Department of Communities and Justice

Risk assessment tools for recidivism have been validated extensively in youth justice populations. These validation studies typically demonstrate good predictive accuracy, leading many jurisdictions to implement structured approaches across custody and community supervision settings. However, previous research has focused primarily on single supervision contexts, with limited examination of whether predictions maintain their accuracy when applied across different settings. This gap has implications for jurisdictions applying similar risk assessment protocols across different supervision settings. Our study examined whether violent recidivism predictions maintain their accuracy across youth custodial and community supervision contexts. We developed a predictive model using data from custodial youth (n = 790) and evaluated its performance among community-supervised youth (n = 766) in New South Wales, Australia. Two-year violent recidivism was higher in custody (37%) than in community supervision (27%). The model showed good discrimination in both custody (AUC 0.771, 95% CI: 0.735–0.807) and community settings (AUC 0.728, 95% CI: 0.685–0.770). While adjusting the baseline hazard improved absolute risk prediction, optimal performance across the risk spectrum required additional adjustments to the linear predictor. This suggests that the strength and pattern of risk factor relationships may vary across different settings; therefore, statistical adjustments may be necessary when applying predictions across various contexts.

The evidence linking human health and housing is overwhelming. However, less focus has been on collaborative action between the sectors to improve health. Focussing on social housing adds an important equity lens to the housing and health partnership literature. Since 2009, a unique formal partnership between a State Health Service, State Social Housing Organisation and a research organisation has existed. The partnership aims to realise potential benefits of integrating health considerations into asset and social planning within the housing sector and to empower residents to lead on outcomes they identify for the community and themselves individually. The partnership has a shared vision of working together to improve the health and wellbeing of social housing communities in one of Australia’s most disadvantaged regions. This paper presents an analysis of over a decade of the partnership work. We present the history of the partnership, overview outcomes to date, and describe key mechanisms and contextual factors that have enabled action. We present our findings using two frameworks, one a practice focussed partnership toolkit and the other a theory informed approach to health focussed governance. Both frameworks enable us to present findings linking the practical successes and challenges of this partnership work with deeper insights from theory.

This study aims to determine whether previous exposure to a total prison smoking ban is associated with a reduction in daily smoking postrelease and to identify the factors associated with daily tobacco use. Demographics, incarceration history, and substance use data were collected from the 2016 National Prisoner Entrants' Blood Borne Virus Survey for 389 Australian prison entrants. Predictors of daily smoking among those exposed to a smoking ban were illicit drug use and Indigenous status. No significant association was found between current daily smoking and recent exposure to a prison smoking ban, suggesting that the majority of Australian prisoners exposed to a tobacco ban while incarcerated will return to smoking once released. Complementary and culturally specific interventions supporting abstinence postrelease should be considered.