Covid-19 Articles: Original Clinical Research Report

Logistical Considerations and Clinical Outcomes Associated With Converting Operating Rooms Into an Intensive Care Unit During the Coronavirus Disease 2019 Pandemic in a New York City Hospital

Mittel, Aaron M. MD^*; Panzer, Oliver MD^*; Wang, David S. MD^*; Miller, Steven E. MD^*; Schaff, Jacob E. MD^*; Hastie, Maya Jalbout MD, EdD^*; Sutherland, Lauren MD^*; Brentjens, Tricia E. MD^*; Sobol, Julia B. MD, MPH^*; Cabredo, Almarie MSN, FNP^†; Hastie, Jonathan MD^*

Author Information

From the ^*Department of Anesthesiology, Columbia University Irving Medical Center, New York, New York

^†Department of Perioperative Services - Nursing, NewYork-Presbyterian Hospital, New York, New York.

Published ahead of print 28 October 2020.

Accepted for publication October 20, 2020.

Funding: None.

The authors declare no conflicts of interest.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Reprints will not be available from the authors.

Address correspondence to Aaron M. Mittel, MD, Department of Anesthesiology, Columbia University Medical Center 622 W 168th St PH 5, 505C, New York, NY 10032. Address e-mail to [email protected].

Anesthesia & Analgesia 132(5):p 1182-1190, May 2021. | DOI: 10.1213/ANE.0000000000005301

Free
SDC
Corresponding Article

Abstract

KEY POINTS

Question: What logistical, structural, and personnel issues must be addressed to effectively convert operating rooms to an intensive care unit and how will this impact clinical outcomes?
Findings: Critically ill coronavirus disease 2019 (COVID-19) patients who received care in the novel operating room intensive care unit (ORICU) at NewYork-Presbyterian Hospital/Columbia University Irving Medical Center (NYP-Columbia) had an unadjusted 41.4% mortality rate and an estimated adjusted probability of survival of 0.61 at 30 days.
Meaning: Survival rates of COVID-19 patients cared for in the novel ORICU were worse in older patients but were similar to those reported by other health care systems, suggesting effective delivery of critical care in an atypical environment.

See Article, p 1179

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and associated coronavirus disease 2019 (COVID-19) illness is a global health emergency that has significantly disrupted normal patterns of health care.^{^1–4} In New York City, the local presence of SARS-CoV-2 was officially confirmed on March 1, 2020.^⁵ COVID-19 subsequently became a regional public health crisis that challenged hospitals’ ability to handle large volumes of critically ill patients.^⁶ The NewYork-Presbyterian (NYP) Hospital (a 7-campus medical center with more than 2600 inpatient beds) and NYP Regional Hospital Network were major local contributors to the COVID-19 response.^⁷ NewYork-Presbyterian/Columbia University Irving Medical Center (NYP-Columbia), 1 of the 2 largest institutions within the NYP Hospital system, began admitting COVID-19 patients in early March 2020. To address the unprecedented need for high acuity care, the institutions implemented a comprehensive response to ensure adequate critical care services for patients admitted to NYP-Columbia.

Before the recognition of COVID-19 in New York City, international centers reported large influx of critically ill COVID-19 patients.^⁴^,^⁸ A centerpiece of NYP’s preparation for COVID-19 thus included the rapid expansion of existing intensive care unit (ICU) capacity. To meet this goal, nonemergency or urgent surgical procedures were placed on hold in March 2020,^⁹^,^¹⁰ increasing hospital capacity, conserving personal protective equipment (PPE), allowing operating rooms (ORs) to be repurposed into alternative care areas, and creating a reserve workforce of clinicians who usually work in the OR suites. Subsequently, intensivist-anesthesiologists and various hospital divisions converted the dormant OR space into an ICU with an 82-bed capacity.

The newly formed operating room intensive care unit (ORICU) soon became the largest ICU at NYP-Columbia. Here, we describe the creation of the ORICU, including repurposing the physical OR space, ensuring appropriate staff coverage, and adapting anesthesia machines for use as ICU ventilators. Following closure of the ORICU, we sought to estimate the survival rates of patients who had been cared for in the ORICIU, and to compare these outcomes to other reports.

