“I, on the other side, have said I’m not disagreeing with the fact anecdotally they might work, but my job is to prove definitively from a scientific standpoint that they do work,”
There is a disagreement. Fauci may not have ice-water in his veins, and Trump knows the meaning of “Show me the money”, but the two are close to opposite poles. Some philosopher, so obscure I can’t Google him up, said, “Truth is found at the meeting of opposites.” If we were flies on the wall, their differences might sound like that.
Medicine restricts “proof” to double blind testing, which means that neither the physician or the patient know whether the actual treatment, or a placebo, is being received. This avoids prejudice in the observations of the clinician. Even subjective observations of the clinician, like “patient seems stronger”, have value in a double-blind study.
What of the patients who are harmed? The ethical rationale is that it saves lives in the long run. Let’s consider a double-blind drug trial which enrolls 1000 patients. Half receive the drug-candidate. Half receive the placebo. Unknown to the experimenters, the drug has one of three actions:
- Promotes recovery.
- Has no effect.
- Harms the patient .
After some period of time, the results are “unblinded.” For the first time, it is possible to see who was getting what, and the likely effects of the drug. Results:
- If the drug is bad, no more than 500 patients will get hurt.
- If the drug has no effect, nobody gets hurt.
- If the drug is good, how many people get hurt?
A double blind trial, with patient consent, is a unique loophole in our legal system, and the exception to the Hippocratic “Do no harm.” Suppose the drug is one which may have a very long use. For the next 50 years, the drug will be the standard of care. The double blind method attempts to ensure the maximum benefit for all future patients. It does this by risking the lives of current patients for the benefit of the indefinite future.
Now I’m done extolling the virtues of double-blind testing. It has these disadvantages:
- It is too slow for a plague situation.
- It may be sacrificial of people’s lives.
- Alternatives exist that, sacrificing scientific purity of thought, would likely save more lives.
See (NCBI) Probability, proof, and clinical significance. We live in a world of probabilities. In a study of 1000 patients, there is a possibility that the study results, good, bad, or indifferent, are due to chance. This chance is called P. Fauci’s acceptable P is probably the conventional choice, 0.05, a 5% chance the study is wrong.
Increasing P means the study can finish sooner, with earlier availability of a drug. This comes with a greater chance the study is wrong and the drug useless or harmful. This is hard to justify if we’re not in a plague situation. But note, compassionate availability of unapproved cancer drugs has been the standard for some years now.
Increasing P can be justified in a plague situation, because of the otherwise certain death of large numbers of patients. It means that regardless of the intent of the drug trial organizers, their choice of P directly affects the survival large numbers of non-study-participants.
This would be evident if a cost/benefit equation, for a drug trial in plague conditions, with the goal of optimizing P, were part of the canon of applied medical statistics. But it’s missing.
The technical resources available to Dr. Fauci could quickly provide the missing equation, minimizing cost by varying P in line with the current crisis.
My guess: P= 0.25. It shortens the required length of the study, with the likelihood (chance always intrudes) of saving the lives of the large numbers of critically ill who are not study participants.