In a little over a month, the Trump administration has started to hollow out America’s federal health agencies. Roughly 2,000 probationary workers have been fired en masse, by virtue of the fact that they were relatively new to their jobs. But the long-term impact of those terminations could pale in comparison to a lesser-noticed spate of departures that has recently roiled the health agencies. In the past two months, the FDA, CDC, and NIH second in commands have all resigned or retired. So have several other prominent officials, including the FDA’s chief operating officer as well as the heads of both its food center and its drug center.

There is always some churn in a new White House, but it typically affects “political staff” who are appointed to serve in a specific administration. Civil servants, meanwhile, are far more likely to stay in their positions regardless of who is president. The continuity of these career officials ensures that agencies can still function as their newly appointed political leaders map out their agendas. As in any other industry, career officials—who dramatically outnumber political staff—do sometimes leave. But many of the top staffers who voluntarily abandoned their positions had previously not shown any sign of being ready to do so. Nirav Shah, the principal deputy of the CDC, is reportedly resigning tomorrow, despite telling a Politico reporter in January that he did not have any “current plans to leave government.” And Jim Jones, the head of the FDA’s food center, was just getting started on a long-term plan to revitalize that office. Shah and Jones, like all the other recently departed health officials mentioned in this story, declined to comment.

[Read: Inside the collapse at the NIH]

The level of attrition happening in the health agencies right now is unprecedented, Max Stier, the head of the Partnership for Public Service, a nonpartisan group that aims to strengthen the federal bureaucracy, told me. “We’re watching a complete sweep of those most senior career experts,” he said. President Donald Trump campaigned on a promise to “demolish” the so-called deep state, which he believes is out to sabotage his agenda, and has repeatedly declared his intent to gut the health agencies. To an extraordinary degree, his administration has already succeeded.

The administration has offered plenty of incentive for government workers to head for the exit. One of the Trump administration’s first moves was offering a buyout to any federal worker willing to abruptly leave their position. Trump has also mandated that all federal workers, including those who live more than 50 miles from their office, work in office five days a week. It’s unclear how many rank-and-file workers have quit because of these efforts, but prominent instances of attrition have not been limited to just the health agencies. A top Treasury Department official recently retired after reportedly refusing to give DOGE access to the government’s system for doling out trillions of dollars each year. So did 21 staffers at the United States Digital Service who had been drafted into working for DOGE.

Jones, the former director of the FDA’s food center, is instructive in understanding what is fueling the public-health exodus in particular. He joined the agency in 2023, and had spent the past several months staffing up areas of the food center that were faltering. But when 89 newly hired probationary employees were fired by the Trump administration earlier this month, he had enough. He did not want to be involved in “dismantling an organization,” he told the health publication Stat. The health agencies were upended by DOGE cuts just as Robert F. Kennedy Jr. was confirmed by the Senate as their boss. The health secretary has his own desires to fire bureaucrats. Had Jones not resigned, it is reasonable to assume that he would have been pushed out of his position: Kennedy had previously implied that everyone at the center Jones ran should be given pink slips.

That’s easier said than done. Although bureaucrats on their probationary period—because they’re either newly hired or recently promoted—can be fired with relative ease, career officials generally cannot be let go without actual cause. But none of that matters if officials resign—whether as a result of their own dissatisfaction or being pressured out. Attrition is a cheat code for thinning the federal workforce. In just a month, the Trump administration’s assault on the federal workforce has managed to push even an ardent reformer like Jones to surrender.

These departures will likely be a cause célèbre for MAGA world. As I wrote in November, public-health officials historically have been the firewall against the political whims of the White House. That is what happened during Trump’s first term. Early in the pandemic, Janet Woodcock, then the head of the FDA’s drug center, reportedly sprang into action to prevent widespread distribution of the unproven COVID treatment hydroxychloroquine over the orders of top Trump officials. What makes the recent resignations so consequential is that they suggest that Trump and RFK Jr. will face less resistance from inside the agencies as they attempt to overhaul public health. Already, Jones has been replaced by Kyle Diamantas, a food and drug lawyer (and hunting buddy of Donald Trump Jr.) who has no previous experience working in the federal government.

