Understanding Variation

In Healthcare

W. Edwards Deming

“If I had to reduce my message to management to

just a few words, I’d say it all has to do with reducing unnecessary variation.”

Numbers We Use Every Day

• Time taken to drive to work• Number of patients in the ED• Number of medication errors• Revenue per month• Average daily census in the

hospital• Number of research studies

approved by the IRB per week.

How to look at data

Registration Times

• Actual time (minutes) it took to register patients in a hospital ED

15 67 4 14 10

12 54 3 7 11

14 83 54 17 20

10 53 What information do these numbers provide?

Does this help?

22.4

29.6

0

5

10

15

20

25

30

Min

ute

s

Monday Tuesday

Does this help?

• Average waiting time is…….

• The mode (frequent number) is……

• The median is…………..

Would any of these numbers help improve the process?

Run Chart of ED Registration Time

0

10

20

30

40

50

60

70

80

90

100

Patient Number

Min

ute

s to

Reg

iste

r

MEAN

Run Chart

• May not give you all the answers, but it well help you ask smarter questions (e.g. What happened when the registration time was much higher?)

• Provides insight into process behavior (Seems like there are two processes, one group registers quickly the other more slowly.)

• No statistical calculations needed

• Use for any type of process and any type of data

Understanding Variation

• Variation exists and permeates all processes. No two things are exactly alike. The problem we face is to be able to measure the extent of the variation and to establish to what degree the variation matters and to whom it matters.

Common and Special Cause Variation

Write the letter “a” eight times on a piece of paper using your

dominant hand. Make all of them the same.___ ___ ___ ___ ___ ___ ___ ___

Are they exactly the same? Why or why not?

Common and Special Cause Variation

Write the letter “a” three times using your

dominant hand, three times with your other hand, then two times with

your dominant hand. Make all of them the same.

___ ___ ___ ___ ___ ___ ___ ___

Are they the same? Why or why not?

• The dominant hand created variation because of the pen and/or paper you used, amount of coffee you’ve had, friction of your hand, etc. etc. This is common cause variation. It is inherent in the process.

• Using the non-dominant hand clearly shows something very different. This is special cause—it does not always happen in the process of writing. Since it so different, you would want to ask “why”.

Common and Special Cause Variation

• Common cause variation – is always present– Is inherent in the process

• Special cause variation – In addition to common cause

variation it is data that signifies the presence of a signal that needs to be investigated.

• Not knowing the difference can create wasted time and effort for all.

Taking ActionAction

NeededYes No

Acti

on

Taken

No

Ye

s

What rules govern your decision?

Actions to TakeCommon Cause

• Recognize the variability inherent in the process

• Responding to this variation is a waste of time

• However, the more variation the lower the reliability

• To change the results, change the process

Special Cause

• Differentiate from common cause variation (rules exist for this)

• Find out what happened

• Eliminate or minimize the impact if negative

• Build positive impact into process

Why Use Run Charts

• “Before and after” data can mask the behavior of a process.

• Run charts allow you to see the behavior of the process.

• Review the following charts that display both “before and after” and run charts.

Evaluating a Change: Displaying Before and After Results

Change Between Weeks 7 and 8

7.385714286

3

0

1

2

3

4

5

6

7

8

WEEKS 1-7 WEEKS 9-14

Weeks

Ave

rag

e

Series1

Change Between Weeks 7 and 8

0123456789

10

1 3 5 7 9 11 13 15

Week

Val

ue Series1

Change Made During Week 8

Week-by-Week ValuesAverages of Before and After

Change During Week Eight

7.285714286

3

0

1

2

3

4

5

6

7

8

WEEKS 1-7 WEEKS 9-14

Weeks

Va

lue

Series1

Change Made During Week Eight

0

1

2

3

4

5

6

7

8

9

10

1 3 5 7 9 11 13 15

Week

Va

lue

Series1

Change Made During Week 8

Week-by-Week ValuesAverages of Before and After

Evaluating a Change: Displaying Before and After Results

Evaluating a Change: Displaying Before and After Results

Change During Week Eight

0

1

2

3

4

5

6

7

8

9

10

1 3 5 7 9 11 13 15

Weeks

Va

lue

s

Series1

Change During Week 8

5.285714286

3

0

1

2

3

4

5

6

7

8

WEEKS 1-7 WEEKS 9-14

Weeks

Va

lue

s

Series1

Change Made During Week 8

Week-by-Week ValuesAverages of Before and After

Evaluating a Change: Displaying Before and After Results

Change During Week 8

3.714285714

3

0

1

2

3

4

5

6

7

8

WEEKS 1-7 WEEKS 9-14

Weeks

Va

lue

s

Series1

Change During Week 8

0

1

2

3

4

5

6

7

8

9

10

1 3 5 7 9 11 13 15

Weeks

Va

lue

s

Series1

Change Made During Week 8

Week-by-Week ValuesAverages of Before and After

Summary

• “Before and after charts” would suggest that the changes made at week 8 resulted in improvements.

• Yet when you look at the run charts, data suggest that some of the improvements began before the action was taken. If this improvement was an expensive proposition, the wrong conclusions will be made.

There are many causes for the variation seen on the run chart. All these must be taken into

consideration. Changing only one may not result in a significant improvement.

Relationship between outcomes and process variables