Power calculation & study design for study with rare safety outcome

Question

An RCT is planned to compare a new/repurposed drug to placebo on the basis of mechanistic findings. Limited safety data exist.

The primary efficacy outcome requires the recruitment of ~200 individuals. However, there is a rare safety outcome that should also be evaluated (which individually powered would require ~2000 individuals).

At present, this has been designed as a two-stage design, first evaluating the primary efficacy outcome - & if that is successful, extending recruitment in a second stage to power the rare safety outcome.

What statistical methodological issues should be considered? Would this count as a co-primary endpoint study with the power calculation modified? Or how should the second stage recruitment be dealt with/powered?

Answer 1

Most trial designs include a separate open-label extension study to evaluate safety. In your example, the participants might be randomized in a double blind fashion to receive experimental therapy or placebo in a 1:1 fashion over 12 weeks, after which they are given the chance to receive the experimental study treatment for a duration of, say, 52 weeks (or longer as needed). This design tends to improve the estimand/inference by recruiting participants who would, as patients, be more likely to take the drug in real-life. By having a longer duration of follow-up, we can reduce the number of patients needed to complete follow-up. Recruiting blinded-treatment naive patients after the first cross-over to receive unblinded therapy can provide important stratification and adjust for various types of experimental bias.

We can't really power trials to detect safety, since we can't be assured of the frequency of the safety outcome. The number needed depends on the rate and also the severity of the safety outcomes. A Data Monitoring Committee (DMC) should review safety data every, say, 6 months or longer/shorter depending on the frequency/occurrence of AEs.