Abstract
A total of 601 clinical isolates of Haemophilus influenzae isolated in six different regions of Sweden were tested for chloramphenicol susceptibility by using agar dilution MIC determinations and disk diffusion tests. For seven strains MICs were 4 micrograms/ml or higher, and for one strain the MIC was 2 micrograms/ml. All eight strains produced chloramphenicol acetyltransferase. For the remaining 593 strains, MICs were less than or equal to 1 microgram/ml, and the MICs for 50% and 90% of the strains were both 0.5 microgram/ml. Disk diffusion tests carried out by using revised interpretive criteria introduced in 1984 by the Swedish Reference Group for Antibiotics correctly identified the 593 strains as susceptible and the 8 strains as resistant. Quality assessments were performed in 29 clinical microbiology laboratories. The revised criteria for chloramphenicol disk diffusion testing gave rise to false resistance results in some laboratories. The interpretive accuracy improved when the interlaboratory variation was compensated for by using adjusted breakpoints. Such revision was possible through peak correction, single-strain regression analysis, and standard curve regression analysis. Peak-corrected breakpoints improved the accuracy from an overall incidence of false-resistant isolates of 4.4% to 2.3%. Single-strain regression analysis and standard curve regression analysis provided laboratory- and species-specific breakpoints which reduced false resistance rates of 0.14% and 0%, respectively.