Preoperative antiplatelet use does not increase incidence of bleeding after major operations

Preoperative antiplatelet use does not increase incidence of bleeding after major operations

ARTICLE IN PRESS Preoperative antiplatelet use does not increase incidence of bleeding after major operations David S. Strosberg, MD,a Todd Corbey, B...

300KB Sizes 0 Downloads 85 Views


Preoperative antiplatelet use does not increase incidence of bleeding after major operations David S. Strosberg, MD,a Todd Corbey, BS,c Jon C. Henry, MD,d and Jean E. Starr, MD,b Columbus and Athens, OH, and Pittsburgh, PA

Background. This study examined the outcomes of patients holding or continuing clopidogrel during the preoperative period. Methods. We reviewed all patients taking clopidogrel who underwent one of 72 different Current Procedural Terminology code procedures, representing major emergency and elective general thoracic and vascular operations from 2009–2012 at a single institution. Demographics, comorbidities, aspirin use, details of coronary stents, and perioperative events were collected. Results. A total of 2,154 major operative procedures were performed on 1,851 patients during the study period. A total of 213 patients (11.5%) were taking clopidogrel at the time of their last office visit or hospital admission and were then instructed to hold or continue the drug prior to an operation. A total of 205 procedures in 200 patients comprised the final study population. Clopidogrel was held in 116 procedures for $5 days prior to operative intervention (56.6%, Group A), and clopidogrel was administered within 5 days of an operation in 89 procedures (43.4%, Group B). There were no differences between the 2 groups regarding estimated blood loss, units transfused, myocardial infarction, stroke, acute visceral or peripheral ischemia, or death within 30 days. Conclusion. We did not identify significantly increased adverse patient outcomes in those patients who received preoperative clopidogrel within this population. We assert that it appears to be reasonable and safe to continue antiplatelet therapy with clopidogrel in this population in elective situations and that preoperative clopidogrel use does not increase the risk of bleeding in emergency circumstances. (Surgery 2016;j:j-j.) From the Department of Surgerya and the Division of Vascular Diseases and Surgery,b The Ohio State University Wexner Medical Center, Columbus; the Ohio University College of Medicine,c Athens, OH; and the Division of Vascular Surgery,d University of Pittsburgh Medical Center, Pittsburgh, PA

DUAL ANTIPLATELET THERAPY WITH ACETYLSALICYLIC ACID (ASA) and clopidogrel (Plavix; Bristol-Myers Squibb, New York, NY) in the early postintervention period after placement of a percutaneous coronary artery is considered the standard of care to prevent in-stent thrombosis.1 Antiplatelet agents are also employed frequently in patients at high risk for sequelae of coronary, cerebrovascular, and peripheral arterial disease. Antiplatelet drug management in patients with pre-existing coronary artery diseases remains an important consideration when these patients require an emergency or major elective operative Presented at the Central Surgical Association 2016 Conference in Montreal, Quebec, Canada, on March 11, 2016. Accepted for publication May 19, 2016. Reprint requests: David S. Strosberg, MD, Department of Surgery, The Ohio State University Wexner Medical Center, Columbus, OH 43026. E-mail: [email protected] 0039-6060/$ - see front matter Ó 2016 Elsevier Inc. All rights reserved.

procedure. Surgeons often face the dilemma of temporarily stopping antiplatelet therapy and risking adverse cardiac outcomes, specifically coronary stent thrombosis, or continuing treatment risking increased operative or postoperative bleeding and its associated consequences.2,3 Because the lifespan of a platelet is 8–9 days, the current standard of care includes stopping antiplatelet therapy for 5–7 days before an elective operative procedure or administering platelet transfusions in patients requiring an emergency operation when indicated. ASA acts by inhibiting irreversibly platelet production, whereas clopidogrel inhibits platelet activation and aggregation. The prescribing information from BristolMyers Squibb recommends holding clopidogrel for 5 days prior to major elective operations to decrease the risk of bleeding4; recommendations may vary in patients with bare stents during the first year after implantation. Several studies have investigated the use of antiplatelet drugs and adverse effects in cardiac surgery; however, few studies have examined the perioperative management of patients undergoing SURGERY 1

