All Content

10 days enoxaparin better than 35 days rivaroxaban for medical inpatient thromboprophylaxis (MAGELLAN)

Clinical question

Is rivaroxaban for 35 days better than enoxaparin for 10 days to prevent venous thromboembolism in medical inpatients?

Bottom line

Enoxaparin for 10 days provides similar protection to rivaroxaban for 35 days against symptomatic venous thromboembolism (VTE) or VTE-related death, and the extended use of rivaroxaban leads to an increase in clinically relevant and major bleeding. Rivaroxaban cannot be recommended for this indication. The larger question is whether we should be routinely anticoagulating these patients at all. Although the 2012 guidelines from the American College of Chest Physicians ( recommend prophylaxis for inpatients at increased risk for VTE, a recent American College of Physicians’ guideline calls this practice into question, noting that for every 4 pulmonary emboli prevented, you cause 9 major bleeding events (Ann Intern Med 2011;155:602). LOE = 1b


Cohen AT, Spiro TE, Büller HR, et al, for the MAGELLAN Investigators. Rivaroxaban for thromboprophylaxis in acutely ill medical patients. N Engl J Med 2013;368(6):513-523.

Study design

Randomized controlled trial (double-blinded)

Funding source





Inpatient (any location)


Patients admitted within 72 hours of an acute medical illness who were older than 40 years and who had reduced mobility were randomized to receive either enoxaparin 40 mg once daily for 10 days plus 35 days oral placebo or rivaroxaban 10 mg twice daily for 35 days plus 10 days of subcutaneous placebo. There were a total of 8101 patients in the study (average age = 71 years). Patients were hospitalized for infectious disease (45%), heart failure (32%), respiratory insufficiency (27%), stroke (17%) or active cancer (7%), and at least 1 day of immobilization was anticipated with decreased mobility for at least 4 days. This study was performed in 52 countries, many of which must keep their patients longer in the hospital than we do in the United States, as the median duration of hospitalization was a whopping 11 days. There was an extensive list of inclusion criteria, with patients having at least one risk factor for VTE and not having any obvious bleeding risks. Groups were balanced at the start of the study, analysis was by intention to treat, and outcomes were blindly adjudicated. Patients underwent ultrasound to detect asymptomatic deep vein thrombosis (DVT) at 10 days and 35 days, and underwent imaging to detect VTE if they were symptomatic at any time. The composite efficacy outcome was a combination of asymptomatic proximal DVT, symptomatic DVT or PE, and VTE-related death at 10 and 35 days, and the safety outcome was major or fatal bleeding at 10 and 35 days. Only approximately 75% of the patients are included in the efficacy outcome, because approximately one fourth in each group failed to have the follow-up ultrasound to detect asymptomatic DVT. Although the authors point to the superiority of rivaroxaban at 35 days (4.4% vs 5.7%; P = .02; number needed to treat [NNT] = 77), this is only because of a decrease in asymptomatic DVTs; that is, DVTs we would never have known about were it not for the mandated study ultrasound. There was no significant difference in the likelihood of symptomatic VTE or VTE-related death. Major bleeding was more common in the rivaroxaban group at 35 days (4.1% vs 1.7%; P < .001; number needed to treat to harm = 42). This includes 7 fatal bleeds in the rivaroxaban group and only 1 in the enoxaparin group. The authors do not perform statistical testing for fatal bleeds (and a number of other outcomes that appear unfavorable to rivaroxaban) but I did and found that it was statistically significant (two-tailed chi-square = 4.7; P = .03). All-cause mortality was similar between groups. The “net benefit” was the composite of the primary efficacy and primary safety outcomes and favors enoxaparin (7.8% vs 9.4%; P = .02; NNT = 62). Rather than stating the obvious (enoxaparin was superior to rivaroxaban for the net benefit outcome), the authors spin this by stating that “the prespecified analysis of net clinical benefit or harm did not show a benefit with rivaroxaban at either day 10 or 35.”

Comment on this Article

Your email address will not be published. Required fields are marked *