Author + information
- Received June 26, 2010
- Revision received August 30, 2010
- Accepted September 14, 2010
- Published online March 8, 2011.
- Rikke Sørensen, MD⁎,†,⁎ (, )
- Steen Z. Abildstrøm, MD, PhD†,
- Peter R. Hansen, MD, DMSc⁎,
- Anders Hvelplund, MD⁎,‡,
- Charlotte Andersson, MB⁎,
- Mette Charlot, MD⁎,
- Emil L. Fosbøl, MD, PhD⁎,
- Lars Køber, MD, DMSc§,
- Jan K. Madsen, MD, DMSc⁎,
- Gunnar H. Gislason, MD, PhD⁎ and
- Christian Torp-Pedersen, MD, DMSc⁎
- ↵⁎Reprint requests and correspondence:
Dr. Rikke Sørensen, Department of Cardiology, Copenhagen University Hospital Gentofte, Niels Andersens vej 65, Post 4210, 2900 Hellerup, Denmark
Objectives The objective of this study was to examine the clinical efficacy of clopidogrel treatment on death and recurrent myocardial infarction (MI) among MI patients revascularized by coronary artery bypass graft surgery (CABG).
Background The benefit from post-operative clopidogrel in CABG-treated MI patients is largely unknown.
Methods All patients admitted with first-time MI between 2002 and 2006, treated with CABG within 180 days after admission, were identified by nationwide administrative registers. Clopidogrel treatment was determined by claimed prescriptions after discharge from surgery. Risk of death or recurrent MI, and of a combined end point of the 2, were assessed by cumulative incidence and Cox proportional hazards model. A propensity score-matched subgroup analysis was done.
Results We included 3,545 patients, and of these, 957 (27.0%) were treated with clopidogrel after CABG. Mean follow-up was 466 ± 144 days. Among patients treated with clopidogrel, 39 (4.1%) died or experienced a recurrent MI, whereas that occurred in 203 (7.8%) patients without clopidogrel (log-rank p = 0.0003). Hazard ratio was 0.59 (95% confidence interval [CI]: 0.42 to 0.85) for patients treated with clopidogrel, with no-clopidogrel as reference. By propensity score, of 945 patients with or without clopidogrel treatment who were matched, death or recurrent MI occurred in 38 (4.0%) patients with clopidogrel and 57 (6.0%) without clopidogrel (log-rank p = 0.05). Corresponding hazard ratio was 0.67 (95% CI: 0.44 to 1.00) for clopidogrel users, with no-clopidogrel as reference.
Conclusions Among MI patients revascularized by CABG, only 27% received clopidogrel after discharge. Clopidogrel-treated patients had a lower risk of the combined end point of death or recurrent MI. Focus on discharge clopidogrel treatment of these patients should be made.
Since 2002, clopidogrel treatment has been recommended for most patients with myocardial infarction (MI) (1–3). For MI patients treated with percutaneous coronary intervention (PCI) and stent implantation, the benefit of clopidogrel is documented, but the benefit from post-operative clopidogrel for MI patients revascularized by coronary artery bypass graft surgery (CABG) is largely unknown. Several studies have reported subgroup analyses examining this clinically important question (4–9), but no trial has been specifically designed to investigate the issue. Focus has primarily been on the necessity of ceasing clopidogrel treatment before surgery to prevent perioperative bleeding, but treatment benefits after CABG in high-risk patients have been suggested (10,11). Even though guidelines since 2004 have recommended that clopidogrel should be continued for a period of 9 to 12 months for acute coronary syndrome patients revascularized by CABG (12–14), concerns have been raised about the actual benefit of clopidogrel treatment for these patients (15). Clinical practices and the effect of clopidogrel treatment in real-life MI patients revascularized by CABG remains unclear. We conducted a nationwide study to investigate the use and the effect of clopidogrel treatment on mortality and recurrent MI among 3,545 MI patients revascularized by CABG.
Five nationwide administrative registers were used and linked on an anonymous and individual level in this study: 1) the Danish National Patient Register, which contains information on all admittances to Danish hospitals since 1978 (classified according to the International Classification of Diseases [ICD]-8 and ICD-10, and to the Danish Board of Health classification of invasive procedures); 2) the Danish Register of Medicinal Product Statistics (the Prescription Register), which holds information about all prescriptions dispensed at Danish pharmacies since 1995 (classified according to the Anatomical Therapeutical Chemical Code, strengths, number of tablets, and date of dispensing); 3) the civil register, which contains information on vital status of all citizens; 4) the register of causes of death, which contains information on causes of deaths (according to ICD-10); and 5) information on yearly income, as provided by Statistics Denmark. Register studies in which individual patients cannot be identified do not require ethical approval in Denmark. This study was approved by the Danish Data Protection Agency (reference 2003-54-1269).
