Cardiac marker sensitivity depends on chest pain duration at the time of sampling. Our objective was to estimate the sensitivity, specificity, and likelihood ratios of early CK–MB and myoglobin assays in patients presenting to the emergency department (ED) with nondiagnostic ECGs, stratified by the duration of ongoing chest pain at the time of ED assessment.

This was a prospective observational study carried out in 10 US and 2 Canadian EDs. Patients >25 years of age with ongoing chest pain and nondiagnostic ECGs were stratified by pain duration (0–4 h, 4–8 h, 8–12 h, >12 h). CK–MB and myoglobin assays were drawn at T = 0 (ED assessment) and T = 1 hr. Patients were followed for 7–14 days to identify all cases of acute myocardial infarction (AMI). ED test results were correlated with patient outcomes.

Of 5005 eligible patients, 565 had AMI. Pain duration was 0–4 h in 3014 patients, 4–8 h in 961, 8–12 h in 487, and >12 h in 543. Marker sensitivity increased with pain duration, ranging from 28%–77% for CK–MB and 39%–73% for myoglobin. The maximal sensitivity achieved by a T = 0 assay was 73%, and this was in patients with 8–12 or >12 h of ongoing pain. No combination of tests achieved 90% sensitivity in any pain duration strata.

Regardless of chest pain duration, single assays and early serial markers (0+1 hr) do not rule out AMI; therefore, serial assays over longer observation periods are required. Likelihood ratios derived in this study will help physicians who use Bayesian analysis to determine post-test AMI likelihood in patients with chest pain.