Age Dependence of Risk Factors for Stroke and Death in Young Patients With Atrial Fibrillation
A Nationwide Study
Background and Purpose—The risk of stroke and death in patients with atrial fibrillation is strongly associated with age and concomitant comorbidities. The aim of this study was to examine the age dependence of risk factors for stroke and mortality in young patients with atrial fibrillation.
Methods—This study is a population-based cohort study of 30- to 65-year-old patients with atrial fibrillation and diagnosed during 2000 to 2011, identified by record linkage between nationwide Danish registries. Cox regression models were used to estimate the risk of stroke and mortality according to risk factors within age groups: 30 to 50, 50 to 65, and 65 to 75 years.
Results—We identified 73 799 nonvalvular atrial fibrillation patients, of which 37 782 (51.2%) were <65 years old (mean age 62.8). A higher modified cardiac failure or dysfunction, hypertension, age 75 (doubled), diabetes, stroke (doubled), vascular disease, age 65–74 and sex category (female) score (CHA2DS2-VASc score) was associated with decreased survival probability in all age groups. The overall incidence of stroke per year for 1 year (5 years) follow-up was 1.2% (0.6%), 3.5% (1.6%), and 5.6% (2.8%), respectively, for the age groups of 30 to 50, 50 to 65, and 65 to 75. Overall, risk factors such as previous stroke, heart failure, vascular disease, diabetes mellitus, and hypertension remained independent predictors of stroke and death in patients <65 years old with nonvalvular atrial fibrillation.
Conclusions—The CHA2DS2-VASc score is an applicable tool for all age groups and in nonvalvular atrial fibrillation patients <65 years old, the same risk factors apply.
The risk of stroke and death in patients with atrial fibrillation (AF) is strongly associated with age and concomitant comorbidities.1,2 The prevalence of AF doubles with each decade of age, reaching ≈9% at age 80 to 89 years3; thus, age is an independent risk factor for the development of AF.4
The risk of stroke and thromboembolism is increased in AF, but this risk is not homogeneous in all patients5 and is dependent on the presence of various stroke risk factors in the individual patient.6 The most common stroke risk prediction score in clinical use is the classic CHADS2 (congestive heart failure, hypertension, age ≥75, diabetes, stroke/transient cerebral ischemia [doubled]) score.7 Given its various limitations,8 such as classifying a large proportion of patients as having intermediate risk and not including important risk factors,8 the CHADS2 score has been refined by use of the CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years [doubled], diabetes, stroke/transient cerebral ischemia/thromboembolism [doubled], vascular disease [previous myocardial infarction, peripheral artery disease, or aortic plaque], age 65–74 years, sex category [female]) score.9 The CHA2DS2-VASc score includes additional common stroke risk factors.10–12 The CHA2DS2-VASc score better identifies the truly low-risk patients with AF who do not require any antithrombotic therapy, so that those with ≥1 stroke risk factors can be offered effective antithrombotic treatment.9
Risk stratification scores for AF have primarily been evaluated in patients with AF with a mean age >65 years. No study has adequately investigated whether risk factors of the CHA2DS2-VASc score have similar impact among younger patients with AF, and secondly, whether the increasing number of CHA2DS2-VASc risk factors is associated with comparable increases in stroke risk in all age groups. We hypothesize that the CHA2DS2-VASc score can be used to assess the risk of stroke in young patients with AF similarly to the general AF population. Using data from nationwide Danish registries, we assessed the rate and risk of stroke and mortality in a cohort of patients with AF, and furthermore, identified which independent CHA2DS2-VASc risk factors increased the risk of stroke, death, and the composite of stroke and death in young patients with AF.
Registry Data Sources
We used the Danish national registries in this study. The Danish National Patient Registry13 has registered all hospital admissions along with diagnoses since 1977 and codes all diagnoses according to the 10th revision of the International Classification of Diseases since 1994. The National Prescription Registry14 keeps data on all prescriptions dispensed from Danish pharmacies since 1994, coded according to the Anatomic Therapeutic Chemical (ATC) Classification System. The Danish Civil Registration System holds information on date of birth, migration, vital status, date of death, and sex of all citizens.15 Data were linked using a unique personal identification number (Central Person Register number), which is used in all Danish national registries. Full data were available up to December 31, 2012. These registries have previously been validated,13,14,16 and the predictive value of the diagnosis of AF was found to be high (99%), with a sensitivity of 88% and a specificity of 88% to 99% in these registries.17 Comparable statistics has been found for ischemic stroke.16
The study population was identified as patients, 30 to 75 years old, discharged with incident nonvalvular AF in the period (January 1, 2000, to December 31, 2011), giving ≥1 year of potential full follow-up, ≥5 years of prebaseline hospital history (start International Classification of Diseases [ICD]-10 code), and medical history. Nonvalvular AF was defined by a discharge diagnosis of AF or atrial flutter, but without a preceding ICD-10 code of valvular disease or mechanical cardiac valve. This is consistent with the definition used in similar cohort studies.10,18 The cohort consisted of both patients treated with anticoagulation and untreated patients. However, patients with a preceding diagnosis of cancer (according to ICD-10) or HIV/AIDS (according to ICD-10) were excluded. Similarly, patients with an invalid Central Person Register number (personal identification number), inconsistent death information (death before AF diagnosis), a date of death equal to the date of AF diagnosis, or immigration within 1 year before AF diagnosis were excluded (Figure 1).
