Detection and classification of suicidal behavior and nonsuicidal self-injury behavior in emergency departments.

OBJECTIVE

The degree of concordance between clinical and standardized assessments in the detection and classification of suicidal behavior and nonsuicidal self-injury behavior in psychiatric emergency department settings was examined.

METHOD

Two hundred fifty-four patients at 3 psychiatric emergency departments between 2007 and 2011 were evaluated by research staff using standardized assessments of suicidal behavior and nonsuicidal self-injury behavior. Of 254 patients, 128 (50%) made a recent suicide attempt, 30 (12%) engaged in recent nonsuicidal self-injury behavior, 20 (8%) made a recent suicide attempt interrupted by self or others, and 76 (30%) had other psychiatric symptoms in the absence of recent suicidal or nonsuicidal self-injury behavior. The classifications derived from the standardized assessments using the Centers for Disease Control and Prevention (CDC) nomenclature were compared to clinical assessments independently extracted from emergency department medical records.

RESULTS

Agreement between clinical and standardized assessments was substantial for both suicide attempts (κ = 0.76, P < .001) and nonsuicidal self-injury behavior (κ = 0.72, P < .001). Importantly, 18% of patients determined to have made a suicide attempt in the past week by standardized assessment were not identified as such by clinical assessment. In addition, as measured by the Columbia Suicide Severity Rating Scale, the potential lethality of attempts for patients classified as making a recent suicide attempt by both clinician and standardized assessments was significantly greater (t120 = 2.1, P = .04) than that for patients who were classified as having made an attempt by the standardized but not clinical assessment.

CONCLUSIONS

The use of standardized assessment measures may improve sensitivity and accuracy of identifying suicidal behavior and nonsuicidal self-injury behavior in psychiatric emergency departments.