AimThe utility of HbA(1c) in diabetes diagnosis is reduced in settings associated with altered haemoglobin glycation. We have studied whether HbA(1c) varies with mean cell volume and mean cell haemoglobin concentration as measures of haemoglobin metabolism. MethodsRandomly selected adults from rural Victoria, Australia, were invited for biomedical assessment. After excluding patients with known diabetes and/or serum creatinine 0.12mmol/l, 1315 adults were included. Demography, arthropometric measurements, oral glucose tolerance test, analyses of full blood count and HbA(1c) were undertaken. ResultsAfter adjusting for age, sex, ethnicity, BMI, town and socio-economic status, there were no significant differences in haemoglobin, mean cell volume or mean cell haemoglobin concentration by glycaemic status (defined by oral glucose tolerance test). HbA(1c) was significantly and independently associated with fasting glucose, town, mean cell haemoglobin concentration, ethnicity, age and BMI among men <50years (R-2=33.8%); fasting glucose, 2-h glucose, mean cell haemoglobin concentration and town among men 50years (R-2=47.9%); fasting glucose, mean cell volume, mean cell haemoglobin concentration, town, 2-h glucose and age among women <50years (R-2=46.3%); fasting glucose, mean cell haemoglobin concentration, mean cell volume and 2-h glucose among women 50years (R-2=51.6%). A generalized linear model showed a gradient from an adjusted mean HbA(1c) of 36 (95%CI 34-38) mmol/mol with a mean cell haemoglobin concentration of 320g/l to 30 (95%CI 29-31) mmol/mol with a mean cell haemoglobin concentration of >370g/l. The gradient across mean cell volume was negative, but only by 1mmol/mol (0.1%) HbA(1c). ConclusionA mean HbA(1c) difference of 5mmol/mol (0.5%) across the mean cell haemoglobin concentration reference range suggests that an accompanying full blood count examination may be required for its use in the diagnosis of diabetes. Further studies are required to confirm this.