Recent heart rate history affects QT interval duration in atrial fibrillation.

QT interval prolongation is associated with a risk of polymorphic ventricular tachycardia. QT interval shortens with increasing heart rate and correction for this effect is necessary for meaningful QT interval assessment. We aim to improve current methods of correcting the QT interval during atrial fibrillation (AF). Digitized Holter recordings were analyzed from patients with AF. Models of QT interval dependence on RR intervals were tested by sorting the beats into 20 bins based on corrected RR interval and assessing ST-T variability within the bins. Signal-averaging within bins was performed to determine QT/RR dependence. Data from 30 patients (29 men, 69.3±7.3 years) were evaluated. QT behavior in AF is well described by a linear function (slope ~0.19) of steady-state corrected RR interval. Corrected RR is calculated as a combination of an exponential weight function with time-constant of 2 minutes and a smaller "immediate response" component (weight ~ 0.18). This model performs significantly (p<0.0001) better than models based on instantaneous RR interval only including Bazett and Fridericia. It also outperforms models based on shorter time-constants and other previously proposed models. This model may improve detection of repolarization delay in AF. QT response to heart rate changes in AF is similar to previously published QT dynamics during atrial pacing and in sinus rhythm.