Article Text
Abstract
Objective: To assess variables associated with the occurrence of atrial fibrillation (AF) and the relation of AF with short- and long-term outcomes and with other in-hospital complications in patients with acute coronary syndromes (ACS) with and without ST-segment elevation.
Design: Pooled database of 120 566 patients with ST-segment elevation myocardial infarction (STEMI) and non-ST-segment elevation (NSTE) ACS enrolled in 10 clinical trials. Multivariable logistic regression and Cox proportional hazards modelling were used to identify factors associated with AF and its relation with clinical outcomes.
Setting: ACS complicated by AF.
Patients: 120 566 patients with STEMI and NSTE-ACS in 10 clinical trials.
Interventions: None evaluated.
Main outcome measure: Short- and long-term mortality.
Results: Occurrence of AF was 7.5% in the overall population (STEMI 8.0% (n = 84 161); NSTE-ACS = 6.4% (n = 36 405)). Seven-day mortality was higher for patients with AF (5.1%) than for those without (1.6%). After adjusting for confounders, association of AF with 7-day mortality was present in STEMI (hazards ratio (HR) = 1.65; 95% CI 1.44 to 1.90) and NSTE-ACS (HR = 2.30; 95% CI 1.83 to 2.90; p interaction = 0.015). Risk of long-term mortality (day 8 to 1 year) was also higher in STEMI (HR = 2.37; 95% CI 1.79 to 3.15) and NSTE-ACS (HR = 1.67; 95% CI 1.41 to 1.99). AF had a larger impact in NSTE-ACS on risk of short-term mortality (p<0.001), stroke (p<0.001), ischaemic stroke (p<0.001) and moderate or severe bleeding (p<0.001).
Conclusions: AF is more common in patients with STEMI. An association of AF with short- and long-term mortality among patients with STEMI and NSTE-ACS was found. Understanding these findings may lead to better care of patients with this common arrhythmia.
Statistics from Altmetric.com
Atrial fibrillation (AF) is the most common cardiac arrhythmia seen in clinical practice.1 It is also a frequent complication of acute myocardial infarction (MI), with an incidence ranging between 5% and 23%.2–4 Some studies have identified increased in-hospital and long-term mortality associated with AF,3–7 while others have not.2 8 9
Although most of the studies of AF in MI have been done in patients presenting with ST-segment elevation (STEMI) acute coronary syndrome (ACS), some studies have targeted ACS patients without ST-segment elevation.3 7 10 There is little information about the association of AF with clinical outcomes across the spectrum of ACS, including both ST-segment elevation and non-ST-segment elevation groups.
We hypothesised that baseline characteristics and outcomes associated with the occurrence of AF differ among patients with STEMI and patients with non-ST-segment elevation (NSTE) ACS. We analysed the variables associated with the occurrence of AF and the relation of AF with in-hospital complications and short- and long-term outcomes in patients with ACS according to whether or not ST-segment elevation was present on the presenting ECG.
METHODS
We conducted these analyses using a pooled database of 120 566 patients with ACS enrolled in clinical trials in which information on AF as an in-hospital complication was available. The STEMI population (n = 84 161) comprised patients from the following trials: Global Utilization of Streptokinase and t-PA For Occluded Coronary Arteries (GUSTO I) (48.5% of STEMI patients),11 Global Use of Strategies to Open Occluded Arteries in Acute Coronary Syndromes (GUSTO IIb) (4.9%),12 Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO III) (17.9%),13 Assessment of the Safety and Efficacy of a New Thrombolytic -2 (ASSENT 2) (19.5%),14 Assessment of the Safety and Efficacy of a New Thrombolytic Regimen-3 (ASSENT 3) (7.3%)15 and Assessment of the Safety and Efficacy of a New Thrombolytic Regimen-3 Plus (ASSENT 3 Plus) (1.9%).16
The NSTE-ACS population (n = 36 405) comprised patients from the following trials: Global Use of Strategies to Open Occluded Arteries in Acute Coronary Syndromes (GUSTO IIb) (22%),12 Platelet IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin Therapy (PURSUIT) (30%),17 Platelet IIb/IIIa Antagonist for the Reduction of Acute Coronary Syndrome Events in a Global Organization Network A (PARAGON-A) (6.3%),18 Platelet IIb/IIIa Antagonist for the Reduction of Acute Coronary Syndrome Events in a Global Organization Network B (PARAGON-B) (14.3%),19 and Superior Yield of the New Strategy of Enoxaparin, Revascularization and Glycoprotein IIb/IIIa Inhibitors (SYNERGY) (27.4%).20
AF was defined as an irregularly irregular rhythm with lack of discernible P waves on an ECG any time during the hospitalisation after randomisation. It was identified as a complication on the patient’s case report form.
