History offers some examples. During the H1N1 pandemic in 2009, the Centers for Disease Control recommended giving the limited vaccinations to pregnant women, caregivers for children under 6 months and health and emergency services workers.

But each emergency will demand its own reasoning, Silverman said. He said transparency is vital.

"When you're drawing up criteria, you want to make sure you have representation from all of the stakeholders," he said. "So especially people in West Africa, you want to make sure you have those voices at the table, because they may have a different value system from Westerners."

One guiding principle, Kass and Silverman both say, is to manage expectations around an experimental treatment. If an untested drug proves ineffective — or harmful — the results can be catastrophic.

"There is already a fair amount of distrust and rumor [in West Africa] surrounding both why people are getting the disease and what the response should be and whether the health care providers are helping or not," Kass said. "If you put into that context a treatment that turns out not to work, there is a potential that you inadvertently create even more distrust. And what is needed right now, much more than a treatment that might be available to 25 or 30 people in the short run, is having as many people as possible cooperating with regard to when they're supposed to assemble and when they're not, whether or not they're supposed to cross borders, how they're supposed to care for people who are sick, how they're supposed to handle the bodies of people who have died."

That the ZMapp might have failed was reportedly a concern of the colleagues of Sheik Umar Khan, who decided last month not to give him the drug. According to The New York Times, a treatment team from Doctors Without Borders and the World Health Organization feared stoking the already considerable suspicion of Western medical institutions.

"What they really didn't want to do was kill Dr. Khan with their attempt at therapy," Dr. Armand Sprecher, a public health specialist at Doctors Without Borders, told The Times. "If word got out that [Doctors Without Borders] killed Dr. Khan, that would have implications for outbreak control."

Khan died — as did the Rev. Miguel Pajares, the Spanish priest who was given ZMapp.

"Before there's going to be any kind of program, there has to be a really intelligently designed communication program that happens alongside it," Kass said.

Kass said exploring the use of experimental drugs is important. But in the current outbreak, she said, the focus should be on existing tools.

"Ultimately, the best hope for containing things is going to be in the public health systems of Africa — which are, in general, woefully inadequate," she said. "Given that a lot of people are not optimistic that the drug will be a magic bullet — it might be helpful, but it's not going to be black and white, just everybody pop a pill and make Ebola go away — maybe there's not enough discussion still about what kinds of work can the United States or other wealthier countries be contributing to, not only in drug development or drug access, but in bolstering the health systems."

The Ebola virus was mischaracterized in an earlier version of this article. It is not a flesh-eating virus.