In this international, non-inferiority, randomized, parallel-group, double-blind trial, we compared rivaroxaban with enoxaparin in patients undergoing non-major orthopedic surgery of the lower limb. The trial was sponsored by the Center Hospitalier Universitaire de Saint-Etienne, France, and by Bayer. The protocol (including statistical analysis plan), available with the full text of this article on NEJM.org, was developed by the authors and approved by relevant regulatory authorities and ethics committees. The steering committee had overall scientific responsibility for the trial, which was managed by contract research organization PSNResearch. An independent data and safety monitoring committee monitored the safety and efficacy data. Analyzes were performed independently by the academic statistician. Bayer had no role in the design or conduct of the trial; the collection, management, analysis or interpretation of data; the preparation or approval of the manuscript; or the decision to submit the manuscript for publication. Assistance with medical writing was funded by the Center Hospitalier Universitaire de Saint-Etienne. The authors vouch for the completeness and accuracy of the data and the fidelity of the trial to the protocol.
Adults who had been admitted to hospital for non-major lower extremity orthopedic surgery and were scheduled to receive thromboprophylaxis for at least 2 weeks (based on the investigator’s assessment of the patient’s venous thromboembolism risk) were eligible for the trial registration. Types of surgery included Achilles tendon repair; knee surgery (including unicompartmental knee prosthesis); surgery involving the tibial plateau or femur (excluding femoral head or neck fractures); tibial or ankle fractures or tibial osteotomy; transposition of the tibial tuberosity; arthrodesis of the knee, ankle or hindfoot; knee ligament repair with planned immobilization or partial weight bearing; ankle ligament repair; or any elective lower extremity orthopedic surgery requiring the use of thromboprophylaxis for more than 2 weeks. Enrollment criteria are outlined in the Supplemental Appendix, available at NEJM.org. All participants provided written informed consent.
Pre-randomization treatment with low molecular weight heparin was allowed for a maximum of 48 hours before surgery (maximum one dose per 24 hours). Randomization (in randomly permuted blocks of four) was performed within 10 hours of surgery and was done centrally in a 1:1 ratio with the use of an interactive web-based response system (ClinInfo) that assigned a unique randomization number to each eligible patient. Randomization was stratified by center and planned treatment duration (2 weeks to 1 month, >1 month to 2 months, or >2 months). The planned duration of treatment was based on medical judgment and aligned with the planned duration of immobilization (cast or recommendation of no weightbearing or partial weightbearing) and country-specific recommendations for the prevention of venous thromboembolism in adults undergoing orthopedic surgery.2.4
Patients randomized to the rivaroxaban group were to receive 10 mg of rivaroxaban orally and a subcutaneous injection of placebo (instead of enoxaparin); patients randomly assigned to the enoxaparin group were to receive a subcutaneous injection of enoxaparin (at a dose of 40 mg [4000 IU of anti-Xa activity] in 0.4 ml of diluent) and an oral tablet of placebo (instead of rivaroxaban) (Fig. S1 in the Supplementary Appendix). The test drug and the corresponding placebo were administered once daily every 24 hours in a window of ± 2 hours. Provided hemostasis was established, the first dose of test drug was given between 6 and 10 hours after surgery if it could be given before 22 hours and at least 24 hours after any preoperative administration of heparin low molecular weight. If the first dose could not be administered by 10 p.m., a postoperative dose of low molecular weight heparin was allowed and the administration of the first dose of the test drug was postponed to the following day. Lists of concomitant medications that were or were not permitted during the trial are provided in the Supplementary Appendix.
Upon discharge, patients received sufficient trial medication for the planned duration of treatment (i.e., until the end of immobilization). All patients underwent routine compression ultrasound at the end of immobilization (i.e. between 15 days and 3 months after randomization) to detect asymptomatic proximal deep vein thrombosis (see Evaluation of compression ultrasound in the supplementary appendix). Patients were contacted by telephone 30 days (within a window of ± 7 days) after the end of treatment to assess the occurrence of venous thromboembolic events.
Symptomatic venous thromboembolic events should be confirmed by objective testing – i.e. compression ultrasound for deep vein thrombosis and CT pulmonary angiography, ventilation-perfusion lung scan or pulmonary angiography for embolism pulmonary. Fatal pulmonary embolism was confirmed at autopsy or was imputed in cases of unexplained death when pulmonary embolism could not be ruled out.
The primary efficacy outcome for major venous thromboembolism was a composite of symptomatic distal or proximal deep vein thrombosis, pulmonary embolism or death related to venous thromboembolism during the treatment period or asymptomatic proximal deep vein thrombosis at the end of treatment. of the treatment. The prespecified secondary endpoints were the safety criteria for major bleeding (fatal, critical or clinically manifest bleeding or bleeding at the surgical site leading to the intervention11), clinically relevant non-major bleeding, overt thrombocytopenia, and death from any cause. Full definitions are provided in the Endpoints section of the Supplementary Appendix. All suspected thrombotic or hemorrhagic events and deaths were adjudicated by an independent central committee whose members were unaware of treatment assignments. An additional post-hoc analysis compared the composite of venous thromboembolism or major bleeding between groups (termed ‘net clinical benefit’).
To determine the non-inferiority of rivaroxaban compared to enoxaparin, the primary analysis was performed in the intention-to-treat population (all patients who were randomized) and in the per-protocol population (all patients meeting the criteria operated, received at least one dose of the test drug, and had no major protocol violation). In the primary intention-to-treat analysis, we used multiple imputation to account for missing data, as described in the Supplementary Appendix. The relative risk and its 95% confidence interval were estimated using a logistic regression model. The non-inferiority margin of the upper limit of the 95% confidence interval of the relative risk comparing rivaroxaban with enoxaparin was set at 1.30. We estimated that a sample size of 4400 patients would provide the trial with 90% power to show non-inferiority at a two-sided Type I error rate of 5% (see Supplementary Appendix).
The protocol specified that if non-inferiority was demonstrated for the primary endpoint, a superiority test would then be performed. To show superiority, a two-tailed Fisher’s exact test at the 5% level of significance was performed, and the resulting P value was reported. Kaplan-Meier curves were constructed.
The hazard ratio for bleeding incidence and other outcomes between the rivaroxaban group and the enoxaparin group was analyzed using the same methods and population as for the primary analysis. Confidence intervals for secondary outcomes were not adjusted for multiple comparisons and therefore inferences drawn from these intervals may not be reproducible.
Statistical analyzes were performed using SAS software, version 9.4 (SAS Institute). The graphs were constructed using R software, version 3.6.0. Further details on the statistical analysis are provided in the supplementary appendix.