Improved sensitivity of point of care troponin I values using reporting to below the 99th percentile of normals

Abstract

IntroductionTroponin point of care tests have been found to have inferior sensitivity to laboratory based tests, when either the 10% CV or the 99th percentile of a healthy population is used as the cut-off. In a prospective study we evaluated a decreased cut-off in the detection of cardiac injury. MethodsWe compared 2 point of care assays (i-stat, Abbott Diagnostics and AQT 90, Radiometer) for troponin I with a laboratory assay for troponin I (ARCHITECT STAT troponin-I assay, Abbott Diagnostics), previously evaluated for diagnosis of acute coronary syndrome (ACS). We used the published 99th percentile and a value that was 50% of that. We investigated these tests in a convenience sample of 195 patients presenting to a suburban hospital. We used chi-square tests for the comparison and a P<0.05 as significant. Clinical outcomes were obtained for patients with elevated levels of the laboratory assay. ResultsAt the 99th percentile both assays did only detect cardiac injury in a percentage of true positives compared to the laboratory test (34 and 51%) with a significant rate of false negative values (19.6 and 14.9%). Using a decreased cut-off (50% of 99th percentile) increased detection of true positives (to 80.9 and 76.5%) with an acceptable rate of false positive results (7.3 and 7.1%) significantly (P<0.01). Clinical review showed POC tests missed 6 of 13 patients with confirmed AMI (a sensitivity of 46%) and that a lower cut-off allowed them to detect all (for the i-stat) or most (4 of 6 for the AQT) of them. DiscussionWe believe that we have described in this study a way to improve the sensitivity of point of care assays for troponin that allows us to identify additional patients without losing the specificity required to identify appropriate patients for discharge from the emergency department.