Rivaroxaban versus enoxaparin after total knee arthroplasty – Authors’ reply

The RECORD programme compared rivaroxaban 10 mg once daily with either the EU or North American approved regimens of enoxaparin (40 mg once daily or 30 mg twice daily). As Nina Raju and colleagues note, factors such as pharmacological target and differences in dose and frequency, as well as timing of initial postoperative drug administration, can affect the safety or efficacy of anticoagulants. However, RECORD4 and the RECORD programme as a whole were not designed to compare different enoxaparin regimens or to provide indirect comparisons on the efficacy of new anticoagulants.

T Stief raises an interesting point regarding enoxaparin dose and the potential undertreatment of patients with prothrombotic conditions. Although the same concerns would apply to any anticoagulant, assessment of an unapproved, investigational, high-dose enoxaparin regimen was beyond the scope of this phase 3 clinical trial. Therefore, it is not possible to draw conclusions about the relative efficacy of such investigational regimens.

The RECORD4 trial, like other phase 3 trials of anticoagulants, was not designed or intended to provide long-term data on surgical outcomes after knee arthroplasty. However, the study did provide results on endpoints that can affect long-term outcomes, such as major bleeding, haemorrhagic wound complications, and postoperative wound infections and wound drainage, and showed low incidences with both drugs.

Strategies for reversal of the anticoagulant effect should be considered for every anticoagulant. However, specific reversal of anticoagulants with relatively short half-lives, such as rivaroxaban, is usually not necessary in clinical practice. Low-molecular-weight heparins also have no specific antidote and have been used extensively for the prevention and treatment of thromboembolic disorders for more than 20 years. If rivaroxaban discontinuation or delay of next dose, or appropriate symptomatic treatment fail to control bleeding, recombinant activated factor VII might be effective.

However, as noted by Umile Longo and colleagues, the risks and benefits of any new anticoagulant can only be fully assessed by long-term clinical monitoring, as is the case for any newly approved medication.

As Antonio Gomez-Outes and colleagues note, the relatively low rates of major bleeding in RECORD4 can be attributed to the definition used, which excluded surgical-site bleeding events associated with a fall in haemoglobin of 20 g/L or more, or requiring transfusion of two or more units of blood; this definition was adopted across the RECORD programme to allow a better assessment of clinically important bleeding events, as opposed to those that are an expected result of the arthroplasty procedure.

We generally agree with Gomez-Outes and colleagues that non-significant results do not imply equality of the treatment groups. On the other hand, the RECORD4 case scenarios presented by Gomez-Outes and colleagues implicitly assume that the relative risk of 2·47, which they calculated using the RECORD major bleeding definition, also applies when using the broader definition of major bleeding (ie, including surgical-site events associated with a fall in haemoglobin of 20 g/L or more, or requiring transfusion of two or more units of blood) used in other contemporary studies.

In fact, in RECORD4, the actual hazard ratio calculated using the broader major bleeding definition that included surgical-site bleeding was 1·67 (95% CI 0·90–3·10).

Furthermore, the US Food and Drug Administration (FDA) ancillary analyses referenced by Gomez-Outes and colleagues used a covariate adjustment (for active treatment duration) that resulted in an overestimate of the rivaroxaban treatment effect on bleeding due to confounding of the treatment group with active treatment duration in RECORD2. This approach was extensively discussed during the FDA Advisory Committee meeting (March 19, 2009) and queried by the independent biostatistician on the FDA Advisory Committee panel.

It should also be noted that, in the sponsor’s briefing document, additional pooled analyses are presented that take into account the different treatment durations of rivaroxaban and enoxaparin in RECORD2.

These results were consistent with the primary analysis results as reported by the sponsor.

AGGT receive honoraria as a member of the steering committee and has served as a consultant to Bayer Schering Pharma. AB and FM are employees of Bayer Schering Pharma.