Atrial Fibrillation Patients Do Not Benefit From Acetylsalicylic Acid

Sara Själander; Anders Själander; Peter J. Svensson; Leif Friberg


Europace. 2014;16(5):631-638. 

In This Article


In this population consisting of 182 678 AF patients, almost one-third were treated with ASA as monotherapy. According to the new European guideline recommendations from 2012, ASA as monotherapy is no longer recommended for stroke prevention in AF, with exception for patients who refuse any form of oral anticoagulation and cannot tolerate a combination of ASA and clopidogrel.

Treatment with ASA was not associated with fewer ischaemic strokes or thrombo-embolic events than if no prophylactic treatment was given. On the contrary, ASA treatment was associated with more ischaemic strokes and thrombo-embolic events than if no antithrombotic treatment was given. This was true for most age groups and CHA2DS2-VASc strata. The higher risk associated with ASA treatment compared with no treatment remained after adjustment for comorbidities and propensity score matching.

The use of ASA for stroke prophylaxis in AF was founded on seven placebo-controlled studies performed from 1989 to 2006.[3,10–16] Of these, there was just one that showed a statistically significant, but modest protective effect.[11] A meta-analysis based on these seven trials showed that there was a 22% (CI 6–35%) relative risk reduction with ASA compared with placebo.[3] However, the reduction mainly concerned TIAs and minor strokes. When only 'disabling strokes' were counted,[10,11,13,14] the protective effect was reduced to a barely clinically relevant 13% (CI −18–36%). The latest trial performed in 2006 was terminated early since ASA was unlikely to reach superiority compared with no treatment. The effect of ASA was insignificantly negative regarding cardiovascular death, symptomatic ischaemic stroke, and TIA.

Recently, data from 12 placebo-controlled studies in the Atrial Fibrillation Investigators database, were reanalysed with the objective to determine the effect of age on the relative efficacy of oral anticoagulants and ASA.[4] This study showed that efficacy of ASA for protection against AF-related ischaemic stroke decreased with age and that the protective effect was none by the age of 75 and became insignificantly negative above that age. Considering that the mean age of the Swedish AF population is 76 years, our findings, although they may seem provocative, are actually in line with the findings in the placebo-controlled trials that once formed the basis for treatment for a majority of all AF patients.

Not only does ASA appear to be almost worthless for protection of ischaemic stroke, it has side effects in the form of increased bleeding risk. A meta-analysis has shown an association between treatment with ASA and increased incidence of gastrointestinal bleedings, major bleedings, and intracranial bleedings.[17] In this study, we could not confirm any association between ASA treatment and an increased incidence of bleeding events.

Our study confirms previous observations that women have higher risk for AF-related stroke than men.[7] We also could confirm findings of lower bleeding risk among women, than among men.[18,19] Since women with AF have higher risk of stroke than men, and lower risk of bleeding, women may possibly benefit more from oral anticoagulation treatment than men.

Our results suggest that patients with AF, who are not suitable for oral anticoagulation, may benefit more from abstaining from ASA, than from using it.


The patients in our study had not been randomized to either receive ASA or to have no prophylactic treatment at all. In spite of our efforts to adjust cofactors by means of propensity score matching and multivariable regression, we admit that it is unlikely that we have succeeded in adjusting for everything that may have affected the outcome. A randomized placebo-controlled study that could give an unambiguous answer to this very important question will however never be conducted given the proven efficacy of oral anticoagulants for patients with risk of AF-related stroke.

Use of registry data has limitations since it is dependent on the accuracy of diagnose registration. Validation studies of the Patient register have shown that most diagnoses have a high positive-predictive value.[6] The extent of under diagnosis is not known, and would require population screening to be determined. It is likely that some comorbidities have not received a diagnostic code in the registers. Therefore, patients may have received lower risk scores than they should have had if all circumstances were known.

Since age is not a continuous variable in age stratification as well as in risk score systems, and since patients treated with ASA in general are older than patients not on antithrombotic treatment, patients treated with ASA are expected to be older inside each age strata. Consequently, age stratification and stratification according to risk score systems cannot completely compensate for age differences between the two different treatment strategies. However, the results were similar after propensity score matching where age was used as a continuous variable.