Incidence, Trends, and Predictors of Ischemic Stroke 1 Year After an Acute Myocardial Infarction
Background and Purpose—Ischemic stroke after acute myocardial infarction is an important complication. It is unknown whether the risk has changed because the treatment of acute myocardial infarction has improved during the past decade. There is also conflicting data about predictors of stroke risk.
Methods—To obtain the 1-year incidence of stroke after acute myocardial infarction, the Register of Information and Knowledge about Swedish Heart Intensive Care Admissions database for the years 1998 to 2008 was merged with the Swedish National Patient Register (NPR). The time trend was studied by dividing the entire time period into 5 separate periods. Independent predictors were identified using a multivariable Cox proportional hazards regression model.
Results—Between 1998 and 2008, 7185 of 173 233 patients with acute myocardial infarction had an ischemic stroke within 1 year (4.1%). There was a 20% relative risk reduction during the study period (1998–2000 versus 2007–2008) relative risk 0.80 (95% confidence interval, 0.75–0.86; P<0.001. Independent predictors of stroke were age, female sex, ST-segment–elevation myocardial infarction, previous stroke, previous diabetes mellitus, heart failure at admission, angiotensin-converting enzyme inhibitor treatment and atrial fibrillation. Reperfusion treatment with fibrinolysis and percutaneous coronary intervention and treatment with aspirin, P2Y12-inhibitors, and statins predicted a reduced risk of stroke.
Conclusions—The risk of ischemic stroke within a year after myocardial infarction is substantial but has clearly been reduced during the studied time period. The major predictive factors found to correlate well with previous investigations. Reperfusion treatment, thrombocyte aggregation inhibition, and lipid lowering are the main contributors to the observed risk reduction.
Ischemic stroke is an important and potentially devastating complication after acute myocardial infarction (AMI). It is associated with great suffering for the patient and costs for society.1 Sustaining an ischemic stroke after an AMI also confers a significantly increased mortality risk.2–4 Despite the seriousness of this phenomenon, our knowledge is limited on the incidence, predictors, and trends over time of post-AMI ischemic stroke. Several investigators have studied the risk of stroke after AMI. However, many of these studies have been conducted in a clinical trial setting, which means that the studied patient populations were strongly selected and not representative of the general population. Many of these investigations have made no distinction between ischemic and hemorrhagic stroke.4–6 Furthermore, the incidence numbers vary widely among studies, a situation that has already been recognized by Witt et al,7 who conducted a meta-analysis in 2006. The 1-year incidence was estimated at 2.1%, of which 1.2% occurred within 30 days after AMI. From previous studies, we know that approximately one half of all ischemic strokes after AMI occur during the first few days,8 and most studies have focused on the in-hospital and 30-day incidence. Our group has recently studied the overall occurrence of ischemic stroke at 30 days after AMI during the period 1998 to 2008. It was 2.1% (95% confidence interval [CI], 2.0%–2.2%) and the occurrence decreased significantly from 2.2% to 2.0% during the study period.9 However, it is of considerable importance to know more about the risk of ischemic stroke after AMI in the longer run. The treatment of acute coronary syndromes (ACS) has undergone dramatic changes during the past decades, which have had positive effects on both morbidity and mortality. Most studies in the field were performed before the broad introduction of dual antiplatelet therapy, statins, angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor blockers, and percutaneous coronary intervention (PCI). In what way the improved treatment of ACS has affected the incidence of ischemic stroke after AMI has not yet been studied. In a registry study from 1997, Mooe et al8 observed a trend toward declining stroke rates after AMI. Another study that compared cerebrovascular events during the prethrombolytic era (1981–1983) versus the thrombolytic era (1992–1994) found no difference between the cohorts.6 A more recent investigation observed declining stroke rates during the 1980s, followed by increasing numbers during the 1990s until 1999, after which the incidence fell again.4 Finally, there is also conflicting data on the identified independent predictors associated with stroke risk after AMI.10
The primary purpose of this study was to make a solid estimate of the incidence of ischemic stroke during the first year after AMI. Second, we intended to evaluate whether the introduction of modern AMI treatment affected the risk of ischemic stroke after AMI. Finally, we aimed to identify independent predictors of ischemic stroke during the first year after an AMI reliably.
