Impact of vascular disease in predicting stroke and death in patients with atrial fibrillation: the Danish Diet, Cancer and Health cohort study
Lars Hvilsted Rasmussen, Department of Cardiology, Cardiovascular Research Centre, Aarhus University Hospital, Aalborg, Denmark.
Tel.: +45 9932 8100 fax: +45 9932 6895.
Gregory Y. H. Lip, University of Birmingham Centre for Cardiovascular Sciences, City Hospital, Birmingham, UK.
Summary. Background: The presence of vascular disease (peripheral artery disease [PAD] and/or myocardial infarction [MI]) may impact on the risk of stroke and death among patients with incident atrial fibrillation (AF). To test this hypothesis, we analyzed data from a large Danish prospective cohort, the Danish Diet, Cancer and Health (DCH) study, to assess the risk of stroke or death among those who developed AF according to concomitant presence of vascular disease. Methods: A prospective cohort study of 57 053 persons (27 178 men and 29 876 women, respectively), aged between 50 and 64 years. The risk of stroke or death for patients with vascular disease was assessed amongst 3315 patients with incident AF (mean age, 67.1 years; 2130 men, 1185 women) using Cox proportional hazard models, after a median follow-up of 4.8 years. Results: Of the subjects with AF, 417 (12.6%) had ‘vascular disease’ (PAD and/or prior MI). The risk of the primary endpoint (stroke or death) was significantly higher in patients with vascular disease at 1-year follow-up (crude hazard ratio [HR] 2.51 [1.91–3.29]), with corresponding crude HRs for PAD and MI being 3.51 (2.40–5.13), and 1.99 (1.46–2.72), respectively. For the secondary endpoints of death or stroke individually, these risk estimates were similar (crude HR 2.48 [1.89–3.26] and 1.77 [1.18–2.66], respectively). After adjustment for risk factors within the CHADS2 score, the adjusted HR for the primary endpoint (stroke or death) in patients with vascular disease was 1.91 (1.44–2.54), which was also significant for death (1.97 [1.48–2.62]). Conclusion: Vascular disease (prior MI and PAD) is an independent risk factor for the primary endpoint of ‘stroke or death’ in patients with AF, even after adjustment for the CHADS2 risk score, although this is driven by the impact on mortality. This reaffirms that patients with vascular disease represent a ‘high-risk’ population, which necessitates proactive management of all cardiovascular risk factors and effective thromboprophylaxis (i.e. oral anticoagulation), which has been shown to significantly reduce the risk of stroke and death in AF.
Atrial fibrillation (AF) is the most common sustained cardiac rhythm disorder, and is associated with a substantial risk of mortality and morbidity from stroke and thromboembolism. Well-established risk factors for stroke in AF patients are prior stroke, advanced age, hypertension, diabetes mellitus and moderate-severe systolic dysfunction [1,2]. These stroke risk factors have been used to formulate stroke risk stratification schema, such as the Cardiac failure, Hypertension, Age, Diabetes, Stroke (Doubled) (CHADS2) score .
Limitations of commonly used stroke risk schemes such as the CHADS2 score have been highlighted  and with the availability of new oral anticoagulation drugs that overcome the limitations of the vitamin K antagonists (VKAs), a paradigm shift has been proposed to be more inclusive (rather than exclusive) of other stroke risk factors in refining stroke and thromboembolism risk stratification for patients with AF . Such a risk factor-based approach has been proposed in recent guidelines  where subjects with no stroke risk factors whatsoever can even be considered for no antithrombotic therapy (as they are ‘truly low risk’) whilst patients with ≥ 1 stroke risk factors should be considered for oral anticoagulation, whether with well-managed VKAs or one of the new agents (e.g. dabigatran) [5,6].
