Critical Factors Affecting Process Validation: One FDA Investigator's View

Editors Note: In this article Dave Vincent (DV), Editorial Advisory Board Member of the Journal of Validation Technology, and Kristen Evans (KE), an investigative engineer at the U.S. Food and Drug Administration, discuss a variety of topics affecting process validation, including the life cycle approach to validation, scientific rationale, qualification, prospective and retrospective validation, change control, protocols, and critical processes.

KE: First a disclaimer: I've been with FDA 12 years, and I have a lot of experience, but what we'll talk about is really just that—1my experiences and my opinion are not FDA policy. It is not the manner in which FDA disseminates policy, but the information discussed in this interview is intended to be helpful and, hopefully, will be taken that way.

DV: What are some of the major process validation problems you have seen during your inspections of manufacturing facilities in the United States?

KE: I think, as a whole, the failure to recognize the life cycle approach to validation. We see many firms, for whatever reason, thinking that once they complete their prospective three-batch validation, that's the end and they're on their way. I like to say that prospective validation is not the end. It's not the beginning of the end; it is hopefully the end of the beginning.

But, clearly, it's an ongoing process. It requires a concerted effort to really maintain confidence in the process and to be able to demonstrate that at any given time. So, when we conduct our inspections, we want to know how the firm gives itself—and, therefore, us—the confidence that a given process on that day is under control. And you're not simply saying, "Well, we validated it a few years ago," or "We're going to do our annual review in a couple of months, and that will show us," but, rather, systems are in place at any given time to show from a big picture that it's validated.

As opposed to general problems discussed the in the previous paragraph, a more specific problem that we see is a lack of scientific rationale in the protocols and acceptance criteria. At least there is a lack of documentation of such rationale, which is what we're expecting to see. We want the process to be there, that you've come up with a scientific study, a protocol—this is what you're attempting to show, and this is why, and this is how its going to be evaluated, and then just simply executing that. That's documentation of the scientific rationale.

DV: So, usually it's the approach to the actual process validation with regard to using logic and scientific rationale for deriving their studies for the process that you're looking at?

KE: Yes, and oftentimes they have it, they just haven't documented it. I've always maintained that if you have a good process of developing protocols, you're going to develop good protocols. In a good process, that means coming from a logical, scientific standpoint, looking at it as a scientific study, and the whole idea that, this is our objective, and this is how we're going to show it and evaluate it. Take that protocol and subject it to internal peer review. Make sure enough people with varied expertise can look at that and say, "Yes, this is a valid study," and then simply execute it. Then, with all those pieces up front, hopefully the execution goes well, and years later, when we look at that particular study or validation, it's clear to everyone what was done, what the objective was, how it was met, and it's a good, complete package.

DV: When you look at validation packages, is there anything that you particularly look for in the overall validation program when you first go in? Are you specifically reviewing the validation activities? Are there certain key elements you look for? The master plan?

KE: If I'm looking strictly at a validation component of an inspection, there are three levels that I could approach on. One would be the master plan, maybe for a new facility or some enormous project. It's not required to have one, but generally there must be some kind of plan to keep everything on track and, more so, prevent things from just slipping through the cracks. So, I may approach it from the top down at the plant, starting with the validation plan. The second approach is looking at validation from the standpoint of the ongoing demonstration of confidence in the process, not simply prospective validation or annual reviews but what else you have to satisfy your own—and our—desire to have confidence.

The third area within validation that I would look at or could start with would be the building block to validation, and that's qualification. I may just simply look at the qualification of the different sub-systems that go into a validated process, such as equipment qualification, of course. But other systems that the GMPs really say need to be qualified, such as personnel, utilities, how you look at and consider process qualification, how you develop your process and show that it's robust, and then that's one component along with the others. Validation, then, is an exercise on top of that. And that is an ongoing demonstration of confidence.

DV: There seems to be some degree of confusion in the industry on terminology. When you say "process validation," it includes IQs, OQs; it's just the overall documentation of the process activities. I see what other people present, and they say process validation means the PQ activity. Do you see the same type of confusion in terminology in the industry?

KE: Clearly there's a lot of confusion on terminology, not only within the industry but even within FDA. I tend to try to stay out of the confusion by simplifying it. I don't look, per se, for IQ or OQ. I look at the qualification for a system and how a firm goes about basically fulfilling the objective of, let's say, a piece of equipment, deciding what kind of equipment they want, what the requirements will be, and documenting the rationale first. This is often referred to as design qualification. But I don't call it that. It's just deciding what you want, demonstrating or documenting that you've got what you want, and then that it works the way you want, the way it's supposed to work. It is part of the initial qualification but, more importantly, it is a fundamental premise of validation that all of the underlying systems are qualified.

