UK liver transplantation allocation algorithm: transplant benefit score Authors' reply

We thank Raj Prasad and colleagues and Zhi Qu for their Correspondences. Prasad and colleagues comment that we considered a restricted set of parameters in isolation and did not mention reduced waiting-list mortality. We focused on hepatocellular carcinoma after failing to understand why the transplant benefit score (TBS) of patients with chronic liver disease fell noticeably if they developed hepatocellular carcinoma. Short-term waiting-list mortality has improved, which is most likely a reflection of multiple changes: national versus regional allocation, TBS-directed transplantation, and clinician-directed transplantation. When including long-term outcomes, waiting-list survival is now worse for patients with hepatocellular carcinoma versus patients with chronic liver disease alone, indicating that risk remains despite stabilising treatments.1Taylor R Downward L Banks J National Liver Offering Scheme: thirty-six month review.https://nhsbtdbe.blob.core.windows.net/umbraco-assets-corp/23351/nlos-36-month-monitoring-report.pdfDate: April, 2021Date accessed: May 25, 2023Google Scholar, 2Collett D Allen E Aluvihare V et al.Fixed term working unit—organ allocation.http://odt.nhs.uk/pdf/advisory_group_papers/LAG/Allocation_System.pdfDate: Oct 15, 2014Date accessed: May 25, 2023Google Scholar Prasad and colleagues argue that comparing different health-care systems has debatable academic merit. We would counter that statement by commenting that strengths and weaknesses only become apparent by comparing approaches. Comparing unadjusted mortality rates is challenging, given differing listing practices and demographics. Prasad and colleagues cite the UK benefit-based system, including need and utility, versus the need-based system in the USA and Eurotransplant regions as fundamental differences. However, we found that TBS is predominantly a need-based score given that utility had minimal effect on TBS. Qu comments that score inflation for patients with hepatocellular carcinoma under systems in the USA and Eurotransplant regions limits access for patients with conditions other than hepatocellular carcinoma. Correctly weighting systems is challenging and undoubtedly requires revisions over time.3Heimbach JK Evolution of liver transplant selection criteria and US allocation policy for patients with hepatocellular carcinoma.Semin Liver Dis. 2020; 40: 358-364Crossref PubMed Scopus (0) Google Scholar Regardless, the policy in these regions is clearly stated. Policy might be challenging to interpret when so-called black box algorithms, which are difficult to understand due to complexity or limitations to access, are used to make allocation decisions. In this case, black box decision making deprioritised patients with hepatocellular carcinoma due to counterintuitive survival predictions. The TBS prediction that hepatocellular carcinoma improves survival does not fit with real-world experience, but algorithms cannot apply common sense. Spurious correlates in training data can only be recognised by systematic testing.4Liu X Glocker B McCradden MM Ghassemi M Denniston AK Oakden-Rayner L The medical algorithmic audit.Lancet Digit Health. 2022; 4: e384-e397Summary Full Text Full Text PDF PubMed Scopus (39) Google Scholar Prasad and colleagues comment that we did not recognise underutilised transplantation options. Rather, we include data that suggest clinicians used livers from donation after circulatory death to compensate for TBS-directed allocation.5Attia A Rowe IA Harrison EM Gordon-Walker T Stutchfield BM Implausible algorithm output in UK liver transplantation allocation scheme: importance of transparency.Lancet. 2023; 401: 911-912Summary Full Text Full Text PDF PubMed Scopus (0) Google Scholar We agree that increasing liver availability requires potential donors and their families to trust and respect the health-care system. Prasad and colleagues raise a concern that our study might affect organ donation. We believe the best way to facilitate confidence in complex systems is for them to be understood, ensuring decisions can be explained. We dispute that organ allocation is a zero-sum game because allocation based on inaccurate risk estimates could reduce the overall benefits of the programme. Ensuring algorithms are transparent, extensively tested, audited, and openly shared might help to generate the maximum benefit with a valuable, limited resource. We declare no competing interests. UK liver transplantation allocation algorithm: transplant benefit scoreWe read with interest the Correspondence by Antony Attia and colleagues1 that fundamentally questions the new liver allocation scheme in the UK, highlighting a disadvantage to patients with hepatocellular carcinoma. The Correspondence considered a restricted set of parameters in isolation and did not mention that the new scheme achieved the short-term objective of reducing waiting-list mortality before the COVID-19 pandemic. The COVID-19 pandemic resulted in a vogue to compare different health-care systems. Full-Text PDF UK liver transplantation allocation algorithm: transplant benefit scoreThe Correspondence by Antony Attia and colleagues1 revealed the limitation of the liver allocation scheme in the UK. Although the model for end-stage liver disease (MELD) score used in the USA and Eurotransplant assigns exception points to patients with hepatocellular carcinoma to reflect the mortality risk of patients, which is established by the progression of the neoplasm rather than by a further decrease in liver function, these additional points lead to a higher proportion of waiting-list dropouts and lower rates of transplantation for patients with liver disease other than hepatocellular carcinoma, resulting in a substantial disparity in access to transplantation. Full-Text PDF