In Africa 70% of people living with diabetes (DM) are undiagnosed. To address this challenge, prediction equations must be tested in Africans. As the A1C-modified Atherosclerosis Risk in Communities (ARIC) DM prediction equation was optimized in African Americans, our goal was to test in Africans: 1) ability of the equation to detect undiagnosed DM; 2) contribution of each biochemical variable to the equation; 3) independent value of each biochemical variable. The equation has 5 clinical and 4 biochemical variables: HDL, TG, A1C and fasting plasma glucose (FPG). Participants were 421 African-born blacks living in the U.S. (66% male; age 38±10y (mean ±SD); BMI 27.5±4.4 kg/m2). OGTT was performed. DM diagnosis required 2h glucose ≥200 mg/dL. Area under the receiver operating characteristic curve (AROC) predicted DM. Youden Index estimated optimal cut points. DM occurred in 6% (27/421). AROC for the full equation was 0.84. Only with FPG excluded from the equation did the AROC decline (P<0.01) (Table 1A). Single predictor variables HDL, TG, A1C had low AROC. However, FPG as a single predictor had AROC 0.89 and was equivalent to the full equation (Table 1B). Thus, DM was identified by the full equation in 78% (21/27) and by FPG alone in 81% (22/27). Optimal cut point for FPG was 100 mg/dL. For the detection of DM, FPG ≥100 mg/dL performed as well as the A1C-modified ARIC equation.


M.F. Horlyck-Romanovsky: None. R. Mugeni: None. J.Y. Aduwo: None. S.M. Briker: None. C. DuBose: None. L. Mabundo: None. S.T. Chung: None. J. Ha: None. A. Sherman: None. A.E. Sumner: None.


National Institute of Diabetes and Digestive and Kidney Diseases; National Institute of Minority Health and Health Disparities

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at