Simplicity of Statistical Sampling: Part 1

It was a VHS Tape (yes, it was that long ago) that I first learned about Statistical Process Control (SPC). Dr. Donald Wheeler was explaining statistical sampling, run charts, and process control. For his example, he used the manufacture of cigarette lighter cylinders for cars (I told it was a long time ago). Specifically, how the indentation around the cylinder could vary in distance from the tip of the cylinder.

The video was brilliant. Simple, easy to understand, and it demonstrated the power of SPC. I was enthralled at the fact you could use control charts to head off trouble; solving problems before they became problems.

I was determined to use this powerful new tool. But there wasn’t a good instance to use it. At the time I was working at the US Navy implementing their Total Quality Leadership (TQL) program (See past article  A Tale of Two Units ) and there just weren’t any processes that were a good fit.

Over the years I always looked for an opportunity to apply SPC. However, I worked primarily in the service sector, and SPC seemed to be best suited to mass production manufacturing. When I found something that looked like I could use SPC, I was hindered by another issue: the complexity.

I no longer had access to the Dr. Wheeling Video, and as I looked at how to set up the SPC, it took me down a statistical rabbit hole. The texts I found were set up for somebody well-versed in statistics. I had to dig out my old stats books to refresh me on the different types of data (nominal, ordinal, discrete, and continuous). Different types of data use different formulas and different control lines; it never seemed that difficult in the Wheeler video.

Thinking I could leverage Excel© I looked for templates and plug-ins. Again, they were overly complex, not user-friendly, and not consistent. Different plugins gave different charts with different results. It was all very frustrating. And all too confusing to me, let alone having me try to explain the workings of SPC to leaders when I briefed results.

So my desire to use SPC languished for a bit longer. And then I stumbled across something that rivaled Dr. Wheeler’s video. Lean guru Mark Graban wrote a book Measures of Success: React Less, Lead Better, Improve More. I knew it had to be good…Dr. Wheeler wrote the Foreword.

I bought the book and was amazed all over again. Mark broke things down: making them simple and understandable. Instead of the variations of control charts based on data type, he showed how you could use one type of chart (Which he calls Process Behavior Charts)

Also, about that time, I was working at a hospital group on a complex project. We were having issues of infection in patients with central lines (Plastic tubes inserted in veins going directly to the heart). Such infections, called CLABSIs (Central Line Acquired Blood Stream Infections) are quite serious as they can quickly spread throughout the body and have a high fatality rate. But while serious, such infections were relatively rare; before having a rash of infections (6 within 5 months) the hospital had not had a CLABSI in over 700 days. There was not enough data on the infections themself to conduct any type of statistical analysis.

Also, tracking infections was a lagging indicator. By tracking infection, it was too late. We needed leading, or predictive, indicators. And we found what we needed.

The Institute of Healthcare Improvement (IHI) has a white paper on using a “bundled” metrics approach to help predict clinical outcomes. From that we created 7 variables that could be used to indicate the possibly of a CLABSI developing. We added an eighth after we began using a disinfecting device called a CUROS Cap. The Variables included:

1. Is the dressing clean, dry, and intact?                   
2. Does the dressing have a date?               
3. Does the dressing have an initial?                          
4. Is there an antimicrobial patch that is correctly applied?              
5. Is the dressing changed within the standard timeframe?              
6. Does EPIC documentation match what's seen bedside?                
7. Is all IV tubing labeled with a day to be changed?                                         
8. Are CUROS caps used?  

Finally, an opportunity to use SPC. Everything was in place to begin.

Next week, in part 2, I’ll tell how we implemented the SPC and the results.

#quality #lean #leansixsigma #operationalexcellence #processimprovement #totalqualitymanagement #storytelling innovation  #lean #leantraining  #leanthinking