Skip main navigation
×

Risk of Stroke in Patients With Heart Failure

A Population-Based 30-Year Cohort Study
Originally publishedhttps://doi.org/10.1161/STROKEAHA.116.016022Stroke. 2017;48:1161–1168

Abstract

Background and Purpose—

The long-term risk of specific stroke subtypes among heart failure patients is largely unknown. We examined short-term and long-term risk of ischemic stroke, intracerebral hemorrhage (ICH), and subarachnoid hemorrhage (SAH) in heart failure patients and in a general population comparison cohort.

Methods—

In this nationwide cohort study (1980−2012), we used Danish population-based medical registries to identify and follow (1) all patients hospitalized for the first time with heart failure and (2) a birth year–, sex-, and calendar year–matched general population comparison cohort. Age-, sex-, and comorbidity-adjusted stroke rate ratios were computed based on Cox regression analysis.

Results—

We included 289 353 patients with heart failure and 1 446 765 individuals from the general population in the analysis. One- and 5-year risks among heart failure patients were 1.4% and 3.9% for ischemic stroke, 0.2% and 0.5% for ICH, and 0.03% and 0.07% for SAH. The 30-day adjusted stroke rate ratio was increased markedly for ischemic stroke (5.08; 95% confidence interval, 4.58–5.63] and was also elevated for ICH (2.13; 95% confidence interval, 1.53–2.97) and SAH (3.52; 95% confidence interval, 1.54–8.08). Between 31 days and 30 years, risk of all stroke subtypes remained positively associated with heart failure (1.5- to 2.1-fold for ischemic stroke, 1.4- to 1.8-fold for ICH, and 1.1- to 1.7-fold for SAH) in comparison with the general population cohort.

Conclusions—

Heart failure was associated with increased short-term and long-term risk of all stroke subtypes, suggesting that heart failure is a potent and persistent risk factor for ischemic stroke, ICH, and SAH.

Introduction

Heart failure, affecting >23 million people per year worldwide, is a leading cause of death.1 Comorbidities associated with heart failure have substantial implications for its prognosis.2 Heart failure may increase the risk of ischemic stroke because of thromboembolic complications and increased activity of procoagulant factors. At the same time, heart failure is associated with low blood pressure, which may protect against stroke.3 Well-known stroke risk factors include disorders associated with heart disease, such as hypertension, coronary artery disease, atrial fibrillation, diabetes mellitus, and obesity.4 However, the role of heart failure as a risk factor for stroke remains less clear.515

A few studies have compared stroke risk among heart failure patients with that of the general population. However, these were limited by relatively small sample sizes (<1500 patients)58 and relatively short follow-up periods (<5 years).57 As well, they were conducted in the era before routine use of angiotensin-converting enzyme inhibitors and beta blockers,8 did not separately examine ischemic and hemorrhagic stroke,68 and did not adjust or stratify for atrial fibrillation.7,8 They found that heart failure patients had a higher 30-day ischemic stroke rate than persons in the general population,5,6,15 but data after this initial follow-up period were sparse and equivocal.5,15

We, therefore, examined short-term (0–1 year) and long-term (1–30 year) risks and temporal trends in risk of ischemic stroke, intracerebral hemorrhage (ICH), and subarachnoid hemorrhage (SAH) in a large cohort of heart failure patients and in a general population comparison cohort. We also assessed how comorbidity affected the relation between heart failure and stroke risk. An understanding of this association could have important implications for future prevention strategies in patients with heart failure.

Methods

Setting and Design

We conducted a population-based nationwide cohort study of Danish-born residents (7 107 236 people cumulatively during the study period).16 In Denmark, all residents have equal access to universal tax-supported health care, including unfettered access to general practitioners and hospitals, and partial reimbursement for prescribed medications. All residents are assigned a unique central personal registry number at birth or on immigration, allowing linkage of data among administrative and medical registries.16

Heart Failure Patients

Patients with a first-time hospitalization for heart failure between January 1, 1980, and November 30, 2012, were identified from the Danish National Patient Registry (DNPR).17 Since 1977, this registry has maintained records on hospital admissions and discharges, including dates and diagnoses coded according to the International Classification of Diseases, 8th Revision through 1993 and 10th Revision thereafter.17 Hospital outpatient clinic and emergency room visits were added in 1995. We used both primary and secondary diagnoses (eg, heart failure diagnosed secondary to myocardial infarction) to identify patients with heart failure. To examine first-time stroke events in our study population, we excluded patients with an inpatient, emergency room, or outpatient clinic diagnosis of transient ischemic attack or stroke before the heart failure admission date. International Classification of Diseases codes used in the study are provided in Table I in the online-only Data Supplement.

