To identify risk factors for excessive bleeding after coronary artery bypass graft (CABG) procedures and to quantify the outcomes related to this complication.

We conducted a case–control study to identify risk factors for hemorrhage following CABG surgery and a historical cohort study to quantify outcomes of hemorrhage.

The cardiothoracic surgery service of a university hospital.

Factors associated with excessive blood loss were recent catheterization (odds ratio [OR] = 0.44; 95% confidence interval [CI95], 0.21 to 0.91); age older than 65 years (OR = 1.94; CI95, 0.96 to 3.93); bypass time of 150 minutes or more (OR = 2.91; CI95, 1.09 to 7.81); and postoperative platelet count of 160,000/mm3 or less (OR = 2.36; CI95, 1.06 to 5.22). The attributable cost of a postoperative hemorrhage was $3,866 (P = .0002) overall, $9,912 (P = .0001) for patients who required reoperation, and $3,316 (P = .03) for those treated medically. The median attributable postoperative length of stay was 1 day longer for cases than for controls (P = .011). Postoperatively, patients who hemorrhaged received significantly larger volumes of packed red blood cells (P < .0001), fresh frozen plasma (P < .0001), platelets (P < .0001), plasminate (P = .007), protamine sulfate (P < .0001), desmopressin acetate (P < .0001), and epsilon-aminocaproic acid (P < .0001) than did controls.

Age, duration of bypass, and postoperative platelet count were associated with excessive bleeding. Hemorrhage after CABG surgery significantly increased the length of stay and cost of care.

To determine the accuracy and cost-effectiveness of a polymerase chain reaction (PCR) for detecting nasal carriage of Staphylococcus aureus directly from clinical specimens.

This occurred in a tertiary-care hospital in Cleveland, Ohio, and included 239 consecutive patients who were scheduled for a cardiothoracic surgical procedure. Conventional cultures and a PCR for S. aureus from nasal swabs were used as measurements.

Data sources were market prices and Bureau of Labor Statistics. The time horizon was the maximum period for availability of culture results (3 days). Interventions included universal mupirocin therapy without testing; initial therapy, with termination if PCR negative (treat-PCR); initial therapy, with termination if culture negative (treat-culture); treat PCR-positive carriers (PCR-guided treatment); and treat culture-positive carriers (culture-guided treatment). The perspective was institutional and costs and the length of time to treatment were outcome measures.

Sixty-seven (28%) of the 239 swabs grew S. aureus. Rapid PCR was 97.0% sensitive and 97.1% specific for the detection of S. aureus. For populations with prevalences of nasal S. aureus carriage of up to 50%, the PCR assay had negative predictive values of greater than 97%. PCR-guided treatment had the lowest incremental cost-effectiveness ratio ($1.93 per additional day compared with the culture strategy). Among immediate treatment strategies, treat-PCR was most cost-effective. The universal therapy strategy cost $38.19 more per additional day gained with carrier identification compared with the PCR strategy.

Rapid real-time PCR is an accurate, rapid, and cost-effective method for identifying S. aureus carriers for preoperative intervention.

To examine the current status of bloodstream infections (BSIs) in a community hospital as part of a 25-year longitudinal study.

Retrospective descriptive epidemiologic study.

Community teaching hospital.

All inpatients in 1998 with a positive blood culture who met the CDC NNIS System case definition of BSI.

Cases were stratified by underlying illness category using case mix adjustment categories (after McCabe) and reviewed for associations among mortality, underlying illness severity, and multiple clinical and laboratory parameters.

Of 19,289 patients discharged in 1998,185 had an episode of infection documented by blood culture (96 cases per 10,000 inpatients). BSI was twice as frequent in patients 65 years and older compared with younger patients. BSIs caused or contributed to the deaths of 22 patients for an overall case-fatality rate of 11.9% compared with 20.7% in 1982 (P = .02). Striking decreases were noted for in-hospital patient mortality in 1998 for BSIs with ultimately and rapidly fatal underlying illnesses (P = .02 and P < .10, respectively). Primary bacteremia decreased compared with 1982. Antibiotic use was vigorous, but resistance was modest in both nosocomial and community-acquired organisms and had changed little from 1982 and 1987.