Statistical methods
Baseline characteristics are presented as medians with 25th and 75th percentiles for continuous variables and frequencies for categorical variables. Categorical variables were compared using the Pearson χ2 test and continuous variables were compared using the Wilcoxon rank-sum test. The large sample size allows for the detection of very small differences between groups. Therefore, when evaluating these comparisons, both statistical and clinically relevant differences were considered.
Baseline factors that were potentially associated with the development of AF during the initial hospitalisation were identified by reviewing published reports.6 7 10 21 22 Characteristics reported in models of AF from the identified papers were used and included: age, sex, history of hypertension, MI, coronary artery bypass grafting (CABG), smoking status, systolic and diastolic blood pressure, heart rate (HR), Killip class, hours from symptom onset to treatment, race, geographic region of enrolment and type of ACS (ST- vs non-ST-segment elevation). The linearity assumption for the continuous measures was evaluated using restricted cubic spline transformations. By applying this transformation in the logistic model of AF, comparison of the linear term alone versus a model with a flexible association of the factor and AF tests the incremental benefit of the non-linear terms. The plot of the spline transformation versus the logit of the probability of AF illustrates the best transformation that may be applied. In most cases, a linear spline adequately accounted for the non-linear shape.
Backwards elimination variable selection was used to determine the main effects associated with developing AF as a complication. The interaction of these main effects with the presence or absence of ST-segment elevation was also considered in this model.
Clinical outcomes were divided into the acute phase (1–7 days post-randomisation) and the chronic phase (day 8 to the end of the follow-up period). These periods were chosen as a reflection of an in-hospital period and a post-discharge period. Since we do not have the date of AF, we could not accurately account for the timing of this complication relative to outcome. To account for this confounding effect of early deaths, we excluded patients who died within the first 24 hours in both the STEMI and NSTE-ACS groups. We also performed the analysis including the early deaths to evaluate the effect of this confounding on the results.
The chronic phase was evaluated beginning at day 8 after randomisation. Thus, events and procedures that occurred during the hospitalisation were before the start of this time period. This allowed us to include this information appropriately when adjusting for covariates.
The rates of all outcomes were calculated by the Kaplan–Meier method, and statistical comparisons of AF versus no AF, STEMI versus NSTE-ACS and the interactions were made with Wald χ2 tests from Cox proportional hazards models. Models for all end points were previously developed and validated for both patients with STEMI and those with NSTE-ACS.23–28 We tested for interactions to determine whether the association of AF with outcomes was similar in patients with STEMI and those with NSTE-ACS. AF was considered a baseline variable for the day 8 and beyond period since AF would generally have occurred before day 8.
The within-study correlation was included in the models of efficacy and safety outcomes. It was also included as a random effect in the prediction of AF. The results from these models were similar to those obtained when the correlation was excluded. Therefore, the simpler models are presented in the results.
All analyses were performed using SAS version 8.2 (SAS Institute Inc, Cary, North Carolina, USA). Results were declared significant for a p value ⩽0.05.
RESULTS
The occurrence of AF during the hospitalisation, regardless of the presence of previous AF, was 7.5% in the overall population, 8.0% in patients with STEMI, and 6.4% in patients with NSTE-ACS.
Baseline characteristics
Table 1 shows the baseline characteristics of all patients with ACS according to the presence of AF. Compared with patients without AF, patients with AF were older, more often female and more likely to be diabetic. They had higher Killip class and history of MI or vascular disease.
Similar differences in baseline characteristics between patients with and without AF were seen among the STEMI and NSTE-ACS groups (table 1).
