Equitably Allocating Resources during Crises: Racial Differences in Mortality Prediction Models.

Area of Science:

• Critical care medicine
• Health equity research
• Biostatistics

Background:

• Crisis standards of care (CSCs) are essential for allocating critical care resources during emergencies.
• Current CSCs often rely on the Sequential Organ Failure Assessment (SOFA) score to predict in-hospital mortality.
• The performance of SOFA and other acuity scores across different racial groups remains under-investigated.

Purpose of the Study:

• To evaluate the prognostic accuracy of the SOFA score and the Laboratory-based Acute Physiology Score version 2 (LAPS2) in predicting in-hospital mortality among Black and white patients.
• To assess potential racial disparities in mortality prediction using these scores.

Main Methods:

• A retrospective analysis of 113,158 patients (24.4% identifying as Black) admitted for sepsis or acute respiratory failure across 27 hospitals.
• Calculation and comparison of discrimination (AUC) and calibration for SOFA, LAPS2, and modified versions, including a SOFA score excluding creatinine.
• Simulation of CSC prioritization based on observed mortality to identify potential racial biases.

Main Results:

• The LAPS2 demonstrated superior discrimination (AUC 0.76) compared to the SOFA score (AUC 0.68).
• Both scores showed calibration issues, underestimating mortality in white patients and overestimating it in Black patients.
• A modified SOFA score without creatinine showed reduced racial miscalibration.

Conclusions:

• The SOFA score, when used in CSCs, may contribute to racial disparities in critical care resource allocation.
• The LAPS2 exhibits better prognostic accuracy and calibration than SOFA across racial groups.
• Development of more equitable mortality prediction scores is crucial for fair resource allocation during crises.