To the Editor: The latest American and European guidelines recommend oral anticoagulation (OA) unless contraindicated, targeting an international normalized ratio between 2 and 3 for patients aged 75 and older with atrial fibrillation (AF) and associated risk factors (arterial hypertension, heart failure, systolic dysfunction according to echocardiography, or diabetes mellitus) and for secondary prevention.1 Aspirin may be used when other risk factors are absent or when OA is contraindicated,1 although even recent cross-sectional studies have shown that only half of patients with AF at a high risk of stroke effectively receive anticoagulants and that this proportion declines with age.2,3
Guidelines are based on a meta-analysis4 of five unblinded randomized controlled trials (Table 1). In these trials, OA resulted in a 36% lower risk of stroke (14–52%) than aspirin. Three randomized trials have been published since the latest guidelines were issued, namely the Atrial Clopidogrel Trial with Irbesartan for prevention of Vascular Event (ACTIVE W),5 the Warfarin Versus Aspirin for Stroke Prevention in Octogenarians with Atrial Fibrillation (WASPO) trial,6 and the Birmingham Atrial Fibrillation Treatment of the Aged (BAFTA) study7 (Table 1). ACTIVE W compared OA with aspirin (75–100 mg) plus clopidogrel and showed that OA was significantly more effective. WASPO focused on primary prevention in octogenarians with permanent AF but was interrupted prematurely. In the BAFTA study, patient screening and recruitment were organized systematically across general practitioner practices in the United Kingdom. Patients could be excluded if their physician thought, in view of their ischemic and hemorrhagic risk factors, that they should, or should not, receive warfarin. How the general practitioners formed their judgment is not specified. The results favored OA.
|Date of publication||1989||1993||1994||1998||1999||2006||2007||2007|
|Number of CONSORT criteria respected (/22)||18||19||19||20||22||16||18||22|
|Design||OA, aspirin, or placebo||OA, aspirin, or placebo||OA or aspirin||OA, aspirin, aspirin+miniAVK, or miniAVK||OA, aspirin, or miniAVK||OA or aspirin+clopidogrel||OA or aspirin||OA or aspirin|
|Particularities||Secondary prevention||2 age groups (< or >75)||<78||Aged 80–90||Aged >75|
|Mode of recruitment||Clinical ECG||NSP||Register ECG||Clinical ECG||Population||NSP||Geriatrics||Population|
|Exclusion by general practitioner||yes||yes||yes||yes||no||NSP||no||yes|
|Patients identified, n||2,546||NSP||NSP||2,820*||1,837||NSP||NSP||4,639|
|Patients included, n||1,007||669||1,100||677||394||6,706||75||9,730|
|Duration of follow-up||NSP||2.3||2.7||2.2||2.7||1.3||NSP||2.7|
|Arm favored by main endpoint||OA||OA||NS||NS||NS||OA||NS||OA|
|Arm favored by stroke results||NSP||OA||NS||NS||NS||OA||NS||OA|
|Arm favored by bleeding results||NSP||Aspirin||Aspirin†||NS||NS||NS‡||NS||NS|
|Patients from current study (n=83) who would have been eligible, n||3||3||10||3||0||12||2||17|
To examine the external validity of these trials with respect to a population of frail elderly patients with AF, they were critically analyzed, with a focus on the inclusion and exclusion criteria, to assess the potential eligibility of a population of geriatric subjects.
During a period of 6 consecutive months, all patients (n=83) admitted to an acute geriatric unit with permanent or paroxysmal AF were identified. Their mean age was 89, and they were at a high risk of ischemic events (mean CHADS2 score8=2.7±1.2) and bleeding (mean HEMORR2HAGES score9=3.0±1.3). Seventeen of these patients (20%) could have been included in at least one of the eight trials, because they fit the published inclusion and exclusion criteria (Table 1). The main factors that would probably have excluded the patients in the current study was their limited life expectancy in 15 cases (11 because of severe dementia), poor adherence to treatment in 13 cases, and progressive cancer in eight cases.
Recruitment was organized systematically in only two of the eight published studies, and the attending physicians were allowed to refuse anticoagulation for their patients, regardless of any explicit exclusion criteria, in at least five of the eight trials (Table 1). These two facts would have resulted in selection biases, the most fragile and oldest patients likely being excluded.
Dans' method was used to assess external validity.10 It consists of posing six questions aimed at identifying a possible difference in the risk–benefit ratio of a treatment between a trial population and a population to which one wishes to apply the trialists' conclusions. Six negative responses are required for the trial results to be applicable to another population. With respect to the population of the current study, the answer would be yes to the third question: “Are there important differences in patient compliance that may diminish the treatment response?” and to the fifth question: “Do my patients have comorbid conditions that significantly alter the potential benefits and risks of treatment?” Dans' approach,10 without casting doubt on the superiority of anticoagulation over aspirin for thrombotic prevention, calls into question the reported difference in the risk–benefit ratios of the two treatments in the geriatric setting.
New trials involving very frail elderly patients are difficult to envisage, for practical and ethical reasons. Centralized registers recording the outcomes of all patients with AF receiving OA could provide clinicians with assistance when weighing the potential risks and benefits of this treatment in elderly patients at the highest risk.
These results thus highlight the poor external validity, with respect to frail elderly patients, of clinical trials comparing OA with aspirin in patients with AF. For such patients, aspirin, or therapeutic abstention, remains an alternative. Although aspirin carries a clear risk of bleeding, its known efficacy and simplicity make it an option for the most fragile patients with AF.