20% of adults are infected in
18 months
20% of adults are infected in
12 months
20% of adults are infected in
6 months
40% of adults are infected in
18 months
40% of adults are infected in
12 months
40% of adults are infected in
6 months
60% of adults are infected in
18 months
60% of adults are infected in
12 months
60% of adults are infected in
6 months
20% of adults infected in …
40% of adults infected in …
60% of adults infected in …
Percentage of occupied hospital beds that would need to be emptied or added
Percentage of occupied hospital beds that would need to be emptied or added
Percentage of occupied hospital beds that would need to be emptied or added
Source: Harvard Global Health Institute estimates of hospital bed capacity and Covid-19 infections
A new Harvard analysis shows that many parts of the United States will have far too few hospital beds if the new coronavirus continues to spread widely and if nothing is done to expand capacity.
In 40 percent of markets around the country, hospitals would not be able to make enough room for all the patients who became ill with Covid-19, even if they could empty their beds of other patients. That statistic assumes that 40 percent of adults become infected with the virus over 12 months, a scenario described as “moderate” by the team behind the calculations.
These numbers are not exact predictions. In many ways, they reflect a worst-case scenario, since they do not take into account the efforts hospitals can make to quickly increase capacity during an emergency. Around the country, hospitals have begun canceling elective operations and speeding home patients with less critical ailments. Those efforts could increase the number of free beds available for coronavirus patients. In a half-dozen interviews, hospital executives estimated that they could increase their capacity between 20 percent and 70 percent.
Yet the Harvard estimates suggest that the coronavirus outbreak could require significantly more resources than that. In hard-hit Italy, hospitals have been rationing care for older patients who are severely ill with the virus.
“If we don’t make substantial changes, both in spreading the disease over time and expanding capacity, we’re going to run out of hospital beds,” said Dr. Ashish Jha, the director of the Harvard Global Health Institute, which produced the estimates. “And in that instance, we will not be able to take care of critically ill people, and people will die.”
Political leaders have talked about unusual emergency options: With policy changes, hospitals for the military and veterans may be able to open their doors to civilians. The military itself may be able to help build new facilities and staff them. Retired doctors and nurses could be relicensed.
But the estimates in this model show that in many realistic scenarios, the scope of the epidemic may overtake the physical and human capacity of the U.S. health system.
See estimates for each hospital region under different scenarios at ProPublica.
The Harvard researchers considered a range of projections, reflecting how much epidemiologists still don’t know about what we should expect. They considered three estimates for how long the disease would take to make its way through the U.S. population. And they considered three credible estimates for how many adults would eventually become infected with the disease, however long it takes to spread.
One thing is very clear from these numbers: The longer the disease takes to spread, the less severe the crunch that hospitals around the country will face. This effort to stretch out the infections over more time is why so many states and cities are closing businesses and schools and canceling large events.
If coronavirus spreads quickly, many hospitals could become overwhelmed.
Estimated peak
hospitalizations
738,000
total U.S.
hospital beds
265,000
unoccupied
hospital beds
If 20% of adults are infected
If 40% of adults are infected
If 60% of adults are infected
Estimated peak
hospitalizations
738,000
total U.S.
hospital beds
265,000
unoccupied
hospital beds
If 20% of adults are infected
If 40% of adults are infected
If 60% of adults are infected
Estimated peak
hospitalizations
738,000
total U.S.
hospital beds
265,000
unoccupied
hospital beds
If 20% of adults
are infected
If 40% of adults
are infected
If 60% of adults
are infected
Estimated peak
hospitalizations
738,000
total U.S.
hospital beds
265,000
unoccupied
hospital beds
If 20% of adults
are infected
If 40% of adults
are infected
If 60% of adults
are infected
Source: Harvard Global Health Institute estimates of hospital bed capacity and Covid-19 infections
You might have seen charts like this that are hypothetical. These use real information about hospital beds in the United States and real estimates about how many people will need them.
Nationwide, the difference between an epidemic that takes six months and one that takes 18 months could mean the difference between health systems that are stretched and those that are completely overwhelmed, even in the model’s more mild scenario, in which 20 percent of adults contract the virus.
The challenge may be especially severe for critical care services, where demand is expected to outstrip capacity at an even greater rate. These types of services are also harder to scale up because they require both specialized staff and expensive equipment such as ventilators.
The crisis could be even worse for intensive care beds.
Estimated peak
I.C.U. hospitalizations
85,000
total U.S.
I.C.U. beds
32,000
unoccupied
I.C.U. beds
If 20% of adults are infected
If 40% of adults are infected
If 60% of adults are infected
Estimated peak
I.C.U. hospitalizations
85,000
total U.S.
