RECaST-Surv: A Calibrated Borrowing Method for Survival Endpoints in Unequal Randomized Trials

Abstract

Randomized trials with limited concurrent control information---including but not limited to unequal-randomization settings---can offer ethical and practical advantages, especially in pediatric and rare diseases, but they often lose power because fewer control patients are available for direct comparison. Borrowing information from external controls may improve efficiency, but can also inflate the Type I error rate when the external and trial populations are not sufficiently comparable. We propose RECaST-Surv, a Bayesian transfer-learning framework for time-to-event outcomes that extends the RECaST method to survival settings. The method learns a structural survival model from external control data and calibrates it to the concurrent control arm of the current trial through a Cauchy random effect. To improve frequentist operating characteristics, we further develop a bootstrap-based procedure to calibrate the testing rule for Type I error control. RECaST-Surv can accommodate multiple external datasets and requires only summary-level information from external sources. Simulation studies show that the method maintains near-nominal Type I error across challenging settings while improving power by roughly 10%--12% over standard analyses. In an Amyotrophic Lateral Sclerosis trial emulation, RECaST-Surv increased power from 82.8% to 95.7% relative to the no-borrowing RCT analysis, while maintaining acceptable error control.

Evaluating prognostic score adjustment and information borrowing in randomized trials with binary endpoints
Zhengfan Wang, Craig S. Parzynski, Aaron Crowley +2
Contemporary clinical trials communications7 Oct 2026Peer-reviewedCC BY-NC-NDFull textStatisticsEurope PMCEurope PMCPMC
Interpreting Hierarchical Composite Endpoints with Survival and Longitudinal Outcomes: Application to Amyotrophic Lateral Sclerosis Trials
Arlina Shen, Dehua Bi, Ruben P. A. van Eijk +2
arXiv23 Sep 2026PreprintCC BY 4.0Full textStatisticsarXiv
Bayesian Methods Incorporating Baseline Covariates for Borrowing Historical Control Data in Randomized Clinical Trials: A Neutral Comparison Study.
David Jesse, Diane Uschner, Markus C Elze +2
Statistics in medicine30 Oct 2026Peer-reviewedCC BYFull textStatisticsEurope PMC
Selective inference for pruned interaction trees with known unequal variances in randomized trials
Tomoshige Nakamura
arXiv1 Nov 2026PreprintarXiv licenceFull textStatisticsarXiv
Calibrated Transfer Learning for Survival Prediction and Treatment Inference
Shuoyang Wang, Wanyu Zhang, Lingli Yang +1
arXiv15 Oct 2026PreprintarXiv licenceFull textStatisticsarXiv

Cite

BibTeX RIS

BibTeX

@misc{bi2026recastsurv,
  title = {{RECaST-Surv: A Calibrated Borrowing Method for Survival Endpoints in Unequal Randomized Trials}},
  author = {Dehua Bi and Arlina Shen and Ruben P. A. van Eijk and Lu Tian and Jiapeng Xu and Guillemette de la Borderie and Nate Bennet and Sarno Maria and Ying Lu},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.19109},
  eprint = {2609.19109},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.19109},
  note = {CC BY 4.0},
}

RIS

TY  - UNPB
TI  - RECaST-Surv: A Calibrated Borrowing Method for Survival Endpoints in Unequal Randomized Trials
AU  - Dehua Bi
AU  - Arlina Shen
AU  - Ruben P. A. van Eijk
AU  - Lu Tian
AU  - Jiapeng Xu
AU  - Guillemette de la Borderie
AU  - Nate Bennet
AU  - Sarno Maria
AU  - Ying Lu
PY  - 2026
DA  - 2026/09/16
PB  - arXiv
DO  - 10.48550/arxiv.2609.19109
UR  - https://arxiv.org/abs/2609.19109
AB  - Randomized trials with limited concurrent control information---including but not limited to unequal-randomization settings---can offer ethical and practical advantages, especially in pediatric and rare diseases, but they often lose power because fewer control patients are available for direct comparison. Borrowing information from external controls may improve efficiency, but can also inflate the Type I error rate when the external and trial populations are not sufficiently comparable. We propose RECaST-Surv, a Bayesian transfer-learning framework for time-to-event outcomes that extends the RECaST method to survival settings. The method learns a structural survival model from external control data and calibrates it to the concurrent control arm of the current trial through a Cauchy random effect. To improve frequentist operating characteristics, we further develop a bootstrap-based procedure to calibrate the testing rule for Type I error control. RECaST-Surv can accommodate multiple external datasets and requires only summary-level information from external sources. Simulation studies show that the method maintains near-nominal Type I error across challenging settings while improving power by roughly 10%--12% over standard analyses. In an Amyotrophic Lateral Sclerosis trial emulation, RECaST-Surv increased power from 82.8% to 95.7% relative to the no-borrowing RCT analysis, while maintaining acceptable error control.
ER  -

Abstract

Related

Cite