The Patient Health Questionnaire (PHQ-9) score ≥ 10 balances best sensitivity and specificity when detecting probable depression in patients. In the general population, different cut-offs are suggested. European studies on general populations validating the PHQ-9 against a diagnostic interview to detect depression are rare.
This was a cross-sectional observational epidemiological survey using multistage household probabilistic sampling to recruit a representative adult sample (N = 1203; age = 43.7 ± 13.6; 48.7% male). Mental disorders including current major depressive episode (MDE) were observer-rated (Mini International Neuropsychiatric Interview). The PHQ-9, quality of life (QoL), and loneliness were self-assessed. We performed validity and reliability tests of the PHQ-9 and receiver operating curve (ROC) analysis.
The Serbian PHQ-9 was internally consistent and correlated in the expected directions with QoL and loneliness. At the cut-off score ≥ 8, sensitivity was .85 and specificity was .91. ROC analysis showed that the area under the curve was .95, indicating that the Serbian PHQ-9 can discriminate very well between persons with/without MDE.
When the PHQ-9 is assessed against the structured diagnostic interview in the general population to detect depression, the cut-off of ≥8 balances best sensitivity and specificity.