COMPARISON OF PULSED-FIELD GEL-ELECTROPHORESIS AND RANDOMLY AMPLIFIEDDNA POLYMORPHISM ANALYSIS FOR TYPING EXTENDED-SPECTRUM-BETA-LACTAMASE-PRODUCING KLEBSIELLA-PNEUMONIAE

The incidence and transmission patterns of extended-spectrum-beta-lact amase (ESBL)-producing Klebsiella pneumoniae in patients admitted to t he intensive care unit (ICU) of a university hospital were investigate d over a 3-year period. K. pneumoniae isolates were characterized by a ntibiotic susceptibility, capsular serotyping, plasmid profiles, and p ulsed-field gel electrophoresis (PFGE) of genome macrorestriction patt erns with XbaI, and the results were compared with those obtained by t yping with the randomly amplified polymorphic DNA (RAPD) patterns. The discriminatory power of RAPD typing was evaluated for three primers. The incidence of isolation of ESBL-producing K. pneumoniae was 2.5 cas es per 1,000 admissions to the ICU versus 0.35 cases per 1,000 admissi ons to other units (relative risk, 7.03; 95% confidence interval, 3.89 to 12.69). Infection developed in 53% of evaluable patients. Thirty-s ix percent of the cases were possibly acquired in other institutions. Isolates from ICU patients were subdivided into six capsular serotypes and into four clonal groups based on antibiotype, plasmid content, an d PFGE and RAPD patterns. Two clones were associated with clusters of cross-infection, involving 5 and 12 patients, respectively. Following implementation of contact isolation precautions, the incidence of noso comial acquisition of ESBL-producing K. pneumoniae decreased from 0.55 to 0.26 cases per 1,000 admissions (P = 0.03). PFGE and RAPD analysis showed concordant results and comparable discrimination for different iation between groups of epidemiologically related strains of ESBL-pro ducing K. pneumoniae. More subclonal variants were determined among ep idemic clones by PFGE analysis than by RAPD analysis. Both methods are useful for typing K. pneumoniae strains in epidemiological investigat ions, although RAPD analysis is more efficient.