Continuing Medical Education Activity in Academic Emergency Medicine

Transcription

1 CONTINUING MEDICAL EDUCATION Continuing Medical Education Activity in Academic Emergency Medicine CME Editor: Corey Heitz, MD Authors: Ingrid Berling, and Geoffrey K. Isbister, MBBS, MD Article Title: The Half RR Rule: A Poor Rule of Thumb and Not a Risk Assessment Tool for QT Interval Prolongation If you wish to receive free CME credit for this activity, please refer to the website: Accreditation and Designation Statement: Blackwell Futura Media Services designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditÔ. Physicians should only claim credit commensurate with the extent of their participation in the activity. Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. Educational Objectives After reading the article, participants should be able to discuss the accuracy of measurements to estimate the QT interval. Activity Disclosures GKI is supported by an NHMRC Senior Research Fellowship This was supported by an NHMRC Program Grant Faculty Disclosures: CME Editor: Corey Heitz, MD has no relevant financial relationships to disclose. Authors: Ingrid Berling, and Geoffrey K. Isbister, MBBS, MD have no relevant financial relationships to disclose. This manuscript underwent peer review in line with the standards of editorial integrity and publication ethics maintained by Academic Emergency Medicine. The peer reviewers have no relevant financial relationships. The peer review process for Academic Emergency Medicine is double-blinded. As such, the identities of the reviewers are not disclosed in line with the standard accepted practices of medical journal peer review. Conflicts of interest have been identified and resolved in accordance with Blackwell Futura Media Services s Policy on Activity Disclosure and Conflict of Interest. Instructions on Receiving Free CME Credit For information on applicability and acceptance of CME credit for this activity, please consult your professional licensing board. This activity is designed to be completed within an hour; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period, which is up to two years from initial publication. Follow these steps to earn credit: Log on to Read the target audience, educational objectives, and activity disclosures. Read the article in print or online format. Reflect on the article. Access the CME Exam, and choose the best answer to each question. Complete the required evaluation component of the activity. This activity will be available for CME credit for twelve months following its publication date. At that time, it will be reviewed and potentially updated and extended for an additional twelve months.

2 ORIGINAL CONTRIBUTION The Half RR Rule: A Poor Rule of Thumb and Not a Risk Assessment Tool for QT Interval Prolongation Ingrid Berling, and Geoffrey K. Isbister, MBBS, MD Abstract Objectives: Measuring the QT interval on an electrocardiogram (ECG) is integral to risk assessment of Torsade de Pointes (TdP). This study aimed to investigate the accuracy of the 1/2 RR rule as a risk assessment tool for drug-induced TdP, comparing it to the QT nomogram, Bazett s corrected QT (QTcB), and Fridericia s corrected QT (QTcF). Methods: The authors calculated sensitivity and specificity of the 1/2 RR rule using a published data set of 129 cases of drug-induced TdP and 316 controls (noncardiotoxic overdoses), compared to the QT nomogram, QTcB > 500 msec and QTcF > 500 msec. To further determine the value of the 1/2 RR rule, its observed positive, and negative agreement were calculated when compared to the QT nomogram for determining an abnormal QT in eight samples of different drugs in overdose. Results: The sensitivity and specificity of the 1/2 RR rule were 88% (95% confidence interval [CI] = 80% to 93%) and 53% (95% CI = 47% to 58%), respectively, compared to the QT nomogram (sensitivity = 97%, 95% CI = 92% to 99%; specificity = 99%, 95% CI = 97% to 100%). It was also less sensitive than QTcB > 500 msec and had a lower specificity than QTcB > 500 msec and QTcF > 500 msec. In drug overdose patients, the 1/2 RR rule had poor observed agreement averaging 41%, which was mainly due to poor positive agreement, except for amisulpride where there was good agreement. Conclusions: The 1/2 RR rule was not as sensitive as the QT nomogram