Background

Atrial fibrillation (AF) is the most commonly sustained arrhythmia and affects at least 1% of the population in Germany [1]. AF prevalence increases markedly with age, resulting in an estimated two- to threefold increase in AF patients in the next two decades [6, 31]. Many AF patients suffer from concomitant conditions including hypertension, vascular disease, heart failure and diabetes mellitus among others. In addition to variable but often relevant symptoms, AF appears to cause every fourth to fifth stroke and is associated with a doubling of mortality [34, 36].

AF management therefore consists of antithrombotic therapy, which is guided by clinical stroke risk estimation [1, 21], rate control therapy to improve left ventricular function and symptoms during AF, and rhythm control therapy to prevent AF recurrences. Despite the widely perceived notion that AF may cause severe complications, rhythm control therapy using common drugs does not prevent deaths in AF patients [11, 26, 33, 39], resulting in recommendations that rhythm control therapy should be pursued in patients who remain symptomatic on rate control [1].

Based on the variable presentation of AF and on slightly differing recommendations in clinical practice, differences in AF management depending on the type of treating physician can be expected [24]. Unfortunately, most registry data so far report that most AF patients are managed by cardiologists or other specialists.

We therefore initiated the prospective German ATRIUM registry to characterize AF management in patients treated by primary care physicians. Here, we report the baseline observations of this registry.

Methods

Design ATRIUM (OutpatientRegistry Upon Morbidity of Atrial Fibrillation) is a prospective, multicenter, epidemiological, non-interventional cohort study. ATRIUM enrolled 3,667 patients in 730 primary care practices in Germany in 2009. Baseline data included current management and information on interventions and complications in the year prior to enrollment. This data set is reported here. The ethics committee of the Technical University Dresden approved the study protocol. All patients gave written consent prior to enrollment.

To draw a random sample of centers, a multi-step procedure was used in which more than 25,000 physicians were contacted (Department of Medical Informatics, Biometry and Epidemiology, University of Bochum). Based on a comprehensive nationwide database of physicians (Schwarzeck-Verlag), a representative sample of 25,000 primary care physicians was drawn by Abteilung für Medizinische Informatik, Biometrie und Epidemiologie, Ruhr-Universität Bochum. These physicians were contacted via letter and informed about the study, and the first 730 respondents were offered participation. The centers agreed to consecutively enroll patients with AF documented by ECG in the 12 months prior to enrollment. No exclusion criteria were defined to minimize selection bias. All data were recorded during an outpatient visit and included information from the patient charts.

Parameters The following baseline parameters were documented: age, sex, body weight, height, blood pressure, risk factors for cardiovascular disease, cardiac history and concomitant diseases. The CHADS2 score was specifically recorded and the CHA2D2SVASc score, which was recently proposed as a refinement of the CHADS2 score [1], was computed using the available information. In addition, we recorded the month of initial diagnosis of AF, type of AF (paroxysmal, persistent including long-lasting persistent, or permanent), type of diagnostic tests performed, suspected triggering factors of AF, therapy in the year prior to enrollment, hospitalizations in the year prior to enrollment and referral to a specialist. Drugs were recorded by drug class. Quality of life was assessed by EQ-5D in its validated German version [9].

Data analysis and statistics

All data were recorded on paper case report forms (CRF), and double-entered by a contract research organization (CRO Dr. Schauerte, Grünwald) into the study database.

A prespecified validation plan was used to check for plausibility. Analysis was done by SAS Institute Inc., version 9.2 (Cary, NC, USA). Continuous parameters are given as means ± standard deviation and categorical parameters as the number of patients and percentages. Continuous parameters were compared between groups using ANOVA, and non-continuous parameters were compared using chi-square test. Throughout the paper, two-sided p values are given.

