Making Sense of Analytics: The Search for Answers Starts with Asking the Right Questions

On the surface, cardiovascular information technology (CVIT) analytics represent a treasure trove of opportunities to improve care, reduce costs and deliver better healthcare experiences for patients and clinicians. While a variety of obstacles have discouraged some health systems from investing in analytics in the first place, the opportunity to unearth their potential is inspiring others to dig deep for creative solutions.

Fertile ground for innovation

A few years ago, Mercy Health in St. Louis instituted a pilot program that used scannable barcodes to track implanted cardiac stents. The team at Mercy, led by principal investigator Joseph P. Drozda, Jr., MD, the health system’s director of outcomes research, used prototype unique device identifier labels that will eventually be required by the FDA on all medical devices.

The device data from the labels and associated device attribution data were joined with electronic health record (EHR) data to create a database that was refreshed weekly and tracked a few thousand patients with cardiac stents for more than a year. In this proof-of-concept study, the new, combined database produced nearly the same results as a randomized, controlled trial would have, Drozda says, and with a much lower cost and time investment.      

Mercy proved its CVIT infrastructure was capable of producing a valid data set, and now it is moving onto the next phase: replicating the demonstration project with two other health systems. This triple-sized network will have data robust enough to evaluate new stent technology coming onto the market, Drozda says. “We’re starting out with cardiac devices,” he explains, “but we’re trying to put together a system that’s generalizable to all sorts of medical devices.”

While the project happened to start in Mercy’s cath lab because stents were a simple device to track and Drozda is a cardiologist by training, cardiology departments are an ideal training ground for innovation in IT analytics. “Cardiology is a big cost center,” he says, “a big revenue generator.”

Communication hurdles

Across the country, clinicians are teaming up with their IT colleagues to use—or try to use—CVIT analytics to improve patient care, reduce costs, achieve efficiencies and enhance clinician satisfaction. The solutions some health systems have developed include dashboards for  tracking lab complications and procedure start times. But, despite gains, obstacles such as the variety of proprietary systems and potentially overwhelming expenses continue to hinder their ability to use CVIT analytics and generate results.

The current landscape of CVIT analytics is, at best, fractured. At Beth Israel Deaconess Medical Center in Boston, an array of different data systems don’t communicate and even have separately maintained dictionaries, says Kalon Ho, MD, MSc, director of quality assurance for the cardiovascular division.

At Centra Health in Lynchburg, Va., Chad Hoyt, MD, medical director for the heart and vascular center, has no difficulty listing examples of  how robust CVIT analytics could yield greater efficiencies and improve processes related to patient care. For starters, the annual accreditation process for imaging labs would be quicker and simpler if only the various imaging systems talked to one another. A list of the cardiomyopathy patients who had an echocardiogram in the last year would be helpful for ensuring that their providers have talked with them about defibrillators. Generating such a list is nearly impossible, Hoyt says, and “that has patient care ramifications.” To ensure carotid stenosis patients without stents don’t fall through the cracks, Hoyt would like to send automatic annual reminders that they are due for repeat studies. But that requires a struggle with Centra’s EHR. “Look at all the businesses that can [send automated reminders],” he says.

[[{"fid":"22856","view_mode":"media_original","type":"media","attributes":{"height":331,"width":577,"alt":" - new-approach-outcomes-research","class":"media-element file-media-original"}}]]

Tough-to-sell investment

The most notable CVIT analytics successes may be with operations and basic tracking, such as counting foreign bodies left in after surgery. At UC San Diego Health System, CVIT analytics are primarily used to monitor supply and product usage in labs, as well as utilization by case and operator, according to Ramesh Sivagnanam, MBA, director of cardiovascular services.

Yet UC San Diego has succeeded in stretching its use of CVIT analytics. When Sivagnanam found data showing on-time starts in the procedure lab were around 64 percent, his team focused on improving the metrics. “We want to get this patient on the table on time,” he says. “We’re going to utilize the room more efficiently as opposed to starting an hour late.” By engaging all the players, from physicians to staff, the team increased on-time starts to 86 percent.

