As surges of COVID-19 cases spread throughout the United States over the past year, hospitals overwhelmed with patients quickly ran out of everything: masks, gloves, beds, space, doctors, nurses. Hospital workers in New York City, one of the first epicenters of the US pandemic, wore garbage bags as protective equipment. Patients in overcrowded California emergency rooms spent hours lying in hallways. Nurses in Missouri worked twice the normal number of shifts to make up for sick colleagues.

The catastrophe was clear early on in the pandemic, and experts gave clear warning that there weren’t going to be enough hospital beds available for everyone who needed one. But even with that warning, there wasn’t enough time to build up capacity before the wave of COVID cases broke and the demand far outstripped supply, particularly in intensive care units. In the end, hospitals were only able to withstand the surge with considerable cost to overworked doctors and nurses.

Now that the immediate emergency is subsiding, those same hospital and ICU beds are getting a closer look. COVID-19 won’t be the last threat to stress-test the country’s fragmented, private health care infrastructure. But getting hospitals ready for the next disaster isn’t just a matter of spending money. Real preparation will require changing the way hospitals work together, changing them from isolated enterprises into a collaborative network — in other words, thinking of hospitals as infrastructure.

With Biden’s infrastructure push gearing up, that rethinking may have to come soon. As it makes the case for the American Jobs Plan, the White House has talked about making hospital networks more resilient, alongside similar efforts in food systems, transportation, and the electric grid. But the only tangible measure detailed in the bill is new funding for VA hospitals, the only public medical facilities directly operated by the federal government. If no new measures are added, the White House may end up missing its chance for the more profound change many experts say is necessary.

“I think that there probably needs to be a re-analysis of the day-to-day healthcare resources of communities, and rethinking those with the idea of a surge in mind,” says Michael Redlener, an associate professor of emergency medicine at Mount Sinai Hospital in New York. “The healthcare system is designed to meet everyday needs. It’s not really conceived of as something for a surge.”

The simplest physical signal of how ready our hospitals were for COVID-19 was the number of beds they had ready for sick people. Before the pandemic, the US had around 2.8 hospital beds per 1,000 people, according to one analysis, which is far fewer than countries like Germany, France, and Japan. Each day, most of those beds were full, which didn’t leave much slack in the system for a sudden influx. Some states had less room than others: Connecticut, for example, only had 0.45 unoccupied beds per 1,000 people, according to the analysis.

Those numbers used to be higher: over the past few decades, the number of hospital beds per 1,000 people in the US has steadily declined. Hospitals have eliminated beds, health care systems have closed down inpatient care centers, and small, independent hospitals have shut down. The trend is particularly acute in rural areas, where over 100 hospitals have closed since 2013.

In large part, that decline is a feature of the US for-profit health system, which pushes hospitals to keep overhead low. Keeping a bed open and ready to go costs money; if there’s not a patient to fill it, it’s seen as a cost waiting to be cut. Building excess capacity can be bad for business.

“In a fee-for-service system, hospitals know what their patient flow looks like, and they’re prepared for that. They’re not always prepared for the worst case scenario,” says Fredric Blavin, a researcher in the Health Policy Center at the Urban Institute who studied hospital capacity at the start of the pandemic.

On the flip side, keeping too many beds open could incentivize hospitals to fill them — giving potentially unnecessary care and costs to patients. Given that model, it isn’t realistic to maintain the space for hundreds of additional patients all the time, says Michelle Mello, who studies health law and health care delivery at Stanford Law School. For Mello, preparedness is more about improving communications between hospitals as they manage the resources they have.

Mello is based in Northern California and says that there wasn’t a good way to route patients to nearby hospitals if the one they showed up at was full. “There wasn’t a great system for sending the things you might need to take care of COVID-19 patients,” she says. The same was true in New York City, Redlener says. “If you talked to someone in Queens during the height of the pandemic, there weren’t enough ventilators for every patient who needed one. But if you looked at the region, there was probably enough capacity to take care of everyone,” he says.

When COVID-19 cases surged in 2020, small hospitals were quickly overwhelmed. They struggled to find larger medical institutions to transfer patients to. There’s no data system that offers easy visibility into the beds and resources available at various hospitals in the US, so hospital staff in many places relied on working the phones to find space for patients. In theory, the hospitals were coordinating with state and local health departments — but in practice, they were mostly operating on their own.

That lack of collaboration also makes economic disparities worse, since wealthy hospitals are more likely to have stockpiles of supplies on which they can rely. “A wealthy hospital might say, ‘I don’t know what’s coming, we could have a surge tomorrow, I don’t want to share.’ On aggregate, that creates huge inefficiencies because everyone’s looking out for their own good,” Mello says. “It cries out for a more cooperative approach.”

A more cooperative plan might be modeled off of something like New York’s regional burn plans, which are designed to respond to a fire or disaster that leaves a large number of people with serious burns, Redlener says. Those plans involve groups like the fire department, burn centers, hospitals, and the Department of Health; they make sure that there’s a way to create beds for burn patients and have specialists available to see them. “A plan exists, and it thinks about everything in a multidimensional way,” he says.

There have been some bright spots of collaboration over the past year, says Nancy Foster, the vice president for quality and patient safety policy at the American Hospital Association: federal and state authorities worked with hospitals to distribute monoclonal antibody COVID-19 treatments, for example. Early evidence showed that the drug, which has to be given to people just after they’re diagnosed with COVID-19, can blunt the severity of illness. Rather than let individual states and hospitals order it, the US Department of Health and Human Services allocated it out to states based on the number of COVID-19 cases they had over the previous week.

“Those kinds of opportunities to sit together and to really appropriately allocate resources in an effective manner are extraordinarily helpful,” Foster tells The Verge. “They’re part of what we have to think about going forward.”

Making sure that spirit of collaboration continues after the acute emergency of the COVID-19 pandemic, though, could be difficult — again, because the US health care system is private and institutions typically act independently. It’d be a challenge to mandate coordination from the federal level, but there could be incentives for organizations to participate, Mello says. “There are already efforts to do these things on a voluntary basis,” she says. That principle holds true for other types of emergency preparedness, like flexible staffing or better supply management, Redlener says. “There’s a balance, and a push-pull between mandates versus incentives to participate,” he says.

But reinvesting in programs that would make the health care system more collaborative — and, therefore, more resilient — benefits everyone. “We truly don’t know what the features of the next disease will be and what the needs will be,” Mello says. “That should create incentives for people to cooperate and hedge their risk.”