Comorbidities were assessed by the components of the CHA2DS2-VASc score at time of AF diagnosis identified using the Danish National Patient Registry and the Danish National Prescription Registry (Table I in the online-only Data Supplement for ICD and ATC-codes).
Data from the Danish National Prescription Registry were used to ascertain the antithrombotic treatment by warfarin (vitamin K antagonist [VKA]) status of each patient. Treatment with warfarin was defined by first prescription and assumed to last until the package was assumed used up with a standard daily dosage (5 mg per day). With repeated prescriptions, dosage was estimated based on time between prescriptions. Treatment was assumed terminated if there were >180 days between end of treatment and next new prescription. If treatment was initiated <30 days after the diagnosis of AF, treatment was assumed to start at the date of the AF diagnosis, assuming that warfarin was handed out from the hospital. The analyses were adjusted for VKA treatment.
According to the CHA2DS2-VASc scheme, patients are given 1 point for congestive heart failure, hypertension, diabetes mellitus, and vascular disease, and 2 points for previous thromboembolism.19 Age group was handled separately in the analysis and therefore not included. Female sex counts for one point in the CHA2DS2-VASc scheme but was not included in this analysis to avoid a score of 0 to consist of males only. This modified CHA2DS2-VASc score is hereafter denoted mCHA2DS2-VASc. According to ESC guidelines 2012, a score of 0 (or a CHA2DS2-VASc score=1 in females) is considered as truly low risk with no antithrombotic therapy needed.
We defined our primary outcome as stroke or death resulting in an ICD-10 code of stroke or a date of death. Information on emigration or death was available from the Danish Civil Registration System, and incident cases of ischemic stroke (ICD-10: I63, I64.9) and systemic embolism (I74) after AF were found in the Danish National Patient Registry.
The study population was stratified into 3 groups according to age at nonvalvular AF diagnosis; age 30 to 50, 50 to 65, and 65 to 75 years. Baseline characteristics and independent risk factors were described separately for each of the 3 age strata with means and standard deviation for continuous measures and proportions for categorical measures.
In each of the 3 age categories, estimates for remaining event-free for up to 5 years after nonvalvular AF were calculated for all patients and stratified by mCHA2DS2-VASc score of 0, 1, 2, and ≥3. Follow-up was censored at time of emigration or end of study (December 31, 2012). Stroke event probabilities were obtained as cumulative incidence rates assuming death as a competing event, whereas risk of death and the composite of stroke and death were based on the Kaplan–Meier estimator. The effect of an increasing mCHA2DS2-VASc score and of individual risk factors was assessed by a Cox proportional-hazards model stratified by separate analyses on age group with no risk factors as reference. The analyses were adjusted for VKA treatment, by including an indicator for treatment, as a time-varying covariate.
Sensitivity analysis was performed by repeating the analysis on the untreated subpopulation, by censoring patients at time of initiated treatment with VKA therapy or at time of radiofrequency catheter ablation during follow-up.
The analyses were performed using SAS/STAT software version 9.2 for Windows (SAS Institute Inc , Cary, NC) and Stata version 11 (Stata Corporation, College Station, TX). A 2-sided P value of <0.05 was considered statistically significant. The study was performed and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations.
No ethical approval is required for anonymous registry studies in Denmark. The study was approved by the Danish Data Protection Agency (File No. 2012-41-063).
The study population comprised 73 799 patients with nonvalvular AF who were 30 to 75 years old. Of these, 37 782 (51.2%) patients were <65 years old and included in our main analyses (Figure 1). For the 5-year follow-up analyses, the median follow-up period was 4.9 years (interquartile range, 2.5–8.0). We had ≥1 year of potential follow-up data on all patients for the entire study period.
Baseline characteristics of the study population are shown in Table 1. The study population had a mean age of 62.8 years; 35.7% were female; and mean baseline stroke risk by CHA2DS2-VASc score was 1.6±1.4. As expected, we found increased comorbidity with increasing age, and a correspondingly higher mCHA2DS2-VASc score with increasing age strata. In all age categories, the most frequent comorbidities were hypertension, vascular disease, and heart failure. Overall, 44.7% of all patients were initiated on VKA treatment within 30 days after or before the AF diagnose, least in the younger age group and most in the older age group. Within each age group, there was no clear trend of increasing proportion for increasing risk.
