Risk assessment in adult psychiatric and criminal justice populations is traditionally one-sided. Protective factors are neglected, and assessment focuses solely violence risk. However, the contribution of instruments that support assessment of protective factors for a full range of relevant outcomes is not well-established.

1) to identify all relevant studies investigating the predictive efficacy of selected tools (START, SAPROF, DUNDRUM 3, DUNDRUM 4, IORNS) for the assessment of protective factors for a range of adverse outcomes, 2) to synthesise available information through meta-analytic procedures

to determine if protective factors are effective in predicting adverse outcomes

A systematic search of five electronic databases for records up to June 2014. A meta-analysis was conducted using the MEANES macro for SPSS (Lipsey & Wilson, 2001; Wilson, 2012).

Seventeen studies (N=2,198) were included in the meta-analysis. Where multiple studies contributed, the largest and smallest mean weighted effect sizes were for violent reoffending and inpatient victimisation respectively. There were no significant differences between effect sizes of protective and risk scales; of the protective scales, the SAPROF dynamic items and total score performed best. Summary judgments based on consideration of each tool's protective and risk scales and other case-specific factors predicted their corresponding outcomes with only two exceptions.

Protective factors perform equivalently to risk factors for prediction of a range of adverse outcomes. Future research should aim to establish if they generate useful indicators about treatment targets, and if their use is beneficial in improving therapeutic relationships.