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


Dr B. I. Eriksson, The Department of Orthopedics, Sahlgrenska University Hospital/Östra, S-41685 Göteborg, Sweden.
Tel.: +46 31 343 4408; fax: +46 31 343 4092; e-mail:


Summary. Background: Ximelagatran and its subcutaneous (s.c.) form melagatran are novel direct thrombin inhibitors for the prevention and treatment of thromboembolic disease. Methods: In a double-blind study, 2835 consecutive patients undergoing total hip or knee replacement were randomized to either melagatran/ximelagatran or enoxaparin. Melagatran 2 mg was started immediately before surgery; 3 mg was then administered postoperatively, followed by 24 mg of oral ximelagatran b.i.d. beginning the next day. Enoxaparin 40 mg, administered subcutaneously o.d., was started 12 h before surgery. Both treatments were continued for 8–11 days. The main efficacy outcome measures were major venous thromboembolism (VTE); [proximal deep vein thrombosis (DVT), non-fatal and/or fatal pulmonary embolism (PE), death where PE could not be ruled out], and total VTE (proximal and distal DVT; PE; death from all causes). DVT was detected by mandatory bilateral ascending venography at the end of the treatment period or earlier if clinically suspected. The main safety outcome was bleeding. Results: The rates of major and total VTE were significantly lower in the melagatran/ximelagatran group compared with the enoxaparin group (2.3% vs. 6.3%, P = 0.0000018; and 20.3% vs. 26.6%, P < 0.0004, respectively). Fatal bleeding, critical site bleeding and bleeding requiring reoperation did not differ between the two groups. ‘Excessive bleeding as judged by the investigator’ was more frequent with melagatran/ximelagatran than with enoxaparin. Conclusions: In patients undergoing total hip or knee replacement, preoperatively initiated s.c. melagatran followed by oral ximelagatran was significantly more effective in preventing VTE than preoperatively initiated s.c. enoxaparin.


Low-molecular-weight heparins (LMWHs) and vitamin K antagonists such as warfarin are the drugs most frequently used for prophylaxis of venous thromboembolism (VTE) in major orthopedic surgery [1,2]. However, the risk of proximal deep vein thrombosis (DVT) is still 5–7% with prophylaxis in both total hip replacement (THR) and total knee replacement (TKR) [1]. These agents have further limitations: the use of LMWHs requires subcutaneous (s.c.) administration and may result in thrombocytopenia [3], and the anticoagulant activity of oral vitamin K antagonists is poorly predictable, making monitoring and dose adjustments mandatory [4,5]. More effective and convenient oral anticoagulants are therefore needed.

Oral ximelagatran (Exanta™; AstraZeneca, Mölndal, Sweden) and its s.c. form, melagatran, provide potent, competitive and direct inhibition of free and clot-bound thrombin [6]. Melagatran exhibits predictable pharmacokinetics and can thus be used without coagulation monitoring [7–11]. In a dose-finding study (METHRO II) in elective joint-replacement surgery, 3 mg of s.c. melagatran started preoperatively followed by a fixed dose of 24 mg of oral ximelagatran b.i.d. was significantly more effective than the LMWH dalteparin in preventing VTE, but showed a somewhat less favorable bleeding profile [10]. To improve the efficacy:bleeding ratio, the postoperative start of prophylaxis with melagatran/ximelagatran was evaluated in the METHRO III study in the same clinical setting [11]. The results of this study concerning major VTE and bleeding demonstrated that the same dosage regimen of melagatran/ximelagatran initiated 4–12 h after surgery was similar to enoxaparin started preoperatively [11]. Thus, the timing of s.c. melagatran administration in relation to the start of surgery seems to influence efficacy, which is consistent with the results obtained with the LMWH dalteparin [12].

To optimize further the efficacy:bleeding profile of melagatran/ximelagatran, we conducted a new study investigating a modified preoperatively started dosing regimen. Because of the advantage of oral administration in terms of convenience, this study was planned to show a non-inferiority of melagatran/ximelagatran in the first instance, and, once the non-inferiority was demonstrated, to test for superiority compared with enoxaparin.



