Patient Risk Factors for Violent Restraint Use in a Children's Hospital Medical Unit

Abstract

Background and objectives: To inform efforts to reduce violent restraint use, we examined risk factors for restraint use among hospitalized children with known behavior concerns.

Methods: We conducted a retrospective cross-sectional study of restraint events in all hospitalizations from 2017 to 2019 on a 10-bed medical-surgical unit with dedicated mental health clinician support. We examined characteristics of restraint events, used adjusted logistic regression models to identify independent risk factors for restraint use, and used an adjusted Poisson regression model to determine the adjusted rate of restraint events per hospital day.

Results: The sample included 1507 hospitalizations representing 1235 patients. Among included hospitalizations, 48% were for a psychiatric indication awaiting transfer to an inpatient psychiatric unit, and 52% were for a primary medical or surgical problem. Sixteen percent had a restraint event. Patient demographic characteristics were not associated with risk of a restraint event. Having a psychiatric indication for hospitalization was an independent risk factor for restraint use (odds ratio: 2.85; 95% confidence interval: 2.06-3.94). Rate of restraint use per day decreased as length of stay increased; hospitalizations lasting 9 days or longer had a 58% lower rate of restraint use per day than 1- to 2-day hospitalizations (P < .001).

Conclusions: Interventions to reduce restraint use may benefit from incorporating information about a patient's psychiatric risk factors, including type and number of diagnoses and reason for hospitalization. Future efforts could investigate whether providing enhanced behavior supports during the first several days of a patient's hospitalization reduces violent restraint use.

FIGURE 1

Histograms revealing the timing and frequency of restraint events. A, This histogram reveals the proportion of restraint events that occurred on each hospital day. B, This histogram reveals the number of violent restraint events per hospitalization.