A meta-analysis of ECG data from healthy male volunteers: diurnal and intra-subject variability, and implications for planning ECG assessments and statistical analysis in clinical pharmacology studies

Abstract

To evaluate the innate variability in key electrocardiography (ECG) parameters from clinical pharmacology studies. Meta-analysis of ECG data from seven clinical pharmacology studies in healthy male volunteers using model building and stepwise multiple regression analyses. Data from 115 male subjects provided over 2,000 observations for all key ECG parameters from baseline (Day-1) and placebo treatment periods (Day 1). Only heart rate and uncorrected QT showed clear and marked changes over the day. QTcB had greater variability compared to QTcF. 1.4% of QTcB and 0.7% QTcF observations were >430 ms and 0.1% of QTcB and 0% of QTcF observations were >450 ms. We estimated that 8.9% of subjects would have at least one out of eight post-observation QTc value in the range 430-450 ms [assuming QTc mean 385 ms, standard deviation (SD) 20 ms] due to intrinsic variability alone. Time-matched within-subjects observations demonstrated that the SD between measurements taken 1 h apart was less than seen with a longer interval, but there was little increase in variability beyond 1 h. The probability of observing an increase in QTc of 30-60 ms in a subject was estimated as 3.0% and 21.8% for one and eight post-dose observations, respectively. The greater the number of observations used to define baseline the narrower the spread; for QTcF the SD of the baseline value was 17.1 ms for a single assessment, 13.3 ms for the mean of three assessments, and 13.2 ms for the mean of all Day-1 assessments. The spontaneous variability in QTc measurements must be taken into account when designing studies and interpreting analyses of ECG data. The categorical analysis of QTc change of 30-60 ms is unlikely to be of any additional value to analyses of central tendency. For standard early clinical pharmacology studies, QTcF should be chosen as the primary correction method, while the mean of three measures taken in the afternoon and evening of Day-1 and pre-dose Day 1 should provide a reliable and representative baseline assessment.