The sample size was calculated based on the Portuguese population over 65 years of age in the 2011 census (2010000 individuals), and assuming that 20% of those requested would not agree to participate in the study. Continuous variables were expressed as mean ± standard deviation, and categorical variables were expressed as percentage or median, as indicated.

The SAFIRA study results revealed an estimated overall AF prevalence of 9.0% in elderly subjects, similar in men (8.9%) and women (9.1%). The prevalence was 9.3% in the subset that underwent Holter monitoring and 10.6% in the subset fitted with an event recorder (Table 2). The estimated prevalence stratified by age was 6.8% among those aged 65-69 years, 11.1% in those aged 70-89, and 15.2% in those aged 80 years or older.

The median CHA2DS2-VASc score in the SAFIRA study was 3.5±1.2. The study revealed that 56.3% of subjects with known AF were not on oral anticoagulation therapy (including 29.8% treated with antiplatelet agents alone), while 43.7% were anticoagulated. The anticoagulation rate was 34.3% among individuals at senior centers and retirement homes. When the anticoagulation rate was stratified based on ischemic risk, the results were as follows: 25.3% for subjects with CHA2DS2-VASc score between 1 and 3; 50.1% with score 4-5; and 18.6% with score 6 or higher.

Among the subjects for whom data on anticoagulation were available, 65.7% were treated with VKAs and 34.3% were on NOACs.

In the former group, TTR was assessed whenever at least five INR values from the last six months were available. The median value was 41.7% (Table 2), this low figure being due in most cases (80%) to periods of subtherapeutic INR. In subjects treated with NOACs, we examined whether the drug, the dose level, and the number of daily doses were in line with the clinical presentation of each subject (including creatinine clearance when data to calculate it were available) and with the drug's Summary of Product Characteristics as approved in Portugal. It was found that 24.7% of the subjects were incorrectly treated (inappropriate dosage, inappropriate number of doses, or creatinine clearance incompatible with the use of NOACs). Using these criteria, 6.1% of anticoagulated subjects were overdosed (87.9% of them on VKAs) or did not have the creatinine clearance necessary to take the NOAC prescribed (93.1% of those treated with dabigatran) or the dosage prescribed (73.4% of them on rivaroxaban).

Overall, only 25.8% of anticoagulated subjects were found to be properly medicated (correct dosage for effective anticoagulation, including a TTR of at least 55% in those treated with VKAs).

The stroke rate in the AF population in the SAFIRA study was 11.2% (25.7% of whom had had two or more previous strokes).

The prevalence of cardiovascular risk factors in this population was very high (hypertension 85.3%, dyslipidemia 75.4%, and diabetes 22.7%) (Table 1). Control of these factors was also assessed, with control defined as blood pressure ≤140/90 mmHg, low-density lipoprotein cholesterol <100 mg/dl and glycated hemoglobin <7%. When these criteria were applied, only 5.6% of patients with AF and data available had all risk factors in the target range and were on appropriate oral anticoagulation.

As part of the SAFIRA study, subjects were also asked some questions concerning their knowledge of the disease, medical treatment, compliance with medication and side effects.

Answers to the question “What is most important to you about the medicines you take?” were as follows: safety (45.1%), efficacy (35.6%), cost (12.8%), and dosing convenience (6.7%).

To the question “What makes you stop taking long-term medication?”, 88.7% responded “side effects”. The main cause of discontinuing anticoagulants was bleeding (93.7%), while cost was mentioned by 1.6% of subjects (reported almost exclusively by NOAC users, in 4.8% of subjects).

Regarding side effects, the rate of any degree of bleeding reported by subjects taking antithrombotics was 4.6% in the anticoagulated population and 12.3% in those treated with antiplatelets. Of these, 9.8% of bleeding cases were classified by the subjects as severe (89.3% of these cases occurred in subjects treated with VKAs), and only 1.9% of cases required hospitalization.

When these prevalence rates are compared with those found by the FAMA study5 (2.5% prevalence of AF in the Portuguese population over 40 years of age), they may seem excessive. However, there are important differences between the two studies.

The FAMA study included subjects aged 40 years or older, recruited from the general population using a random-route method. By contrast, the SAFIRA study focused exclusively on the elderly population (aged 65 years or over) and sought to identify a more clinically varied population, including not only those usually treated in primary health care but also those in the non-governmental sector (senior centers and retirement homes), as well as some users of hospital care such as in day hospitals and emergency departments (although the latter were not used for recruitment of subject of the SAFIRA study). The FAMA study used the ECG as the only method to diagnose AF, whereas in the SAFIRA study, a subset of subjects underwent Holter monitoring or were fitted with an event recorder to improve detection of paroxysmal AF, which these devices identified in nearly three-fifths (58/102) of cases. This population has specific characteristics (Table 3). The results of other European studies, such as the Rotterdam study,6 which reported a prevalence of 8.3%, and a Spanish study by Cea-Calvo7 in which the prevalence was 8.5%, are within the same range as those of the SAFIRA study.

Despite these considerations, the SAFIRA study clearly shows a very high prevalence of AF in this elderly population, with similar figures for both sexes and increasing with age.

