With ACOs having tripled in number since 2011, to almost 500, and costing between $1 million and $4 million up-front, getting the analytics right can mean the difference between success and failure in the shared savings model. So said Naveen Sarabu, in a talk he delivered Oct. 29 at the AHIMA Convention. Many ACOs start out with massive population health analytics projects, he told the audience. Sarabu suggested taking a different tack.
[See also: AHIMA embraces history, looks to future.]
Essentially flip the current approach and build from the ground up, Sarabu, director of product management at Liaison Healthcare Technologies, said. His approach requires constructing the analytics aspect of an ACO with these three building blocks in mind from the get-go.
1. The practice level. “Start with the practice level information,” Sarabu explained. “The easiest thing to begin with is lab results information because that is standardized to a large extent.” And if you have access to orders you get even more information about the patients, such as clinical data and demographics. “The hardest thing is laying the pipes, getting that data into the repository.”
2. Create a longitudinal record. When that plumbing from step 1 has been established, however, the second thing to focus on is creating a repository of longitudinal patient records. To bring in even more information, Sarabu recommended getting claims data, 837s, EHR data, practice management system information. And make sure your physicians know how to document well.
“You have to mash it up with clinical information, medical encounters, allergies,” Sarabu said. “Once you have that you have solved the puzzle. You have the data that is needed for any kind of analytics.”
3. Population health analytics. Figure out what metrics, stratifications, and analytics you want to run. Preparing the analytics requires harmonization of information — which, for an ACO with 500 participating practices, means 500 interfaces, not including the hospitals’ interfaces. “Whether it is claims data or clinical data, somehow you have to put them together for meaningful claims reports and other types of analysis,” Sarabu noted. And it’s not just about meeting ACO requirements, the data can be used in other ways, such as clinical research, or with pharma’s that want to track how effective they are to better understand what’s happening.
It is, of course, one thing to boil the building blocks to interoperable, analytic-savvy ACOs down in an hour-long conference session, Sarabu acknowledged, and something else entirely to actually put them into practice. Bringing multiple organizations together to usable share health information is not easy, and it’s not cheap, but it needs to be done to gain control of care costs.
“Once you have that you have solved the puzzle,” Sarabu said. “You have the data that is needed for any kind of analytics.”