A T RAI N EE ’S GU ID E

TO A Q U A L IT Y
I M P RO V E M EN T PR OJ EC T

A TRAINEE’S GUIDE
TO A QUALITY
IMPROVEMENT PROJECT
Quality Improvement Directorate
March 2021

www.rcseng.ac.uk/standardsandguidance
 Contents

Introduction 3

Stages of a project 4 - 12
1. Identify
2. Objective
3. Baseline measurement
4. Plan, do, study, act
5. Sustainability
6. Share

Conclusion 13

Appendix: Rules to identify special cause patterns

on a run chart 14 - 15

References 16

Bibliography 16

The College would like to thank Daniel Watts, Devan Limbachia (Surgical trainees)
and Namet Surana (GP trainee), for drafting and contributing to the guide.

2
 Introduction
Every hospital environment and healthcare setting contains opportunities to improve
patient care. Existing and novel processes and ways of working require constant
re-evaluation and adjustment to ensure areas of weakness are not overlooked; a structured
quality improvement (QI) project provides a framework to facilitate this. In a healthcare
setting, a QI project can be undertaken by any member of the hospital trust.
As part of their training, junior doctors are often expected to instigate projects, being
uniquely placed to identify areas for improvement ‘on the ground’. Here we present a
simple guide, based on our own experiences of organising a successful QI project from
conception to completion and detail our mistakes and lessons learned in the process.
We hope this guide will serve as a valuable roadmap for anyone planning to embark
upon their own QI journey.

3
 Stages of a project
Sometimes the terminology of QI projects Root cause analysis
can be confusing, but the underlying When improving standards following
principles are quite simple and intuitive. an incident, root cause analysis can be
We recommend a simple six-stage used to identify the underlying cause
approach to a QI project, in line with the to prevent recurrence.
‘model of improvement’ framework.
Three basic types of cause are normally
identified.
1. IDENTIFY ● Physical causes – material items
The first step of any QI project is to identify failed in some way (for example, the
an area that needs improvement. This often online handover list is not functioning,
results from: resulting in an inability to identify
patient caseload).
● personal experience of patient care;
● Human causes – a person has made
● a critical incident; a mistake or not carried out a required
● an audit. task. Human causes can often lead to
Once identified, a couple of tools can physical causes (for example, a task
then be used to explore this further to handed over to a wrong staff member).
understand the underlying issues: ● Organisational causes – a system,
● root cause analysis; process, or policy that people use
to make decisions or do their work
● driver diagram.
is not present or does not work
better system performance (care) as intended (for example, lack of
and better professional development standardised uniforms).
(learning)’.5
There are several basic tools that can be
used to perform a root cause analysis.
These include:
● five whys – by repeatedly asking why
something has occurred you can get to
the root of the problem.
● drill down – split a problem
into sections to better understand
each area.
● cause and effect (fishbone)
diagram – a chart that helps
identify the many possible causes
for an effect or problem by sorting
them into categories.

4
It is usually helpful to ask other members team members can help to generate ideas
of the team what they think causes the and can also help with embedding change
issue and how they think it might be at the end of the project.
improved. Valuing and listening to other

Medication not given Why?

Medication not prescribed Why?

Doctor unaware medication needed prescribing Why?

Task handed to incorrect staff member Why?

Poor recognition of different staff members Why?

Root cause!

Now think of solutions

Figure 1: Example of a ‘five whys’ root cause analysis

Driver diagram A driver diagram is a visual representation

When trying to improve patient experience, of all the factors that could influence
a root cause analysis may not be the a patient’s experience, which then
best option. Instead, a driver diagram or enables structured thinking as to how
alternative tool may be more useful. their experience could be improved.
See example diagram opposite (Figure 2).