I.C.U. beds
32,000
unoccupied
I.C.U. beds
If 20% of adults are infected
If 40% of adults are infected
If 60% of adults are infected
Estimated peak
I.C.U. hospitalizations
85,000
total U.S.
I.C.U. beds
32,000
unoccupied
I.C.U. beds
If 20% of adults
are infected
If 40% of adults
are infected
If 60% of adults
are infected
Estimated
peak I.C.U.
hospitalizations
85,000 total
U.S. I.C.U. beds
32,000
unoccupied
If 20% of adults
are infected
If 40% of adults
are infected
If 60% of adults
are infected
Source: Harvard Global Health Institute estimates of hospital bed capacity and Covid-19 infections
In places already short of beds, a situation that’s challenging during a slow-moving epidemic could be catastrophic during a fast-moving one. Take, for example, the New York City suburb Hackensack, N.J. If the disease reached only 20 percent of the population over the course of 18 months, Hackensack would still need to increase its number of available intensive care beds 140 percent. In a much more dire scenario, in which the disease spread to 60 percent of the population over a year, I.C.U. capacity would need to surge 11-fold.
Daniel Varga, chief physician executive at Hackensack Meridian Health, said his system had already seen its surge capacities tested when about 10 people with suspected coronavirus came to a hospital at once.
“We had to divert to other hospitals for a period of time and throttle back on elective admissions,” he said.
He estimates that, by canceling elective operations and using a new fleet of recently ordered ventilators, his hospitals could increase capacity by about 20 percent. Designating one hospital in the system to work specifically with coronavirus cases may allow them to go even higher.
Several of the areas that may need the most new beds are close to the Washington State coronavirus outbreak, which has already led to 48 deaths and 787 infections.
But even many of the places that could handle a slower epidemic would be overwhelmed in one that infected many people quickly. Look at Manhattan. If the disease eventually reached 40 percent of all adults and took 18 months to spread, the hospitals would need to add or free up about 40 percent of their occupied beds, according to the estimates. If it took only six months, they would need to find nearly double the number that patients with other health problems currently use.
Gov. Andrew Cuomo of New York announced Tuesday that he expected cases in the state to peak in 45 days, faster than any of the estimates shown in this article. His estimates for available hospital and I.C.U. beds are similar to the numbers in the Harvard model.
The American health system has spent years shedding hundreds of thousands of hospital beds, as procedures now typically require shorter stays. Generally, that’s considered a good thing, but it could leave the country vulnerable in this type of crisis.
NewYork-Presbyterian operates 2,600 hospital beds across the city. It began to delay elective operations last Friday, and has set up stretchers and chairs in empty areas near emergency rooms to create additional seating.
The areas that have significant slack in the system, allowing them to handle an influx of patients, tend to be rural and sparsely populated.
Number of hospital regions that may have to free up (or add) …
Between 1% and 100% of occupied beds
Between 1% and 100% of occupied I.C.U. beds
More than 100% of occupied I.C.U. beds
More than 100% of occupied beds
No I.C.U. beds
If 20% of adults
are infected
Over 18 months
231 (out of 305)
Over 12 months
Over 6 months
If 40% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
If 60% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
Between 1% and 100%
of occupied beds
More than 100% of occupied beds
Between 1% and 100
of occupied I.C.U. beds
More than 100% of occupied I.C.U. beds
No I.C.U. beds
If 20% of adults
are infected
Over 18 months
231 (of 305)
Over 12 months
Over 6 months
If 40% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
If 60% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
Between 1% and 100%
of occupied beds
More than 100% of occupied beds
If 20% of adults
are infected
Over 18 months
231 (of 305)
Over 12 months
Over 6 months
If 40% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
If 60% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
Between 1% and 100
of occupied I.C.U. beds
More than 100% of occupied I.C.U. beds
No I.C.U. beds
If 20% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
If 40% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
If 60% of adults
are infected
Over 18 months
Over 12 months
Over 6 months
Note: There are 306 hospital referral areas in the United States, but there was not enough data to include the Panama City, Fla., area in the analysis.·Source: Harvard Global Health Institute estimates of hospital bed capacity and Covid-19 infections
The Harvard team used data on Covid-19 patients in China to estimate which U.S. areas will face extraordinary stress or be relatively well prepared. One big factor is how many people in each region are older than 65 and thus at the highest risk of needing hospital care. (Other research teams use somewhat different assumptions.)
As Dr. Jha described it, there are regions that look like South Korea, where the population is relatively young and where Covid-19 patients have been able to get care. In these places, the fatality rate from the disease has stayed low.