ARTICLE IN PRESS 2 Strosberg et al

general surgery, thoracic surgery, and vascular surgery.3 Previous studies have reported that preoperative clopidogrel use should not delay an operation in spite of the risk of bleeding.5,6 We designed a retrospective review to evaluate whether use of perioperative antiplatelet drugs, specifically clopidogrel, increased the rate of adverse operative outcomes during or after major emergency or elective general, thoracic, and vascular operations. We hypothesized that use of preoperative antiplatelet drugs would increase the rates of operative and postoperative bleeding as well as morbidity and mortality. METHODS Approval by our Institutional Review Board was obtained. We identified 72 different Current Procedural Terminology codes to represent major emergency and elective general, thoracic, and vascular operations. Codes were chosen based on the complexity of the procedure, high likelihood for operative blood loss, and likelihood of being performed in an emergency situation. The institutional database repository was queried for all patients who underwent one of the selected procedures at our large, academic medical center from January 1, 2009, to December 31, 2012. Patients taking clopidogrel at the time of their latest office visit or hospital admission just prior to an operation were included for study. All patients had a history and physical examination within 30 days as required by our institutional policy. We determined if the drug had been stopped or continued by reviewing the surgeon’s note and the medication reconciliation record at the time of hospital admission. If a patient underwent multiple procedures, only the first procedure per hospitalization was included for study. Patients were excluded if it could not be deciphered from the electronic medical record whether clopidogrel was held or administered preoperatively. Demographics, comorbidities, aspirin use (any dose), details of any indwelling coronary artery stents, and perioperative events were collected. A comorbid diagnosis was considered when it was stated clearly by the physician or surgeon in the preoperative history and physical examination. Statistical significance was determined using the Fisher exact test for discrete data and the Student t test for continuous data. RESULTS A total of 72 different Current Procedural Terminology codes included 8 thoracic, 36 vascular, and 28 general operative codes associated

Surgery j 2016

with 550, 651, and 953 procedures, respectively. In total, 2,154 major operations were performed on 1,851 patients during the study period across all hospitalizations. Of these, 213 patients (11.5%) were taking clopidogrel at the time of their last office visit or hospital admission just prior to an operation. Thirteen patients were excluded, because it was unclear from the medical record if clopidogrel was discontinued preoperatively. Five patients underwent 2 of the defined procedures during separate hospitalizations, with a total of 205 procedures in 200 patients to comprise the study population (Table I). Clopidogrel was stopped preoperatively in 116 procedures in 112 patients for $5 days prior to operative intervention (56.6%, Group A), and clopidogrel was administered within 5 days of an operation for 89 procedures in 88 patients (43.4%, Group B; Fig). Patients in Group A were more likely to have aspirin of any dose withheld prior to an operation (P < .001). Group A also had a greater rate of chronic obstructive pulmonary disease (35.3% vs 14.6%, P < .01) and malignancy (40.5% vs 15.7%, P < .01), as well as a lesser prevalence of peripheral arterial disease (62.1% vs 86.5%, P < .01) and presence of a peripheral vascular stent (19.8% vs 39.3%, P < .01; Table II). No patients in Group A received preoperative platelets compared to 2 patients in Group B (2.1%) who received an average of 1.5 units of pooled platelets (P = not significant). Three patients in Group A suffered an acute perioperative myocardial infarction compared to one patient in Group B (2.6% vs 1.1%, P > .6). In Group A, 29.3% of patients required immediate transfusion (within 48 hours of an operation) or delayed transfusion (48 hours from operation to discharge) compared to 38.2% of Group B patients (P = .23). Additionally, there were no significant differences between Group A and Group B regarding intraoperative estimated blood loss (390 mL vs 300 mL, P = .19), number of units transfused (2.7 vs 2.5, P = .81), cerebrovascular accident (0.9% vs 2.2%, P = .58), acute visceral or peripheral ischemia (2.6% vs 3.4%, P = 1.00), or death within 30 days (2.6% vs 3.4%, P = 1.00; Table III). When patients who received preoperative platelets were excluded, there was no statistical difference in perioperative transfusion, estimated blood loss, myocardial infarction, cerebrovascular accident, acute visceral or peripheral ischemia, or death within 30 days (P $ .22 each). Three patients who had clopidogrel withheld suffered a perioperative acute myocardial infarction; one of these patients suffered cardiac arrest

