Objectives To evaluate 90-day cardiovascular outcome in patients after myocardial infarction (MI) in relation to different subtypes of atrial fibrillation (AF) and MI.
Methods We studied 155 071 hospital survivors of MI between 2000 and 2009 in Swedish registries. AF subtypes were defined according to history of AF and in-hospital ECG recordings. Clinical outcomes were evaluated with multivariable Cox models.
Results AF was documented in 24 023 (15.5%) cases. The AF subtypes were new-onset AF with sinus rhythm at discharge (3.7%), new-onset AF with AF at discharge (3.9%), paroxysmal AF (4.9%) and chronic AF (3.0%). The event rate per 100 person-years for the composite cardiovascular outcome (all-cause mortality, MI or ischaemic stroke) was 90.9 in patients with any type of AF versus 45.2 in patients with sinus rhythm, adjusted hazard ratio with 95% CI (HR) 1.28 (1.19 to 1.37). There were no significant differences in the composite cardiovascular outcome between AF subtypes. AF was associated with higher risk of mortality, HR 1.59 (1.41 to 1.80), reinfarction, HR 1.14 (1.05 to 1.24), and ischaemic stroke, HR 2.29 (1.92 to 2.74), respectively. In subgroup analysis, AF was associated with a higher risk of composite cardiovascular outcome in the non-ST-elevation myocardial infarction (NSTEMI) and ST-elevation myocardial infarction (STEMI) cohort, HR 1.24 (1.13 to 1.36) and HR 1.34 (1.21 to 1.48), respectively, with p value for interaction=0.23.
Conclusions AF is common in the setting of MI and is associated with a higher risk of composite cardiovascular outcome and the individual components; mortality, reinfarction and ischaemic stroke, respectively. No major difference in outcome was observed between AF subtypes. No difference in outcome for AF was observed between the NSTEMI and STEMI cohort.
Statistics from Altmetric.com
Atrial fibrillation (AF) is the most frequently sustained arrhythmia with an increasing incidence due to ageing population.1 AF is a common finding in the setting of myocardial infarction (MI) with the incidence reported to vary between 6% and 21%.2 Previous studies have shown that the occurrence of AF in the setting of MI is associated with an increased risk of mortality and ischaemic stroke.3–7 However, there is limited knowledge on the occurrence and the outcome associated with different subtypes of AF in the setting of MI, which is of importance when estimating prognosis of individual patients. In addition, there is limited knowledge regarding the association between AF and the subtype classification of MI into non-ST-elevation myocardial infarction (NSTEMI) and ST-elevation myocardial infarction (STEMI). We aimed to study the occurrence of different subtypes of AF in the setting of MI and the association with 90-day all-cause mortality, MI or ischaemic stroke. We also evaluated outcome in AF patients with NSTEMI and STEMI.
This is a retrospective observational study including all consecutive Swedish patients admitted due to MI in all 72 coronary care units between January 2000 and December 2009. All patients admitted to coronary care units in Sweden are reported in the Swedish Web-system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies register (SWEDEHEART). Key patient characteristics on admission and variables registered during the hospital stay and at discharge are recorded. The quality of data has been monitored yearly with results showing a 96.1% agreement between the data entered in the registry and the patient records.8
Data about mortality were obtained by data linking the SWEDEHEART registry with the Swedish Cause of Death Registry. Information about readmission and history of diabetes mellitus, hypertension, ischaemic stroke, chronic obstructive pulmonary disease, congestive heart failure, MI, peripheral vascular disease, renal disease, liver disease, bleeding events, dementia and cancer, based on the International Classification of Diseases, 10th revision, were obtained by data linking SWEDEHEART with the National Patient Registry (NPR); see online supplementary tables S1–S4. The Swedish Cause of Death Registry includes data on mortality in Sweden from 1961 and the NPR includes diagnoses on all patients hospitalised in Sweden from 1987 and forward. These registries are part of the Swedish National Board of Health and have been shown to have high validity.9 ,10 Data linkage of these registries with SWEDEHEART was approved and performed by the National Board of Health and Welfare using the unique civic registration number of each Swedish citizen.
