On Jul 1, 2001 R H Fitzgerald (and others) published: Prevention of Venous ... including proximal deep-vein thrombosis, in patients undergoing total knee ...
COPYRIGHT © 2001
BY
THE JOURNAL
OF
BONE
AND JOINT
SURGERY, INCORPORATED
Prevention of Venous Thromboembolic Disease Following Primary Total Knee Arthroplasty A RANDOMIZED, MULTICENTER, OPEN-LABEL, PARALLEL-GROUP COMPARISON OF ENOXAPARIN AND WARFARIN BY R.H. FITZGERALD JR., MD, T.E. SPIRO, MD, A.A. TROWBRIDGE, MD, G.A. GARDINER JR., MD, T.L. WHITSETT, MD, M.B. O’CONNELL, PHARMD, J.A. OHAR, MD, AND T.R. YOUNG, DO, FOR THE ENOXAPARIN CLINICAL TRIAL GROUP
Background: Patients treated with total knee arthroplasty are at high risk for the development of venous thromboembolism postoperatively. This study compared the efficacy and safety of two common thromboprophylactic agents, enoxaparin (a low-molecular-weight heparin) and warfarin. Methods: Three hundred and forty-nine patients were included in a prospective, randomized, multicenter, open-label, parallel-group clinical trial. Treatment with enoxaparin (30 mg, administered subcutaneously twice daily) or warfarin (adjusted to an international normalized ratio of 2 to 3) was initiated during the immediate postoperative period, within eight hours after the surgery, and was continued for four to fourteen days. Venous thromboembolism was defined as deep-vein thrombosis documented by contrast venography, symptomatic deep-vein thrombosis documented by lower-extremity ultrasonography, or symptomatic pulmonary embolism confirmed by a positive lung scan or pulmonary angiography. Results: In the all-treated-patients group, eighty (45%) of the 176 warfarin-treated patients had venous thromboembolism: fifty-nine (34%) had distal deep-vein thrombosis; twenty (11%), proximal deep-vein thrombosis; and one (0.6%), pulmonary embolism. Venous thromboembolism developed in significantly fewer (p = 0.0001) enoxaparin-treated patients (forty-four of 173; 25%): forty-one (24%) had distal deep-vein thrombosis, three (2%) had proximal deep-vein thrombosis, and none had pulmonary embolism. The enoxaparin-treated patients also had a significantly lower prevalence of proximal deep-vein thrombosis (p = 0.002). The estimated odds for the development of venous thromboembolism were 2.52 times greater (95% confidence interval, 2.00 to 3.19) with warfarin than they were with enoxaparin. Major hemorrhage occurred in four warfarin-treated patients and nine enoxaparin-treated patients; with the numbers available, this difference was not significant (p = 0.17). Clinically important operative-site hemorrhage occurred in six (3%) of the warfarin-treated patients and twelve (7%) of the enoxaparin-treated patients (p = 0.15). Conclusions: A fixed 30-mg subcutaneous dose of enoxaparin, administered twice daily, with the first dose administered within eight hours after the completion of surgery, was significantly more effective than adjusted-dose warfarin in reducing the occurrence of asymptomatic venous thromboembolism, including proximal deep-vein thrombosis, in patients undergoing total knee arthroplasty. With the numbers available, there was no significant difference between groups with regard to the occurrence of major hemorrhagic complications; however, the rate of overall hemorrhagic complications was higher in the enoxaparin group.
