Data visualization is increasingly important in the communication of quality improvement data, and nearly everyone in the field uses it to some extent. But effective use of data visualization with graphics, dashboards, and other tools requires an understanding of why this approach works and how to optimize its effect.

In the simplest terms, data visualization is taking rows and columns of text and numbers and making it consumable in a picture, explains Greg Horne, global principal for healthcare with SAS, a software analytics company based in Cary, NC.

“Visualization technology demystifies data and presents a story that can be used to improve outcomes,” Horne says. “We all know the mantra that a picture is worth a thousand words, and this holds very true with data visualization — creating a picture to encapsulate many thousands of data points to facilitate better decision-making.”

The most important part of any data strategy is making sure the data can be used by the people who can benefit, he says. In healthcare, this can be as simple as creating a “traffic light” system where clinical staff look for what is red, representing a problem area, he says.

“At the operations level, administrators can review a diagram that identifies care units not meeting goals for outcomes and readmissions. Or they can flip the data to learn about the high-performing units that can teach the others,” Horne explains. “For patient portals that aid in provider choice, data visualization can present clear knowledge to inform decisions.”

Visualization can be taken further by adding a narrative, telling a story with the data and graphics rather than presenting a static representation. For instance, this might be used to tell the story of how a unit addressed hospital-acquired infections by showing where the unit started, what changes were implemented, how the data changed over time, and where the unit currently stands.

All operational departments at University of Pittsburgh Medical Center (UPMC) use data visualization to some extent, and the quality professionals consider it a primary tool for conveying information, says Johanna Bellon, PhD, MS, CFA, senior director of quality analytics and performance at the UPMC Wolff Center for Quality, Safety, and Innovation. The CEO receives a report every morning that uses data visualization for key metrics in the hospital, with others receiving visual representations of the metrics appropriate for their work, down to the floor nurses.

“The delivery and format of those is incredibly important in spurring people to take the actions we hope they will take based on the data,” Bellon says. “Our team focuses on quality and quality metrics, but I work with other data teams like the electronic health record, finance, [and] clinical analytics to help get the best data to users across the enterprise. Visualization is a very important way that we get that data across in a meaningful way.”

Bellon’s department uses an enterprise data warehouse, a system that integrates data from multiple sources and compiles it for analysis, along with a software system that facilitates data analytics and presentations. With vast amounts of data available, the quality department must determine how to present that information in the most useful way possible, and that means converting numbers into something visual and more easily understood.

When preparing data for users within the hospital, Bellon and her colleagues keep in mind several key requirements. The first is what the user needs to know.

“We want to be sure we’re answering their questions and giving them the information they need. If they’re trying to reduce admissions, we want to make sure that information is being presented front and center in our data visualization, and that it’s easy to access,” Bellon says. “After that, we want to be able to drill into the whys. If they are seeing in a chart that they are having an increase in readmissions, part of QI’s task is to delve into the root causes. We want to have drill-through capability in our data visualization so they can access process measures and outcome measures related to that main question.”

For instance, the initial graphic showing an increase in readmissions might lead to another visualization illustrating an increase in patient volume or severity of illness, or a turnover in staff.

The third key ingredient is providing an intuitive information hierarchy. That means Bellon wants UPMC users to see the data visualization and understand it readily.

“We have filters and data specification elements on one side and the data visualization on the other side, with graphs and other elements, making it easy to see the elements they have specified as important for them and the outcomes,” Bellon says.

Bellon’s department also tries to make the data visualization a self-service resource. A few years ago, data visualization was usually flat files, just graphical representations of data with no drill-through capability, she says. But now, users are getting to be much more sophisticated with the use of data.

“We are trying to deliver to their doorstep a data visualization that puts all the answers at their fingertips so they can drill through and not have to keep coming back to us asking for more and more reports that get down to the data they really need to act,” Bellon says.

Documentation also is essential in the data visualization product, she says. It is not the most interesting part of the product, but it must be available to the user to show how the data were calculated. The click-through graphics allow the user to see how the numerators and denominators were determined to support an informed interpretation of the data, she explains.

Users Can Build Own Reports

All UPMC executives, nurse leaders, and frontline staff can access a readmissions dashboard that follows those principles in presenting data visually, Bellon says. The dashboard shows data for the overall hospital, but also broken down by unit, specialty, and other divisions.

“There is another feature called a report builder. If they want to answer questions they may have about why the data is showing something in particular, they can drag and drop data to prepare their own report to answer those questions,” Bellon says. “They can always contact us if they need more support, but it’s more efficient if they can prepare those reports themselves. We’ve had some departments implement new strategies such as risk stratification that resulted in reduced readmissions, so we’re hoping that this kind of data visualization supports that going forward.”

Data visualization also has been used to address sepsis at UPMC. Bellon’s department uses dashboards to show unit leaders and clinicians their volumes of patients with sepsis, along with sepsis mortality and readmissions. “The data visualization helps us provide the feedback loop to say you’re using a power plan over 80% of the time, you’re getting your lactates done in a certain number of patients, and we have seen a dramatic reduction in sepsis mortality across our hospital,” she says. “That’s exciting because it brings together quality improvement efforts with data visualization and feedback loops to support that work.”

Formalize Visualization Principles

To optimize the use of data visualization, Bellon urges close engagement with users. She also says it is important to formalize the key drivers of data visualization, such as answering the user’s questions and providing documentation, rather than assuming quality improvement professionals will just know to include those aspects.

