Author + information
- Received January 10, 2013
- Revision received April 4, 2013
- Accepted May 6, 2013
- Published online September 10, 2013.
- Morten Lamberts, MD∗,†∗ (, )
- Gunnar H. Gislason, MD, PhD∗,‡,§,
- Jonas Bjerring Olesen, MD∗,
- Søren Lund Kristensen, MD∗,
- Anne-Marie Schjerning Olsen, MD∗,
- Anders Mikkelsen, MB∗,
- Christine Benn Christensen, MD∗,
- Gregory Y.H. Lip, MD†,
- Lars Køber, MD, DMS‖,
- Christian Torp-Pedersen, MD, DMS∗,¶ and
- Morten Lock Hansen, MD, PhD∗
- ∗Department of Cardiology, Copenhagen University Hospital Gentofte, Hellerup, Denmark
- †University of Birmingham Centre for Cardiovascular Sciences, City Hospital, Birmingham, United Kingdom
- ‡Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
- §National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark
- ‖Department of Cardiology, The Heart Centre, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
- ¶Institute of Health, Science and Technology, Aalborg University, Aalborg, Denmark
- ↵∗Reprint requests and correspondence:
Dr. Morten Lamberts, Department of Cardiology, Post 635, Copenhagen University Hospital Gentofte, Niels Andersens Vej 65, Hellerup 2900, Denmark.
Objectives The purpose of this study was to investigate the risk of thrombosis and bleeding according to multiple antithrombotic treatment regimens in atrial fibrillation (AF) patients after myocardial infarction (MI) or percutaneous coronary intervention (PCI).
Background The optimal antithrombotic treatment strategy is unresolved in patients with multiple indications.
Methods A total of 12,165 AF patients hospitalized with MI and/or undergoing PCI between 2001 and 2009 were identified by nationwide registries (60.7% male; mean age 75.6 years). Risk of MI/coronary death, ischemic stroke, and bleeding according to antithrombotic treatment regimen was estimated by Cox regression models.
Results Within 1 year, MI or coronary death, ischemic stroke, and bleeding events occurred in 2,255 patients (18.5%), 680 (5.6%), and 769 (6.3%), respectively. Relative to triple therapy (oral anticoagulation [OAC] plus aspirin plus clopidogrel), no increased risk of recurrent coronary events was seen for OAC plus clopidogrel (hazard ratio [HR]: 0.69, 95% confidence interval [CI]: 0.48 to 1.00), OAC plus aspirin (HR: 0.96, 95% CI: 0.77 to 1.19), or aspirin plus clopidogrel (HR: 1.17, 95% CI: 0.96 to 1.42), but aspirin plus clopidogrel was associated with a higher risk of ischemic stroke (HR: 1.50, 95% CI: 1.03 to 2.20). Also, OAC plus aspirin and aspirin plus clopidogrel were associated with a significant increased risk of all-cause death (HR: 1.52, 95% CI: 1.17 to 1.99 and HR: 1.60, 95% CI: 1.25 to 2.05, respectively). When compared to triple therapy, bleeding risk was nonsignificantly lower for OAC plus clopidogrel (HR: 0.78, 95% CI: 0.55 to 1.12) and significantly lower for OAC plus aspirin and aspirin plus clopidogrel.
Conclusions In real-life AF patients with indication for multiple antithrombotic drugs after MI/PCI, OAC and clopidogrel was equal or better on both benefit and safety outcomes compared to triple therapy.
The efficacy and safety of different antithrombotic treatment strategies for patients with >1 indication for antithrombotic treatment is still unresolved. Oral anticoagulants (OAC) are superior to antiplatelets for stroke protection in atrial fibrillation (AF) (1), whereas dual antiplatelet therapy is indicated for acute coronary syndrome (ACS) and after percutaneous coronary intervention (PCI) (2,3). Although triple therapy including OAC, aspirin, and clopidogrel is recommended for AF patients with ACS and/or PCI, evidence stems from small observational studies of mostly single center origin, and American and European position documents categorize recommendations as Level of Evidence: C (4,5). Consequently, guidelines emphasize the need for individualization of treatment in respect to careful assessment of each patient’s risk of thrombosis and bleeding to find the optimal balance between risk and benefit of treatment. A recent study showed that initiation of triple therapy is immediately associated with increased and persistent bleeding risk compared to OAC plus a single antiplatelet agent (6). Therefore, a thorough exploration of a less-is-more approach is crucial as no firm knowledge exists on the effectiveness of adding 2 and not 1 antiplatelet agent on top of OAC.
