Sixteen RCTs (38,747 participants) were included in the review. All studies except one scored the maximum of 5 points on the Jadad scale and were judged to be at low risk of bias. One study scored three points because it was open-label. Follow-up ranged from 35-69 days to 90-100 days.
Patients treated with rivaroxaban had a statistically significantly lower risk of symptomatic venous thromboembolism than those given enoxaparin (RR 0.48, 95% CI 0.31 to 0.75; eight trials, Ι²=5%). No such significant benefit was seen with dabigatran (RR 0.71, 95% CI 0.23 to 2.12; four trials, Ι²=73%) or apixaban (RR 0.82, 95% CI 0.41 to 1.64; four trials, Ι²=40%). The heterogeneity in the dabigatran analysis was not explicable in terms of dose or otherwise. Inclusion of events that occurred during follow-up did not materially alter the results of the analyses.
The results of the secondary analyses that assessed impact on DVT and pulmonary embolism separately gave a more complex picture. Dabigatran was not associated with significant impact on either outcome but both rivaroxaban and apixaban reduced the rate of DVT but not of pulmonary embolism. Apixaban was associated with a numerical increase in cases of pulmonary embolism.
The impact on composite outcomes that included all-cause or thromboembolism-related mortality was similarly complex. Significant benefits were seen for rivaroxaban on both outcomes and for apixaban on the all-cause mortality composite outcome. Dabigatran showed trends towards higher rates of these composite outcomes that were significant for the 150mg dose on the all-cause mortality composite outcome. Full results were reported in the paper.
Rivaroxaban was associated with a statistically significantly increased risk of clinically relevant bleeding compared to enoxaparin (RR 1.25, 95% CI 1.05 to 1.49; eight trials, Ι²=5%). There was no difference between enoxaparin and dabigatran at either dose. There was a statistically significantly lower risk of clinically relevant bleeding with apixaban compared to enoxaparin (RR 0.82, 95% CI 0.69 to 0.98; four trials, Ι²=3%). Other secondary outcomes were reported in the review.
Indirect comparisons between the three new anticoagulants showed that rivaroxaban was associated with the lowest risk for symptomatic venous thromboembolism and apixaban had the lowest risk for clinically relevant bleeding. Net clinical outcomes did not differ between the treatments.
There was no evidence of publication bias. Sensitivity analyses did not materially effect the results of the direct comparison analyses except that author-defined criteria for major bleeding events made rivaroxaban appear statistically significantly superior to enoxaparin. Subgroup analyses showed no difference between knee and hip surgeries. A fixed-effect model made some of the differences between new anticoagulants in the indirect comparison statistically significant.