GXP - The Data Integrity Issue

Data Integrity Background and Frequently Asked Questions

With six guidance documents issued since January 2015 and a stream of FDA warning letters on the topic, Data Integrity has assumed proportions the likes of which have not been see in the pharmaceutical industry in the course of my 30 year career.

In the 1980’s we had the generic drug scandal.  The 1990’s brought the Barr court case, the predecessor to data integrity.  2005 saw data integrity issues which bankrupted Able Laboratories and since then, regulators have realized that there are practices industry-wide that aren't limited to India and China, which impact the reliability of data in regulatory filings.  The industry had not done enough then and speaking for mysef: this was a wake-up call.

Although data integrity bad practices are way beyond computerized systems, the TItle 21 CFR Part 11 regulation came into effect in 1997 and since then the requirement for audit trails has been in place.  Yet in the hundred’s, possibly thousands, of audits I have conducted, I only started to ask to see audit trails about two to three years ago.  I have been asking myself why it didn’t occur to me that just having an audit trail was useless without review and action on findings, and I don’t have a good answer.

In the future, wherever a feature is required by a system, we all need to review that feature from time to time to see if it is being used / implemented as required.  Audits of audit trails have shown me:

(i)         User 1 under every entry – a clear indication that a single password is used on the system

(ii)         Entry deleted -with no explanation as to why and under “user 1”

(iii)        Entry altered – no explanation why and under “user 1”

When asked by IVT Network to conduct a brief datasheet of FAQs for DI, I thought it would beneficial to allow me to crystallize some thoughts.  Overall, as with every aspect of a quality system, it is about clearly defining requirements.  The six guidances referenced, divide into three classes:

  1.  Q&A – so a bit like this article – only inspectors FAQs.  That is the FDA and EMA Data integrity guidances.
  2. True Data Integrity Only Guidance – MHRA – first for GMP and the later draft on GxP.
  3. WHO and PICs on data management and records which essentially are quality system guidances addressing quality culture and metrics with a slew of other topics.

Top management must define acceptable ethical behavior and be sure to avoid pressurizing employees into bad decisions.  Each employee must understand what are acceptable and unacceptable practices, and then be accountable for their actions.  Before firing an employee for a DI infringement, management really should ask themselves if they made it clear that the practice was unacceptable and if they didn’t unconsciously encourage the practice.

With that, here are frequently asked questions with answers which are not necessarily the only answers or even correct – but seem to me to be logical, ethical and practical.  Any comments, corrections and discussion which will help to improve the practices I’ve been disturbed to find, will be welcome.

1.Is it acceptable to shred GMP related documents?

Once a document has reached its archive expiration date (e.g. one year after the product expiration date or seven years from production of the batch or whatever is in your company’s SOP) you can shred the document PROVIDED (at least in the EU GMPS, chapter 4) that document does not contain data submitted in the marketing application. In that case, you need to keep it forever.

2. But what about reconciled, printed packaging materials?
After line clearance close-out, excess materials actually should be destroyed in a controlled manner.

3. So can we have a shredder in production?
I would propose that the shredder should only be available to Quality Unit employees so that you carefully control access and thereby can better control which documents are shredded.

4. What about excess printouts from an HPLC because the printer failed?
Well, not so long ago, I would have said, “part 11 compliant system, prints the number of print outs, then ok to shred if not needed.”  Now, I would advise having an SOP that describes what, how, and who may shred any unwanted printouts, and that the analyst who prints such a document, writes on it why they believe it can be shredded. For example, the printer got stuck.  Then it should get sent to Quality Unit for their review before getting shredded – that’s a controlled procedure.  Certainly the practices that I have seen of stacks of unwanted printouts in the lab being used as scrap paper raises serious concerns as to the ease with which copies are generated.

5. How soon after completion of a manufacturing step should a batch record be signed off by the reviewer?
When asked this question I split the response into several parts:

(a) Do you have an SOP describing the responsibilities of the person who signs:
- performed by
- verified by
- checked or reviewed or approved by
When I ask this, the answer is usually “no.” Employees aren’t necessarily clear that if they sign “verified by,” they need to have been physically present throughout the step and to have observed every part of the activity as if performed themselves.  Whereas “checked” “reviewed” or “approved by” is a data review in which case you must be provided sufficient information to make a reliable assessment of the data before signing.

