SAVI 2026 — Anytime-valid log-rank testing for randomised trials

Poster materials and supplementary content for the SAVI 2026 contribution.

This page hosts materials for the poster Anytime-valid log-rank testing for randomised trials, presented at SAVI 2026 (Sequential Anytime-Valid Inference).

Authors. Joren Brunekreef¹, Renée X. Menezes², Rianne de Heide³,⁴

¹ Netherlands Cancer Institute, Department of Radiotherapy and AI for Oncology, Amsterdam, NL · ² Netherlands Cancer Institute, Biostatistics Centre and Department of Psychosocial Research and Epidemiology, Amsterdam, NL · ³ University of Twente, Enschede, NL · ⁴ Centrum Wiskunde & Informatica, Amsterdam, NL

Poster

Download the poster (PDF)

(Poster file uploaded after the conference; placeholder until then.)

Summary

Anytime-valid (AV) log-rank tests continuously monitor survival trials with Type-I control under optional stopping and continuation. We characterise the fixed-δ AV log-rank by simulation, and deploy it retrospectively on a real-world randomised trial. Under a simple continuation rule — extend the trial when evidence at the classical end time is suggestive but not yet decisive — the AV total rejection rate matches and surpasses classical’s empirically realised power, at a bounded and visible cost.

Method

  • Fixed-δ AV log-rank. A test martingale on the event sequence using a single pre-specified alternative hazard ratio δ, following Ter Schure et al. (2011) Eq. (6).
  • Power-over-time. AV’s rejection probability accumulated through follow-up, generalising classical fixed-n power to a curve through time.
  • Continuation rule. At the classical trial end time, continue monitoring if the e-value exceeds a futility cutoff τ; otherwise stop. Cap continuation at an extended end time. Reject whenever the e-value crosses 1/α.

Real-world deployment

(Trial choice TBD — placeholder for the chosen retrospective.)

Additional materials

These sections will be filled in if time allows after the conference:

  • Code and reproducibility. Scripts that reproduce the simulation results and the retrospective analysis.
  • Extended figures. Per-scenario power-over-time curves; stopping-rule sensitivity sweeps; calendar-date reporting for the retrospective.
  • Notes and discussion. Conversations with poster attendees, follow-up questions, open issues.

References

The poster cites the safe-test / e-value / AV log-rank literature; the most directly relevant references are:

  • Ter Schure, J., Pérez-Ortiz, M.F., Ly, A., & Grünwald, P. (in press). The safe logrank test: error control under continuous monitoring with unlimited horizon.
  • Grünwald, P., de Heide, R., & Koolen, W.M. (2024). Safe testing. Journal of the Royal Statistical Society B.
  • Ramdas, A., Grünwald, P., Vovk, V., & Shafer, G. (2023). Game-theoretic statistics and safe anytime-valid inference. Statistical Science.

Contact

For questions or follow-ups, reach me at j.brunekreef@nki.nl.