Atrial fibrillation (AF) is the most frequently diagnosed sustained tachyarrhythmia. In the United States, a prevalence ranging from 2.7–6.1 million in 2010 to a predicted value of 12.1 million by 2030 has been estimated.1 In Europe, AF affects 8 million people and it is expected an increase of up to 18 million in the year 2060.2 The frequent asymptomatic character of this arrhythmia suggests that these figures would be considerably higher. A targeted approach to prevention based on detection of atrial electrical vulnerability is a reasonable strategy. The electrocardiogram has become a valuable tool for this purpose.

A careful observation of the electrocardiogram allows to realize that the duration of the atrial depolarization manifested as a P wave is not the same in the different recording leads, the phenomenon occurs in both healthy and sick people (with cardiovascular or other pathologies). Based on this fact, in the late nineties Dilaveris et al. proposed that the interleads differences in terms of P wave duration could be expressed through a parameter they called P wave dispersion (PWD), which could predict paroxysmal AF in the absence of comorbidities.3

Multiple investigations demonstrate the utility of PWD in predicting AF (Table 1).3-15 In the pioneering study by Dilaveris et al. a cutoff value of 40ms for PWD separated patients with paroxysmal AF with a sensitivity of 83% and specificity of 85%.3 Yoshizawa et al. in a case-control study showed that PWD was an independent predictor of new-onset AF with values that significantly exceeded those observed in controls (59.6±14.8ms vs. 33.5±12.9ms; P=.001).6 One study found that PWD correctly discriminated AF recurrences in new-onset AF patients undergoing electrical cardioversion, starting at a cutoff point >80.5ms,7 a finding very similar to that found in another investigation.8 Chang et al. found increased PWD values in patients with lone paroxysmal AF compared to controls. 16

Studying the role of inflammation in patients with paroxysmal AF, increases in ultrasensitive C-reactive protein have been found together with increased PWD values, demonstrating that both variables function as independent predictors of the aforesaid arrhythmia.9

Kemal et al. analyzed several parameters of the P wave in 145 patients undergoing electrophysiological studies with normal results and who were divided into 2 groups according to the occurrence or not of AF induced by catheters. The PWD was significantly higher in the group where AF occurred; in the logistic regression analysis this parameter was an independent predictor of AF, a cutoff value of 35ms determined catheter-induced AF with 80% sensitivity and 71% specificity.10 After AF ablation in 126 consecutive patients, Fujimoto et al. observed a significant decrease in PWD at the third and sixth month of proceeding together with the reduction in left atrial diameter and volume, so the authors concluded that PWD could be useful for the evaluation of reverse remodeling after catheter ablation of the AF.17 Walters et al. analyzed the effects on left atrial remodeling prospectively experienced by 38 patients with paroxysmal AF managed medically (group 1) and 20 undergoing AF ablation (group 2). In group 1, deterioration in total left atrial strain and an increase in PWD were observed, while in group 2 there was an improvement in total left atrial strain and reduction in PWD.18 In another study, the use of AF ablation by means of a hybrid technique (catheter and surgical ablation) significantly reduced PWD compared to other techniques.19

PWD can help discriminate patients with different types of diseases prone to developing AF in their lifetime.20 Several examples are presented below.

Different normal values of PWD have been published, observing a wide variability among them. Analyzing 295 patients randomly selected from the original cohort of the Framingham study with a wide age range and without cardiovascular disease or hypertension, diabetes mellitus or obesity, Magnani et al. determined that the median of the PWD was 34ms with a range from the 25–75 percentile of 28–43ms.55 Pérez Riera et al. consider that the normal value is between 20–38ms with an average of 29ms,53 almost equal to that reported in the control group used by Dilaveris et al. that presented a range of 21–35ms (mean, 28ms).3 A large study that included 1353 healthy young men on service in the Greek Air Force observed an average PWD value equal to 38.4ms with a range between 13–80ms.56 In 150 controls with no history of paroxysmal AF, Chang et al. observed that the PWD presented a median of 46.0ms with an interquartile range of 38.5–57.7ms.16

Recently O’Neill et al. reported reference values for PWD in Southeast Asian women and men (women, 22±12ms; men, 25±12ms) and in Caucasians (women, 24±12ms; men, 28±12ms).57 Carmona et al. in their control group observed that the mean value of PWD was 25ms with a standard deviation of 8.61ms,22 very similar to that observed by Scavenius et al. who obtained a value of 27±13ms.58 In 143 healthy volunteers with an average age of 59.17±10.87 years who presented an ergometric test and normal coronary tomography, the PWD was 34.8±12.54ms.34 In 125 healthy controls belonging to a cohort of the REDISCOVER study and with an average age of 51.3±13.6 years, it was observed that the PWD had a mean value of 54.63±21.9ms.21

After analyzing a Cuban pediatric population, Chávez et al. determined that the 95th percentile of the PWD was at 50ms, a value that was similar in both sexes.59 In 232 children between the ages of 7 and 15, Köse et al. found no differences between the sexes for PWD (female 26.8±6.2ms vs. male 27.2±5.3ms; P=.643).60

PWD is determined by subtracting the minimum P wave duration from the maximum P wave duration, taking into account the 12 leads of the electrocardiogram (Fig. 1). Its measurement has not yet been standardized,53 but different methods for obtaining it are recognized, all well accepted, although the different authors highlight some methods over others. PWD have been calculated by measurements on paper or digitized Images61 (with optical scanner or digital records presented by special software) using hand-held calipers, but there are also automated measures. The onset of the P wave is defined as the point of first detectable upward or downward slope from the isoelectric line for positive or negative waveforms, respectively. The return of P wave deflection to baseline is considered the P wave offset.

Fig. 1.