Results

Enrolling centers Of the 730 enrolling physicians (65% males), 63% were primary care phsyicians (“Facharzt für Allgemeinmedizin”), 34% internists with a license and practice in primary care (“Internist in hausärztlicher Praxis”) and 4% practising physicians. Practices were distributed among cities (29%), small towns (30%) and in rural areas (40%, data not recorded in 0.7%) Enrolling physicians were 50 ± 8 years old and worked in their practice for an average of 14 ± 9 years or in a polyclinic (“Medizinisches Versorgungszentrum”) for 9 ± 11 years.

Patient characteristics ATRIUM enrolled 3,667 patients, 58% male, with a mean age of 72 ± 9 years; 80% of the patients were retired (Table 1). The mean age was higher in patients with permanent AF than in those with paroxysmal AF, most likely reflecting the progressive nature of AF and the fact that older age was one of the factors that favored rate control therapy [1]. Paroxysmal AF was present in 994 (26%) patients, persistent including long-standing persistent AF in 944 (27%) and permanent AF in 1,525 (42%); in 204 patients (6%), AF type was not specified. Mean duration since the initial diagnosis of AF was 61 ± 66 months (median 42, interquartile range, 14–88).

Table 1 Patient characteristics
Full size table

The mean CHADS2 score was 2.2 ± 1.3. The mean CHA2DS2VASC score was 3.8 ± 1.7. CHA2DS2VASC score was lower in patients with paroxysmal AF (3.4 ± 1.7) compared to persistent AF (3.7 ± 1.6) or permanent AF (4.1 ± 1.7). Categorical distribution of scores is shown in Fig. 1.

Fig. 1
figure 1

CHA2DS2-VASc score. Score points based on available data for the calculation of the score in 3,667 patients

Full size image

Reported symptoms included palpitations (43%), shortness of breath (49%), fatigue (49%), dizziness (37%) and angina (20%). Most common concomitant conditions were hypertension (84%), heart failure (43%), coronary artery disease (35%), diabetes (35%) and chronic renal dysfunction (20%). Patients with a higher number of risk factors were more frequent in the groups with permanent AF (Fig. 2). Prior myocardial infarction was present in 11%, prior stroke in 10% and prior transient ischemic attack in 10%.

Fig. 2
figure 2

Numbers of concomitant conditions, by AF type. Risk factors: age ≥75 years, arterial hypertension, diabetes mellitus and chronic heart failure. Values are missing for AF type in 204 patients and for concomitant conditions in 18 patients

Full size image

Therapeutic goals Enrolling physicians reported the following therapeutic goals: prevention of thromboembolic events (77%), prevention of hospitalizations (57%), reduction of cardiovascular mortality (61%), rate control (76%) and rhythm control (33%; Table 2).

Table 2 Goals of therapy
Full size table

AF management Antithrombotic medication was used by 93% of the patients (oral anticoagulants 83%, antiplatelet drugs 27%, heparin 4%; Table 3). Contraindications for oral anticoagulants were reported in 6.4%. Of the 3,667 patients, 262 had CHADS2 score 0 and 79 CHA2DS2-VASc score 0 and were therefore rated as not eligible for OAC; 900 (CHADS2) and 240 (CHA2DS2-VASc) patients had a score of 1 and were potentially eligible; 2,486 (CHADS2) had a score ≥2 and thus were eligible for OAC for anticoagulation according to the guidelines in place at the time of the survey. [5, 28] Many patients received OAC despite being not eligible according to the scores (Fig. 3a, b). Further, of the 3,667 patients, 3,329 patients had a CHA2DS2-VASc score of 2 or more, rendering approximately 90% of the surveyed patients eligible for oral anticoagulation according to current recommendations [1]. Most patients at risk for stroke were adequately anticoagulated (Fig. 2) and, especially in patients without an indication for anticoagulation according to CHADS2 score, over-anticoagulation was also found (Fig. 2).