We performed a nationwide study of real-world AF patients who were admitted with myocardial infarction (MI) or undergoing PCI, to evaluate the benefit and safety with multiple antithrombotic drugs. We hypothesized that the combination of OAC and a single antiplatelet (either aspirin or clopidogrel) is equal to triple therapy concerning thromboembolic protection, bleeding, and death.
In Denmark, all residents are provided with a unique and permanent civil registration number that enables linkage among 4 nationwide administrative registries (7). First, the Danish National Patient Registry keeps records of all hospital admissions since 1978 and each hospitalization is coded with 1 primary diagnosis and, if appropriate, 1 or more secondary diagnoses, according to the International Classification of Diseases-8th Revision (ICD-8; until 1994) and 10th Revision (ICD-10; from 1994). The registry also includes surgical and procedural treatment classification codes since 1996 (Nordic Medical Statistics Committees Classification of Surgical Procedures). Second, the Danish Registry of Medicinal Product Statistics (national prescription registry) contains information on all drug prescriptions dispensed from pharmacies in Denmark since 1995. All drugs are classified according to the international Anatomical Therapeutical Chemical (ATC) classification system. Finally, through the National Causes of Death Register and the Civil Registration System, both vital status and causes of death were obtained. The former holds information about the immediate, contributory, and underlying causes of death. These registries are complete for any Danish resident ages 10 years or above on January 1, 1997. (All ICD and ATC codes used are available in the Online Tables 1 and 2, respectively.)
Among subjects with a previous diagnosis of AF, we identified patients hospitalized for MI or PCI between January 1, 2001, and December 31, 2009 (Fig. 1). We excluded patients with an admission for MI/PCI within 1 year before the index date. A 7-day quarantine period was defined at discharge to disregard patients with unmeasured complications during index admission, as done previously (6). The inclusion date was therefore 7 days after discharge, and inclusion criteria were prescription of antithrombotic treatment at baseline in subjects alive, age 30 years or older, and no diagnosis of bleeding, MI, or thromboembolism during the quarantine period. The diagnosis of MI has been validated with a specificity of 93% (8), and the positive predictive value of AF is 99% (9).
Antithrombotic treatment regimens
We collected information on warfarin and phenprocoumon to categorize OAC as these were the only available OACs in Denmark during the study period. Similarly, aspirin and clopidogrel were the antiplatelets used in the majority of cases. Prasugrel, ticagrelor, and newer anticoagulants were not approved for routine use in AF or coronary artery disease patients in Denmark during the study period. We classified the following 4 multiple antithrombotic treatment regimens: aspirin plus clopidogrel (dual antiplatelet therapy); OAC plus aspirin; OAC plus clopidogrel; and OAC plus aspirin plus clopidogrel (triple therapy). Monotherapies with aspirin, clopidogrel, and OAC were also identified. Antithrombotic treatment at baseline was calculated as drug availability within 30 days from discharge, as done previously (10). To only consider patients at risk if treated, for every prescription dispensed the applicable treatment regimen was calculated for each patient. In brief, the daily dosage was estimated for up to 3 consecutive prescriptions. This enabled us to calculate any treatment regimen at any given time from prior claimed prescriptions. Hence, patients were only deemed at risk when exposed to treatment, and their treatment regimen was continually updated (11,12).
Benefit outcomes were defined as MI or coronary death, fatal or nonfatal ischemic stroke (including transient ischemic attack), and all-cause mortality. Safety outcome was fatal or nonfatal bleeding (e.g., intracranial bleeding and gastrointestinal bleeding). (The full list is available in Online Table 3.) Nonfatal events were obtained from admissions in the National Patient Registry, and fatal events were recorded from National Causes of Death Register. If a nonfatal event was followed by death within 1 week, the event was recorded as fatal. Ischemic stroke has been validated in the registries (13). The diagnoses to characterize serious bleeding have previously been used (6,10,11), and bleeding diagnoses have shown a positive predictive value of 89% to 99% in hospital databases (14). Compared to other clinical definitions of bleeding (15), our definition did not include data on hemoglobin levels or transfusions, but also did not include minor bleedings not serious enough to result in hospitalization.