METHODS

The ORICU was created to deliver high-quality critical care in a novel ICU environment to patients suffering from COVID-19. Following closure of the ORICU, we performed retrospective survival analysis of the ORICU cohort. This study was approved by the Columbia University Irving Medical Center (CUIMC) Institutional Review Board (IRB). The requirement for written informed consent was waived by the IRB.

Repurposing the Physical Space

NYP-Columbia has 32 ORs. Eight of these were reserved for urgent surgical procedures that took place during the COVID-19 pandemic. The remaining 23 ORs were converted into the ORICU (1 OR was under renovation). Each OR housed between 3 and 6 patients in individual ICU beds, depending on physical space and the availability of medical gas connections. Generally, 3 or 4 ORs share a central Core, which normally stores procedural supplies. Figure 1 demonstrates the layout of 1 floor of the ORICU, with individual rooms and patient orientation in relation to each other.

Figure 1.:
Representative layout of 1 floor during the creation of the ORICU. A similar figure was used in operation of ORICU for handoff and throughput management. Cores A–C were each formed by a group of 4 ORs; each core had up to 12 total patients. Core G was created from OR 12A and cared for up to 6 patients at a time. Individual patient beds are identified by OR and bed assignment (eg, 1-1) and paired with a specific ventilator, designated “A” if an anesthesia machine and “I” if a typical ICU ventilator. Solid filled green circles denote occupied ORICU beds. Light-green filled green circles denote beds that are awaiting arrival of an already-identified patient. Empty green circles denote beds that are awaiting patient assignment. Empty black circles denote locations being prepared for ICU capability. Green filled red circles denote patients who are receiving CRRT. The second floor of the ORICU, not depicted here, was similar in design but with slight differences in size and number of ORs. CRRT indicates continuous renal replacement therapy; EMR, electronic medical record; ICU, intensive care unit; OR, operating room; ORICU, operating room intensive care unit.

Placing multiple patients into single ORs presented significant challenges. The most immediate of these was the need to continuously monitor individual patients. To address this issue, we made use of the existing connections in the OR for 1 patient and procured separate monitors for others sharing that OR. We coordinated with information technologists to ensure that the monitoring system recognized the added beds and that recorded data were automatically transferred to centralized storage and the electronic medical record. Monitoring display stations were placed in each core so that vital signs could be monitored remotely. Because anesthesia machine alarms were inaudible through the door separating the OR and the central core, ventilator data were incorporated into the monitoring display. Nevertheless, the presence of multiple critically ill patients in a confined space generated noise pollution, which made it difficult to assess the importance of any 1 alarm. Ultimately, we used a staffing approach that included the near-constant presence of a health care worker (HCW) in each OR. A collaborative approach to patient care was adopted, such that clinicians assigned to a given OR would assist with care of all patients in that OR. When a device emitted an alarm, the HCW in the room would evaluate that specific alarm and request assistance as needed.

The adequacy of medical gas supply was also relevant in accommodating a large volume of patients with respiratory failure in our ORs. Although almost all of our ORs had adequate wall supply, a few rooms required splitting of pipeline sources at the location of wall outlets (which remained adequate for patient care). One room did require the biomedical engineers to increase available medical gas supply (simple splitting at wall outlets was insufficient). This same room required use of portable suction systems for patient care, as wall-based sources were of inadequate pressure.

Another challenge in the ORICU was an inability to drain effluent from continuous renal replacement therapy (CRRT) devices. This limitation was particularly relevant given the high rate of acute kidney injury in COVID-19.^¹¹ At first, effluent was drained into large, clean disposal containers. After collaborating with facility engineers, we were eventually able to route effluent into drainage pipes that were previously unused but built into the walls of each OR.

Because ORs are typically positive pressure spaces where air flows from the room itself into external spaces, managing COVID-19 patients in such rooms without increasing the risk of HCW exposure was technically challenging. To reduce the risk of exposure, we worked with NYP engineers to convert each OR into a negative pressure room using antiviral filtered fans to pump air from each OR through a hole created in each OR door. The fans generated additional noise that increased the difficulty of hearing alarms, and hence the importance of central alarm monitoring.

