Aim To determine which prediabetes definition is the best predictor of progression to diabetes in Reunion Island where 10% of the population has treated diabetes.

Methods This follow-up study used data from the REDIA cross-sectional study, a population-based study conducted in two stages. Participants were enrolled in 1999–2001 (REDIA-1) and followed-up years later in 2006–2009 (REDIA-2). Odds ratios (OR) for prediabetes were estimated with their 95% confidence interval (95%CI) from logistic regression models. REDIA-1 participants with no previously identified diabetes in REDIA-1 were assessed for new pharmacologically treated diabetes in REDIA-2. We evaluated several biological definitions of prediabetes, each based on the combination of fasting plasma glucose (FPG), 2-h post-load plasma glucose (2hPG), and/or HbA1c: FPG-ADA (American Diabetes Association), FPG-WHO (World Health Organization), 2hPG, HbA1c-ADA, HbA1c-IEC (International Expert Committee), FPG-WHO/2hPG, and FPG-WHO/HbA1c-ADA.

Results A total of 432 participants met all inclusion criteria. Of these, 102 (23.6%) were classified as prediabetic using the FPG-WHO/2hPG definition, including 58 (56.9%) with isolated impaired glucose tolerance, 17 (16.6%) with isolated impaired fasting glucose, and 27 (26.5%) with both impairments. A total of 54 (12.5%) participants became treated diabetics and in descending order, the ORs for the FPG-WHO/2hPG, 2hPG, FPG-WHO, FPG-WHO/HbA1c-ADA, FPG-ADA, HbA1c-ADA, and HbA1c-IEC definitions were 6.96 [3.72–13.03], 5.91 [3.24–10.77], 5.82 [2.86–11.81], 4.68 [2.38–9.19], 4.37 [2.34–8.17], 3.24 [1.72–6.10], and 2.74 [1.32–5.70], respectively.

Conclusion The FPG-WHO/2hPG definition had the highest strength of association with the progression to treated diabetes, closely followed by the 2hPG and FPG-WHO definitions. Our findings highlight the importance of performing both FPG test and OGTT to diagnose prediabetes in primary care.