Concomitant drugs and procedures
Patients with AF received fewer evidence-based drugs, including aspirin, β blockers and nitrates, but more often received angiotensin-converting enzyme inhibitors and antiarrhythmic drugs at discharge than patients without AF. These results were similar for patients with STEMI and those with NSTE-ACS, except for calcium channel blockers. Greater use of calcium blockers according to AF was seen in patients with NSTE-ACS (p<0.001), but not in patients with STEMI (p = 0.11).
In the NSTE-ACS group, patients with AF received more in-hospital cardiac catheterisation than those without AF (p<0.001). In the STEMI group, no such difference was seen (p = 0.24). Compared with patients without AF, patients with AF received less in-hospital (p<0.001) and 30-day (p<0.001) percutaneous coronary intervention, and more in-hospital (p<0.001) and 30-day (p<0.001) CABG.
Predictors of in-hospital AF
The strongest predictor was older age (odds ratio (OR) = 1.72 per 10 years; 95% CI 1.68 to 1.76), followed by HR ⩾85 bpm (OR = 1.29 per 10 bpm; 95% CI 1.26 to 1.32), receiving care in western Europe (vs North America OR = 0.63; 95% CI 0.60 to 0.67; p<0.001), Killip class IV versus class I (OR = 1.34; 95% CI 1.03 to 1.74), ST-segment elevation versus non-ST-segment elevation (OR = 1.41; 95% CI 1.32 to 1.52), white race versus others (OR = 1.65; 95% CI 1.48 to 1.84) and systolic blood pressure >110 mm Hg (OR = 0.93 per 10 mm Hg; 95% CI 0.92 to 0.95). Other baseline characteristics associated with increased risk of developing AF as an in-hospital complication were higher diastolic blood pressure at randomisation, smoking status, prior MI, history of hypertension, increased HR at randomisation, male sex and no previous CABG.
There were also four significant interactions with STEMI versus NSTE-ACS groups (fig 1). The associations of AF with previous CABG and female sex were stronger in the NSTE-ACS group (OR = 0.59; 95% CI 0.50 to 0.69; OR = 0.75; 95% CI 0.68 to 0.84, respectively) than in the STEMI group (OR = 0.80; 95% CI 0.70 to 0.92; OR = 0.89; 95% CI 0.83 to 0.94, respectively) (fig 1). The decreased risk of AF associated with increasing systolic blood pressure was present only in the STEMI population (OR = 0.90; 95% CI 0.87 to 0.94). Interestingly, this was only seen in systolic blood pressure below 110 mm Hg. For systolic blood pressure above 110 mm Hg, the decreased risk of AF was similar in the STEMI and NSTE-ACS groups. Increased time (>4 hours) from symptom onset to treatment was associated with increased risk of AF only in patients with STEMI (OR = 1.03; 95% CI 1.01 to 1.06). However, below 4 hours, the increased time was associated with decreased risk of AF in STEMI and NSTE-ACS populations.
Association of AF with short- and long-term outcomes
Short-term mortality
Mortality at 7 days was higher among patients with AF (5.1%) than without (1.6%). Similar trends for higher short-term mortality according to the presence or absence of AF were seen in STEMI (5.4×1.8%) and NSTE-ACS (4.2×1.2%) groups. Table 2 shows adjusted associations of AF with short- and long-term outcomes according to the presence or absence of ST-segment elevation. The higher risk of death from 1 to 7 days was stronger in NSTE-ACS (HR = 2.30; 95% CI 1.83 to 2.90) than in STEMI (HR = 1.65; 95% CI 1.44 to 1.90) patients (p interaction <0.015). This higher rate of death was significant in both STEMI (p<0.001) and NSTE-ACS (p<0.001) groups.
Long-term mortality
Mortality from day 8 to 1 year was higher among patients with AF (10.0%) than without (3.3%). After adjusting for confounders, AF was associated with increased mortality in both STEMI (p<0.001) and NSTE-ACS (p<0.001) groups (fig 2); there was a modestly greater risk in the STEMI group (p interaction 0.038). However, when β blockers and calcium antagonists at discharge were included in the model, the interaction of AF with ST-segment elevation was no longer significant (p interaction 0.401). The adjusted HR for long-term mortality was higher in the STEMI group (HR = 2.37; 95% CI 1.79 to 3.15) than in the NSTE-ACS group (HR = 1.67; 95% CI 1.41 to 1.99).M
Short-term stroke
The adjusted HR for short-term stroke (⩽7 days) was three times higher in the NSTE-ACS group (HR = 3.45; 95% CI 2.41 to 4.95) than in the STEMI group (HR = 1.46; 95% CI 1.17 to 1.82). The interaction of AF with ST-segment elevation versus non-ST-segment elevation was significant for short-term stroke (p interaction <0.001).
