1
You have performance data!
Now, what do you do with it?
2
“If I had to reduce my message for management to just a few words, I’d say it all had to do with reducing variation.” W. Edwards Deming
The Problem! 3
Aggregated data presented in tabular formats or with summary statistics, will not help you measure the impact of process improvement efforts. Aggregated data can only lead to judgment, not to improvement.
Percent of A&E patients Seen by a Physician within 10 min
Did we improve? What will happen next? Should we do something?
Source: R. Lloyd
Percent of A&E patients Seen by a Physician within 10 min
100%
95%
90%
85%
80%
75%
70%
65%
60% Change 55% made here
50%
1/9/2008 2/6/2008 3/5/2008
10/3/2007 1/23/2008 2/20/2008 3/19/2008
10/17/2007 10/31/2007 11/14/2007 11/28/2007 12/12/2007 12/26/2007 Did we improve? What will happen next?
Should we do something? Source: R. Lloyd
Change made between week 7 and week 8 Was the change an improvement? 10 9 8 8 7 6 5
Delay Time (hrs) Time Delay 4 3 3 2 1 0 1 2 Before Change After Change (measure on Week 4) (measure on week 11) Was the change an improvement? Case1 12
10
8
6 Delay Time (hrs) Time Delay 4 Make Change 2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 2 10
Delay Delay Time (hrs) 8 Make 6 Change 4 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 3 10 8
Delay Delay Time (hrs) 6 4 Make Change 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Case 4 12 10 8
Delay Delay Time (hrs) 6 Make 4 Change 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 5 10 8 Make Delay Delay Time (hrs) 6 Change 4 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 6 10 8
Make Delay Delay Time (hrs) 6 Change 4 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 2 10
Delay Delay Time (hrs) 8 Random Variation Make 6 Change 4 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 3 10 Headed down before change. 8
Delay Delay Time (hrs) 6 Where begin? 4 Make Change 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Case 4 12 10 Change did not hold 8
Delay Delay Time (hrs) 6 Make 4 Change 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 5 10 Improvement: before change 8 Make Delay Delay Time (hrs) 6 Change (between week 4 & 5) 4 2 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
12 Case 6 10 Week 4 not typical of process 8
Make Delay Delay Time (hrs) 6 Change 4 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 People unclear of the concept!
Percent Hand Hygiene
“…and then another decrease in the percent compliance with hand hygiene this month. But, I have a really good feeling about next month!
The average of a set of numbers can be 10
created by many different distributions
Average Measure
Time
If you don’t understand the variation that lives in11 your data, you will be tempted to ...
• Deny the data (It doesn’t fit my view of reality!) • See trends where there are no trends • Try to explain natural variation as special events • Blame and give credit to people for things over which they have no control • Distort the process that produced the data • Kill the messenger!
Distorting the Data!
“You'll be happy to see that I’ve finally managed to turn things around!”
The 13
to understanding quality performance, therefore, lies in understanding variation over time not in preparing aggregated data and calculating summary statistics!
“A phenomenon will be said to be controlled when, through the use of past experience, we can predict, at least within limits, how the
Dr. Walter A Shewhart phenomenon may be
W. Shewhart. Economic Control of expected to vary in Quality of Manufactured Product, 1931 the future”
“What is the variation in one system over time?” 15 Walter A. Shewhart - early 1920’s, Bell Laboratories
Dynamic View UCL Static View Static
time
Every process displays variation: LCL • Controlled variation stable, consistent pattern of variation “chance”, constant causes
• Special cause variation Static View “assignable” pattern changes over time
Types of Variation
Common Cause Variation Special Cause Variation • Is inherent in the design of the process • Is due to irregular or unnatural causes that are not inherent in
the design of the process • Is due to regular, natural or ordinary causes • Affect some, but not necessarily all aspects of the process • Affects all the outcomes of a process • Results in an “unstable” process • Results in a “stable” process that that is not predictable is predictable • Also known as non-random or • Also known as random or assignable variation unassignable variation
Common Cause Variation
100
90
80
70
60
50
40
30
20
10
0
3/1/2008 3/8/2008 4/5/2008 5/3/2008 6/7/2008 3/15/2008 3/22/2008 3/29/2008 4/12/2008 4/19/2008 4/26/2008 5/10/2008 5/17/2008 5/24/2008 5/31/2008
• Points equally likely above or below center line • There will be a high data point and a low, but this is expected • No trends or shifts or other patterns
Courtesy of Richard Scoville, PhD, IHI Improvement Advisor A Stable Process is Predictable !
