How to Read a Clinical Trial

“Functional unblinding does not bias the observed treatment response, but it potentially biases the specific treatment effect, which is a component of the treatment response.” Szigeti adds that regulators such as the FDA evaluate new drugs mainly by that narrower treatment effect — which is exactly what functional unblinding puts at risk, not a patient’s overall reported improvement.

“With the emergence of psychedelics, the problem is so obvious that we can no longer ignore it.” Szigeti says psychiatry spent decades treating blinding as if it neutralized expectancy effects, even though many trials of conventional psychiatric drugs aren’t truly blind either. By his account, the field still has no agreed way to measure expectancy or to fold functional unblinding into trial analysis — a gap he expects will become a research priority in the coming years.

“There is no single solution, instead we should embrace a pluralistic approach, acknowledging the pros and cons of various methods.” Szigeti is separately raising funds for what he describes as the first psychedelic Zelen-style trial — a proposed, still-exploratory design that is not yet running. Patients would be treated openly but not told what the alternative study arms are, which he says mostly removes the feeling of being randomized into a "better" or "worse" arm, provided the control condition itself is made compelling enough that patients don’t feel short-changed.

Heifets separately points to dose-response as a design feature that strengthens interpretation: the most interpretable trials, he says, include "a dose-response with a meaningful range" — meaning "low-dose but psychoactive doses," not just placebo versus a single full dose.

“The quality and quantity of supportive treatment before and after the treatment session. These factors are very likely to influence the long term efficacy, but are going to be very difficult for the FDA to evaluate, and difficult to get reimbursement for.” He adds that despite those obstacles, this factor is "nonetheless critically important" in his view, and suggests the VA health system — not bound by the same reimbursement constraints — as a natural place to test it.

“Long term follow up while still blinded to treatment arm - does not rule out placebo response, but makes this interpretation less likely and certainly less meaningful. If an intervention produces a year long remission, the question about blinding and placebo seems irrelevant.”

“[...] the patient population should be representative of the disease population - particularly with respect to psychedelic use history.”