“Don’t know” scores should be considered when assessing violence risk for youth in acute institutions

Commonly, relevant information to score violence risk assessment instruments is missing at the time of assessment. While there are indications that lack of information to score items, or “Don’t know” scores, has clinical relevance, these items are commonly omitted or treated as “No” scores in research. The Violence Risk Assessment Checklist for Youth (V-RISK-Y) is a screening instrument designed to identify violence risk in youth aged 12–18. The aim of this study is to assess whether “Don’t know” scores on V-RISK-Y are associated with increased risk for registered violent events for youth during acute institutional stays as compared to “No” scores. This study utilized data from the V-RISK-Y multicenter study, consisting of a sample of 517 youth from child and adolescent psychiatry and residential youth care institutions. The following secondary analyses were performed: (i) the frequencies of “Don’t know” and “No” scores for each item and registered violent events, and (ii) effect sizes from item-level logistic regression analyses of “Don’t know” scores with “No” scores as reference, controlling for sex and type of institution. Findings show more registered violent events for youth during institutional stay when items were scored as “Don’t know” as compared to “No”. Nine of the 12 items had odds ratios (ORs) above 1 for “Don’t know” for recorded violence when controlled for sex and type of institution. One item, “Lack of empathy”, reached significance. The positive ORs indicate that “Don’t know” scores may have clinical relevance despite not reaching statistical significance and that these scores should be considered both in clinical practice and in research.

1 Introduction

Rapid identification of violence risk is often required in acute institutional settings. For instance, in inpatient psychiatric units, a high proportion of violent episodes happen early during the institutional stay (1, 2). Furthermore, in residential care institutions for children and adolescents, youth may be placed in institutions involuntarily due to behavioral issues, which may signify violence risk (3). However, information relevant to determine risk may be unknown at admission point (4, 5). For example, in emergency departments, patient history might be unavailable at intake, and rapid collection of relevant information might be logistically challenging (5). While research assessing the relationship between “Don’t know” scores and violence risk is scarce, Eriksen and colleagues (6) found that “Don’t know” scores were significantly associated with violence in an inpatient psychiatric adult population. These findings were supported in a more recent archival study on V-RISK-10 (7). In research, items with “Don’t know” scores have typically been omitted or treated as missing in statistical analyses (6, 7), and missing data have commonly been excluded (e.g. 8, 9). However, risk assessment scores should be based on all instrument items, and omitting these items can alter the structure of the instrument and impact predictive accuracy (10). Furthermore, it is pointed out that whether lack of information to score items in violence risk assessments impacts the instruments’ performance is currently unknown (11). Thus, “Don’t know” scores may have implications for both clinical practice and research, but there is a scarcity of literature exploring this topic.

1.1 Approaches to violence risk assessment

Currently, there are two main approaches to structured violence risk assessment. First, actuarial risk assessments are conducted by rating the presence of empirically derived risk factors and calculating a score based on these factors (12). The risk assessment is determined solely based on this score, without including subjective evaluations (13). Second, the structured professional judgment (SPJ) approach also relies on validated risk factors, but incorporates clinical judgment (14). Violence risk is typically assessed as “Low”, “Moderate”, or “High” based on present risk factors as well as consideration of other relevant factors like setting and available resources (15). By allowing some professional discretion, SPJ tools allow for consideration of individual and dynamic factors in the risk assessment (16, 17). On the other hand, allowing for this flexibility may introduce bias and increase variability (18). Both types of instruments perform similarly in terms of predictive validity (15, 19). However, in treatment settings, the SPJ approach is typically preferred because the inclusion of individual factors can help inform treatment plans and violence risk management strategies (20, 21). Several violence risk assessments are used with youth populations (22). Most research in this field is conducted in forensic settings, but it is increasingly expected or mandated that identification of violence risk takes place in healthcare and social services (23).

1.2 Violence risk screening in acute institutional settings

While several violence risk assessments have been validated for use in mental health patient populations, they typically require training and lengthy administration (e.g. 24, 25). In acute institutions, however, time and resources to administer comprehensive risk assessments may not be available (26). Accommodating these settings, briefer violence risk screening instruments have been developed for adult populations to rapidly identify violence risk and implement relevant interventions if needed (27, 28). V-RISK-10 (29) is currently the internationally recommended violence risk screener for acute mental healthcare contexts (27). For children and adolescents, comparable screening tools for use in acute settings are largely absent. Therefore, the Violence Risk Assessment Checklist for Youth (V-RISK-Y) was developed.