P
atients treated with total knee or hip arthroplasty are at a high risk for the development of deep-vein thrombosis as well as pulmonary embolism1,2. Both peripheral venous stasis and direct endothelial injury, factors that play an important role in the etiology of venous thrombosis, occur during total hip and knee arthroplasty 3. The prevalence of deep-vein thrombosis following total knee replacement without prophylaxis has been reported to be 50% to 80% in series ranging in size from thirty to sixty-one patients1,4-7. Warfarin is an effective prophylactic agent that is com-
monly prescribed for patients undergoing total hip arthroplasty8-11. Two recently completed clinical trials compared the effectiveness of the low-molecular-weight heparins ardeparin (446 patients) and tinzaparin (535 patients) with that of warfarin in patients treated with total hip or knee arthroplasty12,13. The prevalence of all types of asymptomatic deep-vein thrombosis after total knee arthroplasty was significantly lower when low-molecular-weight heparin had been used (p = 0.004 for ardeparin, and p = 0.02 for tinzaparin). However, the difference in the prevalence of proximal
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 83-A · N U M B E R 6 · J U N E 2001
deep-vein thrombosis was not significant. Enoxaparin is a low-molecular-weight heparin used clinically for prophylaxis against venous thromboembolism associated with hip and knee replacement procedures1. Pharmacokinetic and pharmacodynamic studies have demonstrated little intra-individual or inter-individual variability in the effects of enoxaparin and, in contrast to warfarin, enoxaparin does not require biological monitoring and dosage adjustments14. We report the results of a prospective, randomized, multicenter, open-label, inpatient, parallel-group study performed to compare the safety and efficacy of fixed-dose enoxaparin with that of adjusted-dose warfarin in the prevention of venous thromboembolism in patients treated with total knee arthroplasty. Materials and Methods Study Populations Men and women, thirty-eight years of age or older, undergoing a primary unilateral total knee arthroplasty were eligible for inclusion in the study. Approval by the institutional review board and informed consent by the patients were obtained in each of the twenty-two medical centers involved in the study. Premenopausal women were included only if there was documentation that they were surgically sterile or not pregnant. The all-treated-patients study population consisted of all randomized patients who were scheduled for a primary total knee arthroplasty and had received at least one dose of study medication. Each center was provided with sealed medication kits containing either syringes filled with enoxaparin or warfarin tablets. Randomization numbers generated by the study sponsor were affixed to the exterior of each kit; randomization was performed by the investigator allocating the kits in ascending order. The evaluable-patients population consisted of patients who had undergone primary unilateral total knee arthroplasty; had received the study medication for at least four days; and had an evaluable venogram of the surgically treated lower extremity and adequate contrast venography or ultrasonography of the contralateral extremity, symptoms of deep-vein thrombosis confirmed by ultrasonography or venography, and/or symptoms of pulmonary embolism confirmed by a lung scan or pulmonary angiography. Exclusion criteria included wound hemorrhage continuing for longer than eight hours after wound closure; generalized hemorrhagic disorders or hypercoagulable syndrome, including clinical evidence of chronic or acute deep-vein thrombosis or a documented history of venous thromboembolism; allergy to unfractionated heparin, warfarin, fish or swine products, iodine, or contrast medium; a history of heparin-associated thrombocytopenia or heparin or warfarin-associated skin rash or necrosis; asthma not under medical control; surgery (other than arthroscopy) on the ipsilateral knee within the previous six months or on the ipsilateral hip, contralateral hip, or contralateral knee within the preceding three months; any clinically important disease or requirement for treatment during the study period that could
P R E VE N T I O N O F VE N O U S T H RO M B O E M B O L I C D I S E A S E F O L L OW I N G P R I M A R Y TO T A L K N E E A R T H RO P L A S T Y
interfere with the action, kinetics, or evaluation of the study medications; hepatic disease with a bilirubin level of ≥2 mg/dL (≥34 µmol/L); renal disease with a creatinine level of ≥2 mg/dL (≥177 µmol/L); evidence of current abuse of drugs (excluding tobacco products) or alcohol; surgery involving the eye, spinal cord, or central nervous system within three months before study entry; active ulcerative disease or angiodysplasia of the gastrointestinal tract or active gastrointestinal hemorrhage within the previous six months; hypertension not under medical control (defined as systolic blood pressure of ≥180 mm Hg or diastolic blood pressure of ≥105 mm Hg); stroke or myocardial infarction within the previous three months; and treatment with aspirin, aspirin-containing products, or nonsteroidal anti-inflammatory drugs on a regular basis for the four days immediately preceding hospitalization or regular treatment with these products during hospitalization (products with a short half-life, such as ibuprofen and naproxen, could be continued until two days preceding hospitalization). Sequential compression devices were not permitted, but graduated compression stockings were. Use of a continuous-passive-motion device was permitted for a total of