by Brian Eastwood

How Data Analytics, GIS Helps Dartmouth Atlas Influence Healthcare

Feb 03, 20144 mins
AnalyticsBig DataBusiness Intelligence

The Dartmouth Atlas of Health Care has more than 100 TB of Medicare claims data. This information, combined with peer-reviewed medical research, informs healthcare policy and helps institutions compare their quality metrics. Like any good atlas, it also uses data-driven maps to prove its points.

Over the last 20 years, The Dartmouth Institute for Health Policy and Clinical Practice has examined variations in healthcare delivery across both regions and health systems. The goal is to research the causes and consequences of that variation so patients better understand the health services available in their community.

“In healthcare, geography is destiny. Where you live and where you receive care makes a tremendous difference,” says Dr. David C. Goodman, professor of pediatrics and health policy at The Dartmouth Institute. (Research shows that there’s a relationship between ZIP codes and life expectancy.)

[ Analysis: Can Healthcare Big Data Reality Live Up to Its Promise? ]

[ Also: Healthcare Needs Data Analytics for the ACO Model to Succeed ]

Over time, Dartmouth has amassed the largest non-governmental set of Medicare data in the United States, Goodman says — more than 100 TB in all, and growing. This data, compiled and displayed in the Dartmouth Atlas of Health Care, has measured trends in hospitalization rates, ambulance use and medication use, getting as granular as specific hospitals and, in the case of medication use, subsets of the patient population (such as those with advanced cancer or chronic illness).

With So Much Data, Visualization Becomes Paramount

Not surprisingly, this information, normalized across decades, interests many stakeholders:

  • For healthcare systems, Dartmouth Atlas provides performance data, letting facilities compare themselves to hospitals both near and far.
  • Policymakers, meanwhile, use the data as guidance for defining the future of healthcare, particularly when it comes to clinical quality measures and opportunities to lower costs without compromising that quality, Goodman says.
  • Though the Dartmouth Atlas isn’t necessarily a consumer site, its information has been framed for patients, most notably in Consumer Reports lists of hospital readmission rates and overall hospital rankings. (These lists aren’t without criticism, though, as hospitals that perform expensive procedures aren’t on the measure of cost alone inefficient and ineffective.)

All this information, combined with what Goodman describes as a “foundation of peer-reviewed papers” that’s several hundred strong, undergoes “fairly sophisticated” statistical and spatial analysis, such as the estimated travel time for residents of one area to the nearest critical access hospital. The resulting inferences and conclusions must be presented in a meaningful way, lest viewers get lost amid a sea of charts.

For that purposes, Dartmouth Atlas turns to geographic information systems — specifically, ArcGIS from Esri — for mapping and analysis. Since much of the institute’s data can be referenced at the census tract or zip code level, robust software that can summarize data in map form makes it that much easier to visualize.

[ Feature: 8 Cool Heat Maps That Help You Visualize Big Data ]

[ More: 8 Great Sites for One-Stop Data Visualization ]

“[Maps] stimulate curiosity,” Goodman says. “It makes them read the text to find out more about what they’ve learned. It’s a terrifically powerful inference into the research findings and policy implications of the work.”

Lotus Formula One
This map of tonsillectomies performed in northern New England illustrates what happens when Medicaid claims data and mapping software come together. (Image via Dartmouth Atlas .)

Health Data Analysis Expands Beyond Medicare to Pediatrics, Surgery

While the foundation of the Dartmouth Atlas is raw claims data for the approximately 35 million Medicare beneficiaries in United States, the institute is increasingly gaining access to data sets for patients under 65, Goodman says.

The institute began receiving data on pediatric care in December 2013 and intends to examine “all domains of healthcare” for children, from medication to ER use to surgical procedures, Goodman says. Meanwhile, a project in its early stages aims to present “a unique perspective” on how surgery falls into treatment plans for conditions such as organ failure and joint replacement.

Both endeavors will maintain a geographic focus, Goodman says, since that remains such an important factor in the type of healthcare that an individual receives.

Brian Eastwood is a senior editor for He primarily covers healthcare IT. You can reach him on Twitter @Brian_Eastwood or via email. Follow everything from on Twitter @CIOonline, Facebook, Google + and LinkedIn.