V-RISK-Y is a 12-item violence risk screening instrument developed to identify violence risk in youth aged 12–18 in acute institutions (English version freely available from sifer.no). The interrater reliability of V-RISK-Y was assessed in a case vignette study, which found high agreement between raters for the overall risk assessment (30). The predictive validity of V-RISK-Y was assessed in two studies from a multicenter project in four child and adolescent psychiatric units and four residential youth care institutions. V-RISK-Y had good predictive accuracy for violent behavior during institutional stay (31), with eight individual items as significant predictors of violence (32). In these studies, “Don’t know” scores were weighted in analyses, and item scores coded as “No” = 0; “Don’t know” = 1; “Maybe/Moderate” = 2; and “Yes” = 3. Items should be scored with “Don’t know” in instances where raters do not have sufficient information to conclude on the presence of a risk factor. Thus, items with these scores signify unavailable information but are included in the risk screening. In contrast, missing data stems from unscored items. Linearity of the item variables was established using fractional polynomials. “Don’t know” scores were highly prevalent for several items, and logistic regression analyses indicated that “Don’t know” scorings had substantial effect sizes and significantly predicted violent events for three V-RISK-Y items (Poor insight, Lack of empathy, and Unrealistic planning) (32). In these analyses, however, sex and type of institution were not controlled for. Both were significant predictors of violent events in previous V-RISK-Y studies (26, 31).

1.3 Aim

The objective of this brief report is to explore the impact of “Don’t know” scores on the predictive accuracy of V-RISK-Y by expanding on findings from the V-RISK-Y multicenter project. Specifically, the aim is to compare the predictive validity of “Don’t know” scores (i.e., not enough information to score an item) and “No” scores (i.e., risk factor excluded). It is hypothesized that “Don’t know” scores will be more predictive of violence risk than “No” scores.

2 Methods

2.1 Design and participants

This study leveraged data from the V-RISK-Y multicenter study (see 31 for detailed study and sample description). The multicenter study had a naturalistic, prospective, observational design and included all youth (n = 517) with institutional stays in four acute child and adolescent inpatient units (n = 355) and four residential youth care institutions (n = 162). The sample consisted of 362 girls and 153 boys (missing n = 2), and the mean age was 15.2 years. The duration of data collection was planned for 1 year at each institution, but varied from 12 to 14 months due to logistical challenges caused by the COVID pandemic. While some participants had several institutional stays during the data collection period, only one stay for each youth is included in the study to avoid one person from counting several times in analyses.

2.2 Measures

Data included V-RISK-Y scorings at intake (baseline measure) and registered violent episodes during institutional stay (outcome measure). V-RISK-Y was administered by institutional staff upon admission, and scored interdisciplinary by at least two staff members if possible. Episodes of violence or threats during institutional stay were registered in a separate Violence and Threats (VT) form by staff present during the incident. Violence and/or threats were registered for 59 youth (27 girls and 32 boys). Violence was defined as “attacks against another person with the intent of causing harm” and included verbal and physical threats.

2.3 Ethical approval

Ethical approval for the study was granted by the Regional Committee for Medical and Health Research Ethics (REK ID: 218444) and by the Data Protection Officer at Oslo University Hospital (ID 20/01146). The approval granted exemption from obtaining informed consent from the youth or their guardians to participate in the study. Upon admission, youth and guardians were informed about the project study and given the right to withdraw from the study. There was no further interaction with the study for the youth and/or their guardians.

2.4 Statistical analyses

Power analyses with 5% significance level and 80% power were conducted by a statistician for the multicenter study (31). Analyses were based on data from the V-RISK-Y pilot study (26) and suggested a minimum of 156 participants. All statistical analyses were conducted in SPSS versions 30 and 31.

