The cumbersome oral glucose tolerance test (OGTT) remains the reference standard to diagnose dysglycaemia, while the performance of alternative methods based on glycated proteins in African populations is unknown. We assessed the ability of fasting blood glucose (FBG), HbA1c and fructosamine, singly or in combination to diagnose OGTT-based abnormal glucose tolerance in mixed-ancestry South Africans.Design and method:
Participants were selected from the Bellville-South community in Cape Town. HbA1c, fructosamine, FBG and 2-hour post-OGTT glucose were measured. OGTT values were used to classify participants based on the World Health Organisation 1999 criteria. Robust correlations and segmented regression were used to assess the association between indices of glucose homeostasis, and their ability to detect abnormal glucose tolerance assessed using area under the curve (AUC) analysis.Results:
Of the 840 participants, 113 (13.5%) had known diabetes, 61 (7.3%) had screen-detected diabetes, 137 (16.3%) had prediabetes and 527 (62.9%) had normal glucose tolerance. Correlations of FBG and 2-hour glucose with HbA1c were stronger than those with fructosamine, and significant breakpoints were apparent in the relationship among these markers (Figure 1). The highest AUC for the prediction of any category of abnormal glucose tolerance was recorded with 2-hour glucose and the lowest with fructosamine. For screen-detected diabetes AUCs were 0.996 (95%CI: 0.991–1.000) for 2-h glucose, 0.918 (0.0871–0.966) for FBG, 0.897 (0.841–0.954) for HbA1c and 0.864 (0.811–0.917) for fructosamine (all p < 0.001 for comparison with 2-h glucose). At both optimal and published cut-offs, no combination of FBG, HbA1c and fructosamine did better than 2-hour glucose alone, while FBG alone had substantial advantages over HbA1c and fructosamine on a range of performance measures. At optimal threshold, 2-h glucose correctly diagnosed 97% of screen-detected diabetes. Equivalents figures were 81% for FBG and HbA1c, and 73% for fructosamine. Accompanying accuracy was 98.7%, 91.8%, 85.9% and 83.3%.Conclusions:
Abnormal glucose tolerance in this population is overwhelmingly expressed through abnormalities of 2-hour glucose. No combination of FBG, HbA1c and fructosamine was effective at accurately discriminating participants with OGTT-defined abnormal glucose tolerance. Non-glucose based strategies are unreliable alternatives to OGTT for dysglycaemia diagnosis in this population.