ARTICLE IN PRESS Strosberg et al 3

Surgery Volume j, Number j

Table I. Comparison of Current Procedural Terminology (CPT) codes for patients taking clopidogrel

CPT code 32100 32220 32225 32440 32480 32482 32484 32500 33877 35081 35102 35141 35151 35556 35571 35583 35585 35587 35606 35621 35623 35631 35646 35647 35654 35656 35661 35666 35671 35697 35905 39502 43820 44120 44140 44141 44143 47130 48140 48150 49000 Total

Description Thoracotomy, major with exploration Release of lung Partial release of lung Removal of lung Partial removal of lung, single Bilobectomy Segmentectomy Partial removal of lung, wedge Repair of thoracoabdominal aortic aneurysm with graft Direct repair of abdominal aortic aneurysm Direct repair of abdominal aortic aneurysm involving iliac vessels Repair common femoral artery aneurysm Repair popliteal artery aneurysm Bypass with vein, femoral-popliteal Bypass with vein, popliteal-tibial Bypass with vein (in-situ), femoral-popliteal Bypass with vein, femoral-anterior tibial Bypass with vein, popliteal-tibial Bypass graft, carotid-subclavian Bypass graft, axillary-femoral Bypass graft, axillary-popliteal Bypass graft, aorto-celiac Bypass graft, aorto-bifemoral Bypass graft, aorto-femoral Bypass graft, axillary-femoral-femoral Bypass graft, femoral-popliteal Bypass graft, femoral-femoral Bypass graft, femoral-anterior tibial Bypass graft, popliteal-tibial Reimplant artery, each Excision of infected graft, thorax Repair paraesophageal hernia Gastrojejunostomy Enterectomy, with anastomosis Colectomy, partial, with anastomosis Colectomy, partial, with cecostomy Colectomy, partial, with colostomy Hepatectomy, total right lobectomy Pancreatectomy, distal subtotal Whipple, with pancreaticojejunostomy Exploratory laparotomy

with ST elevation several hours after thoracotomy with lobectomy and ultimately expired; this patient was not taking aspirin preoperatively. Another patient suffered an ST-elevation myocardial infarction on postoperative day 5 requiring cardiac catheterization and succumbed ultimately to cardiogenic shock on postoperative day 7. The last patient in this group suffered a non–STelevation myocardial infarction (NSTEMI) on

Clopidogrel held more than 5 days before operation (n = 116)

Clopidogrel received within 5 days before operation (n = 89)

1 0 1 3 18 1 2 3 1 1 6

0 3 2 1 1 0 0 0 0 2 0

3 1 4 0 2 3 1 1 3 1 1 8 0 2 19 2 4 0 1 1 1 3 2 4 0 1 1 2 6 2 116

3 1 19 1 1 3 0 2 2 0 1 2 1 2 17 5 8 2 0 0 0 0 1 3 1 1 0 0 0 4 89

postoperative day 1 after a left hemicolectomy; cardiac catheterization showed in-stent thrombosis requiring thrombectomy; in both patients, aspirin was held preoperatively. In contrast, one patient who did not have an interruption in clopidogrel or aspirin suffered an NSTEMI after an extra-anatomic vascular bypass; catheterization revealed a previously occluded right coronary artery without new changes.

ARTICLE IN PRESS 4 Strosberg et al

Fig. Flowchart of patients in each group and aspirin use.

Regarding cerebrovascular accidents, one patient in Group A also withheld aspirin preoperatively, and 2 patients in Group B received aspirin. No patients in either group expired from hemorrhagic shock within 30 days of an operation. Emergency operations, defined as operative intervention within 24 hours of initial presentation, occurred in 19.1% of patients in Group B; all those in Group A were elective (P < .01). A subgroup analysis of emergency operative procedures in Group B demonstrated a difference in total number of patients who received preoperative platelet transfusions (11.76% vs 0%, P = .03). There were more patients who required perioperative red blood cell transfusion and had greater estimated blood loss, however, these were not statistically significant (P $ .18 each). There was no difference in the other parameters evaluated between the emergency and nonemergency subgroups (Table IV). DISCUSSION These data represent what we believe to be the largest, albeit retrospective review of general surgical, thoracic, and vascular patients to determine the effect of antiplatelet use, specifically clopidogrel, on adverse operative outcomes. The rate of clopidogrel use in patients undergoing a major operation is considerable, nearly 12% of our sample. There was a slightly greater incidence of immediate or delayed transfusion in those who received clopidogrel preoperatively, but this effect was not significant. Further, there were no differences between the 2 groups regarding intraoperative estimated blood loss, number of units transfused, or specific patient-oriented outcomes, including myocardial infarction, stroke, acute

