Statistical

Process Control
Monitoring Quality
in Healthcare.

Tutorial
Guide.

Information Services Division:

Quality Indicators Team
 Contents
Introduction 1
History of SPC 1
Purpose of Guide 1
1 Understanding Variation 2
1.1 Types of Variation 2
1.2 Tools for Identifying Process Variation 2
2 Types of Statistical Process Control Chart 3
2.1 Run Charts 4
2.2 Run Chart Rules 5
2.3 Shewhart Control Charts 7
2.4 Shewhart Control Chart Rules 11
2.5 Shewhart Control Chart Examples 12
2.6 The XMR chart 17
Summary 18
Further Information 18
Acknowledgments 18
Contact Information 18
Appendix A: Formula for calculating Control Charts limits 19
XmR Chart 19
Xbar and S Chart 19
P Chart 19
C Chart 20
U Chart 20
T Chart 20
G Chart 20

Statistical Process Control

II
Monitoring Quality in Healthcare
Introduction
NHSScotland routinely collects a vast array of data from healthcare processes. The analysis of
these data can provide invaluable insight into the behaviour of these healthcare processes.

Statistical Process Control (SPC) techniques, when applied to measurement data, can be
used to highlight areas that would benefit from further investigation. These techniques enable
the user to identify variation within their process. Understanding this variation is the first step
towards quality improvement.

There are many different SPC techniques that can be applied to data. The simplest SPC
techniques to implement are the run and control charts. The purpose of these techniques is to
identify when the process is displaying unusual behaviour.

The purpose of this guide is to provide an introduction to the application of run charts and
control charts for identifying unusual behaviour in healthcare processes. SPC techniques are
a tool for highlighting this unusual behaviour. However, these techniques do not necessarily
indicate that the process is either right or wrong – they merely indicate areas of the process
that could merit further investigation.

History of SPC
Statistical Process Control (SPC) Charts were first introduced in 1928. Commissioned by Bell
Laboratories to improve the quality of telephones manufactured, Walter Shewhart developed a
simple graphical method – the first of a growing range of SPC Charts.

Understanding the causes of variation within an industrial process proved indispensable as

actions could be taken to improve process and output. In the 1950’s, with the effective use of
SPC, Deming converted post war Japan into the world leader of manufacturing excellence.

This approach is increasingly being applied in healthcare by thinking of healthcare systems as

processes. As well as providing a basis for quality improvement within healthcare, SPC Charts
also offer alternative methods of displaying data.

Purpose of Guide
The purpose of this tutorial guide is to give a brief overview of the different types of SPC chart,
some guidance on the rules that should be applied to the charts, and how to interpret their
output.

An interactive tool has been produced to automate the production of SPC charts for the user.
This is an Excel-based tool which is hosted on the Quality Indicators pages of the ISD website.

Statistical Process Control

3
Monitoring Quality in Healthcare
 1 Understanding Variation

1.1 Types of Variation

Variation exists in all processes around us. For example:

• Every person is different

• No two snowflakes are identical
• Each fingerprint is unique
The two types of variation that we are interested in are ‘common cause’ and ‘special cause’
variation.

Common Cause
All processes have random variation - known as ‘common cause variation’. A process
is said to be ‘in control’ if it exhibits only common cause variation i.e. the process is
completely stable and predictable, with any variation due to regular, natural or ordinary
causes.

Special Cause
Unexpected events/unplanned situations can result in ‘special cause variation’. A process
is said to be ‘out of control’ if it exhibits special cause variation i.e. the process is unstable
due to irregular or unnatural causes that are not inherent in the design of the process.

1.2 Tools for Identifying Process Variation

Now that we know variation exists in all processes we can proceed to identify which type of
variation is present. One method of identifying the type of variation present is by using SPC
charts. Originally developed for use in manufacturing, many applications are now involving
healthcare processes.

Areas of special cause variation may be of interest to the user for quality control or
improvement purposes and can be signals for underlying issues. A signal on a SPC chart does
not necessarily mean that something is wrong, nor do they tell the user what may be causing
the variation. They should instead be used as a trigger for further investigation.

Statistical Process Control

2
Monitoring Quality in Healthcare
2 Types of Statistical Process Control Chart
SPC charts can be applied to both dynamic process and static processes. The type of SPC
chart that should be used depends on the type of data, whether the process is static or
dynamic (as described below), and the underlying probability distribution the data is expected
to follow.

Dynamic Processes
A process that is observed across time is known as a dynamic process. An SPC chart for
a dynamic process is often referred to as a ‘time-series’ or a ‘longitudinal’ SPC chart.