Table 3 Therapy in the previous 12 months
Full size table
Fig. 3
figure 3

Antithrombotic prophylaxis and oral anticoagulation (OAC) in patients with various eligibility categories for OAC by CHADS2 or CHA2DS2-VASc. Score 0 (=no risk factor) = no OAC recommended; Score 1 (=only 1 non-major risk factor) = either ASS or OAC recommended, OAC preferred; Score ≥2 (=at least 1 major or at least 2 non-major risk factors) = OAC recommended. Information on antithrombotic prophylaxis and oral anticoagulation was missing in 409 patients

Full size image

A total of 2,738 patients (75%) received rate control therapy, while 16% received rhythm control therapy either alone (189 patients; 5%) or in combination (404 patients; 11%). Drugs for rhythm and rate control included beta-blockers (75%), calcium antagonists (15%), digitalis (29%), potassium channel blockers including amiodarone (11%), and sodium channel blockers of the type IA (usually quinidine, 1.0%) or IC (usually flecainide or propafenone, 5%). In the year prior to enrollment, 46% of the patients had been cardioverted (23% by drugs and 22% electrically; Table 4). Catheter ablation had been performed in 5%, and 10% received a pacemaker or defibrillator.

Table 4 Pharmacological and electrical conversions
Full size table

A total of 1,602 patients (44%) were hospitalized in the year prior to enrollment, with 772 patients having been hospitalized more than once (Table 5). The mean hospitalization rates were somewhat higher for paroxysmal and persistent AF compared to permanent AF (1.2 vs. 1.1 vs. 0.7 stays during 1 year).

Table 5 Hospitalizations
Full size table

In the surveyed period, only 41% of the patients fulfilled the criteria for stable disease, defined as stable medication without AF-associated interventions.

Quality of life The EQ-5D was obtained from 3,460 patients. The mean EQ-5D index was 0.86 ± 0.19, close to the maximum value of 1. The corresponding VAS, in contrast, showed a mean value of 67 ± 18, indicating reduced quality of life.

Discussion

The data from ATRIUM described here provide information on the type of AF management in a sample of patients in primary care collected through a random sample of primary care physicians. Thereby, the report fills an information gap, as most prior registries include predominantly patients managed by cardiologists and/or in hospitals [2, 4, 17, 24, 27], which likely induced a selection bias based on center selection. Another registry, similar to ATRIUM, enrolled patients managed by German cardiologists [14]. Other information is available from population-wide samples [3, 6–8, 12, 13, 16, 30, 34, 35, 37], but information on the large proportion of patients managed as outpatients in primary care is scarce [19, 20, 22]. The central registry of the Germany AFNET more closely reflects the situation of AF patients managed by different types of physicians through enrollment of almost 10,000 patients from different levels of care [15, 23]. But even in the AFNET registry, only 811 patients were enrolled by primary care physicians (9%), demonstrating the need for further data from this sector of health care [23].

Classification ATRIUM enrolled a large unselected cohort of outpatients with AF in different stages of their disease (paroxysmal, persistent and permanent). We did not differentiate between first diagnosed AF and other forms of AF in this setting, as the classification “first diagnosed AF” is more relevant in acute settings such as emergency rooms or hospitals, as reflected by the rates of first diagnosed AF in the ALFA (26%) [19], Euro Heart Survey (18%) [24] and AFNET [23] (11%) registries.

Patient characteristics in relation to other registries Consistent with the expected flow of management in which patients with AF may be initially seen by a specialist, but later continue their treatment in primary care [1], mean age in ATRIUM (72 years) was higher than in the Euro Heart Survey (69 ± 10 years) or the AFNET registry (67 ± 13 years), and mean AF duration was over 5 years (66 months). Males slightly outnumbered females, comparable to other surveys. As expected, concomitant conditions were common, but there were slight differences: arterial hypertension was more often found in ATRIUM than in Euro Heart Survey (64%) or AFNET registries (69%). Also, coronary artery disease was surprisingly prevalent when compared with the aforementioned surveys [23, 24]. Similar to the AFNET registry, permanent AF was associated with more concomitant conditions.

