Ensuring success and sustainability of a quality improvement project

Learning objectives.

By reading this article you should be able to:

• •Explain the crucial nature of the human element in successful quality improvement.
• •Identify some of the methods that can be used to communicate the focus of a change project.
• •Illustrate the importance of sustainability and spread in ensuring success in a quality improvement project.

Key points.

• •The human element is essential for successful change.
• •It is crucial to include frontline staff from the start.
• •You must involve essential people early in the project.
• •Regular and frequent feedback of the measurement metrics is vital.
• •Do not expect unanimous support for changes that you introduce.

Clinical scenario.

Having noticed that a number of patients are hypothermic in recovery after elective orthopaedic surgery and require active warming, you examine an audit from 2 yr ago and see that the same problem was highlighted then, but the recommendations were not adopted. You decide to conduct a quality improvement project to re-address this issue. You aim to ensure success by eliminating postoperative hypothermia after elective orthopaedic surgery.

In the third article of this series on quality improvement (QI), we discuss how to avoid some of the pitfalls and ensure the success of the project.

All QI projects designed to improve healthcare involve change. This requires not only a change to the physical structure and processes involved, but also an emotional element that is crucial to the successful implementation, spread, and sustainability of that change. Without considering all aspects of the local context and culture, it is likely that any improvements will flounder.

This paper will address these three areas of the human dimension of change, sustainability, and successful spread.

Human dimension

The success of a QI project, its viability, and its potential for repetition in other areas are directly related to the aspects of how the project is set up and how it is run on a day-to-day basis. Key aspects to predicting the success of an intervention are the perceived relative advantage, the lack of complexity, compatibility, ‘trialability’ and observability.¹ All of these elements touch upon psychological aspects and will have a variable degree of appeal to different staff members. A lack of consideration of, and attention to, the contextual factors involving the human dimension is a major cause of an improvement project faltering.

Several different models have been described that include the human side of QI. Deming's ‘profound knowledge’ describes four components: a theory of knowledge, understanding variation, appreciation of the system and psychological aspects.² Medicine is a sociotechnical system, with the emphasis on the social aspect, and the psychological aspects of the system are often neglected when we run an improvement project. Without addressing these aspects we are unlikely to succeed.

Different theories of change have similar concepts, and aspects will appeal to different individuals. These include Lewin's three-stage theory (unfreezing, transition, and refreezing), Bridges' three phases for managing transitions and the more comprehensive Kotter's eight-stage process (Table 1).3, 4, 5

Table 1.

Kotter's eight-stage process for managing change⁵

Stage no.	Process
1	Establish a sense of urgency
2	Create the guiding coalition
3	Develop a vision and a strategy
4	Communicate the change vision
5	Empower broad-based action
6	Generate short-term wins
7	Consolidate gains and producing more change
8	Anchor approaches in the culture

These theories highlight the importance of making the case for change. Resistance to change is a common phenomenon and has a number of underlying causes.⁶ Bridges emphasises the difference between change, which is an external phenomenon, and transition that is an internal process.⁴ He describes three phases for managing transitions: letting go of the current position, the transition zone, and the new beginning. He emphasises the need to make the case for an ending of the current situation before embarking on the next step and that different people go through the transition at different speeds, and this is affected by our past experiences and our degree of involvement in the project. Importantly, what is an ‘obvious’ imperative for change to the instigator of a project is likely to be less of a priority for those involved more peripherally.

Understanding the loss and perceived threat felt by individuals may help to understand the strong emotions (fear, anger, hopelessness, and frustration) that may be involved and potentially derail the project. Other responses are to be defensive and to deny the problem, to overemphasise the benefits of current practice, or to blame others for the problems that have been identified. If the emotional aspects are not acknowledged as being normal and handled well, they may become increasingly destructive as conflict spirals out of control. These emotional responses to change have been characterised by the Kubler-Ross grief-and-loss model.⁷ By discussing these openly with a team and recognising explicitly that the emotions are a normal part of change, teams may be helped to understand any feelings of discomfort with change and to move through the process.

Not all causes of a problem are equally weighted; it is important to prioritise efforts and be able to justify this to all those that will be affected. The tools that may help to make and communicate the focus of a change project include the following:

• (i)Pareto analysis is a method to identify what the largest problem is.⁸ The 80/20 rule states that, for many events, the majority of an effect is the result of a minority of causes. The Pareto analysis, named after the Italian economist, Vilfredo Pareto, utilises this principle in order to identify, categorise and rank the causes of a problem according to relative size and frequency. Pareto charts can be used to visualise these findings. A Pareto chart comprises different observations (x-axis) plotted against their frequency (y-axis), and is enhanced by the inclusion of a cumulative percentage line, which illustrates the contribution of each observation to the overall effect. Figure 1 shows the frequency and cumulative effect of various causes of perioperative hypothermia, and we can see that the first two problems, ‘lack of patient warmers’ and ‘cold patient waiting area’, account for 84% of all cases of hypothermia; these should be addressed first.

