Hospital-acquired legionnaires' disease can be prevented by disinfection of hospital water systems. This study assessed the long-term efficacy of copper-silver ionization as a disinfection method in controlling Legionella in hospital water systems and reducing the incidence of hospital-acquired legionnaires' disease. A standardized, evidence-based approach to assist hospitals with decision making concerning the possible purchase of a disinfection system is presented.

The first 16 hospitals to install copper-silver ionization systems for Legionella disinfection were surveyed. Surveys conducted in 1995 and 2000 documented the experiences of the hospitals with maintenance of the system, contamination of water with Legionella, and occurrence of hospital-acquired legionnaires' disease. All were acute care hospitals with a mean of 435 beds.

All 16 hospitals reported cases of hospital-acquired legionnaires' disease prior to installing the copper-silver ionization system. Seventy-five percent had previously attempted other disinfection methods including superheat and flush, ultraviolet light, and hyperchlorination. By 2000, the ionization systems had been operational from 5 to 11 years. Prior to installation, 47% of the hospitals reported that more than 30% of distal water sites yielded Legionella. In 1995, after installation, 50% of the hospitals reported 0% positivity, and 43% still reported 0% in 2000. Moreover, no cases of hospital-acquired legionnaires' disease have occurred in any hospital since 1995.

This study represents the final step in a proposed 4-step evaluation process of disinfection systems that includes (1) demonstrated efficacy of Legionella eradication in vitro using laboratory assays, (2) anecdotal experiences in preventing legionnaires' disease in individual hospitals, (3) controlled studies in individual hospitals, and (4) validation in confirmatory reports from multiple hospitals during a prolonged time (5 to 11 years in this study). Copper-silver ionization is now the only disinfection modality to have fulfilled all four evaluation criteria.

To measure the association between the disinfection of municipal drinking water with monochloramine and the occurrence of hospital-acquired legionnaires' disease (LD).

One hundred sixty-six U.S. hospitals.

Survey of 459 members of the Society for Healthcare Epidemiology of America (SHEA) for hospital features; endemic- and outbreak-related, hospital-acquired LD; the source of the hospital water supply; and the methods of disinfection used by the hospitals and municipal water treatment plants.

SHEA members representing 166 (36%) of 459 hospitals responded; 33 (20%) reported one or more episodes of hospital-acquired LD during the period from 1994 to 1998 and 23 (14%) reported an outbreak of hospital-acquired LD during the period from 1989 to 1998. Hospitals with an occurrence of hospital-acquired LD had a higher census (median, 319 vs 221; P = .03), more acute care beds (median, 500 vs 376; P = .04), and more intensive care unit beds (median, 42 vs 24; P = .009) than did other hospitals. They were also more likely to have a transplant service (74% vs 42%; P = .001) and to perform surveillance for hospital-acquired disease (92% vs 61%; P = .001). After adjustment for the presence of a transplant program and surveillance for legionnaires' disease, hospitals supplied with drinking water disinfected with monochloramine by municipal plants were less likely to have sporadic cases or outbreaks of hospital-acquired LD (odds ratio, 0.20; 95% confidence interval, 0.07 to 0.56) than were other hospitals.

Water disinfection with monochloramine by municipal water treatment plants significantly reduces the risk of hospital-acquired LD.

To assess the safety and efficacy of a chlorine dioxide water treatment system in controlling Legionella in a hospital water supply.

For 17 months following installation of the system, we performed regular water cultures throughout the building, assessed chlorine dioxide and chlorite levels, and monitored metal corrosion.

Sites that grew Legionella species decreased from 41% at baseline to 4% (P = .001). L. anisa was the only species recovered and it was found in samples of both hot and cold water. Levels of chlorine dioxide and chlorite were below Environmental Protection Agency (EPA) limits for these chemicals in potable water. Further, enhanced carbon filtration effectively removed the chemicals, even at chlorine dioxide levels of more than twice what was used to treat the water. After 9 months, corrosion of copper test strips exposed to the chlorine dioxide was not higher than that of control strips. During the evaluation period, there were no cases of nosocomial Legionella in the building with the system, whereas there was one case in another building.

Our results indicate that operation of a chlorine dioxide system effectively removed Legionella species from a hospital water supply. Furthermore, we found that the system was safe, as levels of chlorine dioxide and chlorite were below EPA limits. The system did not appear to cause increased corrosion of copper pipes. Our results indicate that chlorine dioxide may hold promise as a solution to the problem of Legionella contamination of hospital water supplies.

CDC has estimated that 23% of Legionella infections are nosocomial. When a new hospital was being constructed and a substantial increase in transplantation was anticipated, an ultraviolet light apparatus was installed in the water main of the new building because 27% of water samples from taps in the old hospital contained Legionella. This study reports the rate of nosocomial Legionella infection and water contamination since opening the new hospital.

