Diagnostic accuracy of the G-FAST score for detecting large vessel occlusion in suspected stroke patients: a systematic review and network meta-analysis

Background Early and accurate detection of large vessel occlusion (LVO) is essential for timely endovascular thrombectomy in acute ischaemic stroke. The Gaze-Face-Arm-Speech-Time (G-FAST) scale has been proposed as a potential prehospital screening tool for LVO than the Cincinnati Prehospital Stroke Scale (CPSS). This study systematically evaluates the diagnostic performance of G-FAST and compares it with CPSS. Methods A systematic review and network meta-analysis were conducted. A comprehensive literature search in PubMed, Embase and Scopus was performed up to 31 December 2024. Pooled sensitivity, specificity and area under the receiver operating characteristic curve (AUC) were estimated. Network meta-analysis was used to compare diagnostic performance across G-FAST≥3, CPSS≥2 and CPSS≥3. Results 13 studies involving 12 414 patients were eligible for inclusion in the meta-analysis of G-FAST. G-FAST ≥3 demonstrated pooled sensitivity and specificity of 73.3% (95% CI 66.2% to 79.4%) and 74.3% (95% CI 65.3% to 81.6%), respectively, with an AUC of 0.80. In the network meta-analysis (six studies, 7012 participants), CPSS ≥2 showed the highest pooled sensitivity (79.0%, 95% CI 60.4% to 90.3%) and pooled specificity (66.1%, 95% CI 40.6% to 84.7%), whereas CPSS ≥3 demonstrated higher specificity (83.8%, 95% CI 63.6% to 93.9%) but reduced sensitivity (63.4%, 95% CI 40.1% to 81.7%). G-FAST ≥3 showed a more balanced performance between sensitivity (71.6%, 95% CI 51.4% to 85.7%) and specificity (76.4%, 95% CI 53.8% to 90.0%), with a numerically higher specificity than CPSS ≥2. Overall, there were no statistically significant differences in diagnostic performance among the scales. Conclusions G-FAST demonstrates moderate diagnostic accuracy for prehospital LVO detection, with performance comparable to traditional scales. Its use provides a feasible alternative for pre-hospital stroke triage, though further large-scale validation is warranted. PROSPERO registration number CRD42023477601.