The predictive value of risk categorization in schizophrenia

Background: Risk assessment is increasingly used to inform decisions regarding the psychiatric treatment of patients with schizophrenia and other serious mental disorders. Aims: To examine the theoretical limits of risk assessment and risk categorization as applied to a range of harms known to be associated with schizophrenia. Methods: Using known rates of suicide, homicide, self-harm, and violence in schizophrenia, a hypothetical tool with an unrealistically high level of accuracy was used to calculate the proportion of true- and false-positive risk categorizations. Results: Risk categorization incorrectly classified a large proportion of patients as being at high risk of violence toward themselves and others. Conclusion: Risk assessment and categorization have severe limitations. A large proportion of patients classified as being at high risk will not, in fact, cause or suffer any harm. Unintended consequences of inaccurate risk categorization include unwarranted detention for some patients, failure to treat others, misallocation of scarce health resources, and the stigma arising from patients' being labeled as dangerous.