Surgery j 2016

visceral or peripheral ischemia, or death within 30 days. We found that more patients had peripheral artery disease and the presence of a peripheral intravascular stent in the group whose clopidogrel use was withheld. At our institution, patients undergoing open vascular procedures are instructed generally to continue their antiplatelet therapy given the high incidence of concomitant coronary artery disease in this population. Two patients who did not have an interruption in clopidogrel administration received preoperative platelets due to concern for preoperative bleeding, the emergency nature of the operation, and the choice of the operating team. Platelets were also administered prophylactically due to the presumed impaired function of the circulating platelets, again at the choice of the surgeon. One of these patients presented with an acute gastrointestinal bleed requiring massive transfusion of packed red blood cells and fresh frozen plasma prior to laparotomy, and the other presented with septic shock with concern for gallbladder perforation. To our knowledge, no other studies noted that platelets were administered prophylactically prior to major emergency operations, unless the massive transfusion protocol was initiated. There is adequate consensus that dual antiplatelet therapy should be continued to prevent coronary artery in-stent thrombosis. A review of 91 publications by Brilakis et al7 found dual antiplatelet therapy with aspirin and a P2Y12 inhibitor (Adenosine diphosphate antagonist) decreased the risk of stent thrombosis and subsequent cardiovascular events after percutaneous coronary artery intervention and remains the standard of care. Dual antiplatelet therapy is required for at a minimum of 3 months postplacement of a bare-metal stent and for 12 months after placement of a drug-eluting stent to prevent in-stent thrombosis. Accordingly, in the European literature, coronary artery bare-metal stents are recommended to be placed in patients if an elective operation is required within 12 months of placement. In the semiurgent situation, multidisciplinary consultation should be performed to evaluate the individual risk of thrombotic events and bleeding, and in the emergency situation, operations should be performed under full antiplatelet therapy.8 The investigation of clopidogrel use in the perioperative period has been investigated previously in major cardiac procedures. A study by Karabulut et al9 found the need for reexploration, transfusion of blood products, chest tube output, duration of stay in the intensive care

ARTICLE IN PRESS Strosberg et al 5

Surgery Volume j, Number j

Table II. Comparison of demographic/clinical factors between Groups A and B Demographic/clinical factor*

Group A- clopidogrel withheld (n = 116)

Group B- clopidogrel given (n = 89)

P value

Sex (male) Age, y (mean) Diabetes Cerebrovascular event Coronary artery disease Myocardial infarction Renal disease Hypertension Smoker---current or former COPD Hyperlipidemia Peripheral vascular disease Malignancy Hypercoagulable disorder Presence of coronary artery stent Presence of peripheral vascular stent Emergency operation (within 24 hours of presentation)

75 (65.3%) 64.1 46 (39.7%) 22 (19.0%) 87 (75.0%) 41 (35.3%) 12 (10.3%) 96 (82.8%) 101 (87.1%) 41 (35.3%) 74 (63.8%) 72 (62.1%) 47 (40.5%) 0 47 (40.5%) 23 (19.8%) 0

54 (60.7%) 60.8 37 (41.6%) 12 (13.5%) 61 (68.5%) 20 (22.5%) 12 (13.5%) 78 (87.6%) 70 (78.7%) 13 (14.6%) 62 (69.7%) 77 (86.5%) 14 (15.7%) 2 (2.2%) 28 (31.5%) 35 (39.3%) 17 (19.1%)

.56 .09 .89 .35 .35 .06 .52 .43 .13 <.01 .46 <.01 <.01 .99 .19 <.01 <.01

*Values are n (%) unless specified otherwise; Student t test or Fisher exact test. COPD, Chronic obstructive pulmonary disease.

