Background: Depression is a major burden for the health-care system worldwide. Most care for depression is delivered by general practitioners (GPs). We assessed the rate of true positives and negatives, and false positives and negatives in primary care when GPs make routine diagnoses of depression.
Methods: We undertook a meta-analysis of 118 studies that assessed the accuracy of unassisted diagnoses of depression by GPs. 41 of these studies were included because they had a robust outcome standard of a structured or semi-structured interview.
Findings: 50 371 patients were pooled across 41 studies and examined. GPs correctly identified depression in 47.3% (95% CI 41.7% to 53.0%) of cases and recorded depression in their notes in 33.6% (22.4% to 45.7%). 19 studies assessed both rule-in and rule-out accuracy; from these studies, the weighted sensitivity was 50.1% (41.3% to 59.0%) and specificity was 81.3% (74.5% to 87.3%). At a rate of 21.9%, the positive predictive value was 42.0% (39.6% to 44.3%) and the negative predictive value was 85.8% (84.8% to 86.7%). This finding suggests that for every 100 unselected cases seen in primary care, there are more false positives (n=15) than either missed (n=10) or identified cases (n=10). Accuracy was improved with prospective examination over an extended period (3-12 months) rather than relying on a one-off assessment or case-note records.
Interpretation: GPs can rule out depression in most people who are not depressed; however, the modest prevalence of depression in primary care means that misidentifications outnumber missed cases. Diagnosis could be improved by re-assessment of individuals who might have depression.