Rittirak Othong1, Suttisak Wattanasansomboon1, Thanakorn Kruutsaha1, Douglas Chesson2, Sakda Arj-Ong Vallibhakara3, Ziad Kazzi2. 1. a Department of Emergency Medicine, Faculty of Medicine Vajira Hospital , Navamindradhiraj University , Bangkok , Thailand. 2. b Department of Emergency Medicine , Emory University , Atlanta , GA , USA. 3. c Section for Clinical Epidemiology and Biostatistics, Ramathibodi Hospital, Faculty of Medicine , Mahidol University , Bangkok , Thailand.
Abstract
INTRODUCTION: New QT correction formulae derived from large populations are available such as Rautaharju's [QTcRTH = QT * (120 + HR)/180] and Dmitrienko's [QTcDMT = QT/RR0.413]. These formulae were derived from 57,595 and 13,039 cases, respectively. Recently, a study has shown that they did not experience errors across a wide range of heart rates compared to others. OBJECTIVES: (1) To determine the best cut-off value of QTcRTH and QTcDMT as a predictor of torsade de pointes (TdP) and (2) to compare the sensitivity and specificity using the cut-off value of QTcRTH with those of the QTcBazett (QTcBZT), QTcFridericia (QTcFRD), and QT nomogram. METHODS: Data were derived from two data sets. All cases aged over 18 years with an exposure to QT-prolonging drugs. Group-1, all cases developed TdP. Data in Group-1 were obtained from systematic review of reported cases from Medline since its establishment until 10 December 2015. Group-2 is composed of those who overdosed on QT prolonging drugs but did not develop TdP. This data set was previously extracted from a chart review of three medical centers from January 2008 to December 2010. Data from both groups were used to calculate QTcRTH and QTcDMT. The cut-off values from QTcRTH and QTcDMT that provided the best sensitivity and specificity to predict TdP were then selected. The same method was applied to find those values from QTcBZT, QTcFRD, and QT nomogram. The receiver operating characteristic curve (ROC) was applied where appropriate. RESULTS: Group-1, 230 cases of drug-induced TdP were included from the systematic review of Medline. Group-2 (control group), which did not develop TdP, consisted of 292 cases. After applying all of the correction methods to the two datasets, the best cut-off values that provided the best accuracy (Ac) with the best sensitivity (Sn) and specificity (Sp) for each formula were as follows: QTcRTH at 477 milliseconds (ms), Ac = 89.08%, Sn = 91.30% (95%CI = 86.89-94.61), Sp = 87.33%(95%CI = 82.96-90.92); QTcDMT at 475 ms, Ac = 88.31%, Sn =91.30% (95%CI = 86.89-94.61), Sp = 85.96%(95%CI = 81.44-89.73); QTcBZT at 490 ms, Ac = 86.97%, Sn = 88.26% (95%CI = 83.38-92.12), Sp = 85.96% (95%CI = 81.44-89.73); QTcFRD at 473 ms, Ac = 88.89%, Sn = 89.13% (95%CI = 84.37-92.84), Sp =88.70% (95%CI = 84.50-92.09). We found a significant difference (p-value = 0.0020) between area under the ROC of the QTcRTH (0.9433) and QTcBZT (0.9225) but not QTcFRD (0.9338). The Ac, Sn, and Sp of the QT nomogram were 89.08%, 91.30% (95%CI = 86.89-94.61), and 87.33% (95%CI = 82.96-90.92), respectively, and they were all equal to those of QTcRTH. CONCLUSION: Rautaharju method not only produced minimal errors for QT interval correction but also at QTcRTH 477 ms, it could predict TdP as accurately as QT nomogram and was better than the QTcBZT.
INTRODUCTION: New QT correction formulae derived from large populations are available such as Rautaharju's [QTcRTH = QT * (120 + HR)/180] and Dmitrienko's [QTcDMT = QT/RR0.413]. These formulae were derived from 57,595 and 13,039 cases, respectively. Recently, a study has shown that they did not experience errors across a wide range of heart rates compared to others. OBJECTIVES: (1) To determine the best cut-off value of QTcRTH and QTcDMT as a predictor of torsade de pointes (TdP) and (2) to compare the sensitivity and specificity using the cut-off value of QTcRTH with those of the QTcBazett (QTcBZT), QTcFridericia (QTcFRD), and QT nomogram. METHODS: Data were derived from two data sets. All cases aged over 18 years with an exposure to QT-prolonging drugs. Group-1, all cases developed TdP. Data in Group-1 were obtained from systematic review of reported cases from Medline since its establishment until 10 December 2015. Group-2 is composed of those who overdosed on QT prolonging drugs but did not develop TdP. This data set was previously extracted from a chart review of three medical centers from January 2008 to December 2010. Data from both groups were used to calculate QTcRTH and QTcDMT. The cut-off values from QTcRTH and QTcDMT that provided the best sensitivity and specificity to predict TdP were then selected. The same method was applied to find those values from QTcBZT, QTcFRD, and QT nomogram. The receiver operating characteristic curve (ROC) was applied where appropriate. RESULTS: Group-1, 230 cases of drug-induced TdP were included from the systematic review of Medline. Group-2 (control group), which did not develop TdP, consisted of 292 cases. After applying all of the correction methods to the two datasets, the best cut-off values that provided the best accuracy (Ac) with the best sensitivity (Sn) and specificity (Sp) for each formula were as follows: QTcRTH at 477 milliseconds (ms), Ac = 89.08%, Sn = 91.30% (95%CI = 86.89-94.61), Sp = 87.33%(95%CI = 82.96-90.92); QTcDMT at 475 ms, Ac = 88.31%, Sn =91.30% (95%CI = 86.89-94.61), Sp = 85.96%(95%CI = 81.44-89.73); QTcBZT at 490 ms, Ac = 86.97%, Sn = 88.26% (95%CI = 83.38-92.12), Sp = 85.96% (95%CI = 81.44-89.73); QTcFRD at 473 ms, Ac = 88.89%, Sn = 89.13% (95%CI = 84.37-92.84), Sp =88.70% (95%CI = 84.50-92.09). We found a significant difference (p-value = 0.0020) between area under the ROC of the QTcRTH (0.9433) and QTcBZT (0.9225) but not QTcFRD (0.9338). The Ac, Sn, and Sp of the QT nomogram were 89.08%, 91.30% (95%CI = 86.89-94.61), and 87.33% (95%CI = 82.96-90.92), respectively, and they were all equal to those of QTcRTH. CONCLUSION:Rautaharju method not only produced minimal errors for QT interval correction but also at QTcRTH 477 ms, it could predict TdP as accurately as QT nomogram and was better than the QTcBZT.
Authors: Sharan L Campleman; Jeffery Brent; Anthony F Pizon; Joshua Shulman; Paul Wax; Alex F Manini Journal: Clin Toxicol (Phila) Date: 2020-04-07 Impact factor: 4.467