Win statistics applied to registry-based randomized clinical trials
Rylance, Rebecca LU
; Wagner, Philippe
; Götberg, Matthias
LU
; Mohammad, Moman A.
LU
; Hofmann, Robin
; Fröbert, Ole
; James, Stefan
and Erlinge, David
LU
(2026)
In Trials
27(1).
- Abstract
Background: Win statistics offer an alternative approach to clinical trials that use survival analysis to analyze composite endpoints. Our objective was to re-analyze data from previously published registry-based randomized controlled trials that produced hazard ratios using win statistics to evaluate the correspondence between them. Good correspondence was defined as both results being positive, negative, or neutral. Win statistics were calculated for these trials to encourage transparency, scientific rigor, and possibly validate results. Methods: The win ratio ordered events hierarchically by clinical importance for each trial, with all-cause death regarded as most severe, followed by acute myocardial infarction. Further components,... (More)
Background: Win statistics offer an alternative approach to clinical trials that use survival analysis to analyze composite endpoints. Our objective was to re-analyze data from previously published registry-based randomized controlled trials that produced hazard ratios using win statistics to evaluate the correspondence between them. Good correspondence was defined as both results being positive, negative, or neutral. Win statistics were calculated for these trials to encourage transparency, scientific rigor, and possibly validate results. Methods: The win ratio ordered events hierarchically by clinical importance for each trial, with all-cause death regarded as most severe, followed by acute myocardial infarction. Further components, i.e., other endpoints, were added subsequently to the hierarchy. Each patient in the treatment group was compared with each patient in the control arm in hierarchical order. The total number of wins for each category in the treatment group was added and divided by the total number of wins in the control group. Win odds were calculated as an extension, which incorporate ties into the calculation. Results: The results using win statistics showed good correspondence to the previously reported hazard ratios with their composite endpoints: for the TASTE trial, the results were neutral for both the hazard ratio and win odds. The hazard ratio was 0.86 (0.67–1.10), and the win odds were 1.02 (0.99–1.04). Similar results were found for the iFR-SWEDEHEART, DETO2X-AMI, and VALIDATE trials. The IAMI trial showed better results for the vaccinated group compared to the placebo for both the hazard ratio and win odds. Conclusions: Win statistics offer an alternative approach to traditional survival analysis by harnessing multiple events hierarchically by clinical importance. Win statistics also offer the potential to evaluate a broader range of clinical endpoints, providing a more rounded perspective of treatment efficacy, an important consideration when designing future randomized controlled trials. This is the first multi-registry-based controlled trial reanalysis using win statistics.
(Less)
- author
- Rylance, Rebecca
LU
; Wagner, Philippe
; Götberg, Matthias
LU
; Mohammad, Moman A.
LU
; Hofmann, Robin
; Fröbert, Ole
; James, Stefan
and Erlinge, David
LU
- organization
- publishing date
- 2026-12
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Composite endpoints, Cox ph model, Hierarchical composite endpoints, Registry-based data, Win odds, Win ratio
- in
- Trials
- volume
- 27
- issue
- 1
- article number
- 279
- publisher
- BioMed Central (BMC)
- external identifiers
-
- pmid:41794768
- scopus:105035477000
- ISSN
- 1745-6215
- DOI
- 10.1186/s13063-026-09598-3
- language
- English
- LU publication?
- yes
- id
- 99abe5ce-35b7-4039-b83e-a7bd428e68be
- date added to LUP
- 2026-05-29 10:25:30
- date last changed
- 2026-06-12 11:56:33
@article{99abe5ce-35b7-4039-b83e-a7bd428e68be,
abstract = {{<p>Background: Win statistics offer an alternative approach to clinical trials that use survival analysis to analyze composite endpoints. Our objective was to re-analyze data from previously published registry-based randomized controlled trials that produced hazard ratios using win statistics to evaluate the correspondence between them. Good correspondence was defined as both results being positive, negative, or neutral. Win statistics were calculated for these trials to encourage transparency, scientific rigor, and possibly validate results. Methods: The win ratio ordered events hierarchically by clinical importance for each trial, with all-cause death regarded as most severe, followed by acute myocardial infarction. Further components, i.e., other endpoints, were added subsequently to the hierarchy. Each patient in the treatment group was compared with each patient in the control arm in hierarchical order. The total number of wins for each category in the treatment group was added and divided by the total number of wins in the control group. Win odds were calculated as an extension, which incorporate ties into the calculation. Results: The results using win statistics showed good correspondence to the previously reported hazard ratios with their composite endpoints: for the TASTE trial, the results were neutral for both the hazard ratio and win odds. The hazard ratio was 0.86 (0.67–1.10), and the win odds were 1.02 (0.99–1.04). Similar results were found for the iFR-SWEDEHEART, DETO2X-AMI, and VALIDATE trials. The IAMI trial showed better results for the vaccinated group compared to the placebo for both the hazard ratio and win odds. Conclusions: Win statistics offer an alternative approach to traditional survival analysis by harnessing multiple events hierarchically by clinical importance. Win statistics also offer the potential to evaluate a broader range of clinical endpoints, providing a more rounded perspective of treatment efficacy, an important consideration when designing future randomized controlled trials. This is the first multi-registry-based controlled trial reanalysis using win statistics.</p>}},
author = {{Rylance, Rebecca and Wagner, Philippe and Götberg, Matthias and Mohammad, Moman A. and Hofmann, Robin and Fröbert, Ole and James, Stefan and Erlinge, David}},
issn = {{1745-6215}},
keywords = {{Composite endpoints; Cox ph model; Hierarchical composite endpoints; Registry-based data; Win odds; Win ratio}},
language = {{eng}},
number = {{1}},
publisher = {{BioMed Central (BMC)}},
series = {{Trials}},
title = {{Win statistics applied to registry-based randomized clinical trials}},
url = {{http://dx.doi.org/10.1186/s13063-026-09598-3}},
doi = {{10.1186/s13063-026-09598-3}},
volume = {{27}},
year = {{2026}},
}