Compared with previous studies, case-fatality rates in patients with BSI were substantially lower in rapidly fatal and ultimately fatal underlying illness categories. Antibiotic use was extensive but prompt and appropriate. Microorganism resistance to antibiotics changed little from the 1980s.

To determine the etiology of Pseudomonas aeruginosa surgical-site infections following cardiac surgery.

University teaching hospital.

Those with wound cultures that grew P. aeruginosa after cardiac surgery performed from 1999 to 2001.

Medical records and operating room (OR) records of patients with P. aeruginosa cardiac surgical-site infections from 1999 to 2001 were reviewed. Healthcare workers involved with two or more cases were interviewed and examined. Specimens for environmental cultures were obtained from the ORs and cardiac surgical equipment. Cardiac surgery cases were observed and postoperative care and the cleaning of surgical instruments were investigated. OR air handling system records during the epidemic period were reviewed. Molecular fingerprinting of available P. aeruginosa isolates from infected patients and a healthcare worker was done.

There were five P. aeruginosa cardiac surgical-site infections from January to August 2001, compared with no such infections from 1999 to 2000. All were adult patients. One cardiac surgeon with onychomycosis operated on all five cases. He did not routinely double glove. The involved fingernail grew P. aeruginosa. Three P. aeruginosa patient isolates were available for pulsed-field gel electrophoresis; two were identical to the isolate from the involved surgeon's onychomycotic nail. No environmental OR cultures grew P. aeruginosa. The surgeon's culture-positive nail was completely removed. There have been no P. aeruginosa surgical-site infections among cardiac surgery patients since this intervention.

At least two cases of a cluster of P. aeruginosa surgical-site infections resulted from colonization of a cardiac surgeon's onychomycotic nail.

To assess the frequency of community-acquired methicillin-resistant Staphylococcus aureus (MRSA) infections.

A teaching hospital in Singapore.

Prospectively collected surveillance data were reviewed during a 1-year period to determine the extent and origin of community-acquired MRSA infections.

Whereas 32% of 383 MRSA infections were detected less than 48 hours after hospital admission and would, by convention, be classified as “community acquired,” all but one of these were among patients who had been exposed to outpatient centers including dialysis or chemotherapy clinics, visiting nurses, community hospitals, or all three.

With health care increasingly being delivered in an outpatient setting, community-acquired MRSA infections are often acquired in hospital-related sites and most may be more accurately described as “healthcare acquired.” Infection control measures need to move beyond the traditional paradigm of acute care hospitals to effectively control the spread of resistant pathogens.

To study multicenter antibiotic practices for suspected early-onset sepsis (EOS) with negative blood cultures (NegBCs) and to identify opportunities for reduction of antimicrobial exposure.

Retrospective study.

Thirty academic hospitals (University HealthSystem Consortium) located in 24 states.

Data were from a survey of 790 extremely low birth weight (ELBW) infants. Total antibiotic exposures (antibiotic-days per patient) were calculated.

On admission to the NICU, 94% of 790 ELBW infants had BCs performed and empiric antibiotics initiated. When PosBC and NegBC infants were compared, 47 patients with PosBCs were similar to 695 with NegBCs in birth weight, gestational age (GA), and mortality. Patients with suspected EOS but NegBCs given ampicillin/aminoglycosides were grouped by length of administration and GA. For GA of 26 weeks or younger, 170 infants given a short (≤ 3 days) and 157 given a long (≥ 7 days) course were similar regarding birth weight, mortality, antepartum history, and CRIB scores, but were different (P < .01) in number receiving a third antimicrobial (3% and 17%) and antibiotic-days (23 and 38). For GA of 27 weeks or older, 113 infants given a short and 77 given a long course differed (P < .01) in number receiving a third antimicrobial (2% and 23%) and antibiotic-days (19 and 30).

