A nice comparison of the sensitivities and specificities of the various biomarkers for acute myocardial infarction at ~3.5 hours after symptom onset. Each biomarker was set at the 95th or 99th percentile based on manufacturers definitions for their reference table, and then they also show ROC curves and calculate AUCs for each.
Essentially, none of the biomarkers is completely adequate for ruling out AMI given the constraints of their study. Their best combination, for both sensitivity and specificity, is combining the heart fatty acid binding protein and the troponin – which they state provides an NPV of 95.6%, outperforming the “triple rule out” of troponin, CKMB, and myoglobin at 92.1%. Interestingly, they also state that if they used clinical risk stratification, they could select a population in which HFAP and troponin together get up to 96.9% NPV…showing that regardless the resources we throw at the problem of “low risk chest pain”, it is an absolutely Quixotic quest to definitively rule out every MI in the Emergency Department. 3% of their “very low risk” population that was biomarker negative with the best sensitivity they could muster ended up ruling in for AMI during their subsequent hospital stay. They only way we’re going to prevent healthcare from becoming bankrupt is increasing our levels of acceptable risk.
As a side note, this article gets the award for “best vs. least professional” title so far this year.