Six Tools for Common Cause Variability Reduction for

Six Tools for Common Cause Variability Reduction for Pharmaceutical QA/QC and Manufacturing Lynn Torbeck and Assoc.

Overview of Presentation Linking science and measurement to statistical thinking. n Ten concepts of statistical thinking. n Six tools for variability reduction. n Suggestions for implementation. n Copyright c 2004 Torbeck & Assoc. 2

Measurement is the Essence of Science n “When you can, count. ” ¨ n Francis Galton (1822 -1911) First cousin to Charles Darwin. “When you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind …” Lord Kelvin, 1883. 3

Where there is science … n There is: ¨ Measurement ¨ Data ¨ Variation ¨ Statistics ¨ Statistical Thinking 4

Statistical Thinking, The Book: Improving Performance Through Statistical Thinking ¨ ASQ Statistics Division ¨ Britz, Emerling, Hare, Hoerl, Janis and Shade ¨ ASQ Quality Press, Milwaukee, WI ¨ 2000 5

Statistical Thinking Principles 1. 2. 3. All work occurs in a system of interconnected processes. Variation exists in all processes. Understanding and reducing variation is key to success. n n R. D. Snee, “Statistical Thinking and Its Contribution to Total Quality. ” ASQ Statistics Division Newsletter, Winter 1991. 6

Expanded Concepts: #1 Take a more global view of problems. n Work occurs in systems of processes and sub-processes of interconnected and interrelated steps. n Processes can be modeled with the S. I. P. O. C. view point. n 7

8

Expanded Concepts: #2 Processes can be mapped, flowcharted, studied systematically, understood and improved. Create a 3 -ring binder. n Optimizing each step may cause the whole process to be sub-optimum. n The Goal, 1984, 1986. Eliyahu M. Goldratt n North River Press n 9

Flowcharts 10

Process Mapping n Process mapping expands flowcharting: ¨ Critical factors ¨ Critical responses ¨ Data collection points identified ¨ Statistical analysis and summary of data. n HAACP, FMEA 11

Process Mapping 12

Expanded Concept: #3 Work is performed by teams of people with differing backgrounds, education, expertise, skills, needs and expectations. n Management vs. administration: n The Team Handbook, 1988 n Peter Scholtes, et all. n Joiner and Assoc, Madison, WI n 13

Expanded Concepts: #4 Process outputs vary as a result of both special or systematic causes and common causes or random causes. n Special causes are the result of one or two factors changing. n Common causes are the result of many factors changing more or less at random. n 14

Indication of Control or Lack of Control 15

Common Cause Variation 16

Expanded Concepts: #5 Cause and effect relationships are the foundations of science. n These relationships can be found, studied, quantitated and understood. n Design of Experiments has the goal of describing cause and effects in exact mathematical terms. n 17

18

Expanded Concepts: #6 n Variability is the ENEMY of: ¨ Quality / CGMPs ¨ Efficiency ¨ Productivity ¨ Cycle time ¨ Validation ¨ Profits 19

Expanded Concepts: #7 n Variability can be measured: ¨ Range and Inter-quartile range or IQR ¨ Standard Deviation ¨ % RSD = 100*SD / Xbar Sources of variability can be found and understood. n Variability can be separated out, ANOVA. n 20

Visualizing Variability Graphically n Graphical Tools to Use: ¨ Time Plot or Run Chart ¨ Control Charts ¨ Histograms ¨ Scatter Plots ¨ Pareto Charts 21

Time Plot or Run Chart 22

Control Charts 23

Histograms 24

Scatter Plots 25

Pareto Plots 26

Expanded Concepts: # 8 Statistics is the science of variation. n “Good statistics is not equated with mathematical rigor or purity, but is more closely associated with careful thinking. ” n Professor Robert Hogg: n The American Statistician n November 1991 n 27

Expanded Concepts: #9 Good News! Variability can be reduced. n Variability is not “inherent” or “given. ” n Variability is subject to cause and effect relationships. n There are tools and techniques for reducing variation. n 28

6 Tools for Variability Reduction 1. 2. 3. 4. 5. 6. Work to target. Flexible Consistency. Operational Definitions. Control what can be controlled. Average out the variation. New technology, P. A. T. 29

1. Work to Target The goal is to hit the target every time if at all possible. n T(L, H), 20(15, 30) n The specification limits are not our playground. If for some reason we can not hit the target, the limits allow us to not reject. Individual consistency. n 30

1. Work to Target A company close to one of the hotbeds of biotechnology wanted to do a designed experiment on one of their process/products to improve it. They had a test method that was to be used to test the product. During the planning sessions, it came out that the method was a bioassay and had a %RSD of about 25%. The consultant noted that was large given the objective of the experiment and that to compensate more runs and samples would have to be done. He suggested that rather than spending the money to do many more runs and samples for the one experiment that some of that money and effort be put into reducing the variation of the method. During long pause after that statement, one attendee left the room. The lab manager pointed out that the method had been validated and found to be acceptable. The consultant pressed the issue suggesting that flowcharting the method may highlight points where variability could be reduced. The attendee returned to announce that the developer of the method had been called at the local university. When told the method had a %RSD of 25%, the developer declared “take everybody out for pizza, thatsagood method. ” As part of the visit, the consultant was taken on a tour of the facility including the laboratory. As the tour entered the lab, it was noted that the method was being done at that time. The consultant engaged the analyst in conversation. “Where are you in the method? ’ The analyst replied, “Up to the point where we hold it overnight. ” “What does that mean the consultant replied. “Well, hold overnight, ” was the answer. Doggedly the consultant drove forward. “Is that 8 hours, 10 hours or 12 hours? ” “Yes” was the response. “Do you think the result is effected by how long you hold this step? ” said the consultant fearing the answer. It came, “yes” was the reply after a thoughtful silence. 31