Most suspected EOS infants with NegBCs are given antibiotics, but no antepartum historical risk factors or neonatal clinical signs explained prolonged administration. Discontinuing empiric antibiotics when BCs are negative in asymptomatic ELBW infants can reduce antimicrobial exposure without compromising clinical outcome.

To investigate a possible nosocomial outbreak of tuberculosis (TB).

Retrospective cohort study.

Community hospital.

We reviewed medical records, hospital infection control measures, and potential locations of nosocomial exposure. We examined the results of acid-fast bacilli (AFB) smears, cultures, and drug susceptibility testing, and performed a DNA fingerprint analysis. We observed laboratory specimen processing procedures and bronchoscope disinfection procedures. We also reviewed bronchoscopy records.

In October 2000, three patients had bronchoscopy specimen cultures that were positive for Mycobacterium tuberculosis. Of the three, only one had clinical signs and symptoms consistent with TB and positive AFB sputum smears. The other two did not have signs and symptoms consistent with TB and had no known exposure to individuals with infectious TB. The three M. tuberculosis isolates had matching DNA fingerprints. No evidence of laboratory cross-contamination was identified. The three culture-positive specimens of M. tuberculosis were collected with the same bronchoscope within 9 days. This bronchoscope was inadequately cleaned and disinfected between patients, and the automated reprocessor used was not approved for use with the hospital bronchoscope.

One of the bronchoscopes at this hospital was contaminated with M. tuberculosis during bronchoscopy of an AFB-smear-positive patient. Subsequent specimen contamination likely occurred because the bronchoscope had been inadequately cleaned and disinfected. Patients who subsequently underwent bronchoscopy were also potentially exposed to M. tuberculosis from this bronchoscope.

To choose the most cost-effective option for detecting human immunodeficiency virus (HIV-1) and hepatitis C virus (HCV) among blood transfusion recipients.

Cost-effectiveness analysis. Effectiveness was expressed as the number of HIV-1 or HCV infections detected, regardless of whether they were related to transfusion. To estimate costs, we assumed hospital insurance would cover costs related to detection and compensation, when granted.

A 2,890-bed acute care teaching hospital in Bordeaux, France.

Eight options were defined, from the simplest, which would be to do nothing, to a maximal approach, which would be to keep a serum sample in a serum library for a lookback and perform tests for antibody to HIV-1 and to HCV before and 3 months after transfusion. Data on probabilities and costs were taken from the literature and experiences of French hospitals.

The most cost-effective option was to perform viral antibody testing before transfusions (option 3), which would detect 27 infections per 1,000 patients, for an expenditure of US $1,260 per detected patient Option 6, obtaining a serum sample before transfusion and performing tests for antibody to HIV-1 and to HCV 3 months after transfusion, had a similar cost-effectiveness ratio but detected only 16 infections per 1,000 patients. Performing tests before and 3 months after transfusion (option 4), compared with option 3, would detect 1 additional infection for an additional cost of US $8,322.

The most cost-effective options are not specific to blood transfusion recipients and might be more suited to all hospitalized patients.

To describe Burkholderia cepacia contamination of a cornea bank and the measures taken to identify and eliminate the source of infection.

Cultures were performed to assess the extent and source of contamination, and pulsed-fleld gel electrophoresis was used for molecular typing.

Routine surveillance cultures identified 5 contaminated corneas during a 10-day period. Additional cultures showed that 28 of 88 samples were positive for this organism. Environmental investigation showed that an open bottle of trypan blue used to assess corneal morphology was contaminated with the epidemic strain.

Trypan blue played a major role in this contamination of corneas. This episode shows that microbial contamination can affect transplanted corneas despite ongoing culture surveillance and suggests that new methods may be needed to avoid this risk.

To evaluate the long-term efficacy of urinary Foley catheters (latex and silicone) impregnated with (1) chlorhexidine and silver sulfadiazine (CXS) and (2) chlorhexidine, silver sulfadiazine, and triclosan (CXST) in inhibiting extra-luminal bacterial adherence and to compare their efficacy with that of silver hydrogel latex (SH) and nitrofurazone-treated silicone (NF) catheters.

