BACKGROUND: Evidence suggests that general practitioners (GPs) fail to diagnose up to half of common mental disorder cases. Yet no previous research has systematically summarized the evidence in the case of anxiety disorders. The aim of this review was to systematically assess and meta-analyze the diagnostic accuracy of GPs' assisted (i.e., using severity scales/diagnostic instruments) and unassisted (without such tools) diagnoses of anxiety disorders. METHODS: Systematic review (PROSPERO registry CRD42013006736) was conducted. Embase, Ovid Journals - Ovid SP Medline, Pubmed, PsycINFO, Scopus, Web of Science, and Science Direct were searched from January 1980 through June 2014. Seven investigators, working in pairs, evaluated studies for eligibility. The quality of included studies was assessed with the Quality Assessment of Diagnostic Accuracy Studies tool version 2 (QUADAS-2). The main outcome measures were sensitivity and specificity of clinical diagnoses of any anxiety disorder. We pooled sensitivity and specificity levels from included studies using bivariate meta-analyses. RESULTS: Twenty-four studies were included in the meta-analysis with a total sample of 34,902 patients. Pooled sensitivity and specificity were estimated at 44.5% (95% CI 33.7-55.9%) and 90.8% (95% CI 87-93.5%). GPs' sensitivity was higher when diagnoses were assisted (63.6%, 95% CI 50.3-75.1%) than when unassisted (30.5%, 95% CI 20.7-42.5%) to the expense of some specificity loss (87.9%, 95% CI 81.3-92.4% vs. 91.4%, 95% CI 86.6-94.6%, respectively). Identification rates remained constant over time (P-value = .998). CONCLUSIONS: The use of diagnostic tools might improve detection of anxiety disorders in "primary care."

This study was funded by a grant from the Spanish Ministry of Health, Instituto de Salud Carlos III, FEDER PI10/00530, Fondo de Investigación Sanitaria ISCIII (FI11/00154), and by DIUE Generalitat de Catalunya (2014 SGR 748)