Impact of high-sensitivity cardiac troponin I assay imprecision on the safety of a single-sample rule-out approach for myocardial infarction
Ziwen Li,Yong Yong Tew,Peter A. Kavsak,Kristin M. Aakre,Allan S. Jaffe,Fred S. Apple,Paul O. Collinson,Nicholas L. Mills
DOI: https://doi.org/10.1515/cclm-2024-1011
2024-09-21
Clinical Chemistry and Laboratory Medicine (CCLM)
Abstract:To the Editor, High-sensitivity cardiac troponin (hs-cTn) assays facilitate identification of low-risk patients for immediate discharge and thus may help reduce emergency department (ED) overcrowding. The HiSTORIC trial evaluated implementation of a single-sample rule-out pathway using a hs-cTnI assay in 31,492 consecutive patients [1]. Following implementation, length of stay was reduced by 3 h and discharge from the ED increased by 21 % with no evidence of adverse events [1]. However, assay imprecision at the low hs-cTn thresholds deployed in single-sample rule-out pathways could impact performance [2]. There is presently little guidance on the allowable imprecision at lower concentration thresholds [3]. Here we model the impact of imprecision on misclassification and therefore the safety of a single-sample rule-out strategy for myocardial infarction (MI). In 48,282 consecutive patients with suspected acute coronary syndrome (median age 61 [interquartile range 49–75] years, 47 % female) evaluated using a hs-cTnI assay (ARCHITECT STAT , Abbott Laboratories; Abbott Park, IL, USA) in the HighSTEACS trial [4], we simulated the effect of imprecision for a range of CVs at the rule-out threshold of <5 ng/L: 10 % (±0.5 ng/L), 20 % (±1 ng/L), 30 % (±1.5 ng/L), 40 % (±2 ng/L), 50 % (±2.5 ng/L), 60 % (±3 ng/L), 70 % (±3.5 ng/L), 80 % (±4 ng/L), 90 % (±4.5 ng/L), and 100 % (±5 ng/L). The rnorm function in R was used, setting the mean to the initial cTn concentrations at presentation (ground truth) and the standard deviation to the absolute change corresponding to each CV. The resulting concentrations were then rounded to the nearest whole numbers as used in clinical practice. To model the impact of imprecision, we repeated the simulation 100 times and estimated the proportion reclassified to higher risk (from <5 ng/L to ≥5 ng/L), to lower risk (from ≥5 ng/L to <5 ng/L), and overall, in patients with cTn concentrations below the 99th percentile at presentation. We subsequently calculated the likelihood of (i) a primary outcome of an adjudicated index diagnosis of type 1 MI, type 4b MI or type 4c MI and (ii) a secondary outcome of any myocardial injury on serial testing, in those potentially misclassified as low risk at presentation due to assay imprecision. The 95 % confidence intervals (95 % CI) were estimated using a binomial likelihood with an equal-tailed Jeffreys prior using the binom. bayes function in R. Based on our simulations, the effect of assay imprecision on the proportion reclassified from low to higher risk, which could result in unnecessary serial testing, increased as the CV and cTn concentrations increased from 1 to 4 ng/L (Figure 1). Similarly, the effect of assay imprecision on the proportion of reclassified from higher risk to low risk, which could result in harm, increased as the CV increased, and cTn concentrations decreased from 21 ng/L to 5 ng/L. Overall reclassification was depicted as a cone-shaped gradient. The likelihood of primary and secondary outcomes associated with reclassification from higher risk to low risk had a similar gradient. Estimated proportion of patients with suspected acute coronary syndrome reclassified due to hs-cTnI assay imprecision at the rule-out threshold of <5 ng/L and likelihood of MI or myocardial injury in those misclassified as a proportion of the total number of patients at each measured cardiac troponin concentration. (A) Proportion reclassified as higher risk by assay imprecision (%). (B) Proportion reclassified as lower risk by assay imprecision (%). (C) Overall proportion reclassified by imprecision (%). (D) Likelihood of MI in those incorrectly classified as lower risk by imprecision (%). (E) Likelihood of myocardial injury in those incorrectly classified as lower risk by imprecision (%). Patients with STEMI (n=925) were excluded from the modelling. For example, using an assay with a 10 % CV at the rule-out threshold of <5 ng/L, reclassification to low risk due to assay imprecision would occur in 15.9 % (95%CI 14.4%–17.5 %) and 0.1 % (95%CI 0%–0.4 %) patients with a measured cTn concentration of 5 ng/L and 6 ng/L, respectively (Table 1). The likelihood of having a MI in those reclassified due to imprecision who actually had a value of 5 ng/L when the measured concentration was<5 ng/L would be 0.3 % (95 %CI 0.1 %–0.5 %). Overall imprecision at this level would result in 0.01 % (95 %CI 0.01 %–0.03 %) of all patients evaluated being incorrectly classified as low risk who may have a missed index MI. Estimated proportion of patients with suspected acute coronary syndrome misclassified due to hs-cTnI assay imprecision (coefficient of variation (CV) from 10 to 50 %) at the rule-out threshold of <5 ng/L and the proportion of all patients evaluated (n=47,357) with missed MI or myocardial injury due to imprecision. Whilst the proportion of patients -Abstract Truncated-
medical laboratory technology