Pfizer Reinvents Lean in the Lab

May 4, 2011
At the core of missed opportunities in pharma QC/QA is a subtle misunderstanding of the real intent of Lean.

Since the term “Lean” was first coined in 1990(to describe the manufacturing approach being used in Toyota factories in Japan), the space has become littered with lean programs in their various guises, such as Lean Manufacturing, Six Sigma, and Lean Sigma. These programs are usually high on education, but often lack sufficient emphasis on the necessary paradigm change to make a real, sustained impact on a company’s work practices.

Furthermore, while the literature assures us that Lean Manufacturing principles can be applied to any business process, many lean programs outside of Manufacturing still struggle to make these principles relevant. In the Life Sciences, this failure means that a significant opportunity for cost savings in the business—in particular, Quality Control and Quality Assurance processes—is overlooked.

At the core of this missed opportunity is a subtle misunderstanding of the real intent of Lean. By its original definition, Lean is about “the complete and thorough elimination of wasteful practices.” Unfortunately many companies read this intent as simply “elimination of waste.” This intuitively leads them down the trail of setting up projects which address only one of the three wasteful practices—the one called muda in the original Japanese texts.

There are various types of muda described in the Lean literature: Overproduction, Delays (Waiting), Transporting, Moving, Over-Processing (Rework), Inventory, Making Defective Parts. Chasing down muda is intuitively straightforward. Everyone on a project team can contribute to drawing a Value Stream Map. Indeed, this is a good tool to help everyone to see the Value-Add activities in their process and separate them from the Non-Value-Add (NVA) muda. It is also, then, relatively intuitive to go about addressing the wasteful NVA activities, so that wastes such as excessive movement (of product, samples, batch records, etc.), or causes of rework, invalid tests, and corrections can be reduced or eliminated.

Beyond Muda

However, there are two other wasteful practices to consider: mura (unevenness, i.e. workload volatility) and muri (over-burden, i.e. peaks in workload caused by volatility). Addressing these necessitates the implementation of levelled workloads to enable the creation of product flow. And if we look at the Toyota “House of Lean” (Figure 1), where do we find this aspect of Lean activity? Right at the foundation of the house!

So, we cannot expect waste elimination activities alone to build our House of Lean. Yes, they do contribute to a significant pillar, but this pillar is built on sand unless we address the other wasteful practices which are rooted in volatility. This was always the original intent of Lean. But we have found time and time again that this original intent got somehow lost in translation, whether in translation from Japanese to English or from Manufacturing to other business processes. Therefore, BSM Ireland ( uses the phrase Real Lean when we refer to a back-to-the-real-intent-of-Lean approach.

Companies need to be reminded (at best) and mostly simply made aware (and even then convinced through proven, cost-saving implementations) of the principles of Real Lean. But why is this the case? Despite the fact that addressing volatility is not only critical to a successful Lean implementation, but is clearly at the very foundation of the well documented Lean House!

The reason for this is twofold. The first reason we have already alluded to: Value Stream Mapping and pursuing the “Seven Muda Wastes” is very tangible and intuitive—it is easy to mentally connect why moving a product, sample, or piece of paper does not add value, and how moving it less can therefore save a person time. So when something makes immediate sense, we are likely to do it.On the other hand (and this is the second reason), it’s not as immediately obvious how simply levelling your workload and flow can save you anything. “You mean to tell me that I do the same amount of work on the product, sample or paper, just do it at a different time? How can that save me anything?”
Furthermore, many companies have read about “product flow” and about the wonderful aspiration of “one-piece-flow.” Then, because it seems impossible, they quickly convince themselves that “we are different” and “that won’t work here (or in this part of the business) because . . . .” They don’t go the extra mile to try to understand more fully what it is about their business (or their part of the business) that is different, and make the necessary adjustments.  Soon the principles of levelling and flow are forgotten about. A few months later a beautifully drawn Value Stream Map demonstrates some modest improvements. But the beauty of the poster-sized-color-printed sheet clouds the fact that no real cost savings have been realized.

Pfizer: Paradise Regained

BSM Ireland has partnered with Pfizer on a corporate Lean Lab program, as well as a Lean Batch Disposition project at Pfizer’s Biotech site in Grange Castle, Ireland.

The Pfizer Lean story is an excellent demonstration of the benefits of Real Lean. More specifically it shows how, with some creativity and effort, Lean principles can be adapted to significantly improve business processes outside of Manufacturing—namely, QA and QC.

It’s worth noting that the goals of lean in QC/QA are the same as lean in Manufacturing or anywhere else—i.e. improving performance via the introduction of flow and the elimination of waste. But, labs and batch disposition processes are not the same as manufacturing, so we have to think about our Lean solutions differently.