“Often, data teams are overwhelmed with requests, and people will skip steps to try to push things out faster,” Bellon says. “We have a principle here that says ‘go slow to go fast,’ meaning that if we take the time to follow the important steps with data visualization, you’re going to get more bang for your buck. Even if you don’t have the greatest software tools and you’re working in Excel, you can set up the processing guidelines to ensure that you’re identifying the questions from your users and developing templates so that they can easily use the data you’re providing.”

Healthcare is catching up to the ways other industries have used data visualization, Bellon says. The next frontier may be using data visualization for patient-specific outputs, she says.

“It’s kind of easy to aggregate at the hospital level because that covers up problems with the individual data, whereas if you drill down to the individual doctor or patient, the data has to be really clean. If there is one problem with the data in the background, it can throw off the metrics quickly,” Bellon says. “We’ve been developing reports at the physician level and some techniques to automatically send the report to the physician so they can have that report in their inbox without having to sign into a system. It’s exciting how much data and the different types of data that can be presented with good visualization.”

The general idea of graphs and dashboards is nothing new, of course, and people have tended to think of that level of data visualization as just a routine part of any quality improvement profession, notes Scott Berinato, senior editor at Harvard Business Review in Brighton, MA, and an expert on data visualization. But data visualization is moving way beyond just making a pie chart.

“Data visualization is not just a nice-to-have skill. You have to get good at it because it’s a competitive imperative,” he says. “Other people are starting to get good at it, they’re going to get better, and those who get good at it will find opportunities that others won’t. The people who develop skills in data visualization are going to impress their bosses more and advance their careers.”

Information design is emerging as a profession of its own, Berinato notes. Colleges are beginning to offer information design courses that teach data visualization, and employers are seeking those skills, he says.

The good news is that improving your data visualization skills is not a daunting task, Berinato says. Technology takes care of most of the actual creation of visual representations of data, so the task for quality improvement professionals is to learn what makes good data visualization and how to tailor it to the specific needs of an audience, he says.

Some of the more sophisticated tools have a steep learning curve, but there are many data visualization software programs that are easier to learn, Berinato says.

The only danger lies in falling behind the curve, he adds.

“There were people in the ‘80s who thought Excel was a bad idea, and they didn’t want to put any time and effort into learning it because the ledger system they did by hand worked just fine,” he says. “You can imagine someone now saying they don’t want to use Excel and they’ll just do it by hand. That’s as crazy as how it will be very soon if you balk at using data visualization, or if you don’t understand that it’s more than making a simple bar graph.”

Technology vendors often will offer education on using their software for data visualization, but Berinato says one of the simplest ways to learn more about data visualization is to follow the Twitter hashtag #datavis.

“You will be amazed at how many people are trying this, showing their work, trying out new techniques, and talking about the tools that are available to learn,” Berinato says. “Google ‘data visualization tools’ and you’ll find many free tools that will have data sets to play with, or you can upload your own data sets. There are tools that make it easy to start using data visualization more, and there are good books that will give you an understanding of the principles behind data visualization and how to make the most of those tools.”

Good data visualization will provide much-needed feedback to team members on the frontlines of quality improvement efforts, says Jeanette Ball, BSN, RN, PCMH, CCE, client solution executive with CTG, based in Buffalo, NY.

“It’s hard to run a quality improvement program without feedback that shows people their efforts are bearing fruit. It creates incentives for them when they see that line going up on the graph,” Ball says. “Data visualization also gives you the ability to demonstrate to your executives that you are working on the edicts they’ve given you. It’s important to roll out the data in ways that allow good visualization at different levels of your organization, from top executives to management, middle management, and right down to the provider level so they can gauge themselves against their peers.”

Ball notes that data visualization can be particularly helpful with physicians because they tend to be competitive and will respond when the graphics show how they stand against their peers.

Good data visualization also can be useful with compliance efforts, Ball notes. With compliance dashboards, you can track multiple areas that might be auditable during a compliance period, and data visualization can help manage information from multiple sites, she says.

“If an audit occurs, you have all the documentation at your fingertips and can produce reports from that data,” she says. “That can be hard to do on the spot, especially if you have 10 or 20 sites to account for. When the auditor shows up and wants that data, you can produce reports that are effective in showing that data when you have it all tied in to a dashboard already.”

Data visualization also can improve quality of care and patient safety by presenting information in a way that is quickly accessible to clinicians, says David Williams, general manager of the healthcare provider business unit at Conduent, a process services company based in Florham Park, NJ.

“While a lot of organizations can make information available to clinicians, that information is not always presented in a way that enables clinicians to understand it and make quick decisions,” Williams says.

For data presented to clinicians, standardization of that information can be better than giving physicians the option to customize how they want the information presented to them, Williams says. Some physicians will find the customization to be just one more task and never get around to it, so it can be better to just present them with a format that will work for everyone, he says.

Even with data visualization being adopted much more in healthcare over recent years, Williams sees room for improvement in both how and how much it is used.

“We’ve seen organizations use this tool to improve patient falls, pressure ulcers, and surgical site infections. It can free up time for clinicians who can access this data and act on it quickly without spending a lot of time trying to find and understand the raw data,” Williams says. “We see success when projects are based around patient safety and quality of care, matched up with all the regulatory requirements out there today.”