Crisis Standards of Care

This article by Gershengorn et al. assesses whether resource allocations under crisis standards of care (CSC) during the COVID-19 pandemic were affected by racial or ethnic bias. Using a retrospective cohort study of over 1100 adult patients at 2 US hospitals in Miami, Florida, the study found no significant association between race or ethnicity and daily priority score assigned to patients.

• CSC prioritization is commonly based on a measure of estimated short-term survival (such as SOFA score), and many CSC plans (including at the 2 hospitals in the study) also include assessment of longer-term prognosis based on comorbidities. These approaches, particularly consideration of longer-term prognosis, have been criticized as likely to exacerbate existing inequities because systemic racism and social determinants of health have contributed to higher incidence of comorbidities among communities of color as compared to white populations in the US.
• The study incorporated several potential confounders into its analysis, including sex, primary language, median income of home zip code, primary insurer, age, admission to COVID-19 ward, and which of the two hospitals (tertiary or quaternary) the patient was admitted to. The analysis found no significant differences in maximum or minimum priority group score across race or ethnicity.
• The authors note several possible reasons why there could be an association that the study did not identify, including that 1) the specific study design (focusing on the single most serious comorbidity of a patient and grouping SOFA scores) may have obscured differences, 2) that considering only patients already admitted to a hospital does not captures differences in rates and timing of seeking care that may be related to access, 3) that the sample size may have been insufficient to identify a difference, and 4) that using median zip code income is insufficient to account for confounding by socioeconomic factors.
• Additionally, an attached editorial comment by Manchanda, Molina, and Rodriguez highlights that the pandemic has disproportionately affected Black, Latinx, Indigenous, and other racial and ethnic minority populations and that under-resourced communities more often rely on health care facilities that experience resource constraints and have been hit hardest by infection surges, such as safety-net hospitals, potentially leading to compromised care even before CSC plans are activated. The editorial also notes that the inclusion of “colorblind” SOFA scores in CSC allocation frameworks raise concerns because of racial differences owing to prevalence of specific conditions, such as chronic kidney disease. Lastly, the editorial notes that the primary CSC allocation criteria (SOFA score and life-limiting comorbidities) may be insufficient to assess discrimination because they result in over 70% of patients stratified to the highest priority group. Secondary “tie-breaker” criteria may be required and could introduce bias. Such secondary criteria were not reflected in the study because the hospitals did not actually experience a shortage of ventilators (i.e., CSC priority scores were assigned, but were never used to determine actual allocation).

Using a different approach to the study above, this article by Sarkar et al. (previously covered when it was a preprint) investigates the performance of multiple severity scoring systems using 2 large ICU databases and finds that systematic differences in calibration across ethnicities suggest severity scores reflect bias in their predictions of patient mortality.

• The study considers 3 different scoring systems: Acute Physiology and Chronic Health Evaluation IVa (APACHE IVa); Oxford Acute Severity of Illness Score (OASIS); and Sequential Organ Failure Assessment (SOFA) score. Using data for over 166,000 patient episodes, the study finds evidence that all three models overpredict mortality, particularly for Hispanic and Black patients.
• The authors conclude that the use of any of these clinical prediction scores as part of a triage process should be approached with extreme caution and should potentially be avoided particularly when triaging populations that include patients from different ethnic and socioeconomic backgrounds.

Part of the COVID-19 Policy Playbook II, a report sponsored by the de Beaumont Foundation and the American Public Health Association, this chapter by Gable addresses legal and ethical considerations in allocation of medical resources during times of scarcity. This updates and expands on a chapter by the same author in the prior volume of the Playbook based on an additional 7 months of observations during the pandemic and recommends a series of legal and policy proposals to avoid scarcity and ensure that when CSC must be implemented it is done in a manner consistent with principles of equity.

• Among the crucial observations highlighted in the chapter are the extent to which government leadership failures continued to afflict stockpiling and distribution of scarce resources, the insufficiency of existing CSC plans to address staff shortages (as opposed to supply shortages), and the focus of most CSC plans on worst-case scenarios (e.g., reallocation of life-saving ventilators) rather than more likely circumstances such as extending scarce PPE stocks.
• Few states formally designated CSC implementation, and there have been no explicitly documented cases of triage decisions based on CSC protocols in the US, though there have many anecdotal cases of informal, adaptive efforts to address patient surges that likely effectively changed the standard of care that patients received. Alterations in the standard of care without a formal legal declaration presents heightened risks of potential tort liability.
• The chapter also stresses the need to ensure equity in allocation of scarce resources to avoid exacerbating existing health disparities. As discussed above, existing structural factors produce inequities in morbidity and mortality that may affect patients’ access to scarce resources to the extent such conditions are reflected in prioritization. Additionally, explicit exclusions factors in CSC plans may disfavor persons with physical or intellectual disabilities and have been subject to legal challenges. Even facially neutral frameworks may discriminate by relying on factors such as long-term prognosis or assessed quality of life.