FDA Staff Update on Data Integrity Issues
A panel of FDA staff from CDER and the Office of Regulatory Affairs provided the latest insight into common data integrity inspection findings during a 9/28 session at the annual FDA/Parenteral Drug Association regulatory conference. One common issue is firms’ review of electronic data, such as electronic chromatograms, according to ORA consumer safety officer Sandra Boyd. She said that the agency is seeing firms reviewing electronic data packages, but not at a level that provides assurance that the samples tested were actually the most recent samples.
“If a data packet is turned in on a Friday, what’s the assurance the sample reviewer has that it wasn’t tested on a Wednesday, or the Monday before, if you’re only reviewing the packet that is turned in,” Boyd asked. “So, how do you evaluate all the injections on your system? You can do it numerous different ways, whether you have naming conventions for your samples or your method so you can search and evaluate your electronic data that way. You might control the folders that the analysts have access to so that the review can go on. So it’s just some sort of reconciliation of your samples so you know what’s going on in your system.”
Another data integrity issue, according to Boyd, is documenting investigations contemporaneously, especially with software that people are using to track investigations. “If anybody is looking at an investigation that was started three months ago, the documentation should be more than just, ‘we initiated this investigation and this is what it’s about,’” she said. “And a lot of times that is what I find in the system, or it says ‘ongoing.’ So, somebody has it on their desk and tracking things on a notepad, but it’s not getting into the system and it’s only being uploaded at the very end. All this investigation stuff should be taking place contemporaneously.”
Boyd said that another top data integrity issue is not having the scientific rationale to support decisions. She said she has seen firms thoroughly begin investigations, “but when you dig past that first level, you need the data to support your conclusions and your decision making process. So, if you decide that your facility is not going to test products for micro, then have data to support that. Don't just verbally tell the investigator coming to the door, ‘Well, it has low water activity and we don't have to test for micro.’ Have the historical water data or articles used or documented evidence that you used to make this decision... If you say ‘my media fills are based on routine interventions’ then show me evidence that you have been tracking the number of interventions that took place during production so you would have that scientific evidence to support your decisions. So, make sure that you actually have actual data to support the decisions you’re making.”
CDER branch chief Quallyna Porte told the conference that FDA is still finding a lot of firms not having well-established or existing data governance plans, which is to ensure across a facility that the data is appropriately handled and that all core FDA principles are met. "So, we’re still finding that unique passwords are still not in existence at some firms,” she said.
CDER Manufacturing Quality Guidance and Policy Staff director Tara Bizjak said that over the past five years, anywhere from 21% to 34% of Warning Letters have included citations about unreliable data. “So this is something that we continue to see, and I think we're starting to see another layer as well,” she said. “Not only is it unreliable data that investigators are finding on inspection, but now unreliable data that we’re seeing through remote evaluations.” Bizjak also said that some subsequent inspections that have occurred after a Covid-19-related remote evaluation are finding different data documentation from what was provided during the remote evaluation. “So I think that there are a lot of layers to this, and data integrity continues to be an issue.”