Study design: Cross-sectional study.
Objective: To examine the sensitivity of 2 single-item questions compared with 2 longer questionnaires for screening depression and anxiety among patients with chronic low back pain (CLBP).
Summary of background data: Psychosocial factors are frequently identified as risk factors for developing CLBP and as predictors for treatment, and questionnaires are often used to screen for this. Shorter instruments may be easier to use in clinical practice settings.
Methods: A total of 564 patients with 2 to 10 months of at least 50% sickness absence due to nonspecific low back pain were assessed for depression and anxiety with the Mini-International Neuropsychiatric Interview (MINI). Single-item questions for depression and anxiety from the Subjective Health Complaint Inventory and 2 longer questionnaires, the Hospital Anxiety and Depression Scale and Hopkins Symptom Checklist-25, were compared with MINI results, considered the "gold standard" in this study. Sensitivity and specificity of single-item and longer questionnaires and receiver operating characteristic curves were compared.
Results: According to MINI, the prevalence of anxiety disorders was 12% whereas that of depressive disorders was 4%. The screening questions showed 95% sensitivity and 56% specificity for depressive disorders and 68% sensitivity and 85% specificity for anxiety disorders. The longer questionnaire, Hospital Anxiety and Depression Scale, showed 91% sensitivity and 85% specificity for depressive disorders and 58% sensitivity and 83% specificity for anxiety disorders. Hopkins Symptom Checklist-25 showed 86% sensitivity and 74% specificity for depressive disorders and 67% sensitivity and 87% specificity for anxiety disorders. For 3 of the anxiety disorders and 2 of the depressive disorders, a perfect sensitivity was found between the screening questions and MINI.
Conclusions: A single-item screening question was sensitive for depression but less sensitive for anxiety. The screening questions further performed equal to 2 widely used questionnaires. Validation of these results in other populations and compared with other short-item screeners is needed.