The overall incidence of stroke per year for 1 year (5 years) follow-up was 1.2% (0.6%), 3.5% (1.6%), and 5.6% (2.8%), respectively, for the age groups of 30 to 50, 50 to 65, and 65 to 75 years.
The incidence rate of stroke per year was overall increased by higher mCHA2DS2-VASc score (Table 2). The rate of stroke was <2.7% (1.9%) for a score of 0 or 1 and, with a score >2, it was >12.8% (4.8%) for 1 (5) year follow-up for all age groups. The rate of death was increased by age group and by mCHA2DS2-VASc score. A similar association existed between event rate of combined stroke and death and higher mCHA2DS2-VASc score. The incidence rates of fatal stroke were in all age groups low, especially in the age group of 30 to 50 years.
Figure 2 shows cumulative incidence rates up to 5 years follow-up of stroke and death, respectively, in each age group according to mCHA2DS2-VASc scores of 0 (no risk factors present, including females), 1, 2, and ≥3. The cumulative incidence rates of stroke for mCHA2DS2-VASc score >1 increased steeply shortly after AF in all age groups and leveled off to a constant increase. The death rate curves increased fairly constant for all groups but with increased slope by higher score. Similar figure was made for the composite end point of stroke or death (Figure I in the online-only Data Supplement).
The hazard ratio (HR) for stroke and death of the mCHA2DS2-VASc scores 1, 2, and ≥3 relative to 0 (no risk factors present, including females), and for each risk factors (congestive heart failure, hypertension, diabetes, previous stroke, vascular disease, and female sex) relative to risk factor not present are shown in Figure 3 for each age group.
The end point of stroke was strongly associated with the presence of previous stroke (HR, 25.6; 95% confidence intervals (CI; 18.9–34.6]), vascular disease (HR, 3.8; 95% CI [2.5–5.6]), and diabetes mellitus (HR, 2.8; 95% CI [1.8–4.4]) in the age group of 30 to 50 years (Figure 3). Wide confidence intervals existed in this age group because of few events. In the age group of 50 to 65 years, the strongest associations with stroke were with previous stroke (HR, 10.6; 95% CI [9.7–11.6]), vascular disease (HR, 1.9; 95% CI [1.7–2.1]), and diabetes mellitus (HR, 1.7; 95% CI [1.5–1.9]). These HRs are higher than found in age group 65 to 75 years with previous stroke (HR, 5.7; 95% CI [4.6–5.2]), vascular disease (HR, 1.4; 95% CI [1.3–1.5]), and diabetes mellitus (HR, 1.4; 95% CI [1.3–1.6]). In all age groups, mCHA2DS2-VASc score >1 was resulting in much higher HR than with a score of 1.
The end point of death in the age group of 30 to 50 years was strongly associated with congestive heart failure, diabetes mellitus, vascular disease, hypertension, and previous stroke with HR ranging between 2.7 and 6.3. In the age groups of 50 to 65 and 65 to 75 years, the strongest associations with death were with heart failure, diabetes mellitus, and vascular disease, whereas hypertension was lower. As expected, female sex had a HR below 1, when associated with death in the age groups of 30 to 50, 50 to 65, and 65 to 75 years because females have a longer average life expectancy in general.
The composite end point of stroke or death had the strongest associations with previous stroke, heart failure, followed by diabetes mellitus and vascular disease across all age groups (Figure II in the online-only Data Supplement).
The sensitivity analysis, censoring patients at time of initiated treatment with VKA therapy or at the time of radiofrequency catheter ablation during follow-up, did only lead to minor numeric alterations of the results (Figures III and IV in the online-only Data Supplement).
In this real-world cohort study, we confirmed a low risk of stroke in patients <65 years old, but demonstrated that the risk is different in each age strata. The risk of stroke increased with the risk factors of previous stroke, heart failure, vascular disease, diabetes mellitus, and hypertension in all age categories. Event rates for death were significantly increased in all patients with nonvalvular AF by the presence of these risk factors. Indeed, higher mCHA2DS2-VASc scores were associated with decreased survival probability in all age groups.
Patients with AF who were <65 years old are considered to have a low annual risk of thromboembolic events,9 as supported by the results of this study. The influence of independent risk factors, as used in the CHA2DS2-VASc score, has previously been evaluated in patients with AF who were <65 years old as a homogenous group.20 One study found an independent increased risk of stroke and thromboembolism in association with heart failure, previous stroke, and vascular disease20 consistent with our results. When applying the CHA2DS2-VASc score to young patients with nonvalvular AF, it is still preferable to score these patients and apply the CHA2DS2-VASc score in clinical practice, as recommended in the international guidelines by the European Society of Cardiology.19
Currently, every patient with nonvalvular AF with a CHA2DS2-VASc score of ≥1 is recommended antithrombotic treatment. This broad biphasic decision-making may be too simplistic because not every risk factor is equally important when estimating the risk of stroke as shown in Figure 3. In our study, the risk of stroke was increased considerably by every risk factor except female sex and by higher HRs than in the age group 65 to 75 years. The results support the score of 2 points for previous stroke in the CHA2DS2-VASc score because previous stroke has the highest HR across all age groups.