• (ii) Failure modes and effects analysis (FMEA) is a prospective tool that aims to identify, analyse and prioritise pre-emptively the potential opportunities for your process to fail.⁹ In order to conduct an FMEA, each of the following points needs to be addressed:

• (a)In what ways could your process potentially fail? Consider the potential causes for each potential process failure. Estimate the probability of each potential cause occurring (assign a likelihood score on a 1–10 scale, with 1 being a virtual impossibility and 10 being a virtual certainty).
• (b)Consider the potential effect of failure for each potential cause. Estimate the severity of each potential effect (assign a severity score on a 1–10 scale, with 1 being very benign and 10 being very severe).
• (c)Consider the available methods of detecting failure for each potential cause. Estimate the ease of detecting failure for each potential cause (assign an ease-of-detection score on a 1–10 scale, with 1 meaning that detection is virtually assured and 10 meaning that failure is very likely to go unnoticed).
• (d)Calculate a risk priority number (RPN) by multiplying the scores for probability, severity, and detection scores. The potential cause with the highest RPN should be prioritised.

Fig 1.

Pareto chart of causes of perioperative hypothermia.

These tools provide a factual basis for change. In order for the spread of an idea to be successful, staff requires a vision for how the quality of care will be better. Involving patients at this stage increases the emotional involvement of staff and increases the likelihood of staff seeing a need for change.

Sustainability

Sustainability is the state when a change becomes the norm and can be further built upon. Daft and Noe suggest that up to 70% of attempted organisational change fails.¹⁰ Thus, an understanding of how to ensure sustainability and the factors that affect it is of fundamental importance to prevent wasted effort. These should be incorporated into the project design at the earliest stage of its development. Often, sustainability is only considered part way through a project when initial progress has been achieved.

Buchanan and colleagues describe 11 elements that may affect sustainability (Table 2), whereas the NHS Institute for Innovation and Improvement's sustainability model outlines 10 factors divided amongst three overarching groups: staff, processes, and organisation.11, 12

Table 2.

Factors affecting sustainability¹¹

Category	Outline definition
Substantial	Scale and fit with organisation
Individual	Commitment and expectations
Managerial	Style, approach, and behaviours
Financial	Balance of costs and benefits
Leadership	Setting vision and goals
Organisational	Policies, systems, and structures
Cultural	Shared beliefs, values, and priorities
Political	Stakeholders and influence
Processual	Implementation methods and project management structures
Contextual	External conditions and stability
Temporal	Timing and flow of events

Once these factors are understood, a culture change is needed; the US Institute for Healthcare Improvement suggests six areas that can help sustain improvement:

• (i)supportive management structure;
• (ii)structures to ‘fool proof’ change;
• (iii)robust, transparent feedback systems;
• (iv)shared sense of the systems to be improved;
• (v)culture of improvement and a deeply engaged staff;
• (vi)formal capacity-building programmes.¹³

Within these frameworks, the essential underpinning philosophy of measurement for improvement is that it makes the problem visible and allows feedback of this to staff so that they can see if and when improvements occur. This feedback can take a number of forms and should be targeted to the needs of the specific audience. An invaluable tool for showing improvement is a run chart or Statistical Process Control (SPC) Chart.14, 15 Crucially, this feedback needs to be frequent enough to maintain staff motivation and be delivered by a creditable leader.

Run charts consist of data plotted against time, and will usually display a measure of the average value, usually the median value. All data inherently contain variation. Understanding and reducing variation is the key to quality. It is important to distinguish what is expected variation (common cause) from what is unexpected variation (special cause), and plotting of data can help to determine this. Common-cause variation is a feature of a stable system (although stable is not synonymous with effective).

Run chart data can be evaluated using four rules to determine if the variation within is caused by the intrinsic nature of the process (i.e. common cause, or because of an assignable cause, such as an improvement project).

Consider the following run chart (Fig. 2) showing the results of a year's data collection of postoperative temperature. It can be analysed using four simple rules to find which trends or patterns are features of a special-cause variation.

Fig 2.

Run chart of postoperative temperature.

Identifying special-cause variation from run charts (Fig. 3)

Fig 3.

Run chart rules.

Rule 1: a shift in the process

This is indicated by six or more consecutive points above or below the median. There will be an underlying cause for this trend, which should be sought. The explanation in this series could be that, because of a safety recall of warming mattresses, it was harder to keep patients warm and there was a delay in sourcing a replacement item, leading to a prolonged dip in the data.

Rule 2: five or more points all increasing or decreasing

This indicates a trend and often occurs when a project involves sustained intervention(s) over time. Here, this could be an introduction of a number of new interventions, such as new equipment, education, and directed feedback to staff on their performance against their peers.

