No more guessing: Health systems tap sophisticated data tools to pick expansion sites

Edmond, Okla., an affluent suburb north of Oklahoma City, has drawn the interest of major retailers. St. Louis-based Mercy health system has had its eye on Edmond, too.

With a population of more than 80,000, Edmond is younger and wealthier than Oklahoma City, with a median age of 34.8, median household income of $66,535, and a projected five-year growth rate of 11%.

The 35-hospital Mercy system used a variety of sophisticated data techniques to determine what services and specialties could be successful there. For instance, psychographic data from Buxton, a Fort Worth, Texas-based customer analytics company, showed local healthcare consumption trends. Edmond is “a well-established community, so you've got a little bit of an aging population, but you've also got some younger folks moving in raising families,” said Nikki Viner, Mercy's vice president of market development.

A year ago, Mercy opened the $88 million Mercy Edmond I-35, 10 miles up the interstate from its Oklahoma City hospital. The facility features a sports performance and fitness center, outpatient surgery, diagnostic imaging, pharmacy and laboratory services and a convenient-care center. It has nearly 40 physicians and other providers. In planning a building such as Mercy Edmond, “you have to think like a retailer,” Viner said.

Health systems like Mercy long have studied demographic and marketing data from such sources as the U.S. Census Bureau, health-planning agencies and hospital associations, in deciding where to locate new facilities. Increasingly, they are using new sources of information such as electronic health records, health insurers' billing records and data used by major retailers. Hospital marketing specialists, geo-analytic gurus, health planners and architects mash up the data, overlaying information from a variety of sources to create maps to visualize market opportunities.

Retailers for many years have used geo-analytics—visualization or mapping tools integrating sophisticated geographical and marketing data—to decide where to open stores and restaurants. Healthcare is starting to go that way, too.

Viner said selecting health facility locations is just like planning for a retail operation, moving beyond traditional demographic factors such as age, education and income to the types of data that a Target or Walgreen store tap. For Mercy, it's important to use all the best available data to make the right choices on where to place services such as primary care or a specialty hospital. “Healthcare is becoming more consumer-focused and retail-oriented,” Viner said.

Dr. Este Geraghty, chief medical officer for Redlands, Calif.-based Esri, a mapping technology company, said health organizations generally have similar data needs to those of retailers, including population demographics, population density and market potential. But they may make very different choices based on that data when siting a clinic for low-income people versus picking a spot for a high-end clothing store, for instance. “The data are much the same, but the perspective is going to be totally different,” she said.

Network analysis is especially important in determining locations for healthcare facilities, to understand and overcome barriers to accessing care. “You can analyze multimodal transportation time to a clinic—such as by car, bicycle or potentially even public transportation—during normal compared to peak traffic times,” Geraghty said. “Such considerations are critically important if we are serious about reducing disparities in service access.”

Some Esri clients using its ArcGIS mapping platform include Kaiser Permanente and Children's National Medical Center in Washington.

Bill Stinneford, senior vice president of Buxton, said his firm helps healthcare organizations identify potential new patients by specialty at the household level, looking across their operating areas. The firm helps determine where to place new facilities, helps select the right service lines for each facility and assists with improving marketing campaigns. Using Buxton's Web-based SCOUT analytics platform, clients can map market opportunities and access custom reports. Buxton's clients include Florida Hospital in Orlando and the Marshfield Clinic in Wisconsin.

Stinneford cautioned that a good site for a retailer might not be good for a healthcare facility, which cannot simply piggyback off McDonald's or a Wal-Mart's market research. “Just because it's good for them doesn't mean it's going to be good for you,” he said.

Healthcare organizations need to do their own homework, analyzing their customers from data extracted from EHRs and health insurance claims data to see what kinds of services potential healthcare customers use, where they live and what types of coverage they have. These data can be combined with market data, traffic pattern information and retail location maps.

Analysis of all these data reveal “the patterns in the chaos” to reveal the best locations, Stinneford said. “Maybe they have a different payer type, or are in a different life stage, or for whatever reason they just don't use urgent care that much,” he said. “Once you understand the DNA of what makes a successful location, you can score whatever geography you're looking for.”

Another caveat he offers is that it can be misleading to use traditional demographic data to come up with averages. For instance, the wealthiest neighborhoods in Houston are within a mile of the poorest ones, Stinneford said. “There are different insurance types and they're all mucked together. That's why you have to get down to the individual household, and the data are available to look at how many people are the right kind of people.”