By illustrating the probability of survival according to different CHA2DS2-VASc scores, it is possible to see how much each score influences the survival rate. According to our results, the 5-year survival rate decreased noticeably with a CHA2DS2-VASc score of >1 and with a score of ≥3 being the most hazardous score associated with a lower survival. This indicates that it is reasonable to keep attention on this risk and manage patients with a CHA2DS2-VASc score of ≥1 with antithrombotic therapy, as currently suggested in guidelines. Additionally, the development of risk factors such as hypertension, diabetes mellitus, and cardiovascular disease over time has been associated with an increased risk of stroke,9,12,19,20 emphasizing the importance of reassessment of risk factors for stroke over time. However, the influence and importance of hypertension and female sex as independent risk factors have been discussed.21,22
The age group of 30 to 50 years had a low rate of events and a more diffuse association with the risk factors of the CHA2DS2-VASc score. Therefore, the results in this age group should be interpreted carefully. Further investigations are necessary in these younger age groups because the pathogenesis of ischemic stroke might be different in the younger and older populations, and especially because AF could be related to an electrophysiological abnormality (eg, pre-excitation) or structural heart disease (eg, congenital heart disease).
Because of the observational nature of our nationwide registry study, some results should be interpreted with caution because of the generally low incidence rate of stroke in young patients with nonvalvular AF. We used an end point of stroke and death, and our follow-up was dependent on the National Civil Registration System, where some deaths could be because of undiagnosed stroke. The incidence of stroke was defined by the information extracted from the Danish National Patient Register, and it is not possible to obtain imaging data from the national registers to determine the subtype of ischemic stroke. However, today most diagnoses of stroke are confirmed by either CT or MR scans. Furthermore, we only included ischemic stroke and systemic embolism and not hemorrhagic stroke.
Because this was an observational study, it is not possible to draw direct conclusions on causal relationships of the findings. The frequencies of risk factors in the study population may also be underestimated because patients with heart failure, hypertension, and diabetes mellitus were identified from prescription claims, and thus, we were not able to detect patients treated nonpharmaceutically with diet control and lifestyle interventions alone.
Because of the low number of patients in some age intervals, analyses of these age groups were performed with reduced power resulting in wide confidence intervals. However, this study investigated a contemporary population of patients with nonvalvular AF followed in nationwide registries with limited loss to follow-up, and therefore, the study is unlikely to be subject to selection bias. The major strengths of this study remain the well-validated outcomes and large sample size uniquely possible in this type of cohort study, and almost no patients were lost for follow-up.
In conclusion, the CHA2DS2-VASc score is an applicable tool for all age groups, and in 30- to 65-year-old patients with nonvalvular AF, the same risk factors apply. The weight of some risk factors was higher with regard to the risk of stroke, especially previous stroke, vascular disease, and diabetes mellitus.
Sources of Funding
All authors had full access to all of the data in this study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors contributed to the design, analysis, interpretation of data, drafting the article, or revising it critically for important intellectual content and approved the final version to be published. Dr Lip has been on the Speakers’ Bureau for Bayer, BMS/Pfizer, Boehringer Ingelheim, Daiichi-Sankyo, Medtronic, and Sanofi. Drs Rasmussen and Larsen have been on the speaker bureaus for Bayer, BMS/Pfizer, Roche Diagnostics, Boehringer Ingelheim, and Takeda Pharma. The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.004903/-/DC1.
- Received January 21, 2014.
- Accepted February 27, 2014.
- © 2014 American Heart Association, Inc.
- Hughes M,
- Lip GY
- Kannel WB,
- Wolf PA,
- Benjamin EJ,
- Levy D
- van Walraven C,
- Hart RG,
- Connolly S,
- Austin PC,
- Mant J,
- Hobbs FD,
- et al
- Wolf PA,
- Abbott RD,
- Kannel WB
- Avenue G
- Olesen JB,
- Lip GY,
- Hansen ML,
- Hansen PR,
- Tolstrup JS,
- Lindhardsen J,
- et al
- Lip GY,
- Frison L,
- Halperin JL,
- Lane DA
- Lynge E,
- Sandegaard JL,
- Rebolj M
- Kildemoes HW,
- Sørensen HT,
- Hallas J
- Pedersen CB
- Camm AJ,
- Kirchhof P,
- Lip GYH,
- Schotten U,
- Savelieva I,
- Ernst S,
- et al
- Prescott E,
- Sørensen R