Static Processes
A process that is observed at a particular point in time is known as a static process. An
SPC chart for a static process is often referred to as a ‘cross-sectional’ SPC chart. A cross-
sectional SPC chart is a good way to compare different institutions. For example, hospitals
or health boards can be compared as an alternative to league tables as we will see later.

Choosing the correct chart type is an important step, using the incorrect chart type can lead to
misinterpretation and cause a Type I or Type II statistical error.

Type I Error
Incorrectly flagging a signal when there is no signal present; this is known as a false positive.

Type II Error
Failing to flag a signal when a signal is actually present; this is known as a false negative.

This tutorial guide will focus on dynamic processes only.

Two of the most popular SPC tools in use today are the run chart and Shewhart control
chart. They are easy to construct, as no specialist software is required. They are easy to
interpret, as there are only a few basic rules to apply in order to identify the variation type
without the need to worry too much about the underlying statistical theory.

Run Chart
A time ordered sequence of data, with a centreline drawn horizontally through the chart,
usually based on the median. A run chart enables the monitoring of the process level and
identification of the type of variation in the process over time.

Control Chart
A time ordered sequence of data, with a centreline calculated by the mean. Control charts
bring the addition of control limits (and warning limits – optional). A control chart enables
the monitoring of the process level and identification of the type of variation in the process
over time with additional rules associated with the control (and warning) limits.

Statistical Process Control

3
Monitoring Quality in Healthcare
 2.1 Run Charts
The most basic type of chart that a user can plot is a run chart. A run chart is a simple graph
where a process is plotted against time, with a measurement of central tendency, usually based
on the median. Run charts have no upper or lower control limits, which can make them easier
to produce than typical Shewhart control charts. Their simplicity also means they lack the same
level of statistical sensitivity when compared to a typical Shewhart control chart.

Steps to create a Run Chart

1. Ideally there should be a minimum of 10 to 15 data points.
2. Draw a horizontal line (the x-axis), and label it with the unit of time e.g. year, quarter
etc.
3. Draw a vertical line (the y-axis), and scale it to cover the current data, including
sufficient room to accommodate future data points. Label it with the outcome.
4. Plot the data on the graph in time order and join adjacent points with a solid line.
5. Calculate the median of the data (the centreline) and draw this on the graph.

Example 1
The chart below shows an example run chart: the absolute value of the daily price
movement (%) of the FTSE100 Index from 4 November 2016 to 14 December 2016.

Daily Price Movement for FTSE100 Index

0.02
Price Movement (%, absolute value)

0.018
0.016
0.014
0.012
0.01
0.008
0.006
0.004
0.002
0
Dec 12, 2016
Dec 13, 2016
Dec 14, 2016
Dec 1, 2016
Dec 2, 2016
Dec 5, 2016
Dec 6, 2016
Dec 7, 2016
Dec 8, 2016
Dec 9, 2016
Nov 4, 2016
Nov 7, 2016
Nov 8, 2016
Nov 9, 2016
Nov 10, 2016
Nov 11, 2016
Nov 14, 2016
Nov 15, 2016
Nov 16, 2016
Nov 17, 2016
Nov 18, 2016
Nov 21, 2016
Nov 22, 2016
Nov 23, 2016
Nov 24, 2016
Nov 25, 2016
Nov 28, 2016
Nov 29, 2016
Nov 30, 2016

FTSE100 Price Movements Centreline Shift

Statistical Process Control

4
Monitoring Quality in Healthcare
2.2 Run Chart Rules
The following definitions are useful before proceeding onto the rules for detecting special
variation within run charts and later, control charts.

Useful Observations
Those observations that do not fall directly on the centreline are known as “useful
observations”. The number of useful observations in a sample is equal to the total number
of observations minus the number of observations falling on the centreline.

Run
A sequence of one or more consecutive useful observations on the same side of the
centreline. Any observations falling directly on the centreline can be ignored.

Table 1 below lists the rules which can be applied to the run chart for determining the type of
variation in the process.

Table 1: Run Chart Rules.

Rule Description
Shift A run of 6 or more consecutive data points above or below the baseline.
Points on the centreline neither break nor contribute to a shift.
Trend A run of 5 or more consecutive data points which are increasing or
decreasing. An observation that is the same as the preceding value does not
count towards a trend.
Too many/too few Too many or too few runs: If there are too many or too few runs (i.e. the
runs median is crossed too many or too few times) that’s a sign of non- random
variation. Table 2 provides a guide based on the number of useful
observations in your sample. An easy way to count the number of runs is to
count the number of times the line connecting all the data points crosses the
median and add one.
Astronomical Data A data point which is distinctly different from the rest of the process. As there
Point are no upper or lower control limits, a judgement should be made by the user
on what would constitute an astronomical data point.

