Development and validation of a predictive scoring tool for family violence risk assessment

Abstract

Background: Family violence remains a critical global public health concern, often underreported and undetected in clinical settings. Early risk identification is vital yet hindered by reliance on unstructured clinical judgment and insufficient screening tools. Objective: This study aimed to develop and validate a predictive scoring tool integrating individual, relational, and contextual factors to systematically assess family violence risk, thereby supporting earlier interventions. Methods: A cross-sectional observational study was conducted with 548 participants aged 30–59 years in Phitsanulok, Thailand. Data on personal history, relational dynamics, and community environment were collected through structured questionnaires. Bivariate and multivariate analyses identified key predictors of family violence. A weighted Family Violence Risk Score (FVRS) was constructed and internally validated using k-fold cross-validation and ROC curve analysis. Results: Major predictors included intergenerational exposure to violence (r = 0.77), rigid gender norms (r = 0.79), substance abuse (r = 0.74), emotional distress (r = 0.72), and financial hardship (r = 0.70). The FVRS demonstrated excellent discriminative ability (AUC = 0.97; 95% CI: 0.94–0.99). A cut-off score of ≥31 yielded a sensitivity of 94% and specificity of 72% for detecting any family violence. Conclusions: The developed FVRS offers a robust, evidence-based method for stratifying family violence risk, enabling proactive identification and intervention. Future studies should externally validate the tool across diverse populations and explore its integration into clinical practice to improve violence prevention strategies.