Many of the established stroke risk factors have been derived from non-warfarin arms of (now historical) clinical trial cohorts, and some risk factors (e.g. vascular disease) have not been routinely looked for, nor recorded in clinical datasets [1,2]. Some information can be obtained from cohort studies [7,8] and (eventually) from some contemporary clinical trials [9–11] where the presence of coronary or peripheral artery disease was a potential inclusion risk factor. In anticoagulated patients with AF, coronary artery disease appears to be an independent stroke risk factor for recurrent stroke . Indirect evidence is clearly evident that patients with vascular disease, including myocardial infarction, peripheral artery disease and complex aortic plaque, have a higher thromboembolic risk if they have AF [12–15]. The systematic review that was part of the NICE guidelines identified myocardial infarction as a stroke risk factor in AF patients, and thus ‘vascular disease’ is included within the NICE stroke risk stratification algorithm [2,16]. Finally, ‘vascular disease’ has been included in the CHA2DS2-VASc score, which was proposed as a refinement to complement the CHADS2 score by being more inclusive of risk factors for thromboembolism  and is used in current international guidelines for AF management.
In the original validation study of the CHA2DS2-VASc score from the Euro Heart Survey , MI and peripheral arterial disease were studied as a combined risk factor and not as individual stroke risk factors. This reduces the possibility of assessing the individual risk induced by these risk factors of stroke or death among patients with AF. Furthermore, the event rates for stroke or death were calculated after 1-year follow-up, thereby limiting important information about the time from incident AF to stroke or death. Another limitation was the absence of information on thromboembolic events at 1-year follow-up for more than 30% of patients from the baseline survey . Thus, further evidence on the impact of ‘vascular disease’ on stroke and mortality in AF is needed.
We hypothesized that the presence of vascular disease (peripheral artery disease and/or myocardial infarction) would predict the risk of stroke and death among patients with incident AF. To test this hypothesis, we analyzed data from a large Danish prospective cohort, the Diet, Cancer and Health (DCH) study, to assess the risk of stroke or death among those who developed AF according to concomitant presence of vascular disease.
The Danish Diet, Cancer and Health (DCH) study cohort was established between 1993 and 1997. The study design has been reported in detail elsewhere . The primary objective of this prospective study was to investigate the etiologic role of diet and lifestyle in the development of cancer, and 57 053 participants were enrolled (27 178 men and 29 876 women). The study participants were aged between 50 and 64 years, lived in the urban areas of Copenhagen and Aarhus, and did not have a cancer diagnosis registered in the Danish Cancer Registry at baseline. Participants were, for this study, followed from 1993 until December 2009. The DCH cohort has detailed information on demographics, existing comorbidities, and individual risk factors, including MI and PAD. Cross-linkage between the DCH cohort and the National Registry of Patients provides detailed information on incident AF, stroke and death and specific information about censoring from emigration and death during follow-up until December 2009. The study was conducted in accordance with the Helsinki Declaration II and approved by the regional ethics committees.
The DCH cohort subjects were linked to the National Registry of Patients, dating back to 1976, using the Danish Personal Identification number. This is a unique and national identification number, which is part of the personal information stored in the Civil Registration System. The study population in the present study included participants who developed incident AF during follow-up.
Codes from the International Classification of Diseases (ICD) were used to extract admissions for AF. ICD-8 was used until 1994, and after 1994 ICD-10 was used (see Appendix for details). AF and atrial flutter were coded separately in ICD-8 (codes 427.93 and 427.94), whereas in ICD-10, AF and atrial flutter have one ICD code (I48). Therefore, atrial flutter cases have also been included in the present study. The study population was defined as incident cases of AF after recruitment who had not emigrated before being diagnosed with AF. Cases diagnosed simultaneously with stroke, TE and transient ischemic attack (TIA) or patients who died on the same day they were diagnosed with AF, were excluded from the analysis, as we were particularly interested in a primary prevention cohort (i.e. no previous stroke or thromboembolism) to see if vascular disease was predictive of subsequent events. Components of the CHADS2 score were determined (and score calculated) at the time of AF diagnosis, and only hospital discharge diagnoses prior to the incident AF diagnosis were included.
The exposure variables studied were MI or peripheral arterial disease, either separately or combined as ‘vascular disease’. The primary comparison was ‘vascular disease’ (i.e. MI or PAD) vs. ‘no vascular disease’. Peripheral arterial disease refers to the obstruction of large arteries not within the coronary arteries, aortic arch vasculature or brain. Patients with a history of one of these cardiovascular diseases recognized prior to AF were identified using ICD-8 and 10 codes (ICD-8, 410.09, 410.99, 440.20, 443.99, 444.41 and ICD-10, I21.0-I21.9, I22.0-I22.9, I70.2, I73.9 and I74.5).