Comorbidity and concomitant therapy
Comorbidities of previous bleeding, hypertension, congestive heart failure, ischemic stroke, transient cerebral ischemia, systemic embolus, renal/liver failure, peripheral arterial disease, alcohol abuse, and diabetes mellitus were determined from the registries as previously described (16). Scores of HAS-BLED (hypertension, abnormal renal/liver function, stroke/thromboembolism, bleeding history, labile international normalized ratio, elderly [age >65 years], and drug consumption/alcohol abuse), CHADS2 (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke/transient ischemic attack), and CHA2DS2-VASc (adding vascular disease, age 65 to 75 years, and female sex) were calculated. Of note, data on labile international normalized ratio was unobtainable and use of antiplatelet therapy was omitted from the “modified” HAS-BLED score because it is an explanatory variable. All scores have formerly been confirmed to accurately predict risk of thromboembolism and bleeding in the Danish population (12).
Patient characteristics according to treatment regimen at baseline are expressed as percentages or as means and standard deviations. Crude incidence rates (events per 100 person-years) were calculated within 1 year according to treatment regimen exposure (time-varying exposure). The Cox proportional hazard model was used to estimate the risk between dual and triple therapy regimens. Dual antiplatelet therapy was used as an active comparator for reference as this comprised the largest group and the standard of care for an acute coronary event. In the comparison analysis, we used triple therapy for reference as this is the recommended initial treatment. The analyses were adjusted for age, sex, inclusion year, and inclusion status (MI/PCI); and HAS-BLED and CHADS2 scores also were used for outcomes of bleeding and ischemic stroke, respectively. Subjects were followed up for 1 year, and censored if they died or at end of follow-up on December 31, 2009. The assumptions of the Cox model (proportional hazard assumption, linearity of continuous variables, and lack of interactions) were tested, and the models were found valid unless otherwise indicated. A 2-sided significance level of 0.05 was used.
The following sensitivity analyses were performed. A Cox model where baseline treatment regimen was used as non–time-dependent exposure (intention-to-treat analysis), and a Cox model to investigate the effect of a nonfatal bleeding, ischemic stroke, or MI admission on subsequent mortality. A nonfatal episode was incorporated as time-varying variable; for example, a subject was considered in the nonbleeding group before the date of nonfatal bleeding episode and thereafter included in the bleeding group. This model was adjusted for antithrombotic treatment, age, sex, inclusion year, and inclusion status. As suggested by Ray (17), we decided on a sensitivity analysis using a new-user design, and excluded prevalent users of aspirin and clopidogrel (i.e., any subscription claims within 6 months before index MI/PCI), whereas OAC use was allowed because we did not want to exclude ongoing use in AF patients before inclusion. We also did a Cox model including patients in the quarantine period with 90 days of follow-up. To ensure consistency of the results in the PCI group, we did separate analyses of absolute risk due to the group being considered as “low risk” patients before start of the study. All analyses were performed with SAS version 9.2 (SAS Institute, Cary, North Carolina) and Stata version 11.0 (StataCorp, Chicago, Illinois).
Register-based studies do not require ethical approval in Denmark as patients cannot be identified. The Danish Data Protection Agency approved the study (reference no. 2007-58-0015, international reference: GEH-2010-001).
A total of 12,165 patients was included in the study (Fig. 1) with a mean age of 75.6 ± 10.3 years, and 60.7% were male. Mean CHADS2, CHA2DS2-VASc, and HAS-BLED scores were 1.9 ± 1.2, 4.0 ± 1.6, and 2.0 ± 0.9, respectively. Baseline characteristics are shown in Table 1. Nearly two-thirds were treated with multiple antithrombotic drugs at baseline (62.0%), and a total of 4,659 (38.3%) received OAC. Only 121 (<1%) had a CHA2DS2-VASc score of 0. Figure 2 shows that according to predicted risk of either ischemic stroke (CHADS2) or bleeding (HAS-BLED), the type of antithrombotic treatment regimen prescribed at baseline was fairly consistent regardless of increasing score. During the study period of 1 year, a total of 2,255 (18.5%), 680 (5.6%), and 769 (6.3%) patients were registered with MI/coronary death, ischemic stroke, and bleeding, respectively. A total of 2,356 (19.4%) of the patients died. Table 2 shows crude rates and hazard ratio (HR) for all outcomes. Bleeding rates increased with increased intensity of treatment.
