Risk of transitioning from Out-Of-Home Care into Youth Justice: Can formal risk prediction inform prevention opportunities?

Date

2025

Authors

Malvaso, C.
Montgomerie, A.
Santiago, P.E.R.
Lynch, J.
Pilkington, R.

Editors

Advisors

Journal Title

Journal ISSN

Volume Title

Type:

Journal article

Citation

Journal of Criminology, 2025; 58(4):662-688

Statement of Responsibility

Catia Malvaso, Alicia Montgomerie, Pedro Enrique Ribiero Santiago, John Lynch, Rhiannon Pilkington

Conference Name

Abstract

This article investigates risk of Youth Justice (YJ) supervision by age 18, among children who had experienced Out-Of-Home Care (OOHC) before age 10. Data were drawn from the Better Evidence Better Outcomes Linked Data platform. Children born 1991–1998 who experienced at least one placement in OOHC before age 10 (N = 2,832) were followed to age 18 for YJ supervision. Logistic regression models including child and maternal sociodemographic and perinatal characteristics and child protection characteristics were used to predict the probability of (a) any YJ supervision; (b) any custodial YJ supervision by age 18. Of children in OOHC before age 10, 13.5% (n = 381) experienced any YJ supervision by age 18, and 10.6% (n = 300) experienced custodial YJ supervision. Using all 42 predictors, model discrimination (Area Under the Receiver Operator Characteristic Curve (AUROC) and Area Under the Precision Recall Curve (AUPRC)) was ∼0.8 and ∼0.4, respectively, for both outcomes. We used the top 30% of the predicted probabilities to create a ″high″ risk threshold. At this high-risk threshold, sensitivity was 69.8% and 75.3%, respectively; specificity was 76.5% and 75.5%; and the positive predictive value was 32.3% and 26.8%. These risk prediction models have reasonable discrimination to identify children in OOHC who are at higher risk of transitioning into YJ, and are technically feasible. However, predicting risk implies providing opportunities for early supports that may prevent transitions from OOHC to YJ. There are ethical and practical considerations to using prediction models in this population.

School/Discipline

Dissertation Note

Provenance

Description

First published online October 21, 2025

Access Status

Rights

© The Author(s) 2025. Article reuse guidelines: sagepub.com/journals-permissions.

License

Call number

Persistent link to this record