Early detection of risk factors for seclusion and restraint: a prospective study.

Aim: The study aims to examine the predictive power of static and dynamic risk factors assessed at admission to an acute psychiatric ward and to develop a prediction model evaluating the risk of seclusion and restraint. Methods: Over 20 months, data on demographic and clinical characteristics, psychosocial functioning, level of insight, uncooperativeness, and use of coercive measures were collected prospectively on 520 patients at admission. Logistic regression analysis was used to develop a prediction model. The magnitude of the predictive power of this model was estimated using receiver operating characteristic analysis. Results: The prediction model contained one static predictor (involuntary commitment) and two dynamic predictors (psychological impairment and uncooperativeness), with a high predictive power (receiver operating characteristic area under the curve = 0.83). The final risk model classified 72% of the patients correctly, with a higher sensitivity rate (80%) than specificity rate (71%). Conclusion: Early assessment of patients' psychological impairment and uncooperativeness can help clinicians to recognize patients at risk for coercive measures and approach them on time with preventive and less restrictive interventions. Although this simple, highly predictive model accurately predicts the risk of seclusion or restraint, further validation studies are needed before it can be adopted into routine clinical practice.