COVID-19 challenged critical care in many ways, but also provided a unique opportunity to study the global impact of a single group of patients on healthcare systems. In this month’s issue of Critical Care Medicine, Brinkman et al (1) present data from the Netherlands on 120,393 patients that were hospitalized during the COVID-19 pandemic for non-COVID reasons and compares occupancy and outcomes to a historical prepandemic cohort.

There are several key findings in this article that warrant attention and resonate with the findings of other studies. First, when COVID-19 hospitalizations rose, the number of patients in the ICU with non-COVID conditions decreased—up to 34.2% during pandemic peak periods (1). This is not particularly surprising for hospitals with high volumes of elective procedures but does raise the question in many hospitals about whether patients that would normally receive care in the ICU did not, and also asks us to consider the risks of deferred surgeries (2).

Second, and of particular interest was the finding of a small but clear relationship between surge conditions and mortality. Although the surge-mortality relationship was weaker than that estimated for patients hospitalized with COVID-19 by some authors (e.g., one study showed up to 25% of deaths in hospitalized COVID-19 patients may have been related to the hospital surge conditions) (3) this finding is important, as it points to a key nexus between strain on healthcare resources and safety that goes well beyond COVID-19. It also raises the question: are our patients receiving an expected quality of high-precision care in hospitals routinely strained by unusually high caseloads?

Brinkman et al (1) found that during peak pandemic periods, the adjusted odds ratio (OR) for death compared with the prepandemic cohort displayed a statistically significant elevation at 1.10. Although this may be explained in part by residual confounding despite covariate adjustment (patients hospitalized during peak pandemic weeks appeared somewhat sicker than the prepandemic cohort), the pandemic group also had fewer comorbidities than the prepandemic patients and would be expected to have had better outcomes, strengthening the conclusions that the observed greater mortality risk might be real.

Although strain conditions are described, the situation in the Netherlands clearly did not reach the degree of impact seen in many other countries. The peak occupancy ratio (compared with normal, not maximum ICU occupancy) of two was experienced only for a few weeks early in the pandemic and aside from a few weeks at 1.7 later in the pandemic never exceeded 1.5. Only about one-third of ICU patients during the peak of the pandemic were COVID-19 and over one-third of admissions during these periods continued to be elective, with a median length of stay of 1 day in the ICU. Further, even during peak pandemic weeks, only 45% of ICU patients were mechanically ventilated and over 70% were categorized as “low” Acute Physiology and Chronic Health Evaluation IV mortality risk (1).

Thus, even in ICUs caring for relatively low acuity patients, a clear mortality signal was demonstrated across a national database as occupancy increased. Some other studies have found similar and troubling associations between surge and safety. For example, the U.S. Centers for Disease Control and Prevention estimated 12,000 excess deaths in the subsequent 2 weeks if ICU capacity exceeded 75% and 80,000 when it exceeded 100% (4). A study of the Veteran’s Healthcare databases revealed a doubling of ICU mortality during high-demand versus low-demand periods (5). Significant mortality increases occurred (OR 1.23 vs. baseline) in the United Kingdom as the occupancy of intensive care beds exceeded 85% (6) and in the United States during peak pandemic weeks (OR 1.37 vs. baseline) (7). Investigators have found effects on mortality during surge for specific conditions as well—for example, an increase in non-ST-elevation myocardial infarction mortality of over 50% (8).

However, a surge-mortality association was not found in a pooled analysis of critical care patients from 46 countries (9). This may suggest difficulties discerning signals due to heterogeneity in degrees of surge, baseline bed capacity, level of care, and populations studied rather than the absence of effect.

Clearly, there are many confounders, but there is also clearly risk when our ICU volumes increase. And this higher risk of death is not confined to the ICU, but spills over as higher acuity patients are cared for on lower acuity units or end up boarding in the emergency department and suffering worse outcomes (10,11).

However, surge (increased demand) is only one factor that may explain the poor outcomes. Strain, which can be due to surge or to lack of supplies, staff, and other resources is less easily quantified but may be a critical driver of poor outcomes. Historically, disaster surge planning has focused on the “4S” domains of space, staff, supplies, and systems (12). Interestingly, these domains compare directly to the elements of potential failure that the U.S. Centers for Medicare and Medicaid Services considers in root cause analysis (Fig. 1) (13).

Figure 1.: Comparison of surge capacity and root cause analysis contributing domains. Comparison of U.S. Centers for Medicare and Medicaid Services (CMS) root cause analysis factors versus the 4Ss of surge capacity and resilience: The major categories for causes of a problem outlined by CMS suggested for hospitals to use in root cause analyses [https://www.cms.gov/medicare/provider-enrollment-and-certification/qapi/downloads/fishbonerevised.pdf] closely resemble the 4Ss (supplies, space, staff, and systems) of the surge capacity and resilience framework used in disaster medicine.

The factors that drive improving resilience and capacity during disasters are essentially the same as those that drive safety on a daily basis. With ICUs worldwide under continued strain due to both capacity and staffing issues, COVID-19 has made us aware of the ongoing and significant risk of mortality that strain presents. Yet we have little understanding of the factors that most impact patient outcomes.