Consecutive patients undergoing primary elective unilateral THR or TKR were included. Exclusion criteria were the following: recent (within 1 month) stroke; recent trauma or major surgical procedure; history of intracranial bleeding or of intraocular bleeding within the last year; history of gastrointestinal bleeding within 3 months prior to surgery; endoscopically verified ulcer disease; ongoing malignancy; uncontrolled hypertension; disorders associated with a risk of bleeding; severe renal impairment; known active liver disease, or liver insufficiency. Women of childbearing age were excluded if they were pregnant, nursing or not using effective contraception.

Throughout the study treatment period, concomitant treatment with dextran, anticoagulant, antiplatelet (except for aspirin up to 500 mg daily) or thrombolytic agents was prohibited. However, short-acting non-steroidal anti-inflammatory drugs and cyclooxygenase-2 inhibitors were allowed.

Written informed consent was required from all patients prior to enrolment in the study. The protocol was approved by independent local ethics committees or institutional review boards. The study was conducted in accordance with the Declaration of Helsinki.

Study design

This was a multicenter, randomized, double-blind, parallel-group study conducted at 77 centers. Eligible patients were randomized the day before surgery, using a computer-generated central randomization list. Patients were assigned to receive either melagatran/ximelagatran or enoxaparin. A double-dummy technique was used, i.e. the patients in both study groups received active or matching placebo medications. The melagatran/ximelagatran regimen consisted of 2 mg of melagatran administered s.c. immediately before surgery (for practical reasons a time window of ± 30 min from the start of surgery was allowed) and 3 mg of melagatran administered s.c. the evening after surgery (at least 8 h after the first administration), followed by 24 mg of ximelagatran given orally b.i.d. (hereafter referred to as the ximelagatran group). The first injection of melagatran was to be given after induction of any regional block anesthesia; in the event of a technical problem or associated bleeding complication, the patient had to be withdrawn from study treatment. Oral ximelagatran was to be initiated the day after surgery. Patients unable to take oral medication at that time could be given further s.c. injections of 3 mg of melagatran b.i.d. on the first two postoperative days. Patients randomized to treatment with enoxaparin (Aventis Pharma, Bridgewater, NJ, USA) received enoxaparin according to European standard practice, i.e. 40 mg s.c. o.d., starting 12 h before surgery.

Prophylaxis was scheduled for 8–11 days, until mandatory venography was performed. If a confirmed symptomatic venous thromboembolic event occurred before the scheduled venography, the study drug was to be discontinued and the patient treated according to local practice. Patients underwent a follow-up visit 4–6 weeks after surgery.

Outcome measures

Two primary efficacy outcomes were defined in accordance with current guidelines [13]. The first was confirmed major VTE during the treatment period, defined as proximal DVT, pulmonary embolism (PE) and/or death where PE could not be ruled out [13]. The second was confirmed total VTE during the treatment period, defined as DVT (distal or proximal), PE and/or death from any cause [13].

DVT was diagnosed by bilateral venography using the Rabinov and Paulin technique [14] with minor modifications [15]. All venograms were centrally assessed by two independent, blinded experts and categorized as normal, DVT or non-evaluable. Symptomatic DVT occurring during the follow-up period could also be confirmed by ultrasonography. Symptomatic PE had to be confirmed by ventilation–perfusion lung scanning indicating a high probability of the diagnosis [16], or by pulmonary angiography or spiral computed tomography. In the event of death, an autopsy was performed whenever possible.

Safety variables were bleeding, other adverse events and laboratory parameters. Severe bleeding was defined as fatal bleeding, bleeding involving a critical site (intracranial, intraocular, intraspinal or retroperitoneal), excessive bleeding as judged by the investigator and overt bleeding (except bleeding from the operation wound) leading to transfusions of two or more units of packed red blood cells or whole blood. In practice, ‘excessive bleeding as judged by the investigator’ was based on an observational assessment of more bleeding from the operation wound, either during the procedure or as a drainage, than a surgeon would normally expect to see. In single cases, this category also included gastrointestinal bleeding manifested by melena or hematemesis. Other bleeding events were those reported as adverse events by the investigator and not fulfilling the criteria for severe bleeding. Volumes of blood loss (bleeding volume during surgery plus postoperative wound drainage) and transfusion requirements were measured throughout the treatment period. No formal comparison of bleeding parameters was prospectively defined, therefore these data are presented descriptively with 95% confidence intervals (CIs) but P-values are not given.