General Population Comparison Cohort

We used the Danish Civil Registration System, which has maintained a registry with dates of birth, emigration, and death with daily updates since 1968, to form a general population comparison cohort. We matched each heart failure patient on birth year, sex, and calendar year of heart failure diagnosis with ≤5 individuals drawn from the general population without heart failure.16 We used matching with replacement (ie, individuals from the general population comparison cohort could be matched with >1 heart failure patient).18 We excluded individuals with a previous inpatient or outpatient diagnosis of transient ischemic attack or stroke. Individuals diagnosed with heart failure after the index date were sustained in the general population comparison cohort (to avoid informative censoring). The index date was defined as the inpatient hospital admission date for persons diagnosed with heart failure and the corresponding date of matching for members of the general population cohort.

Stroke

The study outcome was defined as all inpatient hospitalizations for stroke recorded in the DNPR after the index date.17 Stroke included first-time ischemic stroke, ICH, or SAH. In primary analyses, unspecified stroke diagnoses were included in the definition of ischemic stroke because more than 2 thirds of unspecified strokes in the DNPR are ischemic in origin.19

Covariables

We retrieved information from the DNPR on factors associated with heart failure and comorbidities between 1977 and the index date, using all available primary and secondary hospital-based diagnoses except for those made in an emergency room.17 We obtained data on previous myocardial infarction, angina pectoris, atrial fibrillation or flutter, valvular heart disease, hypertension, intermittent claudication, venous thromboembolism, hypercholesterolemia, hypertriglyceridemia, obesity, diabetes mellitus, chronic kidney disease, cancer, chronic pulmonary disease (as an indicator of chronic cigarette exposure), alcoholism-related disease, and dementia.

Statistical Analysis

All heart failure patients and members of the general population comparison cohort were followed from the index date until the hospital admission date for any stroke, emigration, death, November 30, 2013, or 30 years of follow-up, whichever came first. We characterized the cohorts by sex, age categories (<60, 60–69, 70–79, and ≥80 years), index year calendar periods (1980−1989, 1990−1999, and 2000−2012), and the covariables described earlier. Age and person-years of follow-up were reported as medians with interquartile ranges. Characteristics of the cohorts were compared using Chi-square test for categorical variables and 2-sample t test for continuous variables. Cumulative stroke risks were calculated using cumulative incidence curves/risks, accounting for death as a competing risk. We computed standardized incidence ratios as the observed number of stroke cases among heart failure patients divided by the expected number of cases in the general Danish population (assuming that heart failure patients had the same stroke risk as the general population).20 The expected number of stroke cases was calculated using national incidence rates for first-time stroke diagnoses, by sex, age, and 1-year intervals. The 95% confidence intervals (CIs) for the standardized incidence ratio estimates were computed assuming a Poisson distribution of the observed number of stroke cases in the different time periods. Stratified Cox regression analysis was used to calculate unadjusted (controlled only for matching factors by study design) and adjusted stroke rate ratios (aSRRs, specifically hazard ratios) with corresponding 95% CIs, comparing heart failure patients with the general population cohort.21 We adjusted for the variables presented in Table 1 in the regression analysis.

Table 1. Characteristics of Patients Hospitalized With First-Time Heart Failure and Members of the General Population Comparison Cohort, Denmark, 1980–2012

