Deep venous thrombosis and pulmonary embolism are major risks after a total hip arthroplasty7,9,21. Without prophylaxis, the prevalence of deep venous thrombosis is 51 percent (332 of 655 patients)9 and that of fatal pulmonary embolism is 2 percent (forty-one of 2091 patients)15. The percentage of patients in whom proximal or total (proximal and distal) deep venous thrombosis develops after a hip arthroplasty is highest during the fourth, fifth, and sixth days postoperatively, with the relative risk decreasing throughout the two-week period after the operation23,26.
Enoxaparin is a low-molecular-weight heparin approved for use in the prevention of deep venous thrombosis, which may lead to pulmonary embolism, after a hip or knee arthroplasty7,11. A fixed-dose regimen initiated twelve to twenty-four hours after the operation decreases the risk of drug-induced intraoperative bleeding and eliminates the need for routine laboratory monitoring of coagulation times6,11. In a pooled analysis of North American trials12, the prevalence of total deep venous thrombosis was 12 percent (ninety-three of 785 patients) after treatment with enoxaparin (thirty milligrams every twelve hours) compared with 16 percent (eighty-seven of 539 patients) after treatment with unfractionated heparin (15,000 international units each day) and 46 percent (twenty-three of fifty patients) after treatment with a placebo. The prevalence of pulmonary embolism was 0.1 percent (one) in the group managed with enoxaparin, 0.9 percent (five) in the group managed with heparin, and 2 percent (one) in the group that received a placebo. Major bleeding occurred in 4 percent (thirty-one) of the patients managed with enoxaparin, in 6 percent (thirty-two) of the patients managed with heparin, and in 4 percent (two) of the patients who received a placebo.
Therapy with adjusted-dose warfarin has been recommended as an appropriate method of prophylaxis against venous thromboembolic disease after total hip arthroplasty1,2,16,22, with the dose adjusted according to the international normalized ratio8,9,13,18. Bleeding complications associated with treatment with warfarin are related to the prolongation of the prothrombin time, which increases the international normalized ratio.
In the present study, we compared enoxaparin and adjusted-dose warfarin with regard to their safety and their efficacy in the prevention of clinically symptomatic venous thromboembolic disease during hospitalization for elective total hip arthroplasty. We also determined the rates of morbidity and mortality resulting from venous thromboembolic disease after the discontinuation of therapy and within the first three months after discharge from the hospital.
*Although none of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article, benefits have been or will be received but are directed to a research fund, foundation, educational institution, or other nonprofit organization with which one or more of the authors is associated. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was Rhône-Poulenc Rorer Pharmaceuticals.
†Scripps Clinic, 10666 North Torrey Pines Road, La Jolla, California 92037.
‡Sacred Heart General Hospital, 1200 Hilyard Street, 600, Eugene, Oregon 97401.
§Altru Main Clinic, 1000 South Columbia Road, Box 6003, Grand Forks, North Dakota 58206.
#1285 Orange Avenue, Winter Park, Florida 32789.
**University Orthopaedics, 3121 Maryland Parkway, 600, Las Vegas, Nevada 89109.
††202 Loch Alsh Drive, Chalfont, Pennsylvania 18914.
The study was a randomized, open-label, parallel-group clinical trial conducted in 156 centers and divided into two phases: a period of no more than fourteen days during hospitalization after total hip arthroplasty and a follow-up period of three months after discharge from the hospital. Patients had to be at least eighteen years old, had to be scheduled for elective unilateral primary hip arthroplasty, and had to have no history (that is, of blood dyscrasia, heparin-induced thrombocytopenia, or allergy to warfarin or heparin) that would preclude anticoagulant therapy. Prescreening for deep venous thrombosis with duplex ultrasonography or venography was not permitted, and the patients were not allowed to use aspirin for seven days or anti-inflammatory drugs for twenty-four hours before the operation. All patients signed an informed-consent form, approved by the Institutional Review Board at each site, before inclusion in the study. The first patient was managed in October 1993, and the study was completed in April 1996.
