Evaluation of veterans' suicide risk with the use of linguistic detection methods

Abstract

Objective: Many people who die from suicide received recent medical care prior to their death. Suicide risk assessment tools for health care settings focus on a variety of clinical and demographic factors but generally do not examine the text of notes written by clinicians about patients who later die from suicide. This study examined whether clinicians' notes indicated increased use of distancing language during the year preceding patients' suicide. Methods: The linguistic content of clinicians' notes for outpatients of U.S. Department of Veterans Affairs (VA) medical centers was examined in the year preceding suicide of 63 veterans. Approximately half of the veterans had received mental health services. They were matched based on mental health service use with living VA outpatients. Linguistics software was used to construct quantitative themebased categories related to distancing language and to examine temporal trends via keyword analysis. Results: Analysis of clinical notes for outpatients who died from suicide and those who did not revealed a significant difference in clinicians' distancing language. Multiple keywords emerged that also were related to distancing language, and their relative frequency increased in the time approaching the suicide. Conclusions: Linguistic analysis is a promising approach to identify use of distancing language by clinicians, which appears to be a marker of suicide risk. This pilot work indicates that additional analysis and validation with larger cohorts are warranted.