Table III. Comparison of perioperative events between Groups A and B Perioperative event*

Group A- clopidogrel withheld (n = 116)

Group B- clopidogrel given (n = 89)

P value

Total patients who required preoperative platelet transfusions Total patients who required perioperative transfusion Number of packed red blood cell transfusionsy (mean) Estimated blood lossz (mean) Myocardial infarction Cerebrovascular event Acute lower extremity or visceral ischemia Death within 30 days

0 34 (29.3%) 2.65 390.80 3 (2.6%) 1 3 (2.6%) 3 (2.6%)

2 (2.2%) 34 (38.2%) 2.53 300.90 1 2 (2.2%) 3 (3.4%) 3 (3.4%)

.99 .23 .81 .19 .63 .58 .99 .99

*Values are n (%) unless specified otherwise; Student t test or Fisher exact test. yExpressed in units. zExpressed in mL.

unit, and duration of hospital stay were similar in patients who underwent isolated coronary artery bypass grafting with or without perioperative use of clopidogrel. Chen et al10 examined 45 patients who received clopidogrel within 6 days of a cardiac operation and found slightly greater transfusion requirements and lesser periods of controlled ventilation compared to a control group. They were able to predict preoperative platelet dysfunction before heparin administration for cardiopulmonary bypass using adenosine diphosphate aggregometry in this group and developed a strict transfusion algorithm to decrease transfusion requirements in patients undergoing elective coronary artery bypass grafting after recent clopidogrel exposure.

There is a paucity of literature addressing the outcomes of patients taking clopidogrel who undergo noncardiac operations. Ozao-Choy et al5 investigated bleeding complications retrospectively in patients undergoing general operations while on clopidogrel and found substantial bleeding after operations requiring transfusion in patients who took clopidogrel within 6 days before the operation; in this study, however, the sample size was small (28 vs 22), and there were no differences between the groups in operative or postoperative blood transfusions or decreases in hematocrit, hospital stay, intensive care unit stay, late complications, or mortality. A similar study by Chernoguz et al11 at the same institution also noted an increased rate of postoperative bleeding requiring

ARTICLE IN PRESS 6 Strosberg et al

Surgery j 2016

Table IV. Subgroup analysis: Comparison of perioperative events between emergency and nonemergency operations for patients who received clopidogrel within 5 days of an operation (Group B) Perioperative event*

Emergency (n = 17)

Non-emergency (n = 72)

P value

Total patients who received preoperative platelet transfusions Total patients who received perioperative transfusions Number of packed red blood cell transfusionsy (mean) Estimated blood lossz (mean) Myocardial infarction Cerebrovascular event Acute lower extremity or visceral ischemia Death within 30 days

2 (11.76%) 9 (52.9%) 2.78 357.35 0 1 2 (11.8%) 2 (11.8%)

0 (0.0%) 25 (34.7%) 2.44 287.60 1 1 1 1

.03 .18 .68 .63 .99 .35 .09 .09

*Values are n (%) unless specified otherwise; Student t test or Fisher exact test. yExpressed in units. zExpressed in mL.

transfusion in patients who received clopidogrel within 7 days of an abdominal operation. Similarly, their study population size was small (13 vs 8 patients), and they did not find a difference in the need for admission to the intensive care unit or mortality. Regarding thoracic operations, Cerfolio et al6 investigated 33 patients on clopidogrel prospectively who underwent open and minimally invasive thoracic procedures including thoracotomy with lobectomy, video-assisted wedge resection, mediastinoscopy, and Ivor Lewis esophagogastrectomy. They found no intraoperative morbidity or bleeding after a primary thoracotomy; however, patients who underwent redo thoracotomy required transfusion. They also found that overall morbidity, mortality, and duration of stay was no different when compared to a propensity score– matched control group. A study by Saadeh and Sfeir12 examined 647 consecutive major peripheral arterial operations over a 7-year period, comparing cardiovascular and bleeding outcomes of patients who were on dual antiplatelet therapy with patients who were not. The authors found no bleeding-related deaths, a similar incidence of stable hematomas, and similar transfusion requirements. Fujikawa et al13 investigated 519 patients on antiplatelet therapy who underwent gastrointestinal operations and found an increased rate of bleeding compared to patients who were not taking antiplatelet therapy; however, patients on clopidogrel or other thienopyridines represented a small percentage (15%) of their cohort. Our study is not without limitations. Our study is limited by the retrospective design, the relatively small number of patients taking clopidogrel, and the heterogenous grouping of operations, all of which limit our ability to make a robust statistical argument. Not all medical records were complete,

