Drug Makers Take Aim at Variability

Jan. 1, 2004
Six Sigma, Kaizen, and Poka Yoke Are Reducing Error and Improving Product Quality
By Angelo De Palma, Ph.D., Contributing EditorReducing human error has been a goal of manufacturing since tools were made from mammoth tusk. Best implemented as an ongoing process rather than a one-time event, successful error-reduction programs incorporate science, art, psychology, and a big dose of common sense.

Human error extends up and down the pharmaceutical supply chain. The United States Pharmacopeia reported in October, 2003, that healthcare workers often interchange medications with similar names--for example Visicol, used for bowel prep before colonoscopy, and Vicodin, the narcotic painkiller. Serious mistakes occur, even though two different words are staring someone in the face. It's no stretch to assume that errors can occur at every stage in the fuzzy world of manufacturing.

Production managers tend to fall into two camps with respect to human error: Fatalists believe errors are inevitable, while optimists hold that mistakes can and should be reduced. "We subscribe to the second philosophy," says Bonnie Smith, managing director of Lean Sigma Process at TBM Consulting, Durham, N.C.

For Smith, error-laden processes manifest themselves in wasted time, which percolates venomously through everything an organization does, ultimately showing up in the product. Conversely, "when you reduce the time needed to get things done you improve quality and become more competitive by reducing lead time and/or increasing capacity."

Smith cites a biotech company that took thirteen days for fermenter changeover, an operation that adds no value to manufacturing and, necessary as it is, "takes time away from productivity." Smith convinced management to develop new protocols, locate all supplies on a tool trolley before the changeover, and revise the spare parts list so that everything was ready to go. The very next changeover took just four days.

What was management thinking, you might ask, as it squandered nine productive days several times a year? Smith explains the psychology of wasting time by analogy to a simple home improvement job. "Only a fraction of the time it takes to hang a picture in your living room is actually spent hanging the picture," she observes. "You first have to find the hammer, the nail that fits into the hanger, and maybe some wire. Before you know it a five-minute job takes an hour. In pharm and biotech, where production is king, value-added operations are scrutinized, but companies don't spend much time trying to improve the other stuff."

Wisdom from the Orient

Taiichi Ohno, credited with saving Toyota from oblivion in the 1940s, identified seven "wastes" of production: overproduction, non-value-added transportation of goods, non-value-added movement of people, idle time (unprepared workers or operations), non-value-added processing, excess inventory, and defects. Like Bonnie Smith fifty years later, Ohno recognized that value is realized not only by optimizing core activities, but through streamlining day-to-day operations.

During their rise to pre-eminence in manufacturing, the Japanese perfected several techniques for error-proofing and quality management. Kaizen, or continuous improvement, seeks to eliminate operational waste by incorporating the best concepts of lean manufacturing and six-sigma quality practices. Kaizen literally means "school of wisdom" (kai = "school", zen = "wisdom"). Related is poka yoke, literally "error to avoid" which is usually translated as "mistake-proofing."

Kaizen goes hand-in-hand with six-sigma variability reduction--the elimination, as much as is humanly possible, of inconsistencies in processes and in the finished product. Six sigma is not the same as quality control, Smith points out. QC detects substandard product after it comes off line, whereas six sigma seeks to prevent defects from occurring in the first place.

In her practice, Smith comes across companies at every conceivable level of competence and ineptitude. No matter where she finds them, she strives to make them better at what they do.

"That's why it's called continuous improvement. It doesn't matter where you are --you can always get better. Maybe your changeover on a packaging line is "just" one hour. We view that as one hour of wasted time. We've achieved our goal of reducing that time by fifty percent again and again."

Many pharm/biotech companies, says Smith, have been converted to lean manufacturing and six sigma, but the industry still has a long way to go. "Some are using one or the other, but in reality they should be using both because the goal should not be only to eliminate waste, but to reduce variability as well."

