New Diabetes Screening Score May Be Better Than Existing Tools

Laurie Barclay, MD

November 30, 2009

November 30, 2009 — A new, easy-to-use, diabetes screening score seems better than existing methods, according to the results of a cross-sectional study reported in the December 1 issue of the Annals of Internal Medicine.

"National guidelines disagree on who should be screened for undiagnosed diabetes," write Heejung Bang, PhD, from Weill Medical College of Cornell University, Columbia University College of Physicians and Surgeons, and FOJP Service Corporation, New York, NY; and colleagues. "No existing diabetes risk score is highly generalizable or widely followed."

The goal of this study was to develop a new diabetes screening score and compare it with available screening instruments from the Centers for Disease Control and Prevention, American Diabetes Association, and US Preventive Services Task Force guidelines; American Diabetes Association risk questionnaires; and the Rotterdam model.

The new screening tool was developed in US adults 20 years or older by use of the National Health and Nutrition Examination Survey (NHANES) 1999 to 2004 and 2005 to 2006, and the tool was validated in a combined cohort of 2 community studies: the Atherosclerosis Risk in Communities (ARIC) Study and the Cardiovascular Health Study (CHS).

Using the development data set, the investigators designed a risk-scoring algorithm for undiagnosed diabetes, defined as a fasting plasma glucose level of at least 7.0 mmol/L (126 mg/dL) in patients without known diabetes. Identification of participant characteristics that were independently associated with undiagnosed diabetes was accomplished with logistic regression. Standard diagnostic and feasibility measures were used to assess the new algorithm and other methods.

Factors associated with undiagnosed diabetes were age, sex, family history of diabetes, history of hypertension, obesity, and physical activity. Using a cutoff value of 5 or more points, the score selected 35% of persons in NHANES and 40% of persons in ARIC/CHS for diabetes screening. Sensitivity was 79% in NHANES and 72% in ARIC/CHS; respectively, specificity was 67% and 62%, positive predictive value was 10% and 10%, and positive likelihood ratio was 2.39 and 1.89. For the comparison scores, sensitivity ranged from 44% to 100%, specificity from 10% to 73%, positive predictive value from 5% to 8%, and positive likelihood ratio from 1.11 to 1.98.

"This easy-to-implement diabetes screening score seems to demonstrate improvements over existing methods," the study authors write. "Studies are needed to evaluate it in diverse populations in real-world settings."

Limitations of this study include lack of data during pregnancy and inability to incorporate oral glucose tolerance test results.

"A self-assessment method that helps persons decide whether they should seek medical care for diabetes testing may address the lack of interaction with health care facilities and providers, which may underlie the high percentage of the population with undiagnosed diabetes, particularly underserved persons," the study authors conclude. "In addition to its use in clinical encounters, targeted screenings, and health education programs, the screening score can be applied by health plans to existing databases for case finding. The new algorithm may also help identify optimal populations for enrollment in clinical trials that test new strategies to prevent or manage diabetes."

The Clinical and Translational Science Center at Weill Cornell Medical College funded this study. The NHANES were supported by the Centers for Disease Control and Prevention. The ARIC and CHS studies were supported by the National Heart, Lung, and Blood Institute. The conclusions of the study authors do not necessarily reflect the opinions or views of these studies or agencies. One of the study authors has disclosed various financial relationships with Merck & Co. Another study author has received honoraria from Lifetime Healthcare.

Ann Intern Med. 2009;151:775-783.