Charts of all patients with positive Legionella cultures, direct immunofluorescent antibody (DFA), or urine antigen between April 1989 and November 2001 were reviewed. Frequencies of DFAs and urine antigens were obtained from the laboratory.

None of the 930 cultures of hospital water have been positive since moving into the new building. Fifty-three (0.02%) of 219,521 patients had a positive Legionella test; 41 had pneumonia (40 community acquired). One definite L. pneumophila pneumonia confirmed by culture and DFA in August 1994 was nosocomial (0.0005%) by dates. This patient was transferred after prolonged hospitalization in another country, was transplanted 11 days after admission, and developed symptoms 5 days after liver transplant. However, tap water from the patient's room did not grow Legionella. Seventeen (2.5%) of 670 urine antigens were positive for Legionella (none nosocomial). Thirty-three (1.2%) of 2,671 DFAs were positive, including 7 patients (21%) without evidence of pneumonia and 6 (18%) who had an alternative diagnosis.

Ultraviolet light usage was associated with negative water cultures and lack of clearly documented nosocomial Legionella infection for 13 years at this hospital.

To determine incidence rates of hospital-acquired infections and to develop preventive measures to reduce the risk of hospital-acquired infections.

Prospective surveillance for hospital-acquired infections was performed during a 5-year period in the wards housing general and vascular, thoracic, orthopedic, and general gynecologic and gynecologic-oncologic surgery of the University Medical Center Utrecht, the Netherlands. Data were collected from patients with and without infections, using criteria of the Centers for Disease Control and Prevention.

The infection control team recorded 648 hospital-acquired infections affecting 550 (14%) of 3,845 patients. The incidence density was 17.8 per 1,000 patient-days. Patients with hospital-acquired infections were hospitalized for 19.8 days versus 7.7 days for patients without hospital-acquired infections.

Prolongation of stay among patients with hospital-acquired infections may have resulted in 664 fewer admissions due to unavailable beds. Different specialties were associated with different infection rates at different sites, requiring a tailor-made approach. Interventions were recommended for respiratory tract infections in the thoracic surgery ward and for surgical-site infections in the orthopedic and gynecologic surgery wards.

Surveillance in four surgical wards showed that each had its own prominent infection, risk factors, and indications for specific recommendations. Because prospective surveillance requires extensive resources, we considered a modified approach based on a half-yearly point-prevalence survey of hospital-acquired infections in all wards of our hospital. Such surveillance can be extended with procedure-specific prospective surveillance when indicated.

To determine the rates of surgical-site infections (SSIs) after spinal surgery and to identify the risk factors associated with infection.

SSIs had been identified by active prospective surveillance. A case-control study to identify risk factors was performed retrospectively.

University-associated, tertiary-care pediatric hospital.

All patients who underwent spinal surgery between 1994 and 1998. Cases were all patients who developed an SSI after spinal surgery. Controls were patients who did not develop an SSI, matched with the cases for the presence or absence of myelodysplasia and for the surgery date closest to that of the case.

There were 10 infections following 125 posterior spinal fusions, 4 infections after 50 combined anterior-posterior fusions, and none after 95 other operations. The infection rate was higher in patients with myelodysplasia (32 per 100 operations) than in other patients (3.4 per 100 operations; relative risk = 9.45; P < .001). Gram-negative organisms were more common in early infections and Staphylococcus aureus in later infections. Most infections occurred in fusion involving sacral vertebrae (odds ratio [OR] = 12.0; P = .019). Antibiotic prophylaxis was more frequently suboptimal in cases than in controls (OR = 5.5; P = .034). Five patients required removal of instrumentation and 4 others required surgical debridement.

Patients with myelodysplasia are at a higher risk for SSIs after spinal fusion. Optimal antibiotic prophylaxis may reduce the risk of infection, especially in high-risk patients such as those with myelodysplasia or those undergoing fusion involving the sacral area.

In hospital operating rooms (ORs), specially conditioned air is supplied to protect patients from airborne agents that may cause infections. This study investigated whether it is hygienically safe to shut down the air supply at night if measures are taken to ensure a timely restart before surgery is performed.

Experimental study.

Neurosurgical OR of a German university hospital.

The ventilation system was switched off and restarted after 10 hours. Particles suspended in the air near the operating table were counted, OR temperature was measured, and settle plates were exposed and incubated.

In 13 investigations, a median of 1.3 × 104 particles 0.5 μm/m3 or greater (range, 5.8 × 103 to 1.1 × 105) were documented immediately after restart in the morning. After 10 minutes and subsequently, no test showed a particle count exceeding the threshold limit of 1.0 × 104 particles 0.5 μm/m3 or greater recommended by the German Society of Hygiene and Microbiology. Only a few colony-forming units (CFU) were detected per settle plate (median, 0 CFU/60 cm2; range, 0 to 8) and OR temperatures quickly reached normal levels.

