Win Ratios in Ophthalmology-a Hierarchical Metric to Evaluate Adverse Outcomes in Comparative Studies.

PURPOSE: Clinical trials in ophthalmology frequently report multiple efficacy and safety outcomes of varying clinical importance analyzed either in isolation or as simple composites. The Win Ratio (WR) is a hierarchical composite metric originally developed to address such limitations in cardiovascular research by integrating outcomes according to their clinical priority. In this study, we assess its potential utility in ophthalmology outcomes with glaucoma surgery as the exemplar.

DESIGN

Retrospective registry-based study.

SUBJECTS: There are 1257 eyes that underwent a Xen gel stent (Xen) implantation or trabeculectomy (Trab) with a minimum 12 months of follow-up.

METHODS

Propensity score matching yielded two comparable cohorts (212 eyes each) without significant differences in age, baseline intraocular pressure, number of medications, visual acuity, or visual field mean deviation. A predefined hierarchy of eight adverse outcomes was constructed, headed by loss of light perception vision, secondary glaucoma surgery, followed by early postoperative complications (e.g., hypotony and hyphaema) or bleb needling. Each Xen-Trab pair was evaluated sequentially and classified as a win, loss, or tie.

MAIN OUTCOME MEASURES: The overall WR with 95% confidence intervals (CIs) and P value was estimated, and sensitivity analyses examined the robustness of results under alternative outcome hierarchies and random permutations.

RESULTS

The Xen group accumulated fewer wins than Trab (WR 0.83, 95% CI 0.63-1.10, P = 0.19), although the difference was not significant. Differences were driven by higher rates of secondary glaucoma surgery (18.4% vs. 12.7%) and bleb needling within the first 6 months (28.8% vs. 13.2%) in the Xen group, with remaining outcomes occurring infrequently and evenly distributed in each group. Sensitivity analyses with alternative hierarchies yielded consistent results (WR range 0.80-0.98), and random permutations of lower-ranked outcomes clustered around WR 0.77, supporting the consistency of the final outcome.

CONCLUSIONS

The WR offers a clinically meaningful, hierarchical approach to comparing multidimensional treatments that respects clinical priorities. Our findings illustrate the potential utility of this metric in the prospective evaluation of trial outcomes in ophthalmology.

FINANCIAL DISCLOSURES: The authors have no proprietary or commercial interest in any materials discussed in this article.