Batch Failures: Significant Risk Events; Challenging Investigation Opportunities

Jack Garvey |

Today’s (4/25/2012) news from (see full link below) highlights information from FDA’s 2011 Warning Letter dataset.   The article states that:

“In fiscal 2011 the US Food and Drug Administration (FDA) sent biopharma manufacturing-related warning letters to 52 facilities. Close to half of these included an observation about failure to thoroughly investigate batch failure.”

The article goes on to state that:

“The finding reiterates the point, made by FDA speakers at industry conferences, that companies should clearly and thoroughly document their procedures and activities. Other companies were warned last year for lacking written rules for investigating out-of-specification results.”

Looking at the observations cited and considering the FDA speakers’ comments, this would seem to be a reasonably easy issue to resolve.   How hard can it really be to look at what happened during the execution of a batch (i.e, by reviewing the batch record) and determine what happened?   How hard can it be to document procedures and activities for batch execution?    Doesn’t everyone know how to investigate anomalous results?

Oh, were things so simple.

The reality of batch failures, and batch failure investigation, is much more complicated.   Some existing documents don’t help.  Some investigation approaches are often superficial.    Getting to true root cause (no, I mean TRUE root cause) is often neglected, frankly, because it’s very difficult to do.  Given that good investigations may not be as simple as having “rules for investigating out-of-specification results,”  let’s look at two areas that are often not considered in batch failure investigation resolution.


Specification review and analysis should be at the core of good investigations.   The question we often ask is:  why did our batch fail to meet specifications?   We try to find reasons related to subpar raw materials, poor execution, different processing, etc. that made the batch react differently.    We never question our specifications since to do so during an investigation, when product is in play, is a huge red-flag for an FDA Investigator .    The FDA immediately  assumes that financial considerations drove the thinking, and that leads to uncomfortable conversations.

Unfortunately, though the question that is almost never asked outside of an investigation is:   Do our specifications align to our process capability?   In other words, do we have specifications that are disconnected from our process capability?   The answer to this question is critical for a company to understand.  More importantly, a company’s failure to understand the linkage between specifications and process capability is something I’ve seen over and over again as one of the true root causes of batch failures.   Put more simply:  When companies set release specifications that are too tight — essentially using specifications for process management / process control purposes — unless your processes are highly-capable and well controlled — you create the significant risk of failing to meet your specifications, when in fact, you may only be experiencing routine process control variation.

Remember:   the FDA doesn’t create your specifications — you do.   You create the box which you need to live within.   Using specifications for process control leads to many more failures than are necessary.  Many more NCs.  Many more CAPAs.   Specifications should be at the boundary beyond which a company will not release product.  Everything inside of that line should be evaluated and considered through control limits, subject to either monitoring, action, or review and decision.   Specifications, though, should be binary.    Batches either pass or they don’t.   Using well defined specifications with well-understood control limits tied to robust product and process characterization will prevent batch failures, but more importantly save time, energy, and MONEY!


Here’s an opening question:   What is the intended purpose of a batch record?    Answer:   A batch record is intended (by FDA) to provide a documented record of the steps, activities and accompanying data (whether from in-process testing, weigh steps, process parameter settings, etc) that constitute the entire manufacturing record of a pharmaceutical batch’s manufacture.   See 21 CFR 211.188(b).

In essence, a batch record is a key compliance tool to document that a defined process was followed in the production of the batch.    Note that I didn’t say that the batch record was a quality tool.    Why not?    Simply because most batch records don’t provide nearly sufficient information — differentiated information — regarding the one or two key steps in almost any process that if done properly, result in a “good” batch of product, and if done improperly (or something else goes wrong) result in a sub-standard or failed batch of product.

If you look at a batch record, or watch QA people audit and review batch records, each step in the end-to-end process appears equally important.  There is never (well, almost never) any differentiation with respect to criticality or importance.  There is no explanation of linkage to poor step outcomes and overall batch quality.    Further, the tricks and tips that the super-operators use to always get batches perfect, are almost never incorporated into the record.    Why not?   Well, that’s a good question.  It’s likely that it’s just the way it’s always been — our record formats and structure are often derived from legacy products — potentially very old products.   However, that needs to change.    Batch quality is almost always defined by perfect execution of the 20% of steps and activities that are most critical to the quality outcome.  Most everything else — material handling operations, weigh steps, label recording, etc. are oriented towards things like potency and mix-up prevention, admittedly important topics.   But the factors — the steps — the precise activities that have big impacts on characteristics like disintegration, dissolution, bioavailability, etc., are often buried in a much larger step, and often, not well differentiated.

So, back to the original topic:   Industry’s failure to properly investigate batch failures.      Even considering just the two foregoing areas of focus, it is easy to see that developing exceptional (or even just plain good) batch failure investigations is extremely hard work.    There are of course other factors:   Time, number of resources, production pressures, etc. — the list goes on and on.   However, given the focus and clear enforcement tsunami this topic has created, industry should seize the opportunity to look past traditional approaches to specifications, batch and investigation documentation and create new, more rational, more differentiated, and easier to investigate document sets.    Necessary for this is improved product and process characterization and an improved ability to explain what is really happening in batch production — and the occasional failure.   Together, this should result in better interactions with the FDA, and very likely, a reduced cost footprint due to reduction in out-of-spec product and associated investigation costs.

For further information on this or related topics, please contact the author at


Half of FY11 FDA GMP warning letters tied to batch failures

Half the pharma firms sent FDA GMP warning letters sent fiscal 2011 were criticised for failing to thoroughly investigate the failure of a batch.

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.