because our time parameters included records before our electronic medical record system was fully functional at our institution. Outcome parameters, including estimated blood loss and transfusion triggers, are often subjective approximations by the surgeon and may be over- or underestimated by multiple biases. Our study did not investigate preoperative and postoperative hemoglobin, which would have been a more objective measure of blood loss. Further, we did not investigate rates of reoperation specifically for bleeding. At the time the study was executed, only clopidogrel was prevalent; accordingly, our results may not correlate to other anti-ADP antagonists, such as prasugrel, ticlopidine, ticagrelor, or cangrelor. Nevertheless, these limitations are unlikely to have affected our main conclusions, because objective measures, including adverse cardiac events and number of transfusions, were defined clearly and investigated. Our study supports the notion that further investigation is needed to give guidance for this common dilemma for a wide range of patients and surgeons. Ideally, we could institute a randomized control trial to provide level I evidence to support or refute current guidelines regarding antiplatelet use before major operations. At our institution, recent clinical practice guidelines were established to encourage the continuation of low-dose aspirin therapy and to only withhold nonaspirin, antiplatelet therapy if the bleeding risks were thought to outweigh the risk of thrombosis based on type of stent, date placed, and location. This practice is in accordance with the current perioperative guidelines from the American College of Cardiology for ASA use.14 If clopidogrel is taken for prior coronary or vascular intervention, consultation with the physician who implanted the original stent is encouraged. Exceptions to continuing low-dose aspirin

ARTICLE IN PRESS Surgery Volume j, Number j

include intracranial and spinal canal operations, posterior chamber of the eye operations, and certain urologic operations because of the potential ominous consequences of bleeding and difficulty of compressing these operative sites. In our study, aspirin of any dose was included for analysis. Based on the results of our study, we recommend continued antiplatelet administration in patients with planned elective operations and do not recommend platelet transfusion in the emergency setting unless massive transfusion is required.

REFERENCES 1. Huber K. Clopidogrel in coronary artery disease: update 2012. Adv Cardiol 2012;47:31-8. 2. Hall R, Mazer CD. Antiplatelet drugs: a review of their pharmacology and management in the perioperative period. Anesth Analg 2011;112:292-318. 3. Metzler H, Kozek-Langenecker S, Huber K. Antiplatelet therapy and coronary stents in perioperative medicine– the two sides of the coin. Best Pract Res Clin Anaesthesiol 2008;22:81-94. 4. Bristol-Myers-Squibb. Full prescibing information: PLAVIX (clopidogrel bisulfate) tablets [Internet]. Bridgewater (NJ): Bristol-Myers Squibb/Sanofi Pharmaceuticals Partnership; 2015. Available from: com/pi/pi_plavix.pdf. 5. Ozao-Choy J, Tammaro Y, Fradis M, Weber K, Divino CM. Clopidogrel and bleeding after general surgery procedures. Am Surg 2008;74:721-5. 6. Cerfolio RJ, Minnich DJ, Bryant AS. General thoracic surgery is safe in patients taking clopidogrel (Plavix). J Thorac Cardiovasc Surg 2010;140:970-6. 7. Brilakis ES, Patel VG, Banerjee S. Medical management after coronary stent implantation: a review. JAMA 2013;310: 189-98. 8. Korte W, Cattaneo M, Chassot PG, Eichinger S, von Heymann C, Hofmann N, et al. Peri-operative management of antiplatelet therapy in patients with coronary artery disease: joint position paper by members of the working group on Perioperative Haemostasis of the Society on Thrombosis and Haemostasis Research (GTH), the working group on Perioperative Coagulation of the Austrian Society for Anesthesiology, Resuscitation and Intensive Care (OGARI) and the Working Group Thrombosis of the European Society for Cardiology (ESC). Thromb Haemost 2011;105:743-9. 9. Karabulut H, Toraman F, Evrenkaya S, Goksel O, Tarcan S, Alhan C. Clopidogrel does not increase bleeding and allogenic blood transfusion in coronary artery surgery. Eur J Cardiothorac Surg 2004;25:419-23. 10. Chen L, Bracey AW, Radovancevic R, Cooper JR Jr, Collard CD, Vaughn WK, et al. Clopidogrel and bleeding in patients undergoing elective coronary artery bypass grafting. J Thorac Cardiovasc Surg 2004;128:425-31. 11. Chernoguz A, Telem DA, Chu E, Ozao-Choy J, Tammaro Y, Divino CM. Cessation of clopidogrel before major abdominal procedures. Arch Surg 2011;146:334-9. 12. Saadeh C, Sfeir J. Discontinuation of preoperative clopidogrel is unnecessary in peripheral arterial surgery. J Vasc Surg 2013;58:1586-92. 13. Fujikawa T, Tanaka A, Abe T, Yoshimoto Y, Tada S, Maekawa H. Effect of antiplatelet therapy on patients

