Atrial fibrillation (AF) is the most common sustained arrhythmia in clinical practice, with an incidence of 19.2 per 1000 person/years in individuals aged over 65 years. It is a major cause of morbidity due to the condition itself and to the associated risk of thromboembolic stroke, and has been shown to be an independent risk factor for both overall mortality and sudden cardiac death.1

In the Framingham study, the annual risk for stroke attributable to AF was 1.5% in patients aged 50-59 years, rising steeply to 23.5% in those aged over 80.2

The incidence of stroke and associated mortality are higher in Portugal than in most European countries. The annual stroke incidence may be as high as 3.05 per 1000 population in rural areas of northern Portugal and 2.69 in the Greater Porto urban area. These figures are higher than the European average and cannot be attributed solely to other known risk factors for stroke.3

One possible explanation could be that AF is also more prevalent in the Portuguese population. This was the rationale for the FAMA study,4 which included 10447 individuals aged 40 and over of both genders and from all areas of the country. This epidemiological survey found an overall AF prevalence of 2.5%, which was much higher in those aged 70-79 years (6.6%) and in those aged 80 or over (10.4%). Prevalence was, however, low in younger individuals (1.6% for those aged 60-69 years). In the FAMA study, AF was diagnosed by 12-lead electrocardiogram (ECG) and the study sample was selected by random home visits throughout the country.

The prevalence found in the FAMA study was higher than expected considering the results of previous studies in the Portuguese population but lower than in studies in other European countries that served as a reference for it.5 For instance, in the Rotterdam study, which ran from 1990 to 2000 and included 7983 subjects aged 55 years and above (mean age 69.3 years) assessed by 12-lead ECG, baseline assessment indicated an overall prevalence of 5.5%.5 Based on the results of this study and the gender and age distribution of the Portuguese population in 2009, the prevalence of AF could be estimated at around 3.8% for those aged 45 or more and 6.8% for those aged over 60. This discrepancy could be due to the fact that the sample in the Rotterdam study was confined to a small geographical area, while the FAMA study included a much wider population.

One limitation of the FAMA study4 is the fact that heart rhythm was only assessed during a short ECG reading, which will inevitably lead to underestimation of the prevalence of AF if it is paroxysmal. On the basis of the study's results, it became clear that AF is underdiagnosed and undertreated, especially since a non-systematic analysis of clinical practice suggests that paroxysmal AF (PAF) is at least three times as common as persistent AF. There are in fact few estimates of the prevalence of PAF – around 50% in the Framingham study3,4 and 35-60% in other studies – and these refer mainly to symptomatic forms.

Holter 24-hour continuous ECG monitoring is routinely used to assess various symptoms, especially palpitations. It is commonly used to determine the heart rate profile in patients with known AF and to assess the effectiveness of antiarrhythmics or catheter ablation in maintaining sinus rhythm.

Although Holter monitoring is undoubtedly better than 12-lead ECG for assessing the prevalence of AF, since it can identify paroxysmal episodes that would not otherwise be detected, it still estimates prevalence by default. Furthermore, it cannot distinguish with certainty between persistent and long-standing persistent AF, which are considered distinct entities in the ESC guidelines for the management of AF.6 The definition of PAF – AF lasting less than 48 hours – also means that monitoring for only 24 hours may diagnose what is actually PAF as persistent AF.

Nevertheless, the fact that Holter monitoring provides longer ECG recordings means that it enables a more thorough assessment of AF prevalence, identifying individuals with daily paroxysmal episodes who are as vulnerable to thromboembolism as those with sustained AF and a similar risk profile.

The method used in this study, 12-lead Holter monitoring, has advantages over the more common 3-lead Holter, offering a more precise analysis of the patient's rhythm patterns than a conventional 12-lead ECG, with the additional benefit of identifying the onset and end of the episode, transitions between AF and atrial flutter (AFL), and episodes of atrial tachycardia, a condition that is often confused with AF.

The same definitions of AF and AFL can be used for 12-lead Holter monitoring as for the conventional ECG. AF is defined as an arrhythmia with continuous irregular R-R intervals and randomly oscillating baseline with absence of P waves. The most common form of AFL, designated typical AFL, is characterized by predominantly negative biphasic F waves in the inferior leads with a sawtooth pattern and predominantly positive F waves in V1. AFL is termed atypical if the F-wave morphology is different from the common type.7

Holter study is frequently performed routinely, often as part of a cardiological assessment. Individuals who are referred for Holter monitoring by their general practitioner (GP) are fairly typical of the general population, since the reasons for referral are diverse; this reduces selection bias and makes this population a suitable basis for a study estimating AF prevalence which is more likely to give valid conclusions that can be extrapolated to the general population.

The present study aims to determine the prevalence of PAF and, on the basis of its results, to contribute to the development of strategies for the prevention and treatment of AF in Portugal.

The primary objective of this prospective study was to determine the prevalence of PAF in patients aged 40 and above.

