Quality Improvement Zone

A Statistical process control (SPC) chart is a way of understanding how a process is performing over time. It helps you see whether changes in your data are part of normal day‑to‑day variation, or whether something meaningful has changed. 

SPC charts can help you answer questions such as: 

• is this process stable and predictable? 

• is the process capable of achieving the level of performance we want? 

An SPC chart includes: 

• a centre line, which shows the average (mean) of the data 

• control limits, which show the range where most data points are expected to fall if the process is stable 

The control limits are calculated by looking at the amount of variation in the data. If the process is stable, future data points are likely to stay within these limits. 

There are different types of SPC chart, depending on the type of data you are working with. The most common ones are: 

• XmR (or I) chart 
  Used for individual measurements. This is the most commonly used chart and is suitable for many situations. 

• P chart 
  Used for proportions or percentages. 

• C chart 
  Used for counts of events. 
  A U chart is used when counts are expressed as a rate. 

• X and S chart 
  Used for grouped measurement data where an average is calculated at each time point. 

In most improvement work, an XmR chart is a good starting point and will be suitable for many measures. 

If you would like to learn more about different SPC chart types, you can find further detail in The Health Care Data Guide by Provost and Murray. 

Use an SPC chart when you want to: 

• understand how your data is changing over time using a more powerful tool than a run chart 

• identify whether there is common cause or special cause variation (see understanding variation page) 

• see whether your process is stable and predictable, or unstable and unpredictable 

• understand whether your process is capable of meeting the level of performance you need 

SPC charts are especially useful for monitoring the impact of changes and understanding whether improvement is really happening.

Choose the right chart 

Start by identifying what type of data you have so you can select an appropriate chart. Using the correct chart matters because different charts make different assumptions in their calculations. 

If you are unsure, an XmR chart is usually a safe choice. 

Use enough data 

Avoid creating an SPC chart with too few data points. 

XmR charts use probability‑based formulas to calculate control limits, which require 20 data points. With fewer than 20, control limits are unreliable, and your interpretation of stability and capability would not be valid. 

If you have fewer data points, a run chart can be used instead.  

Create the chart 

SPC charts can be created using Excel. There are also commercial add‑ins, such as QI Charts or QI Macros. 

Interpret the chart 

There are a set of SPC rules that help you identify whether special cause variation may be present. These rules look for patterns in the data that suggest something has changed in the process. 

SPC chart rules

If any of the above rules are present, this suggests special cause variation, meaning something has changed in the process. If none of these rules are present, this suggests common cause variation. When an SPC chart shows common cause variation, it tells us that the process is stable and predictable. 

In a stable process, we can reasonably expect future data points to continue to fall within the control limits. The control limits therefore show what the process is currently capable of delivering, based on the variation in the data. 

Be aware that for some chart types, such as P charts, U charts, or X and S charts, the control limits may change over time if the denominators change. 

Use SPC charts to support improvement 

Use your SPC chart to understand whether the changes you are testing are leading to improvement. 

When creating your first SPC chart for a measure, especially if you are using historical data, calculate the mean and control limits using all available baseline data. 

During improvement work, it can be helpful to freeze and extend the baseline mean so changes are easier to detect. Only do this if there is no evidence of special cause variation in your baseline data. 

If you do not have clear baseline data from before your project started, use your judgement to decide whether it is appropriate to set a baseline using the first 20 data points, based on what you know about when change occurred. 

The Understanding variation page explains more about the difference between common cause and special cause variation and how to respond appropriately when using SPC charts.