We identified all patients admitted to Danish hospitals with first-time MI between the years 2002 and 2006 (ICD-10 code I21 or I22). Patients revascularized by CABG (Danish Board of Health classification code KFNA-E) within 180 days and surviving 30 days after discharge from surgery were included in our study. Patients were stratified as clopidogrel users if a prescription of clopidogrel was claimed within 30 days after discharge from surgery. Comorbidity was established according to the modified Ontario acute myocardial infarction prediction rules by identifying discharge diagnoses (ICD-10 codes) during the year before hospital admission for MI (16,17). Because the registers have low sensitivity for the diagnosis of heart failure (18), we used claimed prescriptions of loop diuretics in the period 90 days before until 90 after hospital discharge from MI as a proxy for heart failure (19). In the same way, patients who claimed prescriptions for glucose-lowering medicine were deemed to have diabetes mellitus. Previous bleeding was defined as an admission to hospital with a bleeding diagnosis during a 5-year period before the myocardial infarction (ICD-10 codes I60-62, S06.4-06.6, J94.2, R04, K25.0, K25.4, K26.0, K26.2, K26.4, K27.0, K 27.2, K28.0, K28.2, K 92.0-92.2, R31, D62, D50) (20).
Use of clopidogrel and concomitant drugs was identified by information obtained in the Prescription Register. Baseline for concomitant drug use was determined if a prescription was dispensed within 90 days after discharge from MI, except for statins, for which the period was 180 days. Adherence to clopidogrel treatment was calculated individually, as done previously (19,21). Patients were defined as nonadherent if they had no tablets available for treatment for a period of 30 days or more.
In addition to stratifying the patients according to clopidogrel treatment, we created a propensity score-matched population. The propensity score was estimated as each patient's probability of receiving clopidogrel by multivariable logistic regression analysis conditional on the following baseline covariates: year of admission, age, sex, time to CABG, comorbidities (cerebral vascular disease, diabetes with complications, cardiac dysrhythmias, acute renal failure, chronic renal failure, malignancy, shock, pulmonary edema, previous bleeding), concomitant drugs (beta-blockers, angiotensin-converting enzyme inhibitors and angiotensin-II receptor blockers, statins, loop diuretics, glucose-lowering drugs, vitamin K antagonists, aspirin, proton pump inhibitors, pre-CABG clopidogrel [used after MI but before CABG]), and PCI. The propensity score was used to match comparable patients 1:1 as to who did or did not receive clopidogrel treatment after CABG. The patients were matched according to the propensity score and the year of admittance). The matched patients constituted the propensity score-matched population. The type of hospital where the patients were admitted for MI was classified as local hospital, main regional hospital, or tertiary hospital; the patients were all referred to a tertiary hospital for the CABG operations (5 different departments). Socioeconomic status was determined as tertiles of the average of tax-related income 5 years before admission for MI (22).
The following outcomes were used to evaluate the effect of clopidogrel: 1) all-cause mortality; 2) combined end point of recurrent MI or death from any cause; and 3) recurrent MI (defined as admittance to a Danish hospital with MI [ICD-10 code I21 or I22] >30 days after discharge from CABG). Bleedings that occurred after discharge from CABG were recorded according to the ICD-10 codes listed in preceding text (defining previous bleedings). Finally, the causes of deaths were classified as “definite noncardiovascular death” or “cardiovascular or possible cardiovascular death.” The ICD-10 codes used for these classifications are listed in Table 2 (23).
Baseline characteristics are presented referring to characteristics at the time of MI diagnosis. The Mann-Whitney U test and the chi-square test were used to compare means and proportions of baseline characteristics. McNemar's test and Wilcoxon signed rank-sum test were applied for the propensity score-matched population. Independent factors related to initiation of clopidogrel after CABG was estimated by multivariate logistic regression analysis, including all baseline variables and information on income group and type of hospital. Because of significant interaction between pre-CABG clopidogrel and PCI (p = 0.04), patients with or without pre-CABG clopidogrel were analyzed in separate models. Cumulative incidence was calculated for each end point and illustrated separately for all patients as well as for the propensity score-matched population. Adjusted hazard ratios were calculated for all patients by use of Cox proportional hazard models, which included the following variables: year of admission, sex, age, time to CABG, comorbidities (cerebral vascular disease, diabetes with complications, cardiac dysrhythmias, acute renal failure, chronic renal failure, malignancy, shock, pulmonary edema, previous bleeding), and concomitant drugs (pre-CABG clopidogrel, clopidogrel after CABG, beta-blockers, angiotensin-converting enzyme inhibitors and angiotensin-II receptor blockers, statins, loop diuretics, glucose-lowering drugs, vitamin K antagonists, acetylsalicylic acid, nonsteroidal anti-inflammatory drugs). For the propensity score-matched analysis, the univariate hazard ratio for clopidogrel was estimated using the robust sandwich covariance matrix estimation, as other covariates were used when generating the propensity score. All Cox models were tested for proportional hazards, linearity of continuous variables, and absence of interactions, and were found to be valid. Patients were observed up to 18 months after CABG. Patients included after July 1, 2005, were followed up until December 31, 2006. Statistical analyses were done using SAS version 9.1.4 (SAS Institute, Cary, North Carolina).