six hours per day. Study Evaluations Prestudy evaluations consisted of a medical history including documentation of concomitant medications; a complete physical examination; a twelve-lead electrocardiogram (optional); a chest roentgenogram; bilateral lower-extremity duplex or B-mode ultrasonography (optional); urinalysis; and a complete blood-cell count with differential and platelet count (hematological studies), measurement of activated partial thromboplastin time and prothrombin time (coagulation studies), and determination of levels of total bilirubin, creatinine, aspartate aminotransferase, and alanine aminotransferase (biochemistry studies). On the first study day (that is, the day of the surgery), before the first dose of study medication was administered, surgical, transfusion, and anesthesia data were recorded for each patient, and each patient was assessed for hemorrhage and with hematological studies (excluding white blood-cell differential), coagulation studies (activated partial thromboplastin time for all patients and prothrombin time for warfarin-treated patients only), and a biochemistry profile. Treatment with warfarin or enoxaparin began on the day of the surgery, as soon as hemostasis was achieved and within eight hours after surgical wound closure, and it was continued for a minimum of four days and a maximum of fourteen days. Warfarin therapy was initiated orally, with a dose of 7.5 mg, followed by subsequent daily adjustment of the dose as necessary to maintain the international normalized ratio between 2 and 3. A 30-mg dose of enoxaparin was administered every twelve hours by deep subcutaneous injection in the abdominal area. During the treatment period, evaluations included daily assessment of vital signs, abbreviated physical examination, and assessments for any adverse experiences, concomitant medications, hemorrhage, and blood transfusion. Patients with clin-
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 83-A · N U M B E R 6 · J U N E 2001
P R E VE N T I O N O F VE N O U S T H RO M B O E M B O L I C D I S E A S E F O L L OW I N G P R I M A R Y TO T A L K N E E A R T H RO P L A S T Y
TABLE I Summary of Patient Randomization and Accountability
Number of patients randomized and treated Number of patients treated who did not complete study
Warfarin*
Enoxaparin*
176 (100)
173 (100)
14 (8.0)
15 (8.7)
Adverse clinical event
8
11
Adverse laboratory event
2
1
Other†
4
3
Number of patients evaluable
122 (69.3)
108 (62.4)
54 (30.7)
65 (37.6)
Inadequate vascular assessment
46 (26.1)
57 (32.9)
Venography not done Venography inadequate
20 20
16 40
Venography (other) Ultrasonography not done
3 3
1 0
Study medication error Other‡
2 6
0 8
Number of patients not evaluable
*The values are given as the number of patients, with the percentage in parentheses. †In the warfarin group, the reasons why patients did not complete the study included deviation from the protocol (two patients), withdrawal of consent (one), and cessation of treatment per physician request (one). In the enoxaparin group, the reasons included withdrawal of consent (one patient), administration of only five doses as a result of pharmacy error (one), and lack of effectiveness of the test drug (one). ‡In the warfarin group, six patients were not evaluable because the study medication was given for less than four days. In the enoxaparin group, seven patients were not evaluable because the study medication was given for less than four days, and one was not evaluable because of a previous total arthroplasty on the involved knee.
ical evidence of deep-vein thrombosis were evaluated with bilateral lower-extremity ultrasonography. Patients with a positive ultrasonogram had contrast venography performed. Patients with clinical evidence of pulmonary embolism had a ventilation-perfusion lung scan and/or pulmonary angiography, at the discretion of the investigator. The study medication was discontinued, and all end-of-treatment and follow-up procedures were performed in all patients who underwent venography, irrespective of the outcome, and in all patients with a confirmed diagnosis of pulmonary embolism. Patients who had a negative ultrasonogram continued treatment with the study medication. Patients with pulmonary embolism did not undergo venography. During the treatment period, hematological and biochemistry evaluations were carried out on the second, third, and fourth postoperative days and on alternate days thereafter. For warfarin-treated patients, an international normalized ratio was determined prior to administration of the study drug; on the second, third, and fourth postoperative days; and on alternate days thereafter. Within twenty-four hours following either the last dose or the discontinuation of the study medication, the evaluations included assessment of vital signs; abbreviated physical examination; assessments for any adverse experiences, concomitant medications, hemorrhage (at the operative site and elsewhere), and blood transfusion; bilateral lower-extremity ultrasonography; unilateral venography of the treated extremity in all patients, regardless of the ultrasonographic results; bilateral venography if the untreated extremity had clinical signs and symptoms indicating deep-vein thrombosis or a positive ultrasonogram; and determination of hematological and bio-