Shutting down OR ventilation during off-duty periods does not appear to result in an unacceptably high particle count or microbial contamination of the OR air shortly after the system is restarted. Because substantial energy and cost savings are likely, this should be considered in hygienically safe heating, ventilation, and air conditioning systems. However, normal ventilation should be established at least 30 minutes before surgical activity.

To determine the costs of the interventions aimed at controlling the 4-month outbreak and to determine the attributable length of stay (LOS) associated with infection and colonization with extended-spectrum beta-lactamase-producing Klebsiella pneumoniae.

A retrospective cost analysis was conducted from the hospital perspective. A micro-costing approach was employed. The LOS of four groups of hospitalized patients were compared with each other. National Perinatal Information Center criteria were used to stratify infants for severity of risk. The LOS of each group was compared with that of a national sample of similarly stratified infants.

A level III-IV, 45-bed neonatal intensive care unit.

Infant groups were infected (n = 8), colonized (n = 14), concurrent cohort (n = 54), and prior cohort (n = 486).

The cost of the outbreak totaled $341,751. The largest proportion of costs was related to healthcare worker time providing direct patient care (2,489 hours at a cost of $146,331). Infected and colonized neonates had longer LOS than either the concurrent cohort or the prior cohort (P < .001). Compared with the national sample, infected infants had a 48.5-day longer mean LOS (95% confidence interval [CI95], 1.7 to 95.2), whereas the prior cohort's mean LOS was 6 days shorter (CI95, -9.4 to -2.9).

This study increases the understanding of the burden of these multidrug-resistant organisms. Further research is needed to estimate the societal costs of these infections and the cost-effectiveness of preventive interventions.

To assess changes over time in susceptibility of Enterobacteriaceae from patients in ICUs, compare susceptibility rates of isolates from patients in ICUs with those from inpatients outside ICUs, and explore phenotypic patterns of cross-resistance and co-resistance.

From 1995 to 2000, centers participating in the ICU Surveillance Study tested 100 to 200 consecutive nosocomial gram-negative bacilli by broth microdilution.

Each year, 42 to 97 U.S. hospitals tested isolates.

In all years, imipenem was the most potent agent tested, followed by amikacin and ertapenem. Extended-spectrum beta-lactam and monobactam agents had good activity against Escherichia coli and Klebsiella species, but limited activity against Enterobacter species. Susceptibility to imipenem and amikacin did not fluctuate during the analysis period, whereas susceptibility to ceftazidime, ceftriaxone, and ciprofloxacin decreased 2% to 5%. The decline was most pronounced for susceptibility of Escherichia coli to ciprofloxacin: 98.7% of ICU isolates were susceptible in 1995 versus 93.2% in 2000. Susceptibility of ICU isolates was lower than that of non-ICU isolates, except for ciprofloxacin, for which the reverse was true. Cross-resistance was common among extended-spectrum cephalosporins and penicillins, but uncommon between imipenem and ertapenem. Only imipenem and ertapenem remained highly active against Enterobacteriaceae with a phenotype suggesting possible production of an extended-spectrum beta-lactamase and those with a phenotype suggesting possible Amp C hyperproduction.

Imipenem was the most active agent against nosocomial Enterobacteriaceae. Susceptibility to ciprofloxacin decreased from 1995 to 2000, particularly in Escherichia coli, and, in contrast to other agents, was lower among non-ICU isolates.

To report epidemiologic, bacteriologic, and clinical features of a Chryseobacterium meningosepticum outbreak.

Outbreak investigation.

A neonatal intensive care unit (NICU) of a referral teaching hospital.

During 2 weeks in September 2001, four neonates in the NICU developed sepsis and underwent laboratory investigation. Multiple samples were obtained for cultures from endotracheal tubes, mechanical ventilators and humidifier boxes, infant incubators, parenteral and antiseptic solutions, feeding bottles, sinks, faucets, doors, and healthcare workers.

C. meningosepticum was isolated from the blood cultures of four patients. The first isolate was identified 5 days after the death of the index case. Although all isolates were ciprofloxacin susceptible in vitro, the remaining three patients did not respond to ciprofloxacin therapy given for 6 or 7 days. Therapy was switched to vancomycin and rifampin and all three patients survived, with one having a complication (hydrocephalus). Environmental surveillance revealed C. meningosepticum in the stock lipid solution as the source of the epidemic. The outbreak was controlled after discontinuation of intravenous lipid solution, restriction of further neonatal admissions, and thorough disinfection of the unit and its equipment.

Early identification of an epidemic and its source is important in avoiding morbidity and mortality. A contaminated lipid stock bottle was the source of this outbreak associated with multiple cases and one death.