PROJECT SUMMARY
When a US hospital system is overwhelmed by disaster, Crisis Standards of Care guide the triage teams forced
to choose which patients receive scarce life support treatments. Analogous to an organ allocation system, these
algorithms convert ethical principles into a concrete rank ordering of candidates for Intensive Care Unit (ICU)
treatments with life support allocation scores. Disasters that produce scarcity tend to fall hardest on
disadvantaged communities, especially racial and ethnic minority groups. When designing algorithms to allocate
scarce life support, public health officials should take this context into account.
In an attempt to identify the critically ill patients with the highest likelihood of benefit from treatment,
most US states would prioritize those with low Sequential Organ Failure Assessment (SOFA) scores. But
SOFA was designed for patients already on life support in the ICU, using routinely measured laboratory values,
drug doses, and vital signs to monitor response to treatment. Most patients have low SOFA scores when
critical illness is first recognized, and SOFA cannot accurately predict the risk of death using data before life
support was allocated. We demonstrated how the poor predictive performance of SOFA-based triage protocols
is partially explained by underpredicting the survival of Black patients due to a miscalibrated renal component
of the SOFA score. There is a clear need to develop and validate a novel life support allocation protocol
designed to debias existing scores and save more lives. Place-based disadvantage indices, such as the
Area Deprivation Index (ADI) and the Social Vulnerability Index, offer a potential solution. Using these
validated geographical measures of neighborhood deprivation to allocate scarce healthcare resources
counteracts the risk-increasing effects of social disadvantage, including disadvantage produced by racialized
residential segregation. We hypothesize that a well-designed life support allocation score using place-based
disadvantage indices can save more lives and mitigate healthcare inequity in a crisis.
The overall objective of this project is to develop a life support allocation algorithm that accurately and
equitably allocates scarce ICU treatments in a crisis. In Aim 1, we will use structural equation modeling to
create an Equitable Life Support Allocation (ELSA) score, using place-based disadvantage indices to debias
SOFA. In Aim 2, develop the ICU Crisis Simulation Model (ICSM), a discrete event simulation that models
patient flow and survival, as a testing and evaluation environment for life support allocation protocols. In Aim 3,
we will externally validate ELSA and ICSM in the National COVID Cohort Collaborative Data Enclave, which
currently contains geocoded records from 14 million patients from 74 sites. Our project will address one of
the most pressing challenges in applied public health ethics, producing 1) an empirically derived score to
distribute life support more accurately and equitably in a crisis and 2) open-source simulation software to
evaluate the efficiency and equity of life support allocation protocols.