or QTcB > 500 msec for druginduced TdP. It had poor positive agreement in almost all overdose patients, resulting in over half of patients receiving unnecessary cardiac monitoring and repeat ECGs. ACADEMIC EMERGENCY MEDICINE 2015;22: by the Society for Academic Emergency Medicine Torsade de Pointes (TdP) is a rare but potentially lethal cardiac arrhythmia that is associated with a prolonged QT interval. QT prolongation and the risk of TdP is well recognized in rare congenital long QT syndromes, treated mainly by cardiac electrophysiologists. More commonly, QT prolongation is associated with medications, predominately in drug overdose. Although TdP only occurs in a small proportion of cases, the assessment of patients at risk is important for emergency and critical care physicians, as well as medical toxicologists. Predicting the risk of a patient developing TdP is difficult, but because TdP is always associated with and therefore defined by the presence of a long QT interval, current risk assessment focuses on the length of the QT interval. For this reason, measurement of the QT interval on the electrocardiogram (ECG) is part of the routine assessment of any patient taking medications associated with TdP, whether due to drug interactions or drug toxicity. Numerous methods are suggested and used for measuring the QT interval, correcting for the heart rate and determining what is a long QT interval, so as to determine who is at risk. 1 Methods of measurement of the QT interval are problematic but have been reviewed elsewhere. 1 3 The use of heart rate correction formulae and determining the value of the QT that is abnormal is From the Department of Clinical Toxicology and Pharmacology, Calvary Mater Newcastle (IB, GKI); and the Clinical Toxicology Research Group, University of Newcastle (IB, GKI), Newcastle, NSW, Australia. Received March 17, 2015; revision received May 6, 2015; accepted May 9, GKI is supported by an NHMRC Senior Research Fellowship This was supported by an NHMRC Program Grant The authors have no potential conflicts to disclose. Supervising Editor: Kennon Heard, MD, PhD. Address for correspondence and reprints: Prof. Geoff Isbister; Geoff.Isbister@gmail.com by the Society for Academic Emergency Medicine ISSN doi: /acem PII ISSN

3 1140 Berling and Isbister QT PROLONGATION AND AGREEMENT BETWEEN ASSESSMENT TOOLS the other major problem in the risk assessment of druginduced QT prolongation and for assessment of long QT syndrome. A number of heart rate correction formulae are used to correct the QT with various QTc cutoffs to indicate risk. The most commonly used is Bazett s correction formula (QTcB) with cutoffs ranging from 440 to 500 msec. However, it overcorrects for fast heart rate and undercorrects for slow heart rate. 4 Fridericia s (QTcF) is also increasingly used, but has similar problems. Recently the QT nomogram has been shown to be more accurate than correction formulae for risk assessment, based on an evaluation in a sample of drug-induced TdP cases compared to noncardiotoxic overdose controls. 2,5 The QT nomogram avoids the use of correction formulae and provides a simple graphical method of determining if the QT is abnormal. The QT nomogram line is dotted at heart rates beyond 105 beats/min to represent a mathematically extrapolated line of risk, as in its original evaluation the majority of cases of TdP were below 100 beats/min. 5 However, the risk of TdP at heart rates over 105 beats/min is less clear and possibility protective. Another popular method, because of its simplicity, is the 1/2 RR rule where there is no heart rate correction and an abnormal QT is defined as a QT greater than half of the RR interval. 