Classifying Chest Pain as Typical vs. Atypical Fails to Discriminate Acute Coronary Occlusion and Does Not Predict Mortality

BackgroundWhether symptom descriptors meaningfully change post-test probability for occlusion myocardial infarction (OMI) in the emergency department (ED) remains uncertain, particularly when the reference standard is imperfect and may miss NSTEMI-OMI without ST elevation. MethodsWe retrospectively analyzed 13,262 ED encounters (2021-2025) with chest pain, ECG, and hs-cTnI testing. The index test was a legacy descriptor (type A/B vs C/D). The composite reference was door-to-balloon activation or peak hs-cTnI [≥] 5,000 ng/L, with prespecified sensitivity analyses and a Bayesian latent-class model (BLCA) to correct for reference misclassification. Prognostic value for in-hospital mortality was assessed using patient-clustered multivariable logistic regression with performance and calibration. ResultsIn frequentist analyses, sensitivity was 0.690 (95% CI 0.657-0.722) and specificity 0.500 (0.491-0.509) (LR+ 1.38, LR- 0.62). BLCA that modeled an imperfect reference yielded a posterior LR+ {approx}1.6 and LR- {approx}0.43. "Typical" pain did not independently predict mortality after adjustment (adjusted OR 1.13; 95% CI 0.71-1.79). ConclusionsIn a large, consecutive ED cohort, chest-pain descriptors offer limited diagnostic leverage for OMI, even after correcting for reference misclassification with BLCA and probing potential missed NSTEMI-OMI. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=148 SRC="FIGDIR/small/25330295v2_ufig1.gif" ALT="Figure 1"> View larger version (33K): org.highwire.dtl.DTLVardef@12bad06org.highwire.dtl.DTLVardef@6c8343org.highwire.dtl.DTLVardef@1dbc8borg.highwire.dtl.DTLVardef@455c33_HPS_FORMAT_FIGEXP M_FIG C_FIG