Identifying suicidal ideation in general medical patients.

OBJECTIVES: To describe characteristics of general medical patients with suicidal ideation and to develop an efficient clinical strategy for identifying these patients.

DESIGN: Cross-sectional analysis of data without intervention.

SETTING: A representative sample of community respondents interviewed as part of the National Institute of Mental Health Epidemiologic Catchment Area surveys.

SUBJECTS: A total of 6041 individuals who reported receiving care in general medical settings in the 6 months preceding the baseline interview.

MAIN OUTCOME MEASURE: Suicidal ideation within the year preceding the baseline interview.

RESULTS: A total of 154 patients (2.6%) had experienced suicidal thoughts within the previous year. In multivariate analyses, the following sociodemographic factors were significantly associated with having suicidal ideation: being white (odds ratio [OR], 2.8; P < .001) relative to African American, separated or divorced (OR, 1.8; P = .03) relative to married, and aged 18 to 30 years (OR, 2.3; P = .02) or aged 31 to 50 years (OR, 2.1; P = .03) relative to those older than 65 years. Gender, educational level, and socioeconomic status were not statistically associated with suicidal ideation. Psychiatric disorders significantly associated with suicidal ideation were major depression (OR, 10.3; P < .001), panic disorder (OR, 5.2; P < .001), alcohol disorder (OR, 2.0; P = .04), and phobic disorder (OR, 1.6; P = .02) within the previous year. However, only 34% of respondents with suicidal ideation met criteria for major depression. With the aid of stepwise logistic regression, a four-item screening questionnaire, the Suicidal Ideation Screening Questionnaire, was developed to predict patients with suicidal ideation. Likelihood ratios for suicidal ideation increased from 0.21 to 32.8 with more positive responses to questions concerning symptoms of hopelessness, guilt, depressed mood, and sleep disturbances.

CONCLUSIONS: If the results from the four-item screen are reproduced in a clinical setting, general medical patients with a high likelihood for suicidal ideation may be identified with a brief screening questionnaire. This screen would have greater sensitivity than would be achieved by assessing suicidal ideation only in those patients who meet criteria for a distinct psychiatric disorder.