A different approach to analyzing age-related HbA1c values in non-diabetic subjects

Using appropriate statistical tests and taking into account the analytical performance of hemoglobin A(1c) (HbA(1c)) measurements, is it useful to establish HbA(1c) agerelated values in nondiabetic subjects? Nondiabetic subjects (n=135, 72 women and 63 men) from the neuromuscular department of the PitieSalpetriere Hospital (Paris) were involved in our study. Subjects were divided into two groups related to age: 51 patients under 50 years old and 84 subjects aged 50 years or more. Fasting plasma glucose and HbA(1c) measurements were respectively performed by enzymatic assay using the hexokinase method and highperformance liquid chromatography based on the ion exchange methodology with high precision. We first checked the normality of HbA(1c) distribution using the KolmogorovSmirnov test. Then we compared mean HbA(1c) in the two age subgroups using the Students t test. Mean HbA(1c) was significantly (p<0.0001) higher in the subgroup aged 50 years or more (mean HbA(1c)=5.2%) than in younger subjects (mean HbA(1c)=5.0%). Then plots were drawn to check the relationship between HbA1c and age. Under the hypothesis of linearity, determination coefficients (R-2) were calculated. However, considering their low values, this hypothesis must be rejected and other factors than age must be retained to explain HbA(1c) variability.