Heart Failure Cohort (n=289 353), n (%)Comparison Cohort (n=1 446 765), n (%)P Value
Male150 349 (52.0)751 745 (52.0)1
Age
 <60 y28 760 (9.9)144 113 (10.0)0.723
 60–69 y51 260 (17.7)256 959 (17.8)0.558
 70–79 y96 894 (33.5)484 451 (33.5)0.989
 ≥80 y112 439 (38.9)561 242 (38.8)0.507
Median (interquartile range)77 (69–84)77 (69–83)0.618
Decade of diagnosis
 1980−198992 148 (31.8)460 740 (31.8)1
 1990−199997 377 (33.7)486 885 (33.7)1
 2000−201299 828 (34.5)499 140 (34.5)1
Comorbidities
 Myocardial infarction43 985 (15.2)58 177 (4.0)<0.001
 Angina pectoris42 939 (14.8)72 278 (5.0)<0.001
 Atrial fibrillation or flutter31 001 (10.7)47 136 (3.3)<0.001
 Valvular heart disease11 480 (4.0)11 558 (0.8)<0.001
 Hypertension38 251 (13.2)86 932 (6.0)<0.001
 Intermittent claudication4570 (1.6)7342 (0.5)<0.001
 Venous thromboembolism9291 (3.2)23 728 (1.6)<0.001
 Hypercholesterolemia6041 (2.1)11 656 (0.8)<0.001
 Hypertriglyceridemia1820 (0.6)2667 (0.2)<0.001
 Obesity11 989 (4.1)17 018 (1.2)<0.001
 Diabetes mellitus29 147 (10.1)49 177 (3.4)<0.001
 Chronic kidney disease8191 (2.8)10 505 (0.7)<0.001
 Cancer31 942 (11.0)122 368 (8.5)<0.001
 Chronic pulmonary disease40 766 (14.1)66 023 (4.6)<0.001
 Alcoholism-related disease7609 (2.6)15 827 (1.1)<0.001
 Dementia4808 (1.7)23 003 (1.6)0.005

To investigate associations between heart failure and stroke independent of atrial fibrillation or atrial flutter, we repeated the analyses in heart failure patients and individuals from the general population with and without atrial fibrillation or atrial flutter. To assess temporal changes in stroke risk, we stratified the analyses by calendar periods, and we provided statistics for temporal changes using the statistical basis of meta-analyses.22 We also considered potential interactions in stratified analyses, which we used to examine the risk of stroke by sex, age groups, and in subgroups of heart failure patients. The proportional hazards assumption was assessed graphically in the pooled data set by means of log–log plots and found to be satisfied for the time periods analyzed.

Sensitivity Analyses

We conducted 7 sensitivity analyses. First, to improve the specificity of the stroke diagnosis, we limited an analysis to patients who were diagnosed with stroke and underwent a computed tomography scan or magnetic resonance imaging scan of the brain during the same admission (restricted to patients diagnosed from January 1, 2000, onwards, when these data were available). Second, we separately analyzed patients with unspecified stroke and specified ischemic stroke. Third, since antithrombotic drugs, angiotensin-converting enzyme inhibitors, and beta blockers may be important risk reduction mediators between heart failure and ischemic stroke, we repeated the analyses adjusting for their use within 90 days before the index date, using data from the National Health Service Prescription Database (data available from July 2004 onwards).23 Because the validity of recurrent stroke diagnoses in the DNPR is unknown, in the main analysis, we followed patients only until their first stroke diagnosis. However, to test the sensitivity of this approach, in our fourth sensitivity analysis, we allowed individuals to be at risk of other stroke subtypes after their initial stroke diagnosis. Fifth, we repeated the analyses for patients with first-time outpatient heart failure diagnoses (data available from January 1, 1995, onwards). Sixth, to exclude reverse causality (ie, stroke patients admitted with heart failure), we restricted the analyses to patients with primary heart failure inpatient diagnoses. Finally, because data on heart failure severity (ie, left ventricular ejection fraction) were not available, we stratified our heart failure cohort by intensive care unit admission and length of hospital stay (≤7 days and >7 days) as proxy measures of severity. To avoid conditioning on the future and potential immortal time bias, we changed the index date to 30 days after the admission date, with subsequent new matching at this point in time. We excluded patients who died or had stroke within 30 days in this analysis.

In all sensitivity analyses, a 1- to 5-year instead of a 1- to 30-year follow-up period was applied for long-term risk assessment.

All statistical analyses were performed using SAS version 9.2. According to Danish law, no approval from an ethics committee or informed consent from patients was required for this registry-based study. The study was approved by the Danish Data Protection Agency (record numbers: 1-16-02-1-08 and 2011-41-5755).

Results

The study comprised 289 353 heart failure patients and 1 446 765 individuals from the general population (Table 1). Because of the study design, the distribution of age, sex, and calendar year of the index date was the same for both cohorts. The median follow-up was 1.9 years (interquartile range, 0.2–5.1 years) for the heart failure cohort and 6.7 years (interquartile range, 3.2–11.9 years) for the general population comparison cohort. Competing mortality largely explains the difference in median follow-up. Heart failure patients had a higher prevalence of cardiac and noncardiac comorbidity than people from the general population (Table 1).

