The NHS Institute for Innovation and Improvement website is now being administered by NHS Improving Quality - the Productive Series and other products are still being provided by Delivery Partners and supported by NHSIQ, and all material relating to the Productive Series is still accessible.

The NHS Institute closed on 31 March 2013. If an item you are looking for is not available here, you'll be able to see all publicly available content on The National Archives website: http://webarchive.nationalarchives.gov.uk/*/http://institute.nhs.uk.
X
This site uses cookies to help performance and allow us to improve your browsing experience. You can click here to view the cookies we use on this website along with information on how to restrict cookies using your browser settings. By clicking on the Continue button, you accept the terms of our privacy policy on our website.



   
i
| | |

Variation - An Overview


Back to previous page


Variation - An Overview
 


What is it and how can it help me?

Look around and you will see variation everywhere: different cars, trees, people etc. In a class of children, for example, you will find a range of weights and heights. These natural differences may seem insignificant, but they can affect the way a child is seen in a clinic and how much time they need: an overweight or underweight child may need more advice, and so take up more time.

There are many sources of variation along elective care pathways and these can affect the flow of patients through healthcare systems. Much of the variation is caused by the way we organise and provide services - we call this artificial variation. 

The source of variation is important as this determines what we should do next.

Natural variation
Is an inevitable feature of healthcare systems. Sources of natural variation include:

  • Differences in symptoms and diseases that patients present with
  • The times of day that emergency patients arrive
  • The socio-economic or demographic differences between patients
  • Staff skills, motivation etc

Artificial variation
Is created by the way the system is set up and managed. Sources of artificial variation relevant to reducing waiting times include:

  • The way we schedule services
  • The working hours of staff and how staff leave is planned
  • The order in which we see and treat patients
  • How much work we group and deal with in batches
  • How we manage clinics to deal with priority or urgent cases

Natural and artificial sources of variation are different from 'common cause' and 'special cause' variation. Common cause describes variation that is predictable and expected. Special cause describes variation that is unusual or unexpected. Examples of special cause variation include a large-scale change creating a peak in demand, unexpected weather conditions and flu epidemics. On a small scale, an exceptionally obese or underweight child could trigger a ‘special cause event' in a clinic due to possible social welfare concerns which would involve third parties. 

There is a complementary tool statistical process control which looks at variation using a statistical methodology to help you to identify predictable and unpredictable variation and know how to tackle both.

Linking natural and artificial sources of variation to the way variation behaves (i.e. whether we expect it or not) :

 

Common cause
'predictable or expected variation'.

Special cause
'unusual or unexpected variation'.

Source of variation is natural

Patient's age, gender, disease, condition, personal circumstances.

An exceptionally underweight child turns up at a health clinic triggering social welfare concerns.  It's the first time the clinic has seen this child.

Source of variation is artificial (i.e. comes from the systems we develop)

Some doctors make decisions weekly while others do this daily.

Ordering different tests for the same clinical presentation.

Different systems to manage consultant to consultant referrals from referrals from GPs.

Patients or tests being seen / dealt with out of turn.

A series of things ‘go wrong',
The patient is ‘lost' in the system and waits for three years.
 
Someone coded in the wrong number.

The patient's surname was Smith.

The test results were put in the wrong pile. 
 
The paperwork couldn't be found, the patient moved GP.

 

When does it work best?

Reducing and managing variation are essential approaches to reducing delays in services. There are two reasons:

  • Waiting lists build up because sometimes demand for work exceeds our capacity to do deal with our work. The mismatch is due to variation in both demand for work and variation in our capacity to deal with work. There's a lot of evidence that suggests that our capacity to deal with work varies more than our demand
  • Variation in the way we work and do work, such as the way we deal with paperwork, the timing of decision making along a clinical pathway, the decisions we make, how we organise and manage work, all impact the pace that patients progress and the number and length of unnecessary delays patients experience

This graph illustrates the impact of both natural and artificial sources of variation on how long it takes individual patients to receive treatment after a referral.


Variation - An Overview 1.jpg

 

 

 

 

 



The 18 week target is 126 days and although not shown on this chart, it is just a little above the dotted black line. The dotted black line is the average referral to treatment time. If you look at the chart there are a number of patients waiting longer than reducing waiting times due to the amount of variation. Looking at a chart like this, a natural question is to wonder 'what is the cause of the variation?' and 'what can I do about it?'.

How to use it

This guide helps you identify sources of variation. It links to two other guides that describe what you can do about it, depending upon the source of variation.

Step 1: How much do things vary?
Use the Patient Journey Analyser (PJA) to see how much variation you have in your specialty or sub-specialty.

