Pulled-up race outcomes as welfare-relevant surveillance indicators in United States Steeplechase horses, 2023-2025.

Introduction: Pulled-up race outcomes are routinely recorded but are nonspecific, potentially reflecting fatigue, poor performance, footing, rider judgment, suspected injury, or physiological compromise. We evaluated their association with subsequent participation disruption in United States steeplechase horses. Methods: National Steeplechase Association Race Results Reports from 2023 to 2025 were analyzed retrospectively at the horse-start level. Outcomes were time to the next later-date start, ≥60-day interruption, ≥183-day absence, and no recorded later-date start. Cox models were stratified by race type with horse-clustered robust covariance; binary outcomes used horse-clustered generalized estimating equations. Sensitivity analyses excluded November starts because racing is seasonal from March through November. Results: Among 3,070 starts by 710 horses, 367 (11.95%) were pulled up, involving 277 horses (39.0%). No later-date start was recorded after 37.6% of pulled-up and 21.2% of non-pulled-up starts. Among returners, median time to the next start was 42 versus 35 days; Kaplan-Meier medians for all starts were 148 versus 45 days. Pulled-up status was associated with a lower recorded return rate (hazard ratio 0.589, 95% confidence interval 0.515-0.675). Race-type-adjusted odds ratios were 2.10 (1.67-2.64) for ≥60-day interruption and 2.36 (1.80-3.10) for ≥183-day absence. Associations persisted after excluding November starts. Discussion: Pulled-up outcomes were associated with participation disruption, although clinical reasons were unavailable. They may help prioritize follow-up and longitudinal surveillance but cannot replace veterinary examination or clinical records.