OR WAIT null SECS
Agnes Shanley is senior editor of Pharmaceutical Technology.
Instead of rigidly applying statistical tools, experts suggest that pharma embrace statistical thinking, but focus on reducing variability and adding value for patients.
Six Sigma, both a statistical term and a method developed by Bill Smith, an engineer at Motorola in the 1980s, has allowed companies in a number of industries to improve their business processes, product quality, and overall financial performance. The concept was championed by Jack Welsh when he was CEO of GE, and a number of pharmaceutical companies, including Merck and Johnson & Johnson (J&J) adopted the practice enthusiastically in the past.
The principles of Lean Manufacturing, including its focus on the customer and employees and its emphasis on minimizing waste, enriched Six Sigma’s overall framework of “Design, Measure, Analyze, Improve, and Control” (DMAIC). The resulting “Lean Six Sigma” (LSS) programs allowed companies in a number of industries to achieve significant improvements in efficiency and product quality.
However, LSS programs need senior management support if they are to work, says consultant and statistician Ron Snee, coauthor of Leading Six Sigma - A Step-by-Step Guide Based on Experiences with GE and Other Six Sigma Companies.
In cases where this has happened, improvements have been dramatic, he says. Snee recalls one company outside of pharma whose Six Sigma programs were driven first by three managers, saving a few million dollars a year, snowballing into $30 million in savings over five years, after the CEO assigned a vice-president-level champion to the project.
A decade ago, LSS programs were trumpeted by many pharma companies. Today, efforts are more subdued, and not only in the pharmaceutical industry. “It has been a challenge in other industries as well, to maintain excitement and initiatives around Lean Six Sigma,” says consultant Tara Scherder, a chemical engineer and statistician. Now a partner with Synolostats, Scherder trained employees throughout the enterprise as part of Merck Sigma, Merck’s formal enterprise-level Lean Six Sigma program. Today, LSS programs that once spanned the enterprise live on in a muted fashion, limited mostly to smaller initiatives in manufacturing.
Why has LSS failed to become standard practice at more pharmaceutical companies? For one thing, Snee says, people misunderstand the concept. “Some think it’s only about quality improvement, while others just see it as training. It’s actually about improving the overall performance of an organization.” In addition, he notes, some may be fixated on the definition of Six Sigma, in terms of number of defects.
In the process, Snee says, managers may fail to integrate programs and efforts with the bigger picture, and overall management goals. “Programs must be relevant to the needs of the business,” notes Scherder. “You must be agile and focus on business context and solutions, not statistical tools,” she says, especially because these tools are a relatively small piece of a larger business solution, and they are often not required at all.
That overemphasis on statistics was a weakness in many corporate training programs in the past, says Scherder, at the expense of other beneficial, relevant principles and activities that should be part of LSS. For example, she says, it takes time, and mentoring, for individuals to absorb statistical concepts, and to apply them properly to specific business issues, considering the context of the data involved. “Typically, non-trivial statistical concepts (e.g., design of experiments) come fast and furious at the student without time for practice, or adequate interpretation,” she says. This can lead to a statistical toolbox mentality, in lieu of statistical thinking. Often, she notes, problems won’t even require a statistical solution. “Learning how to assess business and data context and the need and choice of statistical method requires time and contextual practice. This won’t be found in JMP or Minitab or any other statistical software out there,” she says, noting that the “toolbox mentality” naturally leads to non-value-added analysis instead of the simplest solution.
Scherder is also concerned about pharma being compared to other manufacturing industries and criticized for failing to achieve Six Sigma levels. For one thing, notes Snee, Six Sigma has not been achieved for all processes, even within electronics or automotive industries, and it may not always be desirable to achieve it, given business goals. Consider the myriad processes involved in producing a pharmaceutical tablet, Snee says. Which steps would make sense to bring to Six Sigma levels? “Implementing the six sigma improvement process does not mean that a company has attained, or planned to attain six sigma quality in its processes.”
“Admittedly, the industry has room for improvement in the adoption of LSS principles,” says Scherder, “but simple comparisons to other industries are unfair.” Consider, for instance, the relative ease of incremental change. The regulatory burden associated with process changes in pharmaceutical manufacturing is extremely high compared to that in other industries, which impedes motivation for continual improvement, she says.
