The composite primary end point of total venous thromboembolism and death from all causes during treatment was observed in 221 (22.5%) of 982 patients receiving ximelagatran and in 308 (31.9%) of 967 patients receiving warfarin (absolute risk reduction, 9.3%; relative risk reduction, 29.3%; p < 0.001; number needed to treat = 11; Table II) (see Appendix for an explanation of the number needed to treat and sensitivity analysis). The composite end point of total venous thromboembolism and death from all causes in the patients who underwent bilateral surgery performed under the same anesthesia session (4.5% of all of the patients underwent bilateral surgery; see electronic Appendix) was observed in thirteen (31.0%) of forty-two patients in the ximelagatran group and in twenty (43.5%) of forty-six patients in the warfarin group (absolute risk reduction, 12.5%), which was slightly higher than that in the overall efficacy population.
The composite secondary end point of proximal deep-vein thrombosis, pulmonary embolism, or death from any cause during treatment was observed in thirty-eight (3.9%) of 976 ximelagatran-treated patients and in forty (4.1%) of 964 warfarin-treated patients (p = 0.802, Table II). Confirmed symptomatic deep-vein thrombosis occurred in eight patients (0.7%) in the ximelagatran group and in fifteen patients (1.3%) in the warfarin group during the treatment period. Two (0.2%) of 1151 patients in the ximelagatran group and five (0.4%) of 1148 patients in the warfarin group had confirmed symptomatic pulmonary embolism during treatment.
During the follow-up period, symptomatic deep-vein thrombosis occurred in three patients in the ximelagatran group, one of whom also had pulmonary embolism. One patient had symptomatic deep-vein thrombosis and no patient had pulmonary embolism in the warfarin group. All four of these patients had deep-vein thrombosis detected at mandatory venography. An additional two patients had symptomatic venous thromboembolism develop during the follow-up period but did not have mandatory venography. Ten patients died during the study. Seven deaths occurred in the ximelagatran group; four patients died during the treatment period, including one who died because of gastrointestinal bleeding, and three died during the follow-up period. A pulmonary embolism could not be ruled out as a cause of one death during treatment and of two deaths that occurred during the follow-up period. In the warfarin group, three deaths occurred: two patients died during the treatment period, and one died during the follow-up period. The Independent Central Adjudication Committee judged that none of these deaths was associated with venous thromboembolism or bleeding.
In the patients assigned to warfarin treatment, the international normalized ratio was within or above the target range of 1.8 to 3.0 in 640 (67%) of 956 patients at postoperative day 3 and in 690 (73%) of 944 patients at the end of treatment. The mean international normalized ratio on the third day after surgery was 2.4. No appreciable differences were noted in the mean international normalized ratio values between patients with or without venous thromboembolism on either day.
Major bleeding occurred in twelve patients in the ximelagatran group and in five in the warfarin group during treatment (Table III). One patient had a fatal bleeding event and one bled at a critical site during the trial; both patients were in the ximelagatran group. See the Appendix for a list of the bleeding events. A similar pattern was observed in both treatment groups for any bleeding event, i.e., the combination of major and minor bleeding events (see electronic Appendix). No significant differences were detected between the ximelagatran and warfarin groups with respect to mean operative blood loss, postoperative wound drainage, and transfusion volumes.
Wound complications were assessed at three scheduled visits (postoperative day 3, end of treatment, and follow-up). The overall wound appearance was assessed as “as expected” or “better than expected” in the majority of patients in the ximelagatran group (1035 [90.6%] of 1142 patients) and in the warfarin group (1042 [91.3%] of 1141 patients). Three wound hematomas in three patients in the warfarin group required a return to the operating room. In one of them, the surgical site was débrided because of an infection, the implant was removed, and a spacer was implanted; another had irrigation and débridement of the knee and closure of the wound and muscle; and one had irrigation and débridement of the knee wound. Three wound hematomas in three patients in the ximelagatran group required bedside treatment. One of them was managed with elevation of the leg and application of a cold compress, another had surgical débridement, and the third had débridement of the knee wound and removal of an incisional hematoma. Treatment-emergent postoperative complications were similar in the two groups (see electronic Appendix).
