Surveillance of Pseudomonas aeruginosa in intensive care units: clusters of nosocomial cross-infection and encounter of a multiple-antibiotic resistant strain.

Serogrouping (determination of O antigens) and bacteriocin typing (based on susceptibility to one or more of 18 bacteriocins) were employed to survey 210 isolates of Pseudomonas aeruginosa from 201 patients in 8 intensive care units (ICU) during an observation period of 18 months. Eighty-eight isolates (41.9%) were nonserogroupable (NT); most common were serogroups O1, O9, O11, and O3. All except 5 isolates (97.6%) were bacteriocin-typable. However, phenotypic variation of bacteriocin susceptibility, in particular the receptor for bacteriocin No. 13, rendered this typing method presumptive as well. Bacteriocin susceptibility profiles were not predictive of serogroup and vice versa. Workup of 19 isolates from 9 patients disclosed phenotypic variation of antibiotic susceptibility in 3 patients, superinfection by a different strain in 4 patients, and persistence (3 months) of the same strain in 2 patients, respectively. Serotyping and bacteriocin susceptibility test data revealed 15 clusters of putative cross-infection of 2 patients each, 8 clusters involving 3 patients each, one outbreak (serogroup NT, bacteriocin profile 777736) involving 4 patients in the pediatric ICU, one outbreak due to a multiple-antibiotic resistant (MAR) strain in the surgical ICU (4 patients, serogroup O12, bacteriocin profile 30400), and two putative outbreaks in the pneumonology ICU involving 6 patients (serogroup NT, bacteriocin profile 777726) and 9 patients (serogroup NT, bacteriocin profile 777736). Pulsed-field gel electrophoresis (PFGE) macrorestriction analysis (SpeI, XbaI) confirmed the pediatric and surgical ICU strains as singular strains. However, the two putative outbreaks in the pneumonology ICU were due to one particular strain which had infected 13 of the 15 patients as determined with the PFGE genotypic method. Isolates comprising the MAR strain of P. aeruginosa were susceptible only to amikacin, fosfomycin, and polymyxin B; the isolates varied in susceptibility to aztreonam and ceftazidime. This MAR strain was susceptible to the bactericidal activity of 65 vol% of fresh defibrinated human blood from donors B, L, and T. Either amikacin (16 micrograms/ ml) or fosfomycin (8 micrograms/ml) plus blood and amikacin (8 micrograms/ml) combined with fosfomycin (8 micrograms/ml) with and without blood consistently killed isolates of the MAR strain, which thus was amenable to antibiotic therapy.