Science’s COVID-19 reporting is supported by the Pulitzer Center and the Heising-Simons Foundation.
In the wake of the dramatic news of a potentially effective COVID-19 vaccine, President Donald Trump posted a flurry of tweets that claimed its makers, the U.S. Food and Drug Administration (FDA), and Democrats had conspired to suppress the announcement until after the 3 November presidential election. The U.S. company involved, Pfizer, “didn’t have the courage to do it before,” Trump asserted on 9 November. And FDA and Democrats, he wrote, “didn’t want to have me get a Vaccine WIN, prior to the election, so instead it came out five days later.”
Initially, Pfizer and its German partner BioNTech suggested they might have preliminary trial results by late October—a timetable Pfizer’s CEO, Albert Bourla, projected as recently as 29 September. The timeline was based on a plan that called for an outside panel to take a first look at the efficacy data for the vaccine when a total of 32 cases of COVID-19 had accumulated in the vaccine and placebo groups. But the companies and FDA later agreed on a protocol change that nearly doubled that number and delayed that review.
There is no evidence, however, that the decision had anything to do with presidential politics. And the companies flat out reject at Trump’s claims. “What people believe is their business,” Kathrin Jansen, who heads vaccine R&D at Pfizer, told ScienceInsider. “Quite frankly, we had no time and still have no time to deal with politics. We are at this 24/7, thousands of people working diligently to make this work. And for us, it was never about politics, it was always about just the disaster that we were in the middle of, all of us globally, seeing the devastation and the deaths.”
Given that Pfizer and BioNTech only described their results in a press release that had scant details beyond the finding that it had greater than 90% efficacy, ScienceInsider followed up with company representatives and other vaccine experts to address some of the confusion surrounding the vaccine.
What was the initial target for the Pfizer/BioNTech vaccine?
FDA said in a June guideline that it would consider granting what’s known as emergency use authorization (EUA) to a COVID-19 vaccine that demonstrated at least 50% efficacy, a clinical measure of its worth that doesn’t equate to the same level of effectiveness in a real-world situation.
What did the original trial protocol that Pfizer submitted to FDA say about the timing of unblinding the data?
From the outset, company statisticians designed the trial to meet FDA’s target—specifically to determine an efficacy of 52.3% or better after there had been a total of 164 confirmed cases of COVID-19 among trial participants. That translates to 111 cases in the placebo group versus 53 among participants who received the vaccine.
As is standard in blinded vaccine efficacy trials, a data monitoring committee (DMC) of scientists who do not work for Pfizer or BioNTech took scheduled interim looks at unblinded data. The DMC takes these interim peeks in order to protect participants from dangerous side effects, and it can recommend stopping the study if it appears clear the trial will not reach its efficacy target and continuing is futile. On the flip side, if a strong efficacy signal surfaces early, the DMC can recommend unblinding the study and even modifying the protocol to offer the vaccine to all the placebo recipients. Pfizer initially designed the trial to conduct interim analyses after 32, 62, 92, and 120 cases accrued. Fewer cases means less statistical power and therefore greater uncertainty about the vaccine’s impact, so the efficacy goals to meet FDA’s requirements were higher than 50% at interim analyses, ranging from 76.9% for 32 cases to 58.8% for 120.
Some outsiders questioned the planned look at just 32 cases; protocols for several other COVID-19 vaccine efficacy trials have a first peek at about 50 cases. But it also meant Pfizer might obtain the first readout from any of the COVID-19 vaccines now in testing. Then something changed. “After discussion with the FDA,” Pfizer revealed in its press release, it decided to drop the 32-case interim analysis.
What led the company to switch gears?
Ugur Sahin, scientist, CEO, and co-founder of BioNTech, says the initial plan to look at 32 cases stemmed from a conservative assumption about the rate of spread of COVID-19 and the sense of urgency about the need for a vaccine. If the vaccine looked terrific at 32 cases and it was going to take months to get to 62 cases, then waiting seemed like a mistake, he says. “To me, every day counts.” And he had little patience for the debates over when to look at the data. “These protocol discussions are endless, and I’m often leaving the room,” Sahin says.
In mid-October, the companies had yet to confirm 32 cases. But with the epidemic exploding at many of the trial’s locations—which were mainly in the United States—they had second thoughts about FDA’s request that their first interim analysis should have more to support an EUA request. FDA “had strongly recommended to us that we change that, and the pandemic just was spiraling out of control in the United States and elsewhere, and we realized that we probably could get cases much faster than what we had anticipated,” Jansen says.
The math was simple: COVID-19 cases among participants were jumping from one or two per day to up to 10 or more. It became clear that the trial would accrue 62 cases shortly after hitting the 32 mark, and the higher number meant greater statistical power—and fewer debates about the meaning of the data. This 62 cutoff both lowered the efficacy bar the vaccine had to clear, and was also something of an insurance policy: If the vaccine triggered mediocre immune responses and it teetered around 50% efficacy in the trial, it could more easily have been deemed futile at 32 cases because of bad luck.
Did the decision to switch to 62 cases have anything to do with the election?