6 Unfortunately, its simplicity is likely to make it inaccurate, and we hypothesized that the 1/2 RR rule would have poor sensitivity and poor specificity. Routinely in emergency departments (EDs), on Internet blogs sites, and in cardiology texts, the 1/2 RR rule is advocated as a rule of thumb for rapid assessment of the QT interval to determine if there is QT prolongation However, if the rule has poor sensitivity, then it fails as a useful screening tool. There are no studies supporting the diagnostic accuracy of the 1/2 RR rule for risk assessment of drug-induced QT prolongation, despite its increasing and widespread use. It is concerning that a commonly advocated and clinically applied risk assessment rule is not based on evidence, and this should be challenged. The aim of this study was to determine the sensitivity of the 1/2 RR rule as a screening test using a previously developed sample of patients with QT heart rate pairs from documented drug-induced TdP compared to controls and comparing it with the QT nomogram and other QT risk cutoffs. Secondly, the study aimed to determine the positive and negative agreement between the 1/2 RR rule and the QT nomogram in eight samples of drug overdose patients. METHODS This study consisted of two parts. The first was to investigate the sensitivity of the 1/2 RR rule and compare this to other methods of risk assessment. The second was to investigate the observed agreement and in particular the positive agreement compared to the QT nomogram. Ethics approval to use the data extracted from noncardiotoxic drug overdose presentations and the psychotropic drug overdoses was provided by the local human research ethics committee. Determination of the Sensitivity and Specificity of the 1/2 RR Rule for Cases of Drug-induced TdP Study Design and Setting. We calculated the sensitivity and specificity of the 1/2 RR rule using previously collected observational QT heart rate data from 129 cases of drug-induced TdP as positive controls and 316 QT heart rate pairs from an overdose dataset of noncardiotoxic drugs (paracetamol, diazepam, oxazepam, and temazepam) as negative controls. 5 The methods of collection of this sample of QT heart rate pairs are published elsewhere, 5 but include a systematic review of the literature identifying all drug-induced episodes of TdP with documented QT intervals and heart rates. The QT intervals in the control patients were manually measured by trained personnel and the heart rates taken from the ECGs. The data set was previously used to evaluate/validate the QT nomogram and was not used to derive the QT nomogram. 5 Measurements. The 1/2 RR rule defines a QT interval as abnormal (positive outcome) if the measured QT is greater than half the RR interval for the same ECG. This can be shown visually as a plot above the line of abnormality on a graph of QT versus heart rate by plotting QT = RR/2 versus heart rate (Figure 1). The QT nomogram defines a QT interval as abnormal if the QT heart rate pair from an ECG is above the abnormal line on the QT nomogram. 2 The line of abnormality for the QT nomogram was derived from a study by Fossa et al. 11 and has been previously evaluated by Chan et al. 5 We also considered risk assessment approaches using Bazett s and Fridericia s correction formulae where an abnormal QT was defined as greater than 500 msec i.e. QTcB > 500 msec and QTcF > 500 msec (Figure 1). Data Analysis. The sensitivity and specificity analyses are shown visually on a QT versus heart rate plot (Figure 1) and calculated using QT = RR/2 to determine abnormality. The sensitivity was calculated by counting the number of QT heart rate pairs for cases of TdP that were above the QT > RR/2 line (positive outcomes) and dividing it by the total number of TdP cases (positive cases; 129). The specificity was calculated by counting the number of QT heart rate pairs for controls (noncardiotoxic overdose cases) that were below the QT < RR/2 line (negative outcomes), and dividing this by the total number of controls (negative cases; 316; Data Supplement S1, available as supporting information in the online version of this paper). Determination of the Overall and Positive Agreement of the 1/2 RR rule in Determining an Abnormal QT Interval Study Design and Setting. This part of the study was a comparison of risk assessment tools for determining if a QT interval is abnormal and is analogous to comparing diagnostic tests. 12,13 Because the QT nomogram is not the criterion standard, it is an imperfect reference test; the 1/2 RR rule was then compared to the QT nomogram by calculating the observed agreement between tests, the negative agreement, and the positive agreement. Our focus was on the positive agreement as