Thus you can confidently: • Counsel patients about what to expect
• Plan for the future
• Inform management
• Use PDSA testing to improve it!
Courtesy of Richard Scoville, PhD, IHI Improvement Advisor Two Types of Special Causes
Unintentional When the system is out of control and unstable
Holding the Gain: Isolated Femur Fractures Intentional 1200 1000
800
When we’re trying 600 Patient to change the 400
200 Minutes ED to per OR system 0 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 Sequential Patients
Courtesy of Richard Scoville, PhD, IHI Improvement Advisor Point …Variation exists!
Common Cause does not mean “Good Variation.” It only means that the process is stable and predictable. For example, if a patient’s systolic blood pressure averaged around 165 and was usually between 160 and 170 mmHg, this might be stable and predictable but completely unacceptable. Similarly Special Cause variation should not be viewed as “Bad Variation.” You could have a special cause that represents a very good result (e.g., a low turnaround time), which you would want to emulate. Special Cause merely means that the process is unstable and unpredictable.
21 Common Cause Variation Special Cause Variation
Holding the Gain: Isolated Femur Fractures 1200
1000
800
600 Patient 400
200 Minutes ED to per OR
0 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 Sequential Patients
Normal Sinus Rhythm (a.k.a. Atrial Flutter Rhythm (a.k.a. Common Cause Variation) Special Cause Variation)
Appropriate Management Response to 22 Common & Special Causes of Variation
Is the process stable? YES NO
Type of variation Only Common Special + Common
Change the Investigate the origin of the Right Choice process special cause
Treat normal variation as a Change the process Wrong Choice special cause (tampering) Wasted Consequences of Increased making the wrong resources! choice variation! (time, effort, morale, money)
Source: Carey, R. and Lloyd, R. Measuring Quality Improvement in Healthcare: A Guide to Statistical Process Control Applications. ASQ Press, Milwaukee, WI, 2001, page 153.
2 Questions …
1. Is the process stable? If so, it is predictable. 2. Is the process capable?
The chart will tell you if the process is stable and predictable. You have to decide if the output of the process is capable of meeting the target or goal you have set!
23
Understanding Variation Statistically
Unplanned Returns to Ed w/in 72 Hours Month M A M J J A S O N D J F M A M J J A S ED/100 41.78 43.89 39.86 40.03 38.01 43.43 39.21 41.90 41.78 43.00 39.66 40.03 48.21 43.89 39.86 36.21 41.78 43.89 31.45 Returns 17 26 13 16 24 27 19 14 33 20 17 22 29 17 36 19 22 24 22 u chart 1.2
1.0
UCL = 0.88
0.8
0.6
Mean = 0.54 Rate per 100 EDPatients100 per Rate
0.4
0.2 LCL = 0.19
0.0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
STATIC VIEW DYNAMIC VIEW Run Chart Descriptive Statistics Mean, Median & Mode Control Chart Minimum/Maximum/Range (plot data over time) Standard Deviation Bar graphs/Pie charts Statistical Process Control (SPC)
24
The SPC Pioneers
Walter W. Edwards Shewhart Joseph Juran Deming (1891 – 1967) (1904 - 2008) (1900 - 1993)
How do we analyze variation for 26 quality improvement?
Run and Control Charts are the best tools to determine: 1. The variation that lives in the process
2. if our improvement strategies have had the desired effect.
Understanding Variation with Run Charts Resource Article:
The run chart: a simple analytical tool for learning from variation in healthcare processes
British Medical Journal Quality and Safety, 2011, Rocco Perla, Lloyd Provost, Sandy Murray
Elements of a Run Chart
6.00 The centerline (CL) on a
5.75 Run Chart is the Median
5.50
5.25
5.00
4.75 Median=4.610 4.50 ~
4.25 X (CL) Measure Poundsof Red Bag Waste 4.00
3.75
3.50
3.25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Time Point Number Four simple run rules are used to determine if special cause variation is present
How many data points do I need? 31
Ideally you should have between 10 – 15 data points before constructing a run chart 10 – 15 patients • If you are just starting to 10 – 15 days measure, plot the dots and make a line graph. 10 – 15 weeks 10 – 15 months • Once you have 8-10 data 10 – 15 quarters…? points make a run chart.
