Management of preoperative antiplatelet therapy in coronary artery bypass grafting (CABG) is variable among surgeons: guidelines collide with prejudices because replacement of aspirin with low-molecular-weight heparin is still performed because of a presumed minor bleeding risk. This study aims to analyze postoperative bleedings and complications in patients scheduled for elective primary isolated on-pump CABG, depending on preoperative aspirin treatment or its replacement with enoxaparin. In this cohort study, we propensity score matched 200 patients in whom aspirin was stopped at least 5 days before CABG and replaced with enoxaparin and 200 patients who continued aspirin therapy until the day before surgery. Postoperative bleedings and complications were monitored during hospitalization. Among patients who continued aspirin treatment, mean overall bleeding was 701.0 ± 334.6 ml, whereas in the matched enoxaparin group, it was significantly greater (882.6 ± 64.6 ml, p value <0.001); this was associated with reduced postoperative complications, lower values of postoperative C-reactive protein in aspirin takers, and a presumed protective effect for statins. After propensity score adjustment, aspirin treatment carried a protective effect against major postoperative bleeding (odds ratio 0.312, p = 0.001). In conclusion, postoperative bleeding is reduced in patients who continued aspirin, likely due to a reduction in postoperative inflammation. The practice of empirically discontinuing aspirin and replacing it with enoxaparin before CABG should be abandoned. Patients with coronary artery disease referred to CABG should continue antiplatelet medications until the surgical procedure. Those results might be extended to patients under oral anticoagulant therapy requiring CABG.
Coronary artery bypass grafting (CABG) surgery aims to reduce early and long-term risk of myocardial infarction and death in patients with significant coronary artery disease. The continuation of aspirin therapy in patients who undergo coronary surgery, although potentially responsible of intraoperative and postoperative bleedings, can reduce postoperative myocardial infarction, improve oxygenation, and increase survival. Furthermore, platelet inhibition has been shown to reduce the rates of acute bypass graft occlusion in patients undergoing CABG. The problem of bleeding risk usually afflicts cardiac surgeons, and a systematic review concluded that aspirin is associated with an increased risk of postoperative bleeding and greater need for blood products. Therefore, aspirin might be replaced 5 to 7 days before surgery with low-molecular-weight heparin enoxaparin, which exerts its action over 12 to 24 hours, is more easily manageable than unfractionated heparin, and is presumed to be associated with reduced bleeding compared with aspirin because platelet function is not impaired. Although a recent systematic review of recommendations for preoperative antiplatelet discontinuation is remarking that enoxaparin is unable to protect from major cardiac events when used as replacement therapy before surgery, the practice of replacing aspirin with enoxaparin is still in use in many centers worldwide relying more on surgeon’s discretion and concern about intraoperative bleeding rather than scientific evidences or specific randomized clinical trials. Considering the lack of specific reports in this topic, the purpose of this study was to analyze the postoperative bleeding depending on aspirin or enoxaparin therapy in patients scheduled for on-pump CABG and, secondly, to demonstrate a correlation between preoperative aspirin or enoxaparin treatment and postoperative cardiac events.
Methods
This study is based on a propensity score–matched evaluated cohort of consecutive patients who underwent elective primary isolated CABG at our institution (Department of Cardiovascular Surgery, Università Campus Bio-Medico di Roma, Rome, Italy) and treated with aspirin or enoxaparin preoperatively. The primary aim of the study was to analyze postoperative overall bleeding depending on preoperative antiplatelet therapy with aspirin or its replacement with enoxaparin. Secondary aims were the evaluation of postoperative complications in the 2 treatment groups and the possible role of statins in postoperative bleeding and complications.
Data generated by Kamran et al have been used to estimate the sample size, although this calculation is generally used for randomized controlled trials. Considering an expected overall bleeding of 604 ± 348 ml for patients who continued aspirin until surgery, to detect a 20% positive difference assuming 2-sided 5% significance level at 80% power, 131 patients were needed in each group. This number was increased to a recruitment target of 200 patients per group assuming up to 50% noncompliance with the study protocol (such as missing variables) or unexpected variability because this was a cohort study and confounding variables are more difficult to be completely evaluated compared to a randomized clinical trial design. Approval of the study was obtained from the institutional review board.
