I have done my share of complaining about human subjects review by “Institutional Review Boards.” They can sometimes get worried about things that pose no real danger to experimental subjects or survey respondents. But if I were on a human subjects review board myself, I would be skeptical about a challenge trial with a placebo arm. Given the benefits of getting vaccines vetted quickly, I think challenge trials that expose volunteers to a virus after they have been vaccinated can be justified, but exposing volunteers to a virus without vaccinating them seems a bad idea to me. The statistical purity of having the placebo arm where someone randomly doesn’t get the vaccination and then is exposed to the virus is of some value, but to me doesn’t have enough value to warrant exposing unprotected people to the virus.

In “What Fraction of Participants in a Randomized Controlled Trial Should Be Treated?” I look at how a greater cost of treatment as compared to having someone be in the placebo arm could make it optimal to treat less than 50% of the subjects in a double-blind placebo-controlled trial. In a challenge trial, after including the danger to the subject, having someone in the placebo arm is much more costly than giving someone the treatment. As a result, the roles of the treatment arm and the placebo arm in the equation shift. It would be optimal to have far more than 50% of the experimental subjects and have much less than 50% in the dangerous placebo arm.

But even that calculation pretends that only those randomized to the placebo before being exposed to the virus provide information about what happens when someone is not vaccinated. What is happening “in the wild” in the pandemic provides some information about what happens when someone unvaccinated is exposed to the virus. And what happens in vaccine trials that do not involve deliberate exposure to the virus also provides evidence about the placebo. The most crucial input for vaccine development is to know what happens when someone is vaccinated and then exposed to the virus.

In the UK a human subjects review panel is allowing something of a placebo arm, but the amount of the virus the volunteers are exposed to is to be kept very small. Here is the description in the Wall Street Journal news article “U.K. Approves Trials That Will Deliberately Infect Volunteers With Coronavirus”:

With medics and scientists standing by 24 hours a day, researchers will inject closely controlled doses of coronavirus into the noses of quarantined volunteers. This set of volunteers won’t have received vaccines. The idea is to start with the smallest possible amount that allows researchers to gauge infection levels, symptoms and transmission methods while trying to ensure the volunteers’ safety.

If that goes well, a second stage of the trials is planned to include the use of vaccines—researchers haven’t specified which ones—to test how well they protect against symptoms and possibly transmission. Vaccines would be administered to healthy volunteers, who would then be purposely infected with coronavirus, again in closed-off quarters to contain the virus.

One thing I have learned from this pandemic is the importance of the quantity of viral particles one is exposed to. So there may be a quantity of viral particles low enough that a placebo arm for a challenge trial is reasonable, but it would be easy to get that quantity wrong. And if you have to start at very low levels and then go to higher levels, you are using up some of the speed advantages challenge trials. Still, it may make sense.