Diagnostic yield and triage accuracy in a rapid-access transient ischaemic attack (TIA) clinic: a service evaluation

Background: Rapid-access transient ischaemic attack (TIA) clinics have traditionally allocated referrals to high- and low-risk pathways, whereas current UK guidance recommends specialist assessment of all suspected TIA within 24 hours. We evaluated diagnostic yield, triage performance and compliance with local referral-to-appointment targets. Methods: Retrospective single-centre service evaluation of 252 consecutive referrals (January–March 2025). Diagnoses were established by specialist record review. The main classification categories consisted of: CVA-spectrum (confirmed stroke, confirmed TIA, possible TIA) and mimics. Triage category was compared with a reference standard of confirmed TIA or stroke using sensitivity, specificity and logistic regression. Results: 128 referrals (50.8%) were cerebrovascular and 124 (49.2%) were mimics, most commonly migraine/headache and vasovagal causes. High-risk triage had 57.1% sensitivity and 51.8% specificity for confirmed TIA or stroke (p=0.21); 35 of 79 confirmed TIAs (44.3%) had been triaged low-risk. In sensitivity analyses, the association reached significance when possible TIA was counted as cerebrovascular, but discrimination remained weak (Youden's J ≤0.17). Local targets were met by 84.8% of high-risk (≤1 day) and 66.4% of low-risk (≤7 days) patients; 38/113 low-risk patients (33.6%) waited over 7 days. Yield differed between referral sources (p=0.027); eye casualty referrals had the highest CVA-spectrum yield (78.6% vs 47.3% for other referrals; p=0.002). Conclusion: About half of referrals were cerebrovascular. Eye casualty referrals should not be deprioritised. Local two-tier triage distinguished cerebrovascular diagnoses from mimics only weakly, and one-third of low-risk patients exceeded their target. These single-centre findings support reviewing two-tier triage and prospectively evaluating a uniformly urgent pathway.