Secondary analyses were performed on data from the multicenter study. “Moderate” and “Yes” scores were excluded from analyses, so that only “No” and “Don’t know” scores were compared (logistic regression analyses reporting on all scoring options are included in the study assessing predictive validity of V-RISK-Y items in 32). Frequencies of “Don’t know” and “No” scores and corresponding registered episodes of violence and/or threats were extracted from the dataset. Differences in frequency of registered violence were assessed with Chi square or Fisher’s exact test (groups with <5 participants).

Logistic regression analyses were conducted to compare “No” and “Don’t know” scores for each V-RISK-Y item controlled for sex and type of institution (i.e., inpatient psychiatric units and residential youth care institutions).

3 Results

3.1 Descriptives

Number of “Don’t know” scores for V-RISK-Y items ranged from 0 to 12 across ratings. Frequencies of “No” and “Don’t know” scorings and registered violent events for the full sample are displayed in Table 1. The frequency of “No” scores ranged from 15.1% for V10 Future stress to 61.3% for V7 Suspicion. For “Don’t know” scores, the lowest frequency was 15.5% for V7 Suspicion, and the highest was 43.3% for V11 Severe trauma. Differences in rates of registered events of violence and threats between “No” and “Don’t know” scores were significant for V6 Poor insight (χ² = 11.8; p ≤ 0.001), V8 Lack of empathy (χ² = 13.0; p ≤ 0.001), and V9 Unrealistic planning (χ² = 4.37; p = 0.036).

Table 1

Table 1. Frequency of “No” and “Don’t know” scores for individual V-RISK-Y items and registered violent events.

3.2 Logistic regression analyses

Table 2 displays results from logistic regression analyses comparing “No” and “Don’t know” scores, controlling for sex and type of institution. In these analyses, “Don’t know” score for V8 Lack of empathy (OR = 2.1; 95% CI: 1.07, 4.12; p = 0.031) significantly increased risk of registered episodes of violence. In addition to item V8, OR values were above 1 for “Don’t know” scores for item V2 Threats (OR = 1.73; 95% CI: 0.710, 4.19; p = 0.229), V4 Severe symptoms (OR = 1.22; 95% CI: 0.313, 4.75; p = 0.775), V6 Poor insight (OR = 2.07; 95% CI: 0.825, 5.18; p = 0.121), V7 Suspicion (OR = 1.05; 95% CI: 0.449, 2.45; p = 0.913), V9 Unrealistic planning (OR = 1.37; 95% CI: 0.655, 2.87; p = 0.402), V10 Future stress (OR = 1.92; 95% CI: 0.500, 7.34; p = 0.342), V11 Severe trauma (OR = 1.04; 95% CI: 0.385, 2.82; p = 0.936), and V12 Own perception (OR = 1.35; 95% CI: 0.514, 3.54; p = 0.544).

Table 2

Table 2. Logistic regression analyses of “No” and “Don’t know” scores for V-RISK-Y items and registered violent episodes, controlled for sex (ref = girls) and type of institution (ref = residential youth care).

Sex was significant for all items except V11 Severe trauma and V12 Own perception. When type of institution was added to the model, OR was significant for all items except V4 Severe symptoms, V5 Behavior, V7 Suspicion, and V12 Own perception.

4 Discussion

4.1 “Don’t know” scores and registered violence

The proportion of youth registered with violent events during their institutional stay was higher for items scored “Don’t know” as compared to “No” for all items except V1 Violence (see Table 1). The only significant items were V6 Poor insight, V8 Lack of empathy, and V9 Unrealistic planning. Eight of the 12 items had more than 20% “Don’t know” scores, including the three significant items. Particularly high rates of “Don’t know” scores are seen for V11 Severe trauma (43.3%), V9 Unrealistic planning (36.6%), V12 Own perception (31.9%), and V8 Lack of empathy (31.1%). These findings emphasize that several items may be challenging to score conclusively as “No”, “Maybe/Moderate”, or “Yes” at intake, and that this lack of information may be related to increased likelihood of registered violent events. The lack of available information when scoring violence risk assessment instruments in clinical settings has been highlighted in previous literature (4, 5).