We considered eligible for our study only patients in whom aspirin therapy (100 mg/day) was stopped at least 5 days before surgery and replaced with subcutaneous enoxaparin (100 IU/kg twice daily) until the day before surgery and patients in whom aspirin therapy (100 mg/day) was continued until the day before surgery. Preoperative treatment was left to surgeon’s discretion. All patients were given their first oral dose of aspirin (100 mg/day) after about 12 to 24 hours from surgical procedure.
Patients in anticoagulation, for example, due to mechanical heart valve or atrial fibrillation, were excluded from the study; also, we excluded patients who received ticlopidine, patients with a history of bleeding disorders, or patients with significant gastrointestinal bleedings for chronic liver failure. Redo cardiac surgery or off-pump coronary surgery were also exclusion criteria. Patients with a known history of systemic inflammatory diseases, malignancies, hematopoietic disorders, or chronic kidney disease (with a creatinine clearance <30 ml/min) and patients receiving long-term steroid or disease-modifying antirheumatic drugs were also excluded. Patients who stopped aspirin 1 to 4 days before surgery and patients on any dual antiplatelet therapy with clopidogrel, prasugrel, or ticagrelor were not included.
Therefore, from January 2010 to January 2015, we evaluated 200 consecutive patients per group accordingly to inclusion and exclusion criteria, with a total study cohort of 400 patients. Patients received a complete preoperative evaluation; after surgical procedure, bleeding from chest tube, blood chemistry parameters, and complications were recorded as part of the routine procedure. From March 2015 to July 2015, we reviewed and collected data of these patients from the electronic medical records of our institution.
All operations were performed by experienced cardiac surgeons with standard techniques; informed consent was obtained from each patient. Induction and maintenance of anesthesia were similar for all patients. All operations were performed through median sternotomy incision and standard aortic and right-atrium 2-staged cannulation. Cardiopulmonary bypass (CPB) was performed using hollow fiber oxygenator and a Stockert roller pump (Saluggia, Vercelli, Italy). During CPB, hematocrit was maintained from 20% to 25% and pump flows were kept at 2.0 to 2.5 L/min/m 2 to keep a mean arterial pressure of 50 to 70 mm Hg. All patients were cooled to moderate hypothermia (mean 32°C), and cardioplegic arrest was achieved with cold crystalloid St. Thomas cardioplegia (4°C) infused into the ascending aorta. Heparin was given at a dose of 300 IU/kg obtaining an activated clotting time of >400 seconds; on completion of anastomoses, heparin effects were reversed by intravenous protamine sulfate (1 mg/300 IU of heparin) to achieve an activated clotting time similar to preoperative values. All anastomoses were sutured by hand. The perioperative need for blood products was determined on an individual, patient-by-patient basis. In general, blood transfusions were given in the presence of hemoglobin <8 g/dl, unless the patient was clinically considered at risk for decreased oxygen delivery. Bypass time, aortic cross-clamp time, and duration of surgery from skin incision to skin closure were recorded, using a standardized protocol.
The postoperative bleeding was calculated in the first, second, and third hour, and then after 12 hours (which corresponds approximately to the bleeding in the day of the operation), at the end of the first postoperative day (POD), and the second POD. Chest tubes were removed when the drainage output became serosanguineous and <20 ml for 6 consecutive hours, and overall bleeding was also recorded. Patients were followed up for any complication during their stay in the intensive care unit and in the postoperative ward (death, reoperation for bleeding, stroke, pacemaker implantation, myocardial infarction, use of intra-aortic balloon pump, or postoperative atrial fibrillation) as part of the routine procedure. Intraoperative and postoperative use of red packed cells, fresh-frozen plasma, and platelets were recorded for both groups. Routine blood tests were performed preoperatively and therefore according to our center’s protocol. Major postoperative bleeding was defined, in agreement with Mehran et al, as one of the following: perioperative intracranial bleeding within 48 hours, reoperation after closure of sternotomy for the purpose of controlling bleeding, blood loss >600 ml in the operative day, transfusion of ≥5 units of whole blood or packed red blood cells within a 48-hour period, and chest tube output >2 L within a 24-hour period.