5
 Aim Primary Drivers Secondary Drivers Change Idea

Cognition
(e.g. dementia) Memory
Assessments
Language
Barrier Improve Access
Translators
Patient Poor Hearing
Factors Ensure Hearing
Loops Available
Poor Eyesight

Patient leaflets
Lack of
Information

Uniforms Not
Identifiable
Standardise
Uniforms
Lack of Uniforms
Doctors
Staff Resistance Lanyards
to Uniforms
Improve the Staff Name
Staff
identification Infection Control Badges
Factors
of Healthcare
Improve
Cost of Uniforms
Awareness

Communication
Staff Training

Lack of training

High Staff Ensure Staff

Turnover Continuity

Patients
Ward Reduce Patient
Frequently
Factors Movement
Moved

Lack of Signage Ward Posters

Figure 2: Driver diagram highlighting factors that influence the identification of healthcare staff

6
Other tools that can be used to assess how different perspective. Doing this early on
a system works include: can help to shape your project ideas and
● conventional process mapping; confirm that improvement is needed.
● value stream mapping; NB. Some trusts have additional forms
of QI assistance. There may be an
● spaghetti diagram.
established QI champion, QI mentors
OUR PROJECT or a QI hub. Approach these resources
The aim of our project was to improve early for support and advice.
the recognition of healthcare staff on
OUR PROJECT
the hospital ward. While on the ward we
Despite having the ward team onboard,
noticed that both staff and patients were
we were met with resistance from senior
frequently struggling to identify the job roles
managers who were based away from
of other staff members. This led to incorrect
the ward. This occurred at a relatively
handover and miscommunication among
late stage of the project and we facilitated
staff and, most importantly, to confusion
resolution to this by meeting with the
among patients.
director of service improvement.
LESSON LEARNED
LESSON LEARNED
Our initial idea focused only on improving
We should have tried to canvass the full
the recognition of doctors but following
range of opinions from all those affected,
discussions with both patients and staff,
including all service users. Involving an
we expanded our project aim to include all
executive or management sponsor may
clinical healthcare staff.
have facilitated an earlier resolution of the
TIP problems that we faced.
It is important to keep it simple
TIP
– small incremental changes will be
Starting small will improve the chance
more attainable.
of success, as only a smaller group of
people will need to agree to any change.
2. OBJECTIVE The project can be scaled up once the
methodology has proved to be effective.
Having identified the underlying cause,
try to define your project objective clearly
so everyone is aware of what you are 3. BASELINE
aiming to achieve. Using the SMART MEASUREMENT
(specific, measurable, achievable, realistic,
timebound) criteria as a starting guide Successfully measuring any change
is a constructive way of establishing a will be key in demonstrating its efficacy.
project aim. Once your project objective is It is therefore vital that measurements
established, try to build a multidisciplinary are taken correctly to demonstrate your
team around the project. Finding wider project’s success. Before implementing
support from those most affected by any any changes, you will need a baseline
potential changes (stakeholders) will prove measurement as a way of tracking your
essential for any project. Including them in projects progress. Ideally, any baseline
your project team will make things easier should include at least 15 data points to
going forward as it can help to build support allow you to analyse any changes over
and allows others to contribute ideas that time, which will be discussed in more
you may not have thought of by giving a detail later.

7
Donabedian was a physician who came each one leading to improvements in the
up with a framework for evaluating health next, with the model of improvement adding
services and quality of care. His model one more.
described three main types of measure,

Structural
Measures
Describes the structure
within which healthcare
is being delivered
e.g. Staffing numbers,
Theatre capacity,
equipment

Balancing
Measures
Outcome Describe
Measures metrics used
to ensure an Process Measures
Describes the
improvement in one Describes the
effects of healthcare
area isn’t negatively healthcare being delivered
on patient/population
impacting another areas to patients.
outcomes. Funding is often
based around them. e.g. Education, diagnosis,
treatment
e.g. A&E waiting times,
re-admissions rate

Figure 3: Framework for evaluating health services

OUR PROJECT TIP

We created a questionnaire that measured Projects linked to trust or national
how well staff and patients could identify objectives are more likely to obtain
staff roles to gather a baseline before any senior buy-in and therefore have
changes were implemented. a greater chance of success.
LESSON LEARNED
If you are creating a questionnaire, think
closely about what will happen next. It is
important to make sure that you are able to
easily measure any subsequent changes
that do occur. By planning from the start
how the changes will be measured and
analysed over time will make things much
simpler at the end.