4 ACADEMIC EMERGENCY MEDICINE October 2015, Vol. 22, No Figure 1. Plots of QT versus heart rate on the QT nomogram showing a visual representation of the line of abnormality for the QT nomogram (A, black line), 1/2 RR rule (B, green line and dashed for heart rate < 60 beats/min), QTcB > 500 msec (C, dark blue line) and QTcF > 500 msec (D, dark red line) comparing positive cases of drug-induced Torsade de Pointes (red filled circles) and control cases (filled blue circles), as a comparison of the four methods in determining an abnormal QT interval. QTcF = Fridericia s corrected QT. a measure of misclassification of normal cases as abnormal and therefore unnecessary monitoring and intervention being required in these patients. The observed, positive, and negative agreement of the 1/2 RR rule in assessing an abnormal QT interval were calculated in eight different sample sets of QT heart rate pairs from overdose admissions. These were overdoses of different psychotropic medications recruited to a large regional toxicology service that have been previously published. 5,14 20 The eight different drugs were amisulpride, thioridazine, citalopram, quetiapine, venlafaxine, mirtazapine, oxycodone, and risperidone. Measurements. The QT intervals were all measured by trained personnel using a previously published method. 2 In brief, six leads were measured including three chest leads (most commonly V2, V4, V6) and three limb leads (most commonly I, II, and avf or avl). The median of the six individual lead measurements of the QT was then calculated. Disagreement in any measurement of the QT interval of more than 20 msec resulted in review by a third party. 20 If there was more than one ECG during an admission, the ECG with the longest QT was used. The heart rate was taken from the ECG machine for the corresponding QT. Data Analysis. The agreement between the two tests can be seen visually on a QT versus heart rate plot (Figure 2). The positive agreement and negative agreement were calculated according to Feinstein and Cicchetti. 13 Positive agreement = 2a/(2a + b + c), and negative agreement = 2d/(2d + b + c), where a is true-positives, b is false-negatives, c is false-positives, and d is true-negatives. The 1/2 RR rule was compared to the imperfect QT nomogram, so the true-positives (a) were the number of abnormal QT heart rate pairs according to the 1/ 2 RR rule and the QT nomogram, the false-negatives (b) were the number of QT heart rate pairs classified as normal by the 1/2 RR rule that were abnormal according to the QT nomogram, the false-positives (c) were the number of QT heart rate pairs classified as abnormal by the 1/2 RR rule that were normal according to the QT nomogram, and the true-negatives (d) were the number of QT heart rate pairs that were not determined to be abnormal by the 1/2 RR rule and the QT nomogram. Finally, the observed agreement was the number of QT heart rate pairs classified by the 1/2 RR rule as abnormal (true-positives) plus the number classified as normal (true-negatives) divided by the total number of cases. Cohen s kappa was also calculated, which is (p o p e )/1 p e, where p o is the proportion of observed agreement and p e is the proportion of expected agreement by chance. For the first study part, the 95% confidence intervals (CIs) for the sensitivity and specificity were calculated using the Wilson s procedure with a continuity correction. 21 For the second part, the 95% CIs for the positive agreement and negative agreement were calculated from the asymptotic standard errors with the delta method as described by Graham and Bull, 22 using

5 1142 Berling and Isbister QT PROLONGATION AND AGREEMENT BETWEEN ASSESSMENT TOOLS Figure 2. QT versus heart rate plots for all of the QT heart rate pairs showing the large portion of patients under the QT nomogram, but above the 1/2 RR rule line, for amisulpride, thioridazine, citalopram, quetiapine, venlafaxine, mirtazapine, oxycodone, and risperidone. DAG_Stat by Mackinnon. 