OR WAIT null SECS
Without a rigorous discussion of the pros and cons of QbD, its tremendous benefits will be lost.
Quality by Design (QbD) is a promising opportunity for biopharmaceutical companies to capitalize on the current regulatory and industry focus on drug product quality and process control by leveraging science, compliance, and investment for the benefit of patients and stockholders. Companies that recognize the potential benefits have made excellent progress with QbD and have successfully filed applications with the US Food and Drug Administration, QbD's primary advocate. Others, who have used QbD tools informally, are considering more structured cross-functional approaches for product development. Many others, however, are standing on the sidelines waiting for what they see as the latest quality management flavor-of-the-month to run its course. The challenge for QbD implementation is to convince these sideline skeptics. This article discusses reasons for the skepticism and provides suggestions to managers for how to implement QbD in their companies.
Depending on whom you talk to, you will get different descriptions of what QbD is. Some see it as a product approval quid pro quo to raise the regulatory compliance bar or the latest flavor-of-the-month that adds bureaucracy, cost, and time to product development. All QbD really means, however, is that products, and the processes that make them, are designed in advance to meet their product quality specifications and process control requirements.
The truth is that QbD is based on solid science, valid statistical tools, and sound management techniques that have proven their industrial value for many decades. These include product profiling, experimental design, process mapping, risk assessment, and automated process controls. The goal of QbD is to apply these tools in a structured manner.
These techniques are suspected of taking too much time and costing too much compared with the traditional trial-and-error methods that have been prevalent in development. But a better approach is to structure an experimental design that optimizes the available development time and project funds while maximizing the product and process data delivered. If this sounds too good to be true, consider that design of experiments (DOE), the heart of QbD, offers returns that are four to eight times greater than the cost of running the experiments in a fraction of the time that it would take to run one-factor-at-a-time experiments. (See box on DOE). Furthermore, these approaches are the only way to unlock interactions of multiple factors, something trial-and-error and one-factor-at-a-time experiments cannot. When the QbD approach is used, products are purer and more potent and processes are more robust and have higher productivity. Also, technology transfer risks are reduced because fewer start-up runs and less start-up time are needed, which can lower start-up costs and speed the time-to-market.
The univariate trial-and-error approach to product development is traditional and can be safe in the short term. With the time pressure to be first to the market, it is an easy way out. Although this methodology may produce safe and effective products, there is considerable room for improvement and risk reduction. Product quality, process productivity, and profitability can be improved with further in-process refinement and reduction of product variability.
If the traditional approach is ok, why should we be concerned with cost and productivity? Companies, governments, and health maintenance organizations (HMOs) face conflicting budget priorities, expanding demand, and higher costs per person for healthcare. Product cost and the cost of quality and compliance are under increasing pressure to be justified or reduced. In short, improved quality, higher productivity, lower cost, and improved compliance are the roadmap for the future. These goals can be achieved with QbD application. Neway has efficiently made the business case for QbD.1
Most people are willing to change; they just don't like being changed. The desire for change needs to be internal. Culture is a big factor in the adoption of new ideas and approaches. Some companies are open to change and others are resistant. Management must lead the way by setting a good example. Although management support is critical, it can only go so far. At some point, scientists and engineers must understand and accept the reasons for and benefits of changing to a QbD approach. One way to do this is to identify and discuss why they are skeptical and less than willing to change.
Activities like QbD and process analytical technology (PAT) have been standard practice at chemical companies since the mid 1950s. These companies embraced these approaches because it saved them large amounts of money and time. But chemical companies, of course, are not regulated the way the pharmaceutical industry is. Our regulations create a different environment, and we must adapt to that difference. Validation presents the most obvious constraint.
But even validation, now accepted as a basic requirement, was once a new concept. When validation was first expected of companies, a number of articles were published that argued against it.2 Interestingly, despite all the skepticism about QbD today, we have yet to see an article challenging it. Instead, the doubts seems to be expressed more quietly, perhaps out of fear of contradicting the regulators who support it. But there must be a discussion of the pros and cons of QbD. We must be able to justify QbD in real terms of reduced risk to the patient and money and time savings. Without a vigorous discussion, it will quietly fade away and the tremendous benefits will be lost.