The secondary objectives were to determine the overall prevalence of AF/AFL regardless of the type and to compare the population with AF with the general population and patients with PAF with patients with persistent AF, in a population who underwent continuous 24-hour electrocardiographic monitoring.

This was a prospective cross-sectional epidemiological study of individuals referred for Holter study in two reference centers covering the Douro Litoral and Beira Litoral geographic areas, which include all of the Greater Porto urban area.

The individuals under study, who had contacted one of these centers after being referred by their GP for 24-hour Holter monitoring, were assessed and included in the study population regardless of the reason for referral. Patients of both sexes aged 40 and above who had undergone continuous 12-lead ECG monitoring for any reason in the participating centers during the study period and who provided written informed consent were included in the study. Individuals who did not for any reason fulfill the inclusion criteria at the time of the exam, or who refused to provide written consent, were excluded.

The Holter exams were performed by trained staff according to the same protocol and to the same quality standards in both participating centers, both private laboratories, using Mortara 12-lead recorders equipped with solid-state digital memory. Three arrhythmologists oversaw review and validation of the exam and the data were interpreted by two experienced cardiac technicians.

All participants completed a questionnaire providing information on demographics, clinical history, symptoms and medication.

The data were recorded by members of the study team in a dedicated database designed by Keypoint, Consultoria Científica, Lda, in which the Holter readings could be reviewed semi-automatically or manually by the three arrhythmologists, independently and blinded to all other patient data. The diagnosis was recorded in the database; if the assessments of the first two arrhythmologists differed, the third arrhythmologist was automatically notified in order to resolve the conflict.

AF was defined according to the standard ECG criteria described above. An episode of PAF was defined as one lasting longer than 30 s beginning and/or ending during the 24-hour recording, and persistent AF was defined as AF sustained throughout the 24-hour study.

The study was submitted to the National Data Protection Commission (CNPD) and to the Administrative Board of both centers before implementation.

This prospective study is probably the only one ever undertaken using 24-hour 12-lead ECG monitoring for the stated objectives.

Especial care was taken in collecting and interpreting the data, given that the referring GP was not aware that the exam had taken place or where.

The readings were interpreted and recorded in a specially designed dedicated database. Two of the arrhythmologists were blinded to the patient's identity and analyzed the readings independently; the findings were confirmed by a third team member who made the final decision in the event of disagreement.

The data obtained from the questionnaire were those usually collected during patient recruitment.

The overall prevalence of AF and AFL found in this study was even higher than expected (12.4%), which again highlights the importance of the link with stroke and the need for early diagnosis and prevention of thromboembolic events, including referral for invasive treatment such as radiofrequency catheter ablation or cryoballoon ablation.

Although 24-hour ECG monitoring is a well-known technique used routinely throughout the medical community, its usefulness for the detection of paroxysmal AF and AFL, which was the primary objective of this study, is frequently underestimated.

The high figure obtained for the prevalence of AF and AFL may have been affected by bias, the most important source of which is referral by GPs, who will have requested the exam according to their own clinical criteria, which were thus not predefined, although they were recorded for the purposes of the study. As noted above, the referring GPs were not aware of this study, and could not alter their clinical practice.

The prevalence found in this study is clearly different from those of Portuguese studies by Ascenção et al.8 (only 0.5%) and Gomes et al.9 (1.29%), who studied all primary health care users aged over 30 in part of the Greater Porto area. Both studies were based on retrospective analysis of medical records.

The prevalence of PAF in our study, as defined in the Introduction was 2.5%, a similar figure to the overall AF prevalence in the FAMA study. However, we believe that this prevalence is underestimated. If the proportion of patients with paroxysmal AFL is added, the overall prevalence is even higher.

As in the FAMA study, we decided to analyze patients with AF and typical AFL as a single group, since the latter is after all a type of arrhythmia, and one that frequently alternates with AF and carries a similar stroke risk. Although its standard treatment – cavotricuspid isthmus ablation – is different, it is often linked with the development of AF.

The FAMA study was prompted by the lack of data on the Portuguese population and variability in estimates of AF prevalence in European studies. It was a cross-sectional community study that analyzed a representative sample of the Portuguese population aged 40 and over, stratified by gender, age and geographical area, based on a study protocol that included a questionnaire and an ECG applied to a large sample (over 10000 randomly selected individuals from all regions of the country).

The main merit of the FAMA study was the fact that it assessed AF at a particular moment, when the ECG was taken. Its limitations were that the reading was very short (only a few seconds) and many study subjects refused to complete the questionnaire or undergo ECG testing. A total of 10447 subjects were assessed, but the number who fulfilled the inclusion criteria but refused to participate was much higher (15507). The reasons for refusal were not specified, but they may have been due to methodological considerations that are beyond the scope of this study.

In the present study, the participants had a clear reason to undergo 24-hour ECG monitoring, and it was not difficult to obtain their informed consent for an exam that would have been performed anyway.

