UCLA

The COVID-19 pandemic demonstrated the potential for healthcare systems to lack capacity to meet demand for critical care in times of crisis. Recent research suggests a rise in the utilization of intensive care unit (ICU) resources and the potential misallocation of resources to patients without clinical need or hope of benefit. Several authorities have published resource allocation policies to guide healthcare systems, commonly relying on measures of illness severity to determine the priority by which ICU resources would be allocated. This raises concerns about the properties of and the potential for allocation criteria to exacerbate racial disparities in clinical outcomes. Policymakers intend that such policies would maximize the number of lives saved by prioritizing provision of critical care services to patients most likely to benefit. This dissertation examines the interrater reliability of the University of California’s Scarce Resource Allocation Policy (SRAP) in determining the allocation priority of a cohort of consecutively admitted ICU patients at the University of California, Los Angeles (UCLA) Health System. Use of the SRAP had relatively poor reliability in determining allocation priority as laid out within the policy itself. A microsimulation model examined the likely impact of allocation decisions and likelihood of allocation of resources under four defined scenarios of resource constraint, for the outcome of ICU mortality. Mortality differed significantly across tested constraint levels compared to the case of no constraint (i.e., current capacity). Mortality was greater in subgroups with lower priority for constrained ICU resources. A mediation analysis examined if observed differences in mortality risk among racial groups are related to the use of SOFA scores or the selected comorbid conditions. Results suggest that these policy criteria do not mediate the effect of race on mortality. Understanding the projected outcomes related to use of these policies and the policy criteria that drive observed differences among patient groups can better inform policymakers in shaping protocols to maximize lives saved and avoid worsening healthcare disparities.