In principle, caring for an intubated COVID-19 patient receiving mechanical ventilation using a closed system does not require airborne precautions. However, each OR contained multiple intubated patients with additive risks of inadvertent extubation and endotracheal tube cuff leaks. Thus, we mandated airborne, contact, and droplet isolation within the ORs at all times. Stations for donning and doffing of PPE were placed at the entrance and exit of ORICU cores. Our institution adopted a policy requiring that surgical masks be worn over N95 masks, which could then be reused as the exterior was kept clean of droplets. Gloves were changed when moving between patients in a given OR. Gowns were changed when moving between patients with additional contact isolation. These conservation efforts ensured HCWs always had access to new PPE when needed.

The ORICU space did have unique risks not found in typical ICUs. In usual circumstances, patients who carry multidrug resistant organisms (MDROs) are physically isolated from other patient groups.^¹² To maintain this degree of isolation, we did not group patients who carried MDROs on admission with those who did not. Similarly, we did not consider sex when grouping patients, which allowed for expedited transfers. We avoided the potential for misidentification of neighboring patients using highly visible tape, mechanical barriers, and patient-name labels on each patient’s bed. Although patient-name displays may compromise patient privacy, we assessed that the safety benefit outweighed the risk of loss of confidentiality. For extubated patients (often grouped in rooms with intubated patients), we judiciously used privacy screens to minimize visibility of other patients.

Educating and Redeploying the Workforce

The quick expansion of the ORICU was made possible by the early adoption of a pandemic staffing model informed by both the Society of Critical Care Medicine (SCCM)^¹³ and novel work by NYP.^¹⁴ We further modified this model in the ORICU by employing a hybrid staff arrangement, in which duties normally performed by multiple individuals were performed by a single care provider. Examples of this included that of certified nurse anesthetists performing duties incorporating elements of both bedside critical care nursing and the medical team and the development of the Ventilator Specialist Team discussed below.

Each OR contained between 3 and 6 patients and was covered by 1 clinician, termed “First Call.” This role was filled by interns, junior residents, nurse anesthetists, and physician assistants. A “Second Call” role was then filled with senior residents and noncritical care fellows who supervised 3 or 4 First Call team members. Each ORICU core was then supervised by an “ICU Lead” who functioned as a typical critical care attending, managing approximately 12 patients. Because of the extenuating circumstances, critical care fellows at our institution were granted attending privileges and were allowed to act as ICU Leads. ICU Leads were overseen by a critical care–trained “ICU Oversight” intensivist, who typically supervised the care of approximately 40 patients at the peak of the ORICU census.

These roles were consistent during both day and night shifts. However, at night, 1 ICU Lead covered up to 2 cores (depending on patient census). At all times, trainee assignments adhered to duty hour regulations outlined by the Accreditation Council for Graduate Medical Education.^¹⁵ A typical trainee work week would consist at most of 5 shifts, each lasting 12 hours in duration. To maintain continuity of care, attending physicians generally worked 3–4 twelve-hour shifts on consecutive days to balance the need for frequent handoffs with the considerable stress of high acuity critical care management in an atypical environment.

A hallmark of our ORICU staffing model was the incorporation of nonintensivist physicians into ICU Lead roles. A focused didactic program on critical care medicine was designed for ICU Leads to ensure that delivered care met standard-of-care benchmarks. In addition to completing that program, nonintensivist ICU leads also underwent a mini-apprenticeship by rounding with intensivists for 1–3 days. Finally, we created and posted infographics in rounding areas to ensure that treatment adhered to best practices (Supplemental Digital Content 1, Figure, https://links.lww.com/AA/D268).

To facilitate the delivery of supportive ICU care, we formed several specialty teams within the ORICU. A Ventilator Specialist team, composed of anesthesiology trainees and attendings, was formed in response to an institution-wide shortage of respiratory therapists. This team performed various tasks traditionally addressed by nurses, physicians, and respiratory therapists, including suctioning of secretions, measuring airway pressures, changing filters, and adjusting ventilator settings. Additionally, a Prone Positioning Team, composed of pulmonologists, physical therapists, and nurses, physically repositioned patients who met defined criteria for prone positioning trials. Finally, a Surgical Workforce Access Team (SWAT), composed of procedural subspecialists, performed procedures such as vascular access, chest tube insertion, and other bedside interventions. Some of these proceduralists also participated in a dedicated Tracheostomy Team. Unlike the Ventilator Specialists, the Prone Positioning, SWAT, and Tracheostomy teams did not work exclusively in the ORICU but provided consultative care throughout NYP-Columbia. Other dedicated roles, including physical and occupational therapists, pharmacists, nutritionists, and others worked in the ORICU in shifts comparable to those assigned in traditional ICUs throughout NYP-Columbia. Cardiopulmonary arrests were managed by clinicians assigned to care for the patient in question rather than specific “code” teams; staff were required to be wearing proper PPE before assisting with an arrest.