Therapy and interventions In ATRIUM, 46% of the patients underwent cardioversion in the year prior to enrollment. Half of all cardioversions (23% of the total patient cohort) were achieved by drugs, which represents a higher rate than in MOVE (18%) [14], the AFNET registry (3–16% depending on the AF type) [23] or the Euro Heart Survey (3–14% depending on AF type) [24]. Electrical cardioversion was also relatively frequent in ATRIUM (22% in the total AF cohort) compared to AFNET (7–23% depending on AF type), MOVE (18%), or the Euro Heart Survey (3–24%) [14, 23] [24].

The type of rate control therapy was not markedly different from other trials and registries, with the exception of slightly lower use of digitalis glycosides, potentially already reflecting the growing experience that these agents only control heart rate well in sedentary patients [14, 23].

In ATRIUM, almost all patients received (any) antithrombotic therapy (92.5%), suggesting that stroke prevention was a firmly established therapeutic goal in the primary care setting. Furthermore, over 70% of patients with an evidence-based indication for oral anticoagulation received such therapy, a high proportion compared to other surveys. [38] Consistent with other registries [23, 25], a substantial portion of patients potentially ineligible for oral anticoagulation received such therapy (Fig. 3). This may in part reflect the “subconscious” application of a broader indication of anticoagulation in AF patients, as formalized in the CHA2DS2VASc score [1, 21]. Furthermore, the parameter “vascular disease” was somewhat underreported in this survey, as the components atherosclerosis of the aorta and peripheral arterial disease, were not recorded in the CRF.

Frequent hospitalizations despite enrollment of presumably “stable” patients The outpatient setting of ATRIUM also resulted in a high proportion of patients with permanent AF (42%, more than in Euro Heart Survey (29%) or AFNET (33%)), consistent with the AFNET data set showing a higher proportion of patients with permanent AF in outpatient centers [23]. The mean hospitalization rates were higher in paroxysmal and persistent AF compared to permanent AF, which is in contrast to earlier findings in the COCAF study [18]. Despite relatively frequent hospitalizations, the mean quality of life score in ATRIUM was slightly better than in the Euro Heart Survey (EQ-5D men 0.85, women 0.73) [24]. The outpatient setting and the permanent nature of AF render the number of patients with hospitalizations and “unstable disease” in our cohort remarkable, especially when this number was compared to the markedly lower hospitalization rate (about 25–27%) in large, recently published trials in AF patients such as ADONIS/EURIDIS [29], ATHENA [10] and RACE II [32]. While it is conceivable that lower hospitalization rates in trials reflect a selection bias toward “healthier” patients and possibly better overall management owing to the close follow-up regimen in clinical trials, the reasons for hospitalizations in this “all-comer” population with long-standing AF are worthy of further study.

Methodological considerations In selecting centers, care was taken to represent all regions in Germany equally. Nonetheless, it is conceivable that there remained a selection bias for centers with interest and/or expertise in AF management associated with the agreement to participate. Further limitations of this study are possible reporting bias (e.g., underreporting of diseases by physicians), misclassification of disease (e.g., AF type), selection bias of patients (only those willing to participate), neglect of patient-related factors such as treatment compliance and patient recall bias (e.g., on number of procedures or hospitalizations in the previous 12 months). Despite the systematic process for selecting participating centers (see “Methods”), selection bias of participating physicians is also possible. Furthermore, it was not possible to verify consecutive enrollment or the completeness of the information on the paper CRF by source data monitoring.

Conclusion

ATRIUM provides a hitherto unknown insight into details of current AF management in primary care in Germany. The good overall antithrombotic management is remarkable, but the frequent AF-related hospitalizations and the overall, often unstable, course of AF indicate unsolved problems. Challenges in the treatment of AF in these often multimorbid patients (with high rates of coronary artery disease and hypertension, for example) pose challenges to treating physicians.