Oral dabigatran etexilate versus enoxaparin for venous thromboembolism prevention after total hip arthroplasty: pooled analysis of two phase 3 randomized trials

RE-NOVATE and RE-NOVATE II were prospective, double-blind, double-dummy, randomized, multicentre, non-inferiority studies.

Participants were adults aged at least 18 years who were scheduled for primary elective total hip arthroplasty. The two trials had identical study eligibility criteria and were designed to be as similar as possible. Briefly, patients were randomized to treatment with oral dabigatran 220 mg or 150 mg once daily (the latter dose was not used in RE-NOVATE II), or subcutaneous enoxaparin 40 mg once daily started the evening before surgery; in some countries enoxaparin treatment was started postoperatively in accordance with local practice.

The first dose of dabigatran was halved and given 1–4 h after wound closure, provided clinical assessment of perioperative and postoperative bleeding and drainage indicated adequate hemostasis. If administration was delayed until the day after surgery, a full dose was given, followed by a second dose at least 12 h later. Treatment was continued until mandatory bilateral venography at 28–35 days. In both trials, the treatment period was defined as the time from first dose to 3 days after the last dose. Continued VTE prophylaxis was at the discretion of the treating physician. Patients attended a clinical follow-up visit 3 months after surgery. Concomitant administration of low dose aspirin (<160 mg) and selective cycloxygenase-2 inhibitors was allowed during treatment. Elastic compression stockings were permitted, but intermittent pneumatic compression devices were prohibited.

Both studies were approved by National Independent Ethics Committees and conducted according to the Declaration of Helsinki (October 1996 version). All patients gave signed informed consent prior to entry.

The prespecified primary efficacy endpoint of this pooled analysis was identical with that of the individual trials, i.e., the composite of total VTE and all-cause mortality. The endpoint was analyzed in the modified intention-to-treat (mITT) population, comprising all randomized and treated patients who underwent elective total hip arthroplasty and had evaluable adjudicated data on VTE (venographic confirmation in both legs or symptomatic event) or died during the treatment period. The main secondary efficacy outcome was the composite of major VTE (venographic or symptomatic proximal DVT and/or PE) and VTE-related mortality during treatment. Additional predefined secondary efficacy outcomes during the treatment period included total DVT (venographic or symptomatic), proximal DVT (venographic or symptomatic), and symptomatic DVT and/or PE.

All efficacy endpoints were based on assessments made by the same blinded Independent Venous Thromboembolic Event Adjudication Committee. Mandatory bilateral venography was performed within 24 h of the last oral dose, as described previously [3]. Suspected symptomatic DVT during treatment or follow-up was confirmed by ultrasound or venography. Symptoms suggestive of PE mandated confirmation by ventilation-perfusion scintigraphy, pulmonary angiography or spiral chest computer tomography, depending on local center preference. Deaths were considered related to VTE if they were categorized as “VTE related” or “unexplained" by the Independent Adjudication Committee.

All randomized patients who received at least one dose of study treatment were evaluable for safety. The main safety endpoint was the frequency of major bleeding events (which, different to previously performed studies, also included bleeding from the surgery wound site) occurring between intake of the first dose of study medication and 3 days after the last dose. Secondary safety outcomes included the composite of major and clinically relevant non-major bleeding events, any bleeding events during treatment, liver enzyme elevations (≥3 x the upper limit of the normal reference range (ULN) for serum alanine aminotransferase [ALT]) and acute coronary events (defined as confirmed unstable angina, myocardial infarction [MI], and cardiac death). This is in line with a number of pooled analyses from total knee and hip replacement trials with dabigatran, which included more than 8,000 patients [4, 5]. Major, clinically relevant, non-major and minor bleeding events were defined according to accepted guidelines [6], as reported previously [3]. In particular, the definition of major bleeding includes wound site bleeding events, in accordance with recommended guidelines [6]. ALT elevations ≥3 x ULN and any suspected acute coronary syndrome events were reviewed by an Independent Committee who were blinded to treatment allocation. An assessment of causality was provided for each of the reviewed patient cases.

The statistical analysis plan described here was planned before commencement of the RE-NOVATE II trial in accordance with regulatory recommendations [7].

For each trial, the difference in the proportion of patients with an event (efficacy or safety) was compared between dabigatran 220 mg and enoxaparin 40 mg as a risk difference (RD) as this was considered the most clinically meaningful measure. A common RD estimate across the two trials was calculated using a fixed-effects model (maximum likelihood estimation) [8], based on inverse variance weights for combined results from the individual trials, and compared with results obtained with a random effects model (DerSimonian and Laird method). The RD was not expected to differ between the studies. Heterogeneity of the common RD between the studies was assessed using Cochran’s χ2 and the I² statistic; p <0.10 was considered to denote statistically significant heterogeneity and where I² was greater than 50 %, heterogeneity was considered substantial [9]. Pooled event rate data for each treatment group are presented alongside the common RD results obtained from the pooled analysis.

Subgroup analyses were planned to investigate the influence of age, body weight, gender, and renal function (determined from calculated baseline creatinine clearance) on the incidence of the primary efficacy outcome, as well as bleeding events. For these comparisons the odds ratio and 95 % confidence intervals (CIs) for the analyzed subgroups were calculated using a fixed-effects model for the efficacy and safety endpoints.

Sensitivity analyses were conducted to explore the robustness of the results. The impact of missing or non-evaluable venography data, based on imputation of missing values using best and worst case scenarios (all treatment success or all treatment failure), was investigated to ensure that missing data did not affect the power of the trial or bias any estimation of the treatment effect.