In mid-March, given the wide community spread of the virus, NYP issued visitation restrictions which separated patients from their families throughout their hospital stay. Communicating with families was thus challenging given the lean staffing model of the ORICU and the physical environment. To provide updates to the families of ORICU patients, we created a Family Liaison Service, composed of health care providers unable to deliver direct patient care due to their own personal risks of exposure. Family Liaisons were briefed by the ICU Leads daily and were generally assigned to “follow” patients throughout the course of their hospital stay to provide continuity in an otherwise frequently shifting clinical care paradigm.

Adapting Anesthesia Ventilators for Critical Care

The availability of mechanical ventilators was a significant concern during ORICU development.^¹⁶ We increased NYP-Columbia’s ventilator supply by using anesthesia machines as the default ventilator of choice for ORICU patients. A broad range of machines were utilized at our institution including the following models: Drager Apollo and Perseus (Draeger, Inc., Telford, PA), Datex-Ohmeda Aestiva 3000, Aestiva S/5, and Aespire, GE Aisys, Aisys CS2, Avance CS2, Avance CS2 Pro, and a single GE Carestation 650 (GE Healthcare, Chicago, IL). Traditional ICU or transport ventilators were used when the anesthesia machine in a particular location was unable to deliver the mode needed for care of the patient, when the patient needed to be transferred to another location, when space constraints limited the use of large-footprint anesthesia machines, or when specific essential monitoring requirements were unavailable on a given machine.

In all instances of machine use, we performed baseline tests to ensure functionality, removed volatile anesthetic agents, and delivered fresh gas flows greater than 10 L/min. We connected all machines’ gas scavenging system to Waste Anesthetic Gas Disposal outlets to prevent accumulation of pressure in the evacuation system. Bellows-type machines were driven with oxygen. We changed carbon dioxide absorbent when color change indicated exhaustion or when there was clinical concern for rebreathing. We used anesthesia machine gas analyzers or separate bedside carbon dioxide monitors for all patients. Typically, the carbon dioxide absorbent canisters were changed once or twice per day.

We observed some challenges as we included the anesthesia machines into the ORICU workflow. Most importantly, few respiratory therapists were available during the ORICU existence—and all of them were unfamiliar with anesthesia machines. To meet patient care needs, we then formed the aforementioned Ventilation Specialist team to deliver safe and effective ventilatory support to all patients.

Additionally, we encountered logistical difficulties with components of the airway circuit itself. To reduce the potential for viral particles to encounter the machine and preserve circuit humidity, we placed a heat and moisture exchange filter (HMEF) between the endotracheal tube and the Y-piece of the ventilator circuit. (The Gibeck Iso-Gard HEPA Light filter [Teleflex, Morrisville, NC] was the most frequently used device at our institution.) Despite the use of high fresh gas flows, the reaction of carbon dioxide with absorbent generated additional humidity. The HMEF increased dead space ventilation (approximately 80 mL for this filter). We frequently changed the HMEF because it was prone to soilage caused by the combination of mucus production, pulmonary edema, water created by the chemical reaction inherent in the carbon dioxide absorption system, and saline used to mobilize mucus. We did not place the HMEF in the expiratory limb of the patient’s disposable circuit or use multiple filters, because we had limited HMEF supplies and were uncertain of the implications of viral transmission into gas analyzer modules.^¹⁷

