Limited incremental value of balance and plantar-pressure measures beyond baseline CAIT for a 12-month composite ankle-instability outcome in competitive badminton players.

Background: Chronic ankle instability and recurrent symptoms are common after lateral ankle sprain, but prediction is difficult when baseline and follow-up outcomes assess overlapping instability constructs. Objective: To determine whether balance and plantar-pressure measures add predictive information beyond baseline Cumberland Ankle Instability Tool (CAIT) scores for a 12-month composite ankle-instability outcome, and to assess performance among athletes with baseline CAIT >24. Methods: In this prospective multicentre cohort, 326 badminton players were enrolled and 308 completed 12 monthly assessments. Centres 1-6 formed the development cohort (n=280), and centre 7 served as a geographic holdout (n=28). The composite outcome required repeated CAIT ≤24, giving-way or recurrent sprain, and symptoms >3 months; a new sprain was not mandatory. A six-predictor logistic model was compared with baseline CAIT, CAIT plus previous sprain, and CAIT plus the six predictors. Decision curves compared strategies. The model was refitted in athletes with baseline CAIT >24, with sensitivity analyses for index-ankle selection, training exposure, repeatability and bootstrap stability. Results: The outcome occurred in 96/280 development athletes and 11/28 holdout athletes; 68/96 and 7/11 cases, respectively, had baseline CAIT ≤24. The six-predictor model had a development AUC of 0.892 (95% CI 0.839-0.940), versus 0.951 (95% CI 0.926-0.971) for baseline CAIT, with a Brier score of 0.085. Adding the six predictors to CAIT did not improve fit (likelihood-ratio P = 0.940), and decision curves showed no consistent net-benefit gain. Among athletes with baseline CAIT >24, the refitted model had an AUC of 0.680 (95% CI 0.553-0.790) in development and 0.603 (95% CI 0.176-1.000) in the holdout. A five-predictor analysis restricted to unilateral-sprain-defined ankles yielded a development AUC of 0.921 (95% CI 0.855-0.975). Adding training volume did not materially improve fit (P = 0.085). Conclusion: Balance and plantar-pressure measures provided no clinically meaningful incremental prediction beyond baseline CAIT. High apparent performance mainly reflected persistence, recurrence or confirmation of instability already represented at baseline, rather than newly developing chronic ankle instability. The equipment-based model is not supported for pre-season screening or prevention allocation.