The Algorithm That Decides Who Lives

The Committee No One Elected

The formula that decides who gets a liver transplant was written by a committee in 2002. It has been changed 11 times since. Each change saved some lives and cost others. Nobody voted on it. The MELD score—Model for End-Stage Liver Disease—reduces human lives to a single number between 6 and 40, calculated from three laboratory values: bilirubin, creatinine, and INR. In 2023, approximately 11,000 Americans received liver transplants while over 1,500 died waiting. The difference between life and death often came down to a few decimal points in an equation most patients have never heard of.

The Mathematics of Mortality: Deconstructing the MELD Score

The MELD score was originally developed at the Mayo Clinic to predict mortality following transjugular intrahepatic portosystemic shunt (TIPS) procedures. In 2002, the United Network for Organ Sharing (UNOS) adopted it for liver allocation, replacing a system based primarily on waiting time and clinical urgency assessments that varied wildly between transplant centers.

The formula itself appears elegant in its simplicity:

MELD = 3.78 × ln(serum bilirubin [mg/dL]) + 11.2 × ln(INR) + 9.57 × ln(serum creatinine [mg/dL]) + 6.43

Each coefficient was derived from statistical analysis of patient outcomes, but the choice to use natural logarithms—and specifically these three laboratory values rather than dozens of other available markers—represents a series of ethical decisions disguised as mathematical optimization.

[!INSIGHT] The logarithmic transformation means that improvements at lower disease severity carry more weight than equivalent improvements at higher severity. A patient whose bilirubin drops from 3.0 to 2.0 mg/dL gains more priority points than one dropping from 10.0 to 9.0 mg/dL—despite the latter being closer to death.

The Creatinine Controversy

Serum creatinine, a marker of kidney function, carries the highest coefficient weight in the MELD equation (9.57 compared to 3.78 for bilirubin). This mathematical reality means patients with concurrent hepatorenal syndrome—kidney failure secondary to liver disease—receive significantly higher priority.

However, creatinine levels correlate with muscle mass. Women, elderly patients, and those with sarcopenia (muscle wasting common in advanced liver disease) may have deceptively low creatinine values despite severe renal dysfunction. A 2019 analysis published in the American Journal of Transplantation found that women had a 15% lower likelihood of receiving a liver transplant compared to men with equivalent clinical severity, partially attributable to creatinine-based bias in the MELD calculation.

The 2016 introduction of the MELD 3.0 revision attempted to address this by adding sex as a variable and adjusting the creatinine coefficient, but the fundamental question remains: how do we weigh accuracy against fairness when any formula will inevitably disadvantage some patient populations?

Geography as Destiny: The 2018 Allocation Revolution

Until 2018, livers were allocated within 11 geographic regions, some containing transplant centers serving vastly different population densities. A patient in Northern California (Region 5) faced dramatically longer wait times than one in Tennessee (Region 3), despite identical MELD scores. Geographic arbitrage emerged—patients with means relocated temporarily to regions with shorter waitlists.

The 2018 Acuity Circles policy replaced rigid regional boundaries with concentric circles radiating from donor hospitals: 150 nautical miles, 250 nautical miles, then 500 nautical miles. A liver from a donor in rural Pennsylvania might now go to a sicker patient in New York City rather than a healthier patient in Pittsburgh.

The mathematical modeling predicted this change would save 60-100 additional lives annually. Early data suggests the prediction was accurate—but it also concentrated transplants in high-volume urban centers. Smaller transplant hospitals argue the policy threatens their viability, potentially reducing overall transplant capacity.

[!NOTE] The Geographic Disparity Index measures variance in transplant rates across locations. Lower values indicate more equitable distribution. While the 2018 reform reduced geographic inequity, it introduced new concerns about organ viability during longer transport times.

Blood Type: The Biological Lottery

The MELD score determines priority within blood type compatibility groups, but blood type distribution among donors and recipients creates systematic inequities. Type O donors can give to anyone, but Type O recipients can only receive Type O organs. With approximately 45% of the population being Type O but only 38% of donors, Type O patients face longer waits despite identical MELD scores.

This isn't a policy choice—it's biological reality—but the decision not to adjust allocation ratios to compensate for blood type disadvantages represents a deliberate ethical stance: biological fairness over outcome optimization. An alternative system might prioritize by expected years of life saved rather than blood type compatibility, but this would advantage younger, healthier patients over those most urgently ill.

The Exception Paradox: When Rules Bend

The MELD system includes exception points for conditions the formula inadequately captures. Hepatocellular carcinoma (HCC) patients receive automatic MELD scores that may not reflect their laboratory values. A patient with early-stage liver cancer and well-preserved synthetic function might jump ahead of patients with more severe laboratory abnormalities.

Between 2002 and 2019, HCC exception cases grew from 5% to 15% of all liver transplants. A 2020 Journal of Hepatology study found that 23% of HCC exception patients had post-transplant survival worse than non-exception patients with equivalent MELD scores—suggesting the exception system may be allocating organs to patients less likely to benefit.

Each exception represents a lobbying victory by patient advocacy groups and medical subspecialties. The system is technically neutral, but its exceptions encode which diseases have the most effective political advocates.

Living with Algorithmic Mortality

The UNOS liver allocation algorithm will never be fair. Not because the mathematics are flawed, but because fairness itself is not a single objective that can be optimized. Any system must balance competing values: urgency (how sick is the patient?), utility (how much will the transplant help?), equity (are outcomes distributed fairly across demographic groups?), and efficiency (can we minimize organ wastage?).

These values conflict irreconcilably. Prioritizing urgency means transplanting sicker patients who may have worse outcomes. Prioritizing utility means healthier patients get transplants while sicker ones die waiting. Prioritizing equity may require deliberately disadvantaging some groups to benefit others.

The current algorithm represents our collective best attempt at an impossible problem, revised iteratively as we learn more about its consequences. Transparency about the trade-offs—rather than the illusion of mathematical neutrality—may be the most honest path forward.

Sources: United Network for Organ Sharing (UNOS) OptN Policies; Wiesner et al. (2003) "MELD and PELD" Liver Transplantation; Kwong et al. (2019) "Sex Disparities in Liver Transplant Allocation" American Journal of Transplantation; Goldberg et al. (2020) "Geographic Variation in Liver Transplant" Journal of Hepatology; Organ Procurement and Transplantation Network Data Reports 2018-2023