Why It Matters
"In improvement, it is critical to understand that every process has inherent variation that we want to understand."
Processing ...

2 Tools to Understand Variation in Your Improvement Journey

By Zewdie Mulissa | Friday, January 4, 2019

Two Tools to Understand Variation in Your Improvement Journey

Health care organizations at all different levels, from local and regional to national, use data to understand their performance — though they do not always do so effectively.

Data helps answer questions that are fundamental for improvement. In an improvement journey, you need to understand:

  • Where is your performance at the start?
  • What are the effects of your changes?
  • Are you are holding the gains over time?

Moreover, data provides a common reference point to have good conversations about the improvement work all of us are doing and the effect it is having.

There are many ways to present and analyze data, such as summary statistics, tables, statistical tests, and graphs. For improvement efforts, visual displays of data are often the best approach to learn from variation in data. Images are usually easy and quick to prepare, and they make it possible to access nearly all kinds of potential insight from the data. They lead to a systematic view of problems and opportunities. Importantly, visual displays help you identify small but important details that might be missed in other types of analysis. For example, if you are tracking the length of your commute, a summary statistic such as the average (mean) for the month will tell you the average commute time was higher than usual last month. A visual display might show you that this is because on a single day, it took you three times as long as usual – perhaps because your car broke down – while every other day closely followed your usual routine.

The basic types of visual data displays most closely associated with statistical process control are plots showing data over time (e.g., run charts and control charts), plots showing distributions of data (e.g., pareto charts and histograms), and plots showing relationships between different characteristics (e.g., scatter plots). Here we will dive into run and control charts.

Types of variation

Quality improvement requires using data to learn and to predict future performance. In improvement, it is critical to understand that every process has inherent variation that we want to understand. There are two types:

Intended variation is an important part of effective, patient-centered health care. It is also called purposeful, planned, guided or considered variation.

Example: A physician purposely prescribes different doses of a drug to a child and an adult.

Unintended variation is due to changes introduced in to health care process that are not purposeful, planned or guided. They usually create inefficiencies, waste, ineffective care, errors and injuries in our health care system and reducing them usually results in improved outcomes and lower costs.

Example: Without realizing it, a physician prescribes pain medication to one person and does not prescribe it to a second person with the same condition due to implicit bias (subconscious stereotyping) about who needs pain relief.

Variation in a quality measure may result from common causes — expected causes that are inherent in the system. It may also derive from special causes — unnatural causes that are not part of the system but arise due to specific circumstances.

In a stable system, only common causes affect the outcomes. Variation is predictable within statistically established limits. By contrast, in an unstable system, outcomes are affected by both common causes and special causes. In this case, variation is unpredictable.

Plotting data over time

In a quality improvement journey, we use well annotated run charts and control charts to learn from variations in data.

A run chart is a graph of data over time. It helps you determine whether the changes you make are leading to improvement.

Skilled birth run chart

The above run chart depicts skilled birth attendance at a health facility from July 2015 to November 2017. By looking at the chart, we see the changes introduced in August 2016 and March 2017 lead to a improvements in skilled birth attendance, circled in red (a steady upward trend and a huge spike, respectively).

A control chart and helps you distinguish between special and common causes of variation. It includes an upper control limit (UCL) and a lower control limit (LCL) marked above and below the average line. Variation within these limits is expected and attributed to common causes; variation beyond these limits suggests special causes. Use control charts to identify early signs of success in an improvement project and to monitor a process to ensure it is holding the gains from a quality improvement effort.

Syphilis screening control chart

The above control chart depicts syphilis screening in antenatal care from July 2015 to 2018. Again, by looking at the chart we see the changes introduced in October 2016 lead to an improvement in syphilis screening. The rates rise above the upper control limit and the mean shifts upward. Moreover, the new level of performance was sustained after April 2017. This chart demonstrates holding the gains made during improvement.

For more about using run charts, control charts, and other data displays, as well as templates to get you started, download the IHI Quality Improvement Essentials Toolkit.

Zewdie Mulissa is a Senior Monitoring and Evaluation Officer at the Institute for Healthcare Improvement.

first last

Average Content Rating
(0 user)
Please login to rate or comment on this content.
User Comments


© 2022 Institute for Healthcare Improvement. All rights reserved.