And there are places that look like Northern Italy, where the population is older, some patients are being triaged from the highest level of care, and the disease appears to be killing a higher proportion of ill patients.
The research suggests the possibility of a particularly worrisome shortage of beds in intensive care units, a type of hospital bed for patients with the most acute illnesses. Evidence from overseas and from Washington State suggests that many patients with severe illness from Covid-19 require assistance from breathing machines called mechanical ventilators. Hospitals have a limited number of the machines and — perhaps more important — a limited number of nurses and respiratory therapists who are trained to care for the patients who require them.
Percentage of occupied I.C.U. hospital beds that would need to be emptied or added
Percentage of occupied I.C.U. hospital beds that would need to be emptied or added
Percentage of occupied I.C.U. hospital beds that would need to be emptied or added
20% of adults are infected in
18 months
20% of adults are infected in
12 months
20% of adults are infected in
6 months
40% of adults are infected in
18 months
40% of adults are infected in
12 months
40% of adults are infected in
6 months
60% of adults are infected in
18 months
60% of adults are infected in
12 months
60% of adults are infected in
6 months
Source: Harvard Global Health Institute estimates of hospital bed capacity and Covid-19 infections
20% of adults infected in …
40% of adults infected in …
60% of adults infected in …
Source: Harvard Global Health Institute estimates of hospital bed capacity and Covid-19 infections
The estimates of total I.C.U. beds from the Harvard team include specialized beds that are normally devoted to cardiac, surgical and trauma patients. There is some ability for hospitals to increase their numbers. If surgical procedures are canceled, ventilators used in operating rooms may be made available. The federal government has a strategic stockpile of ventilators likely to be deployed in certain regions. In a phone call with governors on Monday, President Trump encouraged governors to also look into buying more ventilators themselves.
But hospitals may face shortages of workers that are more limiting than any shortage of ventilators or physical beds. The United States has fewer doctors per capita than Italy does, for example. And some rural places that could add more beds may not be able to recruit more health care workers.
To deliver I.C.U.-level care, nurses from different hospital departments may need to work under the supervision of those who typically work with critically ill patients.
“Under normal standard of care, we take care of one to three patients,” said Elizabeth Bridges, president-elect of the American Association of Critical Care Nurses. “I think what is unique here is we need to ask, What are the things that other providers can do? Adding them in is an immediate enhancer of capacity.”
In Great Falls, Mont., one of the few areas that would have more than enough beds to handle a moderately damaging coronavirus outbreak, staffing may be a real challenge.
“We could flex up higher, but the hitch is usually about staffing and not beds,” said John Goodnow, the chief executive of Benefis Health System, which operates multiple hospitals in and around Great Falls. “The thing that is worrying me now is, What if the school districts close? Then that’s a lot of people with kids who need to stay home.”
His local school districts still remain open, but Mr. Goodnow has gone as far as searching for empty areas of his hospital campuses that could be turned into temporary care centers for the children of his hospital’s workers.
The Harvard team that calculated the estimates consisted of Ashish K. Jha, Thomas Tsai, Ben Jacobson, Stefanie Friedhoff and Jose Figueroa. They used data from the American Hospital Association and the American Hospital Directory to measure the number of hospital and I.C.U. beds in each region and to obtain their average occupancy. The occupancy rates are an annual average, and tend to fluctuate throughout the year.
To select the infection rates for the three scenarios, they relied on the projections of their colleague Marc Lipsitch, an epidemiologist who has estimated possible trajectories for the epidemic. The team considered infections only among adults, because few children have required hospitalization for the disease so far, and children tend to be treated in pediatric beds, which are not included in the bed counts they used.
Estimates of the proportion of ill patients who would require hospitalization and intensive care come from scholarly literature on the Wuhan, China, outbreak of the disease, as does an estimate of how many days each patient would be likely to stay in the hospital (12 days). Estimates of the effects on a community’s proportion of older residents on hospitalization also come from studies of China.
The maps show 305 of the 306 hospital referral regions in the United States. There was not enough data to include the Panama City, Fla., area in the analysis.
The model’s underlying assumptions
cat | mild | moderate | severe | assumptions |
|---|---|---|---|---|
Infection rate | 20% | 40% | 60% | From Marc Lipsitch, Harvard epidemiologist. |
Adults infected | 49 million | 99 million | 148 million | The number who would be infected out of 257 million adults. |
Adults hospitalized | 10 million | 21 million | 31 million | Assumes hospitalization rate of 19% for those under 65 and 28.5% for those 65+. |
Adults needing ICU care | 2 million | 4 million | 7 million | Assumes 19% I.C.U. rate for those hospitalized under 65 and 28.5% for those 65+. |
