Risk of ischemic stroke and utility of CHA2DS2‐VASc score in women and men with atrial fibrillation

Abstract Background The magnitude of increased risk of stroke in women with atrial fibrillation (AF) remains uncertain. Hypothesis We investigated the risk of ischemic stroke and death in women and men with AF, and the risk associated with individual non‐sex CHA2DS2‐VASc risk factors. Methods Retrospective cohort study of 231 077 (48.1% women) nonselected patients with AF not receiving oral anticoagulation from 2006 to 2014. Data from cross‐linked national Swedish registers. The outcome was the first occurrence of ischemic stroke or death. Median age was 82 and 75 years in women and men, respectively. Mean follow‐up was 2.5 years. Results Hazard ratios, adjusted for non‐sex CHA2DS2‐VASc risk factors, for women vs men were 1.53, 95% CI: 1.49‐1.58 for ischemic stroke and 1.24, 95% CI: 1.22‐1.26 for death, respectively. When divided into age groups the differences in ischemic stroke rates between women and men were attenuated. In patients with only one non‐sex CHA2DS2‐VASc risk factor allotted 1 point, ischemic stroke rates per 100 person‐years were 1.22 in women (n = 9838) and 1.02 in men (n = 15 609), respectively, P < .006. In both women and men, age of 65 to 74 years was associated with higher ischemic stroke risk compared to other non‐sex CHA2DS2‐VASc risk factors allotted 1 point. Conclusions The risk of ischemic stroke was 1.5‐fold higher in women compared to men but this association appears to be the result of confounding by age. In the low risk end, the CHA2DS2‐VASc risk score underestimates the ischemic stroke risk conferred by age 65 to 74 years, while it overestimates the risk conferred by female sex.


| INTRODUCTION
To assess the risk of stroke in patients with atrial fibrillation (AF), international guidelines recommend the use of the CHA 2 DS 2 -VASc risk scoring system to identify those who may benefit from oral anticoagulation (OAC) treatment. 1,2 Studies have shown conflicting results regarding the magnitude of increased stroke risk associated with each of the variables in the CHA 2 DS 2 -VASc score, especially regarding female sex. [3][4][5][6][7][8][9][10][11][12] Further research on the risk of stroke association of female sex and stroke risk is therefore needed and was encouraged by the European AF guidelines. 1 We aimed to (a) investigate the risk of ischemic stroke and death in women and men with AF and (b) assess the risk for ischemic stroke associated with each component of the CHA 2 DS 2 -VASc score separately in women and men with one additional non-sex CHA 2 DS 2 -VASc risk factor.

| Study population
The cohort included all patients with an AF diagnosis registered between 2 December 2005 and 31 December 2014. For each patient, the first day of follow-up was set as 30 days after the first occurrence of an AF diagnosis during the inclusion period. This clearance period is used to prevent an ischemic stroke event that happens before, or at the time of, AF diagnosis being counted as an outcome event, as has been described in previous publications. 13,14 Follow-up ended on 31 December 2014. A flowchart of the included patients is shown in Figure S1. Patients younger than 18 years and those with mitral stenosis or a prosthetic heart valve were excluded. The ICD codes used for exclusion are listed in Table S1. Patients who had filled a prescription for an OAC within 6 months prior to the start date of the study were excluded and patients who filled a prescription for an OAC during the follow-up period were censored at that time. The medication codes used for defining OAC medications are listed in Table S1.

| Outcomes and estimation of CHA 2 DS 2 -VASc risk factors
The outcome was the first of either ischemic stroke or death. CHA 2 DS 2 -VASc risk factors were identified by ICD codes in the Patient Register, with addition of diabetes medication codes in the Dispensed Drug Register. CHA 2 DS 2 -VASc risk factors at baseline were estimated only from diagnoses given before index date.

| Statistical methods
The analysis was performed in several steps. First, unadjusted incidence rates per 100 person-years with 95% confidence interval (CI) were estimated assuming the number of events within specific subgroups model, which might seem more natural for time-to-event data, was that the Poisson approach allows for a direct parametric modeling of the baseline rate and also for analyzing multiple time scales; in our case time from study entry and age. The age time scale was split in 5-year intervals in which the rates were assumed constant. All analyses were done with R, 15 version 3.3.2 using the Epi package. 16  Also, cause-specific hazard ratios (HRs) for the two competing risk outcomes (ischemic stroke and death) for women compared to men are shown in Table 1.

