Performance of glycated albumin as a biomarker of hyperglycemia in pregnancy: Results from the National Health and Nutrition Examination Survey 1999-2004

Aim: We sought to evaluate the performance of glycated albumin (GA) as a measure of hyperglycemia in pregnant women.Methods: We used data from 555 pregnant women aged 20-40 years who participated in NHANES 1999-2004 and did not report a pre-pregnancy diagnosis of diabetes. We used Pearson's correlations and evaluated the area under the curve (AUC) for GA to detect elevated concentrations of random glucose, HbA1c, or fasting glucose (subset). We compared results to 1607 nonpregnant women aged 20-40 without diabetes.Results: In pregnant women, 1.9 % had HbA1c >= 39 mmol/mol (>= 5.7 %), 9.1 % had random glucose >= 5.3 mmol/L (>= 95 mg/dL), and 10.7 % had fasting glucose >= 5.3 mmol/L. In pregnancy, GA was poorly correlated with HbA1c (r = 0.08) and random glucose (r = 0.17). BMI was positively associated with HbA1c (r = 0.33) and random glucose (r = 0.25) but was inversely associated with GA (r =-0.27). GA had poor discrimination for detecting hyperglycemia in pregnant women, defined as HbA1c >= 39 mmol/mol (AUC = 0.634) or random glucose >= 5.3 mmol/L (AUC = 0.628). Similar patterns were observed among nonpregnant women.Conclusions: GA is not a sensitive test to screen for hyperglycemia in pregnancy. GA was inversely associated with adiposity in pregnant women without diabetes. Pregnancy-related weight gain may complicate interpretation of repeated GA measurements.