[Reliability of the bacterial identification tests and of antibiotic sensitivity tests made with automatic methods].

A "MS-2 Microbiology System" (Abbott) has been used to identify and to assess the susceptibility patterns of clinically isolated bacteria from highly compromised patients. Automatically obtained results by the analysis of 300 specimens were compared to those of traditional methods. The automated system was in good agreement with API (81%) while in 4% of the strains there were major discrepancies, and 5% of the experiments were carried out with slow growing bacteria which normally give poor results in automatic systems. Antimicrobial susceptibility testing was satisfactory in 84% of the specimens with 7% of minor discrepancies and 9% of major ones. The evaluation of MIC for slow growing organisms like Haemophilus, Diplococcus and anaerobes belonging to the Bacteroides genus, was the same using either automatic or standard procedures.