Electronic Health Records

Integration & Interoperability News

Mining Electronic Health Records Refines Diabetes Diagnoses

"These discoveries have the potential to dramatically decrease the number of undetected cases of Type 2 diabetes."

By Frank Irving

- UCLA researchers successfully predicted patients’ type two diabetes mellitus (DM2) using an algorithm based on information extracted from de-identified electronic health records. The EHR phenotyping outperformed conventional screening methods, according to study results published Feb. 16 in the Journal of Biomedical Informatics, and could greatly increase the number of correct DM2 diagnoses.

UCLA researchers predicted patients’ type two diabetes using an algorithm based on information extracted from de-identified electronic health records.

The research team also discovered several previously unknown risk factors for diabetes, including a history of sexual and gender identity disorders, intestinal infections and a category of sexually transmitted diseases.

“With widespread implementation, these discoveries have the potential to dramatically decrease the number of undetected cases of Type 2 diabetes, prevent complications from the disease and save lives,” said Ariana Anderson, the study’s lead author and an assistant research professor and statistician at UCLA’s Semel Institute for Neuroscience and Human Behavior, in a public statement.

Anderson and Mark Cohen, a Semel Institute professor in residence, led a team that analyzed electronic records for 9,948 people from hospitals, clinics and doctor’s offices in all 50 states. The records had been scrubbed of patient-identifiable information, but included vital signs, prescription medications and reported ailments, categorized by International Classification of Diseases (ICD) diagnostic codes.

The researchers used half of the patient records to develop an algorithm that enabled them to predict the likelihood of an individual having diabetes. They then tested the pre-screening tool on the other half of the records.

The major findings:

  • Having any diagnosis of sexual and gender identity disorders increased the risk for DM2 by about 130 percent (roughly the same as high blood pressure, a leading risk factor).
  • History of viral infections and chlamydia increased risk for DM2 by 82 percent.
  • History of intestinal infections such as colitis, enteritis and gastroenteritis increased risk by 88 percent.
  • Herpes zoster, previously shown to have a link to diabetes, was confirmed to have a connection to the disease (90 percent increased risk).
  • Chicken pox, shingles and other viral infections grouped under a common ICD code increased DM2 risk as much as high cholesterol.
  • On the opposite end of the spectrum, patients prone to migraine headaches and those taking anti-anxiety and anti-seizure medications had significantly lower risk.

“The overall message is that ordinary record keeping that doctors do is a very, very rich source of information,” Cohen said. “If you use a computerized approach to studying patterns in that data, you can greatly improve diagnosis and medical care.”

The study report notes that additional research will be required to determine the medical reasons that certain factors correlate with greater or lesser risk. Because the analysis was based largely on ICD diagnostic codes rather than actual individual diagnoses, the findings are not detailed enough to explain precisely which conditions are linked to diabetes, the report states.

Nonetheless, the EHR-based pre-screening tool proved to be 2.5 percent better at identifying people with diabetes compared to the traditional method of screening for blood pressure, BMI, age, gender and smoking status. The new tool was also 14 percent better at identifying people who do not have diabetes.

Extrapolating their results, the researchers calculated that if the new tool were used nationally, it would identify an additional 400,000 people with active, untreated diabetes compared to conventional pre-screening models.

“There’s so much more information available in the medical record that could be used to determine whether a patient needs to be screened, and this information isn’t currently being used,” said Cohen, who also is the director of UCLA’s Laboratory of Integrative Neuroimaging Technology. “This is a treasure trove of information that has not begun to be exploited to the full extent possible.”


Continue to site...