All venous thromboembolic events, deaths and bleeding events were adjudicated by an independent adjudication committee, the members of which were unaware of the patients' treatment assignments. Death was classified as fatal PE, death unrelated to PE or unexplained death.

Statistical analysis

The trial was designed with two sequentially assessed primary objectives as described in recent guidelines for drug development [13,17]. The first-stage primary objective was to demonstrate that ximelagatran was non-inferior to enoxaparin in preventing major VTE. If non-inferiority was demonstrated, the analysis could proceed to the second stage, which was to demonstrate that ximelagatran was more effective than enoxaparin in preventing total VTE. Such a procedure is an established statistical instrument known as a closed testing procedure [18].

The planned sample size was based on previous experience of comparing melagatran and ximelagatran with LMWHs [10,11]. Estimates assumed a total VTE rate of 27% and a major VTE rate of 6% with enoxaparin, and predicted a risk reduction of 25% with the melagatran/ximelagatran regimen in both cases. To demonstrate these differences with 90% power at a 5% level of significance would require 1050 patients per treatment group. We planned to enrol approximately 2600 patients to detect the specified risk reduction, anticipating that 25% of patients would not have an evaluable venogram.

Non-inferiority of ximelagatran in comparison with enoxaparin (first stage) was assessed with a 97.5% one-sided CI for the risk difference between treatments, with an absolute margin of non-inferiority (Δ) of two percentage points. A CI below Δ was to be taken as proof of non-inferiority in accordance with current guidelines [13,17]. In the event that not only non-inferiority but also superiority be established, the non-inferiority criteria will be redundant [17]. Superiority of ximelagatran over enoxaparin (second stage) was tested by Fisher's exact test against a two-sided alternative including calculation of the difference in risk with a corresponding 95% CI.

The results are given as proportions with 95% CIs, and as mean ± SD or as the geometric mean with 95% CI, as appropriate.

All analyses were performed according to the intention-to-treat (ITT) principle whereby all patients undergoing the appropriate surgery and receiving at least one dose of study medication were included (referred to as the ITT population). Patients with inadequate or missing mandatory venography who neither died nor experienced venous thromboembolic events were excluded from efficacy analyses.

Role of the funding source

The study was initiated, designed and conducted by a steering committee of 11 people, including four representatives of the sponsor (AstraZeneca). The study data are held in an AstraZeneca database. AstraZeneca provided a study writing committee of 12 people (including three representatives of the sponsor) with full and free access to all data. The authors had full independence in deciding whether to publish.


Study populations

Between April 2001 and February 2002, 2835 patients were randomized and treated, of whom 14 did not receive any study drug and 57 did not undergo primary THR or TKR, leaving 2765 (97.5%) patients in the ITT population (Table 1). Two patients (both randomized to the enoxaparin group) were excluded because of traumatic spinal puncture. Finally, the numbers of patients assessed for first-stage and second-stage primary efficacy were 2316 (81.7%) and 2326 (82.0%) patients, respectively (Table 1). The difference is due to availability of evaluable venograms.

Table 1.  Patients included in the analysis
 Ximelagatran groupEnoxaparin group
  1. ITT, intention-to-treat. *The number of patients with available data for this second-stage primary efficacy outcome was higher than that included in the first-stage primary efficacy outcome population, since visualization of proximal deep veins in both legs was no longer a prerequisite once distal deep vein thrombosis had been observed.

Randomized, n (%)1410 (100)1425 (100)
 Not treated7 (0.5)7 (0.5)
 No appropriate surgery26 (1.8)31 (2.2)
ITT population, n (%)1377 (97.7)1387 (97.3)
Evaluable for first-stage primary efficacy outcome, n (%)1138 (80.7)1178 (82.7)
Evaluable for second-stage primary efficacy outcome*, n (%)1142 (81.0)1184 (83.1)

Baseline and surgical characteristics were similar in the two treatment groups (Table 2). Overall, 2629 (95%) of 2764 patients received the study drug for 8–11 days as planned. Of the 1377 patients in the ximelagatran group, 1294 (94%) and 1355 (98%) received oral study medication within 24 h and within 2 days after surgery, respectively.

