Visualize the Triple Aim

Visualize the Triple Aim

In this section, we present ideas about how to visualize the Triple Aim and value—or improve health and/or patient experience while maintaining or decreasing costs—by continuing our hypertension quality improvement program case study. 

Overview

In this video, Dr. Boutwell explains the next step in using collected data to track Triple Aim measures over time or across sites.

Visualizing the Triple Aim provides powerful images and tools beyond traditional paper reports to analyze and demonstrate value. To introduce how to visualize the Triple Aim and analyze value, this Toolkit section walks you through images created for a hypothetical hypertension quality improvement program.

We show basic images created in Excel and provide Excel files so you can see how we created the basic images. We also show enhanced images created with a data visualization software called Tableau and provide Tableau files for you to interact with (download a free reader here). For each basic and enhanced image, we explain how to interpret the images.

The images are based on real community health center data and are examples intended to help you brainstorm ideas for visualizing the Triple Aim for your topics and measures of interest. We encourage you to take these images and start conversations about how best to measure, analyze, and visualize the Triple Aim for your own priorities.

Remember – as discussed in Step 3, the single measures in the case study below do not represent overall health, patient experience, and cost for all patients. The single measures used for health, patient experience and cost used in the case study reflect progress in a hypertension quality improvement program. The terms “health”, “patient experience” and “cost” could easily signify different measures that apply to different topics or populations.

Case study part 4: Hypertension quality improvement program

The following is a continuation of the case study begun in Step 2. Next, Part 4 of the case study will demonstrate how you can view health, patient experience and cost data to assess value.

To get ready for value based payment, you decided to try out measuring and analyzing value for your hypertension quality improvement program (Step 2). You have chosen measures to demonstrate value (Step 3), and you collected new cost and patient experience data (Step 4).

You now have data for your quality improvement program participants across all nine clinic sites in your health center organization:

  • Health: Percent of patients with hypertension with blood pressure levels under control (<140/90 for patients aged 18-59 or <150/90 for patients aged 60-85).
  • Patient Experience: Average patient satisfaction score (a higher score is higher satisfaction).
  • Cost: Total expenditures per patient.

By combining your community health center’s data on hypertension control and patient satisfaction with health plan data on total cost of care, you generate several visuals to demonstrate value. These visuals allow you to see the hypertension quality improvement program’s progress in health, patient experience, and cost across all clinic sites; how well sites are maintaining or increasing value; how sites compare with each other on value, and much more.

Visualize the Triple Aim across clinic sites

Figures 1a and 1b allow you to assess the quality improvement program’s progress in health, patient experience, and cost across sites.

What the basic image (Figure 1a) shows

Figure 1a includes three Excel graphs: the health measure (top), the patient experience measure (middle), and the cost measure (bottom). Sites are ordered from highest health score to lowest health score. The image shows:

  • For health, site 5 has the highest health scores, and site 3 has the lowest health scores.
  • For patient experience, sites 5 and site 9 have the highest patient experience scores, and site 1 has the lowest patient experience scores.
  • For costs, site 5 has the highest costs, and site 3 has the lowest costs.

By comparing sites to one another for each individual measure, you may be able to take lessons from one site and apply them to other sites. For example, site 5 can share with other sites its unique strategies for controlling blood pressure among patients in the quality improvement program.

Figure 1a. Visualize the Triple Aim across clinic sites -- basic

Figure 1a

Limitations of the basic image (Figure 1a)

Although this image shows how sites are performing across each individual measure, it is difficult to glean how well the sites are doing on value (i.e., improving health or patient experience while maintaining or decreasing cost).

How the enhanced image (Figure 1b) shows value

The enhanced image (Figure 1b) shows health, experience, and cost by site and uses color coding to indicate whether a site’s score is better than average (blue), comparable to average, or worse than average (red).

The colors in the enhanced image help us to identify higher value sites. For example, we see that both sites 4 and 5 have better than average scores on health (blue) and better than average satisfaction scores (blue). However, site 4 has lower costs than site 5 and may prove to be the higher value site.

Figure 1b. Visualize the Triple Aim across clinic sites -- enhanced

Figure 1b

Note: Different mark sizes represent sites with different patient volumes.

The bottom line

We can compare two or more sites on value, and sites can learn from one another. For example, what can site 5 learn from site 4 to bring total costs down without changes to hypertension control or patient satisfaction? Conversely, what might site 4 learn from site 5 about hypertension control and improving patient satisfaction?

Group clinic sites according to value

Figures 2a and 2b are scatter plots; each site is a dot plotted by the health measure along the x-axis and the cost measure along the y-axis. The scatterplot is divided into four quadrants.