Diagnosis of AF was based on physician interpretation of ECG on arrival and at discharge. Data on previous diagnosis of AF, preceding the current hospital admission, were collected in the NPR. Patients were categorised into four groups: new-onset AF with sinus rhythm at discharge; new-onset AF with AF at discharge; paroxysmal AF and chronic AF. New-onset AF with sinus rhythm at discharge was defined as no medical history of AF according to NPR with AF on arrival ECG but sinus rhythm on discharge ECG. New-onset AF with AF at discharge was defined as no medical history of AF in NPR with AF on the discharge ECG, regardless of arrival ECG. Paroxysmal AF was recorded in cases with a known medical history of AF, but with sinus rhythm on either the arrival or discharge ECG. Chronic AF was documented in patients with a history of AF and with AF on both the arrival and discharge ECG.
The primary end point was a composite cardiovascular outcome including all-cause mortality, MI or ischaemic stroke within 90 days from discharge. Results were also analysed for the following individual outcomes within 90 days from discharge; all-cause mortality, MI and ischaemic stroke, respectively. Patients were censored for death when analysing the individual outcomes; MI or ischaemic stroke. No other censoring scheme was applied. Outcome definitions are presented in online supplementary table S4.
Descriptive statistics for baseline characteristics are presented in a tabular format with continuous variables as a median with interquartile range (IQR) and categorical variables as percentages. Kaplan–Meier estimates were reported to illustrate outcome according to subtype classification of AF. Unadjusted and adjusted Cox regression analyses were calculated to identify the relation between occurrence of AF and outcome. In the adjusted Cox-regression analysis, clinical relevant variables included in the CHA2DS2-VASc scoring system11 were individually accounted for and included the following: congestive heart failure, hypertension, age, diabetes mellitus, prior ischaemic stroke, transient ischaemic attack and systemic embolism, vascular disease, gender, and also admission year, hospital (as a γ distributed random frailty effect), in-hospital revascularisation, anticoagulants and antiplatelets at discharge. Continuous variables entered the models as cubic splines to allow for violation of the linearity assumption. Due to enrichment of data from the NPR, no data were missing when adjusting for differences in baseline characteristics. Diagnostics for the proportional hazards assumption was done using Schoenfeld residuals with no significant violations of the assumption observed. Finally, a subgroup analysis was performed on the 69 919 patients undergoing percutaneous coronary intervention (PCI) to assess the association between different subtypes of AF and outcomes in this population.
Results were stated statistically significant for a two-sided p value <0.05. Analyses in this study were done using R statistical program V.2.15.9.
Selection of the study populations is summarised in figure 1. In a primary selection, we included all subjects at their first admission due to MI between 2000 and 2009, n=190 625. As this study focused on long-term outcome of AF in the setting of MI, only the 178 761 hospital survivors were included. Patients with an age <30 years or with missing arrival or discharge ECG were excluded, leaving 155 071 subjects in the present analysis.
AF was recorded in 24 023 (15.5%) patients. The most common type of AF was new-onset AF, n=11 742 (7.6%). About half of these patients had sinus rhythm at discharge, n=5769 (3.7%). The other half had AF at discharge, n=5973 (3.9%). Paroxysmal AF was present in 7633 (4.9%) and chronic AF was recorded in 4648 (3.0%) of the patients.
Baseline characteristics in relation to subtype classification of AF are presented in table 1. Patients with AF, compared with cases without, were more likely to be older and more often female. Among patients with AF, age was highest in those with chronic AF and lower in those with new-onset or paroxysmal AF. Patients with AF were more likely to have a history of diabetes mellitus, hypertension, ischaemic stroke, chronic obstructive pulmonary disease, congestive heart failure, MI, peripheral vascular disease, renal disease, dementia or cancer. The likelihood of these concomitant diseases was higher in patients with chronic or paroxysmal AF than new-onset AF.
