Run charts and control charts
Dr Rod Muir Dr Margaret MacLeod Stephen Yeung 2nd October 2002
Content of the workshop • Run charts and control charts • Examples of use in practice • Discussion
Measurement in the NHS • Report cards • League tables • Reporting • Standards
• Accountability
Traditional approach
• League tables
• Confidence intervals
Example of traditional methods
90 day mortality rate following elective hip replacement
1.8
Mortality rate per 100 patients
1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
C
D
H
Sc G ot la nd
Unit
M
N
A
B
E
J
I
K
F
L
Sources of variation
•People
•Methods •Machines •Materials •Environment •Measurements •Measuring •Collecting •Analysing
•Interpreting
Statistical Process Control (SPC) • Simple graphical way to display data and outcomes • Relatively easy to construct • Easy to interpret • Designed to identify any unusual variation in a process
History of SPC methods • Introduced by Walter Shewhart (Bell Telephone Laboratories 1924) • W.E Deming exported the idea to Japan in the 1950s, where it was successfully applied in industry • More recently have found application in healthcare e.g. Mohammed et al (Lancet, 357; 463-467)
Types of variation • Consider health care systems as processes • All processes have inherent random variability – known as “common cause” variability • Unexpected events / unplanned situations can result in “special cause” variation
SPC definitions • A process is said to be „in control‟ if it exhibits only “common cause” variation. It is completely stable and predictable. • A process is said to be „out of control‟ if it exhibits “special cause” variation. This process is unstable.
Example of a system „in control‟
Run charts • Display of data points plotted in chronological order • Ideally 25 data points are required • Centre line (either mean or median) is included for use in identifying types of variation
Example of a simple time ordered run chart
Example of a run chart
Identifying special cause variation from run charts • 8 or more consecutive points on same side of centre line • Consecutive points alternately going up and down 13 times • 6 successive points all going up or down • A point which is wildly different from all the others • Points following a cyclical pattern
Control charts
• A run chart plus control limits indicating the range of plausible variation • Control limits are set at 3 standard deviations above and below the mean • Control limits provide an additional tool for detecting special cause variation
Example of a control chart
Additional rules for identifying special cause variation from control charts • Any point falling outside the control limits
• 2 consecutive values falling outside
the warning limits that are set at 2
standard deviations from the centre
line
Additional example - comparing units over a similar time period
Control chart of 90 day mortality rates following elective hip replacement
50 40 30 20 10 0 0 500 1000 1500 patients 2000 2500 3000
deaths
Summary • SPC approaches offer insights into work processes
• Simple data presentation – feedback
on performance
• Measurement to inform improvement
More information • Key documents - handout • Website – www.show.scot.nhs.uk/indicators