| RESULTS
Overall, women had a higher risk of ischemic stroke as compared with men. The relative risk increase in women was less pronounced of those aged 65 to 74 years (HR: 1.03, CI: 0.96-1.12 for women vs men) as compared those aged 85 years or older (HR: 1.25, CI: 1.19-1.32).
Overall, women had a higher risk of death (HR: 1.24, CI: 1.22-1.26), although this pattern did not remain when evaluating different age groups, where women actually had a lower risk of death in all age groups (Table 1). These findings were also similar for the composite endpoint of stroke or death, in which women had an overall higher risk than men (HR: 1.28, CI: 1.27-1.30), but when divided into different age groups women had lower risk in all age groups (Table 1). Figure 2 displays the predicted event rate of ischemic stroke and death in relation to age in women and men without any other CHA 2 DS 2 -VASc risk factor. It illustrates the minor differences in event rates of both ischemic stroke and death in women compared to men and also the higher event rates associated with higher age. In women and men at 55 to 70 years of age there were only minor differences in event rates.
3.1 | Ischemic stroke in relation to sex and CHA 2 DS 2 -VASc score Figure 3 shows incidence rates of ischemic stroke for men and women in relation to CHA 2 DS 2 -VASc scores. Men without CHA 2 DS 2 -VASc F I G U R E 1 Cumulative incidence of ischemic stroke and death. Cumulative incidence of ischemic stroke (solid lines) and death (broken lines) in women (red lines) compared to men (blue lines). Shaded area represents 95% confidence intervals. The total group of the 231 077 included patients is shown to the far left and then divided into four different age groups to the right. Follow-up limited to 12 months. M, number of men; W, number of women T A B L E 1 Cause-specific hazard ratios (HRs; with 95% confidence intervals) for the two competing risks outcomes ischemic stroke and death, and for the composite endpoint of ischemic stroke or death for women compared to men

| Risk of ischemic stroke in relation to individual CHA 2 DS 2 -VASc risk factors
Incidence rates of ischemic stroke in patients with only one non-sex

| DISCUSSION
In this study of patients with AF without OAC treatment, the risk of ischemic stroke and death is higher among women compared to men but this association appears to be the result of confounding by age.
The difference between the sexes in risk of ischemic stroke is attenuated in the younger age groups, but the difference in ischemic stroke risk becomes more pronounced above the age of 70. The single point given for female sex in the CHA 2 DS 2 -VASc score thereby overestimates the risk conferred to gender, while the single point given for age 65 to 74 years underestimates the risk conferred by age.
The importance and magnitude of difference in ischemic stroke risk between men and women have been vigorously debated and study results and interpretations have been inconsistent. In a previous observational Swedish cohort study, women had a moderately increased risk of ischemic stroke compared to men with an adjusted F I G U R E 4 Incidence rates of ischemic stroke per 100 person-years with 95% confidence interval in men and women with one additional non-sex CHA 2 DS 2 -VASc risk factor. Incidence rates of ischemic stroke per 100 person-years of follow-up with 95% confidence intervals in patients with one additional non-sex CHA 2 DS 2 -VASc risk factor divided by gender and CHA 2 DS 2 -VASc risk factors. The P-value for a test of no gender difference within each CHA 2 DS 2 -VASc score category was calculated using a Cox regression model including gender as the only covariate. CI, confidence interval; dis, disease; IR, incidence rate; N, number of patients; PY, person-years; SE, systemic embolism HR of 1.18 (CI: 1.12-1.24). 5 When divided by age, the unadjusted HR among those patients aged ≥75 years was 1.24 (CI: 1.18-1.30), while in patients <75 years there was no significant difference in risk. Similarly, a Danish cohort study found no significant differences in stroke/thromboembolic risk in age groups <65 and 65 to 74 years, while among those aged 75 years or older, women had higher risk of stroke/thromboembolism than men (HR: 1.20, CI: 1.12-1.28). 17 In the GARFIELD-AF registry, the stroke risk was elevated from the age of 70 years in women and from 65 years in men. 9 In line with our results, a recent Danish AF cohort study found similar risks of thromboembolic events in women and men without any non-sex CHA 2 DS 2 -VASc risk factors, with annual absolute risks of 0.5% without OAC treatment in both women and men. 11 This study also evaluated a CHA 2 DS 2 -VASc risk score, thus excluding the sex category criterion, but the authors recommended against implementation of this revised risk score since it could potentially confuse prescribing physicians, unintentionally indicating that women in general do not carry excess stroke risk relative to men. An UK AF cohort study reported ischemic stroke incidence rates per 100 person-years of 0.4 in men and 0.2 in women without any non-sex CHA 2 DS 2 -VASc risk factors and not on OAC treatment. 14 In the current study, we found an overall 53% higher risk of ischemic stroke in women compared to men, but when stratified into age groups the difference in risks for women vs men was much lower in all groups. There was also an overall higher risk in the composite of stroke or death in women compared with men, but when divided by age the risk was lower in women compared to men in all age groups indicating that this difference in men and women is confounded by age. Our results are in line with the Framingham Heart Study 4 where female sex was found to be an independent risk factor for stroke but not for a combined endpoint of stroke or death in patients with AF, which the authors assumed to be because of higher risk of mortality from other causes in men. In summary, the higher risk for ischemic stroke in women than in men seems to be lower than reflected by the Another limitation is that patients managed exclusively in primary care were not included in the cohort.

| CONCLUSION
The risk of ischemic stroke was 1.5-fold higher in women compared to men but this association appears to be the result of confounding by age. The CHA 2 DS 2 -VASc risk score underestimates the ischemic stroke risk conferred by age 65 to 74 years, while it overestimates the risk conferred by female sex at the low risk end of the scale. These results support the current European and US AF guidelines giving equal recommendations to both men and women with only one non-sex CHA 2 DS 2 -VASc factor, but the importance of age of 65 to