In relation to classification of MI into NSTEMI and STEMI, patients with AF were more likely to have NSTEMI (68.9%) than STEMI (31.1%). Invasive strategies such as coronary angiography and PCI were less often conducted on patients with AF compared with those with sinus rhythm.
Patients with diagnosis of AF were more likely to be discharged with vitamin K antagonists than patients with sinus rhythm. However, patients with AF less often received antiplatelet drugs. Patients with chronic AF were more often discharged with vitamin K antagonists compared with the other subtypes of AF. Also, patients with new-onset AF and with sinus rhythm at discharge were less likely to receive vitamin K antagonists at discharge than the other subtypes of AF.
The crude event rate per 100 person-years, unadjusted and adjusted HRs with 95% CI (HR) for all outcomes are presented in tables 2 and 3. The event rates are also presented as Kaplan–Meier curves (figures 2⇓⇓–5). The 90-day event rate per 100 person-years of the composite cardiovascular outcome (all-cause mortality, MI or ischaemic stroke) was 90.9 in patients with any type of AF versus 45.2 in patients with sinus rhythm, HR 1.28 (1.19 to 1.37). Similar results for the composite cardiovascular outcome were seen in the NSTEMI and STEMI cohort (p value for interaction=0.23). The highest event rate for the composite cardiovascular outcome was seen in patients with chronic AF followed by new-onset AF with AF at discharge. Lower rates, but still higher than with sinus rhythm, were seen for cases with paroxysmal or new-onset AF with sinus rhythm at discharge. After adjustments for confounding factors, all subtypes of AF were significantly associated with higher risk of the composite cardiovascular outcome than sinus rhythm. However, after multivariable adjustment, no significant difference was seen between the different subtypes of AF.
For the individual end point of all-cause mortality, there was a significantly higher event rate for patients with AF versus sinus rhythm, HR 1.59 (1.41 to 1.80). After adjustment for confounders, AF was associated with similar risk of all-cause mortality in both the NSTEMI and STEMI cohort (p value for interaction=0.21). All subtypes of AF had a significantly higher risk of all-cause mortality than patients with sinus rhythm. The highest event rate of all-cause mortality was seen in patients with chronic AF and new-onset AF with AF at discharge. However, after adjustment for potential confounding factors, only new–onset AF with AF at discharge remained significantly associated with higher risk of all-cause mortality compared with the other subtypes of AF.
The 90-day event rate per 100 person-years for MI was 59.5 in patients with AF compared with 35.9 in patients with sinus rhythm. In a multivariable Cox proportional hazards model, AF compared with sinus rhythm was associated with an increased risk of reinfarction, HR 1.14 (1.05 to 1.24). The adjusted HR was similar for the NSTEMI and STEMI cohort (p value for interaction=0.78). In addition, paroxysmal AF versus sinus rhythm was significantly associated with higher risk of MI within 90 days, HR 1.15 (1.02 to 1.29). A similar finding, with borderline significance, was seen for new-onset AF and chronic AF. No difference was found across the different subtypes of AF.
The event rate for ischaemic stroke within 90 days was higher among patients with AF than with sinus rhythm. Compared with sinus rhythm, the HR for ischaemic stroke for patients with AF was 2.29 (1.92 to 2.74). A similar HR for ischaemic stroke was seen in the NSTEMI and STEMI cohort (p value for interaction=0.52). All subtypes of AF were significantly associated with higher risk of ischaemic stroke than sinus rhythm. The rate of ischaemic stroke was highest among patients with chronic AF than other subtypes of AF, but after adjustment for concomitant factors, no significant difference remained between the different subtypes of AF.
In a subgroup analysis on patients undergoing PCI (n=68 919), similar results as in the primary analysis were found for the composite cardiovascular outcome, all-cause mortality, MI and ischaemic stroke (see online supplementary tables S5 and S6).