Missionary Work

If lean and six sigma are so effective, why doesn't everyone employ them? More and more companies do, says Smith, but even after managers agree to such programs, they must then commit to the discipline of putting them into action. "Implementation is the killer. Management is usually receptive to these ideas but doesn't execute them well. Without management commitment you might as well forget it."

A kaizen program's logic and potential benefits often clash with big egos, one or two of which have been known to reside in pharmaceutical establishments. Smith recalls one manager whom she thought she could challenge with her standard promise to improve throughput by 50%. His response, "Are you telling me I've been doing this wrong all this time?" is not easily countered, especially when the answer is "yes" and you want the guy's business. Another hurdle is the "Ph.D. syndrome," which goes something like this: "We have so many Ph.D.s running around here " why can't we figure it out?" A corollary to the Ph.D. excuse is that kaizen and related methods couldn't possibly work because they're too simple.

Pharmaceutical manufacturing carries enough inefficiency, says Smith, that startling improvements are possible without even touching regulated, value-added operations. Companies can work within Good Manufacturing Practices (GMPs), for example, focusing only on preparation, mechanics, and other day-to-day activities, and still garner many of the benefits of lean manufacturing and six sigma.

"If you don't want to go down the road of changing your process, re-validating, and all that paperwork, look at variability," Smith says. "Work within the system."

Compound Failure

To Gary Gamerman, a consultant with Seraphim Life Sciences Consulting, human error is the number-one cause of manufacturing problems, followed closely by bad process design, which he believes is itself is a factor of human error.

Errors affect manufacturing on at least two levels: process errors caused, say, by not operating a piece of equipment properly, and breakdown in analytics or process/quality control. "You rarely see one type of error without the other," Gamerman observes. "Letting bad product out the door almost always results from compound human failure."

Fixing the second-worst consequences of human error (the first being a death or injury) in a pharmaceutical plant--batch failure--is expensive, but according to Gamerman it usually takes a lot of batch failures before some companies take notice. "They'd rather throw away $100,000 batches than sit down and figure out what's wrong because usually they don't have a good idea of costs related to batch failures. Or they'll say "we don't have the money to fix this" because the losses due to batch failure and the cost of fixing the process come from different budgets."

Errors resulting from failure to perform a routine task are common. FDA cites many such errors in its warning letters. "That's scary enough, but it's hard to imagine all the mistakes that have to be made when, for example, a plasma products manufacturer sends out X units of product contaminated with hepatitis B. That means at least three people messed up."

Gamerman categorizes serious human errors according to three mindsets:

  • I know it's bad process (or operation) but I don't want to make any waves."
  • "This process is too difficult to follow!"
  • "If these three signatures are on it, I'm sure it's ok."

Mindset number one is easy to fall into, given the difficulty, time, and expense of validating a manufacturing process. Only a brave manager or a fool would point out a problem that could delay a project by weeks or months. Mindset number two is related to the confluence of unclear regulatory expectations and a lack of preparedness, which combined can complicate even the simplest tasks.

As an example of what can result from mindset number three, Gamerman cites the mistaken bombing of the Chinese Embassy in Belgrade in 1999 by NATO forces. "This incident had the approval of a dozen people, any one of whom might have consulted a map or telephone directory and prevented a tragedy." More typically, the three-signature excuse allows recommendations for action to go through where something on page one contradicts something in the abstract.

Taking ControlGamerman suggests an error-proofing strategy based on:
  • Designing processes from the ground up with error reduction in mind;
  • Building in error detection whenever possible; and,
  • Regular, periodic review of all operations that includes statistical process control.

The point of hazard analysis is not to eliminate every error, but to reduce risk to a tolerable level. "Hazard analysis tells you that if a hundred things can go wrong, you should focus on, say, ten that provide the best return on your effort." Hazard analysis estimates the probability of a hazard occurring, and the worst possible consequence of that occurrence. Hazards are then tabulated by probability times consequence values. Events above a predetermined threshold are dealt with first.