The antimicrobial spectrum of these catheters was evaluated using a zone of inhibition assay. A novel in vitro urinary tract model was developed to study the potential in vivo efficacy of antimicrobial catheters in preventing extraluminal bacterial colonization. The “meatus” was inoculated daily with Staphylococcus aureus, Staphylococcus epidermidis, Escherichia coli, Enterococcus faecalis, Pseudomonas aeruginosa, and Candida albicans. The “bladder” portion of the model was cultured daily to determine bacterial growth.

Both CXS and CXST catheters had a broader antimicrobial spectrum than SH and NF catheters. In the in vitro model, CXST latex and silicone catheters exhibited significantly better efficacy (3 to 25 days) against uropathogens, compared with CXS (1 to 14 days) and control (0 to 5 days) catheters (P = .01). CXST latex catheters exhibited significantly longer protection against Staphylococcus aureus, Staphylococcus epidermidis, Escherichia coli, and Pseudomonas aeruginosa, compared with SH catheters (P = .01). CXST silicone catheters resisted colonization with Staphylococcus aureus and Staphylococcus epidermidis for a significantly longer period (23 to 24 days) than did NF catheters (9 to 11 days) (P = .01).

Catheters impregnated with synergistic combinations of chlorhexidine, silver sulfadiazine, and triclosan exhibited broad-spectrum, long-term resistance against microbial colonization on their outer surfaces (Infect Control Hosp Epidemiol 2003;24:506-513)

Five cases of primary varicella zoster virus (VZV) were diagnosed among hospital healthcare workers (HCWs). All had complied with a pre-employment VZV screening program and had been considered immune.

To summarize the investigation of VZV among un-immunized HCWs and to provide recommendations for avoiding false-positive serologic tests.

Risk of transmission of VZV to susceptible HCWs is minimized through serologic screening. Varicella vaccine is recommended for susceptible HCWs. A commercially available latex bead agglutination assay (LA) is widely used because it is rapid and easy to perform. LA was compared with the whole-cell varicella ELISA standardized in the Centers for Disease Control and Prevention (CDC) National Herpes Laboratory.

Large inner-city, tertiarycare hospital with a diverse employee population.

In a year, 5 HCWs presented with laboratory-confirmed primary varicella infection. Four had VZV exposures 2 weeks prior to presentation. All had documented positive VZV titers by LA performed at hire. None were offered VZV vaccination. The original LAs were judged false-positives.

Fifty-three consecutive VZV LA samples from the hospital laboratory were retested at the CDC. Forty-four samples concurred. Of the remaining 9, 4 were positive by hospital LA but negative by CDC IgG ELISA. Four were equivocal by hospital LA but negative by CDC IgG ELISA and LA. One was positive by hospital LA but negative by LA and equivocal by ELISA at the CDC.

LA may be prone to false-positive results and inappropriate for screening hospital HCWs.

To assess the way healthcare facilities (HCFs) diagnose, survey, and control methicillin-resistant Staphylococcus aureus (MRSA).

Questionnaire.

Ninety HCFs in 30 countries.

Evaluation of susceptibility testing methods showed that 8 laboratories (9%) used oxacillin disks with antimicrobial content different from the one recommended, 12 (13%) did not determine MRSA susceptibility to vancomycin, and 4 (4.5%) reported instances of isolation of vancomycin-resistant S. aureus but neither confirmed this resistance nor alerted public health authorities. A MRSA control program was reported by 55 (61.1%) of the HCFs. The following isolation precautions were routinely used: hospitalization in a private room (34.4%), wearing of gloves (62.2%), wearing of gowns (44.4%), hand washing by healthcare workers (53.3%), use of an isolation sign on the patient's door (43%), or all four. When the characteristics of HCFs with low incidence rates (< 0.4 per 1,000 patient-days) were compared with those of HCFs with high incidence rates (P = 0.4 per 1,000 patient-days), having a higher mean number of beds per infection control nurse was the only factor significantly associated with HCFs with high incidence rates (834 vs 318 beds; P = .02).