2. Flexible Consistency Sounds contradictory. n In many situations it may not matter which way we do an action even if there is a difference. n The goal is for everyone to do exactly the same thing the same way every time if at all possible. Group consistency. n 32

2. Flexible Consistency A one product company had a test method that determined the content of the batch. The value of the batch was in direct relationship to the result of the test method. Low result, less money. Higher result, more money. Of course batches varied, but they varied more than would be expected based on an understanding of the process. On average, the results were about where they were expected to be. But it was uncomfortable to charge more one time than the other when there was reason to believe they were almost all the same. A consultant was brought in to look at the variability of the data and the historical records. The consultant discussed the issue with the lab manager. The manager indicated that the analysts were following the test procedure and each had been trained. As part of the visit the consultant asked to see the method being done. During a tour in the morning the consultant watched an analyst perform the method. At one point in the method, the analyst was instructed to “cool off the metal box. ” Part of the equipment was home-built including the “box. ” The analyst obtained a chunk of dry ice and a towel and placed it on the box and continued with the procedure. In the afternoon, the consultant and manager were passing through the lab and the consultant stopped to watch a different analyst perform the method. As it happened, the analyst was up to the point in the method to cool off the box. The analyst took an can of CO 2 and gave the box a good long blast and then finished the method. 33

3. Operational Definitions “Sample the tank. ” n We need a full and exact description of n ¨ Who ¨ What ¨ Where ¨ When and ¨ How. 34

3. Operational Definitions A Midwest pharmaceutical company made a bulk product in England shipped it to Puerto Rico for packaging and sale in the United States. The bulk product was tested by QC before it was shipped. Upon receipt, the PR QC group sampled and tested it as well. As might be expected, sometimes a batch that passed in England failed in PR. Unfortunately the practice was to ship the material back to England even though it had passed the English QC test. Of course, the English had no recourse but to test it again and ship it again. Again upon receipt, PR would faithfully test it and it would inevitably pass. If this only happened once in a great while, it wouldn’t have been a problem, but it was happening all too frequently. Clearly the company didn’t need to spend its money shipping its own product back and forth across the Atlantic. A review of the historical data indicated that there was a consistent 5% bias between the two laboratories. Phone interviews with both labs, managers and analysts, yielded little information other than both groups adamantly swore that they followed the test method faithfully without deviations. An experienced chemist was dispatched to England with the instructions, “I want you to watch the method being performed until you have memorized everything that is not written in the test procedure. Then I want you to go to PR and do the same. ” Upon his return, he reported, “I found it. In England they stir before they scoop to weigh a sample and in PR they just scoop without stirring. The material is segregating in PR resulting in the average 5% lower values. ” 35

4. Control the Controllable Build an environment of consistency. n Within economic reason, control everything that can be controlled. Even if not considered critical. n Continuous improvement in control and consistency is a mindset and a company culture. n 36

5. Average Out the Variability Define or redefine the reportable value to be the average of multiple determinations. n Variability decreases by the square root of the sample size. n Averages tend to be Normally distributed. n Spend the money in the laboratory, not in rejected lots of product. n 37

6. New Technology As a last resort, purchase new equipment and new techniques. n The goal behind Process Analytical Technologies, P. A. T. n NIR control of mixing granulations is an example. n 38

Final Concept: #10 n Organizational and financial success is enhanced by using teams for continuous improvement through variation reduction and bringing processes into a state of stability. The tools are simple, the implementation difficult. 39

Implementing Statistical Thinking (Source: Edward Tufte) Don’t be a prisoner of the graphs, charts and tables presented to you. It is only a selection of what is available. n Step back mentally and ask: “What would I really like to see to answer this question or solve this problem. ” Then ask for it. n 40

Implementing Statistical Thinking (Source: Edward Tufte) Always ask to see the data. n Don’t accept folklore and war stories. n Always ask for the data to be plotted. n Put cause and effect on the same page. n Maximize data density per page. n 41

Summary Special cause variation is solved using root-cause analysis. n Common cause variation can be reduced using the six tools presented here. n The tools are simple but effective. n Implementation requires persistence. n 42

Take Back Nugget n We all must ask every day: “Where is the variability coming from and what have you and I personally done today to eliminate it or reduce it? ” 43

References Improving Performance Through Statistical Thinking, ASQ Statistics Division, Britz, Emerling, Hare, Hoerl, Janis and Shade, ASQ Quality Press, Milwaukee, WI, 2000 Mapping Work Processes, Dianne Galloway, ASQ Quality Press, Milwaukee, WI, 1994. The Team Handbook, Peter Scholtes, Joiner and Assoc. , Madison, WI, 1988 The Goal, Eli Goldratt and Jeff Cox, North River Press, Great Barrington, MA, 1984 44