Abstract
BACKGROUND: Allocation of scarce donor organs must balance improving overall outcomes with protecting patients at greatest near-term risk. Urgency-focused systems such as MELD (Model for End-Stage Liver Disease) are efficient in saving lives but can reduce total post-transplant survival, whereas unconstrained utility maximization risks bypassing the sickest patients. METHODS: We propose Harm-Threshold Utilitarianism (HTU), which maximizes expected post-transplant benefit subject to two guardrails: (i) an epistemic threshold requiring sufficient confidence before acting on predicted differences, and (ii) a catastrophic harm threshold that blocks bypassing candidates at high short-term waitlist mortality risk for only marginal or uncertain gains. Using de-identified U.S. liver transplant registry data, we performed proof-of-concept simulations of offer-like pools, comparing HTU with MELD-based selection. Outcomes included Kaplan-Meier curves and 5-year restricted mean survival time (RMST). Sensitivity analyses varied the harm threshold and confidence level. RESULTS: HTU transparently reorders candidates relative to MELD while preserving alignment on broad priorities. In simulated pools, HTU-selected recipients achieved higher post-transplant survival; mean RMST improved by approximately 0.25 years per transplant (about three months) at baseline settings. Varying the catastrophic harm threshold produced a clear urgency-efficiency frontier: tighter thresholds selected more urgent patients with smaller gains, while looser thresholds increased gains but allowed more bypasses of urgent candidates. CONCLUSIONS: HTU operationalizes a tunable, ethically explicit trade-off between benefit and protection of the worst-off. By encoding precaution (confidence threshold) and a non-negotiable floor against catastrophic harm, HTU offers measurable efficiency gains without sacrificing fairness, and reframes policy choices as transparent parameters open to review.