Stopping Diabetes Before It Starts

How can a personalized risk conversation help prevent type 2 diabetes?

More than one-third of U.S. adults have prediabetes, a condition that raises their risk of developing type 2 diabetes.^1,2 Everyone with prediabetes is at elevated risk, but an individual’s odds depend on many factors and vary widely.³ The Diabetes Prevention Program (DPP) showed that an intensive lifestyle program and, to a lesser degree, metformin can substantially lower risk,⁴ and newer medications such as GLP-1 agonists are highly effective for people with prediabetes and obesity.⁵ Despite the benefits these interventions have for individuals with prediabetes, health systems are limited in their capacity to address every single case of prediabetes and to provide services like the DPP to everyone. This makes it essential for providers to identify patients with prediabetes that would benefit the most due to higher diabetes risk and to make that information actionable.

A new study published in Learning Health Systems, a collaborative effort between AMGA research, Tufts Medical Center, the Allegheny Health Network and Highmark Health, examines what happens when providers are equipped to do exactly that, using an electronic health record (EHR)-based personalized risk calculator⁶ at the point of care.

Why Personalized Risk?

In the study’s provider and patient focus groups, providers said they could tell patients only that prediabetes increases their chances of developing diabetes, without the specific number patients wanted. Patients reported that they wanted that number; many spontaneously recalled the ages at which relatives had developed diabetes, showing they were already thinking about their own risk in concrete terms. A personalized estimate gives both sides a shared, factual basis for deciding whether and how aggressively to act, and patients said that knowing their risk was high would be a strong motivator for change.⁷

What the Study Found

An implementation study at Premier Medical Associates (PMA), a 100-provider multispecialty group near Pittsburgh, integrated the risk model into the EHR across seven primary care offices and tracked its use among roughly 2,500 patients with prediabetes seen between 2018 and 2019. An outcome study then followed a subset of those patients for three years, comparing them with a propensity-score-matched group who received usual care elsewhere in the same health system.

The findings were notable on two fronts. The tool sharply increased provider confidence: Before implementation, only 41.6% of providers felt confident estimating an individual patient’s risk, and after the model was available, that figure rose to 92.8%, with similar gains in their confidence tailoring and communicating prevention recommendations.⁷ More importantly, the conversations were linked to better outcomes. Among patients whose providers used the tool, 19.5% developed diabetes within three years, compared with 27.6% of propensity-score matched controls (p=0.042). This benefit held even though the intervention was simply receiving a personalized risk estimate during a discussion, with no required follow-up action.⁷

The Bottom Line for Health Systems

The study’s findings are both encouraging and practical. A simple, scalable change, giving patients a personalized risk estimate during a routine conversation, was associated with meaningfully fewer progressions to diabetes, even without protocol-driven follow-up.

AMGA member health systems looking to strengthen diabetes prevention have an accessible model to build on. As limited program capacity and the rising cost of newer therapies make careful targeting more important than ever, equipping providers to identify and engage the patients at greatest risk offers a high-value path forward.⁷

Elizabeth L. Ciemins, PhD, MPH, MA, is chief research officer for AMGA, and Francis Colangelo, MD, MS-HQS, FACP, CPHQ, was previously the Chief Medical Officer for Premier Medical Associates, and is now a primary care physician for the Allegheny Health Network.