Strosberg et al 7

undergoing gastroenterological surgery: thromboembolic risks versus bleeding risks during its perioperative withdrawal. World J Surg 2015;39:139-49. 14. Abualsaud AO, Eisenberg MJ. Perioperative management of patients with drug-eluting stents. JACC Cardiovasc Interv 2010;3:131-42.

DISCUSSION Dr Michael Dalsing: Thank you, Dr Strosberg and your colleagues, for sending me the manuscript well in advance of the meeting. This is certainly a clear and concise presentation, and it is a clinically relevant topic. Your original hypothesis was that preoperative clopidogrel would increase the rate of surgical bleeding as well as overall morbidity and mortality. Would the null hypothesis have been a more direct statistical benchmark to prove or disprove your study? Did you use a certain reference to define patient comorbidities, for example, hypertension or renal failure? Was the presence of renal disease based on serum creatinine, GFR, or the need for dialysis? Really, without a clear definition, it is hard to tell how these factors will impact your results. You mentioned that some patients will take clopidogrel at the last clinic visit. That was one of your cohorts in group A. Is it always within 5 days? How do you know if they actually stopped clopidogrel? I was surprised that not even 1 patient went back for postoperative bleeding in this high-risk group of patients. I wonder if there was some selection bias. Overall, I think your data support kind of a bias I have. Since I am a vascular surgeon, we almost always operate on clopidogrel, and I do not know if our bleeding rate is worse or better, but it is something we almost always have to do to keep our grafts patent. So I think a lot of us would accept bleeding over a myocardial infarction. And in your patients, where you did have an event, 2 of them died. So I think that would be something we tend to have a bias toward. It was an excellent presentation. Dr David Strosberg: Thank you, Dr Dalsing, for your thoughtful, insightful questions. With regard to your first question, we designed these statistical hypotheses such that there would be no difference between the 2 groups. As all of our P values were above the standard 0.05, we failed to reject the hypotheses, and as such, we concluded there is no difference between the 2 groups. With regard to your second question, patient comorbidities were defined simply by review of the medical record. So in this sense, for example,

ARTICLE IN PRESS 8 Strosberg et al

patients who had stage 3 chronic kidney disease were counted equally as a patient who was on hemodialysis in terms of history of kidney disease. If we had seen a difference between the 2 groups, we should have investigated that to be more thorough and broken that down. But, because we did not, we did not feel that we needed to. Your third question was how could we tell when clopidogrel was held. That was one of the most challenging parts of the study. The majority of our patients were examined before our electronic medical records system had been instituted in 2011. As such, for patients cared for before 2011, we relied on the clinic notes. It was either documented in the plan whether clopidogrel should be held or not in the patient instructions. After our electronic medical record was instituted, we were able to look at the medical reconciliation from when patients were admitted prior to surgery. We could look to see when was the last time they had taken a particular medication. As for your fourth question regarding reoperation, we specifically did not examine rates of reoperation for bleeding. We used transfusion as a marker for blood loss, and we thought that patients who would need to go back for reoperation for bleeding would also have increased transfusion rates. That being said, that is something that we should absolutely consider in future studies. Dr Jason Smith: I noticed that you included both elective and emergency cases within the same cohort, and that sometimes leads to difficulty, because they are often different situations. Why did you decide to do that? Could the outcomes have been different had you changed that? The simple fact that the patients were on antiplatelet therapy often led them to get platelet transfusions. However, your own data show there is no difference in blood loss or any of the other factors that one would associate [with] needing a platelet transfusion. This finding raises the question: Am I giving the patient platelets because I want to feel better, or am I giving the patient platelets because I want them to get better? Dr David Strosberg: Regarding the first question, combining the elective and emergent patients all in the same cohort, we did that simply because we were looking at the complexity of the case as our primary measure to investigate. So we only consider that subgroup analysis of the total. But I think that if we had a larger group, and if we separated them out with clear, defined cohorts, we may have seen a difference between the 2. Your second question asked about transfusing platelets for patients who were on antiplatelet

