"We did a huge investigation, got to root cause, implemented a CAPA for that cause, and then failed the effectiveness check".
As a quality consultant, I do a lot of work with clients to help them optimize their investigation and CAPA processes. I hear this particular story A LOT.
So, what went wrong? There are many paths to this frustrating situation, but one of the most well-trodden is through the phenomenon of Alias. In this case, I don’t mean the alter ego of a Bond villain, but instead, I’m referring to a logic problem that will be very familiar to those of us with some knowledge of Design-of-Experiments principles.
In short, logical alias is when we cannot tell one factor from another, when both factors tend to travel together. If A causes C, but A and B tend to occur together most or all of the time, we cannot easily know from observation alone which relationship is causative (A to C) and which is merely correlative (B to C). If B is more obvious or visible than A, we then end up developing a CAPA for B, and not A. That CAPA is often useless, or less than useless.
Ineffective CAPAs are less than useless when:
- We miss an opportunity to correct true root cause, and the problem occurs again (with its associated disruption and costs)
- We make unnecessary changes to procedures, training, engineering, etc. to fix something that doesn’t need fixing
- We invoke frustration amongst our colleagues and invite criticism and disengagement (“why do we have to have a CAPA for every major investigation?”)
- We waste time and effort (of the whole investigative team) chasing wild geese
- We expose ourselves to critique during audits and inspections
- And the list goes on.
So, what can we do to avoid being bitten firmly on the posterior by Alias? It turns out that one of the most effective and powerful investigative tools, one that is significantly under-utilized, can really help us here. To defeat alias, we need to explore the full picture of commonality and differentiation, guided by technical understanding. The Is/IsNot tool is super powerful for this, when the characterizations (ie. The rows on the table) are defined carefully, with expertise and knowledge. Very frequently, when I’m called in to help resolve the failed CAPA story, careful application of Is/IsNot will crack the case open and shine a spotlight on a root cause candidate that was hiding in plain sight.
Let’s take a step back and ask why does this happen so often? It’s because when we train investigators, we teach them how to find repeat events, how to brainstorm, how to build elaborate fishbone diagrams, and how to hit that 30-day closure milestone (even if they didn’t start until day 24!). However, we don’t teach the Is/IsNot tool enough.
If this sounds familiar, don’t feel too bad, you’re in a big group of your peers. However, if you’d like to learn how to keep your rear end safe from the snapping jaws of Alias, call us. We’d love to help get your CAPA metrics up to where they should be.
Mark Roache, QxP VP of Cell and Gene Therapies, has spent his 30-plus year career in GXP. Mark was the Chief Quality Officer for AveXis (now Novartis Gene Therapies) at the time of Zolgensma launch. He was previously Senior VP of Quality for KBI (a CDMO with cell-therapy capabilities) and has held other senior Quality roles at Novartis, Merck and Bayer.