Selecting a Centerline
Mean? Median? Mode? Why Median Rather Than Mean? Mean = arithmetic average of data Median = middle value of ordered data
(n + 1)/2 = Median Position which leads you to the Median Value
• 8,10,11,14,16,18,20 Mean = 13.8
Median Position = Median = 14
• 8,10,11,14,16,18,95 Mean = 24.5
Median Position = Median = 14
• 1,10,11,14,16,18,20 Mean = 12.8
Median Position = Median = 14
But how do you compute the Median when you have an even number of data points?
The Median with an even number of data points?
(n + 1)/2 = Median Position which leads you to the Median Value
• 8,10,11,14,16,18,20,35 Mean = 16.5
Median Position = Median = 15
• 8,10,11,14,16,18,30,95 Mean = 25.3
Median Position = Median = 15
• 1,10,11,14,14,18,19,20 Mean = 13.4
Median Position = Median = 14
Run Chart (n + 1)/2 How do you find the median? (29 + 1)/2 = 30/2 = 15
6.00 When you slide a piece of paper down, you 5.75 reveal the dots in descending order. When you have revealed the 15th data point you 5.50 have found where the median lives.
5.25
5.00 The Median 4.75 Lives Measure Median=4.610 4.50 here at But, the the 15th Median4.25 Value= 4.6 data point
Poundsof Red Bag Waste 4.00
3.75
3.50
3.25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Point Number How do we analyze a Run Chart
“How will I know what the Run Chart is trying to tell me?”
It is actually quite easy: 1. Determine the number of runs. 2. Then apply the 4 basic run chart rules decide if your data reflect random or non-random variation. First, you need to determine the number of Runs
What is a Run? • One or more consecutive data points on the same side of the Median • Do not include data points that fall on the Median
How do we count the number of runs?
• Draw a circle around each run and count the number of circles you have drawn • Count the number of times the sequence of data points (the line on the chart) crosses the Median and add “1” • The two counts should be the same!
Run Chart: Medical Waste 38 Determine the number of runs on this chart
6.00
5.75
5.50
5.25
5.00
4.75 Median=4.610 4.50
4.25
Poundsof Red Bag Waste 4.00
3.75 Points on the Median
3.50 (don’t count these when counting the number of runs) 3.25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Point Number
Run Chart: Medical Waste 39 Determine the number of runs on this chart
6.00
5.75 5.50 14 runs 5.25
5.00
4.75 Median=4.610 4.50
4.25
Poundsof Red Bag Waste 4.00
3.75 Points on the Median (don’t count these when counting 3.50 the number of runs)
3.25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Point Number
Rules to Identify non-random patterns in 40 the data displayed on a Run Chart
• Rule #1: A shift in the process, or too many data points in a run (6 or more consecutive points above or below the median)
• Rule #2: A trend (5 or more consecutive points all increasing or decreasing)
• Rule #3: Too many or too few runs (use a table to determine this one)
• Rule #4: An “astronomical” data point
Non-Random Rules for Run Charts
A Shift: A Trend 6 or more 5 or more
Too many or too few runs An astronomical data point
Source: The Data Guide by L. Provost and S. Murray, Jossey-Bass Publishers, 2011.
42 This is NOT a trend!
Non-Random Rules for Run Charts
A Shift: A Trend 6 or more 5 or more
An astronomical data point
Too many or too few runs
Source: The Data Guide by L. Provost and S. Murray, Jossey-Bass Publishers, 2011.
45 Rule #3: Too few or too many runs
Use this table by first calculating the number of "useful observations" in your data set. This is done by subtracting the number of data points on the median from the total number of data points. Then, find this number in the first column. The lower number of runs is found in the second column. The upper number of runs can be found in the third column. If the number of runs in your data falls below the lower limit or above the upper limit then this is a signal of a special cause.
