Also, minor errors (any false VX-680 concentration result involving an intermediate result), major errors (false-resistant results) and very major errors (false-susceptible results) were calculated. Statistical analysis Bacterial load in ID broth for GPC and GNR was compared using an independent samples t-test. Results Inoculum of bacteria in ID broth after use of serum separator tubes (SSTs) In total, 134 blood cultures were included,

from 116 patients. The inoculum of GPC in ID broth was on average 3.6 × 107 CFU/ml, whereas that of GNR was 1.8 × 108 CFU/ml, which was a TSA HDAC molecular weight significant difference (95% CI between -1.7 × 108 and -1.2 × 108; P < 0.001). ID of GNR with the direct Phoenix method NSC23766 cost ID with direct inoculation was correct for 95.2% of all tested Enterobacteriaceae. One Escherichia coli strain was incorrectly identified as Salmonella choleraesuis with the direct method. One Serratia marcescens strain could not be identified with the direct method. Identification for Pseudomonas spp. was correct in 71.4%. Both errors in this group involved strains of Pseudomonas aeruginosa that were incorrectly identified as Pseudomonas fluorescens (Table 1). No errors in ID were observed

for the routine method. Table 1 Results of identification of GNR with the direct method   Total no. of strains No. of unidentified strains No. of misidentified strains ID of misidentified strains Enterobacteriaceae         E. coli 26   1 Salmonella choleraesuis K. pneumoniae spp. pneumoniae 8       S. marcescens 4 1     K. oxytoca 1       P. mirabilis 1  

    E. cloacae 1       M. morganii 1       Non-fermenters         P. aeruginosa 7   2 Pseudomonas fluorescens Antibiotic susceptibility testing (AST) of GNR Results of AST were available for 49 strains, one P. aeruginosa strain failed to grow sufficiently in the Phoenix system so no results were available for the direct method. Categorical agreement of the direct method with results of the standard method for GNR was 97.6%. After discrepancy analysis of the results of AST, this percentage rose to 99.0%, with 5 minor errors (0.7%), no major errors, the and 2 very major errors (0.3%) (Table 2). Both very major errors occurred with trimethoprim-sulfamethoxazole in Pseudomonas aeruginosa strains. Categorical agreement of the standard method after discrepancy analysis was 98.4% (table 2). One very major error occurred with trimethoprim-sulfamethoxazole. No antibiotic showed a categorical agreement of <95% (Table 3). Table 2 Agreements and errors for AST of GPC and GNR for the direct and routinely used Phoenix method   Direct vs routinely used method Direct method after discrepancy analysis Routine method after discrepancy analysis GPC (n = 84)       Categorical agreement 93.1% 95.4% 97.3% Minor errors 1.7% 1.1% 0.7% Major errors 4.2% 3.1% 0.8% Very major errors 0.9% 0.4% 1.