Our results emphasize the urgent need to strengthen the microbiologic and epidemiologic capacities of HCFs worldwide to prevent MRSA transmission and to prepare them to address the possible emergence of vancomycin-resistant S. aureus.

To determine the epidemiology of colonization with vancomycin-resistant Enterococcus (VRE) among intensive care unit (ICU) patients.

Ten-month prospective cohort study.

A 19-bed medical ICU of a 1,440-bed teaching hospital.

Patients admitted to the ICU had rectal swab cultures for VRE on admission and weekly thereafter. VRE-positive patients were cared for using contact precautions. Clinical data, including microbiology reports, were collected prospectively during the ICU stay.

Of 519 patients who had admission stool cultures, 127 (25%) had cultures that were positive for VRE. Risk factors for VRE colonization identified by multiple logistic regression analysis were hospital stay greater than 3 days prior to ICU admission (adjusted odds ratio [AOR], 3.6; 95% confidence interval [CI95], 2.3 to 5.7), chronic dialysis (AOR, 2.4; CI95, 1.2 to 4.5), and having been admitted to the study hospital one to two times (AOR, 2.3; CI95,1.4 to 3.8) or more than two times (AOR, 6.5; CI95, 3.7 to 11.6) within the past 12 months. Of the 352 VRE-negative patients who had one or more follow-up cultures, 74 (21%) became VRE positive during their ICU stay (27 cases per 1,000 patient-ICU days).

The prevalence of VRE culture positivity on ICU admission was high and a sizable fraction of ICU patients became VRE positive during their ICU stay despite contact precautions for VRE-positive patients. This was likely due in large part to prior VRE exposures in the rest of the hospital where these control measures were not being used.

To assess the prevalence of and the clinical features associated with asymptomatic Staphylococcus aureus colonization in a healthy outpatient population, and to compare the characteristics of colonizing methicillin-resistant S. aureus (MRSA) strains with those of strains causing infection in our community and hospital.

Outpatient military clinics.

Specimens were obtained from the nares, pharynx, and axillae of 404 outpatients, and a questionnaire was administered to obtain demographic and risk factor information. MRSA strains were typed by pulsed-field gel electrophoresis (PFGE) and evaluated for antibiotic susceptibility. Antibiograms of study MRSA strains were compared with those of MRSA strains causing clinical illness during the same time period.

Methicillin-susceptible S. aureus (MSSA) colonization was present in 153 (38%) of the 404 asymptomatic outpatients, and MRSA colonization was present in 8 (2%). Detection of colonization was highest from the nares. No clinical risk factor was significantly associated with MRSA colonization; however, a tendency was noted for MRSA to be more common in men and in those who were older or who had been recently hospitalized. All colonizing MRSA strains had unique patterns on PFGE. In contrast to strains responsible for hospital infections, most colonizing isolates of MRSA were susceptible to oral antibiotics.

MRSA and MSSA colonization is common in our outpatient population. Colonization is best detected by nares cultures and most carriers of MRSA are without apparent predisposing risk factors for acquisition. Colonizing isolates of MRSA are heterogeneous and, unlike nosocomial isolates, often retain susceptibility to other non-beta-lactam antibiotics.

To determine the prevalence of central venous catheter (CVC) use among patients both within and outside the ICU setting.

A 1-day prevalence survey of CVC use among adult inpatients at six medical centers participating in the Prevention Epicenter Program of the CDC. Using a standardized form, observers at each Epicenter performed a hospital-wide survey, collecting data on CVC use.

Inpatient wards and ICUs of six large urban teaching hospitals.