We obtained our study population from the Register of Information and Knowledge about Swedish Heart Intensive Care Admissions (RIKS-HIA), in which all patients treated for ACS in coronary care units (CCUs) in Sweden are being registered. Patient information is reported on case record forms that include >100 variables. The forms contain information about patient characteristics, diagnosis, medication, and procedures on admission, during hospitalization, and at discharge. Additional details about the RIKS-HIA database, now part of the Swedish Web-system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies (SWEDEHEART), have been described elsewhere.11 In 1998, 58 of 81 hospitals with CCUs participated in the RIKS-HIA registry. At the end of the study period in 2008, 74 of 74 hospitals with CCUs participated, covering ≈100% of Swedish CCU admissions.12 The RIKS-HIA registry is validated annually, which shows a conformity of 93% to 97% between RIKS-HIA records and patient records.13 In the present study, the patients first hospitalization for AMI was included.
Ischemic strokes were identified using the diagnosis codes from the International Classification of Diseases-Ninth Revision (433 and 434) and International Classification of Diseases-Tenth Revision (I63 and I64) for cerebral infarction in the Swedish National Patient Register (NPR), which contains diagnoses at discharge for all hospital stays in Sweden since 1987. The NPR has been validated against the Northern Swedish MONICA, which is a well-validated population-based epidemiological stroke register. The positive predictive value of all stroke events in the NPR was 88.1%. First-ever stroke events had a positive predictive value of 94%.14 The NPR contains patient-related data (personal identity number, sex, age, residence, and municipality), caregiver data (hospital and department), administrative data (admission/discharge dates and duration of admission), and medical data (primary and additional diagnoses, external causes of injury, or poisoning and procedures).15
To identify the occurrence of ischemic stroke within 1 year after AMI, the RIKS-HIA database was merged with the NPR. Specialists at the Department of Epidemiology within the National Board of Health and Welfare performed the amalgamation of the registries. Baseline clinical characteristics and medication on admission and at discharge were extracted from the merged database. All patients for whom data were entered into the RIKS-HIA were informed of their participation in the registry (patients could request to be excluded from the registry) and the long-term follow-up. The National Board of Health and Welfare and the Swedish Data Inspection Board approved the registry. The local ethics committee approved the merging of the registries.
Continuous variables are presented as mean values. Certain continuous variables on baseline characteristics, such as body mass index, blood pressure, serum lipids, and kidney function, had substantial amounts of missing data in the beginning of the study period. For this reason, they are not displayed. Categorical data had no variables with <90% valid cases, and are shown as percentages. Data for patients with and without stroke were summarized as means or percentages and were tested for differences with the t test or the χ2 test, as appropriate.
To study the time trend in the incidence of ischemic stroke after AMI, we divided the entire time period (1998–2008) into 5 separate periods: 1998 to 2000, 2001/2002, 2003/2004, 2005/2006, and 2007/2008. Kaplan–Meier curves were calculated for the occurrence of ischemic stroke within 1 year after AMI during the 5 different time intervals and compared between groups using the log-rank test.
Independent predictors of ischemic stroke after AMI were identified using a Cox proportional hazards regression model. Candidate variables were included in the model if they had proven to predict risk in previous investigations or if they proved to predict risk in a univariate analysis (P<0.25). A total of 150 562 patients were included in the final regression model and only variables with >95% valid cases were included in the multivariate analysis. Patients who died during hospitalization were excluded. Statistical analysis was performed in SPSS versions 19 and 22.
A total of 173 233 subjects with AMI were registered in the RIKS-HIA between 1998 and 2008. Of these, 7185 (4.1%) sustained an ischemic stroke within 1 year (95% CI, 4.01%–4.19%).
Baseline characteristics are presented in Table 1. Patients who experienced an ischemic stroke within 1 year after AMI were older and more often women when compared with nonstroke patients, 76.5 versus 70.9 years and 43.9% versus 36.5%, respectively.