Triple therapy versus dual therapy
Figure 3 shows comparisons of dual antiplatelet therapy, OAC plus clopidogrel, and OAC plus aspirin versus triple therapy. Relative to triple therapy, a nonsignificant benefit for MI or coronary death was seen for OAC plus clopidogrel (HR: 0.69, 95% confidence interval (CI): 0.48 to 1.00), whereas OAC plus aspirin (HR: 0.96, 95% CI: 0.77 to 1.19), or dual antiplatelet therapy (HR: 1.17, 95% CI: 0.96 to 1.42) were comparable. For ischemic stroke only dual antiplatelet therapy showed a higher risk (HR: 1.50, 95% CI: 1.03 to 2.20). When compared to triple therapy, bleeding risk was nonsignificantly lower for OAC plus clopidogrel (HR: 0.78, 95% CI: 0.55 to 1.12) and significantly lower for OAC plus aspirin (HR: 0.69, 95% CI: 0.53 to 0.90), and dual antiplatelet therapy (HR: 0.48, 95% CI: 0.38 to 0.61). OAC plus aspirin and dual antiplatelet therapy were associated with significant increased risk of all-cause death (HR: 1.52, 95% CI: 1.17 to 1.99 and 1.60, 95% CI: 1.25 to 2.05), respectively, whereas OAC plus clopidogrel was not (HR: 0.87, 95% CI: 0.56 to 1.34). Cause of death was equally distributed between the different antithrombotic treatment regimens (Online Fig. 1). Approximately 50% of all deaths were cardiovascular-related deaths.
Baseline treatment regimen as a non–time-dependent variable showed similar estimates of outcomes as in the main analyses (Online Fig. 2). We investigated the effect on subsequent mortality after admission for a nonfatal episode of bleeding, and 181 (26%) of 691 patients with an admission for nonfatal bleeding died compared to 2,175 (19%) of 11,474 without nonfatal bleeding (HR: 2.79, 95% CI: 2.39 to 3.26). Estimates for episodes of nonfatal myocardial infarction and nonfatal ischemic stroke were HR 3.17 (95% CI: 2.84 to 3.55) and HR 3.93 (95% CI: 3.37 to 4.57), respectively. A total of 5,065 patients was examined using the new-user design, and the crude incidence rate and HR of the benefit and safety outcomes were similar to our main analysis (Online Table 3). Including the quarantine period, we followed up 13,007 patients for 90 days and risk estimates were comparable to the main analyses for all outcomes. Relative to triple therapy, no increased risk of MI/coronary death for OAC plus clopidogrel (HR: 0.78, 95% CI: 0.51 to 1.17) and OAC plus aspirin (HR: 1.17, 95% CI: 0.92 to 1.49) were observed. We did not find any interactions between various subgroups; age ≥75 years, <75 years, sex, PCI, or MI status at inclusion. PCI-only patients (without MI) had considerably lower absolute risk for MI or coronary death, with crude incidence rates of 6.2, 3.0, 5.2, and 6.5 events per 100 person-years for triple therapy, OAC plus clopidogrel, OAC plus aspirin, and dual antiplatelet therapy, respectively. The crude incidence rates for ischemic stroke and bleeding in PCI-only patients were comparable to the overall population (data not shown).
Our main finding was that in AF patients after MI/PCI, the combination of OAC plus clopidogrel is comparable to the recommended triple therapy in respect to the prevention of thromboembolic outcomes of MI/coronary death and ischemic stroke while the risk of bleeding was similar. Notably, the risk of all-cause mortality was similar between OAC plus clopidogrel and triple therapy but markedly increased for other therapies, namely, OAC plus aspirin and dual antiplatelet therapy. No beneficial effect was evident for adding aspirin to OAC plus clopidogrel treatment, which challenges current recommendations that favors triple therapy for this population.