Hospitals typically have relied on inpatient data for planning. But the growing shift to outpatient care has created a gap in the availability of useful data because outpatient facilities often don't have to report data to state agencies. Still, the availability of outpatient data is improving as some states and healthcare analytics firms, such as Truven Health Analytics and Sg2, are collecting and analyzing it, said Donald Bellefeuille, a healthcare strategist in the Boston office of NBBJ, an international health architecture firm.

He said the emphasis on data analysis and population health management in the Affordable Care Act is leading to improved data access to plan services and facilities and to help healthcare organizations decide whether to buy existing medical practices or build new ones.

Crystal Run Healthcare, a 350-provider, multispecialty group practice with 32 sites in the lower Hudson Valley, N.Y., used to rely primarily on the addresses of patients served at its main facility in Middletown, N.Y, to plan new sites. “Essentially, we took a leap of faith,” said Dr. Gregory Spencer, chief medical and chief medical information officer at Crystal Run.

But now, using its own data warehouse, applications from Salt Lake City-based Health Catalyst and geo-analytics dashboards developed using Tableau, Crystal Run uses much richer data to make siting decisions. Spencer said his organization now can answer such questions as “Who are these people? What are their conditions? What are their demographics? Which specialists are seeing them?”

Crystal Run applied this data-based approach to plan a $30 million facility scheduled to open in July near a Wal-Mart in Newburgh, N.Y., about 20 miles from the group's main campus in Middletown. Spencer said the practice decided to build the facility after data showed about 20% of its patients came from near Newburgh, population 28,866, a declining, racially diverse city with a median income of $35,731. The new center will feature 20 specialties, an urgent-care center, women's imaging, diagnostic testing and a clinical lab.

Spencer said the analytics give his organization confidence that the Newburgh facility will reach capacity quickly. The days of guessing are over. “You have to use all the information you can just so you don't waste money,” he said.

In booming central Florida, Josef Ghosn, senior vice president of strategic planning for Orlando-based Florida Hospital, is spearheading his organization's building spree. The not-for-profit system, part of Adventist Health System, with eight hospitals and 22 urgent-care centers, just opened a tower in Kissimmee, broke ground on a hospital in Apopka and is building a large outpatient complex with a freestanding emergency department, surgicenter, imaging center and office in Winter Garden.

Ghosn said Florida Hospital asks three key questions before deciding to build: Is there a community need? Does Florida Hospital need to expand its network in that specific market? Is that market projected to grow? “Usually, if the answer is yes to all three, it becomes an attractive market we want to enter,” he said.

Florida Hospital uses state planning data to get a historic view of population and volume of health services based on healthcare utilization and demographics. Its own data provide a snapshot of present conditions. And firms such as Truven, Sg2 and the Advisory Board offer forecasting models to project future demand.

For example, Florida Hospital broke ground this year for the $203 million Florida Hospital Apopka scheduled to open in 2017 with 120 beds, up from 50 beds in the system's current hospital near the city center. The new hospital will offer general surgery, orthopedics and ear, nose and throat care.

The facility's location will reach out to new communities that have developed outside of Apopka in Orange County. The city had a population of 45,587 in 2013, an increase of 9.5% since 2010. Median household income is $59,424, nearly $10,000 a year more than the state median. Ghosn said the facility is aimed at meeting needs in Apopka as well as expanding the system's reach into new markets.

David Crockett, a senior research director at Health Catalyst, whose clients include Kaiser Permanente and Stanford Hospital & Clinics, said healthcare organizations are taking advantage of more intelligent geo-analytic resources. His company is developing tools that determine the geographic service area of a healthcare system and highlight characteristic disease distribution across that population. In addition, the tools provide a hierarchy of care facility levels, and include details of patient drive-time access to needed services such as pharmacies, clinics, hospitals and rehabilitation centers.

Crockett said the trick is to take information generated by the healthcare system itself and merge that with public data sets such as census statistics on population density and income levels. Additional data on disease rates, crime statistics or air quality can be added, along with locations of farmers markets, gyms and parks. “The sheer quantity of data available today allows clients to make more informed decisions than in the past, where educated guesses were the best way to pick locations and services,” he said.

Still, Crockett said, the healthcare industry is “at least 15 years behind the curve” set by the retail, logistics and energy industries.

“Healthcare just hasn't recognized the opportunity. But it's starting,” he said.

Howard Wolinsky is a freelance writer based in Flossmoor, Ill.