P119 Whole list simulation of UK liver transplant allocation

<h3></h3> Demand for liver transplantation exceeds donor organ supply. Both supply and demand are variable over time; allocation complexity is increased by graft:recipient compatibility and time-varying windows of opportunity. Given that controlled trials are impractical, whole-list time-based simulation models are necessary to estimate the effects of allocation algorithms such as the Transplant Benefit Score (TBS) introduced in the UK from 2018. Reference data were NHSBT records reflecting patients listed and organs offered 1995–2020. For simulation, the baseline waiting list was populated by seed defined pseudorandomised sampling of this population. Each simulated day, additional recipient or organs were added from respective NHSBT datasets with daily additions sampled from a Poisson distribution to give a specified rate. Organs were declined/accepted probabilistically by Donor Liver Index (DLI) and allocated by TBS, or alternative, and NHSBT-specified ABO and weight compatibility rules. Pre- and post-transplant survival was estimated using TBS. Cohort and individual outcomes were assessed. Results were combined from analysis with 20 separate seeds; where parameters were changed for investigation, the same seeds were used repeatedly providing consistent recipient and organ characteristics between variations. Simulation in Python; synthetic data using sdv.dev; statistics using Stata. Over 2018, the NHSBT waiting list was ~400 with 674 transplants and a 1-year survival of 83.4%; in 20 simulations from a baseline of 400 simulated patients, outcomes were similar: mean 659 transplants and 1-year survival of 86.6%. In simulation using TBS allocation, increasing organ offer frequency and lowering DLI threshold improved survival with incrementally smaller gains. Predicted survival was greater among women than men (HR 0.91;0.88–0.94); survival was similar by age group but younger patients were transplanted when more unwell. Survival varied by ABO blood group. Death by 1 year was most frequent among re-transplant patients (15%) and least frequent in ‘other’ aetiology (8%); patients with metabolic liver disease were most likely to be transplanted by one year (70%) and least likely in ‘other’ aetiology (20%). Compared to allocation by waiting time alone, allocation by random added 1.1%, UKELD 4.1%, MELD3 4.2%, TBS M1 alone 4.4% and TBS 6.1% population life years. Similar results for all analyses were obtained using synthetic data derived from the base dataset. The presented simulation model simplifies reality but suggests groups who are relatively benefitted by the current allocation schema, estimates a population benefit of TBS over other allocation mechanisms and presents a basis for further simulated experimentation.