Table 2.  Baseline demographic and surgical characteristics of the ITT population
 Ximelagatran group
(n = 1377)
Enoxaparin group
(n = 1387)
  1. ITT, intention-to-treat; THR, total hip replacement; TKR, total knee replacement. *n = 1376;n = 1373;n = 1384.

Age, years, median (range)67 (24–88)67 (20–89)
Weight, kg, median (range)78 (43–137)*79 (43–146)
Previous venous thromboembolism, n (%)47 (3.4)50 (3.6)
Type of surgery, n (%)
 THR914 (66.4)942 (67.9)
 TKR463 (33.6)445 (32.1)
Reason for THR or TKR, n (%)
 Osteoarthrosis1274 (92.5)1297 (93.5)
 Rheumatoid arthritis34 (2.5)20 (1.4)
 Septic arthritis2 (0.1)2 (0.1)
 Post hip or knee fracture34 (2.5)24 (1.7)
 Other33 (2.4)44 (3.2)
Cemented prosthesis, n (%)850 (61.7)847 (61.1)
Type of anesthesia, n (%)
 General538 (39.1)534 (38.5)
 Regional only839 (60.9)853 (61.5)
Duration of surgery, min, median (range)85 (15–220)85 (26–205)
Days until mobilization, median (range)2 (0–28)2 (0–28)


The rate of major VTE was 2.3% in the ximelagatran group and 6.3% in the enoxaparin group, a difference in risk of −4.0% (95% CI −5.6%, −2.4%) (Table 3). Thus, ximelagatran was more effective than enoxaparin in reducing the risk of major VTE (relative risk reduction 63.2%). Similarly, ximelagatran was more effective than enoxaparin in reducing the risk of total VTE, from 26.6% to 20.3%, a difference in risk of −6.3% (95% CI −9.7%, −2.9%; relative risk reduction 23.7%) in favor of ximelagatran.

Table 3.  Major and total VTE by type of surgery and by treatment
 Ximelagatran groupEnoxaparin groupDifference in risk
estimate, %
(95% CI)
n*Risk, %
(95% CI)
n*Risk, %
(95% CI)
  1. CI, confidence interval; THR, total hip replacement; TKR, total knee replacement; VTE, venous thromboembolism. *The number of patients with events/total number of patients.†For a test of superiority using Fisher's exact test vs. the null hypothesis of no difference.

Major VTE
 THR and TKR26/11382.3 (1.5, 3.3)74/11786.3 (5.0, 7.8)−4.0 (−5.6, −2.4)0.0000018
 THR14/7731.8 (1.0, 3.0)45/8235.5 (4.0, 7.3)−3.7 (−5.6, −1.8) 
 TKR12/3653.3 (1.7, 5.7)29/3558.2 (5.5, 11.5)−4.9 (−8.3, −1.5) 
Total VTE
 THR and TKR231/114120.3 (18.0, 22.7)315/118426.6 (24.1, 29.2)−6.4 (−9.8, −2.9)0.0003
 THR99/76512.9 (10.6, 15.5)146/80118.2 (15.6, 21.1)−5.3 (−8.9, −1.7) 
 TKR132/37635.1 (30.3, 40.2)169/38344.1 (39.1, 49.3)−9.0 (−16.0, −2.1) 

The superiority of the ximelagatran regimen was demonstrated in subgroups of patients undergoing THR or TKR (Table 3). In THR, the rate of major VTE was significantly lower in the melagatran/ximelagatran group compared with the enoxaparin group: 1.8% vs. 5.5% (relative risk reduction 67%). Likewise, in TKR the rate of major VTE was 3.3% vs. 8.2% in the melagatran/ximelagatran and enoxaparin groups, respectively (relative risk reduction 60%). The rate of total VTE was also consistently lower in the melagatran/ximelagatran group compared with the enoxaparin group: 13.1% vs. 18.2% in THR (relative risk reduction 28.0%), and 35.1% vs. 44.1% in TKR (relative risk reduction 21%).

