A nationwide analysis of the electronic health records for nearly 10,000 patients by UCLA researchers has uncovered several previously unknown risk factors for Type 2 diabetes, including sexual and gender disorders, intestinal infections and some sexually transmitted diseases.

Using a newly developed screening algorithm and 9,948 electronic records gathered from hospitals, clinics and doctors’ offices in all 50 states, the research team was able to improve screening for the disease by predicting the likelihood of an individual having the disease and then successfully testing the pre-screening tool.

In the process, they believe they have developed a more accurate and less expensive way to identify patients who have undiagnosed Type 2 diabetes.

In fact, based on their calculations, the researchers believe that if their new method were used nationally, it would identify 400,000 people who have not yet been diagnosed with the disease. Their findings were published Tuesday in the Journal of Biomedical Informatics.

Ariana Anderson, assistant research professor and statistician at UCLA’s Semel Institute for Neuroscience and Human Behavior.
Ariana Anderson, assistant research professor and statistician at UCLA’s Semel Institute for Neuroscience and Human Behavior.

“There’s been a fair amount of work done using electronic health records to screen for disease, but there’s a huge problem with these studies because they have a limited patient sample and very narrowly define the patient population,” says Ariana Anderson, the study’s lead author and an assistant research professor and statistician at UCLA’s Semel Institute for Neuroscience and Human Behavior. “What we wanted to do was see whether there was residual information in these typical clinical records that could be used to improve the ability to screen for diabetes.”

According to Anderson, clinicians traditionally screen for the disease based on a limited range of factors, including blood pressure, body mass index, age, gender and whether or not they smoke. However, she says the UCLA team’s pre-screening tool based on the entirety of a patient’s EHR was 2.5 percent better at identifying patients with diabetes than the standard approach, and 14 percent better at identifying those who do not have it.

A statistician by training, Anderson leveraged a few different types of models to predict diabetes while leveraging a free Practice Fusion database of EHRs from 2009-2012. Among newly discovered risk factors for Type 2 diabetes, a diagnosis of sexual and gender identity disorders increased the risk for the disease by about 130 percent—about the same as high blood pressure as a leading risk factor.

“This is based on ICD-9 codes,” says Anderson, who called it “probably the most surprising finding,” but adds “we don’t really have any sort of explanation for it and we’re not sure what’s driving it—there are so many different possible explanations.”

At the same time, other health conditions have shown to be nearly as important risk factors for Type 2 diabetes. For instance, while herpes zoster has previously been shown in the medical literature to have a link to diabetes, researchers confirmed that association finding that it increases the risk by about 90 percent.

In addition, a patient’s history of intestinal infections such as colitis, enteritis and gastroenteritis resulted in an increase in their risk for diabetes by 88 percent increase, and history of viral infections and chlamydia resulted in an 82 percent. By comparison, those having a high body mass index demonstrated a 101 percent increase in their risk for the disease.

Further, some other lesser-known risk factors such as chicken pox, shingles and a range of other viral infections revealed an increased risk for Type 2 diabetes—in fact, as much as high cholesterol.

“Clinicians are spending so much of their time recording all this information. What needs to be done is putting an algorithm within all of the large medical databases to automatically calculate these risk factors directly,” concludes Anderson, who adds that her UCLA team is looking to commercialize its algorithm. “I don’t think it should be done for just diabetes. There are so many other disorders that can have a very harmful impact on patients. Of course, the models are going to be slightly different depending on the patient population.”

Register or login for access to this item and much more

All Health Data Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access