The present recommendation of triple therapy for patients with recent MI and/or PCI and concurrent AF is established because of OAC being indicated for AF patients with stroke risk factors and dual antiplatelet therapy being indicated after ACS or PCI (1,18–20). Nonrandomized studies on triple therapy shows a protection from ischemic events with no clear excess bleeding, compared to other regimens (21,22), but the studies all have limitations of being potentially underpowered, performed in single center settings, and only including patients after stent implantation. Our findings of a potential safer profile of a single antiplatelet added to OAC is in line with Nguyen et al. (23), who followed up 800 patients after coronary stenting but irrespective of presence of atrial fibrillation for 6 months, and did not find a difference between single or dual antiplatelet approach combined with warfarin. Our data also suggest that the favored antiplatelet to be used in combination with OAC could be clopidogrel, and this combination may well be used in AF patients after recent MI and/or PCI. A case-control study of patients after coronary stenting with or without an indication for long-term OAC (being AF in 70% of cases) also found more favorable outcomes for clopidogrel than aspirin in combination with OAC (24), whereas another study found a protective effect of coumarins on a combined endpoint of major adverse cardiovascular events and death compared to non–coumarin-treated AF patients with stent implantation (25). The latter study was limited by being based on post-discharge telephone interviews and post-hoc collection of outcome data.
Results are available from the recently published WOEST (What Is the Optimal Antiplatelet and Anticoagulant Therapy in Patients With Oral Anticoagulation and Coronary Stenting) trial, which was a randomized prospective multicenter study of stented patients on chronic OAC (AF and non-AF patients), testing triple therapy versus OAC plus clopidogrel (26). Albeit with different inclusion criteria and lower numbers (total inclusion of 573 subjects) compared to our study, the WOEST investigators found similar cardiovascular protection for a combined endpoint of death, stroke, MI, target vessel revascularization, and stent thrombosis after 1 year. A significantly lower rate of death and nonsignificantly lower numbers of MI with OAC plus clopidogrel were also present in the WOEST study, which is in concordance with our findings. However, the lower risk for bleeding for OAC plus clopidogrel was primarily driven by minor bleeding events, whereas our study only included bleedings serious enough to warrant hospitalization. Importantly, the poor prognosis found in both the WOEST trial and in the present study with triple therapy could be related to bleeding events, for example, withdrawal of life-saving thromboprophylactic medication or higher case fatality rates than previously assumed. Future studies on the complications of bleeding (minor or major) could provide a better understanding on the use of multiple antithrombotic drugs. While awaiting further studies supportive of the WOEST trial, together with our results, the evidence of benefit when adding aspirin to OAC and clopidogrel seems to diminish.
Another important finding was that antithrombotic treatment prescribed initially was not markedly influenced by either predicted stroke risk or bleeding risk. Every day prescription of antithrombotic treatment is known to show high variability (27), and these findings are suggestive of under-treatment when only approximately 40% received treatment that included OAC. Especially worrisome is it when increased CHADS2 and HAS-BLED scores are clearly related to increased rates of ischemic stroke and bleeding, respectively (Fig. 2). This message of a seemingly random treatment of patients and simple risk scoring schemes that predict risk should be acknowledged by physicians.
As newer OACs (e.g., apixiban, rivaroxiban, dabigatran) are likely to replace vitamin K antagonists in certain AF patients in the years to come, clinicians will soon face new treatment possibilities not fully explored in trials or in large databases. While no head-to-head comparison of these new agents is available, some data on triple therapy with these newer OAC are available. In post–acute coronary syndrome patients (presence of AF not known), the addition of apixaban to dual antiplatelet therapy did no better with thromboembolic outcomes but resulted in increased serious bleeding compared to placebo (28). With rivaroxiban, in very low dosage twice-a-day regimen, there was a reduction of thromboembolic complications but at the cost of more nonfatal bleedings (29). A substudy of the RE-LY (Randomized Evaluation of Long-Term Anticoagulation Therapy) trial showed even further increased bleeding risk with dual antiplatelet on top of dabigatran than a single antiplatelet (30). Unfortunately, no data were reported for the effect of the number or type of antiplatelets for stroke or coronary events, although up to 40% of patients had coronary artery disease.
