Skip to content

Blog Posts

New analysis reveals higher rates of prediabetes progressing to diabetes

According to estimates from the Centers for Disease Control (CDC), for every 100 people, 33 people are living with prediabetes and 30 of them are unaware of it.

The bad news is that once prediabetes progresses to diabetes, it can cause serious complications, especially if left unmanaged. Once diabetes starts, it is much harder to control. For example, diabetics are at higher risk for retinopathy (which can cause blindness) and nephropathy (which can cause kidney failure), as well as stroke and cardiovascular events.

The good news is that research also shows that finding and connecting with prediabetic patients can help prevent or delay diabetes and improve outcomes. How can we get more effective at identification, prediction, intervention and prevention? Big data – using both health and non-health data, machine learning and artificial intelligence – is the key.

Big data, big thinking: Building predictive models to manage diabetes

Allscripts Analytics has analyzed more than 48 million patient records and is combining it with other types of data, including claims data, consumer information and environmental data. We are collaborating with the American Diabetes Association (ADA) to focus on what this data can tell us to help patients, including updating messages to make them more relevant and incorporating best-practice guidelines into EHRs.

We created rules with the CDC criteria that define prediabetes, including patients with HbA1c levels between 5.7% and 6.4%, body mass index (BMI), and age. We applied this rules engine to our EHR data and found more than 3.5 million patients who fit the criteria and were at risk for diabetes.

Next, we followed these patients over four years, and we found that 80% of these patients went on to develop diabetes. This unexpected finding is much higher than the CDC’s findings of one in three patients. But our ability to conduct this large-scale analysis and find a huge progression to diabetes demonstrates that this problem may be grossly underestimated in the United States today.

Given the complexity of disease, it’s worth the effort to uncover insights that can help clinicians predict which patients will develop diabetes. By embedding the CDC’s algorithm into our EHRs, we’ll be able to support clinicians in early identification of prediabetes. We’re continuing to collaborate with the ADA on a consumer app that will use this data to help people assume responsibility for their own care in the real world. Together with patients and clinicians, we’ll continue to work toward better outcomes.

Comments 1

  1. Sagar 03/29/2019

    Thanks for sharing the information, it is very helpful to us

Add a Comment

Scroll To Top