Still, these resignations may not be to the Trump administration’s benefit. Very few individuals have the type of specialized knowledge that comes with decades in government service. “These are the people that you want to do everything possible to hold on to,” Stier, of the Partnership for Public Service, said. Consider Lawrence Tabak, the outgoing principal deputy director at the National Institutes of Health. He served in that role for more than a decade, punctuated only by a short stint running the entire agency. That experience could have been channeled into delivering Kennedy’s promised reforms to the agency, like revisiting the government’s standards on conflicts of interests in research. Jones, though new to the FDA, had previously spent nearly two decades at the EPA regulating the safety of pesticides and other chemicals, which made him one of the few people prepared to deliver on Kennedy’s promise to ramp up the regulation of food chemicals. Indeed, Jones’s resignation from the FDA has shaken the agency’s staff. One employee who works at the food center described the mood to me as “pissed and scared and coping and numb and confused and demoralized.” (I agreed not to name the employee, because they’re not allowed to speak to the press.)

The federal health agencies have real problems: They’re often slow, bureaucratic, and cloistered, as Trump and Kennedy have been quick to point out. That some of the nation’s top health officials have decided to head for the exit may only make matters worse.

Three years ago, when it was trickling into the United States, the bird-flu virus that recently killed a man in Louisiana was, to most Americans, an obscure and distant threat. Now it has spread through all 50 states, affecting more than 100 million birds, most of them domestic poultry; nearly 1,000 herds of dairy cattle have been confirmed to be harboring the virus too. At least 66 Americans, most of them working in close contact with cows, have fallen sick. A full-blown H5N1 pandemic is not guaranteed—the CDC judges the risk of one developing to be “moderate.” But this virus is fundamentally more difficult to manage than even a few months ago and is now poised to become a persistent danger to people.

That didn’t have to be the reality for the United States. “The experiment of whether H5 can ever be successful in human populations is happening before our eyes,” Seema Lakdawala, a flu virologist at Emory University, told me. “And we are doing nothing to stop it.” The story of bird flu in this country could have been shorter. It could have involved far fewer cows. The U.S. has just chosen not to write it that way.

[Read: America’s infectious-disease barometer is off]

The USDA and the CDC have doggedly defended their response to H5N1, arguing that their interventions have been appropriately aggressive and timely. And governments, of course, don’t have complete control over outbreaks. But compared at least with the infectious threat most prominent in very recent memory, H5N1 should have been a manageable foe, experts outside of federal agencies told me. When SARS-CoV-2, the virus that sparked the coronavirus pandemic, first spilled into humans, almost nothing stood in its way. It was a brand-new pathogen, entering a population with no preexisting immunity, public awareness, tests, antivirals, or vaccines to fight it.

H5N1, meanwhile, is a flu virus that scientists have been studying since the 1990s, when it was first detected in Chinese fowl. It has spent decades triggering sporadic outbreaks in people. Researchers have tracked its movements in the wild and studied it in the lab; governments have stockpiled vaccines against it and have effective antivirals ready. And although this virus has proved itself capable of infiltrating us, and has continued to evolve, “this virus is still very much a bird virus,” Richard Webby, the director of the World Health Organization Collaborating Centre for Studies on the Ecology of Influenza in Animals and Birds, told me. It does not yet seem capable of moving efficiently between people, and may never develop the ability to. Most human cases in the United States have been linked to a clear animal source, and have not turned severe.

The U.S., in other words, might have routed the virus early on. Instead, agencies tasked with responding to outbreaks and upholding animal and human health held back on mitigation tactics—testing, surveillance, protective equipment, quarantines of potentially infected animals—from the very start. “We are underutilizing the tools available to us,” Carol Cardona, an avian-influenza expert at the University of Minnesota, told me. As the virus ripped through wild-animal populations, devastated the nation’s poultry, spilled into livestock, started infecting farmworkers, and accumulated mutations that signaled better adaptation to mammals, the country largely sat back and watched.

When I asked experts if the outbreak had a clear inflection point—a moment at which it was crucial for U.S. leaders to more concertedly intervene—nearly all of them pointed to the late winter or early spring of last year, when farmers and researchers first confirmed that H5N1 had breached the country’s cattle, in the Texas panhandle. This marked a tipping point. The jump into cattle, most likely from wild birds, is thought to have happened only once. It may have been impossible to prevent. But once a pathogen is in domestic animals, Lakdawala told me, “we as humans have a lot of control.” Officials could have immediately halted cow transport, and organized a careful and concerted cull of infected herds. Perhaps the virus “would never have spread past Texas” and neighboring regions, Lakdawala told me. Dozens of humans might not have been infected.