In this study, we showed that AF is an arrhythmia occurring in around 16% of patients with MI. Compared with patients with sinus rhythm, patients with AF have an approximately 30% increased risk of a composite cardiovascular outcome (all-cause mortality, MI or ischaemic stroke) within 90 days from discharge. Also, all subtypes of AF were associated with a higher risk of all-cause mortality and ischaemic stroke, with no major difference between AF subtypes. No difference in outcome for AF was observed between the NSTEMI and STEMI cohort.
This is one of the largest studies analysing AF in the setting of MI. Previous studies have shown that AF is common in patients with MI.3 ,12 ,13 Few studies have reported the distribution between AF subtypes in the context of MI. The most common type of AF in this study was new-onset AF accounting for almost half of the AF population followed by paroxysmal and chronic AF. Other, smaller studies have reported previous AF to be more common than new-onset AF during MI.6 ,14 AF in this study was also more common in patients with NSTEMI than in patients with STEMI. These results differ from other studies that have found AF in patients with MI to be more common in STEMI or similar between the STEMI and NSTEMI cohorts.7 ,15 However, it seems reasonable that compared with STEMI, cases with NSTEMI more frequently have AF as these patients are older and have more underlying comorbidities,16 which might contribute to the risk of developing AF.
Acute in-hospital therapy differed between patients with and without AF. Acute reperfusion treatment was less often used in patients with AF, which might be related to differences in age, more comorbidity and concomitant treatments, for example, vitamin K antagonists at entry. Considering that reperfusion treatment is associated with lower risk of mortality after MI, regardless of AF,17 and the larger absolute benefits at higher risk, it is concerning that these treatments were underused in the AF population.
Compared with patients with sinus rhythm, patients with AF were more likely to have concomitant diseases. Despite having a higher risk, the occurrence of AF in patients with MI was independently associated with an increased risk of the composite cardiovascular outcome and the individual outcomes; all-cause mortality and ischaemic stroke, respectively. No difference in cardiovascular risk was seen between the different subtypes of AF. This is in accordance with several previous studies that have shown all-cause mortality and stroke rates to be higher in patients with MI suffering from AF.3 ,5–7 ,18 ,19 In a study by Wong et al,4 new AF was reported in patients with sinus rhythm at enrolment in whom AF developed within 30 days, with new AF resulting in a 64% increased risk of 1-year mortality. In a study by Lehto et al,19 they defined new-onset AF as AF seen on ECG after randomisation but not before and showed an 82% increased risk for mortality during a median 3-year follow-up for patients with new-onset AF than those without. For paroxysmal AF, an increased risk for mortality has been reported in a study by Eldar et al,17 in which patients with paroxysmal AF had a 32% increased risk of mortality within 1 year. These results on mortality are in accordance with those from our study and might reflect on AF as a marker for illness and structural heart disease.20
Few studies have reported stroke risk based on different subtypes of AF in the setting of MI. In a study by Wong et al, new-onset AF was associated with higher risk of stroke, which was also seen in a study by Lopes et al.19 ,21 A small retrospective study by Siu et al22 also showed that patients with in-hospital AF and sinus rhythm at discharge had a higher risk for stroke within 1 year from discharge, which is in line with our results. This implies that a careful evaluation of antithrombotic medication at discharge must be considered for all patients with AF, including those with sinus rhythm at discharge. Despite the known benefit of oral anticoagulants in patients with AF,23 less than a quarter of the patients in our study received oral anticoagulants at discharge.