Rule 3: number of runs

A statistical analysis suggests there will be an expected range of ‘runs’, which are collections of points all above or below the median.¹⁵

Looking at the first 6 months of the year, we can see that there are 11 runs. Referring to a table for the expected number of runs shows us the expected range is between 9 and 18 for 26 data points. If there is more than the expected number of runs, the process is unstable and inconsistent. A reduction in run number shows an increased stability of a system.

Rule 4: astronomical data points

This is a data point that is markedly different to the others in the range. It usually represents a flaw in data collection or definitions; in this case, a new member of staff misread the instructions and counted all the patients with a temperature over 36.5°C rather than 36°C.

SPC charts

For most projects, a run chart will suffice. They are not, however, as sensitive in detecting a special-cause variation as SPC or Shewhart charts.

Walter Shewhart was an engineer and mathematician initially working in the telephone industry in 1920s America

…Through the use of the scientific method, extended to take account of modern statistical concepts, it has been found possible to set up limits within which the results of routine efforts must lie if they are to be economical. Deviations in the results of a routine process outside such limits indicate that the routine has broken down and will no longer be economical until the cause of trouble is removed.¹⁶

SPC charts are similar to run charts in that they are a time-series chart; they also show a mean and upper and lower control limits that delineate the expected boundaries of variation around the mean if only common-cause variation is present.

The upper and lower confidence intervals are calculated and approximated to three standard deviations from the mean. All common-cause variations lie within these. More reliable/less variable processes will naturally have narrower confidence limits than highly variable ones. It allows a rapid assessment of when special-cause variation is being exhibited; where high-stakes processes are involved, this may require immediate investigation and action.

Spread

Spread is the process by which your change intervention is communicated beyond the immediate locality, outside the so-called ‘pockets of excellence’. Barriers to a successful spread may involve the ‘not invented here’ effect and ‘improvement evaporation’, whereby any initial benefits decay over time. The ‘What's in it for me?’ framework may help to identify potential problems in the area of a proposed spread and help to overcome aspects of the ‘not invented here’ effect.¹⁷

The factors that increase the likelihood of spreading an idea successfully include laying the foundation for spread by communicating clearly; enlisting a creditable improvement champion in a position of influence; considering how a successful spread may be achieved including specific steps to do this and refining the process as different staff groups reinvent the improvement to fit their local context.

This often involves local teams going through similar steps in the initial improvement project, and takes time whilst the emotional transition occurs. It is essential that the importance of the local context and culture is not underestimated and that sufficient time and support are given to teams to identify with, and implement the necessary changes to benefit from the spread of the idea.

The human element of spread and how people adopt change is often not given sufficient attention. Rogers's model of diffusion suggests a spectrum from innovators and early adopters (who are the target groups most likely to successfully implement an innovation) through to laggards.¹ Although some suggest that this model is not applicable to areas outside of the commercial market and that psychological variables, such as tolerance of ambiguity, may be more important factors.¹⁸

Spread generally occurs in two ways:

• (i)dissemination, where spread is formal and planned, through vertical hierarchies (e.g. through professional body guidelines, training courses, leaflets, etc.);
• (ii)diffusion, where spread is informal and unplanned, through horizontal peers (e.g. through word of mouth, social networks, local demonstrations, etc.).

In actuality, these two methods are opposite ends of a continuum, and the whole field of ways and means needs to be considered when creating a spread plan to propagate an intervention. In order for an institution to successfully sustain and spread ideas, the underlying principles and skills need to be embedded in the organisational culture.

Summary

It is clear that implementing change can be a difficult endeavour. There are many different techniques that can be used at each stage: defining aims, identifying problems, prioritising problems, and beyond. Importantly, QI tools are a way to help staff take ownership of improving the quality of care and potentially ensure success.

Declaration of interest

None declared.

MCQs

The associated MCQs (to support CME/CPD activity) will be accessible at www.bjaed.org/cme/home by subscribers to BJA Education.

Biographies

Lauren Weekes FRCA is a final year specialty registrar in anaesthesia in the South West Peninsula Deanery. She has undertaken an advanced module in quality improvement and is a member of The Health Foundation's Q Community.

Tom Lawson FRCA is Consultant at Plymouth Hospitals NHS Trust. His interests include quality improvement methodology and basic science for FRCA.

Matt Hill BMedSci, MRCPCH, FRCA, MSc is a Consultant at Plymouth Hospitals NHS Trust and Honorary Fellow at Plymouth University Peninsula School of Medicine and Dentistry, where he teaches Quality Improvement. He has an MSc in Quality and Safety in Healthcare, and has been faculty for the IHI. He was a founding member of The Health Foundation's Q Community and is the Clinical Lead for the safety culture work stream of the National Patient Safety Collaborative.

Matrix codes: 1I02, 2H02, 3A15