23 Cohen s kappa was calculated using Dag_Stat. All statistical and graphical analyses were done in GraphPad Prism version 6 for Windows. RESULTS The number of correctly identified cases of TdP and control cases for each of the four rules are shown in Figure 1 and included in Data Supplement S1. The sensitivity and specificity of the 1/2 RR rule were 87.6% (95% CI = 80.4% to 92.5%) and 52.9% (95% CI = 47.2% to 58.4%), respectively (see Table 1). This is compared to the sensitivity and specificity of the QT nomogram which were 96.9% (95% CI = 91.8% to 99%) and 98.7% (95% CI = 96.6% to 99.6%), respectively. When using the 1/2 RR rule in cases only with heart rates equal to or above 60 beats/min, the sensitivity and specificity improved to 100% (95% CI = 94.6% to 100%) and 49.7% (95% CI = 43.8% to 55.5%), respectively. The sensitivity Table 1 The Sensitivity and Specificity With 95% CIs of the 1/2 RR Rule Compared to the Three Other Methods for Drug-induced TdP Compared to Noncardiotoxic Overdoses Method Sensitivity, % Specificity, % 1/2 RR rule 87.6 ( ) 52.9 ( ) QT nomogram 96.9 ( ) 98.7 ( ) QTcB > 500 msec 93.8 ( ) 97.2 ( ) QTcF > 500 msec 82.2 ( ) 100 ( ) 1/2 RR rule 60 beats/min 100 ( ) 49.7 ( ) Some data from Chan et al. 5 QTcB = Bazett s corrected QT; QTcF = Fridericia s corrected QT; TdP = Torsade de Pointes. and specificity of the QTcB and QTcF being greater than 500 msec are also shown in Table 1. The observed, negative, positive agreement, and Cohen s kappa of the 1/2 RR rule compared to the QT nomogram for abnormal QT intervals in the eight cohorts of different drug overdoses are shown in Table 2. The 1/2 RR rule had poor positive agreement (Table 2 and Data Supplement S2, available as supporting information in the online version of this paper). Figure 2 provides plots of all the QT heart rate pairs for each drug, showing the large portion of patients under the QT nomogram, but above the 1/2 RR rule line. There were 89 cases of amisulpride overdoses, of which 67 (75%) were above the QT nomogram, but 81 (91%) were positive according to the 1/2 RR rule. There were similarly large numbers of abnormal QTs based on the 1/2 RR rule compared to the QT nomogram for thioridazine (50 vs. 16 of 57 total), citalopram (219 vs. 32 of 311 total), quetiapine (166 vs. 10 of 202 total), venlafaxine (269 vs. 22 from 369 total), mirtazapine (51 vs. 0 from 81 total), oxycodone (32 vs. 5 from 50 total), and risperidone (34 vs. 5 from 41 total). DISCUSSION This study shows that the 1/2 RR rule underperforms in comparison to the QT nomogram and more traditional methods using heart rate correction and a cutoff of 500 msec. This results in missed identification of a few patients at high risk with bradycardia and excessive concern for a large number of patients with normal to fast heart rates, but normal QT intervals. When applying the 1/2 RR rule to the eight samples of drug overdoses, it is clear that it results in misclassification of patients without QT prolongation, with the majority of patients ECGs read as abnormal. Of particular concern when looking at the positive TdP cases is the number of patients with heart rates < 60 beats/min who would be missed as at risk using the 1/2 RR rule. This flaw with

6 ACADEMIC EMERGENCY MEDICINE October 2015, Vol. 22, No Table 2 1/2 RR Rule Positive Agreement, Negative Agreement, Observed Agreement, and Cohen s Kappa Compared to the QT Nomogram for Each Drug Cohort Drug No. of Cases Positive Agreement Negative Agreement Observed Agreement Cohen s Kappa Amisulpride (86 95) 53 (32 75) 84 (75 91) 46 (24 68) Thioridazine (34 63) 29 (12 46) 40 (28 54) 10 (2 19) Citalopram (18 32) 49 (43 55) 39 (34 45) 8 (5 12) Quetiapine (5 18) 32 (24 39) 23 (17 29) 2 (1 4) Venlafaxine (9 20) 44 (39 50) 33 (28 38) 4 (2 6) Mirtazapine (43 65) 37 (27 49) Oxycodone (8 46) 57 (43 72) 46 (32 61) 12 (1 22) Risperidone (8 44) 33 (14 51) 29 (17 46) 6 (1 12) the rule has been noted in a study comparing the 1/2 RR rule to Bazett s formula 7 and makes it dangerous to be used even as a rule of thumb for screening. When the 1/2 RR rule is advocated and taught to emergency medicine trainees on the Internet and in lectures, the requirement of a heart rate > 60 beats/min is rarely mentioned. The Life in the Fast Lane (LITFL) blog states, a useful rule of thumb is that a normal QT is less than half the preceding RR interval. 