We also showed that AF in the setting of MI was associated with a 14% higher risk of reinfarction. This was also seen in patients with paroxysmal AF and with a borderline significance in patients with new-onset and chronic AF. Previous studies on AF in the setting of MI,5–7 and recent studies by Soliman et al24 ,25 on patients without history of coronary heart disease, have shown a higher risk of MI in patients with AF than those without. Possible explanations supporting the increased risk of MI in patients with AF might be that both conditions share similar pathophysiological pathways and that AF might be a marker for illness, inflammation and structural heart disease.20 ,26 ,27
Finally, we analysed outcome in relation to AF versus sinus rhythm in an NSTEMI and STEMI cohort. In this analysis, we found that AF in both cohorts was associated with a higher risk of composite cardiovascular outcome, all-cause mortality, reinfarction and ischaemic stroke, with no significant interaction seen between outcome and the two different cohorts. This comparison has to our knowledge only been done in a study by Lopes et al,7 in which they found a higher risk of outcome in patients with AF and NSTEMI than STEMI. Compared with that study, our population consists of real-life patients with a larger proportion of patients with NSTEMI (62.9% vs 30.2%). Also, in our population there is higher occurrence of AF (15.5% vs 7.5%). These differences in populations might be attributed to the different outcomes observed.
There are inherent limitations in retrospective observational studies. Registries running as part of routine care can only contain a limited number of variables. Therefore, unmeasured confounders may not be accounted for in the adjusted analysis. Due to incomplete data regarding ECG records, a number of cases had to be excluded from the final analyses, meaning that the study population might not be representative for the whole MI population. Another limitation is the definition of AF, which in this study was based on information about previous diagnosis of AF and physician interpretation of ECG on arrival and at discharge. This might attribute to some patients who before admission had an undiagnosed AF being misclassified into the new-onset AF group. Also, due to no information regarding follow-up on AF status, some patients with AF at discharge, especially cases with new-onset AF and AF at discharge, might actually have a persistent or permanent AF. However, repeated monitoring of the SWEDEHEART registry has shown high agreement between the data entered and the patient records.8
In conclusion, AF is a common arrhythmia in the setting of MI and independently associated with poor outcome in regard to a composite cardiovascular outcome of all-cause mortality, MI or ischaemic stroke and to the individual components. No major differences in outcome were observed between the AF subtypes or in patients with NSTEMI and STEMI. Oral anticoagulation was underused in patients with AF, with less than one-quarter of the patients receiving warfarin at discharge.
What is already known on this subject?
Previous studies have suggested that atrial fibrillation is present in 6–21% of patients with acute myocardial infarction and associated with an increased risk of mortality and ischaemic stroke after myocardial infarction.
What might this study add?
Atrial fibrillation was found in 15.5% out of 155 071 patients hospitalised for myocardial infarction in Sweden. Compared with those with sinus rhythm, patients with atrial fibrillation have approximately 30% higher risk of a combined cardiovascular end point of all-cause mortality, myocardial infarction or ischaemic stroke, with similar associations seen for different subtypes; new-onset, paroxysmal or chronic atrial fibrillation. Similar outcome estimates were seen among the subgroups of patients with non-ST-elevation myocardial infarction or ST-elevation myocardial infarction.
How might this impact on clinical practice?
This paper has an implication for a broad group of patients in the setting of myocardial infarction. The study indicates that all subtypes of atrial fibrillation, regardless of classification of myocardial infarction, are associated with a higher risk of cardiovascular events. The presence of any type of atrial fibrillation in the setting of myocardial infarction should be considered when estimating the prognosis and considering antithrombotic medication for the individual patient.
The authors thank all the staff in all coronary care units in Sweden for their help and cooperation in contributing data to the SWEDEHEART registry. All participating centres and the details of SWEDEHEART are presented on the registry's website (http://www.ucr.uu.se/swedeheart).
Contributors All authors fulfil all the guidelines for authorship: (1) substantial contributions to conception and design, acquisition of data or analysis and interpretation of data; (2) drafting the article or revising it critically for important intellectual content and (3) final approval of the version to be published.
Funding This work was supported by a grant from the Swedish Foundation for Strategic Research, Stockholm, Sweden (grant number KF10-0024 to Tailoring of treatment in all comers with AMI).
Competing interests None declared.
Patient consent All patients were informed of their participation in SWEDEHEART and the long-term follow-up and had the right to be excluded.
Ethics approval Ethics Committee of Karolinska Institutet.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.