Page 42 - Contributed Paper Session (CPS) - Volume 7

CPS2027 Olayan A. et al.
assessment tool and is frequently used in the Youth Justice System in the
United Kingdom. It is comprised of three broad assessments: (1) physical and
mental health, (2) substance abuse, and (3) neurodisability. These assessments
were designed to be completed within 12, 21 and 30 working days,
respectively. A systematic review[10] evaluated the predictive ability of more
than 70 risk tools to assess the risk of criminal behaviour among approximately
25,000 people;[10] however, concerns were raised regarding the predictive
accuracy of these tools as well as their practical use in real life. Due to their
multicomponency and complex nature, these tools require multidisciplinary
evaluations and may take several hours to be administered; therefore, they
cannot be used universally (i.e. for everyone at clinical or criminal justice
settings) to identify those at high risk of offending. In addition, these tools
have been developed for certain populations such as psychiatric patients and
those with poor cognitive and antisocial behaviours; therefore, their
generalisability was also reported to be questionable.[10]
The primary objective of this study is to develop and validate risk
prediction models to quantify an individuals’ risks of reoffending using a data-
linkage study for individuals who were diagnosed with a severe mental illness
(psychoses) prior to the first offence in NSW, Australia.

2. Methodology
The study design and population has been described in detail
elsewhere.[11] Data linkage was used to identify individuals who were
diagnosed of psychosis from the NSW Ministry of Heath’s Admitted Patient
Data Collection (APDC) and the Emergency Department Data Collection
(EDDC) based on International Classification of Diseases (ICD) nine and ten,
and also who were committed the first offence Research’s Re-offending
Database (ROD) between 2001 and 2012.
7,743 men (%) and women (%) from a retrospective linkage study were
included in this study. Individuals were randomly assigned to either a
development (67%) or an internal validation dataset (33%). The outcome of
interest in this study was an incident of reoffending. All individuals were
followed from their first offence date until their reoffending, death or the 31st
of December 2015, whichever occurred first. We developed the risk prediction
model in two stages: first, we used a split-sample method in order to develop
a risk equation using a weighted-scoring system. The study population was
randomly allocated to either the development (67%) or internal validation
(33%) sample dataset. We used a range of variables as potential predictors of
the reoffending. These included, gender, country of birth, marital status,
Indigenous status, Socio-Economic Indexes for Areas (SEIFA), psychosis type,
age at the first offence, the outcome of the first offence, the type of the first
offence and the status of contact with mental health services. Cox regression

30 | I S I W S C 2 0 1 9