Correlation between glycosylated serum albumin and glycosylated haemoglobin in the southwest Chinese population: Establishment of a regression model.

<bold>Aims: </bold>To correlate glycated albumin (GA) and glycosylated haemoglobin (HbA1c) and establish a novel formula for estimating HbA1c from GA.<bold>Methods: </bold>We retrospectively enrolled 20,381 cases and excluded HbA1c and GA outliers by residual analysis. HbA1c ranged from 4.0-12.0% and GA from 7.5-45%. The HbA1c range of 4.0-8.0% in both sexes was stratified into eight groups with an increase of 0.5%, and the means of GA and HbA1c were compared. HbA1c was divided into 38 groups with increments of 0.1% (range, 4.3-8.0%), and the correlation between HbA1c and GA was investigated.<bold>Results: </bold>There was no significant sex-based difference between HbA1c and GA. The analysis showed that when HbA1c was 6.2% or GA was 12.28%, the linear relationship between the two parameters was not continuous. When HbA1c was <6.2% or GA < 12.28%, we devised the formula: HbA1c = 1.136 × GA - 7.289 (R2 = 0.824). For HbA1c ≥ 6.2% or GA ≥ 12.28%, the equation was: HbA1c = 0.252 × GA + 3.163 (R2 = 0.948).<bold>Conclusion: </bold>A discontinuous linear relationship exists between HbA1c and GA when HbA1c is 6.2% or GA is 12.28%, although with a significant turning point. The GA value can be used to estimate the HbA1c value with the two regression equations to accurately estimate the long-term average blood glucose level of patients. [ABSTRACT FROM AUTHOR]