Mili Natekar

Intra- and interreader variability in QT interval measurement by tangent and threshold methods in a central electrocardiogram laboratory

BACKGROUND The QT interval can be measured by tangent (QT(Tan)) and threshold (QT(Thr)) methods; the better method is the one with lower reader variability. METHODS QT(Tan) and QT(Thr) were measured twice in 100 digital electrocardiograms (ECGs) by 8 experienced readers in a central laboratory. For QT(Thr), the end of the T wave was the point where the T wave reached the isoelectric baseline; for QT(Tan), it was the point where a line from the peak of the T wave through the steepest part of the descending limb intercepted the isoelectric baseline. RESULTS The average absolute intrareader variability ranged from 3.4 to 6.9 milliseconds for QT(Tan) and from 3.5 to 5.2 milliseconds for QT(Thr). By analysis of variance, intrareader SD of QT(Tan) was 7.0 and 7.5 milliseconds for QT(Thr); interreader SD was 13.1 milliseconds for QT(Tan) and 11.9 milliseconds for QT(Thr). QT(Tan) was shorter than QT(Thr) in 96 of the 100 ECGs, it exceeded QT(Thr) in 4 ECGs, which had prominent U waves. CONCLUSIONS For trained readers in a central ECG laboratory using sophisticated on-screen tools for QT measurement in high-quality digital ECGs, between- and within-reader variability are comparable for QT(Tan) and QT(Thr). However, QT(Tan) is consistently shorter than QT(Thr) by up to 10 milliseconds.

Comparison of 5 methods of QT interval measurements on electrocardiograms from a thorough QT/QTc study: effect on assay sensitivity and categorical outliers

INTRODUCTION We studied moxifloxacin-induced QT prolongation and proportion of categorical QTc outliers when 5 methods of QT measurement were used to analyze electrocardiograms (ECGs) from a thorough QT study. METHODS QT interval was measured by the threshold, tangent, superimposed median beat, automated global median beat, and longest QT methods in a central ECG laboratory in 2730 digital ECGs from 39 subjects during placebo and moxifloxacin treatment. RESULTS All 5 methods were able to demonstrate statistically significant moxifloxacin-induced QTcF prolongation. However, lower bound of 95% 1-sided confidence interval of QTcF prolongation did not exceed 5 milliseconds with the longest QT method. More QTcF outliers were observed with the longest QT and tangent methods, whereas the other 3 methods were comparable. QTcF values greater than 500 milliseconds were observed only with moxifloxacin by the tangent method, and with moxifloxacin and placebo by the longest QT method. CONCLUSION The method of QT measurement must be considered when interpreting individual thorough QT/QTc studies.

Effect of Number of Replicate Electrocardiograms Recorded at Each Time Point in a Thorough QT Study on Sample Size and Study Cost

In a “thorough QT/QTc” (TQT) study, several replicate electrocardiograms (ECGs) are recorded at each time point to reduce within‐subject variability. This decreases the sample size but increases the cost of ECG analysis. To determine the most cost‐effective number of ECG replicates, the authors retrospectively analyzed data from the placebo and moxifloxacin arms of a TQT study with crossover design. Six replicate ECGs were recorded at 7 time points on day −1 (baseline day), day 1, and day 3 in 124 normal healthy volunteers who were randomized to receive moxifloxacin or placebo on day 1 and the other treatment on day 3. QT interval was corrected for heart rate by the Fridericia (QTcF) and individual subject‐specific (QTcI) formulas. Within‐subject and between‐subject standard deviations for QTcF obtained by repeated‐measures analysis of covariance were 9.5 and 13.3 milliseconds with 1 replicate; 7.8 and 12.7 milliseconds with 2 replicates; 7.3 and 12.3 milliseconds with 3 replicates; 6.9 and 12.2 milliseconds with 4 replicates; 6.8 and 11.9 milliseconds with 5 replicates; and 6.6 and 11.8 milliseconds with 6 replicates. Within‐ and between‐subject variance with QTcI also declined with increasing replicates. Sample size benefit based on these estimates was negligible beyond 4 replicates. The study cost was least with 3 or 4 replicates, depending on per‐ECG and per‐subject costs.

