Taking the DUNDRUM toolkit to the real world: field validity after implementation in Belgium

Purpose The DUNDRUM toolkit structures and enhances transparency and consistency in decisions about admissions and transfers of forensic psychiatric patients. Although its reliability and validity are supported, most research has been conducted in Ireland by the tool’s authors and focused on actuarial scoring. This study aims to examine DUNDRUM-3 and DUNDRUM-4 in Belgium to evaluate its use in routine clinical decision-making within a structured professional judgement (SPJ) framework. Design/methodology/approach Clinician-rated DUNDRUM-3 and DUNDRUM-4 assessments were collected across different security levels to examine psychometric properties and associations with subsequent transfers. Findings Both scales demonstrated good internal consistency (Cronbach’s alpha 0.82 and 0.83) and discriminated between higher and lower levels of security at discharge using both numerical scores (AUC = 0.62–0.79) and SPJ-informed security categories (AUC = 0.87). SPJ-informed categories generally yielded higher positive predictive values for both positive and negative moves compared with numerical scores. Practical implications These findings extend evidence for the clinical utility of DUNDRUM-3 and DUNDRUM-4, highlighting their potential to support clinical decision-making in forensic psychiatric practice. SPJ tools like DUNDRUM can enhance transparency, consistency and evidence-based decisions regarding patient transfers across security levels. Originality/value This study evaluates the combined use of DUNDRUM-3 and DUNDRUM-4 under naturalistic clinical conditions, providing new insights into their utility for guiding transitions to lower security levels.