U.S. Field Hospitals: A Study on Public Health Emergency Response to COVID-19

Field hospitals are concentrated in states with International airports and high population density

First, we mapped the existing and planned field hospitals to visualize their spatial distribution (Figure 1). It was not surprising to see that the field hospitals clustered around major international airports, such as San Francisco, Los Angeles, Seattle, Chicago, Dallas, and New York, that have connections to China—where the first epicenter was—and to Europe—where most transatlantic import occurred—especially before the international travel restriction was published on the 31^st of March^7-8. In addition to imported cases, the coronavirus then spread out to other cities through domestic travels and community-based transmission. Where the field hospitals indexed the most affected states, their distribution highly resembles that of imported measles outbreak predicted according to 1) international travel volume and port of entry, 2) incidence rate at the origin, and 3) the population of surrounding county⁹. If an international epidemic originating from Europe or Asia occurs in the future, these states should the first ones to start public health emergency response procedures.

• Download figure
• Open in new tab

Figure 1: Visualizing the spatial distribution of field hospitals

1A: Current and planned field hospitals and shelters mapped to their location using Power BI. Each dot represents one field hospital regardless of its capacity. 1B: Density map indicating total number of field hospital beds in each state.

Next, the question becomes, when the local epidemics developed into national pandemic, what determined when and where to build field hospitals? While reading through news reports, it was hard to ignore that the building and opening of field hospitals is a very dynamic process to accommodate the evolving curve of existing confirmed cases. Besides the field hospitals affiliated to and staffed by local healthcare system, recruiting staff can be challenging during the pandemic. Therefore, many field hospitals were constructed but remained closed until necessary to reserve medical resources. Similarly, in states where the outbreak was not as urgent, the potential sites were outfitted but no construction would start until situation worsens; alternatively, if the situation improves the construction would be canceled. Therefore, the decision must be dependent on the live feeds of the growing numbers. For convenience, we collected the U.S. COVID-19 statistics on April 15th, the national peak predicted by IHME, shown in Table 2.

The field hospital opening is more Influenced by accumulated confirmed cases in each state

The timeline histogram shows that the greatest number of field hospitals opened during the 5^th week, where most state met bed demand peaks (Table 1, Fig 2A). Intuitively, the more urgent the COVID-19 outbreak was in the state, the earlier the field hospitals would be built. For COVID-19, the recovery time is calculated from the first day showing symptom or having tested positive to the last day of testing negative for two consecutive days^10-11. During the early phase of highly contagious disease with doubling time of 5 days and recovery time of 10 days, the recovery and death rate is almost negligible, and it is essential to contract trace, identify and isolate the infected individuals to slow down the spreading¹². Therefore, the confirmed cases and the incidence rate, defined as number of infected individuals per 100,000 persons, should be more correlated with the opening time of field hospitals. Because larger population will produce larger number of confirmed cases, it is perceived that the population-independent infected rate would be a more accurate indicator of when to build the field hospitals. However, when we indexed each state with a ranking number, where either accumulative confirmed case number or incidence rate of all 50 state and Washington DC on Apr. 15^th was ranked from largest to smallest, and plotted against the opening date of the field hospitals, a stronger correlation was observed between ranking of accumulative confirmed cases and the field hospital opening date (Fig 2B and D). For example, California had an alarming number of accumulative confirmed cases of 26,686, ranking 6^th in the nation; however, by incidence rate it was ranked 35^th, indicating that the outbreak in the state was not as bad as it appeared even though the incidence rate was exceptionally high in certain cities. At the state level, the additional field hospital beds seemed unnecessary but at city level, they were essential. Consequently, a more case-specific micro-scale analysis is required to describe the discrepancy between the ranking of accumulative confirmed cases and of incidence rate. This observation could also be contributed to the fact that JHU CSSE did not publish the county-level population-corrected COVID-19 outbreak until Apr. 12^th. Even though the incidence rate could be calculated with simple math and more reliable, the mounting number of accumulative confirmed cases may have had a heavier impact on decision-making psychologically.

• Download figure
• Open in new tab

Figure 2: The accumulative confirmed cases is a more important factor determining when and where to build the field hospitals

2A: Histogram of field hospital opening time, binned to weeks into National Emergency. 2B and D: Urgency level, approximated by accumulated confirmed cases or incidence rate ranking, against opening date of field hospitals with linear fit. Radii of circles indicate the capacity of each field hospital. 2C: Most opening dates of field hospitals are clustered around y=0 or the date of predicted peak demand in the state they belong to. The first field hospital in each state on average opened before or by peak demand date. Each dot represents one field hospital. All field hospitals shown in blue, the first field hospital to open in each state shown in orange.

