In 2010, glycated haemoglobin A1c (HbA1c) was officially recommended as a screening tool to diagnose type 2 diabetes mellitus (T2DM) and pre-diabetes, with cut-off points 6.5% and 5.7% to 6.4% respectively. The implications of using the HbA1c criterion, compared to the general diagnostic criteria: fasting glucose test (FPG) and oral glucose tolerance test (OGTT), is however still being debated.
The objectives of this study were to evaluate and compare the pooled prevalence of type 2 diabetes mellitus (T2DM) and pre-diabetes, as measured by the Haemoglobin A1c (HbA1c) test, or the fasting plasma glucose (FPG) and oral glucose tolerance test (OGTT). Secondly, to determine and compare the diagnostic test characteristics (sensitivity, specificity) of these tests.
Published papers, with a cross sectional study design, were selected for a systematic review and meta-analysis. The search strategy was an electronic review of journal articles listed on MEDLINE, PubMed and Google scholar between 1996 and 2012. Reference lists were checked, journals were hand searched and experts were contacted when necessary. Initially all studies related to the validation of HbA1c as a tool to detect pre-diabetes or T2DM in humans, published in English, were examined.
Studies were excluded if they did not meet the above mentioned criteria, and/or were conducted with pregnant women. Further analysis was done if FPG or OGTT was compared to HbA1c. The diagnosis of diabetes had to have been based on ADA or WHO criteria. These criteria are: HbA1c 5.7%-6.4% for pre-diabetes and >=6.5% for T2DM; FPG 5.6mmol-7mmol/l for pre-diabetes and >=7mmol/l for T2DM; OGTT 7.8mmol-11.1mmol/l for pre-diabetes and >=11.1mmol/l for T2DM). The OGTT and FPG tests were used as the reference tests and the prevalence reflected as a positive or negative proportion.
The sensitivity and specificity of HbA1c >=6.5% among cases defined by OGTT or FPG should have been reported, or it was possible to calculate these from the data provided. Study results relating to diagnostic accuracy were extracted and synthesized using multivariate random effects meta-analysis methods. This study focused on patients who were suspected of having T2DM, from two sub-groups (a community-based group and a high-risk group) to compare the detection rate of HbA1c with FPG and OGTT.