At the six medical centers, 2,459 patients were surveyed; 29% had CVCs. Among the hospitals, from 43% to 80% (mean, 59.3%) of ICU patients and from 7% to 39% (mean, 23.7%) of non-ICU patients had CVCs. Despite the lower rate of CVC use on non-ICU wards, the actual number of CVCs outside the ICUs exceeded that of the ICUs. Most catheters were inserted in the subclavian (55%) or jugular (22%) site, with femoral (6%) and peripheral (15%) sites less commonly used. The jugular (33.0% vs 16.6%; P < .001) and femoral (13.8% vs 2.7%; P < .001) sites were more frequently used in ICU patients, whereas peripherally inserted (19.9% vs 5.9%; P < .001) and subclavian (60.7% vs 47.3%; P < .001) catheters were more commonly used in non-ICU patients.

Current surveillance and infection control efforts to reduce morbidity and mortality associated with bloodstream infections concentrate on the high-risk ICU patients with CVCs. Our survey demonstrated that two-thirds of identified CVCs were not in ICU patients and suggests that more efforts should be directed to patients with CVCs who are outside the ICU.

To describe the epidemiology of methicillin-resistant Staphylococcus aureus (MRSA) at a university hospital in Tenerife, Canary Islands, during a 40-month period and to evaluate the effectiveness of the application of control measures.

Laboratory-based surveillance, medical charts and microbiological records review, and characterization of strains by pulsed-field gel electrophoresis (PFGE) were used to describe the epidemiology. Infection control practices were introduced as an intervention.

A 650-bed, tertiary-care university hospital.

Patients with clinical and nasal isolates of MRSA and colonized staff members.

The rate of nosocomial MRSA infections was 32.5% for 1997, 17.9% for 1998, 14.5% for 1999, and 25.6% during the first 4 months of 2000. The major sites of isolation for nosocomial MRSA infection included surgical wounds (25%) and the lower respiratory tract (24%). Intensive care units and surgical specialties had more frequent MRSA cases. Characteristics associated with nosocomial MRSA isolates included prior use of intensive antibiotic therapy, prolonged hospital stays, major underlying illness, invasive procedures, and older age. PFGE type A (subtype A1) was the strain most frequently found and the only PFGE type involved in clusters.

Surveillance cultures and contact droplet precautions were followed by decreased rates for 2 years. Nevertheless, the spread of PFGE subtype Al to many different areas of the hospital and the increase in incidence during the first third of 2000 indicates either that surveillance cultures were not used widely enough or that compliance with isolation measures was suboptimal.

To assess the validity of current estimates of the noncontagiousness of sputum smear-positive respiratory tuberculosis (TB) on treatment.

A descriptive analysis of the mycobacteriologic response to treatment.

A TB inpatient unit of a Canadian hospital.

Thirty-two HIV-seronegative patients with moderate to advanced sputum smear-positive respiratory TB were treated with uninterrupted, directly observed, weight-adjusted isoniazid, rifampin, and pyrazinamide. Each patient's initial isolate was drug susceptible and each patient's sputum mycobacteriology was systematically followed until 3 consecutive sputum smears were negative on 3 separate days.

The time to smear conversion varied remarkably (range, 8 to 115 days; average, 46 days) and was influenced by sputum sampling frequency. Only 3 patients (9.4%) had smear conversions by 14 days and only 8 (25%) had smear conversions by 21 days, the average time it took for drug susceptibility test results to become available. During the first 21 days of treatment, the semiquantitative sputum smear score decreased rapidly and the time to detection of positive cultures doubled. Within the time to smear conversion, virtually all smear-positive specimens (98%) were culture positive and only 34% of the patients had culture conversions (ie, 3 consecutive negative cultures).

Current estimates of the noncontagiousness of sputum smear-positive respiratory TB on treatment (for 14 days, for 21 days, or until smear conversion) are estimates of relative noncontagiousness. They do not signal absolute noncontagiousness (culture conversion). Semiquantitative smear and time-to-detection data suggest that respiratory isolation beyond 21 days of optimal treatment should be selective.

To investigate and control an outbreak of epidemic keratoconjunctivitis (EKC) occurring in a neonatal intensive care unit (NICU) and to determine signs specific for EKC in newborns.

Outbreak investigation and case–control study.

NICU of a 1,600-bed university hospital in Heidelberg, Germany.

