CHORI Staff Directory
CHORI Intranet

 
Revolutionizing Diabetes Diagnosis
CHORI Scientists Develop Novel Model for Differentiating Type 1 and Type 2 Diabetes in African American and Hispanic American Children

March 7, 2012 - Just a few decades ago, type 1 diabetes (T1D), then called “juvenile diabetes,” and type 2 diabetes (T2D) were easy to distinguish. T2d was seen almost exclusively in adults, while T1D was seen in pediatric populations. The advent of the obesity epidemic in the last 10 to 20 years, however, has dramatically changed the clinical landscape for T1D and T2D differentiation, with children as young as eight or nine years old presenting with T2D. CHORI Assistant Staff Scientist and first-author Nancy Keller, PhD and her colleagues, however, have developed a new tool published in the March online issue of the Public Library of Science One (PLoS ONE) that can distinguish T1D and T2D quickly, based on simple observational data, including age, gender, height, and weight.

“Our tool is a model of probability that can predict which type of diabetes a patient is likely to have,” says Dr. Keller. “Our algorithm has the potential to help clinicians accurately diagnose diabetes type right there in the office with 90 percent accuracy. Currently, T2D in pediatric patients is missed 20 to 40 percent of the time.”

 

“Our algorithm has the potential to help clinicians accurately diagnose diabetes type right there in the office with 90 percent accuracy.”



Diabetes affects over 25 million people in the United States, with thousands of new cases of diabetes being reported each year in pediatric populations. While there are lab tests to differentiate type 1 from type 2 diabetes the results can take as long as two weeks to come in.

How children would be treated in the mean time, however, would be quite different depending on their diagnosis: kids with T2D would have treatment for their acute situation and be sent home with oral medication and advice on dieting and exercise, while kids with T1D would be hospitalized for two to three days filled with incredible amounts of information on how to manage their condition.

"If you could tell the difference between the two diagnoses the moment a patient walked through the door, it would not only be easier on the patient and save a lot of money, it would also preserve an incredibly valuable resource: diabetes educator time," says CHORI Scientist and co-author Janelle Noble, PhD. "These educators are already overwhelmed with their load, they don't need to spend their precious time teaching people who don't actually need to be taught about different types of insulin and how to use them."

In addition to saving money and resources, however, this new model could also help save lives by preventing a child with T1D, which is life-threatening if not properly treated, from ever being misdiagnosed as having T2D and sent home without essential care.

“Having a way to predict immediately the moment a child walks through the door whether they are more likely to have type 1 or type 2 diabetes could be incredibly valuable," says Dr. Noble.”
Developed for African American and Hispanic American populations, which suffer from some of the highest rates of T2D, the new tool uses an algorithm that combines three clinical observations: age, gender, and body mass index, or BMI, which is a calculation of weight in proportion to height. When two different algorithmic models based on clinical data for each of the two ethnic populations were compared, each model demonstrated 90 percent accuracy for its specific population, but was less likely to provide the correct diagnosis for the other population.

"What this tells us is that there are specific ethnic differences among populations, so you need to make your algorithm specific for each population," says Dr. Noble, who envisions the development of an application that would allow doctors to input the information, push a button and see a scale of probability for type 1 and type 2 diabetes, with exactly where a particular patient is on that scale.

"The clinical information is amazingly simple, but the astounding thing is that it actually works," Dr. Noble says.

While the data from the current study is quite strong, the participant numbers are small, so Drs. Keller, Noble, and their colleagues are working to develop a much larger and broader population-based survey to confirm their results. The group will then go on to develop models for other patient populations.

Back

Sunday, July 29, 2012 12:44 PM

© 2005 Children's Hospital Oakland Research Institute
5700 Martin Luther King Jr Way • Oakland, California 94609
Phone 510-450-7600 • Fax 510-450-7910
Site MapDisclaimerCHORI Intranet