Stroke patients more often had a history of a previous AMI and ischemic stroke. They also more often had previous heart failure, diabetes mellitus, and peripheral artery disease. However, the differences between the groups on previous diagnoses of renal failure, dialysis treatment, and chronic obstructive pulmonary disease were not statistically significant. During hospitalization, the incidence of atrial fibrillation was almost doubled in patients with stroke versus nonstroke patients. Clinical signs of heart failure (characterized by the finding of pulmonary rales and use of intravenous diuretics) were also more common among patients with stroke. Nonstroke patients were subject to reperfusion treatment with PCI or fibrinolytic treatment more frequently than patients with stroke; however, there was no difference in in-hospital coronary artery bypass graft surgery between the groups.
Table 2 shows baseline characteristics of patients during 5 consecutive time intervals from Table 1. Although previous AMI was declined in both groups during all 5 periods, this finding is explained by the study inclusion criteria (only the first MI event in the registry was included). Previous ischemic stroke remained on the same level. Although atrial fibrillation during hospitalization remained stable in both groups, congestive heart failure during hospitalization also decreased in both groups, which might be explained by improved AMI treatment. PCI increased in both groups, which is attributable to the technical development and establishment of the procedure as the standard of care for ACS in Sweden during the studied time period. Consequently, fibrinolysis has decreased gradually in both groups because PCI has become more important. The incidence of in-hospital coronary artery bypass graft surgery increased over time in both groups.
Tables I and II in the online-only Data Supplement display medication on admission and at discharge in 173 233 patients with AMI during 5 consecutive time intervals. On admission, medication with aspirin, β-blockers, and ACEi remained relatively constant during the studied time period in both groups. All 3 treatments were clearly more common among patients with stroke. In both groups, statin treatment gradually more than doubled during the studied time period and was slightly more common among patients with stroke. At discharge, treatment with aspirin, β-blockers, statins, and ACEi gradually increased in both groups. Similar to medication on admission, statin treatment at discharge gradually nearly doubled during the study period. It was more likely for nonstroke patients to receive these treatments at discharge.
Kaplan–Meier analysis of survival without ischemic stroke revealed a risk reduction over time (Figure 1). The rates of ischemic stroke within 1 year of AMI for each time interval were 4.5% (1998–2000; 95% CI, 0.043–0.047), 4.4% (2001/2002; 95% CI, 0.042–0.046), 4.1% (2003/2004; 95% CI, 0.039–0.043), 3.9% (2005/2006; 95% CI, 0.037–0.041), and 3.8% (2007/2008; 95% CI, 0.036–0.040; overall log-rank test, <0.0001).
In a univariate Cox regression model, the most recent time interval (2007–2008) was associated with a 20% relative risk reduction for stroke within 1 year after AMI when compared with the first time interval (1998–2000; relative risk, 0.80; 95% CI, 0.75–0.86; P<0.001]. A reduced risk of stroke compared with the first time interval was also observed during the third (2003–2004) and fourth (2005–2006) time intervals (relative risk, 0.89; 95% CI, 0.83–0.95; P=0.001 and 0.84; 95% CI, 0.75–0.86; P<0.001, respectively).
The result of the multivariate Cox regression analysis of predictors of ischemic stroke within 1 year after AMI is summarized in Table 3. Age, female sex, ST-segment–elevation myocardial infarction, previous stroke, previous diabetes mellitus, heart failure at admission, atrial fibrillation, and ACEi treatment at discharge were independently associated with an increased risk of ischemic stroke. Reperfusion treatment with fibrinolysis and PCI, and pharmacological treatment with aspirin, P2Y12-inhibitors, and statins were associated with a reduced risk.
Figures 2 and 3 display how the protecting factors have changed over time. There is a sharp increase in the treatment with statins and P2Y12-inhibitors over the study period. ACEi treatment also increased but not at the same rate. Aspirin treatment at discharge increased slightly from an already high level at the beginning of the study period. PCI has increased markedly over the study period and has become the most important reperfusion treatment in AMI, whereas fibrinolysis has decreased as PCI gained ground.