8
4 . P L A N , D O , S T U D Y, A C T
Once you have identified the area requiring improvement, understood the cause of the
issue, and measured the baseline, the next stage is to plan and implement the necessary
intervention(s). Rather than a single large intervention, implementing several small-scale
changes will increase a project’s chance of success. This should be done according to the
plan, do, study, act (PDSA) cycle.

Plan
Plan the next change
or intervention to be
implemented
Act Plan Do
Carry out the test or
change and collect the
data required. This be
based on the measurable
PDSA Cycle outcomes agreed in the
planning stages
Study
Analyse the collected
data compare to predictions
and reflect on what has
Study Do been learned
Act
Plan the next change cycle
or full implementation

Figure 4: PDSA cycle once each change has been tested, measured and acted upon;
the process is then repeated, creating a cycle

Plan These would be the easiest to implement

During the ‘plan’ stage, you should carefully as there were already several leaflets that
plan which changes you are going to could be adapted.
implement and study. The intervention LESSON LEARNED
should be small enough that it takes place Before trying to implement new changes
in a reasonable time scale. The changes it can be helpful to build upon and improve
should be implemented initially as a local existing designs and processes to gain
‘trial run’. traction.
OUR PROJECT TIP
We had a few different ideas for improving Write down everything that is done and
the recognition of staff on the ward. keep a folder with all documentation.
We decided to start with patient leaflets. This is really useful when writing up your
project at the end

9
Do Study
The ‘do’ stage is where you implement Once your data have been collected they
your change(s). Any change(s) should be will need to be analysed. One effective and
tested on a small scale first. The cumulative relatively straightforward method to do this
result of multiple small-scale changes, is using ‘run charts’ (e.g. Figure 4). These
each tested, learned from and fine-tuned, charts allow you to analyse the full impact
should result in a measurable and reliable of any change over a period of time.
improvement. Collecting data accurately Foremost, you need to be sure that the
to show whether the changes being tested change you are observing is due to the
have resulted in improvement will be changes you have implemented. This will
essential in proving your idea’s success. prevent you from reacting unnecessarily
To do this, small datasets should be to one-off changes. There are two basic
collected regularly, 10 data points with reasons that changes occur, and it is
10 samples for each has previously important to differentiate between the two.
been suggested.
● Common cause variation – this is
OUR PROJECT the natural variation that occurs
During our project, we tried to move on to within normal practice and cannot be
subsequent PDSA cycles before completing accounted for by any specific factor
the previous one. The urge to move on (for example, patient demographics).
is understandable but may compound
● Special cause variation – this is the
problems later.
unnatural variation that is secondary
LESSON LEARNED to a specific factor; this should be the
We should have made sure we allocated intervention you are investigating
enough time to complete each PDSA cycle (e.g. introducing patient leaflets).
and collected more data.
You can use run charts to differentiate
TIP between these factors when analysing
Start with the easiest change to implement your results. In Figure 4, for example, the
first, as this can be helpful in identifying any fluctuations largely represent changes
teething problems. resulting from common cause variation,
whereas the overall positive trend
represents changes resulting from special
cause variation (our interventions).

10
 100

80
% of Patients

60

40

% of patients
20 Median

0
1 2 3 4 5 6 7 8 9 10 11 12
Weeks

Figure 5: Run chart showing the percentage of patients who reported that they were able
to recognise staff each week (note that the graph is for illustrative purposes only and does
not contain real data)