Prophylactic treatment with subcutaneous injection of thirty milligrams of enoxaparin (Lovenox) every twelve hours was initiated within twenty-four hours postoperatively, after hemostasis (defined as the cessation of active bleeding as determined by the investigator) had been established. Prophylactic treatment with adjusted-dose warfarin (Coumadin) could be initiated as early as forty-eight hours preoperatively, at the discretion of the investigator, but it had to be started within twenty-four hours after the operation. Warfarin was started at a dose of 7.5 milligrams, and the dosage was adjusted to maintain an international normalized ratio between 2.0 and 3.0. Each treatment was administered until discharge from the hospital.
Preoperative screening consisted of a physical examination, clinical evaluation for deep venous thrombosis or pulmonary embolism, and clinical laboratory testing that included measurement of the prothrombin time, partial thromboplastin time, hematocrit, hemoglobin level, and platelet count. A radiograph of the chest and a twelve-lead electrocardiogram were made if clinically indicated.
During the period of postoperative hospitalization, defined as the day that the first dose of medication was given until the day of discharge from the hospital, the patients were examined daily for clinical signs and symptoms of deep venous thrombosis and pulmonary embolism and for clinically relevant bleeding at operative and nonoperative sites, as is consistent with standard medical practice. Blood-transfusion requirements; use of concomitant medications; vital signs; adverse events; measurements of hematocrit, hemoglobin level, and platelet count; and routine measurement of prothrombin time (for patients managed with warfarin) were documented daily. Patients were dropped from the study if symptomatic deep venous thrombosis was confirmed by duplex ultrasonography or venography or if pulmonary embolism was confirmed by a ventilation-perfusion scan or pulmonary angiography, or both.
Although the occurrence of heparin-induced thrombocytopenia in patients managed with low-molecular-weight heparin is extremely rare compared with that in patients managed with unfractionated heparin, and the prevalence has not been clearly established5,10,25,30, patients were dropped from the study if the platelet count, which was obtained each day for the first three days after the operation, decreased to less than 100 x 109 per liter.
During the period after hospitalization, defined as the day after discharge from the hospital until the follow-up visit three months later, patients were not supposed to receive any antithrombotic or anticoagulant medication. Nonsteroidal anti-inflammatory drugs were acceptable; however, aspirin was not permitted throughout the period after discharge. All patients returned for a follow-up evaluation at six weeks and at three months after discharge. At those times, each patient was examined for clinical signs and symptoms of deep venous thrombosis (pain, inflammation, swelling, and redness of the lower extremity) and pulmonary embolism (chest pain and difficulty breathing). In addition, evidence of bleeding at operative and nonoperative sites, vital signs, adverse events, the use of concomitant medications, and blood-transfusion requirements since the index operation were evaluated. During hospitalization and the three-month period after discharge, use of any pneumatic compression device was not permitted; however, use of elastic compression stockings was acceptable.
The primary parameter to determine the efficacy of the study medications was the prevalence of symptomatic venous thromboembolic disease, including symptomatic deep venous thrombosis of the lower extremity, confirmed by ultrasonography or venography, and symptomatic or fatal pulmonary embolism, confirmed by a ventilation-perfusion scan or pulmonary angiography, or both, during hospitalization and during the three-month period after discharge.
The primary parameter to determine safety was the prevalence of major bleeding. A major bleeding episode was defined as overt bleeding associated with at least one of the following events: death or a life-threatening clinical event; bleeding confirmed to be retroperitoneal, intracranial, or intraocular; postoperative transfusion of more than two units of packed red blood cells or whole blood; or a decrease in the hemoglobin level of more than twenty grams per liter compared with the relevant postoperative level. A minor bleeding episode was defined as an overt bleeding episode that did not meet the criteria for classification as a major bleeding episode. Adverse events and serious adverse events were secondary parameters in the assessment of safety. A serious adverse event was an episode that was fatal or life-threatening, was permanently disabling, necessitated hospitalization (initial or prolonged), resulted in an overdose, or necessitated intervention to prevent permanent impairment or damage.
