CDRH Real-World Evidence Questions Answered
A 10/10 CDRH webinar on using real-world evidence in regulatory decision-making answered some of the questions most often asked about the Center’s 8/31 guidance documents on using real-world evidence. Comments received by the Center fell into four major categories — whether the guidance applies only to certain classes or types of devices, availability of a deterministic score sheet for when real world evidence is applicable to a specific regulatory question, the potential for lowering the evidentiary standard used to make regulatory decisions through use of real-world evidence, and clarification on how the use of real-world evidence affects the IDE process.
The guidance makes clear, officials said, that real-world evidence may be provided as support for any regulatory decision related to a device including diagnostics, software, Class 1 and 2 devices, and pre-market and post-market decisions. CDRH says it cannot develop a scoring tool or pass/fail criteria for using real-world evidence. But it did add additional data on what aspects the agency will consider when evaluating real-world data and real-world evidence.
Regarding the evidentiary standard, officials told the webinar audience that the guidance is explicit that the evidentiary standard for all regulatory decisions remains unchanged. “Using real-world evidence does not lower the bar and we also expanded sections of the document that discuss data quality and applicability to a question…,” the presenters said.
Finally, the guidance includes an expanded section on when the collection of data from real-world use might require an IDE.
To respond to questions about data quality, the presenters said that “information provided to us needs to be fit for [its] purpose. It must be complete, consistent, accurate, and needs to contain all critical data elements to evaluate a medical device and any associated claims. Within any benefit/risk decision, the information must be relevant and reliable such that an evaluation of safety and effectiveness can be undertaken. What this means is that certain decisions will require a higher-level data than others. For instance, the first-of-a-kind Class 3 device may require more information than post-market surveillance for a lower-risk device.”