Premium
Validation of a point of care troponin assay in real life emergency department conditions
Author(s) -
Loten Conrad,
Attia John,
Hullick Carolyn,
Marley John,
McElduff Patrick
Publication year - 2009
Publication title -
emergency medicine australasia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.602
H-Index - 52
eISSN - 1742-6723
pISSN - 1742-6731
DOI - 10.1111/j.1742-6723.2009.01198.x
Subject(s) - medicine , emergency department , troponin , point of care testing , coefficient of variation , myocardial infarction , point of care , correlation coefficient , reproducibility , troponin i , emergency medicine , statistics , pathology , mathematics , psychiatry
Objective: To validate the accuracy of a Point of care (POC) troponin device (Abbott i‐Stat) in real life ED conditions. Methods: Design: A three‐way comparison between troponin I results obtained by experienced POC operators, inexperienced ED staff and central laboratory criterion standard. Participants: Convenience sample of 332 patients presenting to the ED with possible coronary syndromes. Results: Spearman correlation coefficient for experienced versus laboratory was 0.83 (95% CI 0.78–0.87), occasional users versus laboratory was 0.76 (95% CI 0.71–0.81), and experienced versus occasional users on POC was 0.82 (95% CI 0.76–0.87). Using local troponin cut‐off of 0.1 ng/mL, kappa coefficient was 0.94 for occasional users versus laboratory, 0.91 for experienced versus laboratory and 0.94 for experienced versus occasional users. Bland–Altman plots showed good agreement across the range of measured values. The sensitivity of i‐Stat ( vs laboratory as criterion standard) was 92.2% (95% CI 83.8%–97.0%) with the local cut‐off but only 70.1% (95% CI 60.5%–78.6%) using the lowest cut‐off associated with acceptable reproducibility (10% coefficient of variation). Conclusions: The i‐Stat POC device produces similar results in the hands of experienced and occasional operators in ED. There is good agreement between the POC and laboratory at levels used to diagnose infarction by older, more specific criteria. When compared using new lower cut‐offs, the i‐Stat had poor sensitivity.