To determine the degree to which species identification or strain relatedness assessment of successive blood culture isolates of coagulase-negative staphylococci (CNS) may improve the clinical diagnosis of bloodstream infection (BSI).

400-bed community hospital.

Prospective laboratory survey during which all CNS blood culture isolates obtained between mid-August 1996 and mid-February 1997 (study period) were saved and later identified to the species level; selected isolates were genotyped using pulsed-field gel electrophoresis at the Centers for Disease Control and Prevention (CDC). Retrospective review of medical records of 37 patients with multiple cultures positive for CNS.

During the study period, 171 patients had blood cultures positive for CNS; 130 had single positive cultures and 41 had ≥2 positive cultures. Of these 41, 23 (62%) were from patients with signs and symptoms of BSI according to CDC surveillance definitions. Species identification and strain clonality of CNS isolates from patients with ≥2 positives revealed 3 (13%) of the 23 patients did not have a consistent CNS species, and another 3 (13%) did not have a consistent genotype in the ≥2 positive cultures, suggesting that CNS from these patients probably were contaminants. Thus, species identification and strain clonality assessment reduced by 27% the number of patients with BSI diagnosed based on the presence of symptoms and ≥2 positive blood cultures.

Routine species identification and selected strain genotyping of CNS may reduce the misinterpretation of probable contaminants among patients with ≥2 positive blood cultures.

To describe an outbreak of Pseudomonas aeruginosa bloodstream infection (BSD and endotracheal tube (ETT) colonization in a neonatal intensive care unit (NICU), determine risk factors for infection, and make preventive recommendations.

A 15-month cohort study followed by a case-control study with an environmental survey and molecular typing of available isolates using pulsed-field gel electrophoresis.

Neonates in the NICU of a university-affiliated children's hospital.

Improved hand washing and restriction of use of long or artificial fingernails.

Of 439 neonates admitted during the study period, 46 (10.5%) acquired P aeruginosa; 16 (35%) of those died. Fifteen (75%) of 20 patients for whom isolates were genotyped had genotype A and 3 (15%) had genotype B. Of 104 healthcare workers (HCWs) from whom hand cultures were obtained, P aeruginosa was isolated from three nurses. Cultures from nurses A-1 and A-2 grew genotype A and cultures from nurse B grew genotype B. Nurse A-1 had long natural fingernails, nurse B had long artificial fingernails, and nurse A-2 had short natural fingernails. On multivariate logistic regression analysis, exposure to nurse A-l and exposure to nurse B were each independently associated with acquiring a BSI or ETT colonization with P aeruginosa, but other variables, including exposure to nurse A-2, were not.

Epidemiological evidence demonstrated an association between acquiring P aeruginosa and exposure to two nurses. Genetic and environmental evidence supported that association and suggested, but did not prove, a possible role for long or artificial fingernails in the colonization of HCWs' hands with P aeruginosa. Requiring short natural fingernails in NICUs is a reasonable policy that might reduce the incidence of hospital-acquired infections.

To investigate an outbreak of gram-negative bacteremias at a hemodialysis center (December 1, 1996-January 31, 1997).

Retrospective cohort study. Reviewed infection control practices and maintenance and disinfection procedures for the water system and dialysis machines. Performed cultures of the water and dialysis machines, including the waste-handling option (WHO), a drain port designed to dispose of saline used to flush the dialyzer before patient use. Compared isolates by pulsed-field gel electrophoresis.

A hemodialysis center in Maryland.

94 patients received dialysis on 27 machines; 10 (11%) of the patients had gram-negative bacteremias. Pathogens causing these infections were Enterobacter cloacae (n=6), Pseudomonas aeruginosa (n=4), and Escherichia coli (n=2); two patients had polymicrobial bacteremia. Factors associated with development of gram-negative bacteremias were receiving dialysis via a central venous catheter (CVC) rather than via an arterio-venous shunt (all 10 infected patients had CVCs compared to 31 of 84 uninfected patients, relative risk [RR] undefined; P<.001) or dialysis on any of three particular dialysis machines (7 of 10 infected patients were exposed to the three machines compared to 20 of 84 uninfected patients, RR=5.8; P=.005). E cloacae, P aeruginosa, or both organisms were grown from cultures obtained from several dialysis machines. WHO valves, which prevent backflow from the drain to dialysis bloodlines, were faulty in 8 (31%) of 26 machines, including 2 of 3 machines epidemiologically linked to case-patients. Pulsed-field gel electrophoresis patterns of available dialysis machine and patient E cloacae isolates were identical.

Our study suggests that WHO ports with incompetent valves and resultant backflow were a source of cross-contamination of dialysis bloodlines and patients' CVCs. Replacement of faulty WHO valves and enhanced disinfection of dialysis machines terminated the outbreak.

To study vancomycin-resistant Enterococcus (VRE) prevalence, risk factors, and clustering among hospital inpatients.

Rectal-swab prevalence culture survey conducted from February 5 to March 22,1996.

The Veterans' Affairs Medical Center, Atlanta, Georgia.

Hospital (medical and surgical) inpatients.

The overall VRE prevalence was 29% (42/147 patients). The VRE prevalence was 52% (38/73 patients) among patients who had received at least one of six specific antimicrobials during the preceding 120 days, compared with only 5% (4/74) among those who had not received the antimicrobials (relative risk, 9.6; P <.001). The longer the period (up to 120 days) during which antimicrobial use was studied, the more closely VRE status was predicted. Among 67 hospital patients in 28 multibed rooms, clustering of VRE among current roommates was not found.

At this hospital with relatively high VRE prevalence, VRE colonization was related to antibiotic use but not to roommate VRE status. In hospitals with a similar VRE epidemiology, obtaining cultures from roommates of VRE-positive patients may not be as efficient a strategy for identifying VRE-colonized patients as obtaining screening cultures from patients who have received antimicrobials.