Patients admitted to child and adolescent psychiatric inpatient units often are admitted for treatment of acute aggressive behaviors (Baeza et al., 2013; Barzman et al., 2011; Calhoun et al., 2022; Pikard et al., 2018). Caring for patients at risk for aggressive behavior within inpatient settings is notoriously difficult (Tompsett et al., 2011) and can have serious consequences for physical harm, emotional trauma, and further hospitalization (Baeza et al., 2013). Nursing staff in mental health treatment settings experience rates of physical assault three times higher than other nursing positions (Edward et al., 2016). Aggressive patient interactions (APIs) often result in patient seclusion or restraint events (SREs) which are associated with host of negative outcomes including injury and patient death (Chieze et al., 2019; Chung et al., 2023; LeBel and Goldstein, 2005; National Association of State Mental Health Program Directors, 1999). Further, minimal evidence for the benefits of SRE has been documented and the overwhelming consensus is for the reduction and eradication of SREs (Perers et al., 2022; The Joint Commission, 2012). Unfortunately, past research has failed to produce clinically useful tools for identifying the patients at highest risk of needing SREs or causing API events.

Although risk factors for aggression have been identified in adult psychiatric patients (Iozzino et al., 2015), research in child and adolescent populations has yielded inconsistent evidence, particularly in relation to aggressive behavior resulting in injuries to others (Vidal et al., 2020). For example, research has found that both younger (Timbo et al., 2016) and older patient age (Martin et al., 2008) is associated with increased risk of SREs and aggression. Similarly, while some research suggests that male patients are at increased risk of SREs (Delaney and Fogg, 2005; Nielson et al., 2021; Vidal et al., 2020), others have found that female patients are at increased risk of multiple SREs (Furre et al., 2016; Muir-Cochrane et al., 2014).

There is more consistent evidence that certain diagnostic categories are associated with increased risk of aggression and SREs including developmental disorders, disruptive behavior disorders, psychotic disorder(s), and, to a lesser extent, internalizing disorders (Nielson et al., 2021). The evidence is strong for the association between developmental disorders and SREs, with past research suggesting that level of psychosocial functioning is a better predictor of SRE than diagnostic status (Furre et al., 2014, 2016; Stewart et al., 2013). Prior research has consistently found that small numbers of patients account for a large proportion of restraint events and staff injuries (Furre et al., 2017; Nielson et al., 2021). To our knowledge, only one study examined the characteristics of adolescent patients with multiple restraint episodes (Furre et al., 2017). Using a national dataset of 126 beds and 267 adolescent patients who had experienced at least one restraint, they found that 18% of those patients accounted for 77% of the restraint events. Factors significantly associated with multiple restraint episodes were female gender, low global psychosocial functioning, longer hospital stays and repeated admissions, and use of “Pro re nata” (PRN) or “as needed” medication. With this gap, it is critical to better identify patients at high risk of multiple SREs, as these patients are likely to experience negative outcomes such as physical injury, longer hospital stays and repeated admissions (Delaney and Fogg, 2005; Furre et al., 2014; Pogge et al., 2013). Tompsett et al., used a prospective measure of risk of aggressive behavior to distinguish between youth involved in and not involved in restraint incidents (Tompsett et al., 2011). However, their measure had higher specificity (predicting patients not involved in restraints) and only by including additional risk factors and patient information were they able to identify patients involved in a restraint and were unable to predict risk of multiple restraints (i.e., higher-risk patients). Finally, some have argued that SREs are “clustered,” suggesting that one SRE increases the likelihood of further SREs (Leidy et al., 2006); but this hypothesis has not been directly tested in well-designed studies (nor any examining API events).

No research has examined screening tools for identifying within inpatient settings those patients who are at high risk for SREs and APIs. This gap in knowledge limits effective utilization of resources and interventions to mitigate (or prevent) these outcomes. Identifying high-risk patients at admission is necessary to effectively treat those at higher risk (and arguably also those at low risk). Without accurate identification, staff are faced with either providing high-resource interventions to all patients (inefficient) or providing interventions only after an SRE or API event (ineffective). Accurate screening for identifying high-risk patients is critically needed to deploy resources efficiently and effectively to patients likely to engage in aggressive behavior preventing negative outcomes.

We sought to address this critical gap in knowledge by developing a screening tool termed “high-risk notification” (HRN) as part of a quality improvement initiative to identify patients at admission at highest risk of SRE and/or APIs. We report here the results evaluating the utility of the HRN method in relation to SRE and API events. The aims of the project were: 1) to demonstrate use of the HRN method to accurately identify patients at high risk for SREs and APIs in a pediatric hospital; 2) to measure the strength of the association between HRN status and the occurrence of SREs or APIs during a hospital stay; and 3) to examine the influence of HRN at admission, occurrence of SREs, and occurrence of APIs in predicting subsequent SREs and APIs.