Categorical data are displayed as frequencies and percentages, and comparisons were made with the Fisher’s exact test or with the chi-square test, as appropriate. Normality criteria were checked for each continuous variable. To test possible differences between not-normally distributed variables, independent-samples Mann–Whitney test was used, and data are presented as median with its 95% confidence interval in parentheses (95% CI). Paired- or independent-sample t test was used for normally distributed data, which are presented as mean ± SD. To evaluate aspirin effect considering any possible confounding factor and removing the potential bias of the patient’s selection, we performed propensity score analysis using kernel matching on logit regression. We therefore performed the logistic regression with propensity score adjustment and comparison using a paired-sample t test of postoperative overall bleeding between aspirin-treated patients and matched controls. The effect of statin treatment and its association with aspirin was evaluated using a multivariate logistic regression with major postoperative bleeding as dependent variable. Variables were included in the multivariate logistic regression model using a stepwise approach in case of univariate p value <0.15; then statin treatment was added and model fitness improved. Hence, interaction between statin and aspirin was evaluated. The presented model, with aspirin–statin interaction, was generated using 394 observations and is associated with a log-likelihood of −134.95; goodness-of-fit was evaluated using the Hosmer-Lemeshow test (chi-square with 8 df = 3.03, p value = 0.932) and area under the receiver operating characteristic curve was 0.732, thus allowing to assume its reliability. A p value <0.05 was considered statistically significant. Interaction between aspirin or enoxaparin treatment and statin was evaluated with 1-way analysis of variance with multiple pairwise comparisons using Bonferroni’s test, in which a p value <0.01 was considered statistically significant. Statistical analysis was performed using Stata Statistics (version 13.0, 2013; StataCorp, College Station, Texas).
Results
Preoperative characteristics are outlined in Table 1 . There are no statistically significant differences among cardiac risk factors, therapy on admission, and blood test results between groups. All patients received elective primary isolated CABG, with at least 1 arterial graft using the left internal mammary artery; intraoperative and perioperative variables were similar between groups and are outlined in Table 2 .
| Variable | Enoxaparin replacement ( n = 200 ) | Aspirin treatment ( n = 200 ) | P value |
|---|---|---|---|
| Age (years) | 68.1±9.4 | 67.8±9.9 | 0.734 |
| Men | 165 (82.5%) | 165 (82.5%) | 1.000 |
| Hypertension | 193 (96.5%) | 198 (99.0%) | 0.175 |
| Diabetes mellitus | 84 (42.0%) | 73 (36.5%) | 0.306 |
| Insulin therapy | 39 (19.5%) | 50 (25.0%) | 0.229 |
| Smoker | 52 (26.0%) | 39 (19.5%) | 0.152 |
| Past-smoker | 84 (42.0%) | 88 (44.0%) | 0.762 |
| Statins therapy | 155 (77.5%) | 140 (70%) | 0.111 |
| Atorvastatin | 114 (57.0%) | 100 (50.0%) | |
| Simvastatin | 22 (11.0%) | 28 (14.0%) | |
| Rosuvastatin | 15 (7.5%) | 9 (4.5%) | |
| Others | 4 (2.0%) | 3 (1.5%) | |