In research, attempts are often made to minimize the amount of unavailable data, whereas in clinical practice, this is not possible (11). Missing data increases uncertainty in predictions of violence risk (33). Notably, the three items that reached significance are dynamic factors, not historical. Thus, they require observation to determine and may be prone to subjectivity in scoring. For instance, lack of empathy is an intrapsychic phenomenon, which may be difficult for the rater to score based on the first impression of the youth. Consequently, this item may be more prone to a “Don’t know” score. It is possible these items are more challenging for some staff members to score conclusively and rapidly at intake. Simultaneously, given the statistical significance of the “Don’t know” score, these items might be particularly important to consider when assessing risk level and information to score the items is lacking.

4.2 Effect sizes may have clinical relevance

Findings from item-level logistic regression analyses show that, when controlled for sex and type of institution, 9 of the 12 items had odds ratios (ORs) exceeding 1 (see Table 2). Exceptions were items V1 Violence, V3 Substance abuse, and V5 Behavior. Most items have substantial effect sizes, meaning that the odds of a registered violent event is increased when an item is scored “Don’t know” as compared to “No”. In the item analyses conducted by Laake et al. (31), “Don’t know” scores significantly predicted registered violent events for three items. In the secondary analyses performed in the present study, where only “No” and “Don’t know” scores are compared and sex and type of institution are controlled for, only V8 Lack of empathy remained significant. The high ORs for boys show that the effect of sex explains some of the variance in registered violence. “Don’t know” scores may be of clinical relevance despite not reaching statistical significance (e.g. 34). Importantly, risk assessment items scored with “Don’t know” does not necessarily mean that the item is not present (i.e., a “No” score), but that the information needed to score the item conclusively is lacking at the time of instrument administration (6, 7).

4.3 Considering unavailable information in violence risk screening

Findings indicate that items scored as “Don’t know” could be interpreted as indicators of elevated uncertainty and considered as potentially contributing towards risk. While further research is needed, one potential implication of these findings is that they may inform decision support for clinicians conducting violence risk screenings. The nature of SPJ instruments, where Low–Moderate–High risk level is assessed based on item scores and evaluation of other relevant information, rather than a cutoff score, allows for individual evaluation of the relevance of “Don’t know” scores. When SPJ instruments are administered and data to score items are unavailable, clinicians must rely more heavily on subjective evaluation (35). On the other hand, actuarial tools typically have strong predictive validity, objectivity, and consistency (17). If considered in actuarial tools, “Don’t know” would contribute towards the cutoff score for violence risk. However, actuarial instruments have limited ability to capture dynamic and idiosyncratic factors relevant for risk management strategies (13).

4.4 Handling “Don’t know” scores and missing data in research

In addition to the clinical implications discussed above, findings may impact handling missing scores in statistical analyses. While transparency about missing data in violence risk assessment instruments is encouraged (36), handling of missing data is rarely reported (37). To determine whether weighting of missing items is appropriate, the relationship between missing data and “Don’t know” scores must be assessed further. While there are some differences between a missing score (not scored) and a “Don’t know” score (scored based on lack of information), they both signify that information about an item is unavailable. Commonly, studies sum up the available risk factors for risk assessments and dismiss missing items in analyses (10); however, Eriksen et al. (7) found it more accurate to include the “Don’t know” scores in analyses and weight them. Hopefully, findings from this study can further the understanding of how best to deal with missing data in research on violence risk assessment.

4.5 Limitations and future directions

Several limitations should be considered when interpreting findings from this study. First, because this was a naturalistic study, external variables were not controlled for. Specifically, the data collection overlapped with the pandemic, which may have interfered with generalizability due to COVID measures. Furthermore, it is possible that the implementation of V-RISK-Y led to heightened emphasis on violence risk and risk management, potentially preventing violent events in the participating institutions. Moreover, we do not have a way of accounting for subjectivity and staff dependency in violence risk screening evaluations and in registration of violence, and particularly in registration of threats. The generalizability of findings is further limited to the types of settings assessed. The number of observations in each group varies, and some groups in stratified analyses have small sample sizes. Additional limitations of the multicenter study are discussed in Laake et al. (31).

Despite study limitations, findings are a step towards mapping out the impact of “Don’t know” scores in violence risk assessment, which should fuel further exploration in additional clinical and research settings. In particular, the clinical implications of missing information should be mapped out.