Statistical Methods
Conclusions about primary efficacy and safety were based on an intent-to-treat patient analysis14. All patients received at least one dose of a study medication. An interim analysis was conducted after the first 1000 patients had completed the study protocol. Age, weight, height, and blood pressure were summarized with use of descriptive statistics. These continuous demographic and baseline characteristics of the treatment groups were compared between groups with two-way analysis of variance (LogXact-Turbo; Cytel Software, Cambridge, Massachusetts) with treatment and the medical-center effects. For the purpose of analysis, the medical centers were grouped into four categories according to the number of patients in the study who had been managed there. The four categories were: (1) one to twelve patients, (2) thirteen to twenty-four patients, (3) twenty-five to thirty-six patients, and (4) more than thirty-six patients. The categorical characteristics (gender, race, and so on) of the two groups were compared with use of the Cochran-Mantel-Haenszel general association test statistic, controlling for the medical center.
The prevalence of venous thromboembolic disease was determined for the entire study period, for the period of hospitalization, and for the three-month period after discharge. The occurrence of symptomatic venous thromboembolic disease in the two treatment groups was summarized according to age (less than forty, forty to sixty-five, and more than sixty-five years old) and body-mass index (thirty kilograms per square meter or less and more than thirty kilograms per square meter). The cumulative prevalence of symptomatic venous thromboembolic disease at six weeks and at three months after discharge was calculated for the treatment groups. Between-group comparisons were made with use of a logistic model with treatment effects20.
The prevalences of major and minor bleeding episodes in the two treatment groups were compared with use of the Cochran-Mantel-Haenszel general association test stratified by medical center. Descriptive statistics were presented for the number of transfusions occurring in the two treatment groups. Adverse events and serious adverse events were described by body system, according to gender, age, relationship to the study medication, and intensity. Descriptive statistics and the number of patients outside the normal limits were shown for laboratory values.
A total of 3011 patients were randomized to receive a study medication. Enoxaparin was administered to 1516 patients, and warfarin was given to 1495 patients. A total of 2229 patients completed the study protocol, and 782 patients discontinued prematurely. A comparable number of patients in each study group did not complete the study; most of them were dropped because of a deviation from the study protocol involving the use of prohibited medications, the use of compression devices, or the inappropriate receipt of the study medication (Table I). As already stated, the results and conclusions are based on the intent-to-treat analysis14, including all patients who received at least one dose of a study medication.
No cases of heparin-induced thrombocytopenia were reported, but patients in whom the platelet count fell to less than 100 x 109 per liter were not evaluated for heparin-dependent immunoglobulin-G antibodies. Fifty-two patients (twenty-seven who received enoxaparin and twenty-five who received warfarin) had a platelet count of less than 100 x 109 per liter, and nine of those who received enoxaparin and one of those who received warfarin were dropped from the study.
The two groups were comparable with respect to all demographic characteristics at baseline (Table II) except for body-mass index, which was found to be significantly different (p = 0.0055). The group managed with warfarin had more patients (72.5 percent [998 of the 1376 for whom the body-mass index was reported]) with a body-mass index of thirty kilograms per square meter or less than did the group managed with enoxaparin (67.7 percent [961 of the 1420 for whom the body-mass index was reported]). There were fewer patients who had a body-mass index of more than thirty kilograms per square meter in the group managed with warfarin (27.5 percent [378 of 1376]) than in the group managed with enoxaparin (32.3 percent [459 of 1420]).
Most patients were given the study medication for one to seven days. The mean duration of treatment, with either enoxaparin or adjusted-dose warfarin, was 7.3 days. The administration of anticoagulants was supposed to be stopped at the time of discharge from the hospital. However, ninety-three (6.1 percent) of the 1516 patients managed with enoxaparin and seventy-three (4.9 percent) of the 1495 patients managed with warfarin received these medications for one to twenty-one days after discharge. Despite this continued use by a small number of patients, the mean duration of use was 7.5 days for enoxaparin and 7.0 days for warfarin. The mean duration of the operation was similar in both treatment groups.