There is another finding that is at least as significant and challenging: 35.9% of the subjects were unaware that they had AF, which translates into a large population with unknown AF (in the 2011 census, one in five Portuguese were aged over 65). This makes stroke prevention very difficult in this subpopulation.

Despite all the evidence of the efficacy, safety and cost-effectiveness of anticoagulation in the prevention of ischemic stroke, especially among the elderly,1 the SAFIRA study showed that the rate of anticoagulation in subjects with known AF was low, seeing that, in this population aged 65 years or over, no subject had a CHA2DS2-VASc score8 of 0 and very few had a score of 1. Antiplatelet agents would not be recommended for the overwhelming majority of these subjects, and oral anticoagulation would be mandatory for almost all of them. However, the reality revealed by the SAFIRA study is quite different: 29.8% were treated with antiplatelets alone, and only 43.7% were anticoagulated. These figures are especially concerning for individuals at senior centers and retirement homes, even in those with relatively low HAS-BLED9 scores.

Regarding the relationship between CHA2DS2-VASc score and anticoagulation rate, the results were also not encouraging, with a low anticoagulation rate (25.3%) in those with a CHA2DS2-VASc score between 1 and 3, higher (50.1%) in those with a score of 4 or 5, but only 18.6% in those with a score of 6 or more.

Concerning the question of whether those under anticoagulation therapy were properly medicated, two findings warrant particular attention.

Firstly, in subjects treated with VKAs – the majority (65.7%) – the median TTR was only 41.7%, meaning that most of the time they were not properly anticoagulated, almost always due to the use of subtherapeutic doses. These individuals were thus not adequately protected against the risk of stroke,10 which is on average five times higher in patients with AF than in those without and which increases with age and duration of AF.

Secondly, around a quarter of subjects treated with NOACs were inadequately medicated (inappropriate dosage, inappropriate number of doses or creatinine clearance not compatible with the use of NOACs).

It is important to note that the results were much better in subjects treated with NOACs than in those treated with VKAs, even though NOACs were less often used in the elderly subjects of the SAFIRA study (34.3%) than in the general population (about 50%). Overall, only 25.8% of anticoagulated subjects were found to be properly medicated (correct dosing, with effective anticoagulation, including a TTR of at least 55% in those treated with VKAs). This indicates a potential for considerable improvement in this population, and means that the 11.2% stroke rate in the population with AF in the SAFIRA study (25.7% of whom had previously suffered two or more strokes) is hardly surprising.

It is well known that AF does not usually occur alone, especially in the elderly, but is almost always associated at least one cardiovascular risk factor, as clearly demonstrated by the composition of the CHA2DS2-VASc score8 itself. Control of cardiovascular risk is crucial to improving the prognosis for AF.11,12 With this in mind, the authors also assessed the presence and treatment of major cardiovascular risk factors. As expected, their prevalence was very high, as described in other studies.5–7 However, the degree of risk factor control was often inadequate; only 5.6% of subjects had all cardiovascular risk factors within target levels and were receiving appropriate oral anticoagulation.

This may be one of the most important issues addressed by the SAFIRA study, and is one that will be the subject of a separate publication. Here, we highlight some of the main points.

The question of what is most important to patients about their medication is often raised and can lead to heated debate. The ranking revealed by the SAFIRA study (safety [45.1%], efficacy [35.6%], cost [12.8%] and dosing convenience [6.7%]) seems to indicate that, least among the medicated elderly, more importance is attributed to the quality of the treatment than to its cost and convenience. In the future, it would be interesting to ask healthcare professionals or policymakers the same question to see if the results are similar.

Responses to the question of what makes patients stop taking antithrombotic treatments indicate that, as expected, bleeding is at the top of the list, but the rate of any degree of bleeding reported by subjects was relatively low in the anticoagulated population (a third of that reported by those treated with antiplatelet agents, which is all the more important given that antiplatelet therapy is not an appropriate strategy for stroke prevention in the great majority of these patients). The rate of discontinuation due to financial difficulties (found almost exclusively in subjects treated with NOACs, as expected) was lower than might be expected.

With regard to the safety of anticoagulation in the elderly, this is basically a question of the occurrence of bleeding. Subjects in the SAFIRA study classified only 9.8% of bleeding cases as severe (89.3% of these cases occurred in subjects treated with VKAs), and only 1.9% of cases required hospitalization. These results demonstrate the safety of this therapy (especially NOACs), even in these age ranges.

Still on the subject of safety, another finding to note is that according to data from the SAFIRA study, 6.1% of anticoagulated subjects were overdosed (especially those on VKAs) or did not have the necessary creatinine clearance to take the prescribed NOAC (almost always dabigatran) or the prescribed dose (almost always rivaroxaban). There is therefore also room for improvement in this regard.

As with any scientific study, the publication of the SAFIRA study's results and analyses are not the end, but the beginning. These results, with all their nuances and limitations, are not to be overemphasized or downplayed, but more importantly, they should not be ignored. They should prompt analysis of the current situation, which calls for better understanding and positive action, for which there is an obvious need.

This article does not set out to present answers or definitive proposals, but rather to encourage a much-needed discussion of the situation concerning AF among the elderly in Portugal and of the ways in which it will be possible to intervene.