Please see Appendix A for an explanation Act

of rules used to identify special cause To complete the PDSA cycle, you must
patterns in data sets. decide on the next step. This can take three
Our project different approaches:
During our study we did not fully appreciate ● Adopt – fully implement the change
the importance of assessing for a long-term from this PDSA cycle.
trend. We therefore did not collect enough
● Adapt – amend the change from
data points, and so could not demonstrate
this PDSA.
whether the changes we were seeing were
due to common cause or special cause ● Begin the next cycle – start the
variation. planning for the next PDSA cycle.
LESSON LEARNED OUR PROJECT
Smaller, more regular data samples are We adapted our patient leaflets and ward
more important than larger, infrequent posters following feedback from patients
samples. and staff after their introduction on the
ward.
TIP
Try to ensure that enough baseline LESSON LEARNED
measurement are collected so the system Just because your idea was not successful
can be confirmed to be in a steady state does not mean that it cannot be
before any changes are implemented. implemented with adaptations; however,
Ensure that data areas collected at different you should make sure to reassess the new
times on different days of the week to idea fully with a complete PDSA cycle.
reduce the influence of confounding factors. TIP
Try to ensure that changes are simple and
focused or easy to follow.

11
5 . S U S TA I N A B I L I T Y 6. SHARE
One of the biggest challenges with any Once you have completed your project,
QI project is sustaining any change that it is important to consider how you can
has been achieved. Implemented changes share your results. Consider local events
can be forgotten as time goes on, so it in your hospital – grand rounds or QI
may be beneficial to run regular teaching project symposiums can be found in
sessions to refresh staff about the findings most trusts. These are good ways to
of your project. update your colleagues. If your results
This can be compounded as staff have demonstrated a significant change
(particularly juniors) are constantly or improvement, it may have a wider
changing jobs and moving hospitals. relevance and so it is worth considering
While this allows us a great opportunity writing up the project as an abstract
to pick up new ideas, it also gives little presentation or as a journal paper.
time to implement them. This is part of the OUR PROJECT
reason why senior management should be Although our project was not an audit, we
involved early on during the project, as they were able to present our pilot project at a
may be able to recruit others to continue local audit meeting. This helped us to gain
your work when you have moved on. more attention from senior management
Incorporating your changes into and allowed us to secure funding to roll out
standardised frameworks can help the changes more widely.
ensure your changes are sustained, LESSON LEARNED
these can include: A project does not have to be fully
● proformas; completed to be presented. You even use
presentations to gain support and to gather
● check-lists;
further opinions.
● protocols;
TIP
● hospital policy; Aiming to present at a specific meeting can
● guidelines. help to provide a deadline to ensure that
OUR PROJECT you work more efficiently.
Our project took much longer to implement
than originally anticipated. As a result, team
members moved to new trusts before the
project was finished. This meant trying to
recruit new members to join the project on,
to ensure its continuity.
LESSON LEARNED
Incorporating your project within a QI
database can allow hospital management
to ensure that a project continues long after
you have left.
TIP
Make sure that you inform new staff
members of your project at changeover
times (e.g. August). They will often have
an induction that you can present at.

12
 Conclusion
Although the process around QI project can be intimidating, it is a valuable tool for improving
patient care and standards within any environment. With the recent significant change in
clinical practice resulting from the COVID-19 pandemic, this has been made even more
apparent. This simple guide provides the foundations for starting any quality improvement
project and allows readers to learn from the mistakes based on our own experiences.

13
 Appendix: Rules to
identify special cause
patterns on a run chart
Below are some rules you can use to analyse charts objectively to look for special
cause variation.1 If ≥1 rule is met in your run chart, it shows there is non-random variation
i.e. due to your intervention.
Note that we have used a figurative data set to illustrate these rules. A run chart should
include at least 15 points of data for the rules to be applied.2

Rule 1: Shift
100
This pattern observes
a number of consecutive
80
data points, all either
above or below the median.
% of Patients

Data points cannot include 60

those that fall on the median.
Typically, six or more 40
consecutive points % of patients
are needed to prove
20 Median
a shift pattern.

