Leaders in the science of improvement should speak the language of variation.  Knowledge about separating variation of outcomes of a process or system in to common and special causes helps to decide appropriate actions for that process or system.  Inappropriate action may make things worse.

                                                                                    The Improvement Guide (Langley et al, page 81)

Measurements of all outcomes and processes will vary over time. Some variation is intended, for example varying services for people depending on their needs. In contrast unintended variation is due to differences not connected with different needs, for example we would want everyone to have a good experience.  Unintended variation results in poor quality, waste and harm, and commonly forms the focus for improvement. 

Often, especially for accountability, we look at data aggregated over long time periods (for example, by year or by quarter).  This hides the variation in the data, so measuring more often (hourly, daily, weekly or monthly) can be more informative.  

A central aspect of improvement science is looking at charts of data over time, to understand variation and therefore assess whether a change is an improvement. There are two tools for this – run charts and SPC charts. 

In order to tell whether a process or its outcome is improving, it is important to understand whether changes in data are due to actions taken, or simply due to random chance.  Making decisions without understanding the causes of variation can make things worse.

The two types of variation defined by Shewhart and Deming are:

Common causes - those causes that are inherent in the system over time, affect everyone working in the system, and affect all outcomes of the system.  These include random variation.

Special causes - those non-random causes that are not part of the system all the time, or do not affect everyone, but arise because of specific circumstances.

Statistical process control (SPC) charts (Shewhart control charts) are a good way of separating out common cause and special cause variation. 

(Run charts can signal whether the variation present is exhibiting random or non-random patterns, but not special or common cause.  In general – if there are non-random signals in your run charts, it is likely that special cause variation is present and you would react accordingly.)  

SPC charts are designed to minimise the chance of making either of two possible mistakes that can make things worse rather than better:

Mistake 1. Inappropriately reacting to common cause variation, as if it were due to a special cause. This "tampering" may exacerbate the variation.  If the system is stable (only common cause is present) but not working at an acceptable level, then fundamental change is needed – reacting to individual data points is not helpful.

Mistake 2. Inappropriately ignoring special cause variation, treating it as if it were common cause. To get stability and predictability, and therefore an ability to understand the effect of any deliberate changes, we first need to remove special causes.  

Process capability is about predicting what the future values of a measure are likely to be.  The predicted range can then be compared with the desired range to see if the process is capable of meeting its specifications.  SPC methods help us to work out process capability.  For information about how to calculate capability consult the Healthcare Data Guide.  

If a process is stable (only showing common cause variation) but not capable, we need to make a fundamental change if we wish to see an improvement. If we keep on doing what we've always done we'll keep on getting what we've always got.

If we also detect special causes we need to investigate what might be happening and learn from them. Not all special cause variation is bad. If we see a special cause which shows good performance we may wish to investigate, with a view to testing as a change to the process. 

If we seek to improve a process by making a change we are intending to introduce a special cause. However, if we see special causes that are causing poor performance we will want to remove them.

Measurement plan

A measurement plan sets out details for each measure proposed for an improvement project.

Run chart

A run chart is a line graph of data plotted over time. By collecting and charting data over time, you can find trends or patterns in the process.

SPC charts  

Statistical Process Control (SPC) Charts are simple graphical tools that enable process performance monitoring.

Funnel plot

A funnel plot is a chart that helps to understand variation within a system.

Scatter plot

A graph in which the values of two variables are plotted along two axes, the pattern of the resulting points revealing any correlation present.

Pareto chart

A Pareto Chart is a tool to help you understand your system.

Histogram 

A histogram is a plot that lets you discover, and show, the underlying frequency distribution (shape) of a set of continuous data.