Posted on

Mental Illness not a Reliable Predictor of Sex Offender Recidivism

Forensic-Training-AcademyPresence of mental illness is less related to risk and needs but is an important factor for treatment responsivity among sexual offenders. This is the bottom line of a recently published article in the International Journal of Forensic Mental Health. Below is a summary of the research and findings as well as a translation of this research into practice.


Featured Article | International Journal of Forensic Mental Heath | 2015, Vol. 14, No. 1, 10-22. ijfmh

The Relationship between Mental Disorder and Recidivism in Sexual Offenders


Drew A. Kingston, Integrated Forensic Program, Royal Ottawa Health Care Group, Brockville, Ontario, Canada, Institute of Mental Health Research, University of Ottawa, Ottawa, Canada
Mark E. Olver, Department of Psychology, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
Melissa Harris, Integrated Forensic Program, Royal Ottawa Health Care Group, Brockville, Ontario, Canada
Stephen C.P. Wong, School of Medicine, University of Nottingham, Nottingham, United Kingdom, Department of Psychology, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
John M. Bradford, Integrated Forensic Program, Royal Ottawa Health Care Group, Brockville, Ontario, Canada, Institute of Mental Health Research, University of Ottawa, Ottawa, Canada


The importance of mental illness as a risk factor for violence has been debated with significant implications for mental health policy and clinical practice. In offender samples, mental health diagnoses tend to be unrelated to recidivism, although this effect has been questioned recently in sexual offenders. In the present, prospective investigation, the relevance of several mental health diagnoses and relevant co-morbidity is examined as predictors of various types of recidivism in two distinct samples of sexual offenders who were followed up to 27 years in the community. Results indicated that mental health diagnoses were not predictive of recidivism on their own or in multivariate categories, although comorbid substance-use disorders and some personality disorders showed some predictive validity. Results are discussed in the context of a social learning model of crime and in terms of the treatment of sexual offenders.

Keywords: sex offender, mental health, diagnoses, recidivism

Summary of the Research

“The vast majority of studies using offender samples have shown psychiatric diagnoses to be unrelated to recidivism and that the predictors of recidivism are largely shared between mentally disordered offenders and non-disordered offenders. According to the General Personality and Cognitive Social Learning Model (GPCSL), there are eight robust predictors of criminal behavior that reside within the individual or their immediate social learning environment: criminal history, procriminal companions, procriminal attitudes, antisocial personality pattern, education/employment, family/marital, substance abuse, and leisure/recreation. Mental health variables were not considered significant predictors of criminal behavior. Consistent with the research on general offenders, a meta-analysis reported that severe psychological dysfunction (psychosis) was not significantly related to sexual recidivism. Since this meta-analysis, there have been relatively few studies published examining the relationship between serious mental illness and recidivism in sexual offenders” (p. 11)

“In the present study, we examined the relationship between mental disorder and recidivism in a sample of treated Canadian correctional sample of sexual offenders, and replicated our findings in a sample of sexual offenders in an outpatient mental health hospital. We hypothesized that recidivism would be best predicted by factors identified by the GPCSL and previous meta-analyses. Additionally, we predicted that mental health diagnoses would not be associated with recidivism and would fail to show any incremental validity after controlling for needs that are consistent with the GPCSL” (p. 12).

“The vast majority of the sample (95% in study 1, 97% in study 2) had a mental disorder and a significant proportion of offenders had a [substance use disorder] and a [personality disorder], particularly [Antisocial Personality Disorder (ASPD)]. Importantly, these base rates are not representative of general offender populations, but rather reflect a select, broadly high risk sample of sex offenders attending treatment at a corrections-based mental health facility” (p. 15).

