Model Predicts Hypoglycemia Risk in Adults With Diabetes

In a study published online in Current Medical Research and Opinion, a predictive model is presented for hypoglycemia, which combines nearly all known and readily assessed risk factors for hypoglycemia such as infection, non-long-acting insulin, and dementia.

Xiaochun Li, Ph.D., from Indiana University in Indianapolis, and colleagues conducted a retrospective cohort study involving urban adults prescribed a diabetes drug between 2004 and 2013 to develop a hypoglycemia prediction model. Data were included for 38,780 patients with a mean age of 57 years.

The researchers found that hypoglycemia occurred in 8,128 patients; 539 patients were identified by natural language processing only. Infection, non-long-acting insulin, dementia, and recent hypoglycemia were factors positively associated with hypoglycemia. Long-acting insulin plus sulfonylurea and age 75 years or older were factors negatively associated with hypoglycemia. The area under the curve for logistic regression, classification and regression trees, and Random Forest models was similar (89, 88, and 90 percent, respectively, with 10-fold cross-validation).

"Clinicians could use these findings to identify and address important modifiable risk factors. Such a tool could be used by electronic health record systems, to automate the retrieval and presentation of risk factors for clinicians during medical encounters with patients with diabetes," the authors write. "This may lead to immediate counseling of patients or changes to medical practices in pursuit of addressing the risk factors."

Several authors disclosed financial ties to pharmaceutical companies including Merck, which funded the study.

-----------------------------------------------------------------------
Editor's Note:

If you have any suggestion to the content, 

please email: Julia.Zhang@ubmsinoexpo.com