Transient elevations of alanine aminotransferase have been observed in patients in long-term studies of ximelagatran21, but such elevations were rare in the ximelagatran-treated patients in the present study. Further explanation appears in the Appendix. No clinical signs or symptoms were attributed to the alanine aminotransferase elevations.
This study showed that postoperative administration of 36 mg of oral ximelagatran twice daily was safe and superior to postoperative warfarin in reducing the composite end point of venous thromboembolism and death from all causes, according to both independent central adjudication and local site assessment. The incidence of proximal deep-vein thrombosis, symptomatic deep-vein thrombosis, and pulmonary embolism in this trial was low, as in previous total knee arthroplasty trials of venous thromboprophylaxis with ximelagatran22,23.
In randomized trials and cohort studies, both warfarin and low-molecular-weight heparin have been shown to provide effective and safe prophylaxis after total knee arthroplasty. In studies directly comparing low-molecular-weight heparin with warfarin, the prevalence of venographically determined total deep-vein thrombosis ranged from 16% to 45% (mean, 28%) of a combined total of 1263 patients in the low-molecular-weight heparin groups compared with 27% to 55% (mean, 43%) of a combined total of 1207 patients in the warfarin groups1,7-11,24. Bleeding complications and transfusion requirements have generally been higher with low-molecular-weight heparin than with warfarin1,7,9,24. Fondaparinux lowered the risk of venous thromboembolism in comparison with enoxaparin in one trial involving total knee arthroplasty, but it was also associated with a significantly (p = 0.006) higher rate of major bleeding than was the low-molecular-weight heparin25. The advantage of warfarin over low-molecular-weight heparins and fondaparinux is its oral administration, which can facilitate prophylaxis in the hospital and after discharge. Nevertheless, warfarin has major limitations, such as a delayed onset of action, numerous drug and food interactions, and the need for coagulation monitoring and dose adjustment. Warfarin was chosen as the comparator because it is the most commonly used pharmacologic prophylaxis in patients undergoing total knee arthroplasty in North America26,27 and because it received the highest rating in the American College of Chest Physicians guidelines1.
The incidence of major bleeding was low (1.0% in ximelagatran-treated patients and 0.4% in warfarin-treated patients), and, although twice as many events occurred in the ximelagatran-treated patients, the difference was not significant. Major bleeding rates were lower than those reported in previous studies comparing low-molecular-weight heparins and warfarin in patients managed with total knee arthroplasty, which have ranged from 2.1% to 7.9%7,9,10,24.
This study was a multicenter, randomized, double-blind trial performed with a double-dummy technique. A high percentage (84.8%) of the randomized patients completed the protocol treatments and assessments for independent objective evaluation, assuring an accurate and unbiased comparison for both efficacy and safety outcomes. To our knowledge, the present study is the largest trial of warfarin and a comparator in patients managed with total knee arthroplasty only that has been performed to date, with the lowest reported rates of total and proximal venous thromboembolism and very low rates of bleeding events. A 36-mg dose of oral ximelagatran taken twice daily for seven to twelve days without coagulation monitoring or dose adjustment was significantly superior to well-controlled warfarin in reducing the prespecified composite primary end point of total venous thromboembolism (deep-vein thrombosis and pulmonary embolism) and death from all causes after primary total knee arthroplasty. The result was driven by a reduction in calf-vein thrombosis. The rates of major and minor bleeding events with both treatments were low and were not significantly different, although they were numerically greater with ximelagatran. ▪
Appendix Surgical Details of the Efficacy Population
Materials and Methods
Eligible patients were women without childbearing potential or men who were at least eighteen years old, who weighed between 40 and 136 kg, and who provided written informed consent. Only patients undergoing primary total knee arthroplasty were included. The criteria for exclusion were major surgery, stroke, myocardial infarction, or administration of any investigational drug within thirty days before surgery; immobilization for three or more days before surgery; traumatic epidural and/or spinal puncture at surgery; planned pneumatic leg compression; a history of intracranial, retroperitoneal, or intraocular bleeding; a history of gastrointestinal bleeding or other disorder associated with an increased risk of bleeding within ninety days before surgery; an endoscopically verified peptic ulcer disease within thirty days of surgery; uncontrolled hypertension; a malignant tumor receiving cytostatic treatment or being the reason for the knee arthroplasty; an alanine or aspartate aminotransferase level greater than two times the upper limit of normal; renal impairment (defined as estimated creatinine clearance of <30 mL/min)18; thrombocytopenia; a history of drug or alcohol abuse in the past six months; an allergy to venographic contrast media or iodine; and a contraindication to warfarin.
