Navigating the FDA’s Guidance on Real-World Evidence: Insights, Expectations, and Future Trends in Clinical Trials

Scott Swain, PhD, MPH, director of real-world evidence & regulatory science at Cardinal Health, provides context and shares insights on the US Food & Drug Administration’s recently released guidance on the use of real-world data and real-world evidence in interventional studies. 

Transcript: 

Scott Swain, PhD, MPH: I'm Scott Swain, director of real-world evidence and regulatory science with Cardinal Health.

What are your thoughts on the recently released FDA guidance concerning the use of real-world data and real-world evidence, and what are the implications for the industry in 2024? 

Dr Swain: The FDA actually released 3 guidances last year that were at least somewhat related to real-world evidence. So there's one for the design and conduct of external control arms one for decentralized clinical trials. And then there was one for describing using real-world data and real-world evidence for regulatory decision-making. So I assume you're talking about the third one. Interestingly, that was actually an update to a draft guidance that was released in 2021, and it was largely the same as as from 2021. So there were no real surprises in there. I think that overall, this is really good for the industry. It lays out very precisely what the FDA's expectations are when seeing real-world data and real-world evidence that’s submitted in support of a drug efficacy claim. It boils down to a few things. So there’ve been previous guidance is on relevance and reliability this one's a little bit more about processes. So things like transparency with the FDA, making sure that you're there to submit a proposal and then a protocol, and there's opportunity for the FDA to give feedback. The FDA is gonna want the access to raw data there. So, for example, you can't just submit a final analytical data set, or, I mean, worse, just final study results to the FDA. They're going to want to see the patient level data, and they're probably going to want to QA/QC it. 

Part 11 compliance is something that is brought up. A lot of data-extraction tools are HIPAA compliant, but not part 11 compliant. So that's something the FDA has now specific, specifically stated that real world data needs to be extracted in a Part 11-compliant manner.

And then, lastly, just safety monitoring and study reporting, to make sure that you're actually following the procedures as you laid them out in the protocols and other things like that. But I think that at this point the FDA has laid out their expectations on how the entire process should go, not just what not. Just the qualities of what the data and then the generated evidence needs to be. 

Do you expect the FDA’s request for access to raw data to significantly affect research publishing timelines or the time to approval for certain treatments? 

Dr Swain: So based on my experience, I would say that that's not going to be a timeline factor. Assuming that a real-world data study is submitted to them in a timely factor, that's going to be part of the general review process. You certainly wouldn't expect to submit clinical trial data to the FDA and not have them review that.

So I think the expectation that the FDA is going to look at your any real-world data submitted to support safety or efficacy, for that matter, that they're gonna take a look at it. I think that's a pretty reasonable expectation. But no, I don't think that'll affect the timeline.

Looking back at the past year, what are your thoughts on the state of real-world evidence and clinical trials in 2023?

Dr Swain: I think we are transitioning from narrow-use cases to being a bit more creative. So when I was at FDA, most of the real-world evidence submitted was in the external control arms. Like that was just almost everything was external control arms. That is one of many potential use cases for real-world data and real-world evidence. So I think that there is . . . I think we're seeing a transition from just external control arms to other use cases. So, for example, in the last few years several companies have popped up saying that they can use real-world data to either identify patients for decentralized clinical trials, to identify patients for traditional clinical trials, to identify sites to be the clinical sites for clinical trials. So, for example, if you need to know that the patients you're trying to enroll or actually at a clinical site, you can look into data ahead of time and make sure that they're there. One of the biggest problems with clinical trials obviously is recruiting, and by looking at the data ahead of time you can make sure that that's not going to be the, you know, that at least the patients you're trying to recruit are there in the first place.

Other use cases: So there's patient-identification, site identification. I've seen use cases now where FDA is specifically asking for real-world evidence to support a post-marketing commitment or post-marketing requirement.

Also, the other kind of big thing that happened, I guess it's been almost exactly a year, so earlier, in 2022, the FDA released a guidance about having a diversity plan as part of your drug application. And then very late in 2022 that was actually made law. It is now required to have a diversity plan before you submit your pivotal trial, or before you initiate your pivotal trial.

How do you see the use of real-world evidence guiding decision-making moving forward?

Dr Swain: So, taking a step back, we’ve been using real-world data for like as long as it's been there to do a few things. So, formally, real-world data has been submitted to the FDA for like two decades now for long-term drug safety or for pharmacological vigilance. So we've been using real-world data for a long time. The big change that happened a few years ago with 21st century cures is that we can now use real-world data to support efficacy instead of where before that it was primarily safety. But even then, though it hasn't been submitted to the FDA, we've been using real-world data to explore efficacy for a long time, right? So one of the first steps for a drug that's been on the market for a while before expanding a label will be to do a real-world data study and try to figure out you know, who's taking this drug? Are they responding to treatment? Do they have similar safety and efficacy to labeled patients who are taking the drug on label? So we've been using it in that function for a long time, and that's not gonna change. The difference now is you can do a more robust version of that study and submit it to the FDA in support of an efficacy claim.

But then building on what I said earlier, there's . . . I think that you're going to see a lot more use of real-world data to find patients for clinical trials. Not just the sites, but the patients. So that is, I think, just starting to happen. And it's going to become a major use case. Probably in the near future.

Decentralized trials are interesting. It's a form of a pragmatic clinical trial. Pragmatic trials as an idea have been around since the 1960s, but have never really taken off, as the vast majority of clinical trials submitted to the FDA are traditional trials.

I think, though, that the day for an alternative design, such as a decentralized trial, it has to be coming. Part of that is, with cell and gene therapies, for example, that may be curative, the traditional pathway for approval doesn't make any sense, because you may have a really rare disease where you, where it's almost impossible to find patients to recruit in the first place. Or you may have a curative disease where you certainly couldn't randomize somebody to even a standard of care, much less a placebo. So in those cases, building real-world data into the trial design makes a lot of sense.

Beyond decentralized trials there are several different real-world designs that I think we'll see more of in the future. So, for example, I mentioned earlier that external controls are probably the type of real-world evidence that's been submitted to most of the FDA at this point. But external controls so far have almost exclusively believed in historical external controls. I think there's a lot of opportunity for prospective external controls, where you can have your trial at clinical sites, you can randomize most patients to treatment, and then you could have other sites where you are collecting data prospectively, but where you're not actually conducting the clinical trial, but you're still collecting the clinical data for that trial. So that would be a prospective control arm, which is a not quite a decentralized design, but it could be partially decentralized or incorporated into that design. So there's really, there's a, you know, an infinite numbers of way to do that. So another real-world data, pragmatic design, for example, would be to enroll everybody at baseline, and then follow some or all the patients just using a chart review, for example, or electronic medical records depending on what type of clinical data you need. So I think there's a lot of opportunity to incorporate real-world data into both traditional clinical trials and into alternative trial designs. And I do expect that we'll see more of that in the future, especially as we're starting to see kind of this next wave of therapy that is cell and gene therapy. As we see more of that submitted to the FDA and ultimately making it to the market.