chemistry values, urinalysis, measurement of activated partial thromboplastin time, and, for warfarin-treated patients only, determination of the international normalized ratio. At the follow-up visit, approximately three weeks following the final dose of study medication, the evaluations included assessment of vital signs; abbreviated physical examination; assessments for any adverse experiences, concomitant medications, hemorrhage (at the operative site and elsewhere), and blood transfusion; as well as determination of hematological and biochemistry values and urinalysis. Efficacy Assessments The primary efficacy variable was the occurrence of deep-vein thrombosis or pulmonary embolism during the postoperative period. The diagnosis of deep-vein thrombosis in asymptomatic patients was based on the findings of lower-extremity contrast venography. The diagnostic criteria for deep-vein thrombosis included the demonstration of an intraluminal filling defect within a deep-vein segment or an occluded deep-vein segment with a filling defect at one or both ends. If the venographic findings were uncertain, ultrasonography of the lower extremity was used to confirm or rule out the presence of a thrombus. A positive ultrasonogram was sufficient for the diagnosis of symptomatic deep-vein thrombosis. The diagnosis of symptomatic pulmonary embolism required a high-probability ventilation-perfusion lung scan or a positive pulmonary angiogram. In addition to the assessment by the investigator, a blinded, independent review of all venograms and ultrasonograms was carried out by a panel of vascular imaging specialists. The consensus method was used to determine the final diagnosis that was used in the data analysis.
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 83-A · N U M B E R 6 · J U N E 2001
Safety Assessment The primary safety variable was overt hemorrhage, which was categorized as major or minor. An episode of hemorrhage was defined as major if it fulfilled at least one of the following criteria: resulted in transfusion of at least two units of packed red blood cells; resulted in a decrease in the hemoglobin concentration of ≥20 g/L compared with the postoperative hemoglobin concentration before the administration of any study medication; was retroperitoneal, intracranial, or intraocular; or resulted in a serious life-threatening clinical event or death. Statistical Methods For the primary efficacy analysis, the occurrence of venous thromboembolism was assessed with use of data from the all-treated-patients group. A multivariate logistic regression model with explanatory variables for investigator and treatment (without interaction terms) was used to compare the prevalences of venous thromboembolism between the warfarin and enoxaparin groups. The test was two-tailed, at the 5% level. The Fisher exact test was used for the primary safety analysis, with treatment groups compared with regard to the rates of hemorrhagic episodes, adverse events, and selected hematological and biochemistry values of clinical concern. Changes in laboratory parameters and vital signs compared with baseline values were summarized by descriptive statistics; no formal statistical tests of hypotheses were performed. Results Patient Disposition and Demographics A total of 349 patients, thirty-eight to eighty-nine years of age, received at least one dose of study medication: 176 received warfarin and 173 received enoxaparin (Table I). These patients were included in the all-treated-patients analysis. Twenty-nine patients (fourteen in the warfarin group and fifteen in the enoxaparin group) did not complete the study. Premature withdrawal was most commonly due to an adverse clinical event, with eight warfarin-treated and eleven enoxaparin-treated pa-
P R E VE N T I O N O F VE N O U S T H RO M B O E M B O L I C D I S E A S E F O L L OW I N G P R I M A R Y TO T A L K N E E A R T H RO P L A S T Y
tients withdrawing for this reason. None of these adverse events were related to the study medications. Two warfarin-treated patients were withdrawn because of abnormal laboratory findings (decreased hemoglobin concentration or hematocrit and blood in the urine), and one enoxaparin-treated patient was withdrawn because of an abnormal laboratory finding (decreased hemoglobin concentration). Other reasons for withdrawal primarily included protocol deviations and the patient’s withdrawal of consent because of abnormal laboratory findings. Seven patients (four in the warfarin group and three in the enoxaparin group) were treated with heparin after deep-vein thrombosis developed. The all-treated-patients group included 196 women (56%) and 153 men (44%). There was no difference between the warfarin and enoxaparin groups with regard to age, gender, race, height, or weight (see Appendix). The demographic characteristics of the evaluable patients were similar to those of the all-treated-patients group. A minimum of three days of warfarin therapy was required before the international normalized ratio reached a desirable therapeutic range (2 to 3) in an appreciable number of patients (see Appendix). There was a high degree of variability in the international normalized ratios throughout the study period, with a nearly symmetrical distribution of values falling above and below the therapeutic range of 2 to 3. The mean duration from the time that an international normalized ratio of ≥2 was achieved until hospital discharge was 3.3 ± 2.0 days. Two hundred and thirty patients—122 in the warfarin group and 108 in the enoxaparin group—were included in the evaluable-patients analysis. The most common reason for a patient to be classified as not evaluable was that end-of-study vascular examinations were not done or were inadequate. The venogram was inadequate for twenty warfarin-treated patients and forty enoxaparin-treated patients, and venography was not performed for twenty warfarin-treated patients and sixteen enoxaparin-treated patients. Ultrasonography was not performed for three warfarin-treated patients.