During the study treatment period, eight of the 1378 ximelagatran-treated patients (0.6%) and 12 of the 1387 enoxaparin-treated patients (0.9%) experienced a confirmed symptomatic venous thromboembolic event (DVT, PE or both). During the follow-up period, 12 additional patients (six in each treatment group) developed a confirmed symptomatic venous thromboembolic event. During the study treatment period, the rate of confirmed PE was low: two in the ximelagatran group and four in the enoxaparin group. During follow-up, PE was confirmed in one patient in the ximelagatran group and two in the enoxaparin group, and there were two deaths in the ximelagatran group in which PE could not be excluded.


During the study treatment period, there was one case of fatal bleeding in the ximelagatran group (injury to iliac and femoral vessels at the initial phase of the planned surgery) and no cases of bleeding involving a critical site. Adjudicated severe bleeding events were observed in 46 patients (3.3%; 95% CI 2.4, 4.4) in the ximelagatran group and 16 patients (1.2%; 95% CI 0.7, 1.9) in the enoxaparin group (Table 4). They consisted of excessive bleeding as judged by the investigator (3.1% vs. 1.2%, respectively) and non-wound-related transfusions (0.2%, in the ximelagatran group only), and most (77%) occurred on the day of surgery. More transfusions (66.8% vs. 61.7%) and somewhat higher volumes of blood loss (geometric mean volume 1014 mL and 913 mL, respectively) were noted in the ximelagatran group than in the enoxaparin group. However, the intraoperative blood loss was similar in both groups (195 mL and 193 mL, respectively). Patients who underwent THR differed from those who underwent TKR with regard to bleeding. Among THR patients, excessive bleeding as judged by the investigator occurred in 37 (4.0%) of 914 cases in the ximelagatran group compared with 10 (1.1%) of 942 cases in the enoxaparin group. Among TKR patients, there was no difference between the treatment groups: nine (1.9%) of 463 cases and six (1.4%) of 445 cases in the ximelagatran and enoxaparin groups, respectively.

Table 4.  Bleeding complications in the safety population
  1. CI, confidence interval; THR, total hip replacement; TKR, total knee replacement; VTE, venous thromboembolism. *Blood loss was defined as the sum of bleeding volume during surgery and postoperative wound drainage.†Geometric mean volume.

Adjudicated bleeding
 Patients, n13781387915942463445
 Severe bleeding, n (%)46 (3.3)16 (1.2)37 (4.0)10 (1.1)9 (1.9)6 (1.4)
 Other bleeding, n (%)126 (9.2)97 (7.0)87 (9.5)61 (6.5)39 (8.4)36 (8.1)
Blood loss*
 Patients, n13651372906932459440
 Total blood loss, mL (95% CI)1015 (974, 1056)913 (877, 950)1189 (1148, 1232)1029 (994, 1065)741 (675, 814)709 (644, 780)
 Bleeding during surgery,
 mL (95% CI)
197 (176, 220)194 (173, 216)519 (499, 540)478 (459, 496)29 (22, 37)29 (22, 37)
 Postoperative wound
 drainage, mL (95% CI)
645 (617, 673)553 (530, 578)611 (579, 646)511 (484, 539)717 (670, 768)655 (611, 703)
Blood transfusions
 Patients, n13761385913940463445
 Patients, % (95% CI)66.8 (64.2, 69.3)61.7 (59.0, 64.2)70.3 (67.2, 73.3)64.3 (61.1, 67.3)59.8 (55.2, 64.3)56.2 (51.4, 60.8)

During the study period, reoperation due to bleeding complications was required in seven patients in the ximelagatran group and five patients in the enoxaparin group (Table 5). Wound infection occurred in 27 ximelagatran-treated patients and 31 enoxaparin-treated patients. Overall, six patients died: two during the treatment period (one in the ximelagatran group died from bleeding due to iatrogenic injury to iliac/femoral vessels at the beginning of surgery and one in the enoxaparin group died from myocardial infarction on postoperative Day 2) and four patients died during follow-up, all in the ximelagatran group [hemorrhagic gastric ulcer (postoperative Day 20), unconfirmed PE (two cases, postoperative Day 14 and Day 15) and aspiration asphyxia (postoperative Day 23)]. There were no differences in the incidence of any other adverse events between the two treatment groups.

Table 5.  Surgically related complications during the study including the follow-up period
 Ximelagatran groupEnoxaparin group
  • *

    The number of patients with events/total number of patients.

