Exploring heterogeneity among mentally disordered offenders – The key to targeting Interventions

Although the statistical association between mental illness, particularly severe mental illness, and risk of violence and offending is now well established (Fazel et al., 2009), much of the accumulated evidence has ignored the likely heterogeneity within the population of mentally disordered offenders. Subgroups defined on the basis of pathways into violence and other offending behaviours have long been of interest to criminologists and others outside the forensic mental health field, but only recently has such an approach been applied systematically to understanding causal mechanisms emerging over the life-course for those with both a mental illness and a tendency towards offending behaviour.

Simpson et al. (2015) have tested a method of establishing subgroups within a forensic psychiatry patient sample based on the relative timing of illness and offending onset, an approach with considerable theoretical merit. While the subtyping of offenders in general population samples has tended to focus on the age or developmental stage of offending onset, combined with examination of the persistence of offending over time (e.g. to produce the life-course persistent and adolescent-limited offender groups [Moffitt, 1993]), whether offending onset predates or postdates onset of mental illness is arguably a better approach to a mentally disordered offenders typology, although with an acknowledgement of the likely overlap given that the peak period for onset of severe mental illnesses is not very different to that for onset of offending. As others have demonstrated, Simpson et al. find support for subgrouping into early (pre-illness-onset) and late (post-illness-onset) start offenders in terms of differences in socio-demographic, clinical and offending-related characteristics. Criminogenic, antisocial personality and substance use–related factors appeared more important for the early compared to late starters, for example.

Beyond evidence from clinical samples of mentally disordered offenders, differing patterns of association between offending and mental illness have also been found when offending occurs pre-illness-onset compared to post-illness-onset, at a population level. The strength of associations between individual diagnoses across the full spectrum of mental ill health and offending risk was near uniform in a Danish pre-illness-onset study, implicating general rather than diagnosis-specific causal mechanisms (Stevens et al., 2012). However, when post-illness-onset offending was examined in the same setting, the strength of associations varied considerably, particularly for women and particularly in relation to violent offending (Stevens et al., 2015).

The notion that illness-related factors, such as the experience of active psychotic symptoms and the affective consequences of such symptoms, are more important in driving offending in those whose offending occurs after the onset of mental illness appears to be influenced by gender. This is consistent with the striking gender gap narrowing seen in mentally disordered when compared to general offender samples – whatever it is that ordinarily protects women in the general population from acting in a violent or otherwise antisocial manner appears to be eroded by the impact of severe mental illness. In the Danish post-illness-onset study, over one quarter of the population-level risk of violent offending by women was found to be potentially attributable to prior mental illness (Stevens et al., 2015); for men, the figure was around 10%.

Establishing such typologies within the population of mentally disordered offenders offers the promise of both improving aetiological understanding of the association between illness and offending and informing the development of targeted interventions to reduce risk of violence – avoiding a ‘one size fits all’ approach. Simpson et al. argue, for example, that reducing risk of future violence for early starters will not be achieved by treating their active symptoms of mental illness alone, given the importance of additional criminogenic needs and personality pathology. The limited evidence currently available to inform offending reduction strategies from a clinical perspective appears to indicate that need-specific rather than universal interventions are likely to hold the most promise. In an extension of a randomised controlled trial (RCT) of early intervention for first episode psychosis, for example, no effect on future offending was found for the intervention despite benefits seen in the primary and secondary clinical, social and other targets (Stevens et al., 2013). Good clinical practice aimed at effectively treating mental illness will not necessarily translate into improvements in violence or offending risk, at least not for those whose typology indicates that other factors may be the main drivers for such behaviour.

See Research by Simpson et al., 49(11): 1048–1059.