To identify factors associated with an increased occurrence of Klebsiella pneumoniae isolation in urine cultures and infected wounds on a rehabilitation unit and to compare typing methods for K pneumoniae isolates.

Retrospective review of laboratory reports and patient records with case-control study. Analysis of K pneumoniae isolates using capsular serotyping, enzyme electrophoretic typing, ribotyping, and DNA typing.

48-bed rehabilitation unit in an 1,100-bed tertiary care teaching hospital in Winnipeg, Manitoba.

In 1988, 20 (19%) of 106 patients admitted to the rehabilitation unit had K pneumoniae isolated from urine or wound, and in 1989 31 (28%) of 111 patients had Klebsiella isolated. Review of ward practices revealed appropriate written policies but evidence of failure in execution leading to multiple opportunities for transmission among patients. Substantial environmental contamination was not identified, although a common urine graduate may have contributed to some transmission. Individuals with K pneumoniae isolated had a significantly longer duration of stay. Many of these were spinal cord-injured patients and were maintained on intermittent catheterization.

One outbreak strain was identified in epidemiologic typing. Other strains were generally identified in individuals with nonnosocomial acquisition of infection. Comparison of epidemiologic typing methods suggests ribotyping may be the optimal method for typing K pneumoniae strains.

K pneumoniae was acquired frequently by spinal cord-injured patients with extended admissions, re-emphasizing the importance of both patients and stalf following appropriate infection control practices on rehabilitation wards. Ribotyping was the optimal method for typing K pneumoniae isolates.

To investigate trends in plasmid types of methicillin-resistant Staphylococcus aureus over a 3-year period and determine the clinical impact of plasmid typing.

Restriction endonuclease analysis of plasmid DNA (REAP) was used to type 270 clinical isolates.

A Veterans Affairs Medical Center.

Forty-four different REAP types were identified and, of these, 23 appeared only once. Types designed as V1 and V2 were more commonly found in nosocomial cases than in community-acquired cases (p<.05); whereas the K and D types were more common in isolates from community-acquired cases (<c.05). In 30 (11%) isolates, no plasmid was detected. When combined with epidemiological data, REAP typing revealed 4 small outbreaks that would have been missed using traditional epidemiological methods alone. In large outbreaks involving 10 or more cases, REAP typing data revealed unsuspected patterns of transmission.

Multiple plasmid types were present in this endemic setting. Restriction endonuclease analysis of plasmid DNA was a practical and valuable adjunct to traditional epidemiological methods.

To determine the sensitivity and specificity of standard infection control surveillance techniques for the identification of surgical wound infections.

Surveillance data collected by three infection control practitioners (ICPs) was compared to surveillance data collected simultaneously by a gold standard observer.

University-affiliated, tertiary care hospital.

Using standard infection control surveillance techniques (chart review and discussions with patients' nurses and physicians), ICPs collected surveillance data on patients on the General Surgery and Trauma Surgery Services on days 4 and 7 after surgery and then weekly for 30 days or until patients were discharged from the hospital. Simultaneously, a hospital epidemiologist collected surveillance data and examined each patient's wound daily.

Nine hundred twenty-five surgical patients including 537 trauma cases and 388 elective general surgery cases were followed postoperatively. The ICPs identified 67 surgical wound infections, and the hospital epidemiologist identified 80 surgical wound infections for a sensitivity of 83.8% with a 95% confidence interval (CI95) of 75.7% to 91.9%. Specificity was 99.8% with a CI95 of 99% to 100%. The sensitivity was the same for trauma surgery and general surgery, but incisional wound infections were more difficult to identify than deep wound infections. During a second validation period, sensitivity was 92.3% with a CI95 of 62% to 100%.

Standard infection control surveillance techniques have the same sensitivity for detection of surgical wound infections as they do for identification of other nosocomial infections. Accurate data on surgical wound infections can be collected without direct examination of surgical wounds.

To describe an outbreak of Candida tropicalis fungemia in a neonatal intensive care unit (NICU), to evaluate the risk factors associated with this infection and the possible mode of nosocomial transmission.

Descriptive and case-control study

Surveillance cultures were taken from hospitalized patients, personnel, and inanimate objects in the NICU. Six patients with C tropicalis fungemia (cases) were compared with C tropicalis culture-negative patients matched for duration of exposure to the NICU (controls).

During a five-month period, C tropicalis was isolated from 29 blood cultures of six premature infants. The same organism also was isolated from fingernail samples taken from the ward housekeeper, who had a mild onychomycosis, and an asymptomatic nurse. Other potential reservoirs of C tropicalis were not identified among all the other infants or in the hospital environment. The six patients with C tropicalis fungemia were more likely to have received a larger number of antibiotics (4.0 versus 1.8, R<0.001) and to have been subjected to a longer duration of total parenteral nutrition (TPN) therapy (8.5 versus 2.67 days, P= 0.004) than the controls.

The risk of fungemia in this outbreak can be attributed to a larger number of antibiotics and a longer period of TPN administered to the patients. Analysis of events suggests that the outbreak may have been the result of cross-infection between staff and patients.