Abstract 14266: Outcomes of Chest Pain Patients Following Application of a Validated High Sensitivity Troponin Algorithm: Comparison of 30 Day Death/ MI of Algorithm Concordant vs Discordant Discharges

Abstract

Introduction: High sensitivity troponin (hsTn) assays effectively risk stratify patients with chest pain when a validated serial algorithm is utilized. Most of these algorithms require multiple draws for initial values in an intermediate range below a cutoff considered unequivocally consistent with MI. Little is known regarding outcomes of patients who are discharged following deviation from these algorithms. Hypothesis: Minor deviation from a validated hsTn algorithm may yield unacceptable outcomes in chest pain patients. Methods: Retrospective review of outcomes in patients presenting with chest pain to Geisinger Health System emergency departments and discharged based on application of a validated 0/1/3 hour hsTroponinT (hsTnT) algorithm over a 2-year period. We compared 30-day death/MI in patients who had hsTnT results that were consistent with “ruling out” (algorithm concordant) vs those who were discharged despite either equivocal hsTnT results or incomplete application of the algorithm (algorithm discordant). Patients who were admitted or met criteria for MI based on a peak hsTnT > 52ng/L were excluded. Results: Out of 10,342 patients discharged after hsTnT algorithm, 1,166 (11.3%) were algorithm-discordant. Thirty-day death/MI occurred in 13 (0.14%) of patients managed protocol concordant and 16 (1.37%)of patients managed protocol discordant (p<0.001). The 16 discordant pts with death/MI were algorithm-discordant due to either having only a single hsTnT draw (12) or a 0-1hr delta > 5ng/L (4). Comparing all patients with vs without 30day death/MI, univariate predictors included median age (66 vs 50yr, p<0.0001), prior CAD history (41.4 vs 17.5%, p=0.002), hyperlipidemia (44.8 vs 21.8%, p=0.006) and hsTnT algorithm discordance (55.2 vs 11.1%, p<0.001). A multivariable logistic regression model found that among univariate predictors only algorithm discordance was an independent predictor of 30-day death/MI (OR 6.42, p<0.001). Conclusions: Deviation from an hsTnT algorithm was associated significantly with increased 30 day cardiac events. This study highlights the importance of maintaining concordance with algorithm details when using hsTnT assays to triage chest pain patients.