Run Chart

A run chart is a visual tool used to display the variations in a process over time. Run charts can be used to plot any type of data. Examples of process data include the quantity of a specific drug administered in a hospital by shift, the number of patient complaints per month, and the number of walk-in patients in a clinic per week. The major benefit of a run chart is its simplicity. It requires minimal statistical knowledge and can easily be understood by all members of a quality improvement team. The sample run chart in Figure 1 illustrates the number of patient complaints per month over a period of 20 consecutive months.

A run chart helps one determine whether the variation in a specific process has a common or a special cause. Common cause variation is inherent in every process and is a part of the regular rhythm of a process. Processes that demonstrate common cause variation are stable, predictable, and “in control.” Special cause variation, in contrast, is due to irregular or unnatural causes that often occur unpredictably. When special cause variation exists, the process is “out of control,” or unstable. To improve a process, the special causes must first be identified and eliminated. Only after this has occurred should one consider changing or improving the process. Attempting to change processes that contain special causes is futile and a waste of resources. [Page 493]

Figure 1 Sample Run Chart for Patient Complaints by Month

To construct a run chart, the following actions must occur: (a) Collect the appropriate amount of data. (b) Draw a horizontal axis and label it with the unit of time, such as hours, days, weeks, or the sequence of occurrence. (c) Draw a vertical axis and label it with the characteristic, or variable, that is to be plotted. (d) Plot each data point in the order that it occurred. (e) Connect the points on the graph with a solid line. (f) Determine the median for the data, plot it for each data point, and connect the points with a straight line.(g) Count the number of runs on the chart. A “run” is one or more consecutive data points on the same side of the median not counting the data points on the median. (h) Compare the number of runs to a pre-established statistical computation. The statistical computation indicates the lower and upper run limit according to a specified number of data points.

If there are too few or too many runs in the data, a special cause variation exists. There are several other types of runs to include “shifts,” “trends,” and “zigzags,” but each shares the common element of exhibiting a distinctive pattern. Again, if the pattern exceeds the established statistical limit for the number of data points, special cause variation exists. One must attempt to mitigate or eliminate the special cause(s) so that process stability can be achieved. Once stable, the goal is to continuously improve the process by reducing variation.