Simplify and Streamline

Human-related mistakes during production or testing can often lead to loss in confidence in GXP data sets.  Sometimes, these result in product recall, regulatory action, drug shortage, or other consequences we would very much like to avoid.  Most mistakes are concentrated in those areas of our processes that are highly manual and rely on the ability of the human to perform the action “right first time”, often within a high-pressure environment with multiple and continuous distractions.  As a result, we have implemented the concept of “witnessing” or “verification” to reduce the risk that any mistake will go undetected.  In theory, this makes total sense – but in reality, I have my doubts.  Let’s break it down:

First and foremost, in my experience most companies do not have nearly enough resources to honestly perform the amount of witnessing required per their procedures.  As a result, “witnessing” becomes a compliance activity that holds little value, and is the first thing to be skipped when the pressure is on (Friday afternoon?).  When detected by a regulator, this is considered ‘falsification’ or records, a serious charge that questions the documentation throughout the entire site – as it will likely be interpreted as a cultural practice (think ‘iceberg analogy’).

Secondly, a witness does not realistically reduce the risk of a mistake occurring, only the ability to [maybe] detect a mistake as it occurs.  “Maybe” is the reality – due to the first point -> lack of resources.  It is unlikely that the witness has the time/energy to catch the mistake as it happens, therefore witnessing becomes a compliance formality rather than a meaningful activity.  So the question is: what value does it bring to achieving the common aim of the organization?  Not much…  Why expend massive resources to [maybe] detect mistakes vs. investment in mistake prevention instead?  Simplify and streamline, using your existing technology to its full potential: this should be considered ‘plan A’.

Scenario: You might be worried that the operator does not enter the action contemporaneously into the batch record, therefore have a “verifier” e-sign alongside the operator – despite system audit trails capturing the time/date at which the action was recorded in the background, and could be used to re-create the GXP activity if any concern arose that could be linked back to non-contemporaneous record keeping.  This verification makes little sense, unless we are talking about a lack of QA trust in operations staff due to previous mistakes.   This breakdown in trust can easily happen when mistakes occur, so we must be careful and think objectively.  Typically, the preventative action should not be to increase witnessing, but rather to GEMBA the workflow and identify opportunities to reduce mistakes through simplification and streamlining – taking full advantage of your existing technical solutions (in this case eBR): a much more effective CAPA vs. additional workload piled on operations…

In summary: treat witnessing/verification as a last resort.  First try simplify and streamline using your existing technology as the “witness” and see how it works out.  My bet is a notable example of continuous improvement…

Pete