We talk a lot about data integrity, but what happens, in the real world, when US Food and Drug Administration (FDA) data integrity protocols are not met? In December, we got to see this scenario play out.
Between November 22, 2022 and December 2, 2022, FDA inspectors visited a large pharmaceutical manufacturer’s facilities in India 9 times and found a “cascade of failure” that resulted in the issuance of a Form 483.
A Form 483 is essentially a legislative slap on the wrist. In this instance, this 36-page form details where the FDA inspectors found concerning or flagrant violations of data integrity protocols. If this organization fails to meet compliance, the issue could be escalated to a warning letter, followed by a ban from selling in the US.
How Did They Go Wrong?
The Form 483 issued to the organization in question describes a scenario where it appears that employees were scrambling to hide data and paperwork from the inspectors.
On the first day of inspections, November 22, 022, the investigators discovered a truck containing clear, plastic bags, which held shredded documents. There were also black plastic bags which held documents that had been, apparently, torn to pieces by hand. The inspectors found additional plastic bags with ripped up GMP documents stashed under a staircase.
The inspectors did what inspectors do and investigated these suspicious situations further. When they opened the back, they were greeted with a strong chemical smell – and realized the torn and shredded documents were wet.
In turn, their quality control management played off the smell with an explanation about a spill that was wiped up with tissues, which were then thrown away in those bags. The inspectors continued the line of questioning, though, which exposed the concerning truth: while the investigators were initially doing a walkthrough, a quality control employee tore up balance printouts and Auto Titrator spectrums. They threw the pieces out, followed by dousing them in an acid solution. This attempt to destroy evidence revealed he was trying to hide the tests he was working on which showed failing quality specifications.
With the help of some other employees, the FDA investigators pieced together some of the documents. They found the destroyed documents contained pertinent information for products sold in the US.
Additional investigations showed general inconsistencies in data management and recording. The Associated Executive Vice President of Corporate Quality and Compliance reported “the weighing activities recorded on the “Instrument Usage Log” is not a true representation of samples [sic] weight with start and end time for each weighing activities for a specific lot of a product. Further, there is no consistency among the [Quality Control] employees in term [sic] of recording of information in LIMS logbook pertaining to a total number of lots tested. Some [Quality Control] employees may enter this information whereas others may not, leaving no traceability for the exact number of lots tested and their start and end time of analysis”.
On top of consistency failures, environmental monitoring samples were counted inaccurately, either as a result of underreporting or incorrect calculations. There was concern that the incorrect calculations were intentional to ensure samples passed acceptance criteria. Moreover, their quality control department invalidated failed results and did not give a scientific explanation, along with a failure to have data that demonstrated their manufacturing process controlled enough to ensure batch-to-batch consistency.
Why Does Data Integrity Matter?
If you have even an ounce of experience working with highly regulated industries, federal inspections, and data integrity, this whole situation should raise a series of red flags. Unfortunately, many facilities have gotten lax since the pandemic, when the FDA significantly reduced in-person inspections. In the last year, though, they have returned to in-person inspections. In turn, they had to issue 31 Form 483s in 2022, flagging non compliance across the globe. Sadly, 15 of the 31 forms issued during 2022 were to Indian pharmaceutical manufacturers.
While Form 483 does not solely pertain to data integrity, this is one area where many manufacturers lapse. Data integrity is complex, but critical, so it can be easy to make missteps. It is also the area where people might be tempted to lie or hide information in an attempt to pass inspections, as demonstrated in this example.
While intentional “errors” are unethical and break compliance, genuine mistakes in data integrity are frequent, as well. That is why it is important to fully understand 21CFR11.
21CFR11 refers to the 11th part of the FDA’s Title 21 Code of Federal Regulations and it is the portion of the regulations that guide manufacturers on how to handle data that has to be submitted to the FDA. These guidelines outline the importance of ALCOA+, a method for accurately tracking and storing information. ALCOA+ stands for:
Attributable: The author of the data must be very clear, with a signature and date.
Legible: The data must be easily understandable, including explaining any symbols.
Contemporaneous: Data should be written down or otherwise recorded immediately after being generated.
Original: The data should be the original or a certified copy of the original.
Accurate: Data should be an accurate reflection or interpretation of the situation..
Complete: Records should include all information – even testing and re-testing.
Consistent: Data should be entered consistently, time stamps should be in the expected sequence, and data generation should stay the same across the board.
Enduring: Data should only be recorded using invalidated electronic systems (or, if unavailable, controlled worksheet in laboratory notebooks).
Available: For the lifetime of the record, it needs to be readily available for review and audits (i.e., not ripped up, thrown away, and doused in chemicals).
While ALCOA+ is a great place to start for data integrity, truly maintaining data integrity involves equipment designed for compliance. If you would like to avoid human errors, using equipment that tracks and records data and prompts important inputs is the best way to go.
