The main purpose of pre-competitive screening of athletes is to enable early (pre-clinical) identification of pathological conditions associated with increased risk of serious clinical events, including sudden death. Data from Italy show an 89% reduction in the incidence of sudden death in competitive athletes following the inclusion of the electrocardiogram (ECG) in pre-competitive screening.1 In view of this, most European countries currently recommend that pre-competitive screening should include personal and family history, physical examination, and a resting 12-lead ECG.2

Despite this evidence and the many arguments in support of ECG assessment, the inclusion of this exam in pre-competitive screening remains controversial, basically because of disagreement between Europe and the US, where it is not formally recommended.3,4 Of the arguments put forward against ECGs in athletes, the most frequent is the high false-positive rate, which can lead to unnecessary additional diagnostic exams and inappropriate exclusion of healthy individuals from competition. Most false positives result from incorrect interpretation of the ECG, mainly because alterations caused by exercise-induced physiological adaptations of the heart are wrongly classified as pathological.5–7 Thus, the central issue in this controversy is not whether the ECG should be included in pre-participation screening, but rather how the exam should be interpreted.

Various increasingly restrictive criteria have been published with the aim of standardizing interpretation of the ECG in athletes, notably those of the European Society of Cardiology (ESC), the Seattle criteria, and the ‘refined’ criteria.8–11 However, although the application of these criteria has led to a significantly lower number of false positives, they are still underused, and variability in the interpretation of the ECG in athletes remains high.12,13

The purpose of this study was to assess variability in the interpretation of the ECG in competitive athletes, with or without the use of specific criteria, in a sample of cardiologists and cardiology residents in Portugal.

The study included responses from 58 physicians (response rate 31%), 42 (72.4%) of them cardiologists and the remainder cardiology residents; 16 (27.6%) of the physicians stated that they regularly assessed athletes. More than half (32; 55.2%) did not use specific criteria for assessing athletes’ ECGs, and of those who did, the most commonly used were the Seattle criteria (n=13), followed by the ESC criteria (n=8). Table 2 presents the characteristics of the physicians surveyed and the percentages of correctly interpreted ECGs.

Each physician correctly interpreted a mean of 15±2 traces, corresponding to 74% of the ECGs analyzed, ranging between 45% and 100%. For pathological ECGs, the interpretation was correct in 68% of cases, ranging between 22% and 100%, while for non-pathological traces it was correct in 79% of cases (55%-100%) (Figures 2 and 3).

Figure 2.

Proportions of correct and incorrect interpretations of electrocardiograms. ECG: electrocardiograms.

(0.08MB).

Figure 3.

Variability in proportions of correctly interpreted electrocardiograms. ECG: electrocardiograms.

(0.05MB).

There were no statistically significant differences between cardiologists and residents concerning correctness of interpretation (74±10% vs. 75±10%; p=0.724), or between physicians who regularly assessed athletes’ ECGs and those who did not (77±12% vs. 73±9%; p=0.286). There was a trend for a higher rate of correct interpretation using specific criteria (77±10% vs. 72±10%; p=0.092). The intraclass correlation coefficient was 0.972 (95% confidence interval 0.951-0.987; p<0.001), showing that the reproducibility of the study was excellent.

This study demonstrates that assessment of the ECG in athletes remains less than optimal, since around a quarter of the ECGs analyzed were not correctly assessed and variability in interpretation was high. Although most of the physicians surveyed (cardiologists and residents) stated that they did not use specific criteria for ECG interpretations, there was a trend for a higher rate of correct interpretation using such criteria.

These results are consistent with previously published findings, and empirically there is a perception that variability (intra- and inter-observer) in interpretation of the ECG in athletes is high. However, there have been few recent studies on the subject. Berte et al.12 carried out a similar study, using a larger number of ECGs (138) and a smaller number of physicians (seven cardiologists and seven sports physicians), in which the ECGs were interpreted according to established criteria; variability was high, with disagreement on 35% of the ECGs overall. In a study by Hill et al.14 using a similar sample to the present work (18 ECGs analyzed by 53 pediatric cardiologists), the rate of incorrect interpretation was 31%. Analyzing 40 ECGs of athletes interpreted by physicians of various specialties, Drezner et al.15 showed that accuracy improved significantly after participants were provided with and asked to adopt the Seattle criteria; among cardiologists, the proportion of ECGs interpreted incorrectly fell from 15% to 4%.15

Although the difference was not statistically significant, in our study the rate of correct interpretation was lower for pathological ECGs, i.e. the proportion of false negatives was higher than that of false positives. Since the main problem with interpreting athletes’ ECGs is the high false-positive rate, the opposite would be expected. Irrespective of the characteristics analyzed, the proportion of pathological ECGs interpreted correctly was low except with the refined criteria,11 which were only used by a small number of physicians. Since these are the most recent criteria, they are probably more familiar to physicians with more experience and/or more interest in the area, who are thus more likely to interpret the ECG correctly. In previous studies the false-positive rate was similar, with more accurate interpretation seen after the adoption of specific criteria.14–16 For example, an assessment of the diagnostic accuracy of the Seattle criteria showed that the false-positive rate fell from 30% to 9% after they were applied.15 It is thus essential to raise awareness of specific criteria for the interpretation of the ECG in athletes and ensure they are correctly applied.

Variability in interpretation means that many athletes undergo unnecessary additional diagnostic exams, which have a considerable socioeconomic impact. There is also the question of the psychological stress caused by an incorrect reading, as well as the fact that these exams do not identify all causes of sudden death or ensure that athletes have been inappropriately barred from competing. Furthermore, the low prevalence of pathological ECGs in athletes means that they are more difficult to identify, increasing the variability of interpretation (the prevalence effect).15 Another important aspect is the influence of exercise-induced (physiological) electrical adaptations in the heart, which are affected by the athlete's demographic characteristics and the type of sport.5

There are many arguments in favor of including the ECG in pre-competitive screening of athletes, but it is still essential to optimize their interpretation. Possible ways to achieve this include training, standardization of methods, and centralized screening in specialized units. It is also important to raise awareness of the specific criteria for interpreting athletes’ ECGs and of their advantages and disadvantages, ideally through the publication of consensus documents and protocols indicating which criteria to adopt, in order to encourage standardization. However, the current criteria are the product of expert opinion and retrospective studies, and there have been few prospective analyses that can identify the ECG alterations that correlate most closely with clinical events, especially sudden death. Despite the shortcomings of the currently available criteria, their limitations cannot be evaluated, let alone overcome, until they have been applied in clinical practice.

This study has certain limitations that should be pointed out. The sample size was small, which limits the applicability of its results, and fewer than one third of surveyed physicians responded to the questionnaire. However, even so, the final study population was larger than in some previous studies on this subject. The characteristics of the selected athletes were very similar, with few non-Caucasians, females, or older athletes included. Although the questionnaire responses were confidential, physicians who were more interested and experienced in this area may have been more motivated to respond, and this also may have skewed the results. Some of the participants may have consulted the criteria when analyzing the ECGs, which hampers assessment of their existing knowledge; however, this does not affect the main purpose of the study and is in fact to be recommended in clinical practice, particularly for physicians who do not regularly assess athletes.

In the study population, a quarter of athletes’ ECGs were not correctly assessed and variability in interpretation was high. Standardization of ECG interpretation in athletes with the use of specific criteria can improve the accuracy of this exam in pre-competitive screening. However, these criteria are still underused, a situation that could be changed by improvements in medical training in this area.

The authors have no conflicts of interest to declare.