Although anesthesia machines performed adequately for most patients, they were inadequate in patients who had enough improvement in lung mechanics to allow for ventilator weaning. Issues encountered with ventilator weaning in the ORICU included the increased dead space and added resistance caused by HMEFs and the lack of typical weaning modes, such as pressure support ventilation (PSV), on some machines. To reduce the need for heavy sedation and to minimize patient-ventilator dyssynchrony, we used PSV relatively early in the weaning phase. We adjusted inspiratory support as needed to accommodate the increased dead space and airway resistance added by the HMEF. We did not perform T-piece trials to reduce unnecessary patient fatigue and to prevent viral aerosolization. Extubation or tracheostomy occurred per the discretion of the ICU care team. In circumstances where weaning was unsuccessful on an anesthesia machine, ventilators were exchanged for a typical ICU or transport ventilators (Puritan Bennett 840 [Medtronic, Minneapolis, MN], Servo U [Getinge AB, Gothenburg, Sweden], or Carefusion LTV 1200 machines [CareFusion, Yorba Linda, CA]). When ventilators or HMEFs needed to be changed, the endotracheal tube was clamped during the process to avoid alveolar decruitment.

Anesthesia machines typically require a daily system check to assess ventilator function. We elected not to perform these checks, because we did not have sufficient back-up ventilators to take machines offline for a daily check. Continuous use of the machines resulted in occasional flow sensor failures caused by accumulation of condensation. Additionally, we noted that high fresh gas flow rates led to errors in measured tidal volume; volumes could differ by as much as 250 mL/breath when using pressure control mode. Despite these challenges, we identified no critical events related to anesthesia machine failure.

Patient Candidacy and Timeline for Care in the ORICU

Criteria for admission to the ORICU included documented SARS-CoV-2 infection and respiratory failure requiring invasive mechanical ventilation. Patients were transferred to the ORICU from other ICUs, inpatient wards, the emergency department, and other institutions. In the first week of the ORICU’s existence, patients requiring high-dose vasopressors were preferentially not admitted to ORICU. In addition, patients requiring CRRT were not admitted until March 27th because this therapy was not available in ORICU until then due to technical limitations. However, as ICU space became limited, vasopressor doses were not a limitation to ORICU admission. Pulmonary vasodilators and prone positioning were available in the ORICU. Patients with stroke, heart failure, or receiving extracorporeal membrane oxygenation (ECMO) were not admitted to the ORICU but rather to specialized ICUs.

Patients who were admitted to the ORICU remained there until they were able to be transferred to a less acute care area, to another ICU, or died. Approximately 7 weeks after the initial conversion of the OR into ICUs, declining new admission rates coupled with conversion of other inpatient areas to ICU-capable areas permitted a reduction in ORICU census. To allow planning for increased elective surgery, the ORICU was closed before other atypical ICUs. Approximately 50% of ORICU beds were removed from hospital capacity by May 3, 2020. All patients were transferred out of the ORICU by May 14, 2020.

Statistical Methods

All patients who were admitted to the ORICU were included in outcome analysis. We followed patients from the time of their admission to the ORICU and throughout their in-hospital course until death or discharge. Patients who died or were transferred out of the ORICU on the same date as their ORICU admission were considered to have had an ORICU length of stay (LOS) of 0.5 days. We right-censored follow-up time on May 14, 2020, the date the ORICU closed, and created Kaplan-Meier cumulative survival curves. We hypothesized that older and overweight individuals were at greater risk of in-hospital mortality. We were unsure of the potential risk of death in men versus women but were concerned that sex differences may play a role in the risk of death. Thus, we estimated hazard ratios (HR) for death with a Cox proportional hazard regression model using age, sex, and body mass index (BMI) as covariables. We also used a separate Cox proportional hazard regression to compare survival in patients who were <65 years of age versus those who were older, adjusting for age and BMI. All analyses were performed using Stata (version 16.1; StataCorp, College Station, TX).

RESULTS

NYP-Columbia, with a baseline capacity of 117 ICU beds, reached a peak ICU capacity of 257 ICU patients. At peak census, the ORICU contained 79 patients. At 1 point, the ORICU census represented 34% of all patients in ICUs at NYP-Columbia. Figure 2 depicts the ORICU patient census over time, in relation to other NYP-Columbia ICUs.

