Patient Care in Rapid-expansion Intensive Care Units During the COVID-19 Pandemic Crisis

Jade I. Basem; Anna F. Roth; Robert S. White; Virginia E. Tangel; Silis Y. Jiang; Jacky M. Choi; Katherine L. Hoffman; Edward J. Schenck; Zachary A. Turnbull; Kane O. Pryor; Natalia S. Ivascu; Stavros G. Memtsoudis; Peter A. Goldstein


BMC Anesthesiol. 2022;22(209) 

In This Article


Expansion-ICU Management

Our data suggest that it is logistically feasible, and clinically realistic, to rapidly (< 10 days) convert ORs and PACUs into functional ICUs to accommodate significant numbers of critically ill patients. We found no statistically significant differences in patient characteristics (Table 1) or in mortality rates and discharge to home between the Expansion and Referent ICU cohorts (Table 5). There were also no differences in use of vasopressors or prone positioning throughout the entire hospital stay of either cohort for deceased patients; there were, however, notable differences in the use of sedative medications (propofol, ketamine, dexmedetomidine) between the two groups (Table 4), and the implications of this are discussed below.

These results suggest that converting ORs into crisis ICUs allowed for an expansion of ICU capacity under emergency conditions which resulted in non-inferior patient outcomes. Importantly, this was achievable using a minimum of critical care trained personnel. In 2020, global ICU mortality was 35% to 41.6%.[18–20] There was, however, marked differences among countries: from 23.4% in Japan[21] to 57—59.6% in Brazil.[22,23] Among patient receiving mechanical ventilation (and presumably therefore in an ICU), mortality has shown marked variability: United Arab Emirates—20.2%,[24] Netherlands – 38%,[25] Italy – 51.7%,[26] Germany – 52.8%,[27] Russia – 65.4%,[28] United Kingdom – 69%,[29] Mexico -73.7%,[30] and Romania 95%.[31] More locally, ICU mortality in New York City among patients requiring mechanical ventilation has been reported as high as 88.1%.[32] The in-hospital mortality of patients treated in an Expansion-ICU herein is concordant with early reported intra-institutional mortality among patients requiring invasive mechanical ventilation (14.6%),[33] and is markedly less than that reported for other hospitals in the NewYork-Presbyterian system (41%).[34] This is noteworthy as COVID patients in the NYC area have higher levels of comorbidities, longer intubations, and higher rates of kidney injuries compared to other locations.[35]

Recent data indicates that ICU patient load dramatically impacts mortality rates, with lower rates of available ICU beds or increased ICU overflow being associated with increased mortality.[36–40] This may be due to lack of resources and personnel that occurs during times of high clinical burdens, with differing opinions on whether the use of time-dependent changes in clinical practice influenced those outcomes.[38,40] Taccone et al. also found that the proportion of ICU beds available and the number of newly created ICU beds were each independent risk factors for mortality.[40] In contrast, we found that not only were our mortality rates lower than those of similar hospitals, but our ICU load and expansion ICU areas were not associated with inferior outcomes (Table 5). Of note, however, patients in the Expansion-ICU were more likely to be discharged to a subacute rehabilitation facility than those in the Referent ICU (Table 5); this may reflect the fact that the Expansion-ICU discharged cohort had an ICU LoS that was twice as long as those in the Referent ICU [32 (25, 50) vs 17 (9, 30) days, median (IQR)] and as well as time spent requiring mechanical ventilation [33 (23, 43) vs 17 (10, 33) days, median (IQR)] (Table 3). Prolonged LoS in an ICU is associated with ICU-acquired weakness ("deconditioning") which can result in profound functional impairment,[41] and so the higher incidence of discharge to rehabilitation facilities in the Expansion-ICU population might reasonably be expected. Our results are a clear demonstration that "repurposing" of resources can in fact provide safe and effective care, and support in principle the approach advocated by Diaz et al. for repurposing pediatric ICUs to adult critical care units.[42]

We also address the limitations of other studies by providing comparisons of patient characteristics, treatments, and outcomes across different settings. There were a few differences in hospital management for deceased patients between the Referent and Expansion-ICUs (Table 3). The Expansion cohort had a significantly higher ICU LoS and time on ventilator in deceased patients. Those differences could reflect, in part, disease progression and corresponding care-escalation requirements; for example, if a patient required renal replacement therapy (a predictor of disease severity and mortality),[33,34,43–46] they would have been transferred to a traditional ICU.

