Screening for diabetes in Indigenous populations using glycated haemoglobin: sensitivity, specificity, post-test likelihood and risk of disease

Aims Screening for diabetes using glycated haemoglobin (HbA(1c)) offers potential advantages over fasting glucose or oral glucose tolerance testing. Current recommendations advise against the use of HbA(1c) for screening but test properties may vary systematically across populations, according to the diabetes prevalence and risk. We aimed to: (i) characterize the properties of test cut-offs of HbA(1c) for diagnosis of diabetes relative to a diagnosis based on a fasting plasma glucose concentration of 7.0 mmol/l for high-risk Indigenous populations; and (ii) examine test properties across a range of diabetes prevalence from 5 to 30%. Methods Data were collected from Aboriginal and Torres Strait Islander communities in Australia and a Canadian First Nations community (diabetes prevalence 12-22%) in the course of diabetes diagnostic and risk factor screening programmes (n = 431). Screening test properties were analyzed for the range of HbA(1c) observed (3-12.9%). Results In separate and pooled analyses, a HbA(1c) cut point of 7.0% proved the optimal limit for classifying diabetes, with summary analysis results of sensitivity = 73 (56-86)%, specificity = 98 (96-99)%, overall agreement (Youden's index) = 0.71, and positive predictive value (for an overall prevalence of 18%) = 88%. For diabetes prevalence from 5 to 30% the post-test likelihood of having diabetes given HbA(1c) = 7.0% (positive predictive value) ranged from 62.7 to 93.2%; for HbA(1c) < 7.0%, the post-test likelihood of having diabetes ranged from 4.5 to 27.7%. Conclusions The results converge with research on the likelihood of diabetes complications in supporting a HbA(1c) cut-off of 7.0% in screening for diabetes in epidemiological research. Glycated haemoglobin has potential utility in screening for diabetes in high-risk populations.