Intermediate hyperglycaemia to predict progression to type 2 diabetes (ELSA-Brasil): an occupational cohort study in Brazil.

BACKGROUND: The burden of diabetes is increasing worldwide and diabetes can be prevented with intervention in people with impaired glucose tolerance (IGT). Intermediate hyperglycaemia defined without an oral glucose tolerance test as impaired fasting glucose (IFG) and high HbA1c are also used to characterise risk. We aimed to assess the prognostic properties of five definitions of intermediate hyperglycaemia (also known as prediabetes) on the basis of their ability to predict who will progress to diabetes.
METHODS: The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) is an occupational cohort study of active or retired civil servants, aged 35-74 years, recruited from public universities and research institutes in six state capital cities in Brazil. We excluded participants who provided insufficient information to ascertain diabetes status, those without information on relevant covariates, and those with diabetes. We classified type 2 diabetes on the basis of self-report, medication use, measures of fasting plasma glucose (FPG), 2 h plasma glucose, and HbA1c. We used five laboratory definitions of intermediate hyperglycaemia: IGT (2 h plasma glucose ≥7·8 mmol/L [≥140 mg/dL]); IFG based on American Diabetes Association (ADA) criteria (FPG ≥5·5 mmol/L [≥100 mg/dL]); IFG based on WHO criteria (FPG ≥6·1 mmol/L [≥110 mg/dL]); HbA1c based on ADA criteria (HbA1c ≥39 mmol/mol [5·7%]); and HbA1c based on International Expert Committee criteria, IEC-HbA1c, (HbA1c ≥42 mmol/mol [6·0%]). We estimated risk of each definition using Cox regression and overall predictability (area under the receiver operating characteristic curve [AUC]) using logistic regression.
FINDINGS: We recruited 15 105 participants from Aug 18, 2008, to Dec 20, 2010, and followed up for a mean of 3·7 (SD 0·63) years. Diabetes incidence rate was 2·0 per 100 person-years (95% CI 1·8-2·1). Among the 11 199 eligible participants, 6563 (59%) presented with some form of intermediate hyperglycaemia. ADA-IFG (4870/11 199 [43·5%), IEC-HbA1c (1005 [9·0%]), and ADA-HbA1c (2299 [20·5%]) poorly predicted diabetes (3·5-3·6 per 100 person-years). WHO-IFG (1140 [10·2%]) and IGT (2245 [20·0%]) predicted greater conversion (7·5 per 100 person-years and 5·8 per 100 person-years, respectively). All definitions presented either low sensitivity or specificity. Combinations of tests improved prognostic properties, with the combination of IGT or WHO-IFG showing the best, but still insufficient, predictability (sensitivity 67·7%, 95% CI 64·5-70·1; specificity 77·9%, 77·1-78·7). The AUC for the three underlying glycaemic tests was 65·0% (95% CI 63·0-66·9) for HbA1c, 74·6% (72·7-76·4) for FPG, and 77·1% (75·4-78·8) for 2 h plasma glucose, whereas the AUC for a score composed of clinical information was 71·6% (69·8-73·3). When this score was combined with results of an oral glucose tolerance test, the AUC reached 82·4% (80·9-83·9).
INTERPRETATION: IFG based on WHO criteria and IGT predict diabetes progression better than do the other three definitions of intermediate hyperglycaemia, but their sensitivity is low. IFG based on ADA criteria has better sensitivity than the others, but classifies almost half of adults as having intermediate hyperglycaemia and poorly predicts diabetes. Combining glycaemic results with clinical information improves prognostic properties of those at risk. FUNDING: The Brazilian Ministry of Health (Science and Technology Department), the Brazilian Ministry of Science, Technology and Innovation (Financiadora de Estudos e Projetos and Conselho Nacional de Desenvolvimento Científico e Tecnológico), and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-Brasil (CAPES).