Figure 2.:
Stacked area chart showing patient census among different ICU locations at NYP-Columbia during high volume of COVID-19 admissions, March 27 to May 14, 2020. Patients received care in traditional ICU locations, in the ORICU, and in other novel ICU areas other than the ORICU. At 1 point in time, the ORICU cared for 34% of all ICU patients at NYP-Columbia. Time, beginning shortly after ORICU inception, is marked along the x-axis until ORICU closure on May 14, 2020. The blue region indicates the total number of patients admitted to traditional ICU environments. The orange region indicates the total number of patients admitted to novel ICU areas other than the ORICU. The gray region indicates the total number of patients admitted to the ORICU. Each region is additive, such that the total number of all patients admitted to ICUs is indicated along the y-axis for a given date on the x-axis. For example, on April 11, 2020, there were 113 patients in traditional ICU areas, 24 patients in non-ORICU novel ICU areas, and 79 patients in the ORICU. COVID-19 indicates coronavirus disease 2019; ICU, intensive care unit; NYP-Columbia, NewYork-Presbyterian Hospital/Columbia University Irving Medical Center; ORICU, operating room intensive care unit.

A total of 133 patients, 40 of whom were female, were cared for in the ORICU during its existence from March 24 to May 14, 2020. The median age of all ORICU patients was 63 years (standard deviation 12.6). The median LOS in the ORICU was 16 days (IQR = 8–32). Nonsurvivors had a median ORICU LOS of 10 days (interquartile range, IQR = 3–17); survivors had a median of 27 days (IQR = 12–35). Three patients had a <24-hour LOS. Two patients died on the day of admission; 1 required initiation of ECMO and was transferred to another ICU shortly after ORICU admission. At the time of ORICU closure, 27 (20.3%) had been discharged alive from the hospital, 17 (12.8%) were receiving care in a non-ICU inpatient ward, and 34 (25.6%) were receiving care in an ICU (Table).

Table 1. - Characteristics and Clinical Outcomes of Patients Admitted to the NewYork-Presbyterian/Columbia University Irving Medical Center ORICU During COVID-19

	Entire cohort	Survivors	Nonsurvivors
Patients, no.	133	78	55
Median age, y (SD)	63 (12.6)	60.5 (11.2)	68 (13.2)
Female sex, no. (%)	40 (30.1)	26 (33.3)	14 (25.5)
Mean BMI, kg/m² (SD)	29.9 (6.4)	29.9 (6.1)	29.9 (6.9)
Median length of stay, d (IQR)^a	16 (8–32)	27 (12–35)	10 (3–17)
Discharge destination
Outpatient, no. (%)		27 (20.3)	-
Inpatient ward, non-ICU, n (%)		17 (12.8)	-
Inpatient ward, ICU, n (%)		34 (25.6)	-

Abbreviations: BMI, body mass index; COVID-19, coronavirus disease 2019; ICU, intensive care unit; IQR, interquartile range; ORICU, operating room intensive care unit; SD, standard deviation.

^aLength of stay begins on admission to the ORICU. It does not include time in other inpatient wards, ICUs, or the emergency department.

Figure 3.:
Kaplan-Meier survival curve in days since ORICU admission for each patient. Among all patients, the estimated probability of survival 30 d from ORICU admission was 0.61 (95% CI, 0.52-0.69). Time, measured in days since individual patients were admitted to the ORICU, is plotted along the x-axis. The cumulative survival rate at each given time point is plotted along the y-axis. CI indicates confidence interval; ORICU, operating room intensive care unit.

Figure 4.:
Kaplan-Meier survival curves comparing the cohort of patients who were <65 y of age with those who were ≥65 y of age. At 30 d post-ORICU admission, the probability of survival for patients aged <65 was 0.74 (95% CI, 0.62-0.83), compared to 0.47 for patients who were ≥65 (95% CI, 0.34-0.59), P < .001 by log-rank test for equality of survivor function. Time, measured in days since individual patients were admitted to the ORICU, is plotted along the x-axis. The cumulative probability of survival at each given time point is plotted along the y-axis. CI indicates confidence interval; ORICU, operating room intensive care unit.

