS: Data from 2004 to 2017 on 23,630 injurious assaults reported by 747 psychiatric units in 482 general hospitals were analyzed. Odds of seclusion, odds of three restraint types (device, hold, pharmacological), and duration of seclusion and device restraint were modeled as functions of hospital, unit, assaultive patient, and assault characteristics.

RESULTS: Compared with teaching hospitals, nonteaching hospitals had lower rates of seclusion but higher rates of all three types of restraint. Seclusion and restraint rates were lower in government hospitals and hospitals in metropolitan settings. Pharmacological restraint was most common in for-profit hospitals; seclusion was most common in nonprofit hospitals. Episodes of seclusion and device restraint were approximately 20% shorter in teaching hospitals than in nonteaching hospitals and lasted markedly longer in federal government hospitals. Hospitals in metropolitan settings reported 30% longer time spent in seclusion, on average. Involuntary admission was associated with higher odds of seclusion, device restraint, and hold. Female patients were less likely than males to be restrained with a device, and their seclusion and device restraint episodes tended to be shorter. The number of persons injured in an assault predicted odds of seclusion and all three types of restraint. The maximum level of injury sustained predicted odds of seclusion, device restraint, and hold, as well as duration of both seclusion and device restraint. Odds of seclusion/restraint were lower when the most severely injured person was a patient rather than a clinical health care worker.

CONCLUSIONS: Inconsistencies were observed in seclusion and restraint use, which varied by hospital type, patient sex and admission status, maximum level of injury sustained, and type of person most severely injured. Thus, there may be room for improvement in hospital and unit policies and practices. More comprehensive data are needed for further research on use of seclusion and restraint in response to incidents other than injurious assault.

Journal Title

J Patient Saf





First Page


Last Page


Library Record