Improving corrected QT; Why individual correction is not enough

The use of QT-prolongation as a biomarker for arrhythmia risk requires that researchers correct the QT-interval (QT) to control for the influence of heart rate (HR). QT correction methods can vary but most used are the universal correction methods, such as Bazett's or Van de Water's, which use a single correction formula to correct QT-intervals in all the subjects of a study. Such methods fail to account for differences in the QT/HR relationship between subjects or over time, instead relying on the assumption that this relationship is consistent. To address these changes in rate relationships, we test the effectiveness of linear and non-linear individual correction methods. We hypothesize that individual correction methods that account for additional influences on the rate relationship will result in more effective and consistent correction. To increase the scope of this study we use bootstrap sampling on ECG recordings from non-human primates and beagle canines dosed with vehicle control. We then compare linear and non-linear individual correction methods through their ability to reduce HR correlation and standard deviation of corrected QT values. From these results, we conclude that individual correction methods based on post-treatment data are most effective with the linear methods being the best option for most cases in both primates and canines. We also conclude that the non-linear methods are more effective in canines than primates and that accounting for light status can improve correction while examining the data from the light periods separately. Individual correction requires careful consideration of inter-subject and intra-subject variabilities.