# of Useful Lower Number Upper Number Source: Swed, F. and Observations of Runs of Runs Eisenhart, C. (1943) 14 4 “Tables for Testing 12 15 5 12 Randomness of 16 5 13 Grouping in a 17 5 Sequence of 13 18 6 14 Alternatives.” Annals of Mathematical 19 6 15 Statistics. Vol. XIV, 20 6 16 pp. 66-87, Tables II 21 7 16 and III. 22 7 17 23 Two data points on 7 17 24 the median = 27 8 18 25 useful 8 18 So, for 27 useful 26 observations 9 observations 19 we 27 10 19 should have 28 10 20 29 Total data points 10 20 between 10 and 19 30 11 21 runs
Source: Swed, F. and Eisenhart, C. (1943) “Tables for Testing Randomness of Grouping in a Sequence of Alternatives.” Annals of Mathematical Statistics. Vol. XIV, pp. 66-87, Tables II and III.
46
Non-Random Rules for Run Charts
A Shift: A Trend 6 or more 5 or more
Too many or too few runs An astronomical data point
Source: The Data Guide by L. Provost and S. Murray, Jossey-Bass Publishers, 2011.
Rule #4: An Astronomical Data Point 48
25 Men and a Test
25 What do you think about this data point?
20 Is it astronomical?
15
Score 10
5
0 1 3 5 7 9 11 13 15 17 19 21 23 25 Individuals
Total data points = 29 Run Chart Interpretation: Data points on the Median = 2 Medical Waste Number of “useful observations” = 27 (should have between 10 &19 runs) The number of runs = 14 6.00 Number of times the data line crosses the 5.75 Median = 13 + 1 = 14
5.50
5.25
5.00
4.75 Median=4.610 4.50 4.25 Are there any Poundsof Red Bag Waste 4.00 non-random 3.75 Points on the Median patterns (don’t count these as 3.50 “useful observations”) present? 3.25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Point Number
51
So, let’s identify some non-random patterns…
Test #1: % of patients with Length of Stay shorter than six days
Antal patienter med vårdtid < 6dygn i % vid primär elektiv knäplastik (operationsdag= dag1)
90
80
70
60
50
40 Antalpatienter % i
30 % of patients of %
20
10
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 MonthMånad % of patients with Length of Stay shorter than six days
Antal patienter med vårdtid < 6dygn i % vid primär elektiv knäplastik (operationsdag= dag1)
90
80 Rules 1 & 3 70
60
50 Median = 52
40
Antal patienter i % i Antalpatienter % of patients of % 30 Rules 1 & 3 18 useful observations Rule 1: not OK 20 Rule 2: OK
10 Rule 3: 2 runs (6-14 runs), not OK Rule 4: OK 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 MonthMånad 3.5 Average Length of Stay 3
2.5
2
1.5 Median ALOS
1 Touch time in minutes in time Touch
0.5 Average Length Stay of Length Average 0
Month 3.5 Average Length of Stay 3
2.5
2 Rule 2
1.5 Median Median = 2.4 22 useful observations ALOS 1 Rule 1: OK Rule 2: not OK (trend) 0.5 Rule 3: 12 runs (7-17 runs), OK Rule 4: OK 0
Month Analyze this Run Chart
% Timely Reperfusion Date 1/99 2 3 4 5 6 7 8 9 10 11 12 1/00 2 3 4 5 65 7 8 9 10 11 12 Data 32 23 32 38 35 35 40 21 38 26 22 27 23 32 36 29 38 42 39 36 50 48 39 44 Run Chart 60
55 Chg 4,5,6 50 Change 1 Chg 7
45
40 Median = 35 Chg 35 Chg 8,9 percent 30 2,3
25
20
15
2 3 4 5 6 7 8 9 2 3 4 5 7 8 9 1/99 10 11 12 1/00 65 10 11 12 Months Analyze this Run Chart % Timely Reperfusion Date 1/99 2 3 4 5 6 7 8 9 10 11 12 1/00 2 3 4 5 65 7 8 9 10 11 12 Data 32 23 32 38 35 35 40 21 38 26 22 27 23 32 36 29 38 42 39 36 50 48 39 44 Run Chart 60
55 Chg 8 Runs 4,5,6 50 What about the Run Chart Rules? Change 1 Chg 7
45
40 Median 35 35 Chg Chg 8,9 percent 30 2,3
25
20
15