A total of 3,545 patients was included in this study. Of these, 957 (27.0%) claimed a prescription for clopidogrel within 30 days after discharge from surgery. The selection of the study cohort is illustrated in Figure 1. A propensity score-matched population was identified and consisted of 945 matched patient pairs stratified with or without clopidogrel treatment. The C-statistic of the logistic regression model used for calculating the propensity score was 0.69, indicating fair to good discriminative power of the model. Baseline characteristics of the study cohort and the propensity score-matched population are shown in Table 1. There were significant differences (with p < 0.05) in baseline characteristics between patients receiving and not receiving clopidogrel treatment (Table 1). In the propensity score-matched population, the baseline characteristics were balanced by matching, except for time to CABG (difference in matched pairs p = 0.01).
Clopidogrel treatment after CABG
Predictors of higher initiation of clopidogrel after CABG were year of admission, statins, proton pump inhibitors, PCI, and female sex for patients without pre-CABG clopidogrel (n = 2,850); and year of admission for patients with pre-CABG clopidogrel (n = 720). Lower initiation was related to use of vitamin K antagonists, low-income group, and time to CABG for patients without pre-CABG clopidogrel. Corresponding figures for patients with pre-CABG clopidogrel were use of vitamin K antagonists and aspirin. The proportions discharged with clopidogrel varied among the 5 surgical departments: 6.6%, 14.4%, 33.5%, 33.9%, and 44.5%. Adherence to initiated clopidogrel therapy is illustrated by Figure 2. The duration of treatment increased over time, which is illustrated by stratification by year of admission (21). Of clopidogrel-treated patients, 79% were treated 80 days after discharge in 2002, 92% in 2004, and 94% in 2006. The proportions at day 180 were 47% in 2002, 73% on 2004, and 74% in 2006.
During a mean follow-up of 466 ± 144 days, 14 (1.5%) patients died of any cause in the clopidogrel group compared with 136 (5.3%) in the group without clopidogrel (log-rank p < 0.0001). Corresponding numbers for the combined end point of recurrent MI or all-cause mortality were 39 (4.1%) and 203 (7.8%), respectively (log-rank p = 0.0003). Recurrent MI occurred in 26 (2.7%) patients with clopidogrel treatment, and in 76 (2.9%) patients without clopidogrel treatment (log-rank p = 0.81). In the propensity score-matched population, mean follow-up was 452 ± 151 days; 14 (1.5%) patients died in the clopidogrel group, and 36 (3.8%) died in the group without clopidogrel treatment (log-rank p = 0.002). Recurrent MI or death from any cause occurred in 38 (4.0%) and 57 (6.0%), respectively (log-rank p = 0.05), whereas the numbers for recurrent MI were 25 (2.7%) and 23 (2.4%), respectively (log-rank p = 0.78). Cumulative incidence curves are shown in Figure 3. The results of the adjusted analyses for all patients are shown in Figure 4, along with the corresponding analyses of the propensity score-matched population. Table 2 shows the causes of deaths classified as “definite noncardiovascular death” or “cardiovascular or possible cardiovascular death.” Among all patients, fewer cases of cardiovascular or possible cardiovascular deaths were found among clopidogrel-treated patients (p = 0.03) compared with patients without clopidogrel. Corresponding difference for the propensity score-matched population was p = 0.19. Bleedings occurred in 47 (4.9%) patients treated with clopidogrel and 122 (4.7%) patients without clopidogrel treatment (log-rank p = 0.73). In the propensity score-matched population, the corresponding numbers were 46 (4.9%) and 40 (4.0%; log-rank p = 0.52).
