Violence and Suicidal/Nonsuicidal Self-Injury Among Adolescents Undergoing Residential Treatment: An Examination of the Predictive Validity of the SAVRY, START:AV, and VRS-YV

• Published date: 01 July 2023
• DOI: http://doi.org/10.1177/00938548231165531
• Author: Andrew L. Gray,Jodi L. Viljoen
• Date: 01 July 2023
• Subject Matter: Articles

CRIMINAL JUSTICE AND BEHAVIOR, 2023, Vol. 50, No. 7, July 2023, 931 –952.

DOI: https://doi.org/10.1177/00938548231165531

Article reuse guidelines: sagepub.com/journals-permissions

© 2023 International Association for Correctional and Forensic Psychology

931

VIOLENCE AND SUICIDAL/NONSUICIDAL SELF-

INJURY AMONG ADOLESCENTS UNDERGOING

RESIDENTIAL TREATMENT

An Examination of the Predictive Validity of the

SAVRY, START:AV, and VRS-YV

ANDREW L. GRAY

Université de Montréal

JODI L. VILJOEN

Simon Fraser University

Using a retrospective study design, predictive validity of the Structured Assessment of Violence Risk in Youth, Short-Term

Assessment of Risk and Treatability: Adolescent Version (START: AV), and the Violence Risk Scale–Youth Version

(VRS-YV) was examined among 87 adolescents referred to a residential treatment program. With few exceptions, moderate

to high accuracy was achieved for the three measures in predicting violence and suicidal/nonsuicidal self-injury occurring

during the adolescents’ time in treatment. Accuracy of the measures peaked within 90 days for violence and gradually

increased over the 180-day follow-up for suicidal/nonsuicidal self-injury. Dynamic factors were more predictive of repeated

events involving violence relative to static/historical factors, whereas only factors from the START: AV were predictive of

repeated events involving suicidal/nonsuicidal self-injury. These results emphasize the need for further examining the risk of

adverse outcomes beyond violence among adolescents.

Keywords: risk assessment; dynamic risk; adolescence; violence; mental illness; suicide; SAVRY

AUTHORS’ NOTE: It is with great sadness that Andrew Gray reports the passing of his mentor and co-author,

Dr. Jodi L. Viljoen. Jodi supervised Andrew during the completion of his PhD, and he is forever grateful for her

support and guidance. Her contributions have had a substantial impact on the field of adolescent risk assess-

ment, and she will be sorely missed. This article is partially based on the dissertation completed by Andrew

(Gray, 2021) while under the supervision of Jodi. It is dedicated to her memory. Jodi L. Viljoen was the lead

author of the Short-Term Assessment of Risk and Treatability: Adolescent Version (START: AV; Viljoen et al.,

2014). She did not receive any proceeds from its sale but had completed several paid trainings on the START:

AV. She was not involved in data collection and did not conduct any analyses for the current study. All exclusion

criteria and analyses were determined a priori by Andrew L. Gray. This research was supported by a Doctoral

Fellowship awarded to Andrew L. Gray from the Social Sciences and Humanities Research Council of Canada.

The authors wish to thank the British Columbia Ministry of Children and Family Development for supporting

this research. Correspondence concerning this article should be addressed to Andrew L. Gray, Département de

Psychiatrie et d’addictologie, Université de Montréal, 2900 boul. Édouard-Montpetit, Montréal, Quebec H3T

1J4, Canada; e-mail: andrew.gray.1@umontreal.ca.

1165531CJBXXX10.1177/00938548231165531Criminal Justice and BehaviorGray, Viljoen / Violence and Suicidal/Non-Suicidal Self-Injury

research-article2023

932 CRIMINAL JUSTICE AND BEHAVIOR

Adolescent risk assessment is an important and complex process, with mental health

professionals having an ethical and professional duty to assess and manage risk.

Despite substantial strides made in the development of adolescent risk assessment mea-

sures, much of what is known has been drawn from the psychiatric and psychological litera-

tures with the primary focus being the assessment of risk for violence and general reoffending

(Menon, 2013). Although adolescent violence remains an important area of consideration,

assessing the risk of other adverse outcomes such as suicidal behavior and nonsuicidal self-

injury (NSSI) is not only warranted by the empirical evidence but is also a requirement of

mental health law (e.g., involuntary hospitalization) and many mental health agencies.

Rates of suicidal behavior and NSSI are high among adolescents (Labelle et al., 2015), with

suicide being the second most common cause of death worldwide among youth (Hawton

et al., 2012).

Outcomes such as violence and self-injury are especially problematic among youth

accessing residential treatment. For instance, among a large sample of children and adoles-

cents undergoing residential (n = 9,942) and nonresidential (n = 525) treatment, those in

residential care exhibited significantly higher pretreatment rates for behavioral problems

(e.g., violence, aggression; 80.3% vs. 68.6%, respectively), self-injury (28.4% vs. 11.7%,

respectively), and suicidality (i.e., ideation and attempts; 29.5% vs. 12.9%, respectively;

Briggs et al., 2012). When compared to adults, children and adolescents treated within resi-

dential and in-patient settings have a much greater likelihood to harm themselves or others

(Stewart & Hirdes, 2015), and those displaying increased levels of reactive aggression are

at heightened risk to engage in suicidal behaviors (e.g., attempted or completed suicide;

Hartley et al., 2018).

Such findings reinforce the need to identify adolescents at risk of adverse outcomes,

particularly those within residential treatment settings; however, this area of research has

remained largely unexamined. Although there is strong empirical support for assessing risk

in adolescents using structured risk assessment measures (Viljoen, Gray, & Barone, 2016),

many of the available measures meeting criteria for evidence-based tools are designed to

assess risk of violence or reoffending (see A. L. Gray et al., 2019). As time and resources

(e.g., specialized training) are required to administer a risk assessment measure, the ques-

tion remains whether the information gathered (e.g., risk factor ratings) can help inform

clinical decision-making regarding other adverse outcomes (e.g., NSSI). At present, there is

a growing body of evidence suggesting substantial overlap among risk factors for various

adverse outcomes (see Viljoen, Nicholls, et al., 2016, for a review). For instance, violence

and suicide are known to coincide (e.g., murder-suicide; Douglas et al., 2013), and risk fac-

tors for violence such as impulsivity, childhood abuse/neglect, and previous self-harm have

been found to be significantly predictive of subsequent self-injury/suicide attempts (e.g.,

Favril et al., 2020; K. R. Fox et al., 2015; McMahon et al., 2018). In predicting self-harm

among adult forensic psychiatric samples, moderate to large effects have been observed

across multiple research studies examining the accuracy of the Historical-Clinical-Risk

Management-20 (HCR-20; Douglas et al., 2013), an adult risk assessment measure designed

to assess risk of violence (Campbell & Beech, 2018; Daffern & Howells, 2007; Fagan et al.,

2009; N. S. Gray et al., 2003; O’Shea et al., 2014).

This concept of overlapping risk factors across outcomes is best illustrated by the devel-

opment of the Short-Term Assessment of Risk and Treatability (START; Nicholls et al.,

2021; Webster et al., 2004), a structured professional judgment (SPJ) measure comprised of