2 3 4 5 6 7 8 9 2 3 4 5 7 8 9 1/99 10 11 12 1/00 65 10 11 12 Months Percent Week Compliance • Make a run chart with the data 1 79 Measure is the shown in the table to the left. 2 82 3 86 percent • Decide how you want to lay out 4 84 compliance with the X (horizontal) axis and Y 5 85 proper hand 6 79 (vertical) axis. 7 77 hygiene by 8 86 week. • Plot the data points. 9 82 10 74 • Calculate the median. Hint: use 11 85 N = number of the (n + 1)/2 formula to find the 12 74 properly median position first. 13 78 14 83 completed hand 15 81 washings • Then determine the median value. 16 81 17 74 • Determine the number of runs 18 84 D = total on the chart. 19 78 number of hand 20 75 washing • Apply the run chart rules and 21 74 interpret the results 22 68 observations 23 81 • DO NOT use your calculator or 24 84 25 70 Excel!!! 26 85 27 77 Exercise Percent Compliance with Proper Hand Hygiene
Percent Compliance Median = 81 90 (27+1) = 28/2 = 14
85
80
75
Percent Compliance 70
65 How many runs on this chart
60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Week Exercise Percent Compliance with Proper Hand Hygiene
Percent Compliance 90 Median = 81
85
80
75 15 runs Percent Compliance 70
65
60 Week 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Apply theWeek rules and interpret the chart. NOTE: 27 data points with 3 on the median gives you 24 “useful observations.” For 24 useful observations you expect between 8 and 18 runs. A Final Thought! How will we know that a Baseline change is an improvement?
Annotations 1: AMU 2: AMU 3: excl herbal/OTC 4: after pharmacy intervention on AMU Source: Conwy and 5: 15.91 before pharmacy intervention Denbighshire NHS Trust 6: 12.41 before pharmacy intervention Percent Unreconciled 7: 14.47 pre pharmacy intervention Medications 8: 23.26% pre pharmacy intervention 9: 18.6 before pharmacy 10: 27.05 before pharmacy
Deciding if things have changed
Baseline
Extend the centerline (median) Annotations as a reference point 1: AMU 2: AMU for the new results 3: excl herbal/OTC 4: after pharmacy intervention on AMU 5: 15.91 before pharmacy intervention 6: 12.41 before pharmacy intervention Source: Conwy and 7: 14.47 pre pharmacy intervention Denbighshire NHS Trust 8: 23.26% pre pharmacy intervention Percent Unreconciled 9: 18.6 before pharmacy Medications 10: 27.05 before pharmacy
Deciding if things have changed
We have introduced changes which produced a special cause (a run of 7 data points below the median)
Baseline
Annotations Now, we plot the new data 1: AMU 2: AMU and use the run chart 3: excl herbal/OTC rules to determine if a 4: after pharmacy intervention on AMU 5: 15.91 before pharmacy intervention true change has 6: 12.41 before pharmacy intervention occurred. 7: 14.47 pre pharmacy intervention 8: 23.26% pre pharmacy intervention 9: 18.6 before pharmacy Source: Conwy and 10: 27.05 before pharmacy Denbighshire NHS Trust
Percent Unreconciled Medications
Your next move…
…to gain more knowledge about Shewhart Charts (a.k.a. control charts) Why are Shewhart Charts preferred over Run Charts?
Because Control Charts… 1. Are more sensitive than run charts: – A run chart cannot detect special causes that are due to point-to- point variation (median versus the mean)
– Tests for detecting special causes can be used with control charts
2. Have the added feature of control limits, which allow us to determine if the process is stable (common cause variation) or not stable (special cause variation).