Ischemic Stroke

Cumulative incidence curves for the heart failure and general population cohorts are shown in Figure 1. During the first 5 years, patients with heart failure had a slightly higher absolute risk of ischemic stroke than individuals from the general population (Table 2 and Figure 1; and Table II in the online-only Data Supplement). After 5 years, the absolute risk of ischemic stroke was somewhat lower for heart failure patients than for the general population cohort because of competing mortality. In the Cox regression analysis, the 30-day aSRR was 5.08 (95% CI, 4.58–5.63). It declined but remained elevated during 31 to 365 days of follow-up (aSRR=2.08; 95% CI, 1.99–2.18) and during 1 to 30 years of follow-up (aSRR=1.54; 95% CI, 1.51–1.58; Table 2). Standardized incidence ratio estimates agreed closely with unadjusted stroke rate ratios (Table 2), and the associations between heart failure and ischemic stroke persisted in patients without atrial fibrillation or atrial flutter (Table 3).

Table 2. Risk of Stroke in Heart Failure Patients and Members of the General Population Comparison Cohort, by Type of Stroke and Follow-Up Time

Number at Risk/No. of EventsRisk, % (95% CI)SIR (95% CI)Stroke Rate Ratio Controlled for Matching Factors* (95% CI)P ValueFully Adjusted Stroke Rate Ratio (95% CI)P Value
Ischemic stroke
 0–30 days
  CC cohort1 446 765/8860.06 (0.06–0.07)ReferenceReferenceReference
  HF cohort289 353/8830.31 (0.29–0.33)5.30 (4.96–5.67)5.68 (5.16–6.26)<0.0015.08 (4.58–5.63)<0.001
 31–365 days
  CC cohort1 438 164/96770.67 (0.66–0.69)ReferenceReferenceReference
  HF cohort238 274/32271.36 (1.31–1.40)2.19 (2.12–2.27)2.39 (2.29–2.49)<0.0012.08 (1.99–2.18)<0.001
 1–30 y
  CC cohort1 345 483/112 00011.35 (11.29–11.42)ReferenceReferenceReference
  HF cohort176 288/12 2188.73 (8.57–8.89)1.64 (1.61–1.67)1.74 (1.70–1.78)<0.0011.54 (1.51–1.58)<0.001
Intracerebral hemorrhage
 0–30 days
  CC cohort1 446 765/1530.01 (0.01–0.01)ReferenceReferenceReference
  HF cohort289 353/620.02 (0.02–0.03)2.45 (1.87–3.14)2.18 (1.62–2.94)<0.0012.13 (1.53–2.97)<0.001
 31–365 days
  CC cohort1 438 164/14760.10 (0.10–0.11)ReferenceReferenceReference
  HF cohort238 274/3950.17 (0.15–0.18)1.81 (1.64–2.00)1.97 (1.75–2.22)<0.0011.83 (1.62–2.07)<0.001
 1–30 y
  CC cohort1 345 483/14 0241.53 (1.50–1.56)ReferenceReferenceReference
  HF cohort176 288/13270.98 (0.93–1.04)1.38 (1.30–1.45)1.45 (1.35–1.54)<0.0011.37 (1.28–1.46)<0.001
Subarachnoid hemorrhage
 0–30 days
  CC cohort1 446 765/230.00 (0.00–0.00)ReferenceReferenceReference
  HF cohort289 353/150.01 (0.00–0.01)4.24 (2.37–7.00)3.55 (1.83–6.89)<0.0013.52 (1.54–8.08)0.003
 31–365 days
  CC cohort1 438 164/2190.02 (0.01–0.02)ReferenceReferenceReference
  HF cohort238 274/630.03 (0.02–0.03)2.01 (1.54–2.57)1.90 (1.42–2.55)<0.0011.70 (1.24–2.34)0.001
 1–30 y
  CC cohort1 345 483/20800.23 (0.22–0.24)ReferenceReferenceReference
  HF cohort176 288/1850.14 (0.12–0.17)1.23 (1.06–1.42)1.18 (1.00–1.39)0.0561.13 (0.95–1.35)0.178

CC indicates comparison cohort; CI, confidence interval; HF, heart failure; and SIR, standardized incidence ratio.

*Adjusted for matching factors (age, sex, calendar decade of heart failure diagnosis).

Adjusted for matching factors, myocardial infarction, angina pectoris, atrial fibrillation/atrial flutter, valvular heart disease, intermittent claudication, venous thromboembolism, hypercholesterolemia, hypertriglyceridemia, hypertension, obesity, diabetes, chronic kidney disease, cancer, chronic pulmonary disease, alcoholism-related disorders, and dementia.