Statistical Process Control

5
Monitoring Quality in Healthcare
 Table 2: Number of runs which constitute too many/ too
few based on useful observations.
Number of Useful
Too Few Runs Too Many Runs
Observations
15 4 12
16 5 12
17 5 13
18 6 13
19 6 14
20 6 15
21 7 15
22 7 16
23 8 16
24 8 17
25 9 17
26 9 18
27 9 19
28 10 19
29 10 20
30 11 20
31 11 21
32 11 22
33 11 22
34 12 23
35 13 23
36 13 24
37 13 25
38 14 25
39 14 26
40 15 26

Source: Perla, RJ., Provost, LP., Murray, SK. 2011. The run chart: a simple analytical tool for learning from
variation in healthcare processes. [Online]
Available at: http://www.wales.nhs.uk/sites3/Documents/841/Run%20charts%20%28August%202011%29.
pdf [Accessed 3 April 2017].

Statistical Process Control

6
Monitoring Quality in Healthcare
The rules listed above are purely guidelines. Some textbooks may quote different sizes of
trends, and shifts. The above are standard to the work that is carried out within the Quality
Indicators Team and ISD wide but their primary intention was for applications in industry.
Although SPC lends itself well to healthcare processes, healthcare processes deal with
lives. With this in mind, common sense is often the best guideline - SPC charts will illustrate
the variation within your process but if fewer observations in a trend or shift etc is unusual
behaviour in your process then this is just as good an indication of special cause variation and
is therefore worth investigating.

Run charts are a good chart to choose if the user is looking for something quick, simple and
basic, but their lack of statistical power makes them less able to reliably identify areas of
special cause variation (i.e. there will be more type II errors with run charts) than Shewhart
charts, so if the user is looking for something that is more robust, a run chart is unlikely to be
sufficient. However, as a starting point it is always better to plot a run chart first before any
other type of SPC chart.

2.3 Shewhart Control Charts

Shewhart Control Charts enable the monitoring of the process level and identification of the
type of variation in the process over time with additional rules associated with the control (and
warning) limits. The inclusion of upper and lower control limits allows a degree of measurement
of distance from the centreline. Unlike run chart rules, these limits also remove the element of
subjectivity in judging whether a point is an astronomical data point or not as points are either
outliers or they are not.

There are many different types of chart available, and the correct chart to use depends on the
type of data being collected.

Types of Data
Continuous data
Occupy any value within a range. Typically a measurement of some kind, they are
unrestrained by “categories”, and can be infinite in range. Examples of continuous data
would be measurements of height, weight or length; monetary values and anything which
could be considered workload or throughput.

Discrete data
Observations where each possible observation is distinct from one another. It can be
categorical e.g. number of red balloons handed out at a fair; or numeric, typically count
data e.g. the number of falls in a care home each month; or binomial data e.g. how many
red balloons are handed out each hour. Typically, when putting a binomial process into an
SPC chart, the data should be presented as a proportion of a success, or a proportion of
failures.

Statistical Process Control

7
Monitoring Quality in Healthcare
 Type I Error
Incorrectly flagging a signal when there is no signal present; this is known as a false positive.

Type II Error
Failing to flag a signal when a signal is actually present; this is known as a false negative.

Types of Distribution
Poisson
Data that has a Poisson distribution is discrete and is based on events occurring over time
(or space) at a fixed rate on average, but where each event occurs independently and at
random. For example, the number of new hip fracture admissions.

Binomial
Data that has a Binomial distribution is discrete and is based on data with only two
possibilities e.g. the probability of being dead or alive, male or female etc.

Geometric
Data that has a Geometric distribution is discrete and is based on the number of Bernoulli
trials until the first observation with a particular characteristic of interest, where trials are
independent and an occurrence is equally likely on each trial. For example, the number of
days between needles sticks.

A list of chart types and a brief summary of the types of data they serve is found in Table 3 (see
section 2.4 and section 2.5 for examples of their application). This is not an exhaustive list and
is not intended to be prescriptive. A judgement on the type of chart to use should be made
depending on what is appropriate for the data and what measures stakeholders are tracking.
This should be used as a guide only.

