Glycosuric tests should not be employed in population screenings for NIDDM.

BACKGROUND The aim was to evaluate self-testing for glycosuria in screening for diabetes mellitus (DM). METHODS All inhabitants aged 45-76 years in the investigation area were invited (3041 individuals). Participants received two foil-wrapped dipsticks and were asked to examine their postprandial urine. Results were marked on the reply card and returned. Test-positive screenees were offered a fasting blood glucose test at the laboratory, as were a random sample of 143 test-negative screenees. RESULTS The screening detected 15 cases of unrecognized DM. Among 106 test-negative screenees, 3 had DM and 4 had impaired glucose tolerance. Response rate was 76.9 per cent. Sensitivity was 20.80 per cent. Specificity was 99.14 per cent. Positive predictive value was 46.88 per cent. Negative predictive value was 97.17 per cent. Costs per new case of diabetes were 3155 DKK (approximately pounds 332). CONCLUSION The test has such a low sensitivity that it cannot be recommended as a screening test. If screening for DM is to be performed, it should not be on the basis of a glycosuric test. At least 11 of the 15 cases of screening-detected diabetes belonged to a risk group, and might have been detected by selective screening. We suggest that screening be carried out in general practice using blood sugar tests on risk groups.