Given that thromboprophylaxis with oral anticoagulation therapy (but not antiplatelet therapy) has a significant effect on reducing stroke and mortality compared with a control , we defined our primary outcome as the composite of ‘stroke or death’ during follow-up. Given the nature of the dataset, some deaths could have been due to (undiagnosed) stroke, and the coding of ‘stroke’ refers to all strokes, consistent with recommendations on outcomes in AF studies . For the ‘all stroke’ endpoint, no differentiation between ischemic and hemorrhagic stroke was made, because not all subjects had cerebral imaging. Secondary analyses were performed for the components of the primary composite endpoint, that is, stroke and death. Information on emigration or death was available from the National Civil Registration System, and incidences of stroke were found in the Danish Hospital Discharge Register using ICD-8 and 10 (433.09, 433.99, 434.09, 434.99, 436.01, 436.90 and I63.0–I63.9, I64.9, respectively).
To describe the distribution of baseline characteristics we used a descriptive analysis with medians and 10th/90th percentiles. To include accurate information regarding time from incident AF to primary event (stroke or death, combined and independently), Cox proportional hazard models were used in patients with or without vascular disease. The primary exposure variables were categorized into acute MI, PAD, or both. Data were analyzed as crude (no adjustments) or adjusted for the risk factors included in the CHADS2 risk score: congestive heart failure, hypertension, age ≥ 75 years, diabetes, prior stroke or TIA (ICD-8, 427.09, 400, 249–50, 433–34, 435 and ICD-10, I50, I11, I10, I11, I12, I13, I15, E10, E11, E14, I64, I64.9, G45).
Time from diagnosis of AF to the primary outcome as the composite of ‘stroke or death’ was analyzed for the effects of the exposure to vascular disease’ (i.e. MI or PAD) vs. ‘no vascular disease’ (and their individual components, MI and PAD) and additionally, adjusted for the categorical variables included in the CHADS2 score. Results are presented as crude and adjusted hazard ratios (HRs) obtained by Cox proportional hazard models. Analyses were carried out using both 1-year follow-up and the entire follow-up period. Data were analyzed using Stata version 11 (Stata Corporation, College Station, TX, USA). A P value of < 0.05 was considered to be statistically significant.
Of 57 053 subjects in the Diet, Cancer and Health (DCH) cohort, we identified 3325 patients with AF who met the inclusion criteria. None of these had a stroke or had been diagnosed with thromboembolism (TE) on the day of admission; six were excluded because of TIA, and four were excluded from analysis because they died on the day of admission. This left a study population of 3315 patients with incident AF (mean age 67.1 years; 2130 male, 1185 female) (Table 1).
Table 1. Clinical characteristics of 3315 patients with incident atrial fibrillation
|Age, years (at time of AF)||67.1 (59.4;74.2)||66.9 (59.3; 73.9)||68.4 (60.6; 75.5)||69.2 (62.8;75.7)||67.9 (60.5;75.3)|
| Age ≥ 75 years||7.7 (254)||7.0 (203)||12.2 (51)||15.5 (19)||11.5 (38)|
| Age 65–74 years||56.1 (1859)||55.7 (1613)||58.8 (245)||61.8 (76)||58.5 (193)|
|Women||35.8 (1185)||37.5 (1087)||23.5 (98)||25.2 (31)||22.7 (75)|
|Years of follow-up||4.8 (1.0;11.1)||4.9 (1.0; 11.1)||3.6 (1.0;10.8)||4.1 (0.7;9.3)||3.8 (1.0; 11.0)|
|Congestive heart failure|
| Heart failure||13.9 (462)||11.1 (322)||33.6 (140)||35.8 (44)||35.5 (117)|
| LV dysfunction||11.5 (381)||8.8 (256)||30.0 (125)||30.1 (37)||32.7 (108)|
|Hypertension||28.8 (954)||26.5 (768)||44.5 (186)||48.0 (59)||42.4 (140)|
|Diabetes||9.0 (297)||7.49 (217)||19.2 (80)||29.3 (36)||16.1 (53)|
|Previous stroke||8.4 (277)||7.6 (220)||13.7 (57)||22.8 (28)||10.3 (34)|
|Transient cerebral ischemia||3.4 (113)||2.9 (84)||7.0 (29)||6.5 (8)||7.3 (24)|
|Thromboembolism||0.6 (21)||0.4 (12)||2.2 (9)||5.7 (7)||1.5 (5)|
|Peripheral arterial disease||3.7 (123)||0.0 (0)||29.5 (123)||100 (123)||10.9 (36)|
|Myocardial infarction||10.0 (330)||0.0 (0)||79.1 (330)||29.3 (36)||100 (330)|
|Aortic plaque||0.4 (12)||0.1 (3)||2.2 (9)||6.5 (8)||0.9 (3)|
Of this cohort, 417 (12.6%) had ‘vascular disease’ defined as PAD or prior MI. Hypertension was the most prevalent stroke risk factor (28.8%), followed by congestive heart failure (25.4%). MI and PAD were found in 123 (3.7%) and 330 (10.0%), respectively (Table 1). Subjects were followed-up for a median duration of 4.8 years (10th–90th percentiles, 1.0–11.1).
