BACKGROUND: Communities in rural Andhra Pradesh may be at increasing risk of diabetes. We analysed three cross-sectional studies over nine years to estimate the changing prevalence of dysglycaemia (diabetes and prediabetes). METHODS: The 2005 study sampled 4535 individuals from 20 villages. The 2010 study sampled 4024 individuals from 14 villages. The 2014 project of 62 254 individuals sought to include all adults 40 to 85 years old from 54 villages. Blood glucose levels were estimated using a hand-held device in 2005 and 2014 and using glycosylated haemoglobin dried blood spots in 2010. RESULTS: In primary analyses restricted to assays based upon fasting samples (2005 n = 3243; 2014 n = 749) the prevalence estimates for dysglycaemia were 53 . 7% (51 . 8-55 . 7) in 2005 and 62 . 0% (58 . 5-65 . 4) in 2014 (p < 0 . 001). Mean BMI rose from 22 . 2 kg/m2 to 24 . 3 kg/m2 over the same period (mean difference 2.1 95% CI 2.0-2.2, p < 0 . 001). In the secondary analyses using data from all participants (2005 n = 4535; 2010 n = 4024; 2014 n = 62 254) regardless of measurement technique, the estimated prevalence of dysglycaemia was 53 . 9% (52 . 0-55 . 9) in 2005, 50 . 5% (46 . 1-54 . 9) in 2010, and 41 . 3% (40 . 9-41 . 7) in 2014 (p < 0 .001). CONCLUSIONS: The prevalence of dysglycaemia was high at every assessment using every measurement method. Dysglycaemia in this population is most likely to have risen with the rise in BMI. The decline in prevalence suggested by the secondary analyses was likely due to confounding from the different assessment methods.
a1753-0407 (Electronic)