Abstract

AIMS--To evaluate the accuracy and reproducibility of multipoint identification schemes in a multicentre trial. METHODS--Forty two strains of Enterobacteriaceae were distributed to 22 laboratories for identification by routine multipoint methods. Analysis of results enabled inter- and intralaboratory reproducibility of a variety of tests, and the ability of laboratories to identify individual organisms to be determined. RESULTS--Interlaboratory reproducibility of most of the biochemical tests was acceptable. The least reproducible tests, both within and between laboratories, were citrate utilisation, production of urease and beta galactosidase, detection of motility, and decarboxylation of lysine and ornithine. Inconsistent results for these tests were often associated with misidentified strains. Most laboratories performed identifications satisfactorily. Most isolates (72.1%) were identified correctly to species level; 9.6% were incorrectly identified, and 6.4% could not be identified at all. The most difficult organisms to identify were Citrobacter freundii, Enterobacter cloacae, Hafnia alvei and Aeromonas hydrophila. Strains of Enterobacter, Serratia sp, and Providencia sp were difficult to speciate. Several laboratories could not identify organisms exhibiting at least one atypical biochemical reaction. CONCLUSION--This study emphasises the need for quality control of media and reagents for multipoint identification of Gram negative enteric bacilli.