Study strengths and limitations
The main strength of this study is the possibility to evaluate the effect of antithrombotic therapy in real-life patients in a nationwide setting regardless of race, employment, health insurance coverage, and socioeconomic status. We compared clinically used regimens that would not be feasible in a controlled environment or from single center studies. We also were able to update treatment on a continual basis, which is often a concern when estimating bleeding endpoints (31). The diagnoses of AF, ischemic stroke, and MI are all validated in the registries (8,9,13), and using only hospitalized bleeding events, we presume we only collect serious bleeding outcomes. The combination of OAC plus aspirin shows similar estimates for coronary, stroke, and bleeding events compared to OAC plus clopidogrel and triple therapy, but did, however, also show markedly higher risk of death, hence confounding by indication could have affected our results. To estimate the potential effect of confounding by indication, we used a new-user design as a sensitivity analysis (17). The results were consistent with the main analyses, indicating that our estimates were robust and not likely confounded by a “healthy antithrombotic user” effect (Online Table 3). Furthermore, if confounding by indication was related to the prescription of either aspirin or clopidogrel, then results for triple therapy would also be affected. Initial antithrombotic treatment did not seem to be influenced by either predicted stroke or bleeding risk (Fig. 2), and we also found no differences between treatment regimen at time of death and cause of death (Online Fig. 1). Moreover, any bias is most likely to affect our estimates conservatively, namely, intensive recommended triple therapy is most likely to be prescribed to healthier patients with a better outcome profile. Although we have controlled for clinically relevant covariates including validated risk scoring schemes for stroke and bleeding, the possibility of residual confounding from unmeasured variables remains. We did not have access to data on body mass index, smoking status, or international normalized ratio values. Concerning the latter, compared to other countries but in a controlled setting, time-in-therapeutic range is high in Denmark (>70%) (32). The use of over-the-counter aspirin could not be identified in the registries, but as persistence to therapy in a MI population in Denmark is high and because of financial reimbursement, we believe the usage is negligible (10).
In real-life AF patients with indications for multiple antithrombotic drugs after MI/PCI, OAC and clopidogrel was equal or better on both benefit and safety outcomes compared to triple therapy. Our data suggest that triple therapy management regimens might be replaced by OAC and clopidogrel without any additional risk of recurrent thrombotic events and a lower risk of bleeding. Data from large randomized studies are severely needed to estimate overall benefit.
The authors acknowledge Jesper Lindhardsen, MD, PhD, for statistical assistance.
This work was supported by an unrestricted research grant from the Department of Cardiology, Gentofte University Hospital. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- Anatomical Therapeutical Chemical
- confidence interval
- hazard ratio
- International Classification of Diseases
- myocardial infarction
- oral anticoagulant
- percutaneous coronary intervention
- Received January 10, 2013.
- Revision received April 4, 2013.
- Accepted May 6, 2013.
- American College of Cardiology Foundation
- Connolly S.,
- Pogue J.,
- et al.,
- for the ACTIVE Writing Group of the ACTIVE Investigators
- Steg P.G.,
- James S.K.,
- Atar D.,
- et al.
- Lamberts M.,
- Olesen J.B.,
- Ruwald M.H.,
- et al.
- Thygesen L.C.,
- Daasnes C.,
- Thaulow I.,
- Bronnum-Hansen H.
- Sorensen R.,
- Hansen M.L.,
- Abildstrom S.Z.,
- et al.
- Steg P.G.,
- Huber K.,
- Andreotti F.,
- et al.
- Olesen J.B.,
- Lip G.Y.,
- Hansen M.L.,
- et al.
- Ray W.A.
- Faxon D.P.,
- Eikelboom J.W.,
- Berger P.B.,
- et al.
- Rubboli A.,
- Milandri M.,
- Castelvetri C.,
- Cosmi B.
- Nguyen M.C.,
- Lim Y.L.,
- Walton A.,
- et al.
- Karjalainen P.P.,
- Porela P.,
- Ylitalo A.,
- et al.
- Ruiz-Nodar J.M.,
- Marin F.,
- Hurtado J.A.,
- et al.
- Nieuwlaat R.,
- Capucci A.,
- Camm A.J.,
- et al.
- Dans A.L.,
- Connolly S.J.,
- Wallentin L.,
- et al.