The long duration of monitoring considerably increases the likelihood of recording an episode of AF, and so a higher prevalence would be expected. While even longer monitoring periods would further increase the chances of diagnosing AF, this is less practical, due to the effect on patient compliance, and is therefore limited to specific populations.10,11

Study participants who remained in sinus rhythm during the exam were asked if they had previously been diagnosed with AF or PAF, and 59 (1.4%) and 102 (2.4%), respectively, responded in the affirmative. Assuming that these responses were correct and ignoring the obvious possibility of overlap in responses from the same patient, it can be speculated that a further 3.8% should be added to the 12.4% prevalence calculated in our prospective study. This also means that the longer the monitoring period, the higher the prevalence of AF will be. This was demonstrated by Gladstone et al.,10 who studied the relation between AF and cryptogenic stroke in patients aged 55 years or older and recent stroke or transient ischemic attack. ECG monitoring over 30 days was superior to 24-hour monitoring, improving both diagnosis of AF more than five-fold and resulting therapeutic decisions, nearly doubling the rate of anticoagulation treatment at 90-day follow-up. There are, however, problems with prolonged monitoring, including patient compliance and artefacts.

Our study unequivocally demonstrates the value of 24-hour Holter monitoring for the diagnosis of PAF, since only 5.0% of the group diagnosed with PAF reported a previous diagnosis. By contrast, 57.5% of those with persistent AF on 24-hour monitoring had a previous diagnosis of persistent or permanent AF as defined in the international guidelines. This implies that for the latter patients, Holter study would have been requested as part of a rate control strategy as popularized following the results of the AFFIRM trial,12 but it also means that PAF was first diagnosed on the basis of Holter monitoring in 95% of cases, which has significant implications for clinical management of these patients.

The age distribution of individuals with paroxysmal AF or AFL was different from those with persistent AF, the former being significantly more frequent in younger age-groups (7.5% vs. 0.2% in those aged 40-49 and 11.6% vs. 6.6% in those aged 50-59). However, at older ages this pattern reversed, although the absolute numbers of patients diagnosed with PAF remained high and were never significantly lower than those with persistent AF. This means that PAF affects all ages and is not only a problem of aging.

It should be noted that 12-lead Holter monitoring is able to diagnose typical AFL according to the generally accepted definition, which may explain the relatively high prevalence (0.8%) of isolated typical AFL found in our study.

While the thromboembolic risk associated with AFL is similar to that of AF, treatment differs, since catheter ablation of AFL is more often curative than AF ablation, and pharmacological heart rate control with a lasting return to sinus rhythm is notoriously difficult to achieve in these patients.

AFL can coexist with or precede AF, and the relationship between them is complex. Some authors argue that in patients with coexisting AF and AFL, or even those with isolated AFL, catheter ablation of AF should be performed, with the possibility of associating it with cavotricuspid isthmus ablation of AFL.15

The high incidence of stroke found by Correia et al.3 in a geographically similar area to ours may well be linked to the high prevalence of AF found in the present study, in which significantly more patients with PAF/AFL had a history of stroke (p=0.001), with a prevalence of 9.3% in those with any kind of AF or AFL, and even more in those with PAF (p<0.024).

Comparison between patients with PAF/AFL and those with persistent AF shows a stronger association with hypertension in the latter group, although hypertension was significantly more prevalent in both groups than in those in sinus rhythm during the exam. This may mean that hypertension, considered a classic risk factor for AF, is less important in PAF, an early stage of the disease, the onset of which is due to triggering factors that are primarily electrical and centered around the pulmonary veins.13 This is particularly important given that the disease does not behave in the same way in all PAF patients: patients may have long periods without PAF punctuated by occasional symptomatic episodes, which could be due to a weaker trigger in a highly vulnerable atrial substrate, or – as in the individuals identified in our study – to strong and persistent triggers in an atrial substrate that is still able to resist sustainability of the arrhythmia. Bearing in mind that these are younger patients, since PAF is found at all ages, early invasive intervention (pulmonary vein isolation being demonstrably more effective than drug therapy), would appear to be the treatment of choice,10 especially as it is more likely to be successful when patients are referred earlier.14

Finally, as in other studies,4,9 the rates of oral anticoagulation were low: only 29.9% in patients with persistent AF and 12.8% in those with PAF. The explanation for the latter figure is simple: they had not been diagnosed.

The prevalence of PAF found in a population who underwent continuous 24-hour electrocardiographic monitoring for diverse reasons was 2.5% and the overall AF prevalence was 12.4%, five times higher than in the FAMA study. The prevalence of PAF – albeit an underestimation – was 2.5% and its age distribution was more uniform; as such it is more prevalent in younger patients and less associated with risk factors such as hypertension, although with significantly higher rates of stroke. These patients are at an early stage of the disease and may be candidates for catheter ablation, which is frequently curative when applied at this stage.

There was no external funding for the preparation of this article.

This study was promoted by Forpoint – Instituto de Formação e Inovação na Saúde, received editorial assistance from Keypoint – Consultoria Científica, and received an unconditional grant from Sanofi Portugal.

The authors have no other conflicts of interest to declare.