24 If one does assume that the heart rate > 60 beats/min criterion can be remembered and applied when using the ½ RR rule, it would improve the sensitivity of the rule to 100% as shown in Table 1. However, this creates problems because it makes it less of a simple rule of thumb, but more problematic is that it requires another assessment tool for a significant number of drug-induced TdP cases with bradycardia or almost one third of the positive cases (41 of 129) in the data set used here. With or without the heart rate > 60 beats/min criterion, the specificity of 53% still makes the 1/2 RR rule problematic because half the time one would misclassify the patients as abnormal and/or need to measure the QT more adequately to further define risk. It would seem far more sensible to simply use the more accurate QT nomogram from the start, which is almost as simple to use and requires no calculations. The concern regarding rules such as the 1/2 RR rule being advocated on blogs and social media is the lack of peer review that accompanies such advice. Where Internet resources are reformatting previously published information for better education/dissemination, there is no problem. However, when there is little or virtually no evidence to support a statement that then becomes integrated into junior doctor learning as though it where fact, the creation of significant misinformation and poor clinical decision-making occurs. Medical toxicology is a field of limited and often difficult to interpret clinical evidence. Although often the focus of medical blogs for the interest the field generates, the quality of the discussions and experience of the person advising need to be considered. When trying to assess the diagnostic accuracy of a test, it is essential to know the true status of the diseased state in the population you are testing. However, there are few large data sets of TdP, making it difficult to test against a criterion standard. This meant that for the second part of the study we investigated the ability of the 1/2 RR rule to predict (risk-assess) drug-induced TdP occurring by using an imperfect but good reference test the QT nomogram. For this reason agreement was determined for the 1/2 RR rule compared to the QT nomogram, rather than comparing the sensitivity and specificity of the 1/2 RR rule to the QT nomogram. 12 Both the positive and the negative agreement between the two rules were poor, except for amisulpride. However, the positive agreement was worse than the negative agreement for the majority of the drugs in overdose because the RR rule classified such a large number of cases as positive, which the QT nomogram did not. Visually this is represented as the large number of points between the two lines and are patients who would be unnecessarily admitted and monitored. The positive agreement was extremely poor and shows that the 1/2 RR rule of thumb reads a significant number of cases as positive in comparison with the QT nomogram. This results in the majority of ECGs needing to be reread more formally or patients being monitored for TdP inappropriately. The observed agreement between the tests was generally poor, reflecting both the low positive and the negative agreement. This also highlights the importance of reporting both negative and positive agreements and not just overall agreement when comparing tests. LIMITATIONS This study is in part limited by the rarity of cases of TdP, thus the limited sample cases and difficulties in creating sample data to validate tests. Chan et al. 5 were only able to identify 129 cases of drug-induced TdP in the literature, so finding a large data set in prospective studies would be almost impossible. However, the tests examined in this study are not used to diagnose cases of TdP, but rather to assess patients at risk of TdP by identification of a long QT interval. The development of the positive and negative control data used to determine the sensitivity and specificity of all four rules is biased in a number of ways. The positive cases of TdP were extracted from the literature, so there is the potential for publication bias. The negative control patients are overdose patients, as per the positive control cases, but of drugs not expected to prolong the QT interval. If the rules were being tested for diagnostic accuracy of identifying episodes of TdP this would bias the specificity. However, the aim of the tests is risk assessment and the negative controls therefore need to not prolong the QT