Wound infection, n* (%)27/1377 (2.0)31/1387 (2.2)
Wound hematoma, n* (%)24/1377 (1.7)18/1387 (1.3)
Hip joint dislocation, n* (%)15/1377 (1.1)22/1387 (1.6)
Wound dehiscence, n* (%)5/1377 (0.1)8/1387 (0.1)
Wound rupture, n* (%)2/1377 (0.1)3/1387 (0.1)
Delayed wound healing, n* (%)1/1377 (0.1)1/1387 (0.1)
Fistula, n* (%)0/1377 (0.0)1/1387 (0.1)
Reoperation (all causes), n* (%)27/1365 (2.0)31/1380 (2.3)
 Hip joint dislocation9/1365 (0.7)14/1380 (1.0)
 Infection4/1365 (0.3)7/1380 (0.5)
 Other7/1365 (0.6)5/1380 (0.4)
Reoperation (bleeding-related causes), n* (%)7/1365 (0.5)5/1380 (0.4)
 Hematoma4/1365 (0.3)5/1380 (0.4)
 Hematoma and infection3/1365 (0.2)0/1380 (0.0)


In patients undergoing THR or TKR, preoperatively initiated s.c. melagatran followed by oral ximelagatran was associated with a significantly lower rate of both major and total VTE compared with enoxaparin. This was demonstrated for both THR and TKR. Although it is acknowledged that DVT is more resistant to thromboprophylaxis after THR than after TKR [1,2], the relative risk reduction was similar in both types of surgery. Symptomatic VTE was rare. During the study treatment period, numerically fewer patients in the ximelagatran group experienced a symptomatic event, while during the follow-up period the number of such events was similar in both treatment groups. It should be noted, however, that 71% of patients continued on LMWHs during the follow-up period irrespective of previous treatment group allocation.

Recent data indicate that the benefit:risk ratio of LMWHs in patients undergoing joint-replacement procedures is influenced by the timing of their initiation relative to surgery [12]. Two recent trials with melagatran/ximelagatran [10,11] and the present study also suggest this. A combined regimen of s.c. melagatran and oral ximelagatran showed efficacy similar to that of enoxaparin in preventing VTE when initiated 4–12 h postoperatively [11], whereas they were more effective than dalteparin [10] and enoxaparin (the present study) when started immediately before surgery, i.e. ‘knife-to-skin’. These results are consistent with those found with s.c. recombinant hirudin, another direct thrombin inhibitor [19,20].

In comparison with enoxaparin, this dosage regimen of melagatran/ximelagatran was associated with more surgery-related bleeding in patients undergoing THR but not in patients undergoing TKR. In the latter type of surgery, an arterial tourniquet was applied in 91% of patients in each treatment group. Importantly, the two treatment groups did not differ with respect to fatal bleeding, bleeding involving a critical organ and bleeding requiring reoperation. The overall difference in bleeding was not reflected by an increase in postoperative surgical site-related complications, such as wound hematoma with or without wound infection.

In recent trials in patients undergoing joint-replacement surgery, specific anticoagulant agents, which inhibit a single coagulation factor, such as thrombin [19,20] or factor Xa [21–24], were more effective than LMWHs in preventing VTE. Our study, performed with a direct thrombin inhibitor, confirms these findings. While other novel agents must be administered as s.c. injections, this study drug can be used in both parenteral (melagatran) and oral (ximelagatran) forms. In the present study, 94% of the subjects began oral therapy within the first postoperative day. This oral agent, used without coagulation monitoring, is of great interest, particularly in view of the potential for improved patient acceptability and compliance, the trend toward shorter postoperative hospitalization and out-of-hospital thromboprophylaxis in joint-replacement surgery [1].


We thank the coinvestigators and research staff at the centers and the study monitors in the countries involved. This study was supported by a research grant from AstraZeneca, Mölndal, Sweden. AstraZeneca sponsored EXPRESS and provided all study medication and materials.

Conflict of interest statement

Some members of the steering committee received travel grants or honoraria from AstraZeneca, Mölndal, Sweden (B.I.E., G.A., A.T.C., O.E.D., M.R.L., P.M., N.R., P.K.); S.P., C.E. and M.A. are AstraZeneca employees. There are no other conflicts of interest in relation to this study.


