Abstract
We report the in-house determination of the 99th percentile upper reference limit (URL) of the Abbott STAT TnI assay. To calculate the 99th percentile URL, we statistically analysed the results of basal cardiac Troponin I (cTnI) concentrations in samples collected from 198 patients (114 male, 84 female, age range 30–94 y) recruited into a study looking at the cardiac effects of colonoscopy. 1,2
Patients who participated in this prospective cohort study were stratified into three groups according to their risk factors for cardiovascular disease. Group A (44 patients) included patients with documented significant cardiac co-morbidities. Group B (113 patients) included patients with risk factors but no proven heart disease. Group C (41 patients) included patients who had no risk factors for heart disease. All patients underwent 12-lead Holter monitoring to record their electrocardiogram (ECG) during the colonoscopy procedure along with a pre- and 24-hour post-procedure cTnI measurement. The samples were analysed as routine specimens on receipt in the laboratory, all results were blinded with the ECGs being evaluated independently by two cardiologists.
In order to statistically analyse the data, all cTnI results recorded as <0.01 μg/L were transformed to read 0.009 μg/L. In addition, results from three patients (2 from group A and 1 from group B) in which the initial cTnI was ≥0.09 μg/L (cTnI cut-off in effect during the period of the study) were eliminated from the data analysis, further investigation of these patients identified cardiac problems. Statistical analysis of the remaining data by the Mann-Whitney test showed significant difference between group A and groups B and C (P = 0.015 and 0.0162, respectively), but no significant difference between groups B and C (P = 0.4843). Accordingly the data from groups B and C were combined to determine the 99th percentile URL.
After exclusion of one result (cTnI 0.15 μg/L) from groups B and C, there were 153 results remaining (79 male, 74 female, 30–85 y age), of these 82% (126) were <0.01 μg/L (transformed to 0.009 μg/L for inclusion in the statistical calculation) with a range of <0.01–0.03 μg/L. From these results the 99th percentile URL for the Abbott STAT TnI assay was calculated to be 0.03 μg/L (manufacturer's package insert cites 0.028 μg/L).
To determine the decision limit for myocardial injury, we estimated the coefficient of variation (CV) of the assay at cTnI concentrations close to the URL. The All Wales Clinical Biochemistry Audit Group guidelines 3 recommend that the decision limit should be the lowest troponin concentration that can be statistically distinguished (±2SDs) from the 99th percentile URL. The Joint European Society of Cardiology and American College of Cardiology consensus document 4 recommends that the acceptable imprecision, as measured by coefficient of variation (CV), at the decision limit should be ≤10%. The coefficient of variation (CV) of the Abbott STAT TnI assay at 0.023, 0.055 and 0.09 μg/L has been previously reported as 20, 10 and 6% respectively. 5,6 By extrapolation the CV of the assay would be 16% at 0.03 μg/L (±2 SD = 0.02–0.04), 12.5% at 0.04 μg/L (±2 SD = 0.03–0.05) and 10% at 0.05 μg/L (±2 SD = 0.04–0.06) and cTnI results equal to or greater than 0.05 μg/L are significantly raised compared with the 99th percentile URL of 0.03 μg/L.
In conclusion this study determined the 99th percentile URL of the Abbott STAT TnI assay, in an adult population without heart disease, to be 0.03 μg/L. It is also recommended that a clinical decision limit (cut-off) of ≥0.05 μg/L (CV10%) is adopted as the lowest cTnI concentration that can be statistically distinguished from the 99th percentile URL of 0.03 μg/L (CV16%).
DECLARATIONS