Field hospitals in each state opened promptly to meet peak demand

Then we ask the question—were the field hospitals built by the date of peak demand? Knowing that the accumulative confirmed number is the determining factor of building these field hospitals, we plotted the difference between peak demand peak and field hospital opening date against the ranking of accumulative confirmed cases (Fig 2C). A positive y-value indicates that the field hospital opened before demand peak arrived, whereas a negative value suggests otherwise. From the scatter plot, we could conclude that most hospitals were opened around the day of peak demand. When the first hospital in each state to open was highlighted in orange, we could see that on average they were put to service well before or around the day of peak demand. It is noteworthy that some larger field hospitals, such as the Javits New York Medical Station, opened before construction ended to account for the imminent bed shortage (Table 1). In general, field hospitals in each state opened promptly to meet peak demand.

The total number of beds in field hospitals is most correlated to the difference between incidence rate and bed rate in each state

In parallel to the question when and where to build the field hospital is how many additional beds are needed? The classic SIR (Susceptible-Infected-Recovered) model of infectious disease suggests that the maximal infected rate I_max/N can be estimated as where I_max is the maximal number of infected individuals, N the population in the defined area, and R₀ the basic reproduction number of the infectious disease¹³. At first R₀ was estimated to be 3.65, which means that on average, 2.55 individuals will be infected by any given infected person, however after the implementation of social distancing, the effective R₀ is estimated to be 2.55 for COVID-19 in the U.S.¹². Because the number of available hospital beds is directly proportional to population (Fig 3A), and the hospitalization rate (number of patients hospitalized over number of confirmed cases) is known to each state, the maximal number of additional beds required should be more or less proportional to the difference between maximal infected rate and bed rate, calculated as the number of beds per 100,000 persons. When the total number of field hospital beds was plotted against (infected rate – bed rate) (Fig 3B), indeed the correlation is stronger than that against any other statistics, followed by state population and accumulative confirmed cases (Supplemental Fig 3). This phenomenon should be largely credited to the IHME projection model, which provided state-specific guidelines in preparing for the upcoming surge.

• Download figure
• Open in new tab

Figure 3: The number of field hospital beds is mostly correlated to incidence minus bed rate, a proxy for the additional beds required predicting using SIR model

3A: The number of available beds before COVID-19 is proportional to the population in each state. 3B: The number of field hospital beds in each state is mostly correlated to incidence minus bed rate.

Different types of field hospitals were utilized differentially

When the field hospitals had been constructed with the right capacity and ready to receive patients around demand peak, the final question came down to: what purpose should they serve? Should they treat routine non-COVID-19 patients—such as those with kidney failure, heart disease, or cancer—to restore normal operation of the healthcare system, or should they take up COVID-19 patients when there is not enough space in the local hospitals? Should there be intensive care units or for mild-symptom and recovering patients only? While the local hospitals have been toiled by the flood of COVID-19 patients, it seemed like a cost-effective plan to designate field hospitals as overflow facilities to care for mild and recovering COVID-19 patients, where neither extra medical equipment, such as ventilators, nor specially trained staff were required. However, the follow-up news coverage may suggest otherwise (Table 3).

As of end of April, among the 75 surveyed field hospitals, 43 opened, 29 remained closed or under construction, and 3 was canceled; out of the 43 opened field hospitals, 2 had positive news reports, 6 negative, and the rest with no additional follow-up news coverage (Fig 4A). The major concerns included 1) COVID-19 outbreak in a non-COVID-19 facility, 2) low admission rate, and 3) strict transfer rules to accept patients from local hospitals. When the field hospitals were not required any more or could not contain the outbreak, patients would be discharged, and hospital shut down. The relationship between the negative commentary categories is depicted in the Venn diagram (Fig 4B).

• Download figure
• Open in new tab

Figure 4: The effectiveness of decision making and the utilization of resources

4A: The status of all field hospitals. 4B: Reasons of negative news report on the opened field hospitals. No conflicting reports were found. 4C: Positive and negative news coverage assigned to different quadrants of opened field hospitals. Red indicates negative commentaries and green positive. The numerator is the number of reported field hospitals, and denominator the total number of opened hospitals in each category. Each quadrant is labeled with Roman numerals on the upper right corner.

