The UK's NHS is using the National Liver Offering Scheme, a predictive scoring algorithm found to have a fatal error, to help decide who gets a liver transplant
Sarah Meredith was in urgent need of a liver when she found out an algorithm would be making the life-or-death decision
Context & Ripple Effects
The NHS’s use of software in clinical operations has expanded from its pandemic capacity dashboard partnerships to AI-enabled heart screening. This case puts the focus on a far higher-stakes use: a system that helps allocate a scarce treatment resource.
It also fits a documented pattern in transplant technology: a kidney-allocation algorithm was found to disadvantage Black patients, while reporting on the US network identified software mistakes and failures in organ-donation infrastructure.
First-order effects
- Patients awaiting liver transplants and the clinicians advising them face immediate uncertainty over whether allocation recommendations made with the scheme were safe and reliable.
- The NHS’s transplant-allocation process is exposed to heightened scrutiny because an error in a decision-support tool can affect life-or-death prioritisation.
Second-order effects
- The finding increases the need for independent validation, error detection, and clear clinician oversight before predictive systems are relied on in transplant pathways.
- Other health systems and vendors using algorithms for scarce-resource allocation will face stronger pressure to show that their tools are accurate across patient groups, not merely operationally convenient.
Third-order effects
- If such failures recur, clinical AI adoption is likely to shift from deploying tools first to proving traceability, safety, and accountability before they influence irreversible decisions.
- The broader consequence may be a more formal governance category for healthcare algorithms used in allocation: systems whose outputs affect access to treatment will require a higher bar than screening or administrative tools.
The trend: Healthcare providers are moving from using software to monitor and detect risk toward using it in allocation decisions, making auditability and human accountability central deployment requirements.