Response by Allen et al to Letter Regarding Article, “Diagnostic Performance of High-Sensitivity Cardiac Troponin T Strategies and Clinical Variables in a Multisite US Cohort”

We appreciate the interest of Schaefer et al in our analysis.¹ Although we respectfully disagree with the error terminology used in their editorial, we recognize their important discussion points. Regarding their first concern of reporting high-sensitivity cardiac troponin T to the limit of quantification (6 ng/L) instead of the limit of detection (3 ng/L), we realize that countries outside the United States report down to the limit of detection. However, given that this analysis was a multisite US validation of commonly used high-sensitivity cardiac troponin T diagnostic strategies, the STOP CP study (High Sensitivity Cardiac Troponin T to Optimize Chest Pain Risk Stratification) investigators felt that it was most appropriate to abide by the US Food and Drug Administration’s recommendation of not reporting below the limit of quantification. Our intent is not to defend the Food and Drug Administration’s position; in fact, we feel that it would be helpful if the Food and Drug Administration would allow the reporting down to the limit of detection because there is useful prognostic information at this level.

The European Society of Cardiology (ESC) 0/1-hour (0/1-h) algorithm clearly delineates symptom onset timing as a qualification for its use. However, our analysis carefully examined data from both the early and late presenters, finding no substantive difference in algorithm performance on the basis of symptom onset. Schaefer et al requested that we present additional data on the ESC 0/1-h algorithm rule-out criteria that require onset of chest pain ≥3 hours before presentation. To address this concern, we conducted a sensitivity analysis where low risk was defined by 0 hour <6 ng/L in patients with chest pain ≥3 hours or 0 hour <12 ng/L and delta 0 to 1 hour <3 ng/L. The results of this analysis yielded no meaningful differences compared with those reported in the primary article. The negative predictive values (NPVs) for each outcome were unchanged, including a NPV of 99.0% for index myocardial infarction (MI).

Last, Schaefer et al were concerned about outcome selection. We recognize the importance of index MI and reported this outcome in our primary analysis. For index MI, we found the algorithm to have a NPV of 99%, in line with previous studies and guidelines.^2,3 However, a safe discharge strategy for patients with possible acute coronary syndrome in the emergency department is not limited to ruling out MI during the index visit. Providers must also avoid missing major adverse cardiac events within 30 days. Tools such as Emergency Department Assessment of Chest Pain Score, HEART score (history, ECG, age, risk factors, and troponin), and HEART Pathway have all been held to a 30-day major adverse cardiac events standard: evaluating the ESC 0/1-h algorithm using the same standard is reasonable. Furthermore, Dr Mueller’s group has used similar 30-day outcomes in previous ESC 0/1-h algorithm validations.⁴ In that study, the NPV for 30-day cardiac death and MI was high. However, in this cohort we were unable to replicate a high NPV for 30-day cardiac death and MI without the addition of a HEART score. The STOP CP Investigators believe it is important to provide readers with a range of outcomes from index visit through 30 days, so that providers can determine whether the use of the high-sensitivity cardiac troponin T ESC 0/1-h algorithm is consistent with their individual or health system’s risk tolerance.