Violent recidivism among sexual offenders : risk factors and assessment procedures

Abstract: Background: Assessment and management of risk for violent recidivism to prevent future offending among sexual offenders is an important and challenging task within social, correctional, and forensic mental health services. The overall aim of this thesis was to study the usefulness of various models for the assessment of risk for violent recidivism among sexual offenders in Sweden, and to explore the relationship between individual risk factors and violent recidivism. Method: Two different study populations were included in a register-based retrospective follow-up study; 1) all persons subjected to forensic psychiatric evaluation 1988-1990 due to a criminal charge of rape and diagnosed with personality disorder (N = 51) and 2) a cohort of all adult men convicted of a sexual offence and released from Swedish prisons 1993-1997 (N = 1,400). Results: Among five studied assessment procedures, only the two actuarial assessment schemes RRASOR and Static-99 were associated with sexual reconvictions (Paper I-II). When the RRASOR and Static-99 accuracy was explored using alternative outcome measures including imminence, frequency, nature, and severity of sexual reoffending, both checklists showed variability concerning predictive validity with AUC-estimates ranging from "no better than chance" (.40) to "high accuracy" (.94; Paper 111). Prior sentencing dates, nonsexual violence, and any stranger victim were the most prominent individual risk factors for any violent (including sexual) recidivism (Paper 11). Also, psychiatric morbidity during inpatient treatment was related to both sexual and violent (nonsexual) reoffending, with personality disorder exhibiting the strongest association (Paper V). It was more common that the studied sexual offenders were reconvicted for a violent nonsexual than a sexual offence. However, sexual offending characteristics, especially concerning victim preferences, were stable over time among individuals reconvicted for a sexual offence (Paper IV). Conclusions: The purpose of the risk assessment and the decision context should be taken into account when evaluating the relative merits of various risk assessment procedures. The Static-99 could be useful for screening purposes but should not be used as a stand-alone assessment procedure without a subsequent extended evaluation. Also, it seems important that assessments of risk for future violent offending among sexual offenders consider the situational community context into which the offender is released, specifically concerning access to potential victims. Future research should continue to explore dynamic risk factors, stability of sexual offending over time, and elucidate the relationship between psychiatric morbidity and violent recidivism among sexual offenders.