Both Pfizer and BioNTech say no. Sahin and Jansen both are emphatic on this point. Sahin says FDA never gave them any hint of concern about the election timing and the companies’ decision to seek the protocol change had nothing to do with politics, either. “This was our decision,” Sahin says. When he read Trump’s tweets, he shrugged. “This is just not true.”
When did Pfizer’s CEO stop publicly projecting a possible October success and why?
On 16 October, Bourla explained in an “open letter” that though the companies might know “whether or not our vaccine is effective by the end of October,” the new FDA guidelines meant they could not file for an EUA until the third week in November at the earliest. To increase confidence in the safety of these vaccines, the FDA revision—which the Trump administration attempted to block—said that before a company applied for an EUA, 2 months must have passed since at least half of the participants in a trial had received their final shot. Most vaccine-related side effects occur within that period.
Why was there a delay between hitting the 62-case mark threshold and the DMC meeting on 8 November?
When the companies submitted their request for a protocol change, they had yet to accumulate 32 cases. If they had 32 cases before the change was approved, the protocol would have required them to report the results to FDA. In addition, if the results could impact the way investors traded company stock, they may also have been required by the U.S. Securities and Exchange Commission (SEC) to make the results public. They decided to store the nasal swabs taken from participants who had suspected SARS-CoV-2 infections: If they didn’t test the swabs, they couldn’t confirm cases and therefore would avoid a protocol violation. On 3 November, FDA approved the protocol changes and the companies assumed the trial was crossing the 62-case mark. It then took several days to check the stored samples—and they also had new ones coming in that they could test, too. Combined, these samples led to the 94 cases presented to the DMC on 8 November. (The results were publicly announced the next day.)
Did the trial hit 32 COVID-19 cases before the presidential election?
Maybe—but the companies have yet to reveal the timing of the case accruals. They may have had that many cases by the time without knowing it, because not all nasal swabs from participants were immediately tested.
A trial case was defined as someone who had confirmed COVID-19—meaning they tested positive for the causative virus, SARS-CoV-2, regardless of symptom severity—7 days after receiving the second dose of the vaccine. This extra week after the last vaccination would allow the immune system time to build an appropriate response. The press release on 9 November noted the trial had so far tallied 94 cases out of 43,538 participants. At its third-quarter earnings report teleconference on 27 October, Bourla said they had not reached the 32 cases, but the company said it needed only 2000 more participants to complete enrollment and that nearly 36,000 had received the second dose. The trial began to enroll participants on 27 July.
Why did Pfizer and BioNTech issue a press release with few details?
The press release did not specify how many of the 94 people were in the placebo versus the vaccine arms, which is needed to calculate efficacy. It just said efficacy was greater than 90%, without providing a confidence interval that puts upper and lower bounds around the estimate (say, 60% to 95%). That was enough to meet SEC requirements that companies must report any “material and timely information” that could affect whether investors buy or sell stock; the regulations do not require detailed explanations of data.
Sahin and Jansen said they weren’t even given the exact numbers. “About 2 weeks from now, we’ll most likely have the full recruitment of the cases, and everyone will get the numbers,” Sahin says. “So there is no need to provide specifics.” The final analysis, he says, will give a better picture of the overall efficacy, the confidence interval, and critical details about possible differences in the vaccine’s impact in different populations, including the elderly.
Is the efficacy likely to change with the full 164 confirmed cases?
It could, and if it’s lower, that’s one reason the company may not have wanted to share exact figures now: People might be disappointed if, say, the efficacy dropped from 93% at 94 cases to 89% with 164. The efficacy also could go higher. Antibody levels trigged by the vaccine peak at 2 weeks after the second vaccination, and the preliminary analysis of the 94 cases only looked at people who had received their second shot just 7 days before. So the case gap between the placebo and vaccine arms could widen.
When will the companies publicly report more data?
It’s not clear. The companies say they won’t have more data, such as the breakdown of cases by groups, until the study ends. FDA has stated it will convene its vaccine advisory committee, an external group that has no conflicts of interest, if any company submits a request for a COVID-19 vaccine EUA, and those meetings are public. Advisory committee meetings typically include presentations from company scientists that involve detailed descriptions of their data, which are then discussed. The company may also post data to a preprint server or submit a paper to a peer-reviewed journal at any time.
What’s the potential timeline for regulatory approval of the vaccine?
In a best-case scenario, before the end of the year. The advisory committee currently is not scheduled to meet until December, and several possible dates have been floated. The committee will likely make a recommendation to FDA, which the agency typically follows. If the committee gives the vaccine a thumbs-up, it will probably take FDA a few weeks to process the EUA.
“It’s an exciting development, and we look forward to receiving an application and to bringing it through the process ultimately towards the public advisory committee meeting to review the vaccine,” says Peter Marks, who heads the FDA division that oversees vaccines.
*Correction, 12 November: The number of COVID-19 cases among vaccinated trial participants needed to meet Pfizer’s original efficacy goal of 52.3% has been corrected to 53, instead of 85.
*Clarification, 13 November: The story initially said that Pfizer knew it had crossed the 62-case threshold on 3 November. Pfizer has clarified to ScienceInsider that the company assumed it was close to or beyond 62 cases on 3 November, but did not have an actual number of confirmations on that date.