To guide warfarin dosing, international normalized ratios were measured at the study site (with treatment blinding maintained) on postoperative days 1, 2, and 3; the day of venography; and as required. International normalized ratios were measured either at the point of care with use of encrypted devices provided or at local laboratories with a system that prevented access to the international normalized ratios by the local study team. All international normalized ratios were reported by means of an interactive voice-response system, which then relayed back to the local sites either the actual international normalized ratios for the warfarin patients or, for those patients receiving ximelagatran, the sham international normalized ratios generated to mimic values typically observed in warfarin-treated patients. A warfarin-dosing nomogram was provided, but the choice of either the warfarin or the placebo dose was made at the individual investigator's discretion. Treatment compliance was assessed for ximelagatran by the number of doses taken (and missed) during the study as reported in the medication logs and for warfarin by the percentage of patients within the therapeutic range.
Assessment for Deep-Vein Thrombosis and Pulmonary Embolism
The criterion for a diagnosis of deep-vein thrombosis was a consistent intraluminal filling defect visible on at least two images. To be considered adequate for evaluation, venograms needed to show all deep veins except the deep femoral vein, the muscular veins of the calf, and the anterior tibial veins, although these veins were included in the evaluation if visible or if a clot was detected. Venograms were classified as indeterminate if there was a lack of filling of a region of the deep-vein system without a filling defect in the same region. Symptomatic proximal deep-vein thrombosis could be diagnosed by compression ultrasound, but diagnosis of symptomatic distal deep-vein thrombosis required venography. A diagnosis of pulmonary embolism was made if (1) a ventilation-perfusion lung scan revealed one or more segmental pulmonary perfusion defects in at least two views with corresponding normal ventilation, (2) pulmonary angiography showed a persistent intraluminal defect or an abrupt cutoff of a vessel that was >2.5 mm in diameter, (3) a spiral computed tomographic scan demonstrated a distinct filling defect in a large vessel, (4) embolectomy was performed, or (5) it was determined by autopsy.
Bleeding events were classified as “major” if they were clinically overt (defined as clinically apparent bleeding or signs and/or symptoms suggestive of bleeding with confirmatory imaging studies [e.g., ultrasound or computed tomography]) and met one or more of the following criteria: (1) involvement of a critical site (intracranial, retroperitoneal, intraocular, intraspinal, or pericardial); (2) a bleeding index of 2.0 or more (calculated as the baseline hemoglobin level [in grams per liter] minus the hemoglobin level at the end of treatment plus the number of units of packed red-blood cells or whole blood transfused); (3) a need for medical or surgical intervention; or (4) fatal bleeding. Clinically overt bleeding episodes that did not meet any of these criteria were classified as “minor,” and, if bleeding episodes were not clinically overt, they were classified as “criteria for bleeding event not satisfied.”
To ensure that adequate objective efficacy assessments could be made on at least 1720 patients at the end of the study, we planned to randomize approximately 2300 patients, anticipating that, at most, 25% of the venograms would be inadequate for evaluation. The efficacy analyses included all patients who had received at least one dose of study medication and had a venogram adequate for evaluation; had symptomatic, objectively confirmed deep-vein thrombosis or pulmonary embolism; or had died during treatment. Bilateral venography was required, but patients undergoing unilateral surgery were included in the efficacy analysis even if a venogram adequate for evaluation could be obtained only in the leg on which surgery was performed or if a clot was visualized in the contralateral side. For patients who underwent bilateral knee surgery, venograms adequate for evaluation in both legs were required unless deep-vein thrombosis was detected in one leg. Symptomatic venous thromboembolism or deaths occurring within two days after venography or up to day 12 if no venography was performed were deemed to have occurred during the treatment period.