Incidence rates (per 100 person years) for stroke and/or death among 3315 patients after incident AF are shown in Table 2. The presence of vascular disease was clearly associated with significantly greater incidence of the primary endpoint of ‘stroke or death’ (28.8 per 100 person years), as well as stroke (8.6 per 100 person years) and death (19.7 per 100 person years), at 1 year.
Table 2. Incidence rates per 100 person years (95% CI) for stroke and/or death among 3315 patients after incident atrial fibrillation
| Stroke or death||9.5 (8.4;10.7)||8.0 (7.0;9.2)||20.8 (16.5;26.3)||32.7 (22.9;46.7)||17.7 (13.3;23.5)|
| Stroke||5.0 (4.3;5.9)||4.6 (3.8;5.5)||8.6 (6.0;12.4)||10.9 (5.9;20.2)||8.1 (5.3;12.3)|
| Death||9.2 (8.1;10.3)||7.7 (6.7;8.9)||19.7 (15.6;24.9)||30.4 (21.3;43.5)||16.8 (12.7;22.3)|
| Stroke or death||6.0 (5.6;6.4)||5.3 (4.9;5.7)||12.0 (10.4;14.0)||16.2 (12.5;21.1)||11.0 (9.3;13.1)|
| Stroke||1.8 (1.6;2.0)||1.6 (1.4;1.8)||3.4 (2.5;4.5)||4.6 (2.8;7.6)||3.0 (2.2;4.2)|
| Death||4.6 (4.2;4.9)||4.0 (3.7;4.3)||9.8 (8.3;11.4)||12.7 (9.6;16.8)||9.0 (7.5;10.8)|
Univariate and multivariate analyses
The risk of the primary endpoint (stroke or death) was significantly higher in patients with vascular disease at 1-year follow-up (crude hazard ratio HR 2.51 [1.91–3.29]), with corresponding crude HRs for PAD and MI being 3.51 (2.40–5.13) and 1.99 (1.46–2.72), respectively. For the secondary endpoints of death or stroke individually, these risk estimates were similar (crude hazard ratio 2.48 [1.89–3.26] and 1.77 [1.18–2.66], respectively).
After adjustment for risk factors within the CHADS2 score, the adjusted HR for the primary endpoint (stroke or death) in patients with vascular disease was 1.91 (1.44–2.54), which was also significant for death (1.97 [1.48–2.62]). The results for the full period of follow-up were essentially similar to the results for the first year of follow-up (Table 3).