The administration of warfarin could be initiated within forty-eight hours before the operation. Of the 1495 patients who received warfarin, 331 (22.1 percent) received the medication preoperatively and 1164 (77.9 percent) received it at the time of the operation or postoperatively. The mean international normalized ratio on the seventh postoperative day was 1.61 for the patients who had first received warfarin preoperatively and 1.80 for the patients who had first received it at the operation or postoperatively. By the seventh postoperative day, sixty (18.1 percent) of the 331 patients who had first received warfarin preoperatively and 460 (39.5 percent) of the 1164 patients who had first received it at the operation or postoperatively had an international normalized ratio of at least 2.0.
Results of the Efficacy Analysis
During the course of the study, 111 patients (3.7 percent) had venous thromboembolic disease (Table III). These included fifty-five (3.6 percent) of the 1516 patients who received enoxaparin and fifty-six (3.7 percent) of those who received warfarin. Most patients in each treatment group had deep venous thrombosis only.
Overall, twenty-seven patients (0.9 percent) had a pulmonary embolism: fifteen (1.0 percent) of those managed with enoxaparin and twelve (0.8 percent) of those managed with warfarin. Fifteen patients (0.5 percent) had a pulmonary embolism only: six (0.4 percent) of those who received enoxaparin and nine (0.6 percent) of those who received warfarin. Twelve patients (0.4 percent) had both deep venous thrombosis and a pulmonary embolism: nine (0.6 percent) of those managed with enoxaparin and three (0.2 percent) of those managed with warfarin. Most (seventy-one; 64.0 percent) of the 111 patients who had venous thromboembolic disease were more than sixty-five years old and most (sixty-three; 56.8 percent) had a body-mass index of thirty kilograms per square meter or less.
During hospitalization, four (0.3 percent) of the patients who received enoxaparin had venous thromboembolic disease compared with seventeen (1.1 percent) of those who received warfarin (p = 0.0083) (Table IV). Within the first week after discharge from the hospital, ten (0.7 percent) of the patients managed with enoxaparin and fifteen (1.0 percent) of those managed with warfarin had venous thromboembolic disease. Between the first and second week after discharge, seventeen (1.1 percent) of the patients who had received enoxaparin and six (0.4 percent) of those who had received warfarin had venous thromboembolic disease.
Twenty-one patients (0.7 percent) had venous thromboembolic disease during hospitalization (Table V). Seventeen (0.6 percent) had deep venous thrombosis only: two (0.1 percent) of those managed with enoxaparin and fifteen (1.0 percent) of those managed with warfarin. Three patients (0.1 percent) had a pulmonary embolism only: one (less than 0.1 percent) of those managed with enoxaparin and two (0.1 percent) of those managed with warfarin. One patient (less than 0.1 percent) had both deep venous thrombosis and a pulmonary embolism.
During the period after discharge, ninety patients (3.0 percent) had venous thromboembolic disease: fifty-one (3.4 percent) of those managed with enoxaparin and thirty-nine (2.6 percent) of those managed with warfarin (Table V). Sixty-seven patients (2.2 percent) had deep venous thrombosis only: thirty-eight (2.5 percent) of those managed with enoxaparin and twenty-nine (1.9 percent) of those managed with warfarin. Twelve patients (0.4 percent) had a pulmonary embolism only: five (0.3 percent) of those who had received enoxaparin and seven (0.5 percent) of those who had received warfarin. Eleven patients (0.4 percent) had both deep venous thrombosis and a pulmonary embolism: eight (0.5 percent) of those who had been managed with enoxaparin and three (0.2 percent) of those who had been managed with warfarin.
