WSC 2025 | The interpretation of non-inferiority trials in stroke research

We published a review on it recently on stroke. And this type of trials are becoming a lot more popular in stroke because we now have much better care for stroke patients. So it’s becoming harder to find new treatments that are actually a lot more efficacious. And we also are interested in improving other aspects of care, like efficiency, safety, costs. So these trials are becoming way more popular, but interpreting these trials is a lot more difficult than interpreting the traditional superiority trials. There are a lot of distinctions that make them work a little bit different than the superiority trials. And what surveys have shown is that stroke clinicians and clinicians in general, they have a limitation in interpreting these trials correctly. And the reporting of these trials is also not the most adequate in the literature. So we covered this. What we try, of course, there is a lot of details in the publication and also in the presentation. But what I think is the key point is to remember that the non-inferiority trials, they are not the opposite of a superiority trial, so many things that work for a superiority trial they don’t work or they actually work in reverse for a non-inferiority trial, so we are all familiar that in a superiority trial you want to look at the intention-to-treat population, for example, which in a typical superiority randomized clinical trial that will make it harder to prove superiority. So it acts as a safeguard against incorrect conclusions. But in a non-inferiority trial, that might actually make it easier to claim non-inferiority even for a treatment that is actually a little bit inferior. The other aspect that is very important is that in a superiority trial, you are actually comparing the statistical significance of a treatment and you’re comparing whether a treatment and a placebo are different. But in a non-inferiority trial, you are comparing two treatments that you actually need to be sure that they work. And sometimes you cannot be sure anymore that a treatment that was historically effective remains effective in the current setting because healthcare settings changed a lot. A lot of other treatments become available. The patient populations change. So it’s very important to consider that if the controlled treatment in a non-inferiority trial is not that effective as you expected, that might really compromise the interpretation of your non-inferiority. And you might end up claiming non-inferiority for a treatment that actually is meaningfully worse. So yes, of course, there are a lot of details to talk about. But the key is to go a bit beyond statistical significance; we see a lot in presentations and in papers that people interpret a result, they see a difference between an effect or a safety effect, but this difference is not statistically significant, and then they say, “Oh, there is no difference,” but sometimes that includes a very large effect, there just wasn’t enough statistical power to prove that statistically. But it’s very important to consider, can you exclude an effect that is really clinically meaningful or not? And that is the main point of a non-inferiority trial, to prove that two treatments are acceptably safe. And then, in exchange for that, you get some other benefit, be that cost or medication that is easier to administer or has less side effects or something else that is beneficial for the patient or for the healthcare system.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.