Examining Systemic and Interpersonal Bias in Violence Risk Assessments of Patients in Acute Psychiatric Care

The assessment and management of inpatient risk for violence in acute psychiatric care are challenges that introduce the potential for bias. This study aimed to examine inequities based on social determinants of health (SDoH) (e.g., race-ethnicity, gender, or mode of admission to acute care) that may lead to unfair assessment of psychiatric patients.

The authors analyzed electronic health records of 7,424 acute care patients across 12,650 stays (2016-2022) at a large Canadian psychiatric hospital. Risk ratios (RRs) were calculated by SDoH for staff assessments of high risk (perceived risk), for violent incidents (actual risk), and for potentially biased risk assessment (particularly when a patient was assessed as high risk but did not become violent).

In univariate analyses, patients assessed as high risk who did not become violent were more likely to be male than female and to be Black, Indigenous, or Middle Eastern than White. When RRs were mutually adjusted for all variables, the associations for gender and race-ethnicity were attenuated or were no longer statistically significant. Associations with potentially biased risks that remained significant included most psychiatric diagnoses (vs. a depressive or anxiety disorder), supportive or unstable housing (vs. owning a home), and admission by police (vs. self-admission; RR=2.14, 95% CI=1.92-2.40).

Systemic factors, such as admission by police and housing status, and having severe mental illness were the primary drivers of observed inequities in risk assessments of patients from racial-ethnic minority groups. Addressing these systemic factors might be key to improving acute psychiatric care.