  • There is typically more volatility in both volume and mix.
  • Labs have a complex mix of routine and non-routine testing/review activities—e.g., preparation of standards, folders, investigations, and so on.
  • Single-piece flow is generally uneconomical and unfeasible in the lab—e.g., test equipment is most efficiently run with more than one sample.
  • There is generally less of an operational excellence/productivity focus—Manufacturing has dedicated production engineers, labs do not.
  • The significant wastes are different and harder to reduce or eliminate.
The fundamental concepts of Real Lean detailed below are show how Lean can be practically geared to QA and QC processes. This is illustrated best in some of the work being done in Pfizer, Grange Castle.Concept One: The Effect of Volatile WorkloadsWhen incoming workload exceeds capacity, a lab or batch record review process will fall behind and samples/batch release will be late. Because it is usually considered worse to be late than expensive, QC/QA processes with volatile workloads often carry more resources than required to process the mean workload.However, even with some excess resources, they still fall behind during the highest peaks, and conversely suffer from poor productivity during extended periods of low incoming workload.The difference between the staffing required for the mean workload and the actual staffing level represents either excess cost or additional capacity and (here’s the rub) is recoverable if you can level the workload.In Pfizer, Grange Castle, the arrival of Production Batch Records (PBRs) into QA for review gave rise to peaks and troughs in review activity, resulting in batch disposition times that were lengthy and variable, and a very unpredictable workload from day to day.Concept Two: Spending the Benefits of Lean—Speed or ProductivityReal Lean improves flow and reduces waste. When incoming workloads are volatile, a lab generally has to choose between using these benefits to reduce lead times, or alternatively to become more productive.
Improved flow (velocity) can be directly converted into lead time reduction, or, by using it to allow samples or batches to wait in a ‘levelling queue’, it can be used to reduce volatility. Reduced volatility can then be converted into additional capacity or a cost reduction.Our view is that QC labs or QA batch review processes should aim to be as productive as possible within the lead time of the overall value stream constraint. The overall value stream constraint is that part of the end to end process with the longest lead time. This “bottleneck” process may even be outside of the business process you are trying to lean.In Grange Castle, when we were levelling the batch disposition process, the process in the value stream with the longest lead-time was a sterility test which was done in the QC lab. This test took 21 days.The key point here is that there was no net advantage to complete disposition tasks any faster than this slowest part of the overall value stream. And critically, this gave us 21 days within which we could level any volatility in batch disposition activity (reviews, corrections, associated investigations, prep or folders, summary sheets, etc.) resulting in peaks or troughs of batch records arriving from production, and/or production issues.Figure 4 illustrates how Batch Disposition Pack Preparation activities were levelled across this 21-day period, before the final QP Review was completed.Firstly, PBR review activity was taken care of predominantly in real time—that is, QA reviewers toured the manufacturing areas to review documents at various stages of completion. This not only ensured that the review activity on the PBR for a finished batch was minimal, it also improved the right-first-time of the document when QA personnel were regularly meeting with production personnel.QA then carried out the outstanding off-line PBR review during the first 5 days of the 21-day “sterility clock” starting. This left a 14-day period during which Phase 1 of the disposition pack could start being prepared.The critical day was always Day 21, because it was on this day (with the approval of all lab testing) that Phase 2 (the final phase) of the disposition pack could be completed. Achieving this meant that QPs could start their approval as early as possible. And because the project charter had set an overall 26-day target for batch disposition, it gave QPs the maximum remaining 5 days to sign off the batch.Concept Three: Levelling & FlowIn labs, batch review processes (and elsewhere) there is a link between levelling and flow. You cannot flow samples, for example, through a lab unless the short interval workload is level and you generally can’t level volatile workloads unless you flow the samples.
The simplest levelling strategy is to create the ability to process samples or batches at the levelled demand rate quickly (via flow). We achieve this by developing repeating sequences of testing/review that move the samples/batches through all the required tests/reviews quickly. This reduces the throughput time and incoming samples/batches can then be held in a levelling queue at the start of the process.While in the queue, samples/batches can be prioritized or re-prioritized according to customer requirement using a system of Must Start Dates. But when released into the process as part of a level daily workload, they are processed in strict FIFO order.To make this approach simple to manage and control we develop Heijunka devices which we call Rhythm Wheels (for lower levels of volatility) or Trains (for higher levels of volatility). These repeating test sequences are carefully designed to meet the overall workload (including all non-test/review activity) and consistently achieve the lead times required by the business:
  • They allow recovery of much of the volatility losses.
  • Because they control the workload and the mix they allow the design of well balanced productive roles that make good use of people’s time—i.e., they “solve the problem once” and keep using the solution.
  • They eliminate the need for elaborate short term planning and scheduling processes.
  • They facilitate fixed repeating sequences of operation and “economies of repetition.”
Returning to our Pfizer example, all of the QA activity (PBR Real Time and Queued Review, Disposition Pack Prep, etc.) was levelled via levelling queues consisting of folders which were arranged in Must Start Date order (Figure 4 again). These queues clearly indicated what priority batches to work on, while a weekly Rhythm Wheel, based on the calculated, levelled weekly work content told specialists when to do various tasks (see Figure 6).
Concept Four: Standard WorkSome people are naturally good ‘time and task’ managers and will organize and sequence their work in a logical and productive manner. However many people are not!  BSM use a Standard Work approach to develop repeatable analyst roles (possible because we are controlling the workload and the mix). This improves the operation of the rhythm wheel or train and reduces errors and failures. Also, because Standard Work combines tasks and uses people’s time well, it also delivers a significant productivity gain in itself.Each of the roles in Figure 6 has a Role Card which indicates clear short interval targets during the day. So, for example, a QA specialist doing a Phase 1 Disposition Pack Preparation will have a guideline on their Role Card which indicates that by first break they should aim to have (for example) the folder contents list prepared and the autoclave memo, moisture memo and EM report completed, say. By lunch break, four more documents should be completed, and the balance by end of shift. In this way issues are highlighted early in the day, and there aren’t any surprises at the end of the day or at the following day’s meeting.BenefitsShort term volatility (in overall workload and in the mix of samples) is by far the biggest lean opportunity in most labs and QA processes. Successfully levelling a volatile workload will deliver significantly greater benefits than waste elimination alone.
Levelling a volatile workload will typically generate improvements in productivity ranging from 25% to 82% and higher. This can be converted into unit cost reduction and/or used to reduce lead-times (by up to 90% in some previous BSM projects). Annual returns are typically 3 to 5 times the cost of the projects.As can be seen in Figure 7, Pfizer, Grange Castle made significant improvements in its Batch Disposition Leadtime across 2010. This was also achieved with a more productive QA group, doing much more value-add work in terms of daily face to face dealings with the operations people who were producing their documents.
About the Author

Lorcan Mannion | Operations Practice Manager