in Health Care, Tools & Methods, Variation, Videos

By David M. Williams, PhD

Harvardx course Practical Improvement Science in Health Care with the Institute for Healthcare Improvement.

Understanding variation is one of the key lenses of our theory of improvement. As improvers, we aim to reduce the variation of process and create a process that achieves a certain level of reliability. One tool that can help us understand variation is a histogram. A histogram is a chart that can help us understand the distribution of continuous data.

So for example, let’s imagine you work in a federally qualified health center, and you are tasked to improve the patient cycle time in a clinic. The patient’s total time from appointment start time to completion is important to the patient experience and important to the clinic’s efficiency. So using a data collection form, you collected this data for the cycle time of patients in a single day. So imagine here we’ve got our minutes going along the left hand column, and we’ve got the number of patients that had a cycle time within that time interval on the right hand column. And so what we’re going to do is draw the actual histogram. And a histogram looks very much like a bar chart, but the bars are organized by the time interval.

So let’s draw our bar chart. And again, we’re working on cycle time. And along the x-axis we’re going to put in our times here. So we’ve got 10, 20, 30, 40, 50, 60. And then I just lumped anything that’s over 60, because for the most part, we don’t want that. And then on this side, so this is — let me put my time here. It’s good to label your axis. On this side, I’m going to put my number of patients. So I’ll go 1, 2, 3, 4, 5, 6, 7, 8. And again, this is the y-axis. Now I want to label that as well, so people who can see this know what it is that we’re tracking. So now I can come along here with my data, and we just create — we go with each time interval, and we create a bar that corresponds with the number of patients. So for patients that waited only 10 minutes, those lucky few, we’re going to come along, and we will do 2 patients. We’ll make a bar. Let’s just color it in, so it’s a little bit easier to look at. Great. So now as you look at your cycle time data, you can look for the number of patients that had cycle times in each of the various time intervals.

Once you’ve plugged in all the data, you’ll see the distribution of the data and better understand the variation that exists within your process. So two things to look for at the start is, does this histogram show a distribution that appears symmetrical? So if you look at this chart, for example, it looks kind of like a mountain. So it’s pretty symmetrical. Now, it doesn’t mean that you like the variation that exists, and the times that are here, or that you’re happy with the amount of time. But this kind of shows that you’ve got a process that doesn’t have any outliers. And now you can work on trying to figure out how to reduce the variation and reduce the time to whatever’s appropriate for what you’re trying to improve.

Now in some cases, you might encounter an example where, for example, here we came along, and let’s imagine this one’s just off the charts. And it’s actually 10, or 12, minutes, or even higher. And you notice that this particular interval has an outlier or something that just jumps out at you. And it says there’s something else going on here that’s an area that we need to figure out. And so that can be a helpful thing to point out when you’re looking at your histogram to identify if there are some special places that may be in a certain time interval or after a certain point. There’s something unusual that’s outside the symmetry of the process.

If this was helpful, share and include me @DaveWilliamsATX. Sign up here to receive a monthly email from me that includes all my blog posts and other Improvement Science resources I think you’d appreciate.