The first COVID-19 vaccine could arrive before Election Day, Donald Trump avowed in the summer of 2020. But government regulators wanted things to work out differently: “The deep state, or whoever, over at the FDA is making it very difficult for drug companies to get people in order to test the vaccines and therapeutics,” he wrote on Twitter. “Obviously, they are hoping to delay the answer until after November 3rd.”
Regulators did, in fact, end up slowing the process: In the first week of September, the FDA told vaccine makers to extend their clinical trials by several weeks beyond what they’d planned, in order to gather more safety data. That effectively postponed Pfizer’s request for an emergency use authorization of the mRNA vaccine it had developed with BioNTech until after the election. The agency had been under attack from both sides: The president had wanted regulators to move more quickly, while an alarmed public-health community was pushing for the opposite. In the end, the decision to slow things down was meant to bolster the country’s waning confidence in the forthcoming COVID-19 vaccines and restore the FDA’s reputation for integrity and caution. But the move looks very different in hindsight, with full knowledge of the safety and efficacy of the vaccines and of the body count that amassed shortly after. The economist Garett Jones recently opined that Trump’s scuttled hopes to release a COVID-19 vaccine a few weeks earlier “likely would have saved at least 100,000 American lives.”
Could Jones be correct about the human scale of this decision? His colleague at George Mason University, the libertarian economist and blogger Alex Tabarrok, often writes about the “invisible graveyard”: the final resting place for all those killed by the FDA’s bureaucratic foot-dragging. Indeed, one of the more painful lessons we’ve learned about governance during this pandemic is that waiting for more data can be a leading cause of death. But just how deadly was the government’s choice to slow down Pfizer’s and Moderna’s research in the fall of 2020?
Let’s recall the context for the FDA’s decision. During the first year of the pandemic, it was easy to freak out a little bit as the president pushed wishful but unproven COVID-19 therapies along the FDA’s high-speed pathway for emergency use authorization (EUA). In particular, hydroxychloroquine and convalescent plasma were touted as miracle cures and widely distributed to patients even in the absence of convincing data showing that they worked.
As the election approached that fall, and Trump intensified his attacks on the agency, those two messy episodes set the stage for a bigger regulatory standoff. The stakes had seemed much lower for the “miracle cures” because those were typically being given to people already sick with COVID. The new mRNA injections, if authorized, would be delivered to millions of healthy, uninfected Americans, most of whom were unlikely ever to develop severe disease. “We don’t do EUAs for vaccines,” the pediatrician and vaccinologist Peter Hotez told Yahoo News on September 1. “It’s a lesser review, it’s a lower-quality review, and when you’re talking about vaccinating a large chunk of the American population, that’s not acceptable.”
The FDA had already signaled that it would make the EUA process more demanding for the COVID-19 vaccines than it had been for medical products in the past: The vaccine makers would need to show results from a large-scale Phase 3 clinical trial. Pfizer initially expected to have these by the end of October, when its 44,000-person clinical trial would have accrued enough data points for researchers to perform a first statistical analysis. Then the FDA applied the brakes, telling Pfizer and Moderna that it wanted additional safety data. More specifically, it wanted the companies to have tracked the health of at least half of their clinical-trial subjects for 60 days following the second dose. The thinking was that most adverse reactions occur within the first 42 days of vaccination, but out of an abundance of caution, the FDA would extend that period to 60 days, which was still much shorter than the six months of follow-up expected for a full vaccine approval.
FDA scientists touted this new proposal as “EUA Plus.” Marion Gruber, a former head of the agency’s vaccine office, told me it was not a direct response to the president’s actions. “We tried to separate ourselves from whatever politics was going on,” she said when I interviewed her last year. Nevertheless, she was concerned that the public’s confidence in the vaccines was on the decline that fall. Gallup poll numbers indicate that Americans’ willingness to take an approved COVID-19 vaccine had dipped from 66 percent that summer to just 50 percent in September.
Fearful that the FDA bureaucrats could be overruled by the White House, outside scientists locked arms and sent a letter to Pfizer’s CEO, Albert Bourla, on September 25, telling him that the proposed safety delay would be crucial for securing “the utmost trust in the vaccine and the science behind it.” They warned that a “premature application would prolong the pandemic” and “severely damage” Pfizer’s reputation.
In the end, Pfizer did not reveal its trial’s favorable results until November 9—six days after the election. The company had originally planned to consider submitting an EUA request to the FDA with just 32 data points; instead it gathered 94, and it waited another 11 days to accrue the requested safety data, plus even more data showing how well the vaccine worked, before making its filing. With no serious safety issues and vaccine efficacy coming out in the mid-90s, outsiders were more than satisfied. So was the FDA, which finally granted Pfizer’s vaccine an EUA on December 11.