Really, the key component to qualification beyond the initial work is what systems are in place to show that that piece of equipment will remain in a qualified state, so that your validation—your confidence in your process—is going to be built on a solid foundation of well-qualified equipment or processes or people. It's not just the equipment.

I mean in qualification, equipment is one component—the one that gets the most attention—but I tend to apply that qualification concept to all other critical subsystems of a manufacturing operation like people, methods, and procedures. So, once you've established a qualified system, how do you keep it that way? Then change control comes into play in terms of assuring that if you do make changes to a qualified system, they are controlled and don't adversely impact validation or, if they do affect validation, you may indeed need to go back and address that in the process.

DV: I think you brought up a good point regarding training. I think the use of SOPs to support how you write an IQ, OQ, PQ is critical. Having a strong procedural training program is key to the process validation. Is that so?

KE: Certainly. First, before you've even had training, I think the firm needs to decide the process, how they're going to go about demonstrating that equipment is qualified, and then how they're going to demonstrate that a process is validated. Things need to be pretty clear. In today's world, with different companies and subcontractors and a variety of locations that may have to adhere to the same expectations within your company, establishing those standards is critical, as is getting people to understand them and understand the basis for them.

Now, what is the objective of validation and qualification? It's critically important to the point that its become a problem at several facilities, where there's a lack of fundamental understanding of the objectives of validation and, therefore, the protocols. The individual pieces that make up validation are just not well executed, documented, or rationalized. There's no scientific rationale for what they're doing.

DV: When you're conducting a field inspection and looking at process validation as a whole, do you use any statistical tools to determine if a company is in control of their process? Do they routinely monitor their processes? I call it routine monitoring of process, where you use specific statistical tools to see if you are always in a state of control.

KE: FDA does not have any accepted statistical process control program or specification in place that we evaluate a process against. Really, our principal job as investigators is to go ask the firm, "What do you have to satisfy your own desire—and therefore also ours—to demonstrate or have confidence in the process."

So, the first question is, "What do you do to satisfy that ongoing component of validation?" More and more, we are seeing firms come up with statistical, or not necessarily statistical, methods of assessing, in real time, the variability or lack of variability of a process, which indeed we strongly encourage. But our emphasis is to get the firms to sit down and figure out how they're going to do it rather than us pulling out our own SPC tool and evaluating their process.

DV: Let's discuss evaluating batches. Something you don't want to have happen is to have a whole stack of batch production records put on a desk and hear, "You evaluate them for us." It's more that companies should be evaluating and summarizing their own work so we can review the summary. Does that seem to be the case in some instances?

KE: Yes, we don't have the time or resources, and it's not our principal job to make an assessment on the uniformity of consistency or state of confidence of a batch. As I said, the first question is: As a firm, what do you have in place? What systems do you have to assure yourselves that it's under control day to day and, again, not just the prospective validation that was conducted a few years ago or the pending annual review that's coming up or that's often too late to be useful?

So, in order for a firm to satisfy that question, they need to have systems in place that can answer it. That's a very true idea of validation, documented evidence that a process is under control and does consistently produce product that meets specification. The actual definition that's widely published by FDA is a little misleading in that it says documented evidence that a process will produce consistent product. I like to think of it as "does" produce consistent product, not just "will." We'll do three batches and show that it will and then stop. But, really, how does it work from day to day?

DV: That's an excellent point, because I think some companies take that translation as "will" versus "does" it continuously show a state of control throughout the day-to-day activity. I think it's really critical to show that you have a process that's in a state of control. If you're going for a review and you find deficiency in a company's processes, what are the regulatory steps from there?

KE: I guess, if you mean deficiencies in a process, that's really saying you have a process that's not validated.

DV: Exactly. Or attempts were made to validate it, and you found that it took six runs to get to show three reproducible lots.

KE: Yes, well again, with declaring it not validated, there are two instances when we can do that. One is immediately after the prospective validation batches have been conducted, and the second scenario would be down the road, when they have many more batches of experience.

For more information: Terri Kulesa, Institute of Validation Technology, 200 Business Park Way, Suite F, Royal Palm Beach, FL 33411. Tel: 561-790-2025. Fax: 561-790-2065.