Long-term stroke
For long-term stroke (from 8 to 30 days), AF was also associated with increased risk in both patients with STEMI (HR = 3.60; 95% CI 1.51 to 8.52) and those with NSTE-ACS (HR = 2.90; 95% CI 1.62 to 5.20). The interaction of AF with ST-segment elevation versus non-ST-segment elevation for this outcome was significant (p interaction <0.001).
Other ischaemic outcomes
AF was also associated with MI (p<0.001) and death or MI (p<0.001) from 1 to 7 days. The interaction with ST-segment elevation was not significant for either of these outcomes (p interaction 0.381 and 0.719, respectively). The association of AF with these outcomes was significant in both groups, but stronger in the patients with NSTE-ACS (adjusted HR = 2.06; 95% CI 1.81 to 2.34; HR = 1.91; 95% CI 1.70 to 2.14 for NSTE-ACS and STEMI, respectively). AF was also associated with increased risk of in-hospital ischaemic stroke to 7 days for patients with STEMI (p<0.001) and those with NSTE-ACS (p<0.001). The interaction of AF with ST-segment elevation versus non-ST-segment elevation for this outcome was significant (p interaction <0.001).
Bleeding
Moderate or severe bleeding was associated with AF in patients with STEMI (p<0.001) and those with NSTE-ACS (p<0.001). The adjusted HR for this outcome was higher in patients with NSTE-ACS (HR = 4.67; 95% CI 4.26 to 5.13) than in those with STEMI (HR = 2.36; 95% CI 2.19 to 2.54; p interaction <0.001). AF was not associated with risk of in-hospital haemorrhagic stroke in patients with STEMI (p = 0.200) or those with NSTE-ACS (p = 0.054).
Association of AF with other in-hospital complications
Patients with AF had more in-hospital complications than patients without AF. These results appear to be consistent among both STEMI and NSTE-ACS groups, except for chronic heart failure (HF), transient ischaemic attack (TIA) and worse Killip class post-randomisation. AF was associated with more chronic HF and worse Killip class in the patients with NSTE-ACS, but not in the patients with STEMI. AF was not associated with TIA in the STEMI group or the NSTE-ACS group.
DISCUSSION
Our study is one of the largest to examine the occurrence of AF across the spectrum of ACS. AF was not only common (7.5% overall), but it was independently associated with a higher risk of long-term mortality. In this study, AF was analysed as a complication that occurred any time after randomisation and during hospitalisation. Thus, we are not able to define whether the AF was new onset or pre-existing. However, analysing a subset of 40 000 patients where the baseline ECG was available, we found that only 18% of those patients who had AF as a complication also had AF at baseline. Thus, we extrapolate that most of the cases of AF in this study were new-onset AF.
AF was also associated with a higher risk of short-term mortality, death or MI, ischaemic and haemorrhagic strokes and bleeding. This finding was observed in both STEMI and NSTE-ACS groups for all of the studied outcomes. Although some studies have demonstrated the absence of association between AF and short-term (30 days) mortality in acute MI,2 8 9 this study showed a strong association between AF and short-term (7 days) mortality among patients with STEMI and NSTE-ACS. We focused on the first 7 days after ACS, a time period that corresponds to the acute hospitalisation and is the period of highest risk for the early clinical events including stroke.29 Because mortality in the STEMI population tends to occur early (up to 50% of deaths occur within the first 24 hours),30 the relationship of AF and mortality may be blurred by the early deaths that have occurred before development of AF. In fact in this study, 32% of the deaths among the patients with STEMI occurred in the first 24 hours, while only 5% of them occurred within the same period among the patients with NSTE-ACS. As we were not able to determine the exact time of AF, we excluded all patients who died within the first 24 hours, thus estimating the association without the time-related confounding effect. When we included the patients who died in the first 24 hours, the hazard ratio was directionally consistent but the adjusted risk was no longer statistically significant for short-term mortality in the STEMI group (p = 0.998). These findings are consistent with an important confounding effect related to early deaths that was at least partially removed by their exclusion. For all other outcomes, interpretations of the results did not change when the early deaths were included.