The claim that the delay killed 100,000 people assumes that the process got backed up by six to eight weeks, and that the vaccine could have been rolling out by the end of October. Having reported a book on the COVID-19 vaccine race, I find this imagined timeline unrealistic. The FDA was evaluating a new vaccine for use against a new disease, based on a technology—mRNA—that had never before been authorized. Even given the clear sense of urgency, it’s hard to imagine that the review process could have been shortened by more than a week, as happened in the United Kingdom. And even if authorization had been instantaneous, we still don’t know how many doses Pfizer would have had available for distribution by the end of October. Bourla told The Washington Post on September 29 that he expected to have “hundreds of thousands [of doses] ready” in October and “a few million in November”—but production did not ramp up as quickly as expected.
In light of all these facts, the regulatory move probably slowed the rollout by about two to three weeks at most. Without the additional safety requirement, Pfizer could theoretically have stuck with its initial plan and filed for an EUA before the presidential election. Following FDA review, a rollout might have begun in the third week of November.
A quicker process might have kept public confidence in the vaccine depressed for a little while longer, but more than enough people would have still been willing to roll up their sleeve for a jab; after all, it took about four months for vaccine supply to catch up with demand. And with so few doses available for this earlier rollout, they would have had to have been primarily targeted to the most vulnerable: the 1.4 million elderly people in nursing homes, where about 5,000 residents were dying from COVID each week in early December 2020. The vaccines would have cut mortality with the first dose, leading to a steep drop in cases in December, rather than the January decline we saw.
At my request, Claus Kadelka, a mathematician at Iowa State University who has studied COVID-19-vaccination strategies, modeled what would have happened if 2 million bonus doses had been available for nursing homes starting on November 23. Depending on the details of the scenario, he estimated that those extra doses could have saved 6,000 to 10,000 lives.
That doesn’t necessarily mean that the FDA botched its risk-benefit calculus based on what was known at the time, or that the public-health community was out of line in calling for a cautious process. In August and September, when many concerns were raised, COVID case numbers were at a lull, and no one was certain how quickly they would take off again.
And though no major safety events were caught during those extra few weeks of data collection, who’s to say that things couldn’t have worked out otherwise? To argue that the wait was worthless is akin to saying I shouldn’t wear my seat belt tomorrow because I didn’t get in a car accident today. What if a more serious issue had emerged two months after the second dose? It would have harmed not only the uptake of the COVID vaccine (and maybe vaccines for other diseases already on the market), but also the teetering reputation of the FDA. That, too, could have had fatal consequences.
“The truth is that trust in vaccines is critical,” said Peter Marks, the acting head of the agency’s vaccines office, when I reached out to him last week to ask about the claim that the safety delay had cost thousands of lives. “If we didn’t have the data from those extra weeks … we would not have been anywhere near as sure that the vaccines worked—and vaccine hesitancy, which was pretty bad to begin with, would have been even worse.”
The scrutiny of the agency’s shifting goalposts also ignores an important and disturbing fact: An alternative path had been available to the FDA and the vaccine makers all along, one that could have avoided the EUA dilemma while still granting early access to the most vulnerable members of the population.
In October 2020, Deborah Birx, the coordinator of the White House Coronavirus Task Force, was on the ground in North Dakota and alarmed by how quickly COVID was spreading through nursing homes. As I reported in my book, The First Shots, she began to think that the vaccines should be distributed to residents under what’s known as an expanded access, or compassionate use, protocol. This FDA program is typically used by drug companies to provide patients who have rare diseases or cancers with new drugs that are likely safe but have not yet been proved effective in clinical trials. (Prior to the emergency use authorization of convalescent plasma, the Mayo Clinic provided the therapy to more than 70,000 COVID patients via this protocol.)
If a compassionate use program for COVID-19 vaccines had gone forward, doctors would have been able to prescribe them to nursing-home residents, even as the vaccine makers completed their clinical trials with integrity and gathered all the safety data requested under the “EUA Plus” requirements.
According to Marks, Birx asked Anthony Fauci and FDA Commissioner Stephen Hahn to encourage Pfizer and Moderna to apply for that program. (Birx declined to comment for this article; Fauci and Hahn did not respond to inquiries.) If the companies had agreed, and if their applications had been granted, how much time could have been gained is hard to say. Such a complex program would have required close monitoring by the CDC and major tweaks to Operation Warp Speed’s vaccine-distribution plans. In any event, conversations around the topic apparently sputtered, and the companies never applied for compassionate use. (Moderna did not reply to a request for comment on this decision. Pfizer responded with a broadly worded statement that read, in part: “We worked closely with the FDA as we developed our vaccine to determine the most appropriate regulatory pathway to make the vaccine available to the public.”)
Perhaps Birx’s efforts would have gathered more steam with a different president, or if an election were not about to happen, or if her proposal had been floated earlier. Unlike in time-travel stories, the course of history rarely pivots on a single event. The actual timing of the COVID-19 vaccines’ release resulted from a complicated mix of bureaucratic caution, political calculations, and the choices made by vaccine manufacturers. While the benefits of the vaccines have become very clear since then, the precise human cost of that short delay remains a mystery.
When you buy a book using a link on this page, we receive a commission. Thank you for supporting The Atlantic.