Table 3. Hazard ratios of stroke and death among patients with incident atrial fibrillation preceded by peripheral arterial disease and/or myocardial infarction. (A) Risk of stroke or death. (B) Risk of death. (C) Risk of stroke
| Vascular disease (i.e. PAD or AMI)||2.51 (1.91;3.29)||1.91 (1.44;2.54)||2.09 (1.77;2.46)||1.45 (1.22;1.72)|
| PAD||3.51 (2.40;5.13)||2.33 (1.57; 3.46)||2.42 (1.85;3.18)||1.37 (1.03;1.81)|
| AMI||1.99 (1.46;2.72)||1.58 (1.15;2.17)||1.87 (1.56;2.25)||1.39 (1.15;1.67)|
| Vascular disease (i.e. PAD or AMI)||2.48 (1.89;3.26)||1.97 (1.48;2.62)||2.32 (1.94;2.78)||1.81 (1.50;2.18)|
| PAD||3.45 (2.36;5.03)||2.42 (1.64;3.61)||2.64 (1.97; 3.53)||1.76 (1.30;2.37)|
| AMI||1.97 (1.45;2.70)||1.61 (1.17;2.22)||2.07 (1.70;2.53)||1.67 (1.36;2.04)|
| Vascular disease (i.e. PAD or AMI)||1.77 (1.18;2.66)||0.80 (0.53;1.20)||1.87 (1.37;2.55)||0.97 (0.70;1.34)|
| PAD||2.02 (1.06;3.84)||0.73 (0.38;1.40)||2.17 (1.30;3.60)||0.87 (0.52;1.46)|
| AMI||1.63 (1.04;2.57)||0.88 (0.56;1.38)||1.67 (1.18;2.37)||1.01 (0.71;1.44)|
In the non-AF population (n = 52 678, after applying the same exclusion criteria in AF and non-AF cases), the risk of the primary endpoint (stroke or death) was significantly higher in patients with vascular disease, with a crude HR of 3.83 (3.47–4.23), with corresponding crude HRs for PAD and MI being 4.46 (3.75–5.32) and 3.64 (3.25–4.08), respectively (all data not shown). We have complete follow up data on all the patients for the entire study period.
This is one of the largest cohort studies with ‘real world’ patients, which demonstrates clear evidence for previous vascular disease (PAD or MI) as a risk factor for the primary endpoint of stroke or death in patients with incident AF. These associations were also observed even after adjustment for the CHADS2 score, and were independent of the follow-up periods.
The main strengths of the present study are the well-defined population and the prospective study design. Time from incident AF to the primary events was also taken into account in our analysis, and we had full follow-up of the entire population during the study period. Thus, the design of the study reduces the risk of confounding by referral and diagnostic bias. Moreover, we have chosen a primary endpoint that encompasses the main endpoints that are significantly and convincingly influenced by intervention (i.e. oral anticoagulation) . For example, the meta-analysis by Hart et al.  found that adjusted-dose warfarin was associated with a substantial reduction in stroke (64%; 95% CI, 49–74%) and all-cause mortality (26%; 95% CI, 3–43%) compared with a control, whilst the effects of aspirin on reducing stroke (19%; 95% CI, −1% to 35%) and mortality (14; 95% CI, −7% to 31%) were not statistically significant.
AF and vascular disease have many risk factors in common, and many patients with vascular disease also have AF . For example, in the large REduction of Atherothrombosis for Continued Health (REACH) Registry, of 68 236 stable outpatients with established atherothrombosis or ≥ 3 atherothrombotic risk factors, the incidence of AF was 12.5%, 13.7%, 11.5% and 6.2% amongst patients with coronary artery disease, cardiovascular disease and peripheral artery disease, and risk factor-only patients, respectively . Conversely, the cardiovascular risk profiles of stroke, MI and AF per se may have some subtle differences .
In a similar study by Frost et al. , the presence of PAD increased the risk of ischemic stroke by 1.3-fold in both men and women, although this was of borderline statistical significance. Indeed, the presence of complex aortic plaque on the descending aorta in patients with AF undergoing transoesophageal echocardiography is an independent predictor of ischemic stroke [15,24]. The high risk of death associated with PAD in AF has been previously recorded , and in patients with atherothrombosis, the presence of AF confers a poor prognosis [22,25]. For example, of the 6814 patients with AF within the REACH registry , 6.7% experienced cardiovascular death, non-fatal MI or non-fatal stroke within a year, and AF patients had a much higher annual incidence of non-fatal stroke (2.4% vs. 1.6%, P < 0.0001), unstable angina (6.0% vs. 4.0%, P < 0.00001) and cardiovascular death (3.2% vs. 1.4%, P < 0.0001), compared with non-AF patients. In a further analysis from the REACH registry, long-term CV mortality occurred in 5.6% of patients with AF and PAD, compared with only 1.6% of those without AF (P < 0.001). Indeed, AF was an independent predictor of late CV events, with an HR of 1.5 (95% CI, 1.09–2.0) .
