VTE is common after total hip (THR) or knee (TKR) replacement. Anticoagulants reduce the risk of VTE but result in bleeding, requiring assessment of benefit–risk.


To improve estimates of treatment effects on low–frequency events, data were pooled from 4 phase 3 clinical trials (RECORD1–4) of rivaroxaban vs enoxaparin regimens (or enoxaparin/placebo combination in 1 study) for the prevention of deep vein thrombosis and pulmonary embolism after THR and TKR. Studies differed in treatment duration and comparator dose, but pooling was supported by similar study designs, identical endpoints and event ascertainment methods and the same independent, central–blinded adjudication committee. Benefit–risk was assessed by comparing the excess number of outcome events for benefits vs that for harms over the treatment period. Excess number of events was defined as the number of events in a hypothetical population of 10,000 patients treated with enoxaparin minus the number of events in such a population treated with rivaroxaban. A positive value indicates that fewer events occur in patients treated with rivaroxaban. The analysis was undertaken for several clinically comparable pairs of composite benefit and harm outcomes (Table). Pooled Mantel–Haenszel weighted risk differences were used to compute excess numbers of benefit and harm events, and differences between excess numbers of events were used to evaluate net clinical benefit (NCB). Benefits and risks were weighted equally. An assessment was also performed using all treatment–emergent serious adverse events (SAEs) as reported by investigators.


Rivaroxaban is associated with significantly fewer total VTE, major VTE, and symptomatic VTE/all–cause mortality events than enoxaparin, whereas enoxaparin was associated with a smaller number of bleeding events, although only the composite of major + clinically relevant non–major bleeding was significantly different. In each comparison, excess number of bleeding events was less than excess number of VTE–related events by a factor of 4–10. Enoxaparin was associated with an excess of 194 treatment–emergent SAEs compared with rivaroxaban. In all cases, there was a positive NCB for rivaroxaban vs enoxaparin with 95% confidence intervals excluding 0, suggesting that the benefits of rivaroxaban exceed the risks of enoxaparin.


This quantitative benefit–risk approach compares interventions in a clinically relevant population. Using an NCB approach, for a variety of endpoints, benefits consistently exceed harms for rivaroxaban compared to enoxaparin after elective THR/TKR.

Taxonomy of Interventions to Prevent Adverse Events and Readmissions After Hospital Discharge

  Number (#) of events (in 10,000 patients) Excess # of events(E – R) NCB (difference in excess # of events)
  Rivaroxaban (R) Enoxaparin (E)
Outcome* N 95% CI N 95% CI N 95% CI N 95% CI
Total VTE 426 (364, 488) 943 (853, 1032) 504 (399, 608) 440 (320, 560)
Major and CRNM bleeding 319 (274, 363) 255 (215, 295) –64 (–123, –5)    
Major VTE 68 (44, 93) 274 (226, 321) 205 (153, 257) 187 (132, 243)
Major bleeding 39 (23, 55) 21 (9, 33) –18 (–37, 1)    
Symptomatic VTE & all–cause mortality 57 (38, 77) 132 (103, 161) 76 (42, 110) 58 (19, 97)
Major bleeding 39 (23, 55) 21 (9,33) –18 (–37, 1)    
Any treatment–emergent SAE 657 (594, 720) 852 (781, 923) 194 (101, 287)    
Total and major VTE are based on modified intent–to–treat population. Symptomatic VTE, all–cause deaths, bleeding events, and SAEs are based on safety population. Efficacy events are based on treatment duration period and bleeding events on treatment–emergent period. Mantel–Haenszel weighted difference.