A confirmatory study of violence risk assessment tool (M55) and demographic predictors of patient violence

Abstract

Aim: The aims of this study were to confirm the usefulness of Violence Risk Assessment Tool (M55) for prospectively identifying violent patients in medical-surgical units and to explore predictors of Code-55 activation for violent behaviour or physical attack.

Background: A previous retrospective case-controlled study claimed moderate sensitivity and high specificity of the M55 in identifying potentially violent patients. However, the usefulness of M55 needs to be confirmed in a prospective study before wide adoption.

Methods: A prospective cohort study of patients in medical-surgical units of a hospital in southern California was conducted from August 2009 to December 2009. A checklist containing the M55 was completed on admission. Following a violent event or just prior to discharge, the violent event outcome section of the data collection tool was completed. The sensitivity and specificity of M55 for identifying the violent patients were calculated using the original M55 flagging criteria.

Results: Of 2063 patients, 32 (1·6%) had a violent event of Code-55 activation (12), physical attack (26) or both (6). The M55 showed a low sensitivity (41%) and acceptable specificity (99%) for prospectively identifying violent patients. The multivariate logistic regression model indicated higher odds of violent events for patients ≥70 years of age [odds ratio (OR) = 2·3; 95% confidence interval (CI): 0·9-5·7], males (OR = 2·7; 95% CI: 1·2-6·1) or Caucasian (OR = 3·4; 95% CI: 0·8-14·7).

Conclusion: The M55 does not appear to be useful for prospectively identifying violent patients in medical-surgical units because of low sensitivity. Older, male or Caucasian patients had higher odds of becoming violent.