In a perfect world, choosing the correct chart would be very simple. One would follow a process of deciding on what questions need to be answered. Based on that list of questions, one would decide on the data that needs to be gathered. From those two exercises, the type of chart one would use should be determined.

But, we are often in a situation where we have already gathered the data and we need to figure out which questions we can answer with that data, and then which chart we can use to best present the data.

Some of the questions one might ask would be:

Now, let's look at data. There are two general types of data: Attribute and Continuous.

Attribute Data are count data. We are counting the number of times an outcome or event has occurred. For count data, we are working with integers. When we speak of events, we may be talking about such things as Patients being Admitted, Patients receiving Adult Cessation Advice, Falls, or Doses of Medication Administered. These are all things we can count.

If we are working with Attribute Data, we can categorize it as either Percentage Data or Rate Data. Percentage Data are those data where an event results in one of two possible outcomes. Many of our compliance indicators are of this type. For a particular event, we were either in compliance or we were not. A primary example of this would be the Joint Commission Core Measures. AMI 1 measures the percentage of AMI patients admitted who were given aspirin at arrival. The event is the AMI patient was Admitted and was eligible for aspirin. The outcome was either the patient received aspirin or they did not. These data are best charted with a p chart.

You can use the following guidelines to determine if you are working with Percentage Data and would use a p chart:

An example of the p chart is found at AMI 1 Aspirin at Arrival. You can also view more information about the p chart at: p chart. In these and the following examples, clicking on the indicator link will exhibit an Indicator chart from Statit's piMD, while clicking on the chart link will exhibit a window that discusses that particular chart type. It will also provide links to other charts that might be applicable to that data type.

The second type of Attribute Data is Rate Data. Rate Data is best charted using a u chart.

There are a couple of guidelines you can use to help you decide if the Attribute Data you have are Rate Data or not. First, can you count both occurrences and non-occurrences of events? The primary example of this type is Patient Falls. You can certainly count the number of Patient Falls, but can you count the number of times a patient did not fall?

With Rate Data, you are counting some occurrence of an event over some "area of opportunity". While it would be nice to be able to measure the opportunities a patient had to fall, this is not practical. So we would typically look for a surrogate measure that would give us some indication of the magnitude of this area of opportunity. Patient Days is just such a surrogate measure.

We often measure Falls per some multiple of Patient Days. This might be Falls per 1000 Patient Days or Falls per 100 Patient Days. If the latter, we need to be careful to not talk about this in percentage language. We are measuring the rate of patients who fell per 100 patient days.

Thinking about this data a little more, we could see that we would have a bit of a conflict with another Percentage Data guideline, leading us away from that data type. A patient could conceivably fall more than once in a day. With Percentage Data, the numerator could not be higher than the denominator. However, in one patient day, a patient could fall more than once. This leads us to the second Rate Data situation.

A second Rate Data scenario, then, is where we have an event or "area of opportunity" that results in more than one outcome. A good example of this scenario is Medication Errors. One dose of medication could conceivably have more than one error, such as time, patient, dosage, medication etc. Obviously, it is conceivable that our numerator (Med Errors) could be greater than our denominator (doses of medication).

Some guidelines to determine if you are working with Rate Data:

Examples of these two data types are referenced here: Falls per 1000 Patient Days. More information about the u chart can be found at: u chart.

Continuous Data are data that usually come from a continuum, where the discreteness is often a matter of choice. For example, we could measure time in days or days and hours or hours and minutes and seconds. Or you might have a case mix index with two digits past the decimal, where the discreteness is part of the measure.

Continuous Data is usually categorized as either Single Value per Period or Multiple Values per Period. Single Value per Period data are analyzed using an I chart. You have one measurement for a period. Some examples might by Number of Admissions, Case Mix Index, or Budget Variance.

Your data may consist of a data value by month, but you would like to view this with a quarterly period. In this case, you would have three values per period. With Statit piMD Single Value per Period data type, the three monthly values would be added to get the quarterly value on the I chart.

An example is Case Mix Index where we have a monthly case mix index. More information about the I chart can be found at I chart.

The other Continuous Data type is Multiple Values per Period. Relating back to the questions, these are data that you are interested in seeing the average or median value for a period. For these data, the X Bar Chart for the average or Median Chart for the median are appropriate. Average Length of Stay is a good example of the Multiple Values per Period. An example of this chart is Average Length of Stay, with more information on the Multiple Values per Period charts at Xbar chart. In this chart, several individual lengths of stay are average by period.

One final data type is the Rare Events type which is analyzed using the g chart. Rare Events are those events that happen so rarely that other chart types are inappropriate because there are not enough events to fall within the guidelines of the charts. With Rare Events data, the measurement of interest is the time between events. In Statit piMD's rare events analysis, you need to gather the date and time that each event occurred. The g chart then plots the time between events. In this case, exceeding the upper control limit is a good thing because we want the time between events to increase. Information about g charts and the Rare Events data type are at g chart.

If you have been gathering data for some time, you will need to understand the data to decide on the type of chart you can use to analyze the process. If the data can answer the questions given above, the questions then lead you to the appropriate chart to use.

If you would like additional information, please call our Support staff at (541) 752-4100 or send email to .