Baseline Characteristics of the Patients
Within the efficacy population, forty-four (4.5%) of 982 patients assigned to ximelagatran and thirty-one (3.2%) of 967 patients assigned to warfarin had a history of venous thromboembolism. The details of the operation and the course of hospitalization were also similar between the treatment groups. The surgical position was supine for all patients. The surgical approach was medial parapatellar in 888 (90.4%) of 982 ximelagatran-treated patients and 868 (89.8%) of 967 warfarin-treated patients, and a tibial tuberosity osteotomy was performed in only eight (0.8%) of 982 patients in the ximelagatran group and in twelve (1.2%) of 967 patients in the warfarin group. Approximately 130 (13.2%) of 982 patients in the ximelagatran group and 145 (15%) of 967 patients in the warfarin group underwent epidural catheter placement, with a mean duration of use between nine and eleven hours. The mean time to the first dose of medication on the day after surgery was 20.2 hours. The mean time to walking was 1.9 days, and the mean hospital stay was 5.6 days. Most patients (1082 [94%] of 1151 patients in the ximelagatran group and 919 [80%] of 1148 patients in the warfarin group) received study medication for seven to twelve days, with the lower percentage of patients in the warfarin group being primarily due to the need to withhold doses when a therapeutic or elevated international normalized ratio was obtained. Efficacy data adequate for evaluation were available for 1949 (84.8%) of 2299 treated patients. After the treatment period, 30% of the patients in each group received follow-up anticoagulant therapy.
The number of patients needed to treat to prevent one composite primary end-point event with ximelagatran compared with warfarin was eleven patients (95% confidence interval, eight to nineteen patients). Stated another way, the relative risk reduction indicates nine fewer venous thrombosis events per 100 patients undergoing total knee arthroplasty with ximelagatran than with warfarin. In a sensitivity analysis of patients who had total knee arthroplasty that excluded sixty-two patients (thirty-four in the ximelagatran group and twenty-eight in the warfarin group) in whom unilateral but not bilateral venography was performed, similar reductions were achieved (absolute risk reduction, 9.5%; relative risk reduction, 28.9%; p < 0.001).
Local Interpretation of Venograms
The incidence of total venous thromboembolism and death from all causes with use of the local interpretation of venograms was 300 (30.1%) of 996 ximelagatran-treated patients compared with 363 (35.8%) of 1014 warfarin-treated patients (p = 0.007). The superior efficacy of ximelagatran was consistent across all subgroups analyzed, on the basis of gender, age, country, history of venous thromboembolism, body weight, estimated creatinine clearance, type of anesthesia, and time to venography. No significant interactions were found between treatment and any of the subgroup factors (p ≥ 0.304), indicating that the efficacy of ximelagatran (relative to warfarin) was consistent in all of the examined subgroup categories.
The types of major bleeding events included wound hematoma (four patients in the ximelagatran group and three in the warfarin group), gastrointestinal bleeding (four and one, respectively), wound bleeding (three and one, respectively), and a subdural hematoma (an exacerbation of a chronic subdural hematoma due to a recent fall) in one patient in the ximelagatran group. The gastrointestinal bleeding event in one patient in the ximelagatran group was due to a bleeding duodenal ulcer and was judged to be the cause of death in that patient. One additional major bleeding event occurred in each treatment group during the follow-up period. Approximately half (seven) of all thirteen major bleeding events in the ximelagatran-treated patients occurred in the first three postoperative days, while five of the six bleeding events in the warfarin group occurred after day 4.
The first onset of an elevation of alanine aminotransferase levels greater than three times the upper limit of normal was detected in five (0.5%) of 1078 patients in the ximelagatran group and in six (0.6%) of 1071 patients in the warfarin group at the end of treatment and in five (0.5%) of 1091 patients and one (0.1%) of 1083 patients, respectively, at the follow-up visit (more than two days after the last dose of study medication). The alanine aminotransferase values normalized in all of the ximelagatran-treated patients within four weeks of onset. No clinical signs or symptoms were attributed to the alanine aminotransferase elevations.
The membership of the EXULT B Study Group, a figure presenting the cumulative proportion of bleeding events, and tables presenting the surgical details of the efficacy population and evaluation of the surgical wounds are available with the electronic versions of this article, on our web site at jbjs.org (go to the article citation and click on “Supplementary Material”) and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM). ▪
NOTE: The authors are indebted to Mary E. Hardwick, MSN, and Joyce S. Weiner, PhD, for editorial support in the preparation of the manuscript.
A commentary is available with the electronic versions of this article, on our web site () and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).
