LAUSR

AIM To assess agreement between fasting plasma glucose (FPG) and hemoglobin A1c (HbA1c) levels for diagnosis of dysglycemia (diabetes and risk of diabetes), overall and depending on clinical characteristics. METHODS The study enrolled 1020 adult subjects without drug-treated diabetes who underwent a laboratory test at a Spanish health care center. The criteria for dysglycemia of the American Diabetes Association were used. A logistic regression analysis was used to predict de novo diagnosis of dysglycemia based on sex, age, body mass index, anemia, and iron levels. RESULTS Overall prevalence of dysglycemia was 28.04%, and was identified by FPG only in 13.63% of subjects, by both FPG and HbA1c in 7.65%, and by HbA1c only in 6.76% (de novo diagnoses). Independent predictors of de novo diagnoses based on HbA1c were female sex (odds ratio [OR]: 2.119, 95% confidence interval [CI]: 1.133-4.020; p<0.020), age (OR for 42-56 years: 2.541, 95% CI: 0.634-17.140; OR for ≥57 years: 5.656, 95% CI: 1.516-36.980; overall p<0.007), and serum ferritin levels (borderline significance). CONCLUSIONS In this study population, agreement between FPG and HbA1c for diagnosis of dysglycemia was poor, with FPG being the test that identified more subjects. De novo diagnoses based on HbA1c were more common in females and increased with age.