Using predictive modeling to prevent diabetes

Common scenario: A doctor sees a patient who is overweight and complaining of fatigue, increased thirst and occasional blurred vision. The doctor immediately recognizes these signs and symptoms and suggests a screening for diabetes.

Question: What if we could have prevented this doctor appointment altogether?

The majority of healthcare models today are in a reactive model-meaning once a patient starts showing symptoms of type 2 diabetes, the doctor will order the necessary screenings and recommend the appropriate prescriptions.

New recommendations from The United States Preventive Services Task Force (USPSTF) advise healthcare providers to screen all overweight and obese adults between 40 and 70 years old for abnormal blood sugar levels. In other words, once a person has a risk factor-in this case, overweight-then the screening is done.

BeveridgeHowever, some Americans who are at risk won’t have an interaction with a healthcare provider this year. Others who are not classified as overweight, or who are younger than 40, won’t be screened but could still have the abnormal blood sugar levels that characterize prediabetes.

Without intervention, prediabetes is likely to progress to diabetes within 10 years, according to the Mayo Clinic. But with lifestyle modifications, such as weight loss and changes to diet, people with prediabetes can delay or even prevent the onset of type 2 diabetes.

 

Where we are vs. where we’re going

Enter the advanced predictive model of diabetes prevention. This model takes an in-depth look at past health assessments, old medical claims, lab results and existing electronic medical records to anticipate who is at risk for the disease and allows healthcare providers to intervene. So instead of relying on the patient to complain of symptoms, at Humana, we’re getting ahead of the process and using past evidence to inform proactive preventive care.

Our predictive model came about during our partnership with Omada Health, a digital behavioral medicine company, and the Centers for Disease Control and Prevention’s (CDC's) National Diabetes Prevention Program (DPP), when we found we were struggling to identify individuals who were predisposed to diabetes. Of our more than 3 million Medicare Advantage members, we could identify only about 9,000 people who were at risk of diabetes.

So we created a model that uses multiple data points and analytics to find more people at risk for the disease.  Today we use this model to inform as many as 800,000 people of their risk. We then connect them to resources that help them reduce their risk of developing diabetes. The key intervention components include small-group support, personalized health coaching, DPP curriculum, and digital tracking tools. Six months after starting the program, members reported an average weight loss of 8.7% of their total body weight and more than 85%of those participating maintained adequate activity levels.

Others are using predictive models, too. At the state level, health organizations are beginning to implement programs and engage in partnerships that use a predictive model by using historical data to better inform patient outcomes. For example, the Commonwealth of Kentucky (state government) implemented the DPP for its employees, which includes a CDC-approved, evidence-based lifestyle program to prevent or delay onset of type 2 diabetes. More than 680 organizations are enrolled in the program, and it reaches more than 30,000 participants.

Getting ahead of it

Broader screening for prediabetes is a good start, but more is needed. Being able to identify individuals at risk for type 2 diabetes in a timely fashion could, over time, be a catalyst to a smaller population of disease diagnoses. If we can connect the dots between the technology at the fingertips of today’s patient and link it to the medical history, we can make more prescriptive decisions, engage in early intervention and help stop this lifestyle disease in its tracks.

Roy Beveridge, MD, a member of the Managed Healthcare Executive Editorial Advisory Board, is the chief medical officer for Humana.