  • Quadrant 1 contains sites with high health scores and low costs (highest value quadrant)
  • Quadrant 2 contains sites with high health scores and high costs
  • Quadrant 3 contains sites with low health scores and high costs (lowest value quadrant)
  • Quadrant 4 contains sites with low health scores and low costs

What the basic image (Figure 2a) shows

We can compare two quadrants on value so sites can learn from one another.

  • Sites in quadrant 2 (sites 5, 6, and 7) can learn from sites in quadrant 1 (sites 1 and 4) about how to bring costs down without decreasing health.
  • Sites in quadrant 3 (sites 2 and 8) can learn from sites in quadrant 2 about how to improve health while maintaining cost.
  • Sites in quadrant 4 (sites 3 and 9) can learn from sites in quadrant 1 about how to improve health while maintaining cost.

Figure 2a. Group clinic sites according to value -- basic

Figure 2a

Limitations of the basic image (Figure 2a)

Although this image allows you to see how your community health center is performing on value based on health and cost, it does not include patient experience at all.

What the enhanced image (Figure 2b) shows

The enhanced image (Figure 2b) is also a scatter plot with quadrants that allow you to group clinic sites by health and cost. In addition, patient experience is shown for each site using red-green color coding, where green indicates higher patient experience scores and red indicates lower patient experience scores.

Figure 2b. Group clinic sites according to value -- enhanced

Figure 2b

Figure note: Different mark sizes represent sites with different patient volumes

The bottom line

We can compare two quadrants on value so sites can learn from one another. In addition, we can monitor patient experience. For example, sites in quadrant 2 (sites 5, 6, and 7) may learn from sites in quadrant 1 (sites 1 and 4) about how to bring costs down without declines in the percent of people with controlled hypertension or patient satisfaction. Or, sites within the same quadrant (e.g., site 1 and site 4) may learn to improve patient satisfaction while maintaining hypertension control and costs.

Group providers according to value

We can measure the Triple Aim across different providers within a site to compare value at the provider level.

What the basic image (Figure 3a) shows

In this scatterplot (Figure 3a), each provider is represented by a data mark. Health is on the y-axis, and cost is on the x-axis. This example illustrates that the location of the high-value quadrant changes depending on how the measures are plotted. Here, the high-value quadrant is the upper left quadrant (better health, lower cost). Similar to Figure 2a, we can compare two quadrants on value so providers can learn from one another.

Figure 3a. Group providers according to value -- basic

Figure 3a

Limitations of the basic image (Figure 3a)

Although this image allows you to see how providers are performing on value based on health and cost, it does not include patient experience at all.

What the enhanced image (Figure 3b) shows

The enhanced image (Figure 3b) is also a scatter plot with quadrants that group providers by health and cost. The data mark shape indicates the clinic site where that provider primarily practices. Patient experience is shown for each site (ranging from red to blue). The image shows:

  • Providers 6, 22, and 36 in quadrant 1 (upper left) have high health scores and low costs
  • Providers 2, 5, 12, 13, 16, 17, 26, 27, 33, and 39 in quadrant 2 (upper right) have high health scores and high costs
  • Providers 9, 11, 18, 20, 24, 37, and 38 in quadrant 3 (lower right) have low health scores and high costs
  • Providers 3, 15, 19, 23, 31, and 35 in quadrant 4 (lower left) have low health scores and low costs

Figure 3b. Group providers according to value -- enhanced

Figure 3b

The bottom line

You can identify providers delivering high-value care to patients and compare two quadrants on value so providers within and across sites can learn from one another.

Visualize the Triple Aim across community health centers within your region

In addition to the visuals using community health center and health plan data, it is also possible to create visuals using publicly available data. As an example, we developed views using the publically available UDS and Behavioral Risk Factor Surveillance System (BRFSS) data. 

These data sources allow you to look at health, patient experience, and cost in a regional context. They also enable you to develop ideas about how to use public data in the interim while you develop your own data infrastructure. Finally, publicly available data sources can help fill data gaps that you can’t close with your own community health center data.

We selected the following measures from UDS and BRFSS to visualize the Triple Aim:

  • Health: Percent of hypertensive patients with controlled blood pressure, by community health center, 2011-2014 UDS
  • Patient Experience: Percent of the population with a regular source of primary care, by county, 2014 BRFSS
  • Cost: Total cost per patient using operational data, by community health center, 2011-2014 UDS 

Because there are limitations to UDS data, we suggest caution when interpreting these images. Despite data limitations, these images are a good starting place for generating ideas on how to use publicly available data to visualize the Triple Aim and value.