(b) Having understood your responsibility, a person reviewing a batch record is assessing the reliability of the data.  Therefore, one Contractor who told me I couldn’t see my client’s batch record at an audit three weeks after manufacture because their VP Operations hadn’t yet reviewed it was essentially telling me straight out that their data was not very reliable.  They were also giving a sub-conscious message to the production employees that filling in the batch record on real time was not all that important.  The mentality was to keep moving product and to fill in any missing information in the record…LATER.  This type of practice is at the heart of data integrity issues.  If the batch record review has to be completed within 48 hours of performing the step, the record has to be filled immediately and any missing information will be rapidly challenged.  This is a cultural issue.

6. Is it ok to correct an error in a record with initials and date?
Actually, I am never asked that question.  Instead, when during an audit I challenge corrections missing the reason for the change in information I am regularly told “the error is initialed and dated and that’s our SOP.”
This too is cultural.  We shouldn’t hide behind an SOP.  If a number, or any other piece of information is altered from the original paper record (if computer, there should be an audit trail and the audit trail should be provided – electronically or as a printout, with the record), the reason should always be documented.  If I need to release a batch, and someone altered a yield calculation or a time or date on an entry and I don’t know why they did that or on what data they relied…I have a data integrity issue.

7. Do we have to review the audit trail for every batch before release?
Firstly, there are two types of audit trails: system and data related.
The system based audit trail relates to changes in directories and programming changes, the attempts to alter metadata (the data giving context and meaning to the actual data).  Therefore, the system based audit trail needs regular review preferably by the IT department rather than by the user or superuser (the user might be QC analyst or production operator and super user – lab manager or production manager).
The second audit trail is the data specific trail relating to the particular HPLC run or batch production run.  Depending on the data criticality, that data needs to be reviewed if considered critical, prior to or at the time of batch release to be sure that the batch data is reliable…AND (a big and), don’t forget audit trail review of ANY data which is going into a regulatory submission, stability data, toxicological data, QC data – because you really don’t want to find later on that data was deleted or altered or cherry-picked – i.e. deliberately left out from the regulatory submission.  All of the above needs to be covered by SOPs and internal audits and oversight mechanisms.

8. Are paper based systems less susceptible to manipulation than computerized systems?
I have been around long enough to remember (just about) before computers and for many years was highly suspicious of computerized systems.  Today I have undergone a 180 degree turn and firmly believe that ONLY FULLY computerized systems – carefully designed, implemented, validated and with a program for maintenance and ongoing monitoring, can provide the answers we are looking for regarding data integrity. 

As an example: weight printouts.  I always thought those were the ultimate answer to data integrity.  You weigh, you have a printout, you stick it in the notebook and sign – half on, half on the page so you would see if it was removed.  A client pointed out that the analyst can quite easily change the date and time on the balance so that if they don’t like the weight or the result is “just a little out of specification, ” they can make a new printout with a different weight.  Now that might seem like something really deceitful – but if you need to get out to pick up the little kid from nursery and you are really close to the “right” result – in a company where the opportunity exists, you might make a bad choice. However, in a company with a LIMs system where the balance feeds directly into an electronic notebook, there is no opportunity to do this. 

There is a Bible commandment saying to “not to place a stumbling block before the blind” – computerized systems can prevent the opportunities.  Although as an industry we are not very proactive in our computerized systems. When I benchmark on electronic notebooks in QC labs – I rarely get hands raised at all in a group of 2-300 people.  This makes it abundantly clear that the pharma industry still has a way to go.

These were only a few frequently asked questions on data integrity, but there are many other hot topics that will require an updated segment to this piece in the future.

Table 1 – History of Guidance Issuance (final and draft, since January 2015)

11 Aug 2016

EMA: Good Manufacturing Guidance to Ensure the Integrity of Data

10 Aug 2016

PIC/s: Draft Guidance Good Practices For Data Management And Integrity In Regulated GMP/GDP Environments

July 2016

MHRA: GXP Data Integrity Draft Guidance

April 2016

FDA: Draft Guidance: Data Integrity and Compliance with GMP Q&A

Sept 2015

WHO: Draft Guidance: Good Data and Record Management Practices

January 2015

MHRA: Data Integrity Definitions and Guidance for Industry




Product Added Successfully

This product has been added to your account and you can access it from your dashboard. As a member, you are entitled to a total of 0 products.

Do you want access to more of our products? Upgrade your membership now!

Your Product count is over the limit

Do you want access to more of our products? Upgrade your membership now!

Product added to cart successfully.

You can continue shopping or proceed to checkout.

Comments (0)

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
  • Use to create page breaks.
Image CAPTCHA
Enter the characters shown in the image.
Validated Cloud logo