[Read: America’s bird-flu luck has officially run out]

Those sorts of interventions would have at least bought more of the nation time to provision farmworkers with information and protection, and perhaps develop a plan to strategically deploy vaccines. Government officials could also have purchased animals from the private sector to study how the virus was spreading, Cardona told me. “We could have figured it out,” she said. “By April, by May, we would have known how to control it.” This sliver of opportunity was narrow but clear, Sam Scarpino, an infectious-disease modeler and flu researcher at Northeastern University, whose team has been closely tracking a timeline of the American outbreak, told me. In hindsight, “realistically, that was probably our window,” he said. “We were just too slow.”

The virus, by contrast, picked up speed. By April, a human case had been identified in Texas; by the end of June, H5N1 had infected herds in at least a dozen states and more than 100 dairy farms. Now, less than 10 months after the USDA first announced the dairy outbreak, the number of herds affected is verging on 1,000—and those are just the ones that officials know about.

The USDA has repeatedly disputed that its response has been inadequate, pointing out to The Atlantic and other publications that it quickly initiated studies this past spring to monitor the virus’s movements through dairy herds. “It is patently false, and a significant discredit to the many scientists involved in this work, to say that USDA was slow to respond,” Eric Deeble, the USDA’s deputy undersecretary for marketing and regulatory programs, wrote in an email.

And the agency’s task was not an easy one: Cows had never been a known source of H5N1, and dairy farmers had never had to manage a disease like this. The best mitigation tactics were also commercially formidable. The most efficient ways to milk cows invariably send a plume of milk droplets into the air—and sanitizing equipment is cumbersome. Plus, “the dairy industry has been built around movement” of herds, a surefire way to move infections around too, Cardona told me. The dairy-worker population also includes many undocumented workers who have little incentive to disclose their infections, especially to government officials, or heed their advice. At the start of the outbreak, especially, “there was a dearth of trust,” Nirav Shah, the principal deputy director of the CDC, told me. “You don’t cure that overnight.” Even as, from the CDC’s perspective, that situation has improved, such attitudes have continued to impede efforts to deploy protective equipment on farms and catch infections, Shah acknowledged.

Last month, the USDA did announce a new plan to combat H5N1, which requires farms nationwide to comply with requests for milk testing. But Lakdawala and others still criticized the strategy as too little, too late. Although the USDA has called for farms with infected herds to enhance biosecurity, implementation is left up to the states. And even now, testing of individual cows is largely left up to the discretion of farmers. That leaves too few animals tested, Lakdawala said, and cloaks the virus’s true reach.

The USDA’s plan also aims to eliminate the virus from the nation’s dairy herds—a tall order, when no one knows exactly how many cattle have been affected or even how, exactly, the virus is moving among its hosts. “How do you get rid of something like this that’s now so widespread?” Webby told me. Eliminating the virus from cattle may no longer actually be an option. The virus also shows no signs of exiting bird populations—which have historically been responsible for the more severe cases of avian flu that have been detected among humans, including the lethal Louisiana case. With birds and cows both harboring the pathogen, “we’re really fighting a two-fronted battle,” Cardona told me.

Most of the experts I spoke with also expressed frustration that the CDC is still not offering farmworkers bird-flu-specific vaccines. When I asked Shah about this policy, he defended his agency’s focus on protective gear and antivirals, noting that worker safety remains “top of mind.” In the absence of consistently severe disease and evidence of person-to-person transmission, he told me, “it’s far from clear that vaccines are the right tool for the job.”

[Read: How much worse would a bird-flu pandemic be?]

With flu season well under way, getting farmworkers any flu vaccine is one of the most essential measures the country has to limit H5N1’s threat. The spread of seasonal flu will only complicate health officials’ ability to detect new H5N1 infections. And each time bird flu infects a person who’s already harboring a seasonal flu, the viruses will have the opportunity to swap genetic material, potentially speeding H5N1’s adaptation to us. Aubree Gordon, a flu epidemiologist at the University of Michigan, told me that’s her biggest worry now. Already, Lakdawala worries that some human-to-human transmission may be happening; the United States just hasn’t implemented the infrastructure to know. If and when testing finally confirms it, she told me, “I’m not going to be surprised.”