Twenty-two RCTs were included in the review. Follow-up ranged from less than 14 days in nine trials to 90 days in one trial; 12 trials had follow-up of between 30 and 70 days.
Risks of bias varied across criteria. The most common source of bias was incomplete outcome data, which was an issue for at least one assessed outcome in all except three trials; data on DVT and pulmonary embolism were missing for between 3% and 41% of patients. Overall the evidence was considered to be moderate quality.
There were no statistically significant differences between factor Xa inhibitors and LMWH in mortality at the end of treatment (OR 1.27, 95% CI 0.63 to 2.55; Ι²=0%) or at the end of follow-up (OR 0.95, 95% CI 0.55 to 1.63; Ι²=43%). There was no statistically significant difference between treatments in nonfatal pulmonary embolism.
There was a statistically significant benefit of reduced symptomatic DVT in patients treated with factor Xa inhibitors (OR 0.46, 95% CI 0.30 to 0.70; Ι² = 0%) which represented a reduction of three DVT (95% CI 1 to 5) events per 1,000 patients treated for one to five weeks. When the baseline risk from a large cohort study was used, factor Xa inhibitors were estimated to give a benefit of four (95% CI 3 to 6) fewer events per 1,000 treated patients.
Analyses of major bleeding events and bleeding leading to reoperation were not statistically significant but suggested the possibility of harm; analysis of absolute differences showed an increase of two (95% CI 0.98 to 1.65) major bleeding events per 1,000 patients treated for one to five weeks. Subgroup and regression analyses indicated that excess bleeding events resulted from high (OR 2.50, 95% CI 1.38 to 4.53) rather than low or intermediate doses of factor Xa inhibitors (no statistically significant difference) (test for interaction p=0.02).
There was asymmetry in the funnel plots for all outcomes except major bleeding but this may not indicate publication bias since the potentially missing studies would have favoured factor Xa inhibitors.