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Weeks

Rule 2: Trend
100
This pattern observes a
number of data points all
increasing or all decreasing. 80
If the value of two or more
% of Patients

consecutive data points is 60

the same, only the first data
point is counted. Data points
40
can include those that cross
the median. % of patients
Typically, six or more 20 Median
consecutive points
are needed to prove 0
a trend pattern. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Weeks

14
Rule 3: Runs
100
This pattern observes
too few or too many runs
(one or more consecutive 80

data point(s) on one side

% of Patients
of the median). 60
The number of runs can be
calculated using the number 40
of crossings of the median
% of patients
line plus one. An appropriate
number of runs for a given 20 Median
data set can be calculated
using a statistical table. 0
1 2 3 4 5 6 7 8 9 10 11 12
Statistically significant
Weeks
change is signalled by too
few or too many runs. Example: this data set Run Limit

Data points cannot has 10 useful data points No. Data Points Lower Limit Upper Limit
include those that fall (data points from week 5 10 3 8
on the median. and week 7 are not useful 11 3 9
as they fall on the median). 12 3 10
Using the statistical table, 13 4 10
we expect between three 14 4 11
and eight runs. Our data
15 4 12
set shows six runs and
16 5 12
therefore we do not have
special cause variation 17 5 13

according to this rule.3 18 6 13

19 6 14
(Table summary from p.11)

Rule 4: Astronomical point

100
This pattern observes a
data point that is obviously
different from the rest of the 80
points – it is subjective
% of Patients

(unlike the other rules). 60

40

% of patients
20 Median

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Weeks

15
 References
1. ACT Academy. Driver Diagrams. London: NHS Improvement; 2019.
2. Provost LP, Murray SK. The Health Care Data Guide: Learning from data for improvement.
San Francisco, CA: Jossey Bass; 2011.
3. NHS Institute for Innovation and Improvement. A Guide to Creating and Interpreting Run and
Control Charts Turning Data into Information for Improvement. London: NHS Institute; 2009.

Bibliography
ACT Academy. Plan, Do, Study, Act (PDSA) Cycles and the Model for Improvement. London: NHS
Improvement; 2018.
ACT Academy. Developing Your Aims Statement. London: NHS Improvement; 2018.
ACT Academy. A Model for Measuring Quality Care. London: NHS Improvement; 2017.
ACT Academy. Managing Variation. London: NHS Improvement; 2003.
ACT Academy. Driver Diagrams. London: NHS Improvement; 2019.
Batalden PB, Davidoff F. What is ‘quality improvement’ and how can it transform healthcare?
Qual Saf Health Care 2007; 16: 2–3.
Etchells E, Woodcock T. Value of small sample sizes in rapid-cycle quality improvement projects 2:
assessing fidelity of implementation for improvement interventions. BMJ Qual Saf 2018; 27: 61–65.
Foster P. 5 root cause analysis tools for more effective problem-solving [blog post]. EASE, Inc. 30
October 2018 https://www.ease.io/5-root-cause-analysis-tools-for-more-effective-problem-solving
(cited February 2021).
Holmes D. Mid Staffordshire scandal highlights NHS cultural crisis. Lancet 2013; 381: 521–522.
Keogh B. Review into the Quality of Care and Treatment Provided by 14 Hospital Trusts in England:
Overview report. London: NHS England; 2013.
Mid Staffordshire NHS Foundation Trust Public Inquiry. Report of the Mid Staffordshire
NHS Foundation Trust Public Inquiry Executive Summary. Robert Francis QC, Chair. HC947.
London: The Stationery Office; 2013.
Mind Tools. Root cause analysis: tracing a problem to its origins. Emerald Works.
https://www.mindtools.com/pages/article/newTMC_80.htm (cited February 2021).

16
The Royal College of Surgeons of England
The RCS of England produces a wide range of standards and guidance to support
the surgical profession within the areas of team working and leadership, legal and
ethical concerns, personal development and service improvement. To find out more
about our work visit www.rcseng.ac.uk/standardsandguidance.

The Royal College of Surgeons of England

35-43 Lincoln’s Inn Fields
London WC2A 3PE

The Royal College of Surgeons of England

The Royal College of Surgeons of England
@RCSnews
The Royal College of Surgeons of England Registered Charity number 212808

www.rcseng.ac.uk/standardsandguidance