We analysed the effect of heart rate and body position on the complexity of the QRS and T wave quantified by the ratio of 2 nd /1 st eigenvalue from principal component analysis (PCA) (QRS-PCA, T-PCA) . In both positions, the intrasubject variability of QRS-PCA and T-PCA was significantly smaller than the inter-subject variability.
IntroductionPrincipal component analysis (PCA) quantifies the complexity of the ECG waves by defining a set of independent forms (components) with decreasing relative value, which can fully describe their shape. Visibly more complex QRS or T waves with irregularities, notches, etc., which are frequently observed in diseased hearts are described by a greater number of independent components, with increasing relative value of the smaller components.Whilst the diagnostic [1-3] and prognostic [4,5] value of PCA of the T wave has been demonstrated, the effect of physiological factors on the QRS and T wave complexity is unknown. The T wave shape or polarity can be influenced by age, sex, heart rate, body position, autonomic activity, respiration, temperature, electrolyte concentration, food and mental activity [6][7][8][9][10]. It is possible that these factors can also affect T wave complexity. We investigated the effect of heart rate and body position on QRS and T wave complexity using an ECG database recorded in healthy subjects.
MethodsA digital ECG database was previously recoded for another project in 15 subjects (8 men, 7 women, age 28.6±7.5 years, range 21.5 -49.1 years) with negative medical history, normal physical examination and normal resting 12-lead ECG, who were not taking any medications during the project.Data were acquired after 5 minutes of rest in the supine position. In each subject, a continuous digital 12-lead ECG was recorded for 25 minutes (5 minutes supine → 10 minutes standing → 10 minutes supine) with Mason-Likar electrode positions [11] using a PC-based 12-lead ECG recorder (CardioSoft™, GE Medical Systems, Milwaukee, USA, 500 Hz, 4.88 µV resolution). In each individual, the same protocol was repeated after approximately 1 week at the same time of the day (within±1 hour). The local Ethics Committee approved the study protocol and each subject provided a written consent.
Data analysisAll ECG recordings were visually assessed on screen using the CardioSoft™ software programme in order to exclude segments with noise or premature beats. The first 2 minutes and the last 30 seconds of recording in each position were excluded from the study, and 2.5 minutes supine → 7.5 minutes standing → 7.5 minutes supine were analysed. The ECG were preprocessed suppressing power-line interference, electromyogram noise and baseline drift according to our previously published material on Q-onset and T-end delineation [12].QRS detection was performed following the work of Christov [13], then one complex lead was synthesised from the initials 8 leads [14]. The QRS and the T wave boundaries as well as the RR interval were delineated and