Statistical Process Control

8
Monitoring Quality in Healthcare
Table 3: Types of Shewhart Control Chart.
Chart Name Appropriate For
XmR Chart Continuous data, where each data point is an individual measurement and not
an aggregate or average of multiple data points
Xbar & S Charts Continuous data, where each data point is an average of multiple data points
P Chart Discrete data. Data should be a proportion/percentage of successes/failures
C Chart Count data, where the “area of opportunity” is equal for each data point
U Chart Count data, where the “area of opportunity” is unequal for each data point
T Chart For rare events data, recorded as time between unlikely events (continuous).
G Chart For rare events data, recorded as the number of events between unlikely events
(discrete).

Figure 1 is a flow chart which illustrates the decision making process for choosing the correct
chart type.

Steps to create a Control Chart

1. First, select the most appropriate control chart for your data, which is dependent on
the properties of you data. Use the flow chart in Figure 1 and Table 3 to help with
this.
2. Proceed as for the run chart, this time using the mean as the centreline.
3. Calculate the upper and lower control (and optionally warning limits) using the
formulas provided in Appendix 1 (for the appropriate chosen chart)

Statistical Process Control

9
Monitoring Quality in Healthcare
Figure 1: Flow chart illustrating the decision-making process for choosing the most appropriate type of Shewhart Control Chart.

Discrete Are data continuous Continuous

or discrete?

Are data poisson Are data extreme/

(counts) or binomial unlikely events, e.g.
(pass/fail)? HAI occurance?
Yes

No
Are data extreme/ Poisson Binomial Use T Chart
unlikely events, e.g.
HAI occurance?
Is there more than
one data point per
subgroup?

Use P Chart
Yes No
Use Xbar & S Chart Use XmR Chart

No Unequal
Use U Chart

Yes

Use G Chart
Is the Area of Equal
Opportunity equal or Use C Chart
unequal?

Statistical Process Control

10
Monitoring Quality in Healthcare
2.4 Shewhart Control Chart Rules
In order to interpret data from a control chart correctly, a set of consistent rules must be
applied to highlight any special cause variation. These rules highlight data points which may be
of interest in order to prompt the user of the data to investigate further. These rules should be
agreed upon before creating the control chart based on the user’s knowledge of the data, as
creating the chart first could lead to a bias from the user who may feel compelled to choose a
set of rules which show something more in line with what they want to see.

Table 4 contains standard Shewhart Control Chart Rules for identifying special cause
variation agreed for Improvement Programmes in NHS Scotland.

Table 4: Shewhart Control Chart Rules.

Rule Description
Outlier Data point(s) exceeding the upper or lower control limits (at 3 standard
deviations).
Shift A run of 8 or more consecutive data points above or below the centerline.
Trend A run of 6 or more consecutively increasing or decreasing data points. An
observation that is the same as the preceding value does not count towards
a trend.
Outer One – Third Two out of three consecutive data points which sit close to one of the warning
or control limits (within 2 and 3 standard deviations).
Inner One - Third 15 or more consecutive data points that lie close to the centreline (within 1
standard deviation).

These are standard rules which should be sufficient for the vast majority of charts but they can
be adapted accordingly only if it is appropriate for the data and any changes have been agreed
beforehand.

Statistical Process Control

11
Monitoring Quality in Healthcare
 2.5 Shewhart Control Chart Examples
This section will include four examples of SPC charts. Each example will present a dataset, and
follow the flow chart for deciding which chart to use step-by-step.

Example 1

Hospital Standardised Mortality Ratios are calculated for all SPSP-participating hospitals
in Scotland. SPC charts can be very useful for this measure to identify areas of unnatural
variation.

Is the data discrete? – No.

HSMR is a calculated measure which can take any value greater than 0.

Is there more than one observation per subgroup? – No.

There is only one measurement each quarter calculated for each hospital.

As the data are continuous and consist of individual measurements, the preferred chart for
this would be an XmR chart. If HSMR was being calculated and then averaged for multiple
hospitals each quarter, then an Xbar chart would be suitable, with the process being the
average of each measurement across all sites (note: the number of sites should remain
constant).

Control Chart of HSMR

1.8

1.6

1.4

1.2
HSMR

0.8

0.6

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Process Centreline UCL LCL Shift Outlier Trend

The chart above shows that the process looked relatively stable until quarter 19, where
there was a large spike, followed by two outlying quarters. This likely pulled the centreline
up slightly, leading to the downward shift in the centre. So, additional investigation of the
outlying quarters would be necessary.