Data availability statement

The data analyzed in this study is subject to the following licenses/restrictions: Restrictions due to confidentiality. The dataset is kept at an encrypted server. Requests to access these datasets should be directed to JR, am9obm9sckBnbWFpbC5jb20=.

Ethics statement

The studies involving humans were approved by Regional committees for medical and health research ethics (ID: 218444). The studies were conducted in accordance with the local legislation and institutional requirements. The ethics committee/institutional review board waived the requirement of written informed consent for participation from the participants or the participants’ legal guardians/next of kin because All adolescents and their guardians were informed about their participation and right to withdraw. They had no further direct contact with the study.

Author contributions

AL: Conceptualization, Formal Analysis, Methodology, Writing – original draft, Writing – review & editing. JR: Conceptualization, Data curation, Formal Analysis, Methodology, Project administration, Supervision, Writing – review & editing. TH: Supervision, Writing – review & editing. SB: Conceptualization, Writing – review & editing. CG: Data curation, Investigation, Project administration, Writing – review & editing. SG: Data curation, Investigation, Project administration, Writing – review & editing. ØL: Conceptualization, Data curation, Investigation, Methodology, Project administration, Supervision, Writing – review & editing.

Funding

The author(s) declared that financial support was not received for this work and/or its publication.

Acknowledgments

We thankfully acknowledge all the staff at participating institutions and youth involved in the study.

Conflict of interest

The authors declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

1. Lockertsen Ø., Procter N, Vatnar SKB, Færden A, Eriksen BMS, Roaldset JO, et al. Screening for risk of violence using service users’ self-perceptions: A prospective study from an acute mental health unit. Int J Ment Health Nurs. (2018) 27:1055–65. doi: 10.1111/inm.12413

PubMed Abstract | Crossref Full Text | Google Scholar

2. Weltens I, Bak M, Verhagen S, Vandenberk E, Domen P, van Amelsvoort T, et al. Aggression on the psychiatric ward: prevalence and risk factors. A systematic review of the literature. PLoS One. (2021) 16:e0258346. doi: 10.1371/journal.pone.0258346

PubMed Abstract | Crossref Full Text | Google Scholar

3. Grünfeld L, Backe-Hansen E, Guldvik MK, Kjelsaas I, Winje E, Engebretsen L, et al. Institusjonstilbudet i barnevernet. Menon Rep. (2020) 54:2020.

Google Scholar

4. Elbogen EB, Mercado C, Tomkins AJ, and Scalora MJ. Clinical practice and violence risk assessment: Availability of MacArthur Risk Factors. In: Farrington DP, Hollin CR, and McMurran M (Eds.). Sex and violence: The psychology of crimes and risk assessment Routledge (2001). p. 38–55.

Google Scholar

5. Tishler CL, Reiss NS, and Dundas J. The assessment and management of the violent patient in critical hospital settings. Gen Hosp Psychiatry. (2013) 35:181–5. doi: 10.1016/j.genhosppsych.2012.10.012

PubMed Abstract | Crossref Full Text | Google Scholar

6. Eriksen BMS, Bjørkly S, Færden A, Friestad C, Hartvig P, and Roaldset JO. Gender differences in the predictive validity of a violence risk screening tool: A prospective study in an acute psychiatric ward. Int J Forensic Ment Health. (2016) 15:186–97. doi: 10.1080/14999013.2016.1170740

Crossref Full Text | Google Scholar

7. Eriksen BMS, Dieset I, Lockertsen Ø., and Roaldset JO. The risk of not knowing–A predictive validity study of the “Don’t know” scores on a violence screen in acute psychiatry. Psychiatry Res Commun. (2022) 2:100076. doi: 10.1016/j.psycom.2022.100076

Crossref Full Text | Google Scholar

8. Etzler S, Eher R, and Rettenberger M. Dynamic risk assessment of sexual offenders: Validity and dimensional structure of the Stable-2007. Assessment. (2020) 27:822–39. doi: 10.1177/1073191118754705

PubMed Abstract | Crossref Full Text | Google Scholar

9. Nunes KL, Firestone P, Bradford JM, Greenberg DM, and Broom I. A comparison of modified versions of the Static-99 and the Sex Offender Risk Appraisal Guide. Sexual Abuse: A J Res Treat. (2002) 14:249–65. doi: 10.1177/107906320201400305