In the study by Frost et al. , MI increased the risk of ischemic stroke by 1.2-fold, and was significant only in men. For ischemic heart disease without myocardial infarction, the risk of stroke was increased 1.1-fold and was of borderline statistical significance. Nonetheless, symptomatic angina has been found to be an independent risk factor for stroke and thromboembolism in AF , as has myocardial infarction [13,14,27]. For example, in the Copenhagen AFASAK study, previous myocardial infarction was an independent risk factor for development of thromboembolic complications using Cox’s regression model . Also, the occurrence and prognostic significance of AF following acute MI has been reported in numerous studies [13,28–31]. In the TRACE study, the unadjusted 5-year mortality rate was significantly higher in the patients suffering from AF/atrial flutter who had survived hospitalization for their MI . The prognostic implications of new-onset AF or atrial flutter in patients treated for acute MI were also investigated in the GUSTO-III trial . Patients who had AF during hospitalization had greater in-hospital complications, and the presence of AF predicted worse 30-day and 1-year mortality rates independent of both baseline characteristics and pre-AF complications.
Whilst the systematic review undertaken for the NICE guidelines identified MI as a stroke risk factor in AF , the systematic review by the Stroke in AF Working Group  did not identify ‘coronary artery disease’ as a significant stroke risk factor. Of note, this analysis also could not find a significant risk from ‘congestive heart failure’ (the ‘C’ in CHADS2), and was largely driven by the (now historical) original placebo-controlled trial cohorts (and a few epidemiological studies); in the historical trials, female subjects were under-represented and many stroke risk factors were not systematically looked for, nor consistently defined or accurately recorded. None of the studies included in the systematic review were larger than 2000 patients, and the historical studies were very heterogeneous. Furthermore, the mean follow-up time was only half as long as in the present cohort study.
By contrast, the present analysis clearly illustrates the impact of ‘vascular disease’ on stroke and mortality in our cohort, even after adjustment for the CHADS2 score . Of note, high crude hazard ratios were also seen in the non-AF population (although the larger size provides additional statistical power). As mentioned above, ‘vascular disease’ is included within the CHA2DS2-VASc score, which was introduced by the European Society of Cardiology (ESC) guidelines to complement (and not replace) the CHADS2 score . Given that oral anticoagulation therapy can significantly reduce both stroke and mortality in AF, the risk factor-based approach as advocated by the ESC guidelines tries to be more inclusive (rather than exclusive) of common stroke risk factors in AF . This is important, given the recognized limitations of the CHADS2 score in predicting risk in AF and the paradigm shift in our ability to identify the ‘truly low risk’ patients with AF who do not need any antithrombotic therapy, whilst patients with one or more stroke risk factors can be considered for oral anticoagulation therapy, which is highly effective in preventing stroke and thromboembolism [5,6]. Indeed, antiplatelet therapy has a minor impact on stroke prevention in AF (and may not be much safer in terms of major bleeding ), and when confined to the aspirin-only trials, the meta-analysis by Hart et al.  shows a non-significant 19% reduction in stroke with aspirin compared with placebo/control, with no effect on mortality. The combination of aspirin with oral anticoagulation has no effect on stroke, mortality and cardiovascular events, but results in a significant increase in major bleeding; hence, patients with stable vascular disease should be managed with oral anticoagulation alone, without the need for concomitant antiplatelet therapy .
We have used a primary endpoint of ‘stroke or death’ as our follow-up is dependent upon the National Civil Registration System, and some deaths could be due to undiagnosed stroke. Also, the incidence of stroke was defined by the Danish Hospital Discharge Register, and not all stroke endpoints were defined by cerebral imaging. As mentioned, we have chosen a primary endpoint of stroke and mortality, which are the two outcomes significantly influenced by intervention (i.e. oral anticoagulation) [19,33]. In the RE-LY trial, dabigatran 150 mg BID resulted in a significant reduction in stroke (by 36%, P < 0.001) and vascular mortality (by 15%, P = 0.04) compared with warfarin, although the effect on all-cause mortality was only of borderline significance (by 12%, P = 0.05) . Based on a network meta-analysis and indirect comparison analysis, oral anticoagulation with dabigatran 150 mg BID and 110 mg BID significantly reduced all-cause mortality compared with placebo (by 36% and 34%, respectively) . Even if we accept the hypothesis that previous MI and PAD are independent predictors of stroke and mortality, there remains the possibility that some (and probably many) of these events were caused by the concomitant presence of carotid artery stenosis and not necessarily by AF. Indeed, we are also limited by the inability to differentiate between ischemic and hemorrhagic strokes but the primary objective was to see if vascular disease had an impact on ‘all stroke’ (as recommended in recent position papers ) and death, especially because some deaths could have been due to undiagnosed strokes. Also, ‘real life’ data do show that whereas warfarin use for stroke prevention increased over the decade between 1992 and 2002, the rates of ischemic stroke substantially declined over this decade, with no apparent increase in rates of hemorrhagic stroke .
