Wednesday, December 31, 2014

Warning Letter of the Week: No drawers!

From the FDA's list of recently issued warning letters, one to the folks at Novacyl Wuxi Pharmaceutical Co:
Our investigator observed specific deficiencies during the inspection of the API manufacturing facility, including, but not limited to, the following:

1.    Failure to manage laboratory systems with sufficient controls to ensure conformance to established specifications and prevent omission of data.
Our inspection revealed serious deficiencies related to your documentation practices, including missing raw data. It is a basic responsibility of your quality unit to ensure that your firm retains the supporting raw data that demonstrates your APIs meet specifications that they are purported to possess.

For example, during the inspection, our investigator found a chromatogram related to [redacted], API in the trash, dated October 15, 2013, which reported an additional chromatographic peak when compared to the standard. During the inspection, your firm stated that the analyst discarded the chromatogram because it was present in the blank injection. However, the analyst was unable to retrieve the blank chromatogram from the system because it was overwritten by a subsequent injection.

In addition, the inspection documented that your firm made changes to integration parameters for the impurities test without appropriate documentation or justification. Your firm relied upon hand written notes on a chromatogram discovered in a drawer at the laboratory as the documentation for this change. Furthermore, your firm implemented this change without an audit trail that would have captured the date of the change and who made the change.

Other significant deficiencies noted in your laboratory system include:

a)    Failure to have a written procedure for manual integration despite its prevalence.
b)    Failure to use separate passwords for each analyst’s access to the laboratory systems.
c)    Use of uncontrolled worksheets for raw analytical data in your laboratory.
d)    Presence of many uncontrolled chromatograms, spreadsheets and notes of unknown origin found in a drawer.

The lack of controls on method performance and inadequate controls on the integrity of the data collected raise questions as to the authenticity and reliability of your data and the quality of the APIs you produce.
Well, you know, I always have a junk drawer for my chromatograms.  


  1. No data = perfect data? I can't remember the number of times an analyst tried to convince me that the peak I was pointing at didn't exist. The peak wasn't picked up by the approved integration method, so it wasn't "material".
    We try to stuff a method (HPLC) that can perform at best at 1.5 - 2 sigma into a straight jacket system demanding a quasi 6 sigma performance. This a recipe for a perpetual pipe dream. Let's dream on...
    As for the prevalent manual integration... I haven't seen a HPLC trace that didn't need some manual TLC (C = care, not the other C).
    If only nature would finally align itself with regulations...

  2. "If only nature would finally align itself with regulations..." The nature here is human nature. The analysts of the lab inspected apparently have not been required or trained to keep the appropriate records. I'm a QA person and have audited labs in my organization (environmental, municipal government) and have been present when the labs were audited by the state. Good recordkeeping is not difficult but it requires attention to detail and the formation of good habits by the analysts.

  3. Yet another company that realizes they ought to have an LES/ELN only after having an issue with lost documentation

    1. I don't think they want to know (and have records of knowing) what such a system would tell them, let alone pay for it.

      Of course, then, they'd have to admit that their level of effort and diligence is ill-tuned to how much they think it's worth (their dilignece is best tuned to making sand or something more fault-tolerant but sand doesn't sell for the money they want).

  4. 1. I am not happy with the level of diligence described in the FDA letter. Magic drawers and losing raw files cannot be tolerated.
    2. A well functioning system (QA, safety etc.) accounts for and accommodates human nature because this is a good risk reduction strategy.
    3. The inspection revealed defects with some strong causes. Restructuring QMS and re-training should take care of the drawers and records in the short term.
    4. Good risk reduction strategy requires searching for weak causes and their remediation. Without that problems are bound to come back regardless of how many times people are re-trained.
    Some uncomfortable questions to ask include:
    - How well are the direct contributors insulated from business pressures? Is "cheap" more important than "good"?
    - All humans experiment and sometimes deviate from strict procedures. This is how we learn. Are the direct contributors allowed some space to experiment and develop, so that the deviations are minimized where they cannot be tolerated?
    - What is communicated by the upper management? How much message time is spent on profitability and how much on quality/safety/employee development?
    5. Any system that mandates 100% perfection is non-physical and must fail. Risk at all levels can be reduced to some degree (>0%). Reduction of complex risk starts with deciding what level of risk can be tolerated. When regulations require 100% success this discussion is not allowed.
    6. Happy New Year for everyone!