Surgery j 2016

medications. As you know, that is the only way to reverse an ADP antagonist, as it works by competitive inhibition, and so you would need 1 or 2 pools of platelets. We did not find that affected any outcomes. However, that was not a primary outcome. If we wanted to consider that as the primary outcome, we would have to structure the study very differently, and that may be something that we could investigate further. Dr Peter Henke (Ann Arbor, MI): I would be a little bit cautious with this. If you have ever done a big aortic procedure with someone on full Plavix, you know that you can lose up to a couple liters of blood just from oozing. So that leads to a couple of questions. One, did you look at the duration of the operation? Because in the vascular literature, while Plavix is not associated with increased reoperation rates or hematomas, the length of the operation to get them dry was, on average, about 30 minutes longer. More importantly, you did not break down the procedures for which these patients were on Plavix. So for an endarectomy case, not a big deal. They might lose a little bit more blood. But again, an aortic procedure or a long-leg incision with a distal bypass, you can really lose a lot of blood that way. Did you factor type of surgery into your analysis? Dr David Strosberg: In response to your first question, we did not consider duration of operation. That is something that absolutely may be an independent risk factor. We could collect more variables that may be confounding and perform a multivariable analysis to see if any of those are independent risk factors for patients in this group. With regard to your second question, we identified procedures based on 72 CPT codes, which we figured were representative of major cases. So we had 2 physicians that went through procedures and, for example, decided that explantation of an infected thoracic aortic graft was a major procedure and that dialysis access was not. So we did not include procedures that were considered minor. So, of all of the 72 procedures, 2 physicians agreed that all of those were large procedures. That being said, our process was difficult and subjective, especially when we are combining all types of cases together. Dr Peter Rossi (Milwaukee, WI): My question is similar to what Dr Henke was just asking. Those of us that do a lot of open aortic surgery know that very few things bleed like the back wall of an aortic anastomosis of a patient on Plavix. I was wondering exactly how many of the procedures in your major

ARTICLE IN PRESS Surgery Volume j, Number j

vascular category were aortic operations, because that is sort of a separate animal from carotid endarterectomy or thrombectomy and things like that. My second question is how many of the patients that were on Plavix were appropriately on Plavix? At least what I find in my practice is that the vast majority of people that come in on Plavix do not need to be on it. The people who really need to be on Plavix are people that have drug-eluting coronary stents, and those are the ones that are really at high risk for perioperative cardiac events if you stop their Plavix. The ACC has been changing their recommendations on almost a yearly basis for the duration of time that people need to be on Plavix after drug-eluting coronary stents. It has gone from 3 months to 6 months to a year to 2 years, and some people are now saying lifetime. I was wondering if you had the ability to pull any of that data out of your study, because the data can be a little bit subjective if the indications for Plavix are not correct.

Strosberg et al 9

Dr David Strosberg: Thank you, Dr Rossi, for your questions. Those are absolutely considerations for types of cases. I would have to review the data specifically, but there were approximately, for example, under 20 major aortic cases and several hundred peripheral vascular bypasses. If we broke that down separately, say, intrathoracic or intra-abdominal surgery versus peripheral surgery, there may be a difference in outcomes. As for the appropriateness of Plavix, in addition to finding out if they were on Plavix and how long they were held for, was one of the most difficult parts of our study. So, many times, we had a scanned-in note from the electronic medical record system that said this patient was on Plavix due to a history of MI or history of stent. We do not know if this was a drug-eluting stent or bare-metal stent, how long ago it was placed, and what location it was placed in. So if we had more of that information, we could break that down and draw more conclusions.