Also, it was not possible to include detailed data on anticoagulant or antiplatelet therapy, and choice of therapy could be biased by a selective preference for oral anticoagulation or antiplatelet therapy, and variance over time. Indeed, treatment choice regarding type of antithrombotic regimen ultimately relies on balance of benefit and risk estimation, which could introduce selection bias. Including all regimens will provide an overall estimate of risk of stroke or death in the entire cohort. Nonetheless, the calculated incidence rates in our study correspond to values reported from similar populations of patients with AF [10,11]. Indeed, if anticoagulant therapy had been offered to patients with a high CHADS2 score, thromboembolism (and mortality) may have been prevented. Despite this, vascular disease was still predictive of a substantially higher stroke risk in our cohort, even after adjustment for CHADS2 scores. This indicates that vascular disease was predictive beyond not only the CHADS2 score but also beyond the actions taken based on CHADS2 components or even use of the CHADS2 score itself to guide therapy in some patients. Finally, our cohort was also a relatively young population of incident AF patients (mean age 67 years at onset) but would probably be the patients who will benefit the most from more preventive efforts based on more comprehensive (and detailed) risk assessment.
In conclusion, vascular disease (prior MI and PAD) is an independent risk factor for the primary endpoint of ‘stroke or death’ in patients with AF, even after adjustment for the CHADS2 risk score, although this is essentially driven by the impact on mortality. This reaffirms that patients with vascular disease represent a ‘high-risk’ population, which necessitates proactive management of all cardiovascular risk factors and effective thromboprophylaxis (i.e. oral anticoagulation), which has been shown to significantly reduce the risk of stroke and death in AF.
We want to acknowledge The Danish Council for Strategic Research and The Danish Cancer Society for support.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.
ICD codes in the study
|Stroke*||433.09, 433.99, 434.09, 434.99, 436.01, 436.90||I63.0–I63.9, I64.9|
|Transient ischemic attack||435.09, 435.99||G45.0–G45.9†|
|Congestive heart failure||427.09||I50.0–I50.9, I11.0|
|Hypertension||400.09, 400.19, 400.29, 400.39, 400.99, 401.99, 402.99, 403.99, 404.99||I10.0, I10.9, I11.0, I11.9, I12.0, I12.9, I13.0–I13.9, I15.0–I15.9|
|Diabetes mellitus (Type I)||249.00, 249.09||E10.0–E10.9|
|Diabetes mellitus (Type II)||250.08, 250.09||E11.0–E11.9|
|Diabetes with no specification||–||E14.0–E14.9|
|Acute myocardial infarction||410.09, 410.99||I21.0–I21.9, I22.0–I22.9|
|Peripheral arterial disease‡||440.20, 443.99, 444.41||I70.2, I73.9, I74.5,|
|Arteriosclerosis of aorta||440.09||I70.0|
| Coronary artery disease||412.09, 412.99||I25.0, I25.1|
| Amaurosis fugax||377.02||G45.3|
| Pulmonary embolism||450||I26.0, I26.9|
| Deep venous thrombosis||451||I80.1–I80.9|
| Heavy smokers||Questionnaire (> 15 cigarettes per day)||Questionnaire (> 15 cigarettes per day)|
*430, 431 (ICD-8) and I60–I62 (ICD-10) not included (hemorrhagic). †Not inclusive G45.3 (Amaurosis fugax). ‡Peripheral arterial disease; refers to the obstruction of large arteries not within the coronary, aortic arch vasculature or brain.