Sensitivity and specificity of the PHQ-9 for identifying suicidality in medical rehabilitation

Abstract

Suicidality is rarely assessed in patients treated in German rehabilitation centers, although individuals with physical and mental impairments have an increased risk for suicidality. The item 9 of the 9-item depression module of the Patient Health Questionnaire (PHQ-9) asks about the desire to be dead or to harm oneself and could be used as a possible screening for suicidality. The Columbia Suicide Severity Rating Scale (C-SSRS) is a standardized interview to assess suicidality. We assessed the prevalence of suicidality in orthopedic and cardiac rehabilitation using the C-SSRS and the PHQ-9, examined the diagnostic accuracy of the item 9 of the PHQ-9 and of the PHQ-9 total score for identifying suicidality and surveyed the acceptance of the suicidality assessment by the rehabilitants. Study participants were screened with the PHQ-9 and subsequently interviewed using C-SSRS. Sensitivity and specificity of the item 9 of the PHQ-9 and the PHQ-9 total score were tested for the presence of suicidality assessed with the C-SSRS and in a sensitivity analysis for the presence of nonspecific active suicidal ideation (item 2 of the C-SSRS). We calculated the area under the curve (AUC) to predict the ability of the PHQ-9 to discriminate between individuals with and without acute suicidality. Screening and interview were evaluated by the rehabilitants. Among 405 study participants, the prevalence of acute suicidality measured by the C-SSRS was 0.5%. 4% reported nonspecific active suicidal ideation on the C-SSRS. 10.4% reported suicidal ideation on item 9 of the PHQ-9. The sensitivity of item 9 and the PHQ-9 total score for identifying acute suicidality was only 50.0% (95% CI: 1.3% to 98.7%). However, item 9 was sensitive (81.3%, 95% CI: 54.4% to 96.0%) and specific (92.5%, 95% CI: 89.5% to 95.0%) for identifying nonspecific active suicidal ideation. Estimators for sensitivity were highly uncertain because of the low prevalence of acute suicidality. Addressing suicidality was rated as useful and helpful by study participants. If the response to item 9 is positive, the immediate risk of suicide is low. However, PHQ-9 is suitable for identifying patients with non-specific suicidal thoughts. A conspicuous score on the item 9 of the PHQ-9 should be further clarified by a psychotherapist.