Design of Experiments (DOE)
Concerns About the Concept
The skeptics have valid concerns that must be answered clearly. During the webcast of "Pharmaceutical Quality by Design: The Road Ahead," the participants were polled for their concerns about QbD.3 According to the responses, 39% were concerned about lack of clarity on what the concept means; 15% had time constraints (it takes too long) and 13% had budget constraints (it costs too much) for QbD implementation; 13% were concerned about the lack of management support and 11% about the inability of quantifying potential benefits; and 9% had concerns about lack of skilled personal. These concerns are expected and timeless. Every new endeavor faces similar comments.
Addressing the Concerns
The lack of clarity is being addressed by the FDA and industry organizations by presenting conferences, seminars, training, and publications. The FDA has been very proactive in giving talks and being available for discussions. Participants must continue to read, reread, and discuss the available information until the concepts and ideas become clear enough for implementation.
Time is money, and thus the concerns about time and cost can be discussed together. Any new activity is awkward, slow, and inefficient. A student just learning to use a keyboard is frustrated. But with determination and practice the student gains proficiency and speed. Eventually, the keyboard becomes second nature and the attention is on the contents of the document and not the skill of typing. So it is with QbD. At first, it is strange and uncomfortable. The old approach seems faster with fewer trial runs. In the short term, the old ways seem better. But QbD takes a lifecycle approach. The goal is to minimize total costs and time from initial development to product retirement. "Pay me now or pay me later," was once a jingle for motor oil, meaning you could pay to change the engine oil as needed or replace the engine later. The same applies to development. If you short change the development stage, you may pay later for more rejected lots, stability failures, 483 citations, warning letters, and recalls.
Management's support is essential; without it few initiatives prosper and succeed. As W. Edwards Deming so forcibly told us, "It is management's job to work on the system so the employees can work within the system." For example, management needs to make QbD easy to implement. Anything less than full support is worthless.
The inability to quantify potential benefits is a management issue. Cost of quality systems have been used for many decades in the field of quality control.4 Modifications to these systems can potentially be an easy way to initially carry out a cost–benefit analysis. A paper published in the year 2000 titled "The Cost of Non-Conformance: The Linkage Between Quality Performance and Business Results" also proposed an approach to be considered.5 The authors developed the "Cost of Non-Conformance Model" to capture the business impact of non-conformance with quality standards and regulations.
Management should start with the employees that are eager to participate, and encourage, support, and reward them in public. Others will join in as they see the rewards and benefits of participating. Formal and informal training is needed for topics such as statistics and DOE, and the related software.6
Other comments by skeptics have been used to deflect encouragement to implement QbD and DOE. Here are some of the more common ones with replies.
"This is just the latest fad, the flavor-of-the-month. Just watch, this will be replaced by some other new corporate initiative by the next VP or CEO." This could be the case in some companies. Progressive companies that have a long-term view of success, however, recognize the benefits of embedding QbD into the company culture.
"Time-to-market is our primary driver. We need to be first in the market; we will fix any problems later." Time-to-market is important, but being first in the market with a third-rate product and process is not a good route to success either. Again, the concept of lifecycle optimization underlies QbD.
"Our development is more art than science. Designed experiments don't work for this. It takes years and years of experience to develop our products." This is an admission that sources of variability are not known and are not being controlled. Process understanding is a key concept in QbD and PAT. Without control of material and process variabilities, the process capability of any new product is just a guess. A foundation of variation control is fundamental to any development.