A history of previous ischemic stroke stands out as a major risk marker for ischemic stroke after AMI (hazard ratio, 2.59; 95% CI, 2.45–2.75; P<0.001) in the multivariate Cox regression model. Therefore, a sensitivity analysis, in which subjects with a history of previous ischemic stroke were excluded, was performed (Table III in the online-only Data Supplement). In this analysis, patients using aspirin or P2Y12-inhibitors at discharge had some additional risk decrease, but the results were consistent with the main analysis.
This is the first study with a sample size large enough to analyze a potential trend of ischemic stroke rate after AMI over time. Our results indicate that the relative risk of experiencing an ischemic stroke has decreased by 20% during the studied time period. The observed risk reduction coincides with the improved outcome after AMI during the past decades and with the increased use of evidence-based treatment.
The incidence observed in this study (4.1% at 1 year) is nearly twice as high as the incidence found observed in the 2006 meta-analysis by Witt et al7 (2.1% at 1 year), likely because the present study has a more representative patient population. This likelihood is supported by the high mean age of the study population (71.2 years). The estimate is precise, as shown by the narrow CI (CI ±0.09%).
In our multivariate regression analysis, reperfusion treatment with PCI and fibrinolysis was associated with reduced stroke risk. PCI has previously been associated with reduced stroke risk in the Valsartan in myocardial infarction (VALIANT) trial; however, in a post hoc analysis, this association was considered a significant selection bias.16 Although early revascularization has previously been thought to reduce stroke risk, the significance was lost after adjustment for other factors.17 In our study, PCI had the strongest statistical association with a reduced occurrence of ischemic stroke after AMI. Because PCI is an invasive treatment, it has traditionally been thought to be associated with an increased risk of embolic stroke. However, PCI has the potential to reduce the size of damaged myocardium and thereby reduce the risks of heart failure and atrial fibrillation, each of which are associated with an increased risk of ischemic stroke. However, there could be a potential selection bias because elderly patients with comorbidities are more often treated with a noninvasive strategy. Furthermore, one might speculate whether invasively treated patients are generally more likely to be treated according to recommended guidelines because of more careful follow-up and better compliance to treatment.
Prescription of aspirin, statins, and P2Y12-inhibitors at discharge were all associated with a reduced risk of ischemic stroke after AMI. The proportion of patients receiving aspirin, statins, and P2Y12-inhibitors increased markedly during the study period, which coincided with the observed risk reduction of stroke. These drugs are known to reduce the risk of both death and reinfarction in patients with AMI. In the Second International Study of Infarct Survival (ISIS-2), aspirin reduced the risk of stroke, which agrees well with our finding.18 In the Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering (MIRACL) study, which explored the effects of atorvastatin on early recurrent ischemic events in ACS, atorvastatin was shown to reduce the risk of nonfatal stroke in patients with AMI, which supports the finding that statin use is associated with stroke risk reduction.19 However, stroke reduction has not previously been demonstrated with P2Y12-inhibitors. Clopidogrel is well documented as a secondary preventive treatment after an ACS without ST-segment elevation based on the Clopidogrel in Unstable Angina to Prevent Recurrent Events (CURE) study20; however, it did not reduce stroke risk in this category of patients, most of whom did not have a verified AMI. In the Clopidogrel and Metoprolol in Myocardial Infarction Trial/Second Chinese Cardiac Study (COMMIT/CCS-2), patients treated with clopidogrel experienced a nonsignificant reduction of ischemic stroke.21 Furthermore, treatment with clopidogrel in the Clopidogrel as Adjunctive Reperfusion Therapy-Thrombolysis in Myocardial Infarction 28 (CLARITY-TIMI 28) trial also exhibited a nonsignificant trend toward a reduced stroke risk.22 These insignificant findings might be explained by the lack of a loading dose in COMMIT/CCS-2 and a lack of statistical power in CLARITY-TIMI 28. As a secondary preventive treatment after ischemic stroke, clopidogrel has been shown to be equivalent to the combination of aspirin and dipyridamol in the aspirin and extended-release dipyridamole versus clopidogrel for recurrent stroke In the Prevention Regimen For Effectively Avoiding Second Strokes (PRoFESS) trial, which examined aspirin and extended-release dipyridamole versus clopidogrel for recurrent stroke, clopidogrel was equivalent to the combination of aspirin and dipyridamole when used as a secondary preventive treatment after ischemic stroke.23
The independent predictors of ischemic stroke after AMI observed in this study overlap partly with results from previous investigations.3,4,7,8,24 It should now be uncontroversial to consider age, sex, atrial fibrillation, diabetes mellitus, previous stroke, STEMI, and heart failure as independent predictors of ischemic stroke after AMI. In support of our findings, STEMI has been associated with increased stroke risk after AMI in ≥2 previous investigations.3,4 STEMI can be considered a marker of infarction severity, which explains the increased stroke risk. ACEi treatment at discharge was also associated with increased stroke risk, plausibly explained by an association with heart failure.