QTc interval and its variability in patients with schizophrenia and healthy subjects: implications for a thorough QT study

We compared heart rate-corrected QT interval (QTc) and its within- and between-subject variability, in ECGs recorded several days apart for 207 patients with schizophrenia (age range 19-60 yr) with age- and gender-matched healthy controls. Patients had higher heart rates (mean±s.d.) than controls [75±15 beats per minute (bpm) vs. 63±10 bpm; p<0.0001]. QTc by Bazetts formula (QTcB) overestimated QTc interval at high heart rates; consequently QTcB was longer in patients (412±24 ms) than in controls (404±24 ms; p=0.0003). QTc by Fridericias method (QTcF), which was not influenced by heart rate, was comparable (398±22 ms in patients vs. 401±19 ms in controls; p=0.17). Between-subject variability in QTcF was similar in patients (17 ms) and controls (16.2 ms) but within-subject variability was larger (13.1 ms vs. 10 ms, respectively). Thus, a larger sample size is required when thorough QTc studies with a cross-over design are performed in patients with schizophrenia than in healthy subjects; sample size is not increased for studies with a parallel design. Last, QTcF is preferred over QTcB in schizophrenia patients with higher heart rates.

Comparison of Two Methods of Estimating Reader Variability in QT Interval Measurements in Thorough QT/QTc Studies

Two methods of estimating reader variability (RV) in QT measurements between 12 readers were compared.

Baseline and new-onset morphologic ECG abnormalities in healthy volunteers in phase I studies receiving placebo: changes over a 6-week follow-up period.

Morphological ECG abnormalities occur in 5–12% healthy adults participating in early phase clinical trials. We retrospectively analyzed 16,472 12‐lead ECGs recorded at multiple time points in 420 volunteers (282 males, 138 females; aged 18–76 years) randomized to receive placebo from 19 Phase I studies to see if some baseline ECG abnormalities may disappear or new abnormalities may appear during the study. One hundred forty‐four (34.3%) subjects had abnormal baseline ECGs, of which 66 (44.8%) reverted to normal during follow‐up. Of 276 (65.7%) subjects with normal baseline ECGs, 118 (42.8%) developed ECG abnormalities over the next 6 weeks. Common baseline abnormalities included sinus bradycardia, R wave transition abnormalities, right axis deviation, non‐specific T wave changes and atrial premature complexes. On follow‐up ECGs, prolonged QT interval, first‐degree AV block, sinus bradycardia, short PR interval, and R wave transition abnormalities reverted to normal. Common new‐onset abnormalities in subjects with normal baseline ECGs included sinus bradycardia, prolonged QT interval, non‐specific T wave changes, R wave transition abnormalities, and sinus tachycardia. Thus, transient morphological ECG changes may occur in healthy volunteers possibly due to diurnal variation, effect of food, hormones, or autonomic changes. This should be considered when interpreting “treatment‐emergent” ECG changes in clinical trials.

Reader variability in QT measurement due to measurement error and variability in leads selection: A simulation study comparing 2-way vs. 3-way interaction ANOVA model

Reader variability (RV) results from measurement differences or variability in lead used for QT measurements; the latter is not reflected in conventional methods for estimating RV. Mean and SD of QT intervals in 12 leads of 100 ECGs measured twice were used to simulate data sets with inter-RV of 5, 10, 15, 20, and 25 ms and intra-RV of 3, 6, 9, 12, and 15 ms. Six hundred twenty-five data sets were simulated such that different leads were used in Read1 and Read2 in 0, 10%, 20%, 30%, 40% of ECGs by 25 readers. RV was estimated using ANOVA interaction models: three-way model using Reader, ECG and lead as factors, and 2-way model using reader and ECG as factors. Estimates from three-way model accurately matched inter- and intra-RV that were introduced during simulation regardless of percent of ECGs with lead selection variability. The two-way model provides identical estimates when both reads are in same leads, but higher, more realistically estimates when measurements are made in different leads.

Limb Lead Interchange in Thorough QT/QTc Studies

The investigators analyzed 85 133 electrocardiograms (ECGs) recorded in 484 subjects from 5 thorough QT/QTc studies (3 using Holter devices, 2 using 12‐lead ECGs) for inadvertent limb lead interchanges using a dedicated quality control process in a central ECG laboratory. Limb lead interchanges were present in 2919 (3.4%) ECGs in 17.9% of subjects and were more frequent with Holter devices (7.5% vs 0.8%, P < .0001), where leads remain connected for prolonged periods, affecting data from several time points. Left arm—left leg interchange was seen in 54% of 12‐lead ECGs and right arm—left arm interchange in 38%. The ECG device itself could identify 21.7% of interchanges, whereas experienced readers blinded to subject and visit identified 79% of interchanges; 21% of interchanges were identified only during the quality control process. If correctly identified, QT measurement could be performed in a precordial lead. If undiagnosed, incorrect QT interval measurements and morphological diagnosis may confound results.