Prognostic Implications of Single-Sample Confirmatory Testing for Undiagnosed Diabetes: A Prospective Cohort Study

Abstract

Background: Current clinical definitions of diabetes require repeated blood work to confirm elevated levels of glucose or hemoglobin A1c (HbA1c) to reduce the possibility of a false-positive diagnosis. Whether 2 different tests from a single blood sample provide adequate confirmation is uncertain.

Objective: To examine the prognostic performance of a single-sample confirmatory definition of undiagnosed diabetes.

Design: Prospective cohort study.

Setting: The ARIC (Atherosclerosis Risk in Communities) study.

Participants: 13 346 ARIC participants (12 268 without diagnosed diabetes) with 25 years of follow-up for incident diabetes, cardiovascular outcomes, kidney disease, and mortality.

Measurements: Confirmed undiagnosed diabetes was defined as elevated levels of fasting glucose (≥7.0 mmol/L [≥126 mg/dL]) and HbA1c (≥6.5%) from a single blood sample.

Results: Among 12 268 participants without diagnosed diabetes, 978 had elevated levels of fasting glucose or HbA1c at baseline (1990 to 1992). Among these, 39% had both (confirmed undiagnosed diabetes), whereas 61% had only 1 elevated measure (unconfirmed undiagnosed diabetes). The confirmatory definition had moderate sensitivity (54.9%) but high specificity (98.1%) for identification of diabetes cases diagnosed during the first 5 years of follow-up, with specificity increasing to 99.6% by 15 years. The 15-year positive predictive value was 88.7% compared with 71.1% for unconfirmed cases. Confirmed undiagnosed diabetes was significantly associated with cardiovascular and kidney disease and mortality, with stronger associations than unconfirmed diabetes.

Limitation: Lack of repeated measurements of fasting glucose and HbA1c.

Conclusion: A single-sample confirmatory definition of diabetes had a high positive predictive value for subsequent diagnosis and was strongly associated with clinical end points. Our results support the clinical utility of using a combination of elevated fasting glucose and HbA1c levels from a single blood sample to identify undiagnosed diabetes in the population.

Primary funding source: National Institute of Diabetes and Digestive and Kidney Diseases and National Heart, Lung, and Blood Institute.

Figure 1

Cumulative incidence of diagnosed diabetes after ARIC visit 2 (1990 to 1992), adjusted for age, sex, and race–center. ARIC = Atherosclerosis Risk in Communities.

Figure 2

Cumulative incidence of chronic kidney disease, cardiovascular disease, peripheral artery disease, and all-cause mortality after ARIC visit 2 (1990 to 1992), adjusted for age, sex, and race–center. ARIC = Atherosclerosis Risk in Communities.

Appendix Figure 1

Selection of analytic study populations from the original ARIC study cohort, visit 2 (1990 to 1992). ARIC = Atherosclerosis Risk in Communities; HbA_1c = hemoglobin A_1c.

Appendix Figure 2

Scatter plot of HbA_1c and fasting glucose levels in persons with no history of diagnosed diabetes who attended ARIC visit 2 (1990 to 1992). The figure is divided into 4 quadrants (a, b, c, and d) according to diagnostic cut points for fasting glucose and HbA_1c levels. On the basis of the number of persons in these quadrants, the overall percent agreement is 95%, defined as 100% × ([b + c]/[a + b + c + d]). The positive percent agreement is 37.5%, defined as 100% × (b/[a + b + d]). The Pearson correlation coefficient is 0.72; the Spearman correlation coefficient is 0.43. ARIC = Atherosclerosis Risk in Communities; HbA_1c = hemoglobin A_1c.