Febuxostat versus allopurinol: the hazard changes with time on treatment

Design

one-stage individual participant data meta-analysis of four studies, data reconstructed from published Kaplan-Meier curves, mixed-effects Cox model with treatment-by-time interaction

Population

19,090 patients with gout in randomised trials and propensity-score-matched cohorts, searched to November 2025

Primary outcome

time-varying cardiovascular risk and all-cause mortality, febuxostat versus allopurinol

Effect

cardiovascular HR 0.89 (0.79-0.99) at 0-12 months, 0.58 (0.50-0.68) at 12-24, 0.86 (0.74-0.99) at 24-48, 1.09 (0.81-1.46) beyond 48 months

Whether febuxostat carries a cardiovascular penalty against allopurinol has never settled, because the major trials disagreed. This analysis reconstructed individual patient data from published Kaplan-Meier curves in four studies — randomised trials and propensity-score-matched cohorts, 19,090 patients in total — and fitted a mixed-effects Cox model with a treatment-by-time interaction over pre-specified intervals.

The effect was not constant. Against allopurinol, febuxostat was associated with lower cardiovascular risk at 0-12 months (HR 0.89, 95% CI 0.79-0.99), markedly lower at 12-24 months (0.58, 0.50-0.68), lower again at 24-48 months (0.86, 0.74-0.99), and no different beyond 48 months (1.09, 0.81-1.46). All-cause mortality followed the same shape and further from 1.0: 0.70, then 0.39, then 0.75, then 1.02.

Two cautions before this changes anything. The patient-level data were reconstructed from published curves rather than obtained from the trials, and pooling randomised with propensity-matched observational studies imports confounding that randomisation was supposed to remove. A time-varying hazard can also be produced by depletion of susceptibles rather than by the drug. What the analysis does support is that a single hazard ratio averaged over years hides the shape of the risk, and that the fear of an early cardiovascular penalty with febuxostat is not what these pooled curves show.

• Do not withhold febuxostat from a patient who cannot tolerate or has failed allopurinol on cardiovascular grounds alone
• Record the reason for choosing either drug, and the date started — the risk profile is time-dependent
• Continue to treat cardiovascular risk factors in gout in their own right, whichever agent is used
• Febuxostat remains the practical option in significant chronic kidney disease and in allopurinol hypersensitivity
• In India, check HLA-B*5801 testing availability locally before starting allopurinol in at-risk groups rather than assuming it

The statistics, in plain English

A hazard ratio below 1.0 favours febuxostat. The 0-12 month interval (0.89, upper limit 0.99) only just clears 1.0, so that estimate is fragile; the 12-24 month figure (0.58, 0.50-0.68) is the only one clearly separated from no effect. The interval beyond 48 months (1.09, 0.81-1.46) crosses 1.0 in both directions, which means no detectable difference rather than proven equivalence, and it rests on the fewest patients still under follow-up. Reconstructing individual data from published curves recovers the shape of the survival experience but not the covariates, so adjustment is limited to what the original reports show.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.