Secondary stroke prevention in atrial fibrillation: a challenge in the clinical practice

The high risk for recurrences in cardioembolic stroke due to atrial fibrillation (AF) renders secondary prevention mandatory [1]. Oral anticoagulation (OA) has been proven as an effective treatment for this condition [1]-[5]. In recent investigations new tools for risk stratification have been developed [6]. Accordingly, current guidelines for patients with cardioembolic stroke and AF recommend oral anticoagulant treatment [7]. A favourable risk/benefit ratio using this regimen was also proven in elderly patients [8]. However, registry data shows an underutilisation of this treatment option [9]-[11]. Lack of knowledge of current guidelines, concern for the risk of bleeding, socio-demographic factors (increased age, ability to cope with everyday tasks) and clinical factors (neurological deficits, dementia, and previous bleedings) are suspected to influence the low rates in the usage of OA in the clinical practice.

In this study, we investigated factors which might have impact on the decision to use OA in stroke patients with AF with a favourable clinical outcome after the acute event. For this reason, we analysed prescriptions documented nationwide in an insurance dataset in the Federal state of Hesse, Germany, together with baseline data, derived from a large comprehensive stroke registry.

The presented study is based on data generated by the study of two different databases and where patient information records overlapped:

Within the GQH dataset 6261 documented cases were identified as stroke patients suffering of atrial fibrillation with a mRS ≤ 3 and no subsequent referral to another department. Out of these 6261 patients, 2101 were identified within the insurance's dataset. 1828 patients were included in the final analysis as they fulfilled the inclusion criteria (30 days free interval of further hospitalisation and no recurrent stroke with the subsequent 90 days after discharge and complete 90 day follow up). For details see also Figure 1.

The mean age of the study population was 77.6 years, 1064 (58.2%) were female. In 639 (35%) patients a transient ischemic attack (TIA) was documented as index stroke event. 1243 (68%) patients were treated in a neurological department and in 205 (11.2%) patients documentation for dementia was evident. In 827 patients (45%) a prescription for OA was identified. With increasing age the proportion of patients treated with OA decreases; while in the age category below 75 years the patients with prescription for OA represented the majority, above this age non-prescription predominated (Figure 2).

Factors such as advanced age, female sex, TIA, worse disability status on discharge, no treatment in a neurological department and documented dementia were associated with no prescription for oral anticoagulants (Table 1). For identifying independent factors, parameters associated in the univariate analysis were entered into a logistical regression analysis (age, sex, disability status as assessed on discharge, treatment in a neurological department and documented dementia). As a cerebrovascular event classified as TIA might directly influence the clinical outcome, which is indicated by the factor mRS, the parameter TIA was not included in the logistical regression analysis. Independently associated with a non-prescription decision for oral anticoagulants in the secondary stroke prevention were: increased age (OR: 0.54, CI: 0.46-0.63; P < 0.0001), female sex (OR: 0.77, CI: 0.63-0.94; P < 0.011), worse disability status on discharge (OR: 0.88, CI: 0.81-0.96; P < 0.006) and documented dementia (OR: 0.54, CI: 0.39-0.73; P < 0.001), (Table 2). Treatment in a neurological department was associated with a prescription decision (OR: 1.47, CI: 1.19-1.81; P < 0.003), (Table 2).

In studying a group of stroke patients with AF with a best case scenario, we detected a low rate (45%) of prescriptions for OA. This is a surprising result, considering the selected patients were most suitable for OA therapy; after the acute stroke they all had a favorable disability status (mRS ≤ 3 as assessed on discharge). With increasing age the proportion of patients treated with OA decreased (Figure 2). While advanced age, female sex, worse disability status and dementia were associated with a non-prescription, treatment in a specialized neurological department facilitated the therapeutic decision for OA in the secondary prevention.

Considering the low rate of prescriptions for OA in our study and the bulk of evidence, indicating the necessity for this therapy, the situation in the broad care delivery seems to follow particular rules [2],[4],[5],[14],[15]. Even though the evidence provided in the Birmingham Atrial Fibrillation Treatment of the Aged (BAFTA) indicates superiority of OA against aspirin in elderly patients, increased age was associated with non-prescription for OA [8]. However, in comparison to recent trials, which investigated the efficacy of OA in AF, patients selected for OA in our study were older (median 76 versus 70-71 years) [2],[4],[14]. Nevertheless, 55% of the patients seen were not prescribed OA which might be explained by the generally advanced age range (median 79 years) in patients encountered in daily practice.

A further aspect which needs to be taken into consideration is the risk for bleeding. This risk increases with increasing age [16],[17]. Particularly in the case of intracranial hemorrhages, the risk increases substantially for patients over 75 years in age [15],[18]. This clearly influences the practitioner's decision-making when selecting treatment options. Whereas, in the age category from 75 to 79 years half of the patients are treated by OA, in individuals ≥80 years the decision for an OA is considerably less frequent (Figure 2). Weighing all the arguments for OA for secondary stroke prevention in relation to the safety concerns, the reluctance to prescribe OA for elderly patients is understandable. A clinician's treatment decisions are guided by balancing effectiveness against risk to harm. As indicated by our results, safety concerns prevail. In order to minimize difficulties in the adequate selection of patients for OA, clear evidence on efficacy and safety among patients occurring in the real world are required. Future studies should address treatment decisions for elderly patient care. Furthermore, studies proving treatment effects in the secondary stroke prevention are necessary.

However, age was not the only determinant which contributed to the low usage rate of OA among stroke patients with AF in the clinical practice. An additional independent factor in facilitating the prescription for OA therapy was the treatment of the acute event in a specialized neurological department. Therapeutic recommendations from specialized departments seem to be of considerable relevance for the subsequent secondary stroke prevention emphasizing the need to establish such units. Furthermore, specific educational programs for general practitioners and specialists who are involved in the treatment of stroke patients with AF, appears necessary.

Concerning the disability status, there is no conclusive reason for not prescribing OA in our study. All selected patients were not severely disabled (mRS ≤ 3 as assessed on discharge), which implies the patient's ability to walk independently as well as minor assistance needed in coping with daily tasks. However, even in this range a slight deterioration results in the decision for an alternative therapy to OA.

The low rate of OA prescriptions in patients with documented dementia is not surprising, given to the need for monitoring this treatment precisely. However, there is no credible argument to withhold an effective therapy to patients with cognitive decline. Given the availability of newer anticoagulants with slightly less treatment monitoring requirements, these patients can benefit from OA usage in the future.

The identified factors might not be sufficient to entirely explain the low rate of patients selected for OA. The influence of further factors including cerebral imaging findings, the individual preference or acquisition of medication other than by prescription was not captured in our study. Thus, their relevance on the decision for or against OA remains unclear.

In the delivery of patient care, more than half of the patients with AF after stroke did not receive OA medication; 50% of these patients were above 81 years of age. In balancing effectiveness versus risk to harm, safety concerns prevail regarding therapeutic decisions. This is predominantly due to the high proportion of elderly patients receiving care. The factor of advanced age is correlated with a decision against OA. Patients with dementia were also less likely to receive OA. Even though all our subjects were only slightly affected by stroke (mRS ≤ 3), we noticed a decreased use of OA as there was an increased worsening of the disability. In contrast, patients treated in specialized neurological departments were more likely to receive OA as a treatment option. As a result, it would be beneficial providing additional education programs to non-specialists treating stroke patients make them more aware of its clinical benefits over other treatment options.