“Consistent with extant findings, non-substance related mental disorders were not predictive of recidivism outcomes either on their own or in multivariate categories. The Violence Risk Scale-Sexual Offender version dynamic score significantly predicted sexual, violent, and general recidivism. The only diagnostic categories that were associated with risk and recidivism were personality disorders (including ASPD), substance use disorders (SUD), and dual diagnoses involving substance use disorder comorbidity. Even in these instances, most diagnostic categories were marginal predictors of sexual violence, although they tended to be more robust predictors of general violence and any criminal recidivism.” (p. 15)

Translating Research into Practice

“A number of studies have shown that severe mental illness, particularly schizophrenia and other psychotic disorders are associated with crime. Such studies have been used to support specific policy recommendations and the initiation of diversion programs (e.g., mental health courts) that are intended to treat mental illness so that the likelihood of recidivism is reduced. Although diversion programs have been successful in reducing the rates of incarceration and increasing general access to mental health services, the ability of these programs to reduce recidivism has been mixed, at best. The evidence for mental health courts is particularly weak for those programs weighted more heavily toward mental health models, as opposed to criminal justice–based models” (p. 19)

“Addressing these dynamic risk factors or criminogenic needs is an important component of treatment programs that adhere to the principles of effective correctional intervention (i.e., risk, need, responsivity). A number of programs based on the principles of risk, need, and responsivity have been developed and applied to general offenders and sexual offenders with some degree of empirical support. It is our sense that some diagnoses contain or embody dynamic risk factors, which may be why they contribute to the prediction of recidivism outcomes. Given comprehensive measure of sexual violence risk that assesses static and dynamic domains, diagnostic information may be immaterial from a risk and need standpoint, although may be an important responsivity consideration” (p. 19)

“Despite the fact that mental illness is not a reliable predictor of recidivism, such diagnoses may be best conceptualized as a responsivity factor as noted above, such that some individuals with active mental illness may find it difficult to engage in treatment or attend to the treatment content. Therefore, targeting such symptoms may be an important first step in treatment process. In turn, addressing mental health symptoms may help to promote improvement on an individual’s identified criminogenic needs. For example, managing mental health symptoms may allow one to make more adequate use of leisure time, resisting urges to turn to substance use to manage symptoms, and to obtain employment, all of which have been shown to reduce the likelihood of recidivism” (p. 20).

Other Interesting Tidbits for Researchers and Clinicians

In study 2, SUDs, “demonstrated a unique association with outcome, particularly for general violence and any kind of reoffending upon release. It is worth noting that the sample had a fairly low base rate of NSMD, but a high base rate of personality disorder and SUD. Interestingly, SUD did not covary with VRS:SO posttreatment dynamic score. That is, individuals with or without an SUD bore little relation to their actuarial level of sexual violence risk; however, there was a clear differentiation between the two groups in terms of nonsexual outcomes, and this was additive beyond the VRS:SO. In this context, SUD may have served as a proxy for a general risk variable, given that the VRS:SO contains diverse item content, of which one domain includes general criminality” (p.15).

“After controlling for Static-99 score, few of these diagnoses (SUDs, PDs) predicted any outcome, save for a few interesting exceptions. This reflects the high degree of shared variance between some of these categories with a risk assessment tool such as the Static-99, explaining the higher scores observed in most groups that were positive for a given diagnosis (except for SGIDs and NSMDs). Thus, although some of these diagnostic entities have risk variance, there is little beyond that already captured by the Static-99 and they tend to offer little incrementally, at least from a pure prediction perspective. One interesting exception was Sexual Sadism, which provided additional predictive information beyond the Static-99, which likely captures in part the prominent risk-need domain of sexual deviance from the broader extant literature as a salient marker of risk for sexual violence” (p. 19).

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!


Authored By: Megan Banford

Megan is a graduate student in the Forensic Psychology program at John Jay College. She graduated in 2013 from Simon Fraser University with a B.A. (Honors) and hopes to attain her PhD in clinical forensic psychology. Her main research interests include violence risk assessment and management, juvenile offenders and public policy.


2 thoughts on “Mental Illness not a Reliable Predictor of Sex Offender Recidivism

  1. Apreciada Dr. Banford:
    Cordial saludo. Soy un académico dedicado a la Psicología Jurídica en Bogotá, Colombia. Agradecería que usted el envío de su artículo. Gracias

Leave a Comment :)

This site uses Akismet to reduce spam. Learn how your comment data is processed.