  • Referral to treatment times - statistical process control (SPC) chart
  • Lowest performing statistical process control chart

Step 2: Is the source of the variation mainly natural?
Identify patient characteristics - age, gender, clinical presentation - to find patterns that may be associated with delays or resources being wasted (e.g. patients not attending). For example, there is some evidence that patients in older age groups are less likely to push for earlier treatment and so don't ‘jump the queue'. There could be seasonal patterns relating to referrals or demand.

Identify patient characteristics that may be associated with procedure or care pathways required. For example, in one hospital a study showed that six procedures made up 52 per cent of the work. If you are able to identify specific patient characteristics to link them to these procedures then you are using ‘natural variation' to an advantage. Some people call this segmentation.

Present what you find graphically. You may wish to talk to your Information Department to help you to do some of this. 

Step 3: Is the source of the variation mainly artificial?
Most of the variation in healthcare is due to our ways of working and the systems we have set up. Quite often we do things that add variation unintentionally. To reduce this, you need to identify the sources to reduce the impact to delays in patient care.

Select a patient journey, procedure, or administrative process and map them out to identify sources of variation (see example below).
Here are a few techniques:

  • Mapping techniques that involve the team will help you collectively spot opportunities for improvement and redesign that will reduce variation along the whole pathway
  • Compare journey times using case files.  a source of information along key stages of the last 10 patients who received treatment (see example below) or shadow a patient to collect information
  • Compare pathway time for 10 bits of paperwork / 10 case files using a tracer approach - a form people fill whenever they ‘touch' a patient's notes

Analyse variation in capacity and demand
Take a complementary approach to looking at variation in capacity and demand. Plot out the underlying numbers. There's a short introduction to capacity and demand and a more detailed approach to carrying out capacity and demand analysis.

Some examples for both capacity and demand include:

  • To identify variation in demand, look at patterns in the time of day, day of week, weekly and monthly demand. How much does it vary?
  • To identify variation in capacity, look at patterns in annual leave, staff sickness, skill mix and rooms / equipment. The trick here is to look at mismatches in demand and capacity and link this to the activity which is the actual work done in a day, week or month

Present what you find graphically. You may wish to talk to your Information Department to help you to do some of this. 

Step 4: Do you understand the real cause of variation?
Your aim is to uncover the why and not the what.  Five whys and cause and effect are just two of the tools available to help you to identify the root cause of artificial variation.

Examples

How to identify both natural and artificial causes of variation for did not attends (DNAs).

Mapping what happens to patients, test results, paperwork against key stages can highlight difference. For example, the diagram below shows the locally agreed timescales for the current colonic pathway accompanied by an illustration for the last ten patients who went through this pathway under the care of three consultants. It shows variation from referral to treatment of 21 to 167 days. This is enough to have a really useful discussion to try and understand the causes of the differences and then to identify potential solutions.

Variation - An Overview 2.jpg


 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 




Source: Cancer Services Collaborative Improvement Partnership 'How to Guide Achieve Cancer Waiting Times' 2006.

What next?

Evidence suggests that we and the way we work are the source of a lot of variation that causes delays to patients.

This means at the heart of reducing or managing variation you will need to engage with other members of staff. The example above is quite convincing. There are a couple of tools that may help you: clinical engagement and staff perception.

The source of variation is important as this determines what we should do next:

  1. Aim to reduce artificial variation (or variation caused by the way that we work)
  2. Aim to manage natural variation

Sometimes there is so much variation, say in referral to treatment times, that things seem quite chaotic. Spend a bit of time understanding why. It is possible you are looking at two or more very different clinical pathways. If so, you will want to look at the different clinical pathways separately to understand where to focus improvement efforts. An alternative explanation is that there is no set agreement for the way things are done; one improvement focus could be to show staff the impact this has on patient care. 

Other useful tools and techniques on this website

Statistical process control is a complementary tool which looks at variation using a statistical methodology to help you to identify predictable and unpredictable variation and know how to tackle both.
Reliable design shows how reducing variation is key to higher reliability.

Additional resources

Documents:
The 10 High Impact Changes highlights the importance of reducing artificial variation in admission and discharge processes (change 3 and change 4)). Clinicians' Guide - 10 High Impact Changes.

Background

The principle of looking at variation and keeping a steady flow of work originates from Deming.  Many improvement methodologies have reducing variation at their core, for example Lean, Six Sigma, clinical systems improvement and reliable design.

Acknowledgements / sources

Sections of text extracted and sourced from ‘Improvement Leaders' Guide: Improving Flow' NHS Institute for Innovation and Improvement.

© Copyright NHS Institute for Innovation and Improvement 2008