President Trump stood before a televised audience Wednesday and proclaimed that “results are very good” for vaccines targeting the novel coronavirus. A day later, Moderna and Pfizer, two front-runner drug companies developing a shot, released the full rule books for their studies, revealing that no one yet knows conclusively whether a vaccine is safe and effective — not even company executives.

Trump’s imprecise, extemporaneous comments about vaccines have frequently clashed with messages from government officials, outside scientists and companies. That discord has intensified concerns that political pressure will force a vaccine to be prematurely approved but also has sown public confusion as important public health messages have become entangled with politicians’ appeals to voters and companies’ communications to shareholders.

“We need trust as much as we need efficacy,” said Andrew Pavia, an infectious-diseases specialist at the University of Utah School of Medicine. “We can’t afford to do anything that reduces the trust. It’s not just the trust of the public. Experts in the field have to see enough of the data to feel comfortable to recommend it.”

Leaders of Moderna and Pfizer cited the need for greater transparency than usual in covid-19 clinical trials as the reason behind their decision to release the full documents describing how their studies will measure safety and effectiveness.

The documents confirm that study participants, physicians running the trials and the companies are “blinded,” unable to tell who received a real vaccine and who received a placebo.

Moderna’s trial contains preprogrammed milestones — when there are 53, 106 and 151 cases of covid-19, the illness caused by the coronavirus, among all participants — that allow an independent committee to analyze data to assess if there’s a signal of effectiveness. A key metric in vaccine studies involves examining who falls ill during the course of a trial and whether that person received a shot of protection or a placebo. It’s only then that anyone will know how well the vaccine is protecting people from becoming sick.

After reviewing the infection findings, the data committee will then make recommendations to an oversight group on whether to continue the trial. According to Moderna’s document, the first analysis is projected to be completed five months after the trial began in late July, but the timeline will vary depending on how many people get sick while participating in the trial.

“I have seen no data since the Phase 3 [trial] started,” Moderna chief executive Stéphane Bancel said in an interview. He predicted the company may have enough cases to detect whether the vaccine is effective in November, and it is possible but unlikely there could be a signal in October. Bancel predicted it would take a week or two to analyze and clean up data and submit it to regulators.

“We really have to wait for the data,” Bancel said. “We need to wait for the science to cook in the oven.”

Trump predicted Wednesday a vaccine would probably be available by mid-October, an apparent reference to the Pfizer vaccine.

Pfizer chief executive Albert Bourla said he expects there will be enough data for a “conclusive result by the end of October” during an event reviewing the company’s research portfolio, although he told CNBC he did not expect the data to come before a scheduled Oct. 22 meeting of a vaccine advisory committee for the Food and Drug Administration.

But experts have said they believe at that time, the most advanced trials in the United States will be administering the second booster shot needed to trigger immunity to the last participants — or waiting for participants’ immune systems to muster a response before they can begin to count cases.

“Mid-October — that all defies the math of the enrollment, the follow-up, the second booster shots. It doesn’t even compute, if it’s done right,” said Eric Topol, a cardiologist at Scripps Research Translational Institute.

Pfizer spokeswoman Amy Rose said the company traditionally does not share the full protocol of its trials until results are published but is doing so now in recognition that “the covid-19 pandemic is a unique circumstance and the need for transparency is clear.”

Pfizer’s trial protocol shows that interim analyses of data are planned at regular intervals, when there are 32, 62, 92, 120 and 164 cases of covid-19 among study participants.

Late-stage testing of an experimental vaccine from AstraZeneca is paused in the United States after a possible adverse event in a British trial, even as other countries have restarted their trials. Physicians have been calling for transparency about that event to shore up public trust in the vaccine.

Trump’s comments have often made it appear that people already know the vaccine works and getting it approved is a formality that could occur at any time.

But as the documents provided by both companies show, the reasons for stopping a trial early have been carefully delineated. The data analyses by independent committees allow those panels to recommend whether to stop the trial early — because the vaccine is so effective it would be unethical to continue or because it is clearly not working or it is harmful.

The possibility that the vaccine trials will be stopped early has been endlessly examined by politicians and the media, but experts said it is important to remember that stopping a trial early is a rare outcome, not the norm.

David L. DeMets, a University of Wisconsin at Madison biostatistics expert who has served on data safety and monitoring committees for nearly half a century, said the overwhelming majority of trials are not stopped early.

“If we stop too soon, we won’t have as reliable an answer as we set out for,” DeMets said. “If you stop early, not because the data are mature enough but because of other pressures, then you might not have the same confidence in the answer you want to know or needed.”

While the criteria for stopping a trial are heavily dependent on statistics, that should not be the only factor, Topol said.

“If you stop a trial early, it’s not just a statistical assertion; it’s saying it’s unethical to proceed because of overwhelming efficacy,” Topol said. “You sure want to know it works, and you could say it’s unethical not to proceed, to get more compelling and complete data.”

Approving or authorizing any vaccine on incomplete data would spark a cascade of follow-on questions, including whether the people in the placebo arm of the trial should be given the vaccine right away and how future vaccine trials should be structured. Most experts say multiple vaccines will be needed to end the pandemic, but it might become harder to recruit volunteers, or even unethical, to test a vaccine against a placebo if another has been deemed safe and effective by regulators.

Trump indicated the vaccine would be available to the general public right away after it is authorized. His coronavirus adviser, Scott Atlas, said high priority groups would probably be vaccinated by January.

Prioritizing the people who need the first doses of the vaccine is a huge part of the planning process, and those guidelines have not even been finalized. Discussions have focused on prioritizing doses for those at greatest risk for infection, including front-line health-care workers and older people with conditions that increase their risk for serious illness.

Peter Hotez, dean of the National School of Tropical Medicine at Baylor College of Medicine, said the only way to convince the public of the safety and effectiveness of vaccines that utilize new technologies and are likely to get a form of authorization that is short of full approval is through clear and open communication.

“The world is littered with very good vaccines that never get used due to public perception. It could easily happen here,” Hotez said. “We still don’t know if these vaccines are going to work, we have no evidence they’re going to work [yet]. … The only way you’ll be able to sell this to the scientific community is through complete transparency.”