Table 3. Risk of Stroke Among Heart Failure Patients With and Without Atrial Fibrillation or Flutter, by Stroke Subtype

Risk, % (95% CI)Fully Adjusted Stroke Rate Ratio* (95% CI)P Value
No atrial fibrillation or atrial flutter
 Ischemic stroke
  0–30 days0.29 (0.27–0.31)5.49 (4.95–6.10)<0.001
  31–365 days1.26 (1.21–1.31)2.18 (2.09–2.28)<0.001
  1–30 y8.48 (8.31–8.64)1.52 (1.49–1.55)<0.001
 Intracerebral hemorrhage
  0–30 days0.02 (0.02–0.03)2.57 (1.86–3.55)<0.001
  31–365 days0.16 (0.14–0.18)1.78 (1.58–2.02)<0.001
  1–30 y0.95 (0.89–1.01)1.33 (1.25–1.41)<0.001
 Subarachnoid hemorrhage
  0–30 days0.01 (0.00–0.01)4.09 (1.99–8.38)<0.001
  31–365 days0.03 (0.02–0.03)1.95 (1.42–2.66)<0.001
  1–30 y0.14 (0.11–0.16)1.08 (0.92–1.28)0.343
Atrial fibrillation or atrial flutter
 Ischemic stroke
  0–30 days0.41 (0.35–0.49)2.84 (2.13–3.78)<0.001
  31–365 days2.13 (1.96–2.31)1.40 (1.26–1.57)<0.001
  1–30 y11.11 (10.48–11.77)1.07 (1.00–1.13)0.041
 Intracerebral hemorrhage
  0–30 days0.02 (0.01–0.04)0.66 (0.23–1.90)0.442
  31–365 days0.22 (0.17–0.28)1.24 (0.88–1.74)0.216
  1–30 y1.39 (1.13–1.70)1.05 (0.87–1.27)0.584
 Subarachnoid hemorrhage
  0–30 days0.01 (0.00–0.02)0.75 (0.09–6.54)0.798
  31–365 days0.03 (0.02–0.06)1.27 (0.51–3.14)0.606
  1–30 y0.19 (0.12–0.30)1.55 (0.93–2.59)0.095

CI indicates confidence interval.

*Adjusted for matching factors, myocardial infarction, angina pectoris, valvular heart disease, intermittent claudication, venous thromboembolism, hypercholesterolemia, hypertriglyceridemia, hypertension, obesity, diabetes mellitus, chronic kidney disease, cancer, chronic pulmonary disease, alcoholism-related disorders, and dementia.

Figure 1.

Figure 1. Cumulative 30-year incidence curve for ischemic stroke, ICH, and SAH in patients with incident heart failure compared with the general population comparison cohort. ICH indicates intracerebral hemorrhage; and SAH, subarachnoid hemorrhage.

Hemorrhagic Stroke

During the first 5 years of follow-up, absolute risks of ICH and SAH were similar for heart failure patients and the general population comparison cohort. After 5 years, absolute risks decreased for the heart failure cohort because of competing mortality (Figure 1). The 30-day aSRRs of ICH and SAH were increased (2.13 [95% CI, 1.53–2.97] and 3.52 [95% CI, 1.54–8.08], respectively) and remained 1.1- to 1.8- fold increased from 31 days to 30 years of follow-up (Table 2).

Stroke Risk Over Time

Temporal changes in stroke risk are illustrated in Figure 2. For ischemic stroke, a slight increase was seen in the 30-day aSRR over the 3 decades. In contrast, the aSRR slightly decreased for the 31- to 365-day and 1- to 5-year follow-up periods. For ICH, the aSRR remained stable during the 3 decades.

Figure 2.

Figure 2. Short- and long-term temporal trends in the adjusted stroke rate ratio (SRR) for heart failure patients compared with the general population during 1980 to 2009, with 95% confidence intervals (CIs). Adjusted SRR for 0 to 30 days omitted for subarachnoid hemorrhage because of insufficient numbers.

Subgroup and Sensitivity Analyses

Analyses stratified by age, sex, and cardiac comorbidity are presented in Table III in the online-only Data Supplement. For ischemic stroke, the aSRRs were similar for men and women. Although the aSRR for ischemic stroke decreased with age, the age-stratified results were consistent with the pattern reported for the main analysis (Table III the online-only Data Supplement).

The results were not appreciably different in any of the sensitivity analyses (Tables IV–X in the online-only Data Supplement). Within the first year of follow-up, the association between heart failure and ischemic stroke was stronger for patients admitted than for those not admitted to the intensive care unit and for those with length of stay >7 days than for those with length of stay ≤7 days (Table X in the online-only Data Supplement).

