Can you really improve quality and reduce cost?

Posted Aug 22, 2018

Ben Higgins, Lead, Health System Analytics

From my place in the world, it seems that any change to improve the quality of healthcare is going to increase cost. Can you help me understand how improvements can maintain, or better yet, lower the cost of healthcare delivery?

Geralyn, HQCA Patient and Family Advisory Committee Member

Hi Geralyn,

What a great question, and one with an answer I am very passionate about.

In order to answer this for you, it is important to understand variation, since reducing variation is one way to improve quality and lower cost.

What is variation? Put simply, it can be a different form or version of something. Or, it could be a change or slight difference we can observe or measure. For this conversation, we are going to talk about variation we can observe or measure. Everything around us that we observe or measure, varies. Variation is expected and some variation in healthcare is desirable, even essential, since each patient is different and should be cared for uniquely.

When healthcare decision-makers are trying to decide what to focus their attention and effort on, how do they know what variation is worth investigating?

To do this effectively, they need to understand the difference between common and special causes of variation.

Before we get into some healthcare examples, let me offer an analogy to help understand these ideas. Let’s think about variation and your commute to work.

You know about how long it takes for you to get to work each day. Let’s say it’s usually between 30-35 minutes to get from home to your work site. There is no way you can pinpoint the exact amount of time it will take because of common causes of variation (public transit arrival time, traffic light timing, a short distraction with your child while getting ready, etc.). You just expect that you will arrive sometime in that 30-35 minute range and accept that variation.

But then, one day, it takes 55 minutes to get to work. There was a lot more traffic on the roads and it turns out there was a broken traffic light at a huge intersection that caused a major back-up. This longer commute is well outside of that “typical” range and the traffic light problem is an example of what we would call a special cause of variation.

It is SPECIAL CAUSES of variation that healthcare decision-makers should be investigating.

Let’s look at an example from here in Alberta’s hospitals, from the HQCA’s FOCUS website. Below is a graph showing the length of patient hospital stay compared to the Canadian average for the Queen Elizabeth II Hospital in Grande Prairie. When you take a quick glance at this graph, does anything stand out as “unusual”?

Did you say the data points from October 2016 and January 2017? These look like really sharp spikes, right?

However, guess what? The lines for the upper and lower control limits actually show us the range for normal, common cause variation. Control limits are a statistical calculation that tells us what the boundaries are for when variation is in control (inside the limits) or out of control (outside the limits).

As we can see here, while those two spikes (October 2016 and January 2017) appear unusual, they are still inside the limits and would be actually considered a normal and expected amount of variation. So, while it might be tempting to make changes to processes after observing these spikes, we would not recommend investigating this variation further. Doing so would not likely improve the length of hospital stay (because the variation observed is within the expected limits for this system) and would lead the team in an unhelpful direction, wasting time, money, and other resources (e.g. people, supplies, equipment, etc.).

Special cause variation is where we recommend decision-makers focus their efforts on improving quality to make the most positive impact. Here are three of the most useful things we look for in the data before investigating special cause:

1. A single point outside of the control limits

2. Eight consecutive points above or below the mean line

3. Six consecutive points increasing or decreasing

When you see any of these three things happen with data, it’s worth looking into further.

To come back to your question, this example of the length of stay in the hospital is also a great one to talk about the relationship between quality and cost. When these hospitals see special cause variation for the better (what we see in both the examples for the Peter Lougheed Centre and the Grey Nuns Community Hospital above), they have an opportunity to investigate further, hopefully find out why there was an improvement, and then take steps to keep achieving the positive results (by sustaining, replicating, or scaling what caused the change). In this example, if these hospitals decrease their hospital occupancy rate (how many beds are in use) by shortening the length of stay, this could reduce the overall cost to the hospital and, ultimately, the system.

On the flip side of that coin, like in the early part of the Rockyview General Hospital example, when there is a special cause in the undesirable direction, they have an opportunity to investigate and correct the situation (by stopping or avoiding what happened). As you can see from their data, either improvements were made or circumstances shifted to where they have a second string of at least eight consecutive points below the mean. This sustained improvement offers another opportunity for investigation to learn from this and try to ensure the positive trend continues.

In fact, the Rockyview General Hospital data shows that this improvement was sustained for long enough that we could actually consider this to be a new normal for operations at this hospital. Considering this is the new normal, we recalculate the control limits to reflect this:

So, what does this mean? Well, let’s break it down in a way similar to how a quality improvement team at the hospital might look at this. The measure for “length of patient hospital stay compared to the Canadian average” is made up of two pieces:

  1. The length of hospital stay for patients at Rockyview General Hospital and
  2. The expected (average) length of hospital stay for similar patients in Canada.

When the measure is broken down into these pieces, we discover that the length of hospital stay for patients at Rockyview General Hospital has decreased slightly. Over this same period of time, the average for similar patients in Canada has increased slightly. Together, these changes lead to the shift you see at the Rockyview General Hospital.

From a cost perspective, it is likely that the slight decrease in length of hospital stay for patients at Rockyview General Hospital resulted in a lower average cost per patient (because they would have spent less time in the hospital) during the period of time from March 2017 to March 2018. And, if the hospital also saw a reduced occupancy rate during that same period, the overall cost to the hospital may have gone down as well.

So, as I mentioned before, it is very easy to get caught up in the “hot” issue of the day and start trying to fix or change something that was unusual and seemed like a big deal. This is why looking more closely at variation over time with these simple rules of thumb can lead to more meaningful quality improvement and uncover potential cost savings for Alberta’s health system. Those cost savings can be realized by focusing staff on chasing more variation where there is a real opportunity for improvement, not just where we have a “hunch” change is needed.

Hopefully this answered your question and helps you see why we find data so interesting here at the HQCA. If you have any other questions for us, please just let us know. Also, if you would like to read more about variation and get a bit more technical, I recommend checking out these great resources from the National Health Service (NHS) in England: here and here.

All data presented here is available on the HQCA’s FOCUS on Emergency Departments website, which is updated quarterly. If you would like to sign-up to receive notifications when new data is available, please click here.

For HQCA Matters updates, follow @HQCA on Twitter.

HQCA Matters is published monthly and presents HQCA representative perspectives on topics or issues relevant to healthcare in Alberta.