New York wasn’t the first state to tinker with reforming Medicaid, but its attempt may have been the most ambitious. A case study from Staten Island presented Tuesday at HIMSS shows what’s possible when the right players—including EMS—are brought into the game.
What that effort achieved, said Joseph Conte, PhD, executive director of the Staten Island Performing Provider System, was a reduction by half of preventable ED visits by impacted beneficiaries—who include roughly 40% of those living on the island. Accomplishing it took a cross-sectional effort under a program called DSRIP (the Delivery System Reform Incentive Payment program), the goal of which is to fundamentally restructure the healthcare delivery system by reinvesting in Medicaid, with a primary target of reducing avoidable hospital use by 25% over five years.
Linking Medicaid claims and patients’ EHRs was the first step toward this, but not the last. SIPPS worked to populate an entire EDW, or enterprise data warehouse, with live data from new partners. This had the unanticipated benefit of bringing competing local healthcare entities together, noted Raj Lakhanpal, MD, CEO of a company called SpectraMedix, which worked with SIPPS to help harmonize the data from disparate sources.
As the array was built, project managers used advanced analytics to inform its strategic direction, gauge pay-for-performance risk, and develop specific plans. Hotspotting and geomapping proved invaluable tools—revealing, for example, that behavioral health resources were clustered toward the northern end of the island, while many of those who needed them were to the south. “It’s important to bring in those outside data sources,” said Lakhanpal. That included EMS data, which helped identify superutilizers of 9-1-1 and emergency department services, and other data from sources like schools, housing authorities, and more.
Broadly, the data was pulled from three places: public sources for publicly available figures; the state health department for claims and other aspects; and additional repositories like managed-care organizations.
Using deterministic matching, number-crunchers improved match accuracy among records from 83% to 98% despite the challenges of different institutions using different patient identifiers and even some patients having multiple identifiers within single organizations.
Just 1.2% of SIPPS Medicaid enrollees were ultimately classified as high-risk, but they accounted for 40% of the system’s preventable readmissions. To fix that, Conte stressed, “you have to work with partners outside the clinical realm.” For diabetic patients, that meant partnering with the City Harvest program, which offers food and nutritional assistance to the needy; for child asthmatics it was home visits looking for triggers and opportunities to work with schools. EMS data showed alcohol abuse as the top driver of 9-1-1 misuse, followed by psych issues.
The outcomes were reductions of 10% in potentially preventable readmissions; 9% in potentially preventable ED visits; and 7% in potentially preventable behavioral-health visits from the same six-month period in 2015 to 2016. The bottom line, Conte said, was that data can drive reform, an approach that can work across the concept of value-based care.