To examine the robustness of our findings, we performed additional analyses, First, the additional end point revascularization in combination with recurrent MI (PCI and recurrent CABG occurring anytime after primary CABG) was analyzed. Among all patients, 54 (5.6%) with clopidogrel had recurrent MI or revascularization compared with 126 (4.9%) without clopidogrel (log-rank p = 0.26). In the adjusted Cox model, with no clopidogrel treatment used as reference, the hazard ratio for recurrent MI or revascularization with clopidogrel was 1.18 (95% CI: 0.85 to 1.65). The corresponding figures for the propensity score-matched population were 51 (5.4%) with clopidogrel and 36 (3.8%) without clopidogrel (log-rank p = 0.10). The hazard ratio was 1.43 with clopidogrel (95% CI: 0.94 to 2.20), with no clopidogrel treatment used as reference. Second, we added income group and type of hospital to the logistic model for the propensity score. The C-statistic score changed from 0.685 to 0.687. By use of the new propensity score (in addition to year of admission), 940 patients with and without clopidogrel were matched, and the risk estimates were obtained. The hazard ratios were nearly the same (and without interactions): 0.38 (95% CI: 0.20 to 0.70) for death, 0.65 (95% CI: 0.44 to 0.97) for death or MI, and 1.13 (95% CI: 0.66 to 1.95) for recurrent MI. In the same way, among all patients, the hazard ratios were unchanged when income group and type of hospital were included in the Cox model. The hazard ratios were as follows: for death, 0.34 (95% CI: 0.19 to 0.60); for recurrent MI or death, 0.59 (95% CI: 0.41 to 0.83); and for recurrent MI, 0.92 (95% CI: 0.58 to 1.46).
In a nationwide cohort of real-life MI patients revascularized with CABG within 180 days after MI, we found that 27% were treated with clopidogrel after surgery. Of these, fewer died or experienced the combined end point of recurrent MI or death compared with patients not treated with clopidogrel. We found similar rates of recurrent MI among patients treated with or without clopidogrel. The numbers of bleedings after discharge were comparable between the groups.
The use and effect of post-surgery clopidogrel in acute coronary syndrome patients revascularized by CABG have previously been studied as subgroup analyses of randomized trials (1,7–9), and in small observational studies (4–6). No trial has been specifically designed to investigate this clinically important question. Particular benefit from clopidogrel treatment has been proposed for high-risk patients (8,9,11), and current guidelines recommend that patients with non–ST-segment elevation acute coronary syndrome undergoing CABG should be treated with clopidogrel in addition to aspirin 9 to 12 months after the procedure (12–14), a statement that has been questioned (15).
In our study, we found that only 27.0% of the MI patients claimed a prescription of clopidogrel after CABG. For patients without pre-CABG clopidogrel, predictors of lower initiation were use of vitamin K antagonists, time to CABG, and the lowest income group. For patients with pre-CABG clopidogrel, use of vitamin K antagonists and aspirin predicted low initiation. Use of vitamin K antagonist and low-income group have previously been related to low initiation of clopidogrel (21) or other cardiovascular drugs (22). Once started on a regimen of clopidogrel, adherence with treatment improved over time, with 47% continuing treatment for 180 days in 2002 and 74% in 2006.
Concerning efficacy of clopidogrel after CABG, we found an association between reduced mortality and fewer patients with the combined end point of recurrent MI or death among patients treated with clopidogrel compared with patients not treated with clopidogrel. Surprisingly, we found similar risks of recurrent MI in the 2 groups. The results were consistent across the propensity score-matched population. Exploring all data on mortality, we found more cases of definite noncardiovascular deaths and fewer cases of cardiovascular or possible cardiovascular deaths among clopidogrel-treated patients (p = 0.03). There was a trend toward confirmation of these differences in the propensity score-matched population (p = 0.19). The lack of effect on recurrent MI may be explained by this, since all patients in our study had advanced cardiovascular disease (treated with CABG). A new ischemic event might, therefore, often be fatal, and thus recurrent MI may be hidden in the mortality rates. An indication of this suggestion was found previously in the CAPRIE (Clopidogrel Versus Aspirin in Patients at Risk of Ischemic Events) trial, in which the strongest benefit from clopidogrel was seen in the end point of vascular death (9).