PubMed Abstract | Crossref Full Text | Google Scholar

10. Perley-Robertson B, Babchishin KM, and Helmus LM. The effect of missing item data on the relative predictive accuracy of correctional risk assessment tools. Assessment. (2024) 31:10731911231225191. doi: 10.1177/10731911231225191

PubMed Abstract | Crossref Full Text | Google Scholar

11. Buchanan A, Binder R, Norko M, and Swartz M. Resource document on psychiatric violence risk assessment. Focus. (2015) 13:490–8. doi: 10.1176/appi.focus.130402

Crossref Full Text | Google Scholar

12. Scurich N. An introduction to the assessment of violence risk. In: Singh JP, Bjørkly S, and Fazel S (Eds.). International perspectives on violence risk assessment. Oxford: University Press (2016). p. 3–15.

Google Scholar

13. Monahan J and Skeem JL. The evolution of violence risk assessment. CNS spectrums. (2014) 19:419–24. doi: 10.1017/S1092852914000145

PubMed Abstract | Crossref Full Text | Google Scholar

14. Borum R. Assessing violence risk among youth. J Clin Psychol. (2000) 56:1263–88. doi: 10.1002/1097-4679(200010)56:10<1263::AID-JCLP3>3.0.CO;2-D

Crossref Full Text | Google Scholar

15. Heilbrun K, Yasuhara K, and Shah S. Violence risk assessment tools: Overview and critical analysis. In: Otto RK and Douglas KS (Eds.). Handbook of violence risk assessment. Routledge (2011). p. 11–28.

Google Scholar

16. Bowden J, Logan C, Robinson L, Carey J, McDonald J, McDonald R, et al. Clinicians’ use of the structured professional judgement approach for adult secure psychiatric service admission assessments: A systematic review. PLoS One. (2024) 19:e0308598. doi: 10.1371/journal.pone.0308598

PubMed Abstract | Crossref Full Text | Google Scholar

17. Hart SD and Logan C. Formulation of violence risk using evidence-based assessments: The structured professional judgment approach. In: Sturmy P and McMurran M, (Eds.). Forensic Case formulation. John Wiley & Sons, Ltd. (2011), 81–106.

Google Scholar

18. Ling X, Li H, Li W, Wang S, Zhang Q, and Cai W. A review of progress in violence risk assessment methods. Forensic Sci Res. (2025) 10:owaf014. doi: 10.1093/fsr/owaf014

PubMed Abstract | Crossref Full Text | Google Scholar

19. Singh JP, Grann M, and Fazel S. A comparative study of violence risk assessment tools: A systematic review and metaregression analysis of 68 studies involving 25,980 participants. Clin Psychol Rev. (2011) 31:499–513. doi: 10.1016/j.cpr.2010.11.009

PubMed Abstract | Crossref Full Text | Google Scholar

20. Johnstone L and Gregory L. Youth violence risk assessment: A framework for practice. In: Rogers A, Harvey J, and Law H, editors. Young People in Forensic Mental Health Settings: Psychological Thinking and Practice. UK: Palgrave Macmillan (2015). p. 96–122. doi: 10.1057/9781137359803_5

Crossref Full Text | Google Scholar

21. Pendlebury G, Anderson J, Hales H, Harding D, and Lewis A. Violent behaviour in adolescents: assessment and formulation using a structured risk assessment tool. BJPsych Adv. (2023) 30:147–55. doi: 10.1192/bja.2023.13

Crossref Full Text | Google Scholar

22. Väätäinen L, Björkqvist M, Li Y, Pelto-Piri V, Ferreira A, and Lantta T. Instruments for short-term (24 h) violence risk assessment and strategies for managing violence risk among adolescents with risk for violent behaviour: A systematic review. Int J Ment Health Nurs. (2025) 34:e70110. doi: 10.1111/jpm.12905

PubMed Abstract | Crossref Full Text | Google Scholar

23. Buchanan A, Binder R, Norko M, and Swartz M. Psychiatric violence risk assessment. Am J Psychiatry. (2012) 169:340–0. doi: 10.1176/appi.ajp.2012.169.3.340

PubMed Abstract | Crossref Full Text | Google Scholar

24. Borum R, Lodewijks HP, Bartel PA, and Forth AE. The structured assessment of violence risk in youth (SAVRY). In: Handbook of violence risk assessment. New York, NY: Routledge (2020). p. 438–61.

