While reading a trade magazine recently, I was struck with the detailed information on how to design and implement a Corrective and Preventive Action (CAPA) system. ISO 9001:200, Clause 8.5.2 states, “…the organization shall take action to eliminate the cause of nonconformities encountered…” For medical devices, 21 CFR Part 820.100 states, “…each manufacturer shall establish and maintain procedures for implementing corrective and preventive action…” Similar wording appears in 21 CFR Part 211 for drug companies. Do we really need all these regulations to ensure we make the product as we are “supposed to” be making it?
Now, it strikes me as interesting that we still consider (under cGMP, of course) the NDA (New Drug Application) as “perfect” after all our trial, production-sized batches. What did you say? We only make three batches? But surely, numerous hours of physical and chemical design went into designing the product. And don’t forget how smoothly we went from pilot-size to production-size. What? A full-time staff of troubleshooters, you say? That many people are needed?
So, based on that information, we submit our “perfect” formulation and a plan of what to do when it fails to work. Hmmm…what’s wrong with this picture? An online source says, “Implementing corrective action and preventive action systems for FDA-regulated manufacturers is a necessity to guarantee quality and ensure compliance with Current Good Manufacturing Practices of the Quality System (QS) Regulations.”  Well, I’m just a humble country chemist, but (already with the “but”) if we did our homework and actually knew what we were doing, why do we need a corrective action plan?
I remember saying something recently about GMPs stating that relevant measurements should be taken during the process to ensure good product (back in the 1970s, even). I must have missed the part that said to run blindly until we go out of spec then do an investigation to find out why. From speaking with associates who routinely do OOS (Out Of Specification) investigations, there seem to be two categories of findings: the obvious (error in the lab, error in weighing in the plant) and the “como sabe (who knows?)” type, which leads to never knowing.
It is a fact that many, many lots of products simply go out of expiry in the warehouse, waiting for a cause of failure to be found. The problem with current production (under cGMP) is that if we do not know what the critical steps are, how can we find where the product made a wrong turn? We’re not even sure when the product makes a correct turn! We approach the production of pharmaceutical products much like using directions from MapQuest.com. Whether or not a bridge is out, we proceed down a road. Why? Because we went down that road for the first three NDA batches, that’s why.
PAT is the moral equivalent to using common sense: if we see that a bridge is out, we take a detour. In cGMP, we don’t look for the bridges; we simply say, “It worked last year, so let’s keep driving…we’re covered by cGMPs (read: SOPs).” Check the particle size of the raw materials? Wasn’t in the filing; forget about it. What about polymorphism? Nope, wasn’t in the document specifications 10 years ago. How about flowability, compactability, pore size or density? Nope, nope, nope and nope. I guess it’s much easier to conduct several (dozen?) OOS investigations each year than understand the process. [I have a button that states, "Several months in the lab can save an hour in the library." Not exactly the same, but close.]
Now, please don’t misconstrue what I am saying. There is no reason not to have a plan, should some disaster strike a production run. Not having such a plan would be akin to leaving your house without insurance. However, to stretch a comparison, that doesn’t mean you can leave a pot on the stove and go to the store. Having a “disaster drill” and “evacuation plan” doesn’t mean not taking all precautions to prevent the disaster.
Assuming that because a product ran smoothly three times in a row (possibly years ago), it is not necessary to keep a vigilant watch on all current incoming materials and process parameters is the ultimate fantasy (yes, even including that website). Under the “traditional” way of manufacturing, using the “three strikes, er, uh, batches and you’re out” approach, it is impossible to do business without a well-thought-out CAPA in place.
What I am saying is that the idea of a CAPA has to morph into something new under the QbD and PAT manufacturing paradigms. In lieu of a change-control based approach, designed to find the problem with a single batch (“Round up the usual suspects”), we will have an ongoing CAPA, using “design space” mini-adjustments throughout the process. Since I already used a driving analogy, I’ll try another: current CAPA documents wait until we go up a curb, then we try to find where we hit a pothole to make us lose control. Using a PAT set-up, we monitor the road constantly and avoid the potholes before we hit the curb.
All we are saying is give PAT a chance… (apologies to John Lennon). We can’t do it sort of, part way, keeping some elements of the process static and changing others. That’s like being “partially pregnant.” DO IT!
- © Copyright 2000-2006 MasterControl. All rights reserved.