| Statins: average dosage (mg) | 40.2 (36.7 – 43.7) | 43.3 (39.4 – 47.1) | 0.212 |
| Bronchodilators | 10 (5.0%) | 5 (2.5%) | 0.292 |
| Previous percutaneous transluminal coronary angioplasty | 36 (18.0%) | 45 (22.5%) | 0.320 |
| Extracardiac arteriopathy | 55 (27.5%) | 70 (35.0%) | 0.131 |
| Recent acute coronary syndrome | 64 (32.0%) | 63 (31.5%) | 1.000 |
| Previous acute coronary syndrome | 39 (19.5%) | 37 (18.5%) | 0.899 |
| Body mass index (kg/m 2 ) | 26.8±3.2 | 26.9±3.6 | 0.753 |
| Body mass index > 30 kg/m 2 | 37 (18.5%) | 32 (16.0%) | 0.597 |
| Creatinine clearance (mL/min) | 82.1±26.7 | 83.3±30.6 | 0.670 |
| Preoperative ejection fraction (%) | 54.5±7.2 | 55.1±6.2 | 0.408 |
| Ejection fraction < 35% | 10 (5.0%) | 6 (3.0%) | 0.445 |
| Preoperative haemoglobin (g/dL) | 13.5 (13.3 – 13.7) | 13.5 (13.2 – 13.7) | 0.989 |
| Preoperative haematocrit (%) | 40.5 (39.9 – 41.0) | 40.3 (39.7 – 40.9) | 0.738 |
| Preoperative platelet count (/mm 3 ) | 213 000(205000-222000) | 211 000(203000-218000) | 0.857 |
| Creatinine (mg/dL) | 0.96 (0.93 – 1.00) | 0.99 (0.95 – 1.04) | 0.544 |
| International normalized ratio (INR) | 1.09 (0.98 – 1.20) | 1.02 (1.01 – 1.03) | 0.564 |
| Partial thromboplastin time (PTT) (s) | 30.7 (29.8 – 31.6) | 30.9 (30.0 – 31.7) | 0.300 |
| Fibrinogen (mg/dL) | 369.7(354.8 – 384.6) | 367.5(354.6 – 380.4) | 0.807 |
| C reactive protein (mg/L) | 8.6 (5.9 – 11.3) | 9.5 (6.5 – 12.5) | 0.313 |
| White blood cell count (10 9 /L) | 7.4 (7.1 – 7.7) | 7.4 (7.1 – 7.7) | 0.948 |
| Troponin I (ng/L) | 0.3 (0.1 – 0.4) | 0.5 (0.3 – 0.7) | 0.435 |
| Variable | Enoxaparin replacement ( n = 200 ) | Aspirin treatment ( n = 200 ) | P value |
|---|---|---|---|
| Mean EuroSCORE-2 | 1.6±0.4 | 1.6±0.5 | 1.000 |
| SYNTAX Score | 0.530 | ||
| ≤ 22 | 19 | 26 | |
| 23-32 | 123 | 120 | |
| ≥ 33 | 58 | 54 | |
| Number of grafts | 0.344 | ||
| 2 grafts | 88 | 91 | |
| 3 grafts | 93 | 82 | |
| 4 grafts | 19 | 27 | |
| Cardiopulmonary bypass time (min) | 69.2±19.2 | 65.9±21.7 | 0.106 |
| Aortic cross-clamp time (min) | 50.5±16.0 | 53.3±18.6 | 0.113 |
| Duration of surgical procedure (min) | 203.0±60.3 | 193.5±47.9 | 0.162 |
| Oro-tracheal intubation (hours) | 10.7 (9.4 – 11.9) | 9.4 (8.7 – 10.0) | 0.204 |
| Intensive care unit stay (days) | 1.3 (1.1 – 1.4) | 1.1 (1.1 – 1.2) | 0.341 |
| Chest drain removal (POD days) | 2.0±0.5 | 1.9±0.7 | 0.101 |
| RPC: transfused patients | 142 (71.0%) | 130 (65.0%) | 0.238 |
| RPC: transfused units per patient | 1.8±1.0 | 1.7±0.8 | 0.402 |
| PLT: transfused patients | 6 (3.0%) | 2 (1.0%) | 1.000 |
| FFP: transfused patients | 8 (4.0%) | 2 (1.0%) | 0.105 |
| Hospital discharge (POD days) | 6.5 (5.9 – 7.2) | 6.5 (6.1 – 6.9) | 0.150 |
Postoperative bleeding was significantly less in patients who continued aspirin compared to patients who replaced it with enoxaparin in the first hour from surgery, in the next 12 hours, in the first POD, and considering overall bleeding (p value <0.001). Data about postoperative bleeding are summarized in Table 3 and Figure 1 . Duration of surgical procedure was similar between groups, indicating that observed findings cannot be related to differences in the time spent for hemostasis, and no differences were found in the number of patients transfused with red packed cells, platelets, or fresh-frozen plasma ( Table 2 ).
