By Michael H. Crawford, MD, Editor

SYNOPSIS: Among patients with newly diagnosed atrial fibrillation age 66-74 years without other CHA2DS2-VASc risk factors for thromboembolism, the older they are in this age range, the more likely they are to experience a stroke.

SOURCE: Abdel-Qadir H, Singh SM, Pang A, et al. Evaluation of the risk of stroke without anticoagulation therapy in men and women with atrial fibrillation aged 66 to 74 years without other CHA2DS2-VASc factors. JAMA Cardiol 2021; May 19:e211232. doi: 10.1001/jamacardio.2021.1232. [Online ahead of print].

Most physicians use the CHA2DS2-VASc score (congestive heart failure, hypertension, age ≥ 75 years, diabetes, stroke, vascular disease, age 65-74 years, female sex) to decide which patients with atrial fibrillation (AF) to treat with anticoagulants (AC). The exact score cutpoint is controversial, especially in the group age 66-74 years with no other CHA2DS2-VASc risk factors.

Abdel-Qadir et al sought to determine the stroke risk in this population and the factors that influence this risk in a retrospective, population-based, cohort study using administrative databases in Ontario, Canada. They included community-dwelling individuals with newly diagnosed AF between 2007 and 2017. Patients with a prior diagnosis of AF, long-term care residents, and those on AC were excluded. The primary outcome was hospitalization with a stroke including intracerebral bleeds, but not bleeds outside the brain (e.g., subdural). Follow-up was limited to one year. Patients were censored if they were started on ACs. The subgroup who started AC within two months of AF diagnosis without an intervening stroke were the AC cohort. The occurrence of hospital-diagnosed bleeding was determined in this group.

The 16,351 individuals who met the inclusion/exclusion criteria were a mean age of 70 years. Women were 49% of the subjects, and there were 4,145 in the AC cohort. During the one-year follow-up from the onset of AF, there were 183 hospitalizations for stroke, including 19 with intracerebral bleeds, and 347 hospital-diagnosed noncerebral bleeds.

Using death as a competing risk (1,484 deaths), the risk of stroke in those not on AC was 1.1% and the risk of death without stroke was 8.1%. The stroke risk was not different between men and women, and the stroke risk in those younger than age 70 years was 0.7%.

The only baseline characteristic that was significantly associated with stroke was age. The risk of stroke more than doubled from age 66 years to age 74 years (0.7% to 1.7%, respectively). Warfarin was given to 1,936 patients, and a direct-acting oral AC was given to 2,209 patients. In the year after starting AC, there were 89 hospital-diagnosed bleeds (2.1%), which was not related to age. The authors concluded among new AF patients age 66-74 years with no other CHA2DS2-VASc risk factors, older patients are more likely to gain net clinical benefit from AC therapy than younger ones.


The CHA2DS2-VASc score at which AC is recommended varies from 1 to 3, depending on which societies’ guidelines one considers. The biggest discordance is among patients age 66-74 years with no other CHA2DS2-VASc risk factors. This controversy occasioned this population study. Part of the reason for conflicting recommendations is the risk of stroke has decreased since the pre-2005 data used to develop the CHA2DS2-VASc score. In those early studies, the risk of stroke was approximately 4% to 6% per year. Current data suggest it is more like 2% to 3% per year. The reasons for this drop are unclear, but it is probably at least partially explained by better control of stroke risk factors.

Considering the annual bleeding risk on AC of about 2% shown in the Abdel-Qadir et al study, only the higher-risk patients are going to reap a net clinical benefit from AC therapy. What the authors clearly showed is that in the age 66-74 years group without other risk factors for AF, that age should be considered since the older patients in this group are more likely to experience a net clinical benefit from AC therapy.

Patients age 66-74 years represented 6.5% of the Abdel-Qadir et al study population with new onset AF, and there are no randomized, controlled studies for this group. Thus, this retrospective, observational study is of interest and features several strengths. It was relatively large and contemporary. The problem of the competing risk of death was considered because if one does not, the estimated stroke risk is overestimated. In this study, the competing risk of death was substantial at about 8%. Also, there is the problem of treatment-switching, because about 38% of the subjects were started on AC. The authors addressed this by censoring those patients at this point and transforming this group into the AC cohort, which provided data on bleeding risk.

However, censoring only works if the decision to start AC was random. This raises the issue of unmeasured confounders. For example, those treated with AC may have had fewer comorbidities. Also, the types of AF were not considered (e.g., paroxysmal, permanent), nor was race or socioeconomic status. Finally, the CHA2DS2-VASc score usually is calculated once, and then determines lifelong AC therapy. This is an oversimplification that needs more study.

Perhaps, as these authors suggested, age should be a continuous variable rather than a categorical value, as age is included in the CHA2DS2-VASc score. For now, in borderline cases, I am going to consider age as an overriding factor for any decisions on AC therapy.