[This is an answer to the question, “How can we run drug trials much more quickly than we managed in 2020?”]
2020 phase 3 vaccine trials averaged 107 days between starting phrase 3 and submitting data for approval. We can reduce this to 30 days (perhaps fewer) if we make it legal to infect volunteers in a controlled environment. That will reduce infection and death rates, as vaccination can start earlier
Broadly speaking, you need 100 people to “get it” to seek approval. You then look at the odds-ratio for those who got the placebo, and those who got the treatment.
Imagine 4 groups of X volunteers, each in separate safe environments (cruise ships are cheap to rent). Half of each group get the vaccine, and half the placebo.
Group A (half treated, half placebo) get the virus swabbed into their noses immediately. They would be PCR tested daily from this point. Within 5 days we will know whether one dose of the vaccine provides any degree of instant immunity. If sufficient, the drug company can apply for approval immediately. Even a vaccine with 30% efficacy would be (much) better than nothing. We can also assess whether it reduces the severity for those who still get the disease.
Group B (half treated, half placebo) get the virus swabbed into their noses after (say) 21 days and are PCR tested from then onwards. This would test whether one dose is effective after 21 days.
Group C would be given a second dose after 21 days, and the virus swabbed into their noses immediately after that. Again, they would be PCR tested from that point. This would test 2 doses.
The efficacy of B and C would be apparent within 30 days.
Group D would be given the second dose, and virus swabbed after a further 21 days. The efficacy of this option would be known within 50 days.
How many people do you need? At a first approximation, you need 100 infections to apply for approval. Since the vaccine could be 100% effective, you therefore need 100 placebo people to get infected. Each of groups A-D therefore need 200, plus an allowance for the non-guaranteed nature of the nasal swab infecting procedure. If you think that only 1 in 4 nasal swabs will take, you need 400 placebo, and 400 vaccine in each group (total 3200) plus a margin of error. This is an order of magnitude lower than the 43,000 in the Pfizer test.
Vaccines tend to increase efficacy with two doses and a delay, but it is unlikely that Groups B, C and D would show an effect without Group A showing some effect. In general, therefore, companies would be able to apply for approval within a week. I understand that the Pfizer vaccine shows about 60% efficacy for Group A, 90% for Group B, slightly higher for Group C and 96% for Group D.
Pfizer would therefore have submitted their vaccine for approval on August 3, within a week of the trial commencing, 110 days earlier than under the current system. They would have reported c. 60% effectiveness. In the absence of anything better, even if there was no effect on transmission, we would want that drug.
Pfizer would, however, continue with Groups B, C and D. By the end of August, they would be showing Group B’s 90% efficacy, as well as reporting on the effects on severity of illness. We would be even keener on the drug.
When the second dose raised efficacy to 96%, Pfizer would file again, seeking approval for a 2-dose approach.
This approach has other advantages.
1) It allows us to capture the asymptomatic cases much more accurately than field trials, where many will have it, but not know it.
2) We need far fewer volunteers, although they must be willing to be on a cruise ship for between 14 and 60 days.
3) Because the numbers needed are smaller, and the results faster, it would be possible to do proper RCTs on more plausible drug candidates, and certainly on different doses
4) It would be easier to monitor side effects in a confined setting.
5) It is particularly well-suited to a hard-to-administer vaccine (e.g. one requiring very cold storage) as all doses will be in one place. Training to administer is also easier.
People sometimes argue that it is unethical to infect people in this way. They might say that if 3200 volunteers are infected and if the vaccine is ineffective, 50 might die. That number is not valid. First, people with strong co-morbidities would not be accepted. It is not ethical to infect such people. Second, everyone will be monitored daily. They would have PCR tests daily, and anyone found to be positive would be observed and treated accordingly. We would not expect anyone to die although it is possible that a few will die. Without this proposal, 50m people who did not volunteer got it, of whom more than 800,000 died. It is better that there is a small chance that 1 or 2 volunteers might die, than that 800,000 should perish.
(Note that if a disease had a high R and a high death rate, the optimal ethical policy would be widespread vaccination with every plausible vaccine we could manufacture, after only a phase 1 study to establish safety and plausible efficacy. It is never ethical to infect people with a disease like Ebola).
259-11