Predictive validity of the START with intellectually disabled offenders
| DOI | https://doi.org/10.1108/JFP-12-2012-0029 |
| Published date | 04 February 2014 |
| Date | 04 February 2014 |
| Pages | 78-88 |
| Author | Andy Inett,Grace Wright,Louise Roberts,Anne Sheeran |
| Subject Matter | Health & social care,Criminology & forensic psychology,Forensic practice |
Predictive validity of the START with
intellectually disabled offenders
Andy Inett, Grace Wright, Louise Roberts and Anne Sheeran
Andy Inett is a Forensic
Psychologist, based at
Tarentfort Centre, Learning,
Dartford, UK.
Grace Wright is a Forensic
Psychologist in Training and
Dr Anne Sheeran is a
Consultant Clinical and
Forensic Psychologist, both
are based at Tarentfort Centre,
Dartford, UK.
Louise Roberts is a Trainee
Clinical Psychologist, based at
Research Department of
Clinical, Educational and
Health Psychology, University
College London, London, UK.
Abstract
Purpose – Offenders with intellectual disability (ID) have been largely neglected in past forensic literature on
assessment of dynamic risk factors. The purpose of this paper is to evaluate the predictive validity of the
Short-Term Assessment of Risk and Treatability (START), in a sample of males with IDs in a low-secure
hospital (n ¼28).
Design/methodology/approach – A prospective analysis was conducted, with START scores as the
predictor variables, and the number of recorded aversive incidents as the outcome measure.
Findings – Receiver operating characteristic analysis demonstrated that total START risk scores had
a significant high predictive accuracy for incidents of physical aggression to others (area under the curve
(AUC)¼0.710, p o0.001) and property damage/theft (AUC¼0.730, po0.001), over a 30-day period,
reducing to medium predictive validity over a 90-day period. Medium predictive validity was also identified
for incidents of verbal aggression, suicide, self-harm, and stalking and intimidation. STARTstrength scores
were also predictive of overt aggression (AUC ¼0.716), possible reasons for this are explored.
Research limitations/implications – The small sample size limits the generalisability of the findings, and
further research is required.
Practical implications – The paper offers preliminary support for the use of the STARTwith ID offenders
in low-secure settings. Given the lack of validation of any previous dynamic risk assessment tools, multi-
disciplinary teams in such settings now have the option to use a tool which has potentially good validity with
an ID population.
Originality/value – This study represents the first attempt to examine the predictive validity of the
START with ID offenders, and a step forward in the understanding of dynamic risk factors for violence in
this population. The significant predictive relationship with incidents of physical aggression and property
damage offers clinicians a preliminary evidence base supporting its use in low-secure settings.
Keywords Intellectual disability, Risk, Offenders, Protective factors
Paper type Research paper
Introduction
Risk management should be based on a plan to reduce the risk of harm occurring and increase
the potential for a positive outcome. (Department of Health, 2007, p. 14).
Professionalsworking in forensic intellectualdisability (ID) settings frequentlyface the paradoxical
challenge of developing the strengths of service-users whilst simultaneously managing the risks
that they pose to themselves and others. Whilst risk assessment and management within
mainstream offender and psychiatric populations has received much research attention in
recent years, offenders with ID represent a subgroup of mentally disordered offenders that
some have suggested are largely ignored in the literature (Barron et al., 2004). The use of
Structured Professional Judgement (SPJs) tools is widely considered to represent best
practice whenassessing risk of violence acrossa variety of settings. Department of Health (2007)
Guidelines Best Practice in Managing Risk advocate the use of SPJs, such as the HCR-20
(Webster et al., 1997). The HCR-20 has consistently shown to be a good predictor of future
violence across a range of offender populations. Yang et al. (2010) published a meta analysis
PAGE 78
j
JOURNAL OF FORENSIC PRACTICE
j
VOL. 16 NO. 1 2014, pp. 78-88, CEmerald Group Publishing Limited, ISSN 2050-8794 DOI 10.1108/JFP-12-2012-0029
To continue reading
Request your trial