Industry and regulators recognize the benefit associated with reducing this burden over the lifecycle of a product. Signs of progress include the 2016 draft FDA guidance for comparability protocols, and the drafting of a guidance (ICH Q12) for lifecycle management of post approval changes by the International Council for Harmonization (ICH).
In addition, in pharma, manufacturers do not always have customer-based specifications, such as tolerance for a car component, or the dimensions of a semiconductor wafer, says Scherder. Instead, specs are often derived from process performance. This is particularly true for some biopharma attributes, where the mechanism of action to the patient cannot be simply described, she says. In these cases, the sigma quality level is essentially bounded to three, she explains, because the specifications are based on the expected distribution (the mean +/- 3 standard deviations, or similar statistical interval).
Historically, Scherder notes, when variability has improved over time, regulators have required update of the specifications to reflect the tighter performance even though the tighter specifications were not required for patient safety or efficacy. This practice caps the sigma quality level at three or less, she says, because the specifications bracket only the new performance. This type of specification adjustment is not prevalent in other industries.
The use of specification ranges that simply bracket the expected process variability in lieu of true customer derived specifications must be considered in any valuation or comparison of pharma industry sigma quality performance. Clinically relevant specifications (customer derived in terms of LSS), are receiving more attention within the industry. The topic is discussed in a paper by Yu and Kopcha (1) (see Sidebar), and the International Society of Pharmaceutical Engineers has established a working group to focus on this issue, she says.
Although Merck’s CEO, and a few other CEOs in pharma, supported Six Sigma and Lean Sigma, “pharma has not yet had a Jack Welch figure,” notes Snee, while companies may abandon an approach that focuses on individual projects.
But problems with some past programs may also be to blame, says Scherder. At some companies, she says, there may have been too much emphasis on training large numbers of Black Belts who went forth with an arsenal of complex statistical tools to solve problems. Over time, at some companies, the title Black Belt may have become synonymous with ‘someone who complicates things.’ “Today, business priorities have become highly focused,” says Scherder. “You have to prove yourself to be relevant and effective in moving product along the lifecycle,” she says, or improvement programs will be considered a cost that can be eliminated.
Scherder sees a need for pharma’s LSS programs to focus on the fundamental concept of understanding and controlling variability and driving that throughout the organization. She also sees a need to incorporate the best aspects of Lean thinking in training. “The concentration on many statistical tools taught too quickly comes at the price of holistic problem solving. Instead, we need to develop problem solvers who will drive statistical thinking throughout an organization, to understand variability, reduce waste, and improve processes,” she says.
“Everyone needs to have a line of sight from product/process development to commercial supply and back. Application of statistical thinking across an organization would enable the connections needed to optimally develop and continually improve processes,” she says. What would happen, she asks, if process and analytical development professionals were made aware of long term variability that could affect process capability and continual improvement?
Resource prioritization should be given to activities that will provide value or protection to the patient.“If a process exhibits high process capability, there’s no patient benefit to chasing variability that has negligible safety or efficacy implications. In such cases, patient needs are better served by resources spent on new product development, and less capable processes,” she notes.
Finally, she says, there needs to be a focus on activities that will add value to the patient. Some companies have developed LSS approaches incorporating the best elements of both DMAIC and Lean. Amgen, for example, took an approach that focused on process monitoring, incorporating elements of Lean, and setting targets for process capability to improve individual process performance and reduce cycle time as well as overall waste (2).
1. L. Yu and M. Kopcha, International Journal of Pharmaceutics, 528 (2017), pp. 354-359 (June 2017).
2. M. Van Trieste, “The Journey from Good to Great: Process Monitoring Leads to Improving Product Quality,” paper presented at the Second Annual FDA/PQRI Conference, October 5, 2015.
Vol. 41, No. 10
When referring to this article, please cite it as A. Shanley, “Reinventing Lean Six Sigma for the Pharmaceutical Industry," Pharmaceutical Technology 41 (10) 2017.