In support of their research or preparation of this manuscript, one or more of the authors received grants or outside funding from AstraZeneca LP. In addition, one or more of the authors received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity (AstraZeneca LP). No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, educational institution, or other charitable or nonprofit organization with which the authors are affiliated or associated.
* The members of the EXULT B (Exanta Used to Lessen Thrombosis B) Study Group are listed in the electronic Appendix.
1. , Heit JA, Clagett GP, Pineo GF, Colwell CW, Anderson FA Jr, Wheeler HB. Prevention of venous thromboembolism. Chest. 2001;119(1 Suppl): 132S-75S.
2. , Insall JN, Williams GW, Ghelman B. Deep-vein thrombosis following total knee replacement. An analysis of six hundred and thirty-eight arthroplasties. J Bone Joint Surg Am. 1984;66: 194-201.
3. , Steadman CA, Hedges AR, Thomas EM, Morley TR, Kakkar VV. Deep vein thrombosis after elective knee surgery. An incidence study in 312 patients. J Bone Joint Surg Br. 1989;71: 492-7.
4. , Mohr DN, Silverstein MD, Petterson TM, O'Fallon WM, Melton L J 3rd. Predictors of recurrence after deep vein thrombosis and pulmonary embolism: a population-based cohort study. Arch Intern Med. 2000;160: 761-8.
5. , Turkstra F, Buller HR, MacKinnon B, Magier D, Hirsh J. Postthrombotic syndrome after hip or knee arthroplasty: a cross-sectional study. Arch Intern Med. 2000;160: 669-72.
6. , Hirsh J, Julian J, Vander LaandeVries M, Magier D, MacKinnon B, Gent M. Prevention and treatment of postphlebitic syndrome: results of a 3-part study. Arch Intern Med. 2001;161: 2105-9.
7. , Berkowitz SD, Bona R, Cabanas V, Corson JD, Elliott CG, Lyons R. Efficacy and safety of low molecular weight heparin (ardeparin sodium) compared to warfarin for the prevention of venous thromboembolism after total knee replacement surgery: a double-blind, dose-ranging study. Ardeparin Arthroplasty Study Group. Thromb Haemost. 1997;77: 32-8.
8. . RD heparin compared with warfarin for prevention of venous thromboembolic disease following total hip or knee arthroplasty. J Bone Joint Surg Am. 1994;76: 1174-85.
9. , Spiro TE, Trowbridge AA, Gardiner GA Jr, Whitsett TL, O'Connell MB, Ohar JA, Young TR; Enoxaparin Clinical Trial Group. Prevention of venous thromboembolic disease following primary total knee arthroplasty. A randomized, multicenter, open-label, parallel-group comparison of enoxaparin and warfarin. J Bone Joint Surg Am. 2001;83: 900-6.
10. , Raskob G, Pineo G, Rosenbloom D, Evans W, Mallory T, Anquist K, Smith F, Hughes G, Green D, Elliott CG, Panju A, Brant R. A comparison of subcutaneous low-molecular-weight heparin with warfarin sodium for prophylaxis against deep-vein thrombosis after hip or knee implantation. N Engl J Med. 1993;329: 1370-6.
11. , Lensing AW, van der Meer J, Smid WM, van Ooy A, Hoek JA. Subcutaneous low-molecular weight heparin or oral anticoagulants for the prevention of deep-vein thrombosis in elective hip and knee replacement? Fraxiparine Oral Anticoagulant Study Group. Thromb Haemost. 1995;74: 1428-31.
12. , Gent M, Hirsh J, Geerts WH, Ginsberg JS. The incidence of symptomatic venous thromboembolism during and after prophylaxis with enoxaparin: a multi-institutional cohort study of patients who underwent hip or knee arthroplasty. Canadian Collaborative Group. Arch Intern Med. 1998;158: 873-8.
13. , Bredberg U, Gislen K, Johansson LC, Frison L, Ahnoff M, Gustafsson D. Pharmacokinetics and pharmacodynamics of ximelagatran, a novel oral direct thrombin inhibitor, in young healthy male subjects. Eur J Clin Pharmacol. 2003;59: 35-43.
14. , Carlsson S, Rak J, Liao P, Vlasin M, Stafford A, Johnston M, Weitz JI. The benefit-to-risk profile of melagatran is superior to that of hirudin in a rabbit arterial thrombosis prevention and bleeding model. J Thromb Haemost. 2003;1: 587-94.
