Predictive validity of SVR-20, RSVP, HCR-20, PCL-R, and clinician summary risk ratings among adult males convicted of sexual offending: nationwide prospective field study

Structured professional judgement (SPJ) checklists are widely used to assess reoffending risk in individuals convicted of sexual offending (ICSOs), but a few prospective studies have evaluated real-world predictive accuracy. We present a Danish prospective field study of clinician SPJ checklist ratings (SVR-20, RSVP and HCR-20), the PCL-R and clinician summary risk ratings (CSRRs) among 96 male ICSOs assessed 1997–2001. The findings suggested moderate-to-high SVR-20 and RSVP predictive validity for sexual reoffending at 2- and 5-year follow-ups (AUCs = 0.69–0.72), but declining accuracy for 10-year predictions (AUCs = 0.63–0.64). The HCR-20 and PCL-R predicted sexual reoffending less accurately (p < 0.05) than the SVR-20. All measures had moderate-to-high predictive accuracy for any, severe and recurrent violent reoffending (AUCs = 0.67–0.85). Notably, CSRRs performed well (AUCs = 0.67–0.80) across outcomes and periods. Despite overestimation of absolute sexual reoffending, SVR-20 and RSVP were supported for ICSO reoffending risk within 5-year follow-ups.