Evaluating the reference range of a new high-sensitivity troponin assay using retrospective review of laboratory data
Abstract Rapid diagnosis of myocardial infarction (MI) in the emergency department (ED) is critical for proper patient management and optimal patient outcomes. High-sensitivity plasma troponin assays may help identify cardiac injury sooner by having better analytical sensitivity. When our institution first adopted the Beckman-Coulter high-sensitivity troponin I assay (hsTnI), cardiologists were hesitant to use the upper reference limit (URL) provided by the manufacturer (< 0.0136 ng/mL for female patients, < 0.0198 ng/mL for male patients), due to concerns regarding assay specificity at the URL. To help transition to the new high-sensitivity assay and its potential benefits, we began using the new hsTnI assay, but maintained the URL of the previous automated immunoassay, which was based on the 99th percentile of a healthy population (< 0.04 ng/mL = negative; 0.04 - 0.4 ng/mL = suspected MI, above the 99th percentile; > 0.4 ng/mL = likely MI). The purpose of the current study was to assess the suitability of the manufacturer’s URL for the new hsTnI assay for our health system’s patient population. To do this, we extracted patient encounter ICD-10 diagnosis codes, diagnosis-related groups (DRGs), and hsTnI results from our institution’s enterprise data warehouse and the laboratory information system for patients seen between Jan 2, 2020 and Aug 31, 2020. We included all patients who presented to the ED and received hsTnI testing. Each patient encounter was subsequently categorized as having MI or no-MI, based on diagnosis codes and DRGs. The first troponin test for each encounter was selected for analysis. The URL recommended by the manufacturer identified 92% and 100% of female and male patients with MI, respectively. Furthermore, 5.1% of 1947 female patients without MI and 9.9% of 1976 male patients without MI were above the manufacturer’s URL (female = 0.0136 ng/mL; male = 0.0198 mg/mL) but below 0.04 ng/mL. These results are lower than the false-positive rate of 12% observed in a separate study cited by the assay manufacturer. These data suggest that the manufacturer’s range is applicable to our patient population.