Nineteen patients (0.6 percent) died during the study period: nine (0.6 percent) of the patients who had received enoxaparin and ten (0.7 percent) of those who had received warfarin. Four deaths (two patients in each treatment group) were classified as resulting from venous thromboembolic disease; however, no autopsy was performed to confirm the cause of death of three of these patients. The cause of death of one of these patients, who had received enoxaparin, was listed as massive pulmonary embolism due to deep venous thrombosis, but the study investigator reported that the patient had had no apparent signs or symptoms of deep venous thrombosis and no diagnostic tests had been performed. One of the two patients in the warfarin group who died had a deep venous thrombosis of the entire left femoral and popliteal vein, which was confirmed by ultrasound, but the death was listed as due to undetermined natural causes. The other patient who had received warfarin died because of a presumed diagnosis of pulmonary embolism; however, no autopsy was performed. A pulmonary embolism was confirmed at autopsy as the cause of death of the fourth patient, who had been managed with enoxaparin. Fifteen deaths (seven patients who had received enoxaparin and eight who had received warfarin) were due to other events, such as myocardial infarction, chronic obstructive pulmonary disease, and pneumonia.
Results of the Safety Analysis
Major or minor bleeding occurred in 262 patients (8.7 percent): 152 (10.0 percent) of the 1516 patients managed with enoxaparin and 110 (7.4 percent) of the 1495 managed with warfarin (Table VI). Eighteen (1.2 percent) of the patients who received enoxaparin and eight (0.5 percent) of those who received warfarin had major bleeding (p = 0.055). One hundred and forty-three (9.4 percent) of the patients managed with enoxaparin and 106 (7.1 percent) of those managed with warfarin reported minor bleeding (p = 0.021).
A major bleeding episode occurred in twenty-six patients (0.9 percent). These included eighteen (0.1 percent) of the 1516 patients who were managed with enoxaparin (nine [0.6 percent] had major bleeding only and nine had both major and minor bleeding). Fourteen of them had received medication from zero to twelve hours postoperatively, and sixteen were more than sixty years old. Eight (0.5 percent) of the 1495 patients who were managed with warfarin also had major bleeding (four [0.3 percent] had major bleeding only and four had a combination of major and minor bleeding). Five of them had received medication from zero to ten hours before the operation, and all eight were more than fifty-five years old.
Nineteen of the twenty-six major bleeding episodes occurred at the site of the operation: fourteen episodes were in patients who were managed with enoxaparin, and five were in patients managed with warfarin (Table VII). Other sites or manifestations of major bleeding included the gastrointestinal tract and ecchymosis; no patient in either group had subconjunctival, retroperitoneal, or intracranial bleeding.
A minor bleeding episode occurred in 249 patients (143 patients who received enoxaparin and 106 who received warfarin); 152 episodes (61.0 percent) occurred at a site other than that of the operation, and ninety-seven (39.0 percent) occurred at the operative site (Table VII). Minor bleeding at the operative site occurred in sixty-two (43.4 percent) of the 143 patients who received enoxaparin and in forty-five (42.5 percent) of the 106 patients who received warfarin (Table VII).
The mean operative blood loss was 549 milliliters in the patients who received enoxaparin and 554 milliliters in those who received warfarin. A total of 1908 patients had operative blood loss of less than 500 milliliters: 976 patients (64.4 percent) who were managed with enoxaparin and 932 (62.3 percent) who were managed with warfarin.
A total of 1943 patients (64.5 percent) received transfusions because of operative blood loss: 1014 (66.9 percent) of the patients managed with enoxaparin and 929 (62.1 percent) of those managed with warfarin. Sixty-five patients (2.2 percent) received transfusions for replacement of postoperative blood loss: forty-three (2.8 percent) of those managed with enoxaparin and twenty-two (1.5 percent) of those managed with warfarin. A total of 1673 patients (55.6 percent) received between one and two units of blood, and 270 patients (9.0 percent) received more than two units.
Adverse Events
An adverse event occurred during the study period in 1921 (63.8 percent) of the 3011 patients: 987 (65.1 percent) of those managed with enoxaparin and 934 (62.5 percent) of those managed with warfarin. The most commonly reported adverse events were fever (486 patients; 16.1 percent), anemia (391 patients; 13.0 percent), and nausea (292 patients; 9.7 percent). A serious adverse event occurred in 301 patients (10.0 percent): 167 (11.0 percent) of the 1516 patients who received enoxaparin and 134 (9.0 percent) of the 1495 patients who received warfarin. There were no serious events that occurred in more than 5 percent of the patients.
