Home / Resources / Articles / Are We Doing a Better Job of Detecting Diabetes?

Are We Doing a Better Job of Detecting Diabetes?

Mar 10, 2018

While detection of type 2 diabetes for specific subgroups in the population has slightly increased, overall detection rates remain low among US adults.

Is diabetes still underdiagnosed even in the new era of medicine, which provides us with ample resources at our fingertips to combat any single disease state? The latest study conducted by Linda Geiss and colleagues with improved research methods proves just that. Formerly, studies that explored detection rates of diabetes did so indirectly by approximating the prevalence of undiagnosed T2D. However, data from a study published last year by Ralph Brinks et al. labels the use of proportion of diagnosed cases to detect diabetes as misleading. Instead, authors advise to detect rates of diagnosis by calculating the number of undiagnosed patients amid the population that has not been diagnosed with diabetes. As a result, investigators in this study assessed trends in detection of diabetes on a national level by employing the new metric of detection.


Researchers analyzed data from the National Health and Nutrition Examination Surveys from 1999 to 2014, and included a little over 16,6500 participants over the age of 18 into the trial. Individuals who have been diagnosed with diabetes in the past and pregnant women were excluded from the participation in the study. Undiagnosed diabetes was characterized in individuals if their plasma glucose levels were above 126 mg/dL and their HbA1c was greater than 6.5%. Study participants were classified into subgroups based on age, sex, race or ethnicity, education level, and by poverty-income-ratio (PIR).  To estimate diabetes detection rates, investigators calculated probability of finding one undetected case of diabetes among population of individuals not diagnosed with T2D; higher probability of finding an undetected case corresponded to poorer detection rates.

Probability of discovering an undiagnosed T2D individual was 3% in 1999 and 2000; this number decreased to 2.8% by years of 2013 and 2014. The decreased probability was not statistically significant as shown with p-value of 0.52. When subjects were compared according to their sociodemographic groups, Mexican-American individuals were less likely to be diagnosed with diabetes. While the detection rates for this group of individuals were 3.7% in 1999–2002, this number climbed to 6.0% during 2010 to 2014. Improvement in detection was found in persons above age 65, white individuals, and adults in the highest PIR subgroup, p-values of 0.04, 0.02, and 0.047, respectively. From 2011 to 2014, detection rates of diabetes were higher for younger adults than they were for geriatric individuals; there was a 1.3% probability of finding an undiagnosed case of T2D in adults aged 18 to 44, compared to 5.6% in individuals over 65 years of age. Males were more likely to be undiagnosed with T2D compared to females; from 2011 to 2014, probability of finding an undetected T2D among males was 3.7%, while the probability was lower by over 1% amongst females. Similar values were seen in following subgroups as well: Mexican-Americans, African-Americans, lower education level, and lowest poverty-income-ratio. Overall, no significant evidence that shows improved detection trends was seen throughout the years of the study.

Although the researchers used the new and improved metric to evaluate the rates of diabetes detection, they are still not perfect. Using incidence rate for discovering the true trends of detection would be the most appropriate measure, however, that is nearly impossible to do because a true number of new diabetes cases is mostly unidentified.  Regardless, the findings made by Geiss and colleagues are alarming. By focusing on individuals in vulnerable subgroups, such as Mexican-Americans, less educated individuals, and persons of a lower socioeconomic status, rates of detection can be increased, and in turn interventions can be started early on in the disease state to avert long-term complications. Moreover, the slight increase of detection trends in certain subgroups of highest income levels, greatest education, and white individuals can be explained by those persons having better access to healthcare and, in turn, a more likelihood of being tested for diabetes.

Practice Pearls:

  • Despite advances in the medical field, diabetes remains poorly detected amongst adults in United States.
  • Mexican-Americans have a highest probability to not be diagnosed with T2D.
  • Throughout the years, improvements in detection rates were seen in elderly, white individuals, and subjects with the highest poverty-income-ratio.


Linda Geiss, Kai McKeever Bullard, Ralph Binks, et al. “Trends in type 2 diabetes detection among adults in the USA, 1999-2014.” BMJ Open Diabetes Research & Care. 2018.  http://drc.bmj.com/content/6/1/e000487.  Accessed on Jan 2018.

Ralph Brinks, Annika Hoyer, Deborah Rolka, et al.  “Comparison of surveillance-based metrics for the assessment and monitoring of disease detection: simulation study about type 2 diabetes.” BMC Medical Research and Methodology. 2017.  https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-017-0328-2. Accessed on Jan 2018.

Lamija Zimic, PharmD(c), University of South Florida, College of Pharmacy