Google Scholar

25. Hoge RD. The Youth level of service/Case management inventory. In: Handbook of violence risk assessment. New York, NY: Routledge (2020). p. 191–205.

Google Scholar

26. Roaldset JO, Gustavsen CC, Lockertsen Ø., Landheim T, and Bjørkly SK. Validation of a violence risk screening for youth in psychiatric inpatient care-A pilot study of V-RISK-Y. Front Psychiatry. (2023) 14:1210871. doi: 10.3389/fpsyt.2023.1210871

PubMed Abstract | Crossref Full Text | Google Scholar

27. Anderson KK and Jenson CE. Violence risk–assessment screening tools for acute care mental health settings: Literature review. Arch Psychiatr Nurs. (2019) 33:112–9. doi: 10.1016/j.apnu.2018.08.012

PubMed Abstract | Crossref Full Text | Google Scholar

28. Rotter M and Rosenfeld B. Implementing a violence risk screening protocol in a civil psychiatric setting: Preliminary results and clinical policy implications. Community Ment Health J. (2018) 54:245–51. doi: 10.1007/s10597-017-0226-4

PubMed Abstract | Crossref Full Text | Google Scholar

29. Bjørkly S, Hartvig P, Heggen F-A, Brauer H, and Moger T. Development of a brief screen for violence risk (V-RISK-10) in acute and general psychiatry: An introduction with emphasis on findings from a naturalistic test of interrater reliability. Eur Psychiatry. (2009) 24:388–94. doi: 10.1016/j.eurpsy.2009.07.004

PubMed Abstract | Crossref Full Text | Google Scholar

30. Laake ALW, Roaldset JO, Husum TL, Bjørkly SK, Gustavsen CC, and Lockertsen Ø. Interrater reliability of the violence risk assessment checklist for youth: a case vignette study. BMC Psychiatry. (2024) 24:303. doi: 10.1186/s12888-024-05746-8

PubMed Abstract | Crossref Full Text | Google Scholar

31. Laake ALW, Roaldset JO, Lossius Husum T, Bjørkly SK, Chudiakow Gustavsen C, Grenabo ST, et al. Predictive accuracy of the Violence Risk Assessment Checklist for Youth in acute institutions – A prospective naturalistic multicenter study. Eur Psychiatry. (2025) 68:1–33. doi: 10.1192/j.eurpsy.2025.3

PubMed Abstract | Crossref Full Text | Google Scholar

32. Laake ALW, Roaldset JO, Husum TL, Bjørkly SK, Gustavsen CC, Grenabo ST, et al. Assessing the predictive validity of the 12 V-RISK-Y items: A prospective naturalistic study in acute institutions for youth. Psychiatry Res Commun. (2025) 5:100224. doi: 10.1016/j.psycom.2025.100224

Crossref Full Text | Google Scholar

33. Connors MH and Large MM. Calibrating violence risk assessments for uncertainty. Gen Psychiatry. (2023) 36. doi: 10.1136/gpsych-2022-100921

PubMed Abstract | Crossref Full Text | Google Scholar

34. Jacobson NS and Truax P. Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. In Kazdin AE (Ed.), Methodological issues & strategies in clinical research. American Psychological Association. (1992). pp. 631–648.

Google Scholar

35. Garrington C and Boer DP. Structured professional judgement in violence risk assessment. In: Wormith JS, Craig LA, and Hogue TE (Eds.). The Wiley Handbook of What Works in Violence Risk Management: Theory, Research and Practice. Wiley Blackwell. (2020). p. 145–62.

Google Scholar

36. Singh JP, Yang S, and Mulvey EP. Reporting guidance for violence risk assessment predictive validity studies: the RAGEE Statement. Law Hum Behav. (2015) 39:15–22. doi: 10.1037/lhb0000090

PubMed Abstract | Crossref Full Text | Google Scholar

37. Logan C. Reporting structured professional judgement. In: The forensic psychologist’s report writing guide. New York, NY: Routledge (2017). p. 66–77.

Google Scholar