What the enhanced image (Figure 4) shows

First, users select counties of interest on the left. The image shows whether community health centers in the counties of interest have scores that are better than average (green), comparable to average (gray), or worse than average (red).

Then, users select comparison counties on the right, which are used to calculate the blue diamond mark estimates.

The state average reference line is shown for each measure using data reported by all federally qualified health centers in California reporting to the UDS.

The bottom line

You can compare community health centers in one county versus another county (or several counties simultaneously) on value.

Figure 4. Visualize the Triple Aim across health centers within your geographic region – publicly available data

Figure 4

Note: For our example image, we have selected San Francisco and San Mateo counties on the left and Alameda, Contra Costa, Marin, Napa, Santa Clara, and Santa Cruz counties on the right for comparison.

What the enhanced image (Figure 5) shows

You can assess community health center performance relative to other community health centers in the county or neighboring counties or relative to other community health centers with similar patient volumes. There are three main parts to the image.

The scatterplots on the right allow you to compare community health centers on average patient costs and blood pressure control over time.

The map on the left depicts community health centers within counties.

  • Counties are color coded for primary care access: red means a smaller percent of the population has primary care access, and blue means a larger percent of the population has primary care access.
  • Community health centers are the dots, color coded for the health measure: yellow means a lower percent of patients with controlled blood pressure and green means a higher percent of patients with controlled blood pressure. Dot size indicates the size of the community health center’s patient population.
  • The table at the bottom shows the health, experience, and cost score for the community health centers in the selected county and with the selected patient volumes.

The bottom line

Figure 5 highlights the variation that can exist among community health centers in the same county. It also highlights community health centers with strong performance on health despite the challenging context of relatively low primary care access in the county.

Figure 5. Visualize the Triple Aim across health centers within your geographic region -- mapping publicly available data

Figure 5

Note: On the far right panel, users can select counties of interest and health centers’ patient volumes.

Tailor images to your own needs

These images all demonstrate that it is feasible to measure and visualize the Triple Aim with your own and publicly available data. These ideas are intended to help you brainstorm ways to tailor images for your own needs.

For example, we show images using sites, providers, and community health centers within regions. Instead, you could focus on different quality improvement programs that use similar measures.

Or, instead of hypertension control, your community health center may be interested in—or excel in—another clinical area like diabetes management. You can create images to show value for different clinical areas that are relevant for your patient population or that are your community health center priorities.

Instead of using the colors and patterns that we chose, you could use any colors or shapes that appeal to you and best communicate results to your audiences. 

Avoid common pitfalls

Creating quality visuals for the Triple Aim can be a challenging task.  Here are some important considerations for both newcomers and seasoned visualization veterans.

Association is not causation. This common statistical theme applies here. For example, our site-by-site comparisons only illustrate that some sites have higher scores in certain areas. The reasons for those differences could be related to the region, organization (e.g., local policy, leadership, patient and family engagement efforts), patient (e.g., age, baseline health status), or many other factors.

It is important to remember not to imply causal relationships among cost, patient experience, and health when viewing all three aims together. Combine these views with clinical insight and organizational experience to stimulate discussions about improvement and goal setting.

The quality of the visuals depends on data quality. As data quality evolves, the quality of the visuals will change too. Thus, it is important to make sure your stakeholders and audiences can interpret the visuals and understand their limitations.

Make sure that benchmarks are calculated correctly. For example, the lines representing averages calculated in Figure 1b are calculated from original respondent level data. Sometimes only aggregated data are available (i.e. measures are given for each site in a summary table).  If Figure 1b had been constructed from aggregated data, then the line representing the average may not account for the patient volume at each clinic and would not represent the real average across all respondents. 

Explore how others visualize their data. Take time to explore how others analyze and visualize health, patient experience, and cost data. Sometimes data visualization blogs (such as the one here) can provide good insights about what not to do in data visualization.  

Tailor visuals to specific audiences. Some may want many visuals that show fewer features per visual (e.g., one graph at a time). Others may want fewer visuals that highlight multiple features at once (e.g., a visual that shows multiple graphs on a single dashboard). Conveying information is not one size fits all.

Once you have created these visualizations, take action. Operationalize the information and use the visuals to make improvements and to communicate with stakeholder audiences.

Resources

We provide Excel files so you can see how we created the basic images. We also provide Tableau files for you to interact with (download a free reader here).

**These downloads work best in Google Chrome and Firefox browsers. Please right click on the file you would like to download and click “Save as” in order to access the file.