Community screening for non-insulin-dependent diabetes mellitus: self-testing for post-prandial glycosuria

Recent requirements of the GP contract have included the testing of random urine samples from all patients aged 16-75 years for glycosuria. As a means of detecting diabetes this will be both inefficient and wasteful of resources, because of the method employed and age group screened. We examined a simple method for screening large populations for non-insulin-dependent diabetes mellitus using self-testing for post-prandial glycosuria, and compared its performance to that using fasting plasma glucose and haemoglobin A1. The study involved five general practices, screening over 13,000 subjects aged between 45 and 70 years. Of the 13,795 subjects screened, 10,348 responded (75%). Glycosuria was detected in 343 subjects (3.3%), of whom 330 (95.9%) attended for oral glucose tolerance testing. Of these 99 (30%) had newly diagnosed diabetes, 56 (17%) had impaired glucose tolerance. A further 65 subjects were identified as having diabetes but were not on a register of cases. Prevalence of diabetes in this age group rose from 2.26% to 4.1%. In a normal population control group (442 subjects from Ely), this system gave a sensitivity of 43%, specificity of 98% and positive predictive value of 53%, and compared favourably to the use of a fasting plasma glucose of 6 mmol/l, which gave a sensitivity of 65%, but only a specificity of 64% and positive predictive value of 9%. These results support the use of this simple system in large population screening for diabetes.