We have a pressing need for both better data as well as research definitions. How do we define ICU care? It cannot be defined by location alone as critical care spills over into other areas of the hospital—although understanding the risk when this does happen is key to appreciating the importance of trained staff, lower staff-to-patient ratios, and dedicated equipment. Using markers of patient acuity such as the use of pressors and mechanical ventilation across the hospital and comparing to baseline is an important factor that Brinkman included. This should be expanded to look at other interventions such as use of high-flow and positive pressure (continuous positive airway pressure, bilevel positive airway pressure) therapies among others. Accounting for changes in staffing including ratios and level of training is also important but very difficult to capture. Many data “lakes” exist currently with a wealth of patient data that may be helpful to discern safety risks, but are usually not coupled with the temporal data on volumes, nor with their hospital of origin (14). Similar to the aviation industry, healthcare needs to agree on submitting essential elements of information to a third party that without finding fault can help shape our understanding of the relationship between strain and safety and allow us to both monitor conditions and focus on defining the tipping points at which adverse events substantially affect patient outcomes (15). This could allow multiple interventions.

At the bedside, once key vulnerabilities are identified we can proactively evaluate risk mitigation with tests of change—for example, new staffing models, “smart” technology that alerts providers and automatically adjusts infusions according to hemodynamic parameters, or artificial intelligence that recognizes deteriorating or at-risk patient faster than human caregivers with saturated bandwidth.

A surge index that incorporates case severity, resources, personnel, and baseline bed capacity could be developed and used to predict and compare strain across hospitals in a region. Most of this information is available within electronic health records in real-time and could be transferred to an aggregating/analyzing entity with specific rules to govern access and use. This approach might identify centers at risk of surge-related harm and enable within-hospital comparisons in surge and excess deaths over time, direct available support to hospitals and regions that need it most, and direct or transfer patients to facilities that are best positioned to meet their care needs. Homogenization of elements and metrics to define “strain” might also allow easier aggregation of data for global burdens and impacts of surge strain in the future.

ICU care is intently focused on avoiding adverse outcomes associated with indwelling device infections, pressure ulcers, and other documented safety concerns. But perhaps we are missing the forest for the trees. With ICUs often at capacity on a daily basis even outside the pandemic, there very likely are deaths occurring that are preventable and the risk increases with the strain on our resources. We owe it to ourselves and our patients to improve our understanding of the magnitude and the locus of the excess risk. Data are abundant. Determining which data elements may have highest correlation with adverse outcomes (likely volume, acuity, and staffing primarily) and then agreeing on definitions, sourcing, and analysis are the first steps toward improving access to the lowest-risk care that we can provide regardless of the external demand.

1. Brinkman S, de Keizer NF, de Lange DW, et al.: Strain on Scarce Intensive Care Beds Drives Reduced Patient Volumes, Patient Selection, and Worse Outcome: A National Cohort Study. Crit Care Med 2024; 52:574–585 2. Sundt TM: Managing aortic stenosis in the age of COVID-19: Preparing for the second wave. JAMA Netw Open 2020; 3:e2020368 3. Kadri SS, Sun J, Lawandi A, et al.: Association between caseload surge and COVID-19 survival in 558 US Hospitals, March to August 2020. Ann Intern Med 2021;174:1240–1251 4. French G, Hulse M, Nguyen D, et al.: Impact of hospital strain on excess deaths during the COVID-19 pandemic—United States July 2020-July 2021. MMWR Morb Mortal Wkly Rep 2021; 70:1613–1616 5. Bravata DM, Perkins AJ, Myers LJ, et al.: Association of intensive care unit patient load and demand with mortality rates in US Department of Veterans Affairs Hospitals During the COVID-19 Pandemic. JAMA Netw Open. 2021; 4:e2034266 6. Wilde H, Mellan T, Hawryluk I, et al.: The association between mechanical ventilator compatible bed occupancy and mortality risk in intensive care patients with COVID-19: A national retrospective cohort study. BMC Med 2021; 19:213 7. Bela P, Robert E M, Siddharth K, et al.: Surge in Incidence and Coronavirus Disease 2019 Hospital Risk of Death, United States, September 2020 to March 2021. Open Forum Infect Dis 2022; 9:ofac424 8. Glance LG, Joynt Maddox KE, Shang J, et al.: The COVID-19 pandemic and associated inequities in acute myocardial infarction treatment and outcomes. JAMA Netw Open 2023; 6:e2330327 9. Greco M, De Corte T, Ercole A, et al.; ESICM UNITE-COVID investigators: Clinical and organizational factors associated with mortality during the peak of first COVID-19 wave: the global UNITE-COVID study. Intensive Care Med 2022; 48:690–705 10. Roussel M, Teissandier D, Yordanov Y, et al.: Overnight stay in the emergency department and mortality in older patients. JAMA Intern Med 2023; 183:1378–1385 11. Mohr NM, Wessman BT, Bassin B, et al.: Boarding of critically ill patients in the emergency department. Crit Care Med 2020; 48:1180–1187 12. Institute of Medicine and National Research Council. 2005. Public Health Risks of Disasters: Communication, Infrastructure, and Preparedness: Workshop Summary – Chapter 3. Washington, DC: The National Academies Press. https://doi.org/10.17226/11201 13. Centers for Medicare and Medicaid Services. Quality Assurance and Performance Improvement. How to use the fishbone tool for root cause analysis. Available at: https://www.cms.gov/medicare/provider-enrollment-and-certification/qapi/downloads/fishbonerevised.pdf. Accessed November 30, 2022 14. EPIC Cosmos. Available at: https://cosmos.epic.com/. Accessed November 30, 2022 15. Hick JL, Toner ES, Hanfling D, et al.: Data and disasters: Essential information needed for all healthcare threats. Health Secur 2023 Sep 20. [online ahead of print]