WHAT THE STUDY ACTUALLY SAYS

Aspirin was not as good as anticoagulants at preventing clots after joint replacement

In an 18,883-patient trial, aspirin failed a non-inferiority test against warfarin and rivaroxaban for preventing clots — but the anticoagulants in turn were not safer than aspirin on bleeding.

Death or symptomatic venous thromboembolism through 6 months (per-protocol)Aspirin 81 mg twice daily: 1.26%; Warfarin (INR 2.0): 1.23%; Rivaroxaban 10 mg daily: 0.94%0%1%2%Aspirin 81 mg twice daily1.26%Warfarin (INR 2.0)1.23%Rivaroxaban 10 mg daily0.94%
Death or symptomatic venous thromboembolism through 6 months (per-protocol)
GroupValue (%)
Aspirin 81 mg twice daily1.26
Warfarin (INR 2.0)1.23
Rivaroxaban 10 mg daily0.94
Death or symptomatic venous thromboembolism through 6 months (per-protocol) Primary effectiveness event rate in each drug group; there was no placebo arm [s1]. Source: The New England Journal of Medicine

Low-dose aspirin is widely used to prevent blood clots after hip and knee replacement, on the view that it is cheap, simple and roughly as effective as prescription anticoagulants. A large head-to-head trial published in The New England Journal of Medicine now complicates that view: aspirin was not as effective as warfarin or rivaroxaban at preventing clots — but the anticoagulants were not safer than aspirin either [s1].

What the trial did

PEPPER was a comparative-effectiveness, non-inferiority trial in patients undergoing elective hip or knee arthroplasty [s1]. They were randomly assigned to one of three regimens for 30 days after surgery: aspirin at 81 mg twice daily, warfarin dosed to an international normalised ratio of 2.0, or rivaroxaban at 10 mg once daily [s2]. The composite primary effectiveness outcome was death from any cause or symptomatic venous thromboembolism, and the primary safety outcome combined bleeding, reoperation for wound problems, and myocardial infarction; both were assessed through 6 months in a blinded fashion [s1]. The pre-specified non-inferiority margin for the risk difference was 0.69 percentage points [s1].

A total of 18,883 participants underwent randomisation, of whom 15,317 — 7,214 with a hip replacement and 8,103 with a knee replacement — were included in the primary per-protocol analysis [s1].

What it found

The clotting events were uncommon in all three groups. A primary effectiveness event occurred in 77 of 6,090 participants (1.26%) in the aspirin group, 52 of 4,231 (1.23%) in the warfarin group, and 47 of 4,996 (0.94%) in the rivaroxaban group [s1].

The question the trial was built to answer was not whether those numbers look close, but whether aspirin could be declared no worse than the anticoagulants within the 0.69-point margin. It could not. In the pairwise comparisons, aspirin ran 0.30 percentage points above rivaroxaban (1.24% versus 0.94%; 95% confidence interval −0.20 to 0.81) and 0.23 points above warfarin (1.24% versus 1.01%; 95% CI −0.34 to 0.81), while warfarin and rivaroxaban were close to each other (1.13% versus 1.10%; difference 0.03 points, 95% CI −0.57 to 0.63) [s1]. Because the upper bounds of those intervals exceeded the margin, aspirin was not shown to be non-inferior to either drug for preventing death or symptomatic clots [s1].

Safety ran the other way. A primary safety event occurred in 177 aspirin patients (2.91%), 142 warfarin patients (3.36%), and 174 rivaroxaban patients (3.48%) [s1]. In the pairwise safety comparisons, rivaroxaban was 0.68 points worse than aspirin (3.70% versus 3.02%; 95% CI −0.04 to 1.41) and warfarin 0.77 points worse (3.65% versus 2.88%; 95% CI −0.04 to 1.58), while rivaroxaban and warfarin were indistinguishable (3.59% versus 3.61%; difference −0.02 points, 95% CI −0.89 to 0.85) [s1]. Here too, neither anticoagulant could be declared non-inferior to aspirin [s1]. Intention-to-treat analyses gave similar results [s1].

How to read it

The clean interpretation is a trade-off rather than a winner. Aspirin was not as good as the anticoagulants at preventing clots, and the anticoagulants were not as safe as aspirin on the bleeding-and-wound composite [s1]. Both of those are statements about a formal non-inferiority test, which asks whether a difference can be ruled out within a pre-set margin — not whether the groups differed by a clinically dramatic amount. The absolute event rates were low across the board, which is worth holding in mind alongside the statistical conclusions [s1].

A few limits frame it. The trial tested specific regimens — twice-daily 81 mg aspirin, warfarin to an INR of 2.0, and 10 mg rivaroxaban — so it speaks to those, not to every dose or drug a surgeon might choose [s1]. It was funded by the Patient-Centered Outcomes Research Institute, a public body, rather than a drug maker [s1]. And it combined hip and knee patients, whose baseline clot risk differs; the overall result does not tell an individual patient which way their own balance of clot risk against bleeding risk tips.

What to watch

The practical question clinicians will press on is whether the modest effectiveness gap with anticoagulants outweighs their modest safety cost for a given patient, and whether guideline committees read PEPPER as a reason to narrow aspirin's role or to keep it as a reasonable option for lower-risk patients. For now, the finding is specific and symmetric: after hip or knee replacement, aspirin prevented fewer clots than warfarin or rivaroxaban, and those drugs caused more of the trial's safety events than aspirin did [s1].

This article describes clinical-trial results and is not medical advice. Decisions about clot prevention after surgery should be made with the treating clinician.

Sources

  1. Prevention of Venous Thromboembolism after Total Hip or Knee Replacement — The New England Journal of Medicine , November 12, 2026
  2. Comparative Effectiveness of Pulmonary Embolism Prevention after Hip and Knee Replacement (PEPPER) (NCT02810704) — ClinicalTrials.gov

More on

Related coverage