Panels A and B show identical electrocardiographic traces taken from a healthy patient to exemplify the PWD measurement technique using a 12-lead simultaneous recording and hand-held caliper. Pmax and Pmin were measured in the same beat to avoid the phenomenon of P wave lability over time. The measurements were made with a sweep speed of 50mm/s and amplification of 20mm/mV. Pmax, maximum P wave duration; Pmin, minimum P duration; PWD, P wave dispersion.

Digital measurements stand out above manual made on paper with a reduction of the relative intraobserver error from 24% to 13% and of the interobserver error from 30% to 14%61 with very similar results reported by others.62 Evaluating a cohort of the Framingham, Magnani et al. determined that the intraclass intraobserver and interobserver coefficient for the measurement of PWD was 0.82 and 0.70, respectively, concluding that it was excellent for the first and good for the second.63 Dilaveris et al. consider that the simultaneous recording of 12 leads is mandatory to avoid the phenomenon of P wave lability over time.64 Taking the measurement with a paper velocity of 50mm/s and an amplitude of 20mm/mV has become the trend for many researchers, but additionally magnifications can be performed. Some studies have used a minimum of 8 to 9 leads as an inclusion criterion for calculating the PWD.61

Although this parameter has been available for more than 20 years for clinical use, the factors that give rise to it have not yet been fully defined. Two main theories have been proposed to explain the origin of PWD. The first of these is based on the inhomogeneous atrial propagation of the electrical impulse, which would lead to temporal discrepancies in the stimulus conduction between the different areas of the atria, that is, areas with slower or faster conduction.3,20,41,53 The theory of inhomogeneous conduction (local theory) maintains that the leads of the electrocardiogram, when arranged in such a way that they capture the electrical activity of the heart from 12 different sites, might be able to record P waves of different durations depending on the atrial area explored. Those regions of the atria with slower conduction zones would be responsible for longer duration P waves while the opposite would happen if the zone explored by a given lead presented faster conduction, these duration differences would give rise to the PWD phenomenon.

Using 2-dimensional strain rate in patients with idiopathic dilated cardiomyopathy, Badran et al. found that PWD is directly correlated in these patients (r=0.45; P<.00001) with the quantification of the electromechanical delay of the left atrium, but this result was not found in the control group.65 In addition, higher PWD values have been found and right and left interatrial and intra-atrial electromechanical delays measured by 2-dimensional strain rate in non-dipper hypertensive patients compared to dipper.66 Demir et al. observed that in patients with type 2 diabetes mellitus PWD was directly and significantly correlated (r=0.428; P<.001) with the intraleft atrial electromechanical delay interval estimated by tissue Doppler.67 In hypertensive patients without a history of AF the PWD was an independent predictor of intra left atrial conduction times (B=0.306; P=.04) and interatrial conduction times (B=0.483; P=.05) measured by tissue Doppler, in addition there was a positive correlations between interatrial (P=.722), intra right (P=.643) and intra left (P=.722) atrial conduction times and PWD.68 Ermis et al. also found that PWD was significantly correlated with intra and interatrial conduction times in prehypertensive patients, using the same measurement method.69

The second theory (global, projection or vectorial theory) has hardly been mentioned. It is based on the basic principles of vectorial electrocardiography, which consists in projecting the atrial depolarization vector (of X spatial direction) on the axis system that has the leads of the frontal and horizontal plane according to standardized configurations.70 Applying basic principles of representation of coordinates in 2-dimensional planes we have that, without neglecting the magnitude of a vector, the more parallel its direction is with respect to a coordinate axis, its projection will be completely or almost complete on it, which for an electrocardiographic leads would be a P wave of maximum or almost maximum voltage and inscribed throughout its duration without parts of that wave manifesting.

At the other extreme, if a given vector of atrial depolarization followed a direction perpendicular to any coordinate axis, its projection on it would be impossible, and the equivalent in electrocardiography is the lack of inscription of the vector in the lead that represents the coordinate axis.71 We must emphasize that, although in practice it is recommended to determine the average vector, there are actually multiple instantaneous vectors72 that are manifested as the atrial depolarization proceeds. In this way each of the vectors that are generated (instantaneous vectors) can have different directions in time (± parallel or ± perpendicular) and therefore different projection relationships with the different leads of the electrocardiogram, since each of these vectors is common to all leads at any given time. An instantaneous vector would have a global effect on all leads. Because the P wave in any given lead is registering as the electrocardiograph can detect the multiple vectors that make it up, there will be P waves of greater or less duration if in certain leads there are more or less perpendicular projections of a common vector during certain fractions of time. This tends to manifest mainly at the beginning and end of atrial depolarization, where the generated potentials have lower magnitudes and are more likely not to be reflected in the P wave inscription process. As a result, isoelectric or very low amplitude intervals would occur, only detectable at high magnifications.

The consequences of this phenomenon were highlighted by Zimmer et al., who measured the PWD in 94 patients with 2 different magnifications, the first at paper speed of 50mm/s and amplification×8, the second at 200mm/s and magnification×128–256.73 In the first case, a mean value of PWD equal to 45.1ms was obtained while at maximum amplification the mean was 1.2ms. However, to our knowledge, there is only this work demonstrating the effect of the hidden portions of the P wave as a plausible and alternative explanation of the PWD phenomenon.

PWD has consistently been shown to predict the occurrence of AF. Its use has been extended to different clinical settings in which it would be desirable to detect vulnerable individuals to the development of AF in the future. PWD is considered by most researchers to be the result of inhomogeneous conduction of the atrial electrical impulse, but a vectorial explanation may play a role in its genesis. The large body of evidence surrounding PWD supports its use in clinical practice.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profict sectors.

Authors declare no conflict of interest related to this article.