Abstract

Asymptomatic urinary tract infections (UTI) are frequent in hospital patients. Two prospective studies were carried out to evaluate the reagent strip method used to detect these infections. The diagnosis of UTI rested on cytobacteriological examination of urine. Detection by reagent strip was performed either in the microbiology laboratory (study 1:976 samples) or at the patient's bedside and in the laboratory (Study 2:312 samples). The strips were examined visually and by an automated device. There was no significant difference between the results obtained with these two reading methods. Study 1 showed that the reagent strip is more useful to exclude the diagnosis of UTI by using the leucocyte zone alone or combined with the nitrite zone (negative predictive value 98 percent) than to assert this diagnosis. Study 2 showed that the nitrite zone was less sensitive when it was examined at the laboratory than at the patient's bedside. Analysis of discordances between reagent strip and cytobacteriology in the light of data supplied by the patient's medical records made it possible to improve the negative predictive value of the tests. (100 percent). Thus, 60 percent of cytobacteriological examinations might be avoided if this type of examination was performed only in urine with positive leucocyte or nitrite zones. This means a 15 percent saving in B-coded examinations performed in hospital microbiology laboratories.