The following is a summary of “HARM models: Predicting longitudinal physical aggression in patients with schizophrenia at an individual level,” published in the May 2023 issue of Psychiatry by Watts et al.

In forensic psychiatric settings, predicting and preventing aggression in individuals with schizophrenia remains a top priority. While risk assessment methods are well-established in forensic psychiatry, no individual-level instruments are available to predict longitudinal physical aggression in patients with schizophrenia in forensic settings. In the present study, the researchers used evidence-based risk and protective factors and variables related to treatment course assessed at baseline to predict future physical aggression (4-month, 12-month, and 18-month follow-up) among 151 patients with schizophrenia in the forensic mental healthcare system.

The balanced accuracy (sensitivity + specificity/2) of predicting physical aggressive incidents in patients with schizophrenia ranged from 59.73 to 87.33 % at 4-month follow-up, 68.31 to 80.10% at 12-month follow-up, and 46.22 to 81.6% at 18-month follow-up, respectively, across their HARM models. In addition, they created distinct models based on clinician-rated clinical judgments of short-term and immediate violent risk as a comparison metric. Several modifiable evidence-based predictors of prospective physical aggression in schizophrenia, including impulse control, substance abuse, impulsivity, treatment non-adherence, mood and psychotic symptoms, substance abuse, and inadequate family support, were identified. 

To the best of their knowledge, the researchers’ HARM models are the first to predict longitudinal physical aggression on an individual level in schizophrenia patients in forensic settings. As these machine learning models were developed in forensic settings, it is essential to note that they may not apply to all individuals with schizophrenia. Moreover, the testing cohort exhibited a low baseline rate of physical aggression (6.0% to 11.6% across time points). Consequently, larger cohorts will be necessary to ascertain the reproducibility of these results.

Source: sciencedirect.com/science/article/abs/pii/S0022395623001012