3. Can be used to define process capability.
4. Allow us to more accurately predict process behavior and future performance. Elements of a Shewhart Control Chart
50.0 (Upper Control Limit) 45.0 An indication of a UCL=44.855 A 40.0 special cause
B 35.0
C 30.0 CL=29.250
C 25.0 X (Mean) B 20.0
A 15.0
Number of Complaints Numberof LCL=13.645 Measure 10.0 (Lower Control Limit)
5.0 Jan01 Mar01 May01 July01 Sept01 Nov01 Jan02 Mar02 May02 July02 Sept02 Nov02 Time Month The choice of a Control Chart depends on the Type 68 of Data you have collected Continuous (Variables) Data Time, money, scaled data (temperature, length, volume), workload or productivity (throughput, counts)
Attributes Data Nonconforming Units Nonconformities Defectives Defects (count) data are counted, not measured. Must (classification) be whole numbers. percent that meet a particular criteria (OK vs (e.g., number of errors, falls or not OK) incidents) % of staff who receive QI training) % of new inpatients with a skin assessment completed within 12 hours)
There Are 5 Basic Control Charts
Variables Charts Attributes Charts
• C chart • I chart (number of defects) (individual measurements) • U chart • X & S chart (defect rate) (average & SD chart)
• P chart (proportion or percent of defectives)
Source: R. Lloyd. Quality Health Care: A Guide to Developing and Using Indicators. Jones and Bartlett, 2004, Chap.6 Type of data
Attributes Data Continuous (variables) Data
Numbers of items that passed or Measurement, on some time of scale failed Time, money, height/weight, throughput Data must be whole number when (workload, productivity) originally collected
Classification Each dot on the chart Each dot on the chart Count consists of a single consists of multiple observation of data Either/or, pass/fail, data values 1,2, 3, 4 etc. (errors, (i.e., cost for one yes/no falls, incidents) procedure, waiting X-bar plots the average Numerator can be time for one patient or of all the data values Percentage or greater than the total number of S plots the standard proportion denominator clinic visits for each deviation of the data day) values
Equal area Unequal Subgroup Equal or unequal Equal or unequal of area of size of 1 subgroup size subgroup size (n>1) opportunity opportunity (n=1)
X bar & S c-chart u-chart p-chart I chart chart Six consecutive points increasing (trend up) or decreasing (trend down) Rules for Detecting Special Causes
Two our of three consecutive points near a control A single point outside the control limits limit (outer one-third)
Eight or more consecutive points above or below Fifteen consecutive points close to the centerline the centerline (inner one-third)
71 Using a Control Chart (Wait Time to See the Doctor)
Xm R Char t F e b r u a r y Ap r il 3 0 . 0 2 7 . 5 Intervention 2 5 . 0
2 2 . 5 Where 2 0 . 0 will the 1 7 . 5 process UCL = 1 5 . 3 1 5 . 0 M inut es A go? B 1 2 . 5 C CL = 1 0 . 7 1 0 . 0 C B 7 . 5 A Baseline L CL = 6 . 1 5 . 0 Period 2 . 5 Freeze the Control Limits and Centerline, extend them and 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 2 4 6 8 10 12 14 16 18 20 compare22 24 the26 new28 process30 32 performance to these reference 16 Pat ient s in Febr uar y and 16 Patlines ient to s determine in Apr il if a special cause has been introduced as a result of the intervention. Using a Control Chart (Wait Time to See the Doctor)
Xm R Char t F e b r u a r y Ap r il 3 0 . 0
2 7 . 5 Freeze the Control Limits and compare 2 5 . 0 Intervention the new process performance to the 2 2 . 5 baseline using the UCL, LCL and CL from the baseline period as reference lines 2 0 . 0
1 7 . 5
UCL = 1 5 . 3 1 5 . 0 M inut es A B 1 2 . 5 C A Special Cause is CL = 1 0 . 7 1 0 . 0 detected C B A run of 8 or more 7 . 5 Baseline A data points on one L CL = 6 . 1 side of the centerline 5 . 0 Period reflecting a sift in the 2 . 5 process 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 16 Pat ient s in Febr uar y and 16 Pat ient s in Apr il
Using a Control Chart (Wait Time to See the Doctor)
Xm R Char t F e b r u a r y Ap r il 3 0 . 0
2 7 . 5
2 5 . 0 Intervention Make new control limits for
2 2 . 5 the process to show the
2 0 . 0 improvement
1 7 . 5
UCL = 1 5 . 3 1 5 . 0 M inut es A B 1 2 . 5 C CL = 1 0 . 7 1 0 . 0 C B 7 . 5 A Baseline L CL = 6 . 1 5 . 0 Period 2 . 5 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 16 Pat ient s in Febr uar y and 16 Pat ient s in Apr il This really is child’s play!
1:36:37 The Charts Don’t Tell You…
• The reasons(s) for a Special Cause.
• Whether or not a Common Cause process should be improved (is the performance of the process acceptable?)
• How the process should actually be improved or redesigned.
76