For all outcomes other than long-term mortality, the risks associated with AF were higher in the NSTE-ACS group than in the STEMI group. This may relate to the fact that the patients with NSTE-ACS are older, had more history of hypertension, diabetes mellitus, chronic HF, MI, stroke, and TIA than the patients with STEMI. After the acute phase (7 days) of ACS, AF was also associated with long-term (day 8 to 1 year) death in both the STEMI and NSTE-ACS groups. These findings highlight the importance of AF in the ACS population as a predictor of short- and long-term mortality.
AF was also associated with other in-hospital complications such as HF, acute mitral regurgitation, sustained hypotension and cardiogenic shock. This shows that AF after ACS is associated not only with short- and long-term outcomes but also with in-hospital complications.
We identified several important baseline predictors of AF in the setting of ACS. In general, our findings confirm prior studies.6–10 However, this study is the first to analyse differences in predictors of AF between patients with STEMI and those with NSTE-ACS. We found interesting differences in some predictors, such as hypotension, between the STEMI and NSTE-ACS groups that appeared to be more related to risk of AF in the STEMI population for unclear reasons. Delay from symptom onset to initiation of the study drug was associated with an increased risk of AF only in the patients with STEMI, perhaps related to adverse consequences in delay to reperfusion in this population.
Patients who received care in Western Europe compared with other regions were less likely to develop AF after ACS. This finding may reflect differences in baseline clinical characteristics, ascertainment, variation in patient management and hospital type across the world.31 32 Compared with other races, white race was associated with an increased risk of developing AF, and the reason for this is not clear. These findings are consistent with previous studies.6 7
Female sex and prior CABG were associated with decreased risk of in-hospital AF. This finding is concordant with findings of other studies.7 10 21 An HR >85 bpm and diastolic blood pressure >70 mm Hg were associated with an increased risk of AF. It has been described previously that higher HR is associated with increased risk of AF7 10 and higher diastolic blood pressure may be associated with chronic hypertension that is associated with higher risk of AF.
Study limitations
This study has some limitations. First, it is an observational study; thus, unmeasured confounders may be present and cause and effect relations cannot be determined. Second, because those patients who had prior chronic AF may be more likely to have been classified as having AF as a complication, our reporting of incidence may in fact be a combination of incidence and prevalence. Third, we were unable to determine the timing and duration of AF and whether or not patients had a history of AF predating the ACS. Fourth, we were unable to establish a temporal relation between the time of AF and the occurrence of in-hospital complications. Fifth, the extent of MI and left ventricular dysfunction may be key determinants of development of AF. Unfortunately, since these data are available only in a small subset of patients they cannot be used in the multivariable modelling. However, we were able to assess the relationship of Killip class, a reflection of the degree of acute left ventricular dysfunction, with the complication of AF. Sixth, since our population is derived from clinical trial databases, it is a population at lower risk than in unselected clinical practice. Finally, we could not analyse the data of AF and atrial flutter separately as these two supraventricular arrhythmias were collected as a single variable in most of the trials. However, AF usually coexists with atrial flutter, and probably has similar clinical consequences.33
CONCLUSIONS
AF is common in both patients with STEMI and those with NSTE-ACS. While AF is associated with worse short- and long-term outcomes in ACS, this study highlights important differences in the clinical correlates and outcomes of patients with STEMI and NSTE-ACS complicated by AF. These differences were more in the magnitude of the association with outcomes than in the direction. In general, the results were similar between the two groups. Older age was the strongest predictor of in-hospital AF, followed by tachycardia, in both the STEMI and NSTE-ACS groups. Thus, AF is common, independently associated with worse outcomes across the spectrum of ACS, and is a predictor of increased mortality. These results were consistent across the trials, although the trial populations were different, with different age, prior risks factors, and performed at different times. Understanding these findings may ultimately lead to better care of patients with this arrhythmia that occurs commonly in the setting of ACS.
REFERENCES
Footnotes
Competing interests: None declared.