There were some differences in laboratory findings (Table 2). The Expansion-ICU cohort had significantly lower median levels of ferritin, LDH, and white blood cell count (WBC) within the first 24 h of admission to an ICU. In contrast, increased procalcitonin, C-reactive protein, IL-6, ferritin, LDH, and D-dimer levels associated with severe or fatal COVID-19 infections (> 1 μg•mL−1) were observed, consistent with prior reports.[43,46–54] A recent meta-analysis found that compared to patients discharged from the hospital, those that died had higher WBC, ferritin, C-reactive protein, D-dimer, LDH, and IL-6 levels with decreased levels of lymphocytes, hemoglobin, and albumin compared to those that were discharged from the hospital.[50] While the lower levels of ferritin, LDH, and WBC in the Expansion cohort may point to a less severe disease progression, other lab results were similar between cohorts suggesting that the overall severity of illness was comparable.

When considering the co-morbidity of the two groups, the incidence of hypertension, diabetes, or obesity as the most reported comorbidities in both cohorts was comparable to that reported within and outside of the NewYork-Presbyterian system.[33,34,43–48,53] The high rates of cardiovascular disease, pulmonary disease, older age, elevated IL-6 and D-dimer levels that we observed in our cohorts (Table 1 and Table 2) are common features in COVID-19 patients, and are potential predictors of mortality within the wider NYC cohort[34] and elsewhere.[44–46,53,55] Previous studies have demonstrated that use of any vasopressors, particularly for extended time periods, was associated with disease severity or mortality.[33,43] In contrast, however, we found no significant difference in disposition between patients who did and did not receive vasopressors.

Prone positioning has been reported to improve oxygenation in spontaneously breathing non-intubated patients with hypoxemic acute respiratory failure[56] as well as in patients with acute respiratory failure in the setting of COVID-19 respiratory failure.[57] Other studies have also suggested benefits of prone positioning for ARDS and COVID-19, but with its benefits often limited to early use in patients not requiring mechanical ventilation.[58–61] The apparent lack of benefit of proning reported here suggests that its application late in the course of COVID-19 respiratory failure (i.e., once invasive mechanical ventilation has been initiated) is not indicated. Whether this is true requires an appropriately designed prospective trial.

There were several differences in hospital management for discharged patients between the Expansion and standard ICU cohorts. While vasopressors and prone positioning remained similar in the discharged group as with the deceased group, the Expansion-ICU cohort had significantly higher rates of use of propofol, dexmedetomidine, and ketamine. Dexmedetomidine administration is associated with improved oxygenation in morbidly obese patients with restrictive lung disease compared to a placebo group.[62] Of direct relevance here, dexmedetomidine, when administered to adult patients with COVID-19 who were admitted to an ICU and required sedation, was associated with a significant increase in the PaO2/FiO2 (PF) ratio 4–12 h following dexmedetomidine administration (PF at baseline; 17 ± 6 vs 21 ± 5 at 6 h, P < 0.001).[63] Critically, dexmedetomidine administration in a different cohort of patients with COVID-19 who required invasive mechanical ventilation had significantly lower 28-day mortality than those who did not receive it (respectively, 27.0% vs 64.5%, relative risk reduction 58.2%, 95% confidence interval 42.4–69.6).[64] The observed survival benefit in patients who received dexmedetomidine is consistent with our results wherein dexmedetomidine administration was administered more often to patients who were discharged from the hospital as compared to those who died (Table 4). The mechanism(s) through which dexmedetomidine might confer a survival benefit are not known with certainty, but may include: reduced agitation and increased ventilator compliance, enhanced hypoxic pulmonary vasoconstriction (HPVC), and improvement in the ventilation/perfusion ratio ([64] and references therein). Thus, the higher rate of dexmedetomidine administration in the Expansion-ICU cohort may have may have had a beneficial effect on their overall outcomes.


Our findings are associative and will require additional research to define their value in relation to the COVID-19 pandemic. With a much smaller Expansion-ICU cohort compared to the Referent cohort, a larger dataset may have been helpful for detecting small differences in therapeutic interventions between the two groups. As with many retrospective studies, the analyses are only as reliable as the available data, and documentation gaps in the EMRs hampered our ability to perform analyses. Similarly, the study was limited by the fact that some patients were transferred to our care from outside hospitals, and we were unable to obtain full laboratory results and management interventions prior to arrival. Within-hospital transfer between units occurred as well, either due to space limitations or as warranted by disease severity, so this offers an additional confounding variable when analysing patient outcomes between cohorts. Finally, our capacity to rapidly expand services in a large, urban, tertiary care medical center may not be generalizable to smaller hospitals with fewer resources.