15. , Andersson TB, Frison L, Thuresson A, Johansson S, Eriksson-Lepkowska M, Larsson M, Eriksson UG. Ximelagatran, an oral direct thrombin inhibitor, has a low potential for cytochrome P450-mediated drug-drug interactions. Clin Pharmacokinet. 2003;42: 765-77.
16. , Bergqvist D, Kalebo P, Dahl OE, Lindbratt S, Bylock A, Frison L, Eriksson UG, Welin L, Gustafsson D; Melagatran for Thrombin Inhibition in Orthopaedic Surgery. Ximelagatran and melagatran compared with dalteparin for prevention of venous thromboembolism after total hip or knee replacement: the METHRO II randomised trial. Lancet. 2002;360: 1441-7.
17. , Agnelli G, Cohen AT, Dahl OE, Mouret P, Rosencher N, Eskilson C, Nylander I, Frison L, Ogren M; METHRO III Study Group. Direct thrombin inhibitor melagatran followed by oral ximelagatran in comparison with enoxaparin for prevention of venous thromboembolism after total hip or knee replacement. Thromb Haemost. 2003;89: 288-96.
18. , Berkowitz SD, Comp PC, Lieberman JR, Ginsberg JS, Paiement G, Peters GR, Roth AW, McElhattan J, Colwell CW Jr; EXULT A Study Group. Comparison of ximelagatran with warfarin for the prevention of venous thromboembolism after total knee replacement. N Engl J Med. 2003;349: 1703-12.
19. , Ekman S, Lindbratt S, Eriksson BI, Pauli U, Zachrisson BE, Close P. Percentage of inadequate phlebograms and observer agreement in thromboprophylactic multicenter trials using standardized methodology and central assessment. Thromb Haemost. 1996;76: 893-6.
20. , Paulin S. Roentgen diagnosis of venous thrombosis in the leg. Arch Surg. 1972;104: 134-44.
21. , Olsson R, Lewis JH, Keisu M, Auclert L, Sheth S. Hepatic findings in long-term clinical trials of ximelagatran. Drug Saf. 2005;28: 351-70.
22. , Agnelli G, Cohen AT, Dahl OE, Lassen MR, Mouret P, Rosencher N, Kalebo P, Panfilov S, Eskilson C, Andersson M, Freij A; EXPRESS Study Group. The direct thrombin inhibitor melagatran followed by oral ximelagatran compared with enoxaparin for the prevention of venous thromboembolism after total hip or knee replacement: the EXPRESS study. J Thromb Haemost. 2003;1: 2490-6.
23. , Davidson BL, Berkowitz SD, Lotke PA, Ginsberg JS, Lieberman JR, Webster AK, Whipple JP, Peters GR, Colwell CW Jr. Ximelagatran versus warfarin for the prevention of venous thromboembolism after total knee arthroplasty. A randomized, double-blind trial. Ann Intern Med. 2002;137: 648-55.
24. , Geerts WH, Desjardins L, Laflamme GH, L'Esperance B, Demers C, Kassis J, Cruickshank M, Whitman L, Delorme F. Prevention of venous thromboembolism after knee arthroplasty. A randomized, double-blind trial comparing enoxaparin with warfarin. Ann Intern Med. 1996;124: 619-26.
25. , Eriksson BI, Lassen MR, Turpie AG; Steering Committee of the Pentasaccharide in Major Knee Surgery Study. Fondaparinux compared with enoxaparin for the prevention of venous thromboembolism after elective major knee surgery. N Engl J Med. 2001;345: 1305-10.
26. , Brand RA, Iorio R, Gradisar I, Heekin R, Leighton R, Thornberry R. Venous thromboembolic disease management patterns in total hip arthroplasty and total knee arthroplasty patients: a survey of the AAHKS membership. J Arthroplasty. 2001;16: 679-88.
Copyright 2005 by The Journal of Bone and Joint Surgery, Incorporated
27. , Hirsh J, White K, Fitzgerald RH Jr; Hip and Knee Registry Investigators. Temporal trends in prevention of venous thromboembolism following primary total hip or knee arthroplasty 1996-2001: findings from the Hip and Knee Registry. Chest. 2003;124(6 Suppl): 349S-56S.