Some studies of patients with MI undergoing subacute CABG have found a reduced risk of adverse cardiac events (recurrent MI, coronary reintervention, and cardiac death) with post-surgery dual-antiplatelet therapy (5). Likewise, subgroup analyses of the CURE (Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Events) trial and the CREDO (Clopidogrel for the Reduction of Events During Observation) trial reported risk reductions of cardiovascular death, stroke, and MI for patients who underwent CABG and were treated with aspirin plus clopidogrel compared with aspirin monotherapy; but it is noteworthy that these reductions were statistically insignificant, probably due to the low number of patients (7,8). Opposed to this, Sanon et al. (4) found no reduction in 30-day and 4-year mortality associated with dual treatment with aspirin plus clopidogrel after CABG compared with aspirin alone, but merely 43% of their study population had acute coronary syndrome. More recently, a systematic review of the available evidence also found no beneficial effect of clopidogrel monotherapy or clopidogrel in combination with aspirin after CABG (15). Study differences between the proportions of patients with high-risk non–ST-segment elevation acute coronary syndrome have likely contributed to these divergent results.
In the present study, the safety of post-operative clopidogrel treatment was evaluated by hospital admissions with a bleeding diagnosis (20). We found no excess of bleeding among the patients treated with clopidogrel compared with patients not treated with clopidogrel, but this result should be interpreted with caution as more patients in the group without clopidogrel received vitamin K antagonists (Table 1). In the propensity score-matched population, the difference in use of vitamin K antagonists was balanced, and the outcome with no difference in bleedings was confirmed, as it has been in other studies as well (5,7,24).
Study strengths and limitations
The main strength of our study is the completeness of data, with a nationwide unselected cohort of patients followed up in a real-life setting and with complete data on clopidogrel use. The data in the registers have previously been validated (25,26). The main limitations are inherent in the observational nature of the study; lack of randomization prevents us from establishing a true causal relationship between clopidogrel use and outcomes. The information on clopidogrel use in the period between the diagnosis of MI until CABG operation, used in the multivariate logistic regression analysis (for the propensity score) as well as in the Cox model, is associated with some uncertainty because some patients stayed in hospital during this period and some patients were discharged. Our information on clopidogrel treatment only covers the period after discharge from hospital. We had no information on the decision of the physicians to prescribe clopidogrel treatment after discharge or on the type of follow-up, but we found large variation in discharge prescriptions of clopidogrel between the departments performing CABG, indicating that the choice of therapy was driven by local habits rather than by systematic selection. Selection bias cannot be fully ruled out, although the results were consistent across the propensity score-matched population. Finally, the number of fatal events was low, and thus makes it difficult to draw final conclusions. Clearly, clarification of the causal relationship, the effect and safety of post-operative clopidogrel after CABG, in an adequately powered randomized controlled clinical trial is warranted.
In a cohort of real-life, first-time MI patients revascularized with CABG, only 27% were treated with clopidogrel after surgery. Fewer of these patients died or experienced a combined end point of recurrent MI and death compared with patients without clopidogrel treatment. A focus on discharge clopidogrel treatment of these patients should be made.
This study was funded by the Danish Heart Foundation (08-4-R64-A1885-B641-22470) and the Danish Medical Research Council (271-06-0572). The authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- coronary artery bypass graft surgery
- confidence interval
- International Classification of Diseases
- myocardial infarction
- percutaneous coronary intervention
- Received June 26, 2010.
- Revision received August 30, 2010.
- Accepted September 14, 2010.
- American College of Cardiology Foundation
- Anderson J.L.,
- Adams C.D.,
- Antman E.M.,
- et al.
- Bassand J.P.,
- Hamm C.W.,
- Ardissino D.,
- et al.
- Sanon S.,
- Lee V.V.,
- Elayda M.,
- Wilson J.M.
- Gurbuz A.T.,
- Zia A.A.,
- Vuran A.C.,
- Cui H.,
- Aytac A.
- Fox K.A.,
- Mehta S.R.,
- Peters R.,
- et al.
- Saw J.,
- Topol E.J.,
- Steinhubl S.R.,
- Brennan D.,
- Berger P.B.,
- Moliterno D.J.
- Bhatt D.L.,
- Chew D.P.,
- Hirsch A.T.,
- Ringleb P.A.,
- Hacke W.,
- Topol E.J.
- Budaj A.,
- Yusuf S.,
- Mehta S.R.,
- et al.
- Dunning J.,
- Versteegh M.,
- Fabbri A.,
- et al.
- Tu J.V.,
- Austin P.C.,
- Walld R.,
- Roos L.,
- Agras J.,
- McDonald K.M.
- Gislason G.H.,
- Rasmussen J.N.,
- Abildstrom S.Z.,
- et al.
- Sorensen R.,
- Hansen M.L.,
- Abildstrom S.Z.,
- et al.
- Rasmussen J.N.,
- Gislason G.H.,
- Rasmussen S.,
- et al.
- Ebrahimi R.,
- Dyke C.,
- Mehran R.,
- et al.