| Postoperative bleeding (ml) | Enoxaparin replacement ( n = 200 ) | Aspirin treatment ( n = 200 ) | P value | ||
|---|---|---|---|---|---|
| Median | 95%CI | Median | 95%CI | ||
| 1 st hour | 70.0 | 60.0-76.7 | 60.0 | 50.0-60.0 | 0.005 |
| 2 nd hour | 35.0 | 30.0-40.0 | 30.0 | 25.0-35.0 | 0.096 |
| 3 rd hour | 30.0 | 25.0-30.0 | 25.0 | 20.0-25.0 | 0.050 |
| Next 12 hours | 362.5 | 323.3-390.0 | 270.0 | 245.0-280.0 | < 0.001 |
| 1 st POD | 330.0 | 310.0-366.7 | 255.0 | 245.0-271.7 | < 0.001 |
| 2 nd POD | 65.0 | 50.0-70.0 | 45.0 | 35.0-70.0 | 0.057 |
| Overall bleeding | 797.0 | 738.3-885.0 | 636.0 | 580.0-665.0 | < 0.001 |
Postoperative complications are discussed in Table 4 , and a higher number of all postoperative complications were observed in patients who used enoxaparin as replacement therapy.
| Enoxaparin replacement ( n = 200 ) | Aspirin treatment ( n = 200 ) | P value | |
|---|---|---|---|
| Deaths | 1 (0.5%) | 1 (0.5%) | 1.000 |
| Reoperation for bleeding | 7 (3.5%) | 2 (1.0%) | 0.175 |
| Pacemaker implantation | 3 (1.5%) | 0 (0.0%) | 0.248 |
| Stroke | 5 (2.5%) | 1 (0.5%) | 0.215 |
| Myocardial infarction | 1 (0.5%) | 1 (0.5%) | 1.000 |
| Intra-aortic balloon pump | 2 (1.0%) | 1 (0.5%) | 1.000 |
| Overall complications ∗ | 19 (9.5%) | 6 (3.0%) | 0.012 |
| Atrial fibrillation | 53 (26.5%) | 38 (19.0%) | 0.095 |
| Major post-operative bleedings | 36 (18.0%) | 14 (7.0%) | 0.001 |
∗ Deaths, reoperations for bleeding, pacemaker implantation, stroke, myocardial infarction, intra-aortic balloon pump.
Major postoperative bleedings occurred in 36 patients in group 1 and in 14 patients in group 2 (p value = 0.001). In univariate logistic regression analysis, raw odds ratio (OR) for aspirin treatment in determining major postoperative bleeding was 0.343 (95% CI 0.179 to 0.658, p value = 0.001). After propensity score adjustment, aspirin still carries a statistically significant protective effect with an adjusted OR of 0.312 (95% CI 0.156 to 0.623, p value = 0.001). Postoperative overall bleeding was compared between aspirin-treated patients and propensity score–matched controls using paired-sample t test. Among patients who continued aspirin treatment, mean overall bleeding was 701.0 ± 334.6 ml, whereas in the matched enoxaparin group, it was significantly greater (882.6 ± 64.6 ml, p value <0.001).
We therefore focused on inflammatory reaction after CPB to elucidate some potential association with the incidence of major bleeding. C-reactive protein values tended to be lower in patients in group 2 than those in group 1 in the first and third PODs. Results are summarized in Table 5 . Considering our results, we may hypothesize that the inflammatory component might play a role in the determinism of bleeding. Besides aspirin, a strong anti-inflammatory effect is also played by statins and multivariate logistic regression models were used to evaluate its role; results are presented in Table 6 . The combined treatment of aspirin and statin proved to have a beneficial effect that exceeded single results, with an OR of 0.212 (95% CI 0.048 to 0.937, p value = 0.041). Lastly, the effect of preoperative aspirin or enoxaparin treatment and statins in overall postoperative bleeding was evaluated ( Table 7 ). In post hoc analysis, results indicate that statin carries a protective effect only when used in combination with aspirin.