Another success has been in using CVIT analytics to save the health system money, Sivagnanam says. Because cardiovascular labs are expensive, cost data are particularly important. By knowing, for instance, how many stents were used per procedure and throughout the health system, the hospital has been in a better position to renegotiate price agreements with vendors.

The ability to leverage a health system’s CVIT analytics to reduce costs is crucial, according to Hoyt, because the investment in analytics itself—from buying the software to funding additional workforce—could reach into the double millions. “It’s like building a new building,” he says. “It’s a lot of money.”

With slim operating margins, health systems all over the country struggle with whether to make a significant investment in CVIT analytics. “When you’re looking at a system where the operating margin is a couple percent, it’s very hard,” Hoyt says. “There are tough choices to make.” But, he adds, when health systems are facing the possibility of penalties or incentives from payers, the cost of the investment becomes easier to swallow.

Besides the technology itself, the likely larger cost is the people-power to create and maintain CVIT

analytics systems. Software isn’t hard to find, Ho says, but the greater need is for analysts who can unearth, extract and analyze the data. But, when pitched to human resources, a data analyst can be a tough sell. “It’s hard to convince the powers-that-be to fund these sorts of positions,” he has found. Administrators want to ensure a return on their investment, but it’s impossible to demonstrate the return without first making the investment.

Without analysts, cardiologists find their own ways to examine the data. Centra has a weekly operations dashboard to which 10 staff members contribute. They log about 40 hours of work per week to generate a detailed, three-page report with data elements including how long patients spend in the waiting room, new visits and physician productivity. Centra also uses separate dashboards for quality and financial information. Hoyt is “not pleased with how much staff time it ties up.”

Focus on the questions, from start to chart

There are other ways to find bits of information in CVIT systems, says Ho, whose cath lab at Beth Israel Deaconess maintains a dashboard devoted to tracking complications, case costs and more. “Part of the challenge there is to have the report automated and to update automatically,” he notes.

Another challenge is asking the right questions, one of the reasons clinician involvement is critical. “You have to start with a question,” Ho says. “You don’t want to start with a data set.” They’ve had cases of analysts with no clinical perspective coming up with things like coronary artery disease as a predictor of bypass surgery. “Not very helpful since they don’t know what they’re looking at,” he explains. “Someone has to ask the right questions first, even if they don’t know the data elements going into answering the question.”

Ho is partial to spark charts and sparklines for their efficiency in showing data in little space. The dashboard consists of a few sheets of paper, each page with eight to 10 spark graphs. Trends are easier to spot in graphs of varying colors, rather than tables, he says. There are courses and seminars devoted to efficient ways to graph data, including thinking outside the traditional bar and pie charts to display more information-rich material.

With so far to go, it might seem tempting to scrap everything and invest in a huge new CVIT analytics system. But at Centra, they’ve engaged a cardiovascular IT consultant to map the current systems and help them figure out what they have and what they need to make them work together.

[[{"fid":"22857","view_mode":"media_original","type":"media","attributes":{"height":331,"width":577,"alt":" - seeing-data-differently","class":"media-element file-media-original"}}]]

Big potential, but no guarantees

The full potential of CVIT analytics will be realized when clinicians can obtain real-time information, Hoyt says. Not just dashboards, he emphasizes, but instant risk calculators on a mobile phone, feedback on performance and support from clinical decision tools to allow modification of treatments. “Delayed information is just not going to cut it,” he says.

Meanwhile, analytics can provide a sense of what should be happening, Ho explains, but the model becomes obsolete once something changes. He urges cardiologists to be cautious about applying descriptive models in a proscriptive way because there always will  be outliers.

“We don’t collect all the data, so we can’t have a perfect model,” Ho says. “It’s like when you hear a commercial for a particular stock. Past performance [of the model] is no guarantee of future performance for a specific patient.”