Group practice uses predictive models to improve care, gain savings
Better use of patient data enabled Palmetto Primary Care Physicians to improve care transitions, education and quality.
The South Carolina-based organization worked with Optum Analytics to mine patient care data and track physician performance through provider scorecards.
The multi-site practice, with 36 clinics and 124 providers, created a competition to see which could most effectively use data-based insights. The result was improved care and savings of $4 million across Palmetto’s accountable care organization contracts.
Competition aside, the core goal was to identify patients at risk before they become patients with acute conditions, says Terry Cunningham, CEO at Palmetto.
Using artificial intelligence and machine learning predictive models to support early outreach to patients, the practices worked with OptumIQ to conduct risk stratification analyses of patients to identity those who “ride the line on pre-diabetes and are on their way to being introduced to diabetes,” Cunningham explains.
Other providers are piloting test cases for patients at risk for artrial fibrillation, and patients believed to be at risk for a chronic condition also now are going through the risk stratification and identification process, says Steve Griffiths, senior vice president and chief operating officer of enterprise analytics at Optum.
Analytics is the beginning of an effort to identify and engage at-risk patients, but if an organization does not have adequate change management processes, it can’t drive provider performance initiatives, says Griffiths.
Getting buy-in for the program was a challenge, Cunningham acknowledges. There always is apprehension with change, and some clinicians did not initially trust the data they were being given. To encourage competition, clinician scorecards were made public within the organization.
Clinicians were given education on why the program was being done. In addition, they received training in the best approaches for explaining to patients why the clinician is doing a certain procedure or why the clinician is asking certain questions, which better inform the clinician and engage the patient.
For example, Cunningham says, if a patient mentions “prediabetes” during the questions, the clinician will offer the patient access to a class on diabetes at Palmetto’s cost.
Organizations preparing for population health technologies should focus first on obtaining physician buy-in before they start looking at products. When technology is implemented, healthcare organizations need tech-savvy physician champions to guide peers and get them ready for the changes, Cunningham advises.
“We created a quality committee with 12 champions covering areas such as templating, electronic health records, health screening and the need for an annual wellness visit,” he explains.
As technology advances, Palmetto Primary Care Physicians is looking at how the use of artificial intelligence and bots can pull abstractions or records quickly from the electronic health record system. This is data that physicians trust because it’s already in the EHR and it’s already been used.
The next steps for Palmetto are focusing on social determinants of health to better understand patients and how they live; giving physicians EHR and claims data to support care coordination; using provider performance analytics; signing risk arrangements with insurers; and improving the targeting of at-risk patients.