TABLE II Occurrence of Venous Thromboembolic Disease by Treatment Group in All-Treated-Patients and Evaluable-Patients Populations All Treated Patients*
Evaluable Patients*
Outcome
Warfarin (N = 176)
Enoxaparin (N = 173)
Warfarin (N = 122)
Enoxaparin (N = 108)
Venous thromboembolic disease
80 (45.5)
44 (25.4)†
72 (59.0)
41 (38.0)‡
Deep-vein thrombosis Distal
59 (33.5)
41 (23.7)
56 (45.9)
38 (35.2)
4 (2.3)
0 (0.0)
1 (0.8)
0 (0.0)
Proximal and distal
16 (9.1)
3 (1.7)
15 (12.3)
3 (2.8)
Total proximal
20 (11.4)
3 (1.7)§
16 (13.1)
3 (2.8)#
1 (0.6)
0 (0.0)
0 (0.0)
0 (0.0)
Proximal only
Pulmonary embolism
*The values are given as the number of patients, with the percentage in parentheses. †P = 0.0001 for warfarin compared with enoxaparin. ‡P = 0.0043 for warfarin compared with enoxaparin. §P = 0.002 for warfarin compared with enoxaparin. #P = 0.01 for warfarin compared with enoxaparin.
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 83-A · N U M B E R 6 · J U N E 2001
P R E VE N T I O N O F VE N O U S T H RO M B O E M B O L I C D I S E A S E F O L L OW I N G P R I M A R Y TO T A L K N E E A R T H RO P L A S T Y
TABLE III Occurrence of Hemorrhage by Treatment Group in All-Treated-Patients Population Outcome Any clinically important hemorrhage Major hemorrhage Clinically important operative-site hemorrhage Minor hemorrhage
Warfarin* (N = 176) 41 (23.3)
Enoxaparin* (N = 173) 58 (33.5)†
4 (2.3)
9 (5.2)‡
6 (3.4)
12 (6.9)§
37 (21.0)
49 (28.3)
*The values are given as the number of patients, with the percentage in parentheses. †P = 0.04. ‡P = 0.17. §P > 0.05.
Primary Diagnosis and Surgical Data Osteoarthritis was the primary diagnosis for the majority of patients (see Appendix). Most patients underwent a tricompartmental arthroplasty with cement under general anesthesia. The mean duration of surgery, the mean duration of tourniquet use, the performance of lateral retinacular release, and the use of intramedullary alignment guides were similar in the warfarintreated and enoxaparin-treated patients. The intraoperative blood loss averaged 211 mL overall (231 mL in the warfarin group and 191 mL in the enoxaparin group). An average of 594 mL of salvaged blood was reinfused in thirty-five warfarintreated patients, and an average of 617 mL was reinfused in thirty-five enoxaparin-treated patients. Efficacy In both the all-treated-patients and the evaluable-patients efficacy analysis, the prevalence of venous thromboembolism associated with the use of enoxaparin was significantly lower than that associated with the use of warfarin (Table II). In the alltreated-patients population, eighty (45%) of the 176 warfarintreated patients had venous thromboembolism: fifty-nine (34%) had only distal deep-vein thrombosis, twenty (11%) had only proximal or proximal and distal deep-vein thrombosis, and one (0.6%) had pulmonary embolism. In comparison, a significantly smaller (p = 0.0001) fraction of the enoxaparintreated patients (forty-four of 173; 25%) had venous thromboembolism. None of the 173 enoxaparin-treated patients had pulmonary embolism, whereas forty-one (24%) had distal deep-vein thrombosis and three (2%) had proximal and distal deep-vein thrombosis. The estimated odds for the development of venous thromboembolism in the warfarin group was 2.52 times greater (95% confidence interval, 2.00 to 3.19) than those for the enoxaparin group. Similar results were observed in the evaluable-patients population. Venous thromboembolic disease occurred in seventy-two (59%) of the 122 warfarintreated patients and forty-one (38%) of the 108 enoxaparintreated patients (p = 0.0043; odds ratio, 2.26; 95% confidence interval, 1.71 to 2.98). In the all-treated-patients population, twenty of the seventy-nine deep-vein thrombi documented in the warfarin group were in the proximal venous system (popliteal, femoral, or iliac veins), whereas only three of the forty-four deep-vein thrombi in the enoxaparin group were in the proximal venous system. Similarly, in the evaluable-patients population, proximal deep-vein thrombosis developed in sixteen of the