Fault-tree analysis, a related approach, recreates process steps on paper and asks what can go wrong at each juncture. Fault-tree lacks the quantification of hazard analysis, but provides a visual, more intuitive picture of potential pitfalls.

Tangled Up in GMPs

According to Steven Bizar, an attorney with Philadelphia-based Buchanan Ingersoll PC, definitional gaps in GMPs make mistakes almost inevitable. "GMP means different things at different stages of development. The regulations are more defined and better understood late in development, but during Phase I it's much less clear what GMP means, and I don't understand why that is."

Regulatory uncertainty, says Bizar, has bred an industry of high-priced consultants who advise companies on what FDA wants, but whose counsel is often shot down by the agency. GMPs should be explicit enough that companies can interpret and execute them with a reasonable level of assurance, Bizar says.

Part of the fallout from overly complex GMPs is companies trying to get drug approvals on a shoestring by cutting corners. Because FDA resources are generally over-committed, serious irregularities sometimes go unnoticed while trivial problems cause long, unanticipated delays. FDA, says Bizar, should have more resources and show "greater consistency in enforcement, which definitional clarity would go a long way toward achieving."

GMPs are adequate for manufacture of finished dosage forms ---"patient-ready drugs," as Bizar calls them. Definitive standards for bulk manufacturing, or processing of bulk/formulated material into final dosage forms, are fuzzier. "People are operating in these activities, and FDA is approving drugs coming out of these operations, by feel. This creates unanticipated costs for manufacturers." Potential problem areas cited by Bizar include litigation with toll manufacturers over GMP compliance, uncertainty in moving a drug forward, and not knowing, after all the Ts are crossed and Is are dotted, if the agency will approve a manufacturing process.

"Streamlining GMPs would be nice also, but the most important error-reducing thing from a manufacturer's perspective is to have a body of practice, well-defined from the start. Once they have such a target, reputable companies will do their best to hit it, but where there's uncertainty there are hidden costs, which naturally breed errors."

But major drug makers take pride in manufacturing integrity, Bizar adds. "Companies have manufacturing processes down pat. If they know what their targets are, they'll spend anything to hit the nail on the head every time."

I, Stopper

Chatting with West Pharmaceutical about rubber stoppers is like revisiting Leonard Read's 1956 essay,"I, Pencil" on what it takes to make a number two writing implement. "Stoppers must perform multiple functions," says Don McMillan, vice president of marketing for Americas. West's products, which are used by virtually every major pharmaceutical company worldwide, must provide a multi-use, non-leaching, non-porous barrier for the shelf life of the products they serve. "Consequently, our manufacturing is heavily regulated. All our rubber formulations are contained in master files at FDA." 

With process controls rivaling those of drug makers, West's separate factories for producing "these silly rubber things" and metal seals operate under GMPs. Eliminating human error is a constant concern, because once product reaches filling lines it's too late for excuses. "Our customers have enough to worry about with reducing errors in blending, manufacturing, filling, and packaging," McMillan says. "They prefer that we worry about variability in depyrogenation, sterility, and integrity."

In the late 1980s West instituted a train-the-trainer program at its metals operations facility in Clearwater, Fla. The program, says vice president Fred McCleery, is admired throughout the company but more importantly, by West's customers. West turns out hundreds of combinations of materials, colors, and coatings for the "billions and billions" of stoppers and seals it manufactured in 2003. The company's reputation is built on error-free delivery of components that no one notices, but without which modern injectible drugs could probably not exist.

West's training program evolved over the past 15 years to include job-certified trainers charged with instructing new employees on any number of West's nearly 250 standard operating instructions (SOIs), a number West is trying to cut down, according to McCleery. Each new worker is assigned a trainer who works with them until they're certified, by experience and a written test, to carry out their SOIs. "New employees work full time with their trainer, take breaks with their trainers, and are almost attached to them." Workers who make the cut must take a re-certification exam every year on every task they perform.