Putting healthcare on the map
"Geospacial analytics" may sound futuristic, but it's actually being used to enable data-driven decisions in population health management.
For example, Children's National Health System in Washington harvests EHR-related data and combines it with geomapping technology to help identify regional variations in health and healthcare.
"In my own field in pediatrics, we take care of a lot of respiratory illnesses. Things like daycare settings, weather changes, pollen counts, air pollution and smoking in the home are all things that clearly affect healthcare conditions in children," said Brian Jacobs, MD, who serves as both CMIO and CIO at Children's National. "Understanding geographic and co-variable distributions can tremendously impact the epidemiology of what's going on for a patient in that population and what you can do as a healthcare system or as a patient's family member to prevent illness."
Jacobs described his health system's geomapping initiatives during the Physician's IT Symposium on Sunday at HIMSS15.
The process involves identifying EHR-specific data fields, writing data queries to extract that data, filtering and cleaning the data, de-identifying it and geocoding it — combining it with co-variables such as the location of restaurants and grocery stores, educational facilities, healthcare clinics and hospitals, as well as population and economic demographics. A geographic information system merges the data and uploads it to create analytics and graphical maps.
Children's National currently geomaps children who have suffered scalds and contact burns to determine geographic distribution in terms of cases per square mile in the Washington area. The health system also uses its GIS to study childhood obesity, sickle cell disease and influenza outbreaks.
One of the biggest challenges of implementing the technology is the way doctors and nurses document findings. In its obesity study, Children's National discovered that up to 20 percent of the EHR-extracted data on straightforward metrics such as height and weight was either erroneous or vague.
"We need to clean the data to move forward," said Jacobs, "but we're talking about population health here. It's not really critical that every last piece of data is completely accurate. It's okay to discard some data in order to get to the greater good of understanding the population itself."
For instance, out of a total population of 400,000 children within a 50-mile radius in the greater Washington area, "we have the ability to easily get 50,000 clean records through a simple extraction process," Jacobs said.
Children's National put the technology to use in testing physicians' theory that fast-food restaurants were a major cause of childhood obesity. The health system mapped the distribution of restaurants against the patient population and surprisingly found no correlation.
"We saw it visually as well as statistically that there was no relation between the fast-food restaurants and childhood obesity in the D.C. area," Jacobs said. "So the thought of challenging the restaurants' menus and trying to keep kids away was mistaken. It showed how we could use the tool to manage a health condition in the population."