1
$\begingroup$

I have a question regarding sample size (power) calculations when considering time-to-event data. When there is previous data for an RCT that showed non-inferiority (e.g. a simple two group set-up, treatment A and B, followed-up for 5 years) with a given hazard ratio, e.g. 0.7, [CI 0.4-1.0], what would be the best way to calculate the required sample size to be able to detect whether treatment A is superior to treatment B? I understand that there is a need to define a superiority margin, but have no idea how to decide on this? Thank you!

$\endgroup$

1 Answer 1

1
$\begingroup$

A hazard ratio alone is not enough to determine sample size for a clinical study. The hypothesis testing framework is irrelevant. Instead you need to understand the event rate accrual, at least in the control arm. This can be expressed as the median event time. Most software assumes an exponential event distribution, then there is a 1-1 correspondence between the median rate time and the rate parameter $\lambda$, so that the event rate in the active arm can be solved from the hazard ratio if not provided explicitly. The other inputs or considerations for a sample size calculation is the participant attrition (loss to follow-up), the accrual rate of participants, and the total study duration. Each of these is a consideration for design, and not provided externally. Statistical software such as EAST can calculate this, or you can just as easily design a simulation to identify the required sample sizes.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.