Colorado 5M WebEx Variation, Run Charts, and Control Charts Beth A' Katzenberg, EdM, MBA, CPHQ Direc

1
Colorado 5M WebExVariation, Run Charts, and
Control ChartsBeth A. Katzenberg, EdM, MBA,
CPHQDirector, Corporate Quality
ComplianceColorado Foundation for Medical Care
2
Types of variation

• Common cause
• Always present
• Inherent in process
• Can predict performance with a range of variation
• Cannot tell what specifically causes variation

• Special cause
• Abnormal, unexpected
• Due to causes not inherent in process
• Can be identified (e.g., change in shift,
  weather, process)

3
You must understand the type of variation that is
occurring as this will determine how you address
the problem.
4
Variation
5
Pitfalls

• If only common cause variation and treat as
  special cause (tampering), leads to greater
  variation, mistakes, defects
• If common cause and special cause, and change the
  process, leads to wasted resources because the
  change wont work

6
Tools to identify variation
7
Run charts
8
Run chart
Graph of data over time Track performance Display
identify variation
9
Run chart analysis Common cause variation only
Common cause variation around the median Only
common cause variation present. Output may or
may not meet customer/ patient requirements
10
Run chart analysis Runs

• Run one or more consecutive data points on the
  same side of the median
• Excludes data points on the median

11 runs
11
Expected number of runs
12
High probability of special
cause variationToo few runsToo many runs
0.05)
(
13
Run chart analysis Run length
Special causerun length lt20 data points (not
on median) A run of 7 data points on one side
of the median (either above or below) 20 data
points (not on median) A run of 8 data points
on one side of the median
14
Run chart analysis Trends
Special causetrends Consecutive points all
going up or all going down. May cross the
median. Ignore 2 consecutive points that are
the same.
(Pyzdek, 2003)
15
Run chart analysis Freaks
Freaks The presence of more than one or two
dramatic spikes suggests the process is out of
control. Run charts not as sensitive in
identifying, thus may fail to detect.
16
Run chart analysis Cycling
Cycling A zigzag or saw-tooth pattern with 14
points in a row alternating up or down.
17
Run charts tips

• How many data points?
• 15-20 minimum is preferable
• Median 50/50 split point
• Precisely half of the data set will be above the
  median and half below it

18
Control charts
19
Control chart
Run chart with control limits Determines type of
variation Is process stable? Predictable?
20
Dividing control chart into zones
UCL
Each zone is 1 sigma wide
X
LCL
21
Identifying special causes

• Apply independently to each side of the center
  line
• 1 point outside the 3 sigma limit
• 2 out of 3 consecutive points in zone A or beyond
• 4 out of 5 consecutive points in zone B or beyond
• lt20 total data points 7 consecutive points in
  zone C or beyond on one side of center line
• 20 total data points 8 consecutive points in
  zone C or beyond on one side of center line
• (continued)

22
Identifying special causes, cont.

• Apply this test to entire chart
• lt21 total data points 6 or more points in a row
  steadily increasing or decreasing
• 21 total data points 7 or more points in a row
  steadily increasing or decreasing
• 14 consecutive points alternating up and down in
  saw-tooth pattern
• 15 consecutive points in zone C (above and below
  center line)

23
Deciding which control chart to use
24
(No Transcript)
25
Control chart example 1
Common cause variation only
26
Control chart example 2
snowstorm
new hire
27
Control chart example 3
Common cause variation only can predict will
stay within control limits, if no changes
28
Control chart example 4
Out of control, unpredictable
29

• Just because a process is under control (common
  cause variation only), it does not mean that the
  process is meeting expectations.
• It just means that the process is predictable and
  you are getting consistent performance.

30
Control charts tips

• Control limits are not specifications limits
  (specification limits related to customer
  requirements)
• After removing special causes and recalculating
  chart, continue to plot new data on this chart,
  without recalculating control limits.
• Recalculate control limits only when a permanent,
  desired change has occurred in the process and
  only using data after the change occurred

31
Share the data

• Team meetings
• Post in break-rooms
• Newsletters
• Intranet

32
Examples of Software

• QI Macros www.qimacros.com
• StatSoft www.statsoft.com
• Minitab www.minitab.com

33
References

• Carey, R.G. Lloyd, R.C. Measuring Quality
  Improvement in Healthcare A Guide to
  Statistical Process Control Applications, Quality
  Resources, 1995.
• Pyzdek, R. The Six Sigma Handbook A Complete
  Guide for Green Belts, Black Belts, and Managers
  at All Levels, 2003.
• The Six Sigma Memory Jogger II, GOAL/QPC, 2002.

34

• Beth Katzenberg, EdM, MBA, CPHQ
• Director, Corporate quality compliance
• Colorado Foundation for Medical Care
• bkatzenberg_at_cfmc.org