Why Preventing ‘Orphan Data’ is the Key to Ensuring Data Integrity
This article was originally published in The Gold Sheet
Executive Summary
The light bulbs go off when clients realize the importance of establishing controls to prevent orphan data, consultant says.
The key to preventing data integrity failures in pharmaceutical manufacturing is to prevent the occurrence of “orphan data,” Crystal Mersh, president of Quality Executive Partners, told “The Gold Sheet” in an interview.
Mersh, who had spoken about data integrity in March at the International GMP conference in Athens, Ga., said she had found that it’s really important for clients to understand the problem with orphan data.
“We see light bulbs go off as soon as people understand ‘OK here’s what I’m trying to prevent, is this orphan data residing there.’ Then they understand why the controls are important and how they have to work together.”
She often sees people “who have a false sense of security around audit trails.” They don’t realize that there’s more to it than turning on audit trails, which provide a record of activity in electronic systems.
“It’s not just enough to have the controls in place, but you have to have those controls integrated, including audit trails, security controls, file structure. They have to all be there but they also have to be all working in concert.”
So for example, she said, the value of an audit trail depends on when and how it’s reviewed.
“The primary concern is that they want to make sure that there are not OOSs [out-of-specification results] that are being generated and then not being made visible to the quality assurance unit during the batch disposition process. And as a result that’s why the orphan data is so critical.”
What orphan data is
Orphan data is information that never made it into the quality system, Mersh explained. “It can be data that is test injections, it can be in folders labeled as ‘wash,’ it can be just labeled as the same batch number, but it never gets processed and then processed forward and then ultimately into the QC data package that goes to QA for batch release.”
For an example of orphan data, Mersh pointed to a warning letter FDA sent Jan. 30 to Apotex Inc. in Toronto, Canada, concerning the company’s manufacturing facility in the Bommasandra Industrial Area on Jigani Link Road in Bangalore, India.
An inspection that ended July 1, 2014, “found that 2,803 of 44,643 injection results were not processed or reported in the official data folder for dissolution analysis,” according to the warning letter
.
The test injections “had never been reported in the formal QC data, so they were just residing in the computer system,” Mersh explained. “And FDA’s concern is that there could be OOSs in there related to a batch that really never had the view of the quality assurance team when they did the batch disposition.”
If people find orphan data, they need to check their systems – their audit trails, their file structures – to find out what’s allowing it, and make sure going forward that the systems work together to prevent it.
It’s also important to check orphan data to see whether any out-of-specification product needs to be recalled from the market.
How to prepare for inspections
Companies should expect that every inspection will have a data integrity component and prepare accordingly.
Mersh advises clients to “make sure that you can sit down and proactively explain your controls and how those controls prevent orphan data.”
She said it’s important to develop confidence in the controls built into the system “before you go out into the lab with the investigator to actually sit at a computer.”