RESUMO
RATIONALE: Acute coronary syndrome (ACS) requires improved diagnostic accuracy through useful, safe and easy-to-apply tools. OBJECTIVES: To obtain an assessment scale for the diagnosis of ACS in patients with chest pain and nondiagnostic electrocardiogram and troponin concentrations. METHODS: A prospective cohort study included 286 patients treated in the emergency department for chest pain, with normal electrocardiogram and troponin levels. Using multiple logistic regression, we obtained the independent predictors for the diagnosis of ACS. The assessment scale's discriminative power was assessed with the area under the ROC curve. RESULTS: The diagnosis of ACS was confirmed in 103 patients (36%). The final predictive model included 3 endpoints: a history of coronary artery disease, hyperlipidaemia and a score≥6 points on the Geleijnse scale. The area under the ROC curve for the final model was 0.90 (95% confidence interval [95% CI] 0.85-0.93). A threshold of 5 points achieved a sensitivity of 76.7% (95% CI 68-84), a specificity of 91.8% (95% CI 87-95), a positive likelihood ratio of 9.36 (95% CI 5.70-15.40), a negative likelihood ratio of 0.25 (95% CI 18.00-36.00) and an overall diagnostic accuracy of 86.4% (95% CI 82-90). The predictive model was superior to the Geleijnse scale alone. CONCLUSIONS: The final scale showed good discriminative capacity for diagnosing ACS and could therefore be of interest for identifying ACS in emergency departments. Nevertheless, the scale needs to be validated in larger multicentre studies.