Case-patients were defined as individuals who had adenoviral antigen detected by ELISA or by PCR from conjunctival swabs or who had clinical signs of conjunctivitis. Twelve newborns from the NICU who had EKC between October and November 1998 were compared with 11 who had bacterial conjunctivitis.

Control measures included cohorting patients in contact isolation, using gloves and gowns for patient care, and using a hand rub and disinfectants with virucidal activity.

Thirteen patients, 6 healthcare workers, and 11 relatives of the patients had EKC, according to our case definition. Case-patients were more likely than control-patients to develop lacrimal swelling (P < .001), eye redness (P = .004), and lacrimation (P = .037) involving both eyes (P = .002). Prior examination by an ophthalmologist was a significant risk factor for EKC (P = .004). For diagnosis or treatment of retinopathy, premature newborns were seen by an ophthalmologist from a nearby eye care center where an EKC outbreak was ongoing. No new cases were diagnosed more than 10 days after the implementation of control measures.

In newborns, lacrimal swelling, eye redness, and lacrimation in both eyes should evoke suspicion of EKC. Ophthalmologists who have had contact with known EKC cases should use antiseptics and disinfectants with virucidal activity before contact with newborns or abstain from examining newborns (Infect Control Hosp Epidemiol 2003;24:514-519)

To describe a group of infants with complex congenital heart disease (CCHD) who had candidemia postoperatively and to perform a case–control study of risk factors.

Descriptive and case–control study.

Tertiary-care medical centers.

Infants with CCHD who underwent cardiac surgery and developed candidemia. Controls were matched for congenital heart disease lesions, age, and postoperative hospital stay.

Of 95 infants younger than 6 months with CCHD who underwent cardiac surgery between January 1999 and April 2001, 6 (6.3%) developed candidemia with 5 different species. The candidemia was prolonged (range, 12 to 32 days; median, 17 days). The interval between cardiac surgery and onset of candidemia was 12 to 57 days (median, 24 days). All had a central venous catheter inserted 8 to 50 days prior to the onset of candidemia. The mortality rate was 83%, compared with 13% for the group without candidemia. A univariate analysis of the potential risk factors revealed that duration of total parenteral nutrition (TPN), duration of antibiotics, intraoperative cardiopulmonary bypass time, and documentation of an intravascular thrombus were associated with candidemia. In multivariate analysis, long duration of TPN and documentation of a thrombus were associated with candidemia.

Candidemia following cardiac surgery for infants with CCHD was persistent and associated with high mortality. The variety of species indicates that this was not a common-source outbreak. Risk factors associated with candidemia were duration of TPN and documentation of an intravascular thrombus.

To examine the reliability of the data produced by an automated system for the surveillance of nosocomial infections.

A 906-bed, tertiary-care teaching hospital.

Three surveillance techniques were concurrently performed in seven high-risk units during an 11-week period: automated surveillance (AS) based on the prospective processing of computerized medical records; laboratory-based ward surveillance (LBWS) based on the retrospective verification by ward clinicians of weekly reports of positive bacteriologic results; and a reference standard (RS) consisting of the infection control team reviewing case records of patients with positive bacteriology results. Bacteremia, urinary tract infections, and catheter-related infections were recorded for all inpatients. The performances (sensitivity, specificity, and time consumption) of AS and LBWS were compared with those of RS.

Of 548 positive bacteriology samples included during the study period, 229 (42%) were classified as nosocomial infections. The overall sensitivity was 91% and 59% for AS and LBWS, respectively. The two methods had the same overall specificity value (91%). Kappa measures of agreement were 0.81 and 0.54 for AS and LBWS, respectively. AS required less time to collect data (54 seconds per week per unit) compared with LBWS (7 minutes and 43 seconds per week per unit) and RS (37 minutes and 15 seconds per week per unit).

Our results confirm that the retrospective review of charts and laboratory data by physicians lacks sensitivity for the surveillance of nosocomial infections. The intranet-based automated method developed for this purpose was more accurate and less time-consuming than the weekly, retrospective LBWS method.