warfarin-treated patients and three of the enoxaparin-treated patients. The prevalence of proximal deep-vein thrombosis was significantly reduced in the enoxaparin-treated patients in both the all-treated (p = 0.002) and the evaluable (p = 0.01) patient populations (odds ratios of 7.3 and 5.3, respectively). The distribution of deep-vein thrombosis by treatment group was examined according to the primary diagnosis, type of surgery, type of anesthesia during surgery, use of surgical cement, use of compression stockings, use of alignment guides, race, gender, age, body-mass index (body weight in kilograms divided by height in meters squared), and history of tobacco use or of a specific medical condition (obesity, diabetes mellitus, congestive heart failure, varicosities, and cancer). The occurrence of deep-vein thrombosis within each subgroup with at least fifteen patients in each treatment group was similar to the overall occurrence in each treatment group. Male patients and patients with varicose veins had a higher prevalence of thromboembolic disease. Among all patients diagnosed with venous thromboembolism, forty-seven (59%) treated with warfarin and twentyone (48%) treated with enoxaparin received treatment and/or had extended hospitalization because of the thromboembolic disease. Twenty-six warfarin-treated patients and twenty enoxaparin-treated patients had their hospital stay prolonged as a result of the thromboembolism. Safety Within the all-treated-patients population, major hemorrhagic episodes occurred in four (2%) of the 176 warfarin-treated patients and in nine (5%) of the 173 enoxaparin-treated patients; this difference was not significant (p = 0.17), with the numbers available (Table III). The prevalence of major and minor hemorrhagic episodes in the warfarin-treated patients (forty-one of 176; 23%) was lower than that in the enoxaparin-treated patients (fifty-eight of 173; 34%) (p = 0.04). A clinically important hemorrhagic episode occurred at the operative site in six (3%) of the warfarin-treated patients and twelve (7%) of the enoxaparin-treated patients (p > 0.05). However, no patient required a reoperation, and only one (enoxaparin-treated) patient had to stop using the study medication, on the fourth day of the study. A superficial infection at the incision site developed in this same patient, forty days after discontinuation of use of the study drug. The major hemorrhagic episodes in the warfarin group included ecchymoses of >5 cm in three patients (2%) and ret-
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG VO L U M E 83-A · N U M B E R 6 · J U N E 2001
roperitoneal hemorrhage (resulting in death) in one patient (0.6%). In the enoxaparin group, the major hemorrhagic episodes included clinically important wound drainage in three patients (2%), wound hematoma in three patients (2%), ecchymoses of >5 cm in two patients (1%), and hematuria and tracheotomy-site hemorrhage in one patient (0.6%). Minor hemorrhagic events in the warfarin-treated group included ecchymoses not at the operative site in twenty patients (11%), hematuria in ten patients (6%), gastrointestinal bleeding in three patients (2%), and other hemorrhagic episodes in twelve patients (7%). In the enoxaparin-treated group, ecchymosis not at the operative site was seen in twenty-five patients (14%); hematuria, in thirteen patients (8%); gastrointestinal hemorrhage, in five patients (3%); and other hemorrhagic episodes, in eleven patients (6%). Three patients in the warfarin group died; one death was due to multiple organ failure secondary to retroperitoneal hemorrhage, one was due to postoperative respiratory failure, and one was due to unknown causes. One patient in the enoxaparin group died of postoperative respiratory and cardiac complications. Thrombocytopenia, defined as a platelet count of