The prevalence of deep venous thrombosis associated with elective hip arthroplasty without prophylaxis has ranged from 45 to 63 percent15,27,28, with a rate of 51 percent (332 of 655 patients) reported in an analysis of pooled data from thirteen studies9. Low-molecular-weight heparins and warfarin are the most effective antithrombotic agents for patients managed with total hip arthroplasty7,9. According to the American College of Chest Physicians Consensus Conference on Antithrombotic Therapy, studies have shown that prophylaxis with enoxaparin reduces the risk of deep venous thrombosis after hip arthroplasty by 71 percent (382 [15 percent] of 2571 patients), whereas prophylaxis with warfarin reduces the risk by 61 percent (127 [20 percent] of 637 patients)9. As far as we know, the present study is the first to compare enoxaparin and warfarin for the prevention of clinically symptomatic venous thromboembolic disease during hospitalization after total hip arthroplasty. In addition, enoxaparin and warfarin were compared with respect to the morbidity and mortality from venous thromboembolic disease after discontinuation of the medications and within three months after discharge from the hospital.
In the present study of 3011 patients, enoxaparin was found to be significantly better than warfarin for the prevention of clinically symptomatic venous thromboembolic disease during the period of hospitalization (p = 0.0083). Four (0.3 percent) of the 1516 patients managed with enoxaparin had venous thromboembolic disease compared with seventeen (1.1 percent) of the 1495 patients managed with warfarin.
During the period after discharge from the hospital, fifty-one (3.4 percent) of the patients managed with enoxaparin and thirty-nine (2.6 percent) of the patients managed with warfarin had venous thromboembolic disease (p = 0.2248). The benefit provided by enoxaparin during the period of hospitalization was lost during the period after hospitalization, when the drug was no longer administered. It could be hypothesized that a seven-to-ten-day course of enoxaparin or warfarin does not give total protection against venous thromboembolic disease and, in fact, a longer course does not provide such protection either3,23.
Eighteen (1.2 percent) of the patients managed with enoxaparin and eight (0.5 percent) of those managed with warfarin had either major bleeding only or a combination of major and minor bleeding. We found that the between-group difference in the occurrence of major bleeding episodes approached significance (p = 0.055), with fewer major bleeding episodes in the patients managed with warfarin.
It became apparent at the end of the study that the data-collection tool designed to capture overall bleeding events was neither sensitive nor specific enough to delineate bleeding events induced by the study medication from those caused by a concurrent illness or operative procedure. No definitive conclusion can be drawn with regard to actual bleeding events related to either enoxaparin or warfarin other than that caution should be exercised when either of these medications is administered with other medications possessing anticoagulation properties.
One of the benefits of enoxaparin is that monitoring of the prothrombin time or the activated partial prothrombin time is unnecessary, but periodic monitoring of platelet counts is recommended in the labeling approved by the Food and Drug Administration. During a short course (seven to ten days) of low-molecular-weight heparin, we monitor the platelet count as well as the hemoglobin level and the hematocrit on the first and second postoperative days and the platelet count only on the third and fourth postoperative days. The platelet count often is monitored on additional days when patients are managed with a longer course of enoxaparin.
The timing of the dose of the enoxaparin had a notable effect on the occurrence of major bleeding. Fourteen of the eighteen patients who had major bleeding in association with enoxaparin therapy were administered the medication from zero to twelve hours postoperatively. Enoxaparin rapidly achieves peak levels within three to five hours after being administered19. Because hemostasis is not established in most patients until twelve to twenty-four hours postoperatively, delaying the administration of the first dose of enoxaparin until this point ensures that the wound has stabilized. The current information on dosage and administration approved by the Food and Drug Administration recommends that enoxaparin be started twelve to twenty-four hours postoperatively.
Additional investigations related to the administration, monitoring, and stabilization of these two therapies are imperative before an economic analysis can be performed. Recent reports in the literature4,17,29 have appropriately identified the financial impact of the costs of the medication, laboratory tests, labor, education, readmissions, duration of hospitalization, and other factors as key attributes for consideration. The variables that occurred during the present study also require additional investigation before any conclusion on the financial impact can be drawn.