Ask the Expert

Quality by Design for Analytical Laboratories

Zenaida Otero Gephardt, PhD, is an associate professor of chemical engineering at Rowan

Rachel Muenz

Zenaida Otero Gephardt, PhD, is an associate professor of chemical engineering at Rowan University (Glassboro, NJ), where she has served as director of engineering and assistant dean. Dr. Gephardt is also a consultant through Otero Associates. Her research focuses on optimization, development, and mathematical modeling of chemical processes and laboratory techniques. She teaches on-site courses for industry and provides analysis and experimental design support. Dr. Gephardt is a fellow of the American Institute for Chemical Engineers.

Q: For those unfamiliar with the term, what is Quality by Design (QbD)?

A: Basically, what QbD entails is the strategic and efficient use of analytics. It’s taking analytical tools and putting them together to empower the scientist or engineer to analyze data and reach better conclusions with more precision. That’s what the bottom line is, in addition to using risk assessment and focusing on product and process understanding, which are essential elements in QbD analyses. It’s one more tool for the scientist and engineer to make better decisions and improve product quality and safety.

Q: What are the key benefits QbD provides for laboratories?

A: First of all, [it provides] standardization of data analysis. As scientists and engineers, we are really good at standardizing processes, assays, and laboratory procedures. QbD is the same thing in terms of data analysis. What that does is it improves efficiency—you make better decisions because you’re looking at the experimental condition space in a multidimensional way, so you’re going to identify optima and optimal conditions faster and more efficiently, and you’ll be able to reach better conclusions that are more likely to be correct. You can be more certain of the conclusions that you reach and the recommendations that you make because they’re based on an intelligent design of experiments that avoids many of the pitfalls, interactions among dependent variables, and the inability to detect signal from noise that sometimes can be part of one-at-a-time-type experimentation and data analysis.

Q: Has QbD for labs changed recently?

A: There are a couple of things that have changed. There’s a wider range of use of QbD. Originally, a lot of these techniques—Six Sigma, QbD—were for improved manufacturing and production. QbD is more focused on laboratory-type work in part because of the involvement by the FDA in streamlining approvals and applications submitted using QbD techniques and experimental design. But more people are using QbD for more applications. The development of software is another big change for QbD. There is now more software and some of it is [designed for] specific applications that really help people use QbD, so you don’t have to do many of the calculations yourself that were tedious or difficult to carry out [in the past].

Q: What are some other key trends in QbD for labs?

A: One of the most interesting things is the new academic view of the importance of analytics and seeing analytics being included in undergraduate scientific and engineering education. I think, long term, this is the way we’re going to need to go and this is the way we should go because it minimizes the amount of labor associated with experimentation, it’s safer, and it’s more environmentally responsible. Every time we do experiments, we have solvents and waste, so doing fewer experiments, being more strategic, and squeezing every bit of information out of data is responsible in an environmental capacity also. In the future, you really want to develop scientists and engineers from the onset who use these techniques regularly and have these techniques as part of their standard operation in how they look at experimentation and data analysis. Those of us who have been working in this field for a long time have long been advocating for that. It’s very good to see programs both in science and in engineering include more data analysis and experimental design.

Q: What are some of the most common challenges that lab professionals tend to run into when implementing QbD in their labs?

A: One of the most important things is you have to take time at the onset to really understand the question you are trying to answer and then select the analytics tool that best answers that question. Sometimes, I think people who may be applying these techniques and may not use them every day or may not be familiar with them will apply techniques that aren’t useful for the questions they want to answer. For example, there are techniques for small data sets and for large data sets. If you have a very small sample size, techniques that are better for larger sample sizes won’t yield the best results. Because there’s so much software, people tend to apply it and not necessarily look into what the software is doing. In the past couple of years, that’s been the thing I’ve seen the most in terms of situations that create difficulties in the lab for researchers.

Q: What other advice do you have for lab professionals who are looking to implement QbD in their labs but have no idea where to start?

A: I think it depends on how you learn. If you’re a visual type of learner, there are some really outstanding review articles that you can look at to get yourself started and there are some books that are not [completely focused on] QbD, but they [deal with] experimental design and data analysis for scientific applications. You should look at those [resources] from the onset. If you’re more of an auditory type of learner, there are so many short courses and conferences throughout the country that, depending on what your specific area of work is, you can find one that looks at applications specific to what you’re doing. Understanding that QbD is something you do from the very beginning is important. One of the things that I bump into on a regular basis is the idea that, “Well, I can’t use QbD” or “I can’t use experimental design because I don’t really understand the process at all.” That’s actually when you most need experimental design—QbD in general—to really take a look at something [where] you may not know what variables are significant. You can then narrow those variables down and use QbD techniques to gain process and product understanding that can go hand in hand with your experimental program. Many folks in laboratories have not had a course in any technique of this kind. They go into labs as analytical chemists, organic chemists, or engineers, and there’s so much to do. In modern laboratory practice, people are really busy and they’re responsible for more, so when they think of QbD, they think, “Oh no, that’s just one more thing” and don’t realize how much easier their lives can be if they include these techniques as part of their laboratory operation. So, look at QbD as a real helper, as a tool that will help you be more effective, more productive, safer, and more cost-effective. Think of it in terms of something you do from time zero when you start a set of experiments, because it will really and truly help you understand your product or your process faster, more efficiently, and more completely.

Q: How do you expect QbD for labs to change, farther into the future?

A: I think it will expand to basically reach most, if not all, aspects of most laboratory operations. Even small laboratories that do custom manufacturing or custom production for bigger companies are beginning to get on the bandwagon because they see how important it is. It will be, more and more, an inherent part of production and manufacturing. The main reason I think it will move in this direction is because one of the important aspects of QbD is it makes that connection between the analytical laboratory and production and manufacturing more seamless. It connects those two areas where it’s so important to have a strong connection so that production and manufacturing can get the strong analytical support they need and the laboratory gets the samples in the way they need to do their very best job. The other thing that’s happening already is the development of more software so that more people can make use of these techniques. The development of that software will continue and be very prevalent in the next five to 10 years.

Q: Did you have anything more to add?

A: I think it’s really important to encourage laboratory management to see QbD as a critical tool in their operations and that there is some responsibility there, in terms of professional development of your employees, to treat this as something that is important that you should be training your employees to do and not necessarily depend on them finding a way to get this information [themselves]. I think laboratories and companies in general should treat QbD as an important aspect of effective operation and make it more available, easier to use, and more in tune with what their applications actually are.