7 1144 Berling and Isbister QT PROLONGATION AND AGREEMENT BETWEEN ASSESSMENT TOOLS interval. A control cohort of human volunteers or general ED admissions should be tested in the future to better identify the typical baseline population QT interval lengths and the interindividual variability. The measurement of the QT interval is highly variable, with multiple methods existing, but this is unlikely to bias this study as a consistent method was used and the same QT length applied to all four rules. The eight samples of cases used for comparison between the 1/2 RR rule and the QT nomogram were unlikely to be biased by knowledge of the type of drug and underlying concerned risk for TdP, as at the time of QT measurement the investigators were blinded to the drug involved. A further limitation is that the data set has been analyzed retrospectively. However, all data were collected prospectively at the time of presentation and investigators blinded to any details. CONCLUSIONS This study demonstrates that the QT nomogram is far superior to the 1/2 RR rule, and the latter should not be used in clinical practice. There is also support for the diagnostic accuracy and usefulness of the QT nomogram in previous studies. References 1. Isbister GK. How do we assess whether the QT interval is abnormal: myths, formulae and fixed opinion. Clin Toxicol (Phila) 2015: Isbister GK, Page CB. Drug induced QT prolongation: the measurement and assessment of the QT interval in clinical practice. Br J Clin Pharmacol 2013;76: Isbister GK. Risk assessment of drug-induced QT prolongation. Australian Prescriber 2015;38: Davey P. How to correct the QT interval for the effects of heart rate in clinical studies. J Pharmacol Toxicol Methods 2002;48: Chan A, Isbister GK, Kirkpatrick CM, Dufful SB. Drug-induced QT prolongation and torsades de pointes: evaluation of a QT nomogram. QJM 2007;100: Phoon CK. Mathematic validation of a shorthand rule for calculating QTc. Am J Cardiol 1998;82: Vieweg WV, Wood MA, Fernandez A, Beatty- Brooks M, Hasnain M, Pandurangi AK. Proarrhythmic risk with antipsychotic and antidepressant drugs: implications in the elderly. Drugs Aging 2009;26: Monroe Community College. Understanding the QT/ QTc Measurement. Available at: Accessed Jul 26, Grauer K. A 1st Book on ECGs. Gainesville, FL: KG/ EKG Press, Garcia T. 12-Lead ECG: The Art of Interpretation, 2nd ed. Burlington, MA: Jones & Bartlett Learning, Fossa AA, Wisialowski T, Magnano A, et al. Dynamic beat-to-beat modeling of the QT-RR interval relationship: analysis of QT prolongation during alterations of autonomic state versus human ether a-go-go-related gene inhibition. J Pharmacol Exp Ther 2005;312: Feinstein AR, Cicchetti DV. High agreement but low kappa: I. The problems of two paradoxes. J Clin Epidemiol 1990;43: Cicchetti DV, Feinstein AR. High agreement but low kappa: II. Resolving the paradoxes. J Clin Epidemiol 1990;43: Page CB, Calver LA, Isbister GK. Risperidone overdose causes extrapyramidal effects but not cardiac toxicity. J Clin Psychopharmacol 2010;30: Berling I, Isbister GK. Mirtazapine overdose is unlikely to cause major toxicity. Clin Toxicol 2014;52: Berling I, Whyte IM, Isbister GK. Oxycodone overdose causes naloxone responsive coma and QT prolongation. QJM 2013;106: Friberg LE, Isbister GK, Duffull SB. Pharmacokinetic-pharmacodynamic modelling of QT interval prolongation following citalopram overdoses. Br J Clin Pharmacol 2006;61: Isbister GK. Electrocardiogram changes and arrhythmias in venlafaxine overdose. Br J Clin Pharmacol 2009;67: Isbister GK, Murray L, John S, et al. Amisulpride deliberate self-poisoning causing severe cardiac toxicity including QT prolongation and torsades de pointes. Med J Australia 2006;184: Berling I, Isbister GK. Prolonged QT risk assessment in antipsychotic overdose using the QT nomogram. Ann Emerg Med 2015;66: Newcombe RG. Two-sided confidence intervals for the single proportion: comparison of seven methods. Stat Med 1998;17: Graham P, Bull B. Approximate standard errors and confidence intervals for indices of positive and negative agreement. J Clin Epidemiol 1998;51: Mackinnon A. A spreadsheet for the calculation of comprehensive statistics for the assessment of diagnostic tests and inter-rater agreement. Comput Biol Med 2000;30: Cadogan M, Nickson C. Life in the Fastlane. QT Interval. Available at: ecg-library/basics/qt_interval/. Accessed Jul 25, Supporting Information The following supporting information is available in the online version of this paper: Data Supplement S1. The 2 x 2 tables for each of the four different risk assessment tools, 1/2 RR rule. Data Supplement S2. The 2 x 2 tables for all eight sample drug overdose cases.