At NICHQ, we’re obsessed with run charts. Why? Because they are easy to construct and interpret and an elegant way to display data that makes performance visually legible. At a quick glance, you can determine whether the changes you make result in improvement, and even further, if those improvements are being sustained. (We primarily use run charts to answer the second question in the Model for Improvement: “How will we know that change is an improvement?”)
When you’re constructing your run chart (also sometimes called a trend chart or time series), usually the horizontal axis is a time scale and the vertical axis is the measure or characteristic that you are studying. Once you have collected the data and plotted the data points, there are some key rules and tips that will not only better help you understand the effect of the changes that you are testing, they will also make your run chart more legible and compelling to others—say, members of your team who could use some encouragement to buy in to your quality improvement initiative.
Seems pretty straightforward, right? The hard part is testing the change; in comparison, understanding your run chart is a breeze.
Here are some tips and rules for creating and interpreting run charts.
Creating a Run Chart
- To get started, calculate and place the median of the data on the run chart (the median being the number in the middle of the data set when the numbers are reordered from highest to lowest)
- Next, identify the runs. A run is defined as “one or more consecutive data points on the same side of the median, ignoring points exactly on the median.” Runs will be useful in assessing patterns, see below, rule three.
- Add the goal line, which will help you and others visually process progress.
- Annotate the run chart by labeling points in time when you tested changes—write those changes right into the chart.
- If your vertical axis is displaying a percent, add the "N” somewhere on the chart to give your viewer or reader a sense of the numbers being assessed. For example, 20 percent of five is very different than 20 percent of 500.
- If it’s not clear to the casual observer, add an arrow that signifies the direction of goodness. For example, when the vertical axis is minutes from triage to treatment in the emergency room, lower numbers are better.
Interpreting a Run Chart
- A shift in the process is indicated by six or more consecutive points above or below the median.
- A trend is indicated by five or more consecutive points all increasing or decreasing.
- Too many or two few runs indicate a nonrandom pattern.
- An astronomical data point is a pretty good signal of a nonrandom pattern.
These are just basic guidelines for using and interpreting run charts, and how to understand if the changes you are making result in improvement. When you want to understand if the variation in your data is due to a common cause or a special cause, that is the job for control charts. But that is a different story.