When looking at the Table 3 more closely, a pattern could be found between the property of field hospitals and the commentary they received. Therefore, we divided up the opened field hospitals into four quadrants depending on the patients they received (COVID-19 vs non-COVID-19), and the affiliation (military vs non-military). Then the reported field hospitals are assigned to each quadrant (Fig 4C). The non-military-built field hospitals are mostly affiliated or sponsored by local healthcare systems and receive direct transfers whereas the larger scaled military-built field hospitals are mostly freestanding facility accepting patients from the entire region. Therefore, it is an indicator of how well the field hospitals are integrated into the local healthcare system.

Javits Medical Center and TCF Center fell into quadrant I (COVID-19, military) and they both had low admission rate due to strict transfer rules. Even though many patients requested transfer, only a small portion was granted admission. The two USNS cruises and CenturyLink belong to quadrant II (non-COVID-19, military) and they all had low admission rate. Both USNS cruises had staff or patients tested positive for COVID-19. USNS Comfort discharged all patients and left New York after 3 weeks of service and USNS Mercy dispatched crew members to assist local healthcare facilities. CenturyLink in Seattle received 0 patients 3 days after opening and was soon closed. San Diego Convention Center Shelter in quadrant IV also had 2 confirmed COVID-19 cases among the homeless. The two well utilized field hospitals—Central Park Field Hospital and Boston Convention Center—both fall into quadrant III (COVID-19, non-military). Central Park Field Hospital, especially, is more like an overflow ward of Mt. Sinai Hospital with full capacity to provide intensive care. It admitted 142 patients one week after opening and hospitalized at least 50 patients at all time.

In general, the non-COVID and/or military-build field hospitals appeared less utilized than planned. However, from the perspective of public health, the low utilization rate can have mixed implications.

Low utilization rate of field hospital could be due to various factors

In response to the COVID-19 pandemic, many countries have built makeshift hospitals. Like many field hospitals in the U.S., the NHS (National Health Service) Nightingale Hospitals in the UK also treated very few patients. The Nightingales are also designed as overflow hospitals when the surge does occur. Therefore, the underutilization demonstrate that the local health systems had coped with the extra pressure COVID-19 brought¹⁴. Similarly, the low utilization rate in the U.S. field hospitals, which are also built as the last resort during the medical crisis, could be due to that public health interventions, such as social distancing and wearing masks, are effective to “flatten the curve” and the infected population did not grow to the predicted value¹⁵. According to the IHME model, the estimated maximal daily infections in the U.S. were 260K on March 29^th, however the actual number peaked on April 8^th with 32K confirmed cases⁵. Consequently, the field hospitals prepared for the surge were no longer needed.

On top of the lowered number of infected cases, procedural and socioeconomic factors further decrease the utilization of field hospitals. CDC encourages patients with mild symptoms to practice self-isolation and recover at home, which decreases the proportion of infected cases entering the medical system¹⁶. In addition, loss of health insurance due to unemployment and confusion of coverage also discourage patients to seek medical attention when symptoms are mild^17-18. When patients are hospitalized but do not need intensive care, the transfer qualification, specifically in Javits Medical Station and TCP Center Field Hospital, again strictly limits the number of patients admitted to the field hospital even though large volume of patients requested for transfer^75,78.

COVID-19 intensive care should be prioritized in the epicenter

The most significant difference between the U.S. and UK field hospitals is that the NHS prioritized the COVID-19 patients needing critical care over mild and recovering patients, with the non-COVID-19 patients coming last. The most important medical center, the ExCel Nightingale, specifically was commissioned with a purpose to take up unconscious, ventilated patients who could not take direct referrals from the community. It was not a “step down” facility, but rather an alternative. Whereas in the U.S., most COVID-19 field hospitals only take mild and recovering patients and had low admission rate. In contrast, the Central Park Field Hospital, equipped with intensive care unit have filled the beds with patients and are running around the clock. Comparatively speaking, it may be a better idea to accentuate the resources on smaller number of intensive care beds instead of on large amount of non-intensive care beds, when most patients can recover without hospitalization¹⁹.

Non-intensive COVID-19 field hospitals can serve as entry instead of step-down facilities

The FangCang field hospitals built in Wuhan, China were also designed for mild and recovering COVID-19 patients but had high occupancy. The major difference is that instead of taking patient outflow, FangCang field hospitals triage and quarantine COVID-19 positive patients before the more severe individuals are transferred to designated healthcare facilities²⁰. Not only can this workflow free up the space in local hospitals, the confirmed cases can be properly quarantined to prevent further in-person transmission. Whereas many COVID-19 patients with mild symptoms were self-quarantined at home and may not strictly follow the rules, group quarantine patients in field hospital appears a more effective way to utilize the resources.