The use of design of experiments (DOE) in quality by design (QbD)

ANNE MILLEY: For those getting started, how would you explain the role of experiments in Quality by Design?

JULIA O'NEILL: One of my favorite areas to work in is development of control strategy, and with Quality by Design, you begin with the end in mind. So you say, first of all, what specifications should be in place for this product? And that's why I have this passion for setting specifications more rationally.

You start with, here are the specifications for the product attributes that have to be there at the end. Then you work backwards and say, what are the process parameters we can control or the elements, the components of the process design that we can select to try to make this product look that way repeatedly?

But then you also have to consider, moving back upstream, what are the starting materials we're going to have to deal with? And in many of today's processes, those starting materials are just full of variation by nature. They're coming from human blood, from fecal matter, things where you can't control the variation. But you have to figure out a way to cope with it.

In Quality by Design, the idea is to modernize the pharmaceutical industry to do a better job of controlling quality from the design of the process to become less reliant on inspection of the final product. I think we're probably always going to inspect the final product, but the pharmaceutical industry has lagged behind others in moving towards more building quality in. Yes, we inspected the end and checked that things worked, but—

ANNE MILLEY: It could move further upstream by understanding the variation before it's messing up your end product.

JULIA O'NEILL: Yes, and one of the most productive areas with DOE is robust design, where we have known variability in the inputs. By making really good choices in the parameter settings for process steps, we can mitigate that variability. Or sometimes it even becomes an advantage, where we might be able to increase yield. That is a classic application of DOE, which I've seen work really well.

ANNE MILLEY: That's very cool, and Quality by Design has implications to other industries as well that they can take those same concepts and benefit from applying them.


ANNE MILLEY: Great. Well, some have said that the adoption of DOE in pharma has trailed other manufacturing sectors, as you were just kind of intimating. Why do you think that is? And have you seen a wider adoption of regulatory influence to use QbD, Quality by Design?

JULIA O'NEILL: Yes, things are really moving in what I think is a good direction. When I first got into pharmaceuticals 15 years or so ago, I was—I won't say shocked, but it was very noticeable that DOE was not mainstream. There were still lots of other approaches being used. But when I go to conferences now that are pharmaceutical-oriented, the scientists, engineers, biologists that are presenting, many of them now are using DOE. It has become well-accepted, and there are rumors that some regulators are looking for DOE in a submission and will not be favorable to one unless it uses a methodical DOE approach.

Now like anything with regulatory agencies, there's no one view or one opinion. There are tens of thousands of people who work as reviewers, so that is not a blanket view. But there's definitely a shift towards not only accepting the use of DOE, but expecting it.

ANNE MILLEY: Good. Well, it's so strategic, and it's been underutilized in my view. There are so many places where it could be used, and it's not being used. So we'll take progress.