A Longstanding Clinical Oversight? Protective Factors May be Central to Accurate Violence Risk Assessment

Clare Neil, NHS Forth Valley
Suzanne O’Rourke,  University of Edinburgh and The State Hospital, Edinburgh
Nuno Ferreira, University of Nicosia
Liz Flynn, NHS Lothian and Forensic Mental Health Services Managed Care Network

Research and practice in violence risk assessment in forensic mental health primarily focuses on risk factors; however consideration of protective factors may improve the accuracy and utility of assessments. Using a pseudo-prospective design, the predictive and incremental validity of protective factors was explored using the Structured Assessment of Protective Factors (SAPROF) and Historical Clinical Risk Management-20 (HCR-20V3) in 75 male inpatients in a secure setting. Over a 12-month period, protective factors significantly predicted the absence of inpatient (institutional) violence and risk factors, particularly dynamic factors, predicted the presence of violence. Hierarchical logistic regression did not establish the incremental validity of the SAPROF. Preliminary evidence for the predictive and incremental validity of the Integrative Final Risk Judgment was found with individuals judged high risk being almost seven times more likely to engage in violence than those assessed as moderate risk. High risk ratings were associated with fewer protective factors and more risk factors. Therefore, whilst dynamic risk factors are clear targets for risk management, consideration of protective factors may contribute to overall estimates of risk and provide additional targets for intervention.

Violence risk; assessment; protective factors; SAPROF; HCR-20

“The accurate assessment and management of violence risk is a core task in forensic mental health settings…Institutional violence within patient settings has personal consequences and considerable organizational costs…There is therefore a need for defensible practice and decision making to effectively prevent future violent behavior…A recent study by Singh et al. (2014) indicated some of the most widely used and evidence based tools in violence risk assessment practice ere based on the structured professional judgment (SPJ) approach…The SPJ approach involves identifying the presence and relevance of risk factors in the individual case, integrating these into a risk formulation and scenarios, and using these to inform risk management…An overall level of risk…is given using the assessor’s professional judgment…Although a number of SPJ risk assessment tools exist, most focus on risk factors…rather than protective factors…” (p.84).

“One risk assessment tool designed to specifically assess protective factors for violence risk is the Structured Assessment of Protective Factors (SAPROF – 2nd version…)…the SAPROF is an SPJ tool developed for use alongside other tools which assess risk factors for violence…The protective factors are primarily dynamic in nature and conceived as being associated with an absence of violence…The present study sought to explore the predictive and incremental validity of the SAPROF for inpatient violence in a secure forensic mental health setting. It was hypothesized that the SAPROF would predict the absence of violence behavior and the SAPROF and SPJ risk estimate of overall level of risk would have incremental validity over the assessment of risk factors. Risk factors were assessed using the most recent version of the Historical, Clinical, and Risk Management-20 (HCR-20V3; Douglas et al., 2013); a secondary aim of the study was therefore to explore the predictive validity of the HCR-20V3 and to consider the utility of the SAPROF when combined with the HCR-20V3“ (p.85-86).

“The State Hospital provides a high secure forensic mental health service for Scotland and Northern Ireland…A total of 129 male patients were detained in the hospital at the beginning of the data collection period (April 2014)…Seventy five patients (58.1%) met the inclusion criteria…The results supported the hypothesis that the presence of protective factors predicts the absence of inpatient violence. SPJ risk estimates of the overall protection offered by the protective factors (the FPJ [Final Protection Judgment]) were also significantly associated with the absence of all types of violence and disruptive behavior” (p.86-97).

“…The External scale of the SAPROF was a poor predictor of absence of most types of violence; this is likely to be due to the limited variability in scores within this domain as three of the items are rated the same for all patients due to the nature of the secure environment. The predictive validity of the total and domain scores in the HCR-20V3 varied in the current study…some preliminary observations are noted…it was apparent from analysis of the HCR-20V3 domains that the Historical scale was a particularly poor predictor whilst the dynamic Clinical and Risk Management scales fared better…” (p.97).

“The current study hypothesized that the SAPROF would have incremental validity in relation to predictive accuracy when added to the HCR-20V3…the SAPROF did not significantly add to the predictive validity of the HCR-20V3 for violence and disruptive behavior within the hospital setting. This may reflect a degree of overlap in the content of some of the items within the HCR-20V3 and SAPROF…[however], consideration of protective factors may have other benefits in terms of informing treatment by highlighting targets of intervention or facilitating engagement in those being assessed…” (p.98).

“The present study found moderate to high correlations between HCR-20V3 dynamic risk factors and SAPROF protective factors. This, coupled with the SAPROF’s lack of incremental validity, suggests that the HCR-20V3 may already be capturing some of the protective factors in the SAPROF…The HCR-20V3 total-SAPROF total variable significantly predicted all types of violence except sexual violence, however, confidence intervals were large and this variable accounted for little variance in violence outcome. This would suggest that the relationship between risk and protective factors and underpinning mechanisms between protective factors and violence is complex and not adequately captured by simply subtracting the numerical scores of protective factors from those risk factors” (p.98).

“There is clear interest in protective factors and the SAPROF has been implemented in practice despite the relatively limited empirical evidence base. The present study therefore adds to the existing research on the SAPROF as well as highlighting areas for further research. In particular, further research is required to establish whether additional structured assessment of protective factors is necessary or whether existing risk factors can be reconceptualized to also capture protective factors…” (p.99).

“…Our results provide evidence to support that a structured professional judgment approach, in the form of the SAPROF’s Integrative Final Risk Judgment, results in greater predictive accuracy in the assessment of inpatient violence risk…Given the apparent inter-dependence of many existing risk and protective factors, it is suggested that consideration could be given to considering the role of protective factors when using existing tools focusing on risk factors. This would ensure violence risk assessments and overall risk judgments are comprehensive, predictive validity is maximized, a range of targets for intervention can be identified, and a more inspiring treatment framework and management plan for patients could potentially be developed” (p.100).

“None of the SAPROF subscales or total score significantly predicted the absence of sexual violence. This may have been due to the relatively low base rate for sexual violence within the sample and broad definition used which included relatively minor forms of inappropriate sexualized behavior. It is also possible that sexual aggression is qualitatively different and may require consideration of different protective factors that are not included in the SAPROF. A SAPROF manual focusing specifically on protective factors associated with sexual violence is currently in development” (p.97).

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