HbA1c Of 6.5% To Diagnose Diabetes Mellitus — Does It Work For Us?

Discussion

Type 2 diabetes is increasing in developing countries due to growth and aging of the population, and urbanisation with its introduction of an unhealthy diet and increasing obesity and sedentary lifestyles. In a recent article Bradshaw et al estimated that 5.5% of all South Africans ≥30 years had diabetes [10]. Levitt et al studied the prevalence of diabetes in 974 residents from the mixed ancestry (coloured) community of Mamre, a rural town near Cape Town in the Western Cape. The age standardized prevalence of type 2 diabetes in the age group 30–65 years, South African mixed ancestry population group was 10.8% and that of impaired glucose tolerance (IGT) 10.2% [11]. Age, physical inactivity, family history of diabetes and waist circumference was all identified as independent risk factors [11]. Another study that specifically examined the prevalence of diabetes in elderly coloured subjects found a prevalence of 28% for men and 29% for women [12], [13]. A population of Indian origin studied in sub-Saharan Africa found prevalences of diabetes of between 12 and 13%. This increasing incidence of diabetes with its subsequent complications in Sub-Saharan African countries such as South Africa places an even greater burden on health care systems already buckling under the challenges of diseases such as malaria, tuberculosis and HIV [14]. The incidence of diabetes in this study was more than 20% using either fasting blood glucose or an OGTT for diagnosis. For this reason, a quick and simple diagnostic test such as HbA1c would be advantageous.

HbA1c is formed by the attachment of glucose to various amino groups. The Diabetes Control and Complications Trial (DCCT), which determined HbA1c using a precise HPLC method, showed that a reduction of HbA1c led to a reduction in diabetic complications [15]. This opened the door for HbA1c standardization, as there are numerous analytical methods available for its determination [16]. In 2009, the International Expert Committee of the ADA issued a statement proposing an HbA1c value of 6.5% (48 mmol/mol) as a diagnostic level for the diagnosis of diabetes. This value was chosen, as it was found to be the value after which the incidence of retinopathy, a common complication that often is present before the actual diagnosis of diabetes is made, is increased [7]. Westgard expressed concern about using the HbA1c value as a diagnostic test, as even though the test has been “standardized”, CAP proficiency testing still shows significant biases between methods [17].

Previous studies proposed using HbA1c as a screening tool for the detection of diabetes, but their cut-off values differed [18]–[21]. In the present study, we have shown a cut-off of 6.1% (43 mmol/mol) as optimal for all mixed ancestry ages groups from South Africa. Our findings are similar to those reported by Bennett et al who performed a meta-analysis of nine studies and found that most recommended an HbA1c cut off of 6.1% (43 mmol/mol) as optimal [19]. A recent study by Kumar et al assessed the validity of HbA1c as a screening and diagnostic test for diabetes. They found a value of 6.1% (43 mmol/mol) had optimal sensitivity and could thus be used for screening and 6.5% to have optimal specificity and could thus be used as a diagnostic test [21]. In 2008, Saudek et al found an HbA1c cut off of 6.0% (42 mmol/mol) to be optimal for screening and 6.5% (48 mmol/mol) for diagnosis of diabetes [20]. Rowley et al found that a value of 7% (53 mmol/mol) had the best specificity and could therefore be used as a diagnostic test. However, this study was published in 2005 [18].

Our study has the following strengths: (1) the number of participants is large and from the same population group and (2) the diagnosis of diabetes was made using either fasting blood glucose or an OGTT and compared.

However, our study also has some limitations: (1) the haemoglobin and iron status was not determined simultaneously, and these can affect the red blood cell survival and thus render HbA1c levels unreliable; (2) the presence of renal impairment can also affect HbA1c levels and was not investigated; (3) HbA1c levels are affected by various haemoglobinopathies and thalassaemias and these were not determined; (4) the extent to which haemoglobin glycation occurs also varies between individuals and may be affected by environmental parameters, such as lipid peroxidation and hereditary factors, and (5) medication use, specifically those such as antiretrovirals which may affect glucose metabolism, was not examined Although we did not look for haemoglobinopathies and thalassaemias, their incidence is fortunately low in this population. Another possible shortcoming, but not necessary limitation, of this study was that HbA1c was not determined on an HPLC-based method as used in the DCCT and may thus be more prone to interferences. However, the assay is NGSP- certified and had acceptable CV's.

In conclusion, we recommend an HbA1c value of 6.1% (43 mmol/mol) as an optimal cut off to screen for diabetes in our local population. A cut off of 6.5% (48 mmol/mol) would be a good diagnostic tool with its high specificity (95% using fasting blood glucose and 96% using OGTT), however the low sensitivity limits the use of this value (50% using fasting blood glucose and 46% using OGTT).

Our study emphasizes the need for evidence based values to be established in various population groups.