Fallacies of Using the Win Ratio in Cardiovascular Trials

Summary The win ratio was introduced into cardiovascular trials as a potentially better way of analyzing composite endpoints to account for the hierarchy of clinical significance of their components and to facilitate the inclusion of recurrent events. The basic concept of the win ratio is to define a hierarchy of clinical importance within the components of the composite outcome, form all possible pairs by comparing every subject in the treatment group with every subject in the control group, and then evaluate each pair for the occurrence of the components of the composite outcome in descending order of importance, starting at the most important and progressing down the hierarchy if the outcome does not result in a win in either pair until pairs are tied for the outcome after exhaustion of all components. Although the win ratio offers a novel method of depiction of outcomes in clinical trials, its advantages may be counterbalanced by several fallacies (such as ignoring ties and weighting each hierarchal component equally) and challenges in appropriate clinical interpretation (establishing clinical meaningfulness of the observed effect size). From this perspective, we discuss these and other fallacies and provide a suggested framework to overcome such limitations to enhance utility of this statistical method across the clinical trial enterprise.

However, it is difficult to define all adjustment variables prospectively that require adjustment and, therefore, this technique is not often used. Another strategy is to use a stratified WR approach. The stratified approach involves identifying major risk factors for the outcome in advance and dividing participants into strata based on these risk factors before performing pairwise comparisons between each stratum. 10 The stratum specific win proportions are then combined across the strata to estimate the overall WR for the study.
The stratified WR was used in the ATTR-ACT trial where individuals were stratified according to TTR status (variant or wild-type) and baseline NYHA functional class. 6 In a randomized controlled trial, baseline risk is generally balanced between the study arms and, therefore, an unmatched WR is acceptable.   Table 1. In a situation where ties are minimal, the WR will approximate the WO; however, in most cases the WO should be preferred over the WR. In the event of a significant amount of censoring before the end of the study, there is no clear consensus on whether the WO or WR is superior.
When a patient is lost to follow-up, they are censored from the study. Because each pair can only be compared for their shared study period, if an event occurred in the patient that remains after the other drops out, or an event occurs in the subject that drops out after they leave the study, this cannot be adequately accounted. This type of censoring makes the assessment of a win, loss, or tie unreliable. Some methods to adjust the win statistic for censoring have been proposed, but require further evaluation. 14 If many dropouts are encountered in a WR analysis, a sensitivity analysis that adjusts for censoring should be mandated. The second major consequence of ignoring ties is that the WR only reflects nonidentical outcomes between the groups and can be misleading about the similarity of the distribution of outcomes in the study arms. This finding is particularly relevant in noninferiority trials, because

THE WR
The demonstration that an intervention has achieved its intended effects is an important legal and ethical standard that must be achieved before an intervention is introduced into general clinical practice.
The statistical method by which this efficacy is established must undergo the highest level of scrutiny. We have described 5 fallacies of using the WR that highlight the complexity within this seemingly simple analytical method. As use of the WR becomes more common in cardiovascular trials, to complement the FDA's recent guidance on the analysis of composite endpoints, we believe that the time has come to convene key stakeholders to outline consensus guidance on how (and when) the WR should be used in the analysis and interpretation of cardiovascular trials. We offer several suggestions for the optimal use and clinical interpretation of the WR within cardiovascular trials. A clinician should not stop interpretation at the WR statistic outcome alone.
They must seek to understand the overall treatment effect (by accounting for ties and dropouts) and to have sufficient information from the analyses to ascertain a number needed to treat or harm. Thus, an assessment of aggregate clinical benefit is necessary to translate information from trials that demonstrate wins using the WR to not only appreciate the magnitude of effect, but temporal relationships of outcomes, their durability, and finally the overall value (or cost effectiveness). Similarly, what drives the overall win must be clearly depicted and whether a placebo effect may exist ( Figure 2).
We also offer some important solutions to overcome the inherent challenges described in the various fallacies of using the WR or when to avoid its use.
First, we recommend the traditional frequentist statistical methods approach over the simplistic WR in select scenarios. For instance, if the treatment effect   In summary, although the WR offers a novel method of depiction of outcomes in clinical trials, its advantages may be counterbalanced by several potential fallacies in its conduct and clinical interpretation for which we have provided a framework to overcome such limitations and enhance the value of such utility across the clinical trial enterprise. @MRMehraMD.