Background: Glycated hemoglobin (HbA1c) is an indirect method for estimating mean glycemia, but several factors may interfere with glucose and HbA1c measurement.
Aim: To evaluate the association between estimated average glucose (eAG) and fasting plasma glucose (FPG) in outpatient population from our hospital and to assess possible interfering factors in that association.
Methods: Retrospective study including samples with concomitant measurement of hemoglobin (Hb), HbA1c and FPG from outpatient adults followed in Centro Hospitalar São João. Only the last sample of each patient was included in analysis. FPG was performed by an enzymatic-ultraviolet assay and HbA1c by ionic exchange high-performance liquid chromatography, according to international standards. The eAG (mg/dL) was calculated by the formula 28.7xHbA1c(%)-46.7.
Results: We included 24 380 patients with simultaneous measurement of Hb, HbA1c and FPG, 55.9% female, mean age 59.96±15.96 years, Hb 13.48±1.75 g/dL, HbA1c 6.25±1.20%, eAG 132.71±37.45 mg/dL and FPG 115.56±47.76 mg/dL. Male, patients > 65 years, with anemia and those with stage 3 or 4 chronic kidney disease had significant higher HbA1c, eAG and FPG levels. We reported a strongly positive correlation between AG and FPG (r=0.711; p<0.001). The correlation was stronger in female (r=0.729; r=0.688; z=6.391; p<0.001), patients < 65 years (r=0.738; r=0.664, z=11.293; p<0.001) and in those without anemia (r=0.738; r=0.626; z=13.53; p<0.001). In multivariate analysis model including age, sex, Hb and glomerular filtration rate as potential confounding factors, the FPG was a strong predictor of eAG (β=0.697; 95%CI 0.540-0.554; p<0.001); for each 1 mg/dL increase in FPG there was an increase of 0.697 mg/dL in eAG, keeping constant the other variables in the model. FPG, Hb, age and GFR allowed to explain 50% of the eAG variation (r2=0.519).
Discussion: This is one of the largest observational studies regarding the association between eAG and FPG. To our best knowledge, this is the first Portuguese study concerning this association. Although several interfering factors, FPG was a strong predictor of eAG, even in multivariate analysis. However, eAG and HbA1c are difficult to predict from FPG values alone.

Estimated average glucose and fasting plasma glucose in outpatient adults from a tertiary Portuguese Hospital