All authors, except P.K., were members of the steering committee that designed the study and supervised and reviewed the progress of the trial. All authors were members of the writing committee that interpreted the data, wrote the report and decided to submit the paper for publication. B.I.E. chaired the steering committee. B.I.E., G.A., A.T.C., O.E.D., M.R.L., P.M. and N.R. recruited patients. P.K. was responsible for the central review of all venograms. M.A. performed the statistical analyses.


The EXPRESS Study Group consisted of the following members. Steering Committee: B. I. Eriksson (Chairman), G. Agnelli, M. Andersson, A. T. Cohen, O. E. Dahl, C. Eskilson, A. Freij, M. R. Lassen, P. Mouret, S. Panfilov, N. Rosencher. Writing Committee: B. I. Eriksson (Chairman), G. Agnelli, M. Andersson, A. T. Cohen, O. E. Dahl, C. Eskilson, A. Freij, P. Kälebo, M. R. Lassen, P. Mouret, S. Panfilov, N. Rosencher. Safety Committee: A. Planes, K. Svärdsudd, H. Wedel. Venography Central Reading Committee: B. Zachrisson, P. Kälebo. Outcome Adjudication Committee: U. Angerås, H. Eriksson, P. Kälebo, G. Sandgren, J. Wallin, B. Zachrisson. Statistical Analysis: M. Andersson. Centres—Austria (225 patients): C. Wurmig (Wien), H. Niessner (Wiener Neustadt), K. Zhuber (Wels); Belgium (52 patients): J. Bellemans (Pellenberg), J. P. Simon (Pellenberg); Denmark (370 patients): P. Seest Jørgensen (Hellerup), B. Rønno, W. Duus (Copenhagen), A. Borgwardt (Fredriksberg), M. Rud Lassen (Hillerød), E. Hørlyck (Silkeborg), H. Schwarz Lausten (Herlev); Finland (128 patients): H. von Bonsdorff (Jyväskylä), O. Suomalainen (Kuopio), M. Hämäläinen (Oulu); France (118 patients): N. Rosencher (Paris), V. Souron (Annecy), D. Baylot (Saint-Etienne), E. Gaertner (Strasbourg), N. Vochelle (La Rochelle), M. Delecroix (Lille), J. J. Pinson (Saint Hernlain Cedex); Germany (547 patients): P. Mouret (Frankfurt), W. Birkner (Rheinfelden), H. Schmelz (Bad Mergentheim), G. Salzman (Wiesbaden), A. Halder (Sommerfeld), U. Schietsch (Halle), A. Kurth (Frankfurt), C. Hendrich (Würzburg), H. M. Fritsche (Garmisch-Partenkirchen); Hungary (134 patients): T. Mészáros (Szeged), K. Tóth (Kesckemét), A. Sárváry (Budapest), E. Sántha (Székesfehérvár), G. Dósa (Gyula); Italy (123 patients): F. Sonaglia (Perugia), P. Parize (Gubbio), M. Silingardi (Reggio Emilia), F. Piovella (Pavia), D. Biagini (Perugia), V. Mera (Varese); Norway (398 patients): O. Dahl (Oslo), J. H. Solhaug (Olslo), V. Finsen (Trondheim), P. Borgen (Baerum), V. Punsvik (Aalesund), T. Bjørang (Toensberg), P. Ståhl (Bodoe), E. Nesse (Kongsvinger), J. Midjord (Moss); Poland (290 patients): K. Modrzewski (Lublin), M. Synder (Lodz), S. Czyrny (Tarnow), J. Skowronski (Bialystok), A. Zygmunt (Rzeszow), A. Widawski (Krakow), S. Tkaczyk (Bydgoszcz); South Africa (79 patients): R. McLennan-Smith (Durban), S. Cunningham (Johannesburg), D. Adler (Johannesburg); Sweden (210 patients): B. I. Eriksson (Göteborg), B. Edshage (Kungälv), A. Folestad (Mölndal), S. Ponzer (Stockholm), S. Lind (Jönköping), P. Hansson (Halmstad), H. Tropp (Norrköping), L. Ahnfelt (Uddevalla), C. Andersson (Linköping), L. G. Petersson (Kalmar); United Kingdom (90 patients): A. T. Cohen (London), P. Grigoris (Glasgow), K. Miligan (Belfast).