All patients cared for in the ORICU were included in survival analysis. As of May 14, 55 (41.4%) of ORICU patients had died. Among all patients, the estimated probabilities of survival at 10, 20, and 30 days from ORICU admission were 0.82 (95% confidence interval [CI], 0.73-0.87), 0.67 (95% CI, 0.58-0.74), and 0.61 (95% CI, 0.52-0.69), respectively (Figure 3). The median length of follow-up for all patients was 33 days (IQR = 12–42), 41 days for survivors (IQR = 34–45), and 11 days for nonsurvivors (IQR = 3–19). Increased age was significantly associated with risk of mortality, such that there was a 5% increase in the expected risk of death relative to a 1-year increase in age (HR = 1.05, 95% CI, 1.03-1.08, P < .001). Male sex (HR = 1.68, 95% CI, 0.90-3.14) and BMI (HR [for a 1-unit increase] 1.02, 95% CI, 0.98-1.07) were not significantly associated with mortality. Patients who were ≥65 years were an estimated 3.17 times more likely to die than younger patients (HR = 3.17, 95% CI, 1.78-5.63, P < .001) when adjusting for sex and BMI. At 30 days post-ORICU admission, the probability of survival for patients younger than 65 was 0.74 (95% CI, 0.62-0.83), compared to 0.47 for patients ≥65 (95% CI, 0.34-0.59), P < .001 by log-rank test of equality of survival function, as displayed in Figure 4.

DISCUSSION

Despite the challenges inherent to the creation of the ORICU, our preliminary outcomes are consistent with reported data from other centers.^¹¹^,^{^18–24} When adjusting for sex and BMI, patients admitted to the ORICU had an estimated 61% probability of survival to 30 days. Our unadjusted overall mortality rate of 41.4% was calculated at the time of data censoring, which coincided with ORICU closure, and represented a median follow-up time of 33 days after ORICU admission. Preliminary reports from ICUs in other centers with a high volume of COVID-19 admissions have identified mortality rates ranging from 17% to 62%, with higher mortality in older persons.^¹¹^,^{^18–24} In our own institution, the overall in-hospital mortality rate in all critically ill COVID-19 patients was similar to that of the ORICU (39%).^¹¹ However, these data encompass outcomes for 2 locations (NYP-Columbia and an affiliated community-based hospital) and the observation periods overlap with ORICU patients by only 9 days (March 24 to April 1). Furthermore, pre-ORICU care in other ICUs may have encouraged identification of patients who were stable and therefore favorable for transfer to the ORICU. Further analysis of disease severity and processes of care (eg, use of CRRT) would be necessary to directly compare mortality rates.

Importantly, we may have offset some of the challenges associated with ORICU care with the adoption of our hybrid staffing model, such that multiple clinicians had overlapping skill sets and were able to provide care outside of their more narrowly defined traditional roles. Ultimately, the ORICU’s outcomes are comparable to those reported elsewhere, suggesting the ORICU delivered effective critical care despite the logistical challenges.

Other authors have discussed creation of an ORICU, though without inclusion of clinical outcomes.^²⁵ We recognize that local needs will always guide an institution’s response; outcomes in the context of extreme resource strain will be a function of both patient characteristics and the way critical care is delivered. The ORICU was created to address the large patient surge experienced in New York City with the expectation that this would be a temporary space. We observed significant patient movement between sites as well as care in novel areas, including other wards that were converted to ICU environments. Thus, aggregated clinical outcomes of patients cared for in the ORICU reflect both the influence of care in multiple sites within NYP-Columbia as well as the uncertain implications of resource scarcity during this public health crisis. These factors limit our ability to further characterize the influence that ORICU care had on patient outcomes.

Ultimately, our experience and outcomes suggest the creation and use of this novel ICU was an effective approach in a time of crisis. Other centers may benefit from incorporation of some elements of our approach if future scenarios require rapid ICU expansion.

ACKNOWLEDGMENTS

This paper is dedicated to Desmond Jordan, MD (1953 - 2020), an inspiring educator, a friend to all, and a master clinician. Desmond was committed to the betterment of all who were fortunate to meet him. He was a pivotal piece of both the ORICU and our Department. We will miss him dearly.

DISCLOSURES

Name: Aaron M. Mittel, MD.

Contribution: This author helped with ORICU creation, clinical care of ORICU patients, data collection, data analysis, and manuscript preparation.

Name: Oliver Panzer, MD.