Discussion

In this nationwide cohort study, heart failure was associated with increased risks of ischemic stroke, ICH, and SAH over both the short and long term, and risks did not differ over 3 decades of follow-up. The associations persisted in patients without atrial fibrillation or flutter, across age groups, and sex, and remained robust in sensitivity analyses.

In accordance with previous studies, we found that heart failure is a strong risk factor for ischemic stroke, especially over the short term. A US cohort study of 630 heart failure patients reported a 17-fold elevated 30-day ischemic stroke risk compared with the general population, which persisted over 5 years.6 These findings were supported by a UK study reporting 2- to 3-fold higher odds for prevalent stroke in heart failure patients compared with the general population.7 Similar to these results, a Danish study of 1239 heart failure patients in the Diet, Cancer and Health Cohort reported an ischemic stroke rate ratio of 2.3 (95% CI, 1.8–3.0) and a 30-day relative risk for the composite outcome of death and all strokes of 35.7 (95% CI, 27.5–46.4). Although the association leveled out, it persisted over time (6 months to 14 years).15 Similarly, a Dutch cohort study of 1247 heart failure patients found that the rate of ischemic strokes was elevated in the first 6 months after a heart failure diagnosis. In contrast to our findings, the risk then converged to or became even lower than the risk of the general population.5 The Danish Diet, Cancer and Health Cohort study also reported elevated hemorrhagic stroke risk (adjusted hazard ratio, 1.8; 95% CI, 1.0–3.3) in heart failure patients,15 while another study found a decreased hemorrhagic stroke risk (hazard ratio, 0.80; 95% CI, 0.37–1.76)5 among heart failure patients relative to the general population. Our study complements and extends knowledge about long-term stroke risk and risk of stroke subtypes among patients with heart failure.

Several mechanisms are thought to underlie the increased risk of ischemic stroke in heart failure patients. One is formation of thrombi in the dilated, hypokinetic left ventricle because of wall-motion abnormalities and in the left atrium because of atrial fibrillation. In addition to shared stroke risk factors, heart failure is also associated with increased activity of procoagulant factors, aggregation of thrombocytes, and endothelial dysfunction.3 Changes in cardiovascular risk factors over time could be a part of the causal pathway to a subsequent stroke, and thus, they were not adjusted for in the analyses. Clinical pathways leading to the increased risk of hemorrhagic stroke are less well characterized and likely multifactorial, but may in part reflect a higher use of antithrombotic drugs in the heart failure cohort than in the general population comparison cohort during follow-up.

Use of antithrombotic agents to reduce ischemic stroke risk among heart failure patients in sinus rhythm has been debated during recent years.2 Because of null findings in randomized trials, anticoagulants have not been included in international treatment recommendations for heart failure patients without atrial fibrillation.2 However, the WARCEF substudy (Warfarin Versus Aspirin in Reduced Cardiac Ejection Fraction) of heart failure patients in sinus rhythm reported that longer time in the therapeutic range among patients allocated to warfarin reduced the risk of the primary outcome (ischemic stroke, ICH, or death) and death alone and also improved net clinical benefit.24 Heart failure patients with particularly high stroke risk include those with severely impaired left ventricular ejection fraction9,12,13 and with high risk scores for atrial fibrillation.25 The potential benefit of ischemic stroke prevention, including use of anticoagulants in heart failure patients without atrial fibrillation who are at high risk of stroke, and the role of non–vitamin K oral antagonists remains to be elucidated.

Our study benefitted from a large sample size, nationwide coverage, and virtually complete follow-up for 30 years. The risk of selection bias was, thus, minimized. The positive predictive value of diagnoses of ischemic stroke in the DNPR is 97% (using medical records as reference).17 However, the positive predictive values are somewhat lower for heart failure (~80%–100%),26 ICH (74%), and SAH (67%).17 Because recording of stroke subtypes is likely independent of the presence or absence of heart failure, any misclassification would be nondifferential and, thus, would bias our results toward the null.20 We adjusted for a range of confounders, but cannot exclude unmeasured confounders, such as physical activity. We lacked data on left ventricular ejection fraction. We could, therefore, not separately assess the potential differences in stroke risk among heart failure patients with reduced left ventricular ejection fraction and in those with preserved ventricular ejection fraction. However, analyses stratified by proxy measures of heart failure severity—intensive care unit admission and length of hospital stay—suggested that stroke risk may indeed be greater among patients whose left ventricular ejection fractions are reduced.