Statistical Process Control

12
Monitoring Quality in Healthcare
 Example 2
You have a large jar of jelly beans, and from the jar you take 25 scoops of beans. You
measure the number of green beans in each scoop and wish to monitor this process with
an SPC chart.

Is the data discrete? – Yes.

Are there only two possible outcomes? – Yes (green or not green).

As the data are binomial (acceptable/unacceptable), then the best chart for this type of data
is a P chart. The best way to plot this process is with the proportion of successes to the
number of overall beans. In this case, the process will be the proportion (percentage) of
green beans in each scoop.

Control Chart of Proportion of Green Beans Chosen

45

40
Rate of green beans chosen (%)

35

30

25

20

15

10

0
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

Process Centreline UCL LCL Shift Outlier

The variable control limits come from the variable size of each sample. Scoops with more
beans will have tighter limits as the predicted proportion of green beans obtained by
random chance should tend towards the true mean (which the centreline is an estimate of)
when the sample size increases.

The above P chart shows no outliers, but does show a downward shift. There is also no
lower control limit as the calculated limit was less than 0, and a negative percentage is
impossible to achieve in this context.

Statistical Process Control

13
Monitoring Quality in Healthcare
 Example 3
A factory that produces computer accessories takes a sample of 10 keyboards every hour
and counts the number of loose keys per keyboard, but they take a random sample across
all models of keyboard they produce.

Is data discrete? – Yes.

Are there only two possible outcomes? – No. The data is a count of loose keys, not a
classification of whether or not the keyboard had a loose key or not.
Is the area of opportunity variable? – Yes. Different keyboards have different numbers of
keys which could be loose, and each sample has a different combination of potentially
loose keys.

As the data in this example are poisson counts, a C or U chart would be suitable, and
as there is room for one keyboard to have more loose keys than another, the chart used
should be a U chart.

Control Chart of Loose Keys per Sample

0.35

0.3
Number of Loose Keys per
Sample of 10 keyboards

0.25

0.2

0.15

0.1

0.05

0
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

Process Centreline UCL LCL Shift Outlier

The mean number of defects (unadjusted for sample size) found per sample is 2.7 (~0.1 per
100 keys per sample). Most samples fluctuate closely around this number and no samples
are out of control, however there is a downward shift from sample 9 to sample 16,
indicating that further investigation is required to determine what was happening there.

Statistical Process Control

14
Monitoring Quality in Healthcare
 Example 4
A care home monitors how many of its residents have had a fall each month, and wants to
produce an SPC chart to ensure the issue is kept controlled.

Is data discrete? – Yes.

Are there only two possible outcomes? – No. The data is a count of residents who have
had a fall.
Is the area of opportunity variable? – No. The number of residents in this example is
constant each month.

As the data in this example are Poisson counts, a C or U chart would be suitable, and as
the number of residents in the home is constant month-to-month, a C chart is appropriate
here.

Control Chart of Number of Falls per Month

7

6
Number of Loose Keys per
Sample of 10 keyboards

0
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

Process Centreline UCL LCL Shift Outlier

The mean number of defect falls found per sample is around 2. Most samples fluctuate
closely around this number or below, however an upward shift is present in the final 6 data
points.

NOTE: these data could also be presented as a percentage (number of residents who fell/
total number of residents).

Statistical Process Control

15
Monitoring Quality in Healthcare
 The table below provides further examples.

Data Flow Chart Process Type of SPC to use

Number of patients that go Is data discrete? No. Data XmR chart.
through a GP surgery each week are throughput which, by
convention, are considered
continuous.
Is there more than one data
point per subgroup? No. Data
are single measurements.
Percentage of patients that died Is data discrete? Yes. P chart.
within 30 days of admission to
Is data poisson or binomial?
hospital
Binomial (patients can only be
dead or alive after 30 days)
Number of morbidities patients Is data discrete? Yes. U chart.
report when admitted to A&E
Is data poisson or binomial?
departments each day
Poisson (data are counts of
morbidities)
Is the area of opportunity equal?
No. The number of patients may
vary day-to-day and the number
of morbidities each patient can
report is theoretically infinite.
Average daily spend on Is data discrete? No. Monetary Xbar & S chart.
antibiotics prescribed by GP values are continuous.
practices across a health board
Is there more than one datapoint
per subgroup? Yes. Each GP
practice contributes towards the
health board average.
Percent of hospitals across the Is data discrete? Yes. P chart, but data must first be
UK which make the target of expressed as a proportion.
Is data poisson or binomial?
95% of patients seen within 4 Binomial.
hours in A&E
The number of days between Is data discrete? No. Data are T chart.
hospital acquired infections time measurements.
(HAI) for a single hospital site
Are data unlikely events? Yes.
The number of patients Is data discrete? Yes. G chart.
prescribed a certain drug
Are data unlikely events? Yes.
between two adverse drug
events

