Screening for diabetes mellitus in general practice using a reflectance meter system. The Islington Diabetes Survey

The feasibility of using a blood-glucose reflectance meter to identify subjects with abnormal glucose tolerance in a community-based screening programme for diabetes mellitus has been assessed. 530 (51.0%) of 1,040 screened subjects had a finger-prick capillary blood-glucose level estimated 2 h after a 75 g oral glucose load. Blood-glucose estimations were performed on this sample by an automated glucose-oxidase method and by a reflectance meter system (Reflocheck, BCL (London) Ltd., Lewes, East Sussex, UK). Blood-glucose values measured by the meter were found to be higher than those assayed by the glucose-oxidase method by a mean ratio of 1.16:1. This overestimation could not wholly be explained by the meter being calibrated to measure plasma- rather than whole blood-glucose. This ratio was constant in specimens assayed by the glucose-oxidase method within 24 h of sampling or within 7 days of sampling. The meter system gave a good approximation of the prevalence of DM in the sample (14 "diabetics" identified by the meter, 15 by the glucose-oxidase 2hBG). The optimal test characteristics of the meter 2hBG as a screening test for DM were at a cut-off level of 9.7 mmol/l (sensitivity and specificity 98.4%, CI 85.8-100%, for sensitivity, 97.3-99.5% for specificity). However, in order to be 100% sensitive for DM, confirmatory GTTs would be necessary on all subjects with meter 2hBG greater than 9.5 mmol/l, providing a specificity of 98.2% (CI 97.0-99.7%). The unit cost of detecting 1 case of DM using a single meter became cheaper than using the laboratory glucose-oxidase method when 143 subjects had been screened.