Conclusions

In this nationwide cohort study, heart failure was associated with increased hazard of ischemic stroke, ICH, and SAH, especially in the short term but also in the long term, suggesting that heart failure is an important risk factor for all types of stroke. This finding highlights the importance of clinical attention to stroke risk among heart failure patients. Further studies on potential prevention strategies are warranted.

Footnotes

The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.016022/-/DC1.

Correspondence to Kasper Adelborg, MD, Department of Clinical Epidemiology, Aarhus University Hospital, Skejby, Olof Palmes Allé 43–45, DK-8200, Aarhus N, Denmark. E-mail

References

  • 1. Roger VL. Epidemiology of heart failure.Circ Res. 2013; 113:646–659. doi: 10.1161/CIRCRESAHA.113.300268.LinkGoogle Scholar
  • 2. Yancy CW, Jessup M, Bozkurt B, Butler J, Casey DE, Drazner MH, et al.; American College of Cardiology Foundation; American Heart Association Task Force on Practice Guidelines. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines.J Am Coll Cardiol. 2013; 62:e147–e239. doi: 10.1016/j.jacc.2013.05.019.CrossrefMedlineGoogle Scholar
  • 3. Haeusler KG, Laufs U, Endres M. Chronic heart failure and ischemic stroke.Stroke. 2011; 42:2977–2982. doi: 10.1161/STROKEAHA.111.628479.LinkGoogle Scholar
  • 4. Sacco RL, Benjamin EJ, Broderick JP, Dyken M, Easton JD, Feinberg WM, et al.. American Heart Association Prevention Conference. IV. Prevention and Rehabilitation of Stroke. Risk factors.Stroke. 1997; 28:1507–1517.CrossrefMedlineGoogle Scholar
  • 5. Alberts VP, Bos MJ, Koudstaal P, Hofman A, Witteman JC, Stricker B, et al.. Heart failure and the risk of stroke: the Rotterdam Study.Eur J Epidemiol. 2010; 25:807–812. doi: 10.1007/s10654-010-9520-y.CrossrefMedlineGoogle Scholar
  • 6. Witt BJ, Brown RD, Jacobsen SJ, Weston SA, Ballman KV, Meverden RA, et al.. Ischemic stroke after heart failure: a community-based study.Am Heart J. 2006; 152:102–109. doi: 10.1016/j.ahj.2005.10.018.CrossrefMedlineGoogle Scholar
  • 7. Pullicino PM, McClure LA, Wadley VG, Ahmed A, Howard VJ, Howard G, et al.. Blood pressure and stroke in heart failure in the REasons for Geographic And Racial Differences in Stroke (REGARDS) study.Stroke. 2009; 40:3706–3710. doi: 10.1161/STROKEAHA.109.561670.LinkGoogle Scholar
  • 8. Kannel WB, Wolf PA, Verter J. Manifestations of coronary disease predisposing to stroke. The Framingham study.JAMA. 1983; 250:2942–2946.CrossrefMedlineGoogle Scholar
  • 9. Freudenberger RS, Hellkamp AS, Halperin JL, Poole J, Anderson J, Johnson G, et al.; SCD-HeFT Investigators. Risk of thromboembolism in heart failure: an analysis from the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT).Circulation. 2007; 115:2637–2641. doi: 10.1161/CIRCULATIONAHA.106.661397.LinkGoogle Scholar
  • 10. Szummer KE, Solomon SD, Velazquez EJ, Kilaru R, McMurray J, Rouleau JL, et al.; VALIANT Registry. Heart failure on admission and the risk of stroke following acute myocardial infarction: the VALIANT registry.Eur Heart J. 2005; 26:2114–2119. doi: 10.1093/eurheartj/ehi352.CrossrefMedlineGoogle Scholar
  • 11. Mahaffey KW, Harrington RA, Simoons ML, Granger CB, Graffagnino C, Alberts MJ, et al.. Stroke in patients with acute coronary syndromes: incidence and outcomes in the platelet glycoprotein IIb/IIIa in unstable angina. Receptor suppression using integrilin therapy (PURSUIT) trial. The PURSUIT Investigators.Circulation. 1999; 99:2371–2377.CrossrefMedlineGoogle Scholar