"Statistics was my most difficult class in college." Unfortunately, many university statistics classes are too theoretical and are often intended for mathematics majors. The field of statistics was born at the junction of biology, genetics, and mathematics in the mid-to-late 1800s as practical way to deal with variability and large amounts of data. Applied statistics courses, however, can be very helpful, particularly if they are sufficiently pragmatic and data oriented. Most applied courses require only algebra. The root cause of the problems is that most students are not exposed to any statistical concepts until they are adults in college. If the process development staff lack a solid foundation in statistics, management must support inhouse training or send staff out for applied courses.
"If my major professor didn't think this (QbD, DOE) was important enough to teach it to me, then it must not be anything I need." This is understandable, regrettable, and self-perpetuating. There can be two reasons for the professor's failure to promote QbD. First, the professor's goal is to teach the science subject, not statistics. Like knowing algebra or grammar, the professor assumes the student has the background or will get it later. The second reason is that for most courses, the focus is on theory and not on applying or implementing the concepts using data.
Finally, we must acknowledge two human failings—the failure to get expert help and the lack of willingness to endure failure. It can be difficult for successful people to ask for help in a subject they don't know. They feel that they should be able to master the topic themselves. Again, management needs to step in with support and encouragement. Encouragement and money are further needed to weather the inevitable failures.
Not all endeavors succeed and not all experiments are breakthroughs, but we learn from the failures. Experimentation should be seen as guided learning that builds the knowledge base of the company and thus competitive advantage. QbD is a process, not an event. Knowledge gained in one set of experiments is used to refine and design the next set of experiments. QbD is an investment, not an expense.
It is human nature to resist new and different ways of thinking and working. Given past management fads, skeptics have a legitimate concern about the long-term sustainability of QbD. But the 50-plus years of its application in the major chemical companies is an assurance that it can be implemented successfully. QbD offers future benefits to companies and management willing to invest in the time and effort to be competitive.
If QbD really is the competitive future, how can we learn more about it? The key industry organizations delivering education and training in the principles and tools for successful QbD implementation are the American Society for Quality, the Parenteral Drug Association, the American Association of Pharmaceutical Scientists, and the International Society for Pharmaceutical Engineering. The courses, seminars, and workshops sponsored by these organizations have demonstrated that projects using QbD approaches can be more efficient, less costly, and data rich in the development of new products. Process scientists have also learned how to mine data using these tools to improve existing products and processes.
Take advantage of this opportunity. Learn more about QbD and use the tools to enhance the products delivered to patients and also improve the financial bottom line.
1. Neway J. How to make the business case for Quality by Design. BioPharm Int. 2008:21(12):42–47.
2. The Problem of Process Validation," J. R. Sharp, The Pharmaceutical Journal, January 11, 1986.
3. PharmaQbD web cast; 2008 Oct 14. Available from www.PharmaQbD.com.
4. Gryna FM. Quality and Costs. Juran JM. Godfrey AB, editors. Juran's Quality Handbook. 5th ed. New York: McGraw Hill; 1999. p. 8.1–8.26.
5. Dwyer T, Keresty G, Sherry B. The cost of non-conformance: the linkage between quality performance and business results. Pharmaceutical Eng. 2000 Sept/Oct:8-18.
6. Design Ease, Inc. Design Expert, Version 7.1. Available from: www.statease.com.
7. Fisher RA. The arrangement of field experiments. J Ministry Agriculture, England. 1926;33:503–513.
8. Torbeck LD, Branning RC. Designed experiments-a vital role in validation. Pharm Technol. 1996;20(6):108-114.
9. Fisher RA. The design of experiments. London: Oliver and Boyd;1935.
10. Box GEP, Wilson KB. On the experimental attainment of optimum conditions. J Royal Statistical Society. 1951(13):1–45.
11. Rathore AS, Branning R, Cecchini D. Quality: design space for biotech products. BioPharm Int. 2007;21(4):36–40.
Lynn Torbeck is a statistician at Torbeck and Assoc., Evanston, IL, 847.424.1314, Lynn@Torbeck.org and Ronald Branning is the vice president of corporate quality assurance at Gilead Sciences, Inc., Foster City, CA, 650.522.5282, email@example.com