Among the predictors associated with increased risk of stroke, the occurrence of atrial fibrillation and heart failure during hospitalization decreased over time, which might have contributed to the reduced stroke risk. Interestingly, the reduced rate of atrial fibrillation over time observed in the present study with an AMI population is in contrast to the 2.5-fold increased proportion of atrial fibrillation over time among patients primarily hospitalized because of stroke.25 The explanation of the observed reduced rate of heart failure during hospitalization is multifactorial. Improved reperfusion treatment likely plays an important role. However, it should not be forgotten that the proportion of STEMI has diminished during the study period, leading to less severe myocardial damage.
The RIKS-HIA only includes patients with AMI hospitalized in CCUs. The registry, therefore, contains ≈90% of patients aged <80 years and 65% of patients aged ≥80 (compared with the NPR, which contains all patients with a hospital diagnosis of AMI).26 This might lead to an underestimation of the stroke rate because patients with AMI outside a CCU, based on our findings in this study and previous studies, are exposed to greater stroke risk than patients admitted to a CCU.
Some of the variables in the analyses had a low number of valid cases. This was especially true in the early time intervals for heart rate, blood pressure, P-glucose, and angiotensin receptor blockers, and throughout the entire period for body mass index, P-cholesterol, and P-low-density lipoprotein. When trying to establish independent risk factors with a Cox multivariate model, there is always the risk that confounders will not be taken into account because they were not recorded as variables in the database.
The risk of ischemic stroke within 1 year after AMI is substantial but has decreased over time. Older age, female sex, STEMI, previous stroke, diabetes mellitus, heart failure during hospitalization, and atrial fibrillation increase the risk of ischemic stroke, whereas reperfusion treatment with either PCI or fibrinolysis, as well as antiplatelet and lipid-lowering drugs, reduces the risk.
Sources of Funding
This study was granted financial support from the Research and Development Unit in Jämtland.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.005770/-/DC1.
- Received April 9, 2014.
- Revision received August 22, 2014.
- Accepted August 22, 2014.
- © 2014 American Heart Association, Inc.
- Budaj A,
- Flasinska K,
- Gore JM,
- Anderson FA Jr,
- Dabbous OH,
- Spencer FA,
- et al
- Tanne D,
- Gottlieb S,
- Hod H,
- Reicher-Reiss H,
- Boyko V,
- Behar S
- Mooe T,
- Eriksson P,
- Stegmayr B
- Kajermo U,
- Ulvenstam A,
- Modica A,
- Jernberg T,
- Mooe T
- Jernberg T,
- Attebring MF,
- Hambraeus K,
- Ivert T,
- James S,
- Jeppsson A,
- et al
- Sampson UK,
- Pfeffer MA,
- McMurray JJ,
- Lokhnygina Y,
- White HD,
- Solomon SD
- Van de Graaff E,
- Dutta M,
- Das P,
- Shry EA,
- Frederick PD,
- Blaney M,
- et al
- Schwartz GG,
- Olsson AG,
- Ezekowitz MD,
- Ganz P,
- Oliver MF,
- Waters D,
- et al
- Yusuf S,
- Zhao F,
- Mehta SR,
- Chrolavicius S,
- Tognoni G,
- Fox KK
- Chen ZM,
- Jiang LX,
- Chen YP,
- Xie JX,
- Pan HC,
- Peto R,
- et al
- Wienbergen H,
- Schiele R,
- Gitt AK,
- Schneider S,
- Heer T,
- Gottwik M,
- et al
- Jernberg T