  • 12. Loh E, Sutton MS, Wun CC, Rouleau JL, Flaker GC, Gottlieb SS, et al.. Ventricular dysfunction and the risk of stroke after myocardial infarction.N Engl J Med. 1997; 336:251–257. doi: 10.1056/NEJM199701233360403.CrossrefMedlineGoogle Scholar
  • 13. Dries DL, Rosenberg YD, Waclawiw MA, Domanski MJ. Ejection fraction and risk of thromboembolic events in patients with systolic dysfunction and sinus rhythm: evidence for gender differences in the studies of left ventricular dysfunction trials.J Am Coll Cardiol. 1997; 29:1074–1080.CrossrefMedlineGoogle Scholar
  • 14. Sampson UK, Pfeffer MA, McMurray JJ, Lokhnygina Y, White HD, Solomon SD; VALIANT Trial Investigators. Predictors of stroke in high-risk patients after acute myocardial infarction: insights from the VALIANT Trial.Eur Heart J. 2007; 28:685–691. doi: 10.1093/eurheartj/ehl197.CrossrefMedlineGoogle Scholar
  • 15. Lip GY, Rasmussen LH, Skjoth F, Overvad K, Larsen TB. Stroke and mortality in patients with incident heart failure: the Diet, Cancer and Health (DCH) cohort study.BMJ Open. 2012; 2:pii: E000975. doi: 10.1136/bmjopen-2012-000975.CrossrefGoogle Scholar
  • 16. Schmidt M, Pedersen L, Sørensen HT. The Danish Civil Registration System as a tool in epidemiology.Eur J Epidemiol. 2014; 29:541–549. doi: 10.1007/s10654-014-9930-3.CrossrefMedlineGoogle Scholar
  • 17. Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sørensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential.Clin Epidemiol. 2015; 7:449–490. doi: 10.2147/CLEP.S91125.CrossrefMedlineGoogle Scholar
  • 18. Stuart EA. Matching methods for causal inference: a review and a look forward.Stat Sci. 2010; 25:1–21. doi: 10.1214/09-STS313.CrossrefMedlineGoogle Scholar
  • 19. Krarup LH, Boysen G, Janjua H, Prescott E, Truelsen T. Validity of stroke diagnoses in a National Register of Patients.Neuroepidemiology. 2007; 28:150–154. doi: 10.1159/000102143.CrossrefMedlineGoogle Scholar
  • 20. Rothman KJ, Greenland S, Lash TL. Modern Epidemiology. 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2008.Google Scholar
  • 21. Therneau T. The Cox Model., Dietz K, Gail M, Krickeberg K, Samet J, Tsiatis A, eds. In: Modeling Survival Data: Extending the Cox Model. New York: Springer-Verlag; 2000:44–48.Google Scholar
  • 22. Ken Rothman’s Episheet, Modern Epidemiology [online]. http://www.krothman.org/episheet.xls. Accessed January 30, 2017.Google Scholar
  • 23. Johannesdottir SA, Horváth-Puhó E, Ehrenstein V, Schmidt M, Pedersen L, Sørensen HT. Existing data sources for clinical epidemiology: The Danish National Database of Reimbursed Prescriptions.Clin Epidemiol. 2012; 4:303–313. doi: 10.2147/CLEP.S37587.CrossrefMedlineGoogle Scholar
  • 24. Homma S, Thompson JL, Qian M, Ye S, Di Tullio MR, Lip GY, et al.; WARCEF Investigators. Quality of anticoagulation control in preventing adverse events in patients with heart failure in sinus rhythm: Warfarin Versus Aspirin in Reduced Cardiac Ejection Fraction trial substudy.Circ Heart Fail. 2015; 8:504–509. doi: 10.1161/CIRCHEARTFAILURE.114.001725.LinkGoogle Scholar
  • 25. Melgaard L, Gorst-Rasmussen A, Lane DA, Rasmussen LH, Larsen TB, Lip GY. Assessment of the CHA2DS2-VASc score in predicting ischemic stroke, thromboembolism, and death in patients with heart failure with and without atrial fibrillation.JAMA. 2015; 314:1030–1038. doi: 10.1001/jama.2015.10725.CrossrefMedlineGoogle Scholar
  • 26. Adelborg K, Sundbøll J, Munch T, Frøslev T, Sørensen HT, Bøtker HE, et al.. Positive predictive value of cardiac examination, procedure and surgery codes in the Danish National Patient Registry: a population-based validation study.BMJ Open. 2016; 6:e012817. doi: 10.1136/bmjopen-2016-012817.CrossrefMedlineGoogle Scholar