Statistical Process Control

16
Monitoring Quality in Healthcare
2.6 The XMR chart
There is an alternative school of thought around Shewhart chart selection which proposes only
an XmR chart is necessary. This section lays out the arguments for and against this approach.
The decision over which approach to take is entirely down to the user and the decision should
be made before producing any SPC chart.

Wider range of Shewhart Charts XmR Charts

Advantages When applied properly, these They are simple to understand
charts are correct and provide more
In most improvement situations, the
statistically appropriate results
results can be deemed ‘appropriate
The control limits vary according to enough’
the size of each individual sample
Does not require any assumptions
about underlying data distributions
Quick and easy to apply to practical
situations
They are recommended by influential
members of the quality improvement
communityi
Disadvantages Assumptions about underlying data They are not correct for all types of
distributions may not be metii data
Lack of understanding of these A large portion of the quality
probability distributions can lead to improvement community does not
errors or lack of user confidence in recommend them over the wider
choice of chart range of Shewhart Charts.
The greater complexity of these charts
makes understanding and automation
more difficult

These advantages and disadvantages are for information purposes only and are not intended to
try and sway a user towards one approach or the other. The decision to use just an XmR chart
is down to the user(s) and should be agreed upon before production of any chart begins.

i Stauffer (2010), Balestracci (2014)

ii Wheeler (1996)

Statistical Process Control

17
Monitoring Quality in Healthcare
 Summary
Choosing the correct chart is an important step in monitoring a process, and ensures that
incorrect conclusions around process stability are not drawn. However, chart selection is not
an exact science and there is room for manoeuvrability. It is also important that any decisions
regarding chart selection and rules are made prior to producing the chart, as making these
decisions retrospectively can lead to a bias towards a chart which is more agreeable with what
the user would like to see.

Run charts and Shewhart control charts are not the only type of SPC chart available, and a
process over time is not the only type of data which can be put into an SPC chart. Data which
are point-in-time comparisons, e.g. crude mortality rates by hospital for one quarter in Scotland
could be put into an appropriate funnel plot to be compared against the national average. Like
Shewhart control charts, there are a number of approaches to funnel plots (based on Xbar+S, P
or U charts) depending on the data.

Further Information
There are a wide range of online resources on the use and calculation of SPC charts. As
mentioned in the introduction, the Quality Indicators team have produced an interactive tool in
producing SPC charts. The tool takes the user through the selection process step-by-step and
decides what the best chart is to use.

The Quality Improvement hub has pages dedicated to Shewhart control charts and run charts.
The Healthcare Data Guideiii is an excellent reference guide for this type of analysis.

Acknowledgments
This document was prepared by the Quality Indicators Team from both knowledge of the topic
within PHI and reference material. This document was circulated to individuals with Quality
Indicators and wider PHI for comments and feedback. Their input was much appreciated
and we are grateful for their advice. Any error or lack of clarity that remains is, however, our
responsibility.

Contact Information
If after reading this document you have any unanswered queries on SPC charts, or would
like to discuss any aspects further, please contact the Quality Indicators Team, via NSS.
isdQualityIndicators@nhs.net.

We would also welcome any feedback from you on this document.

iii See http://eu.wiley.com/WileyCDA/WileyTitle/productCd-0470902582.html

Statistical Process Control

18
Monitoring Quality in Healthcare
Appendix A: Formula for calculating Control
Charts limits
XmR Chart

Where xi is the ith datapoint and m is the number of datapoints.

Xbar and S Chart

Where xij is the ith datapoint from group j and is the number of groups and n is the number of
data points per group.

P Chart

Where xij is the ith datapoint from group j and m is the number of groups and n is the number
of data points per group.

Statistical Process Control

19
Monitoring Quality in Healthcare
 C Chart

Where xi is the total number of “defects” from group j and m is the number of groups.

U Chart

Where xi is the total number of “defects” from group j, is the number of groups and n is the
sample size for each group.

T Chart

Where ti is the time between the ith event and n is the number of events.

G Chart

Where gi is the number of occurrences between the ith event and k is the number of events.

Statistical Process Control

20
Monitoring Quality in Healthcare