To identify factors associated with developing nosocomial bloodstream infection (NBI) and to identify factors associated with fatal outcome of NBI.

Prospective matched and unmatched case-control studies.

Songklanagarind Hospital, a 640-bed medical school and referral center in Songkla, Thailand.

During a two-year study period, 277 patients of 17,829 total admissions in the hospital services of medicine, surgery, pediatrics, and orthopedics acquired bloodstream infections during hospitalization. For the purpose of identifying factors associated with mortality attributed to NBI, patients who died from NBI were assigned as cases; patients who survived the infections were assigned as controls. For the purpose of identifying risk factors for acquisition of NBI, patients with NBI were considered as cases; patients with the same primary diagnosis as a case but without NBI were eligible to be case-matched controls.

Case fatality rate was 37.2% based on deaths attributed specifically to bloodstream infection. Patients who died from NBI, compared with those who survived such infections, were associated with inappropriate antibiotic treatment, infection caused by Klebsiella pneumoniae, and the medicine service. When comparison was made between 239 patients with bloodstream infections and 598 diagnosis-matched controls with no bloodstream infection, infection was found to be associated with old age, number of comorbidities, number of prior infections, neutropenia, duration of immunosuppressive drugs, and duration of indwelling intravenous catheter.

Awareness of the factors predisposing to NBI may permit better surveillance and better care. Although most of the factors associated with development of NBI or death from NBI are not subject to control, some factors are duration of immunosuppressive drugs, duration of indwelling intravenous lines, and choice of antibiotic treatment.

In many hospitals, the only sterile precautions used during the insertion of a nontunneled central venous catheter are sterile gloves and small sterile drapes. We investigated whether the use of maximal sterile barrier (consisting of mask, cap, sterile gloves, gown, and large drape) would lower the risk of acquiring catheter-related infections.

Prospective randomized trial.

A 500-bed cancer referral center.

We randomized patients to have their nontunneled central catheter inserted under maximal sterile barrier precautions or control precautions (sterile gloves and small drape only). All patients were followed for 3 months postinsertion or until the catheter was removed, whichever came first. Catheter-related infections were diagnosed by quantitative catheter cultures and/or simultaneous quantitative blood cultures.

The 176 patients whose catheters were inserted by using maximal sterile barrier precautions were comparable to the 167 control patients in underlying disease, degree of immuno-suppression, therapeutic interventions, and catheter risk factors for infections (duration and site of catheterization, number of catheter lumen, catheter insertion difficulty, reason for catheter removal). There were a total of four catheter infections in the test group and 12 in the control group (P=0.03, chi-square test). The catheter-related septicemia rate was 6.3 times higher in the control group (P=0.06, Fisher's exact test). Most (67%) of the catheter infections in the control group occurred during the first 2 months after insertion, whereas 25% of the catheter infections in the maximal sterile precautions group occurred during the same period (P<0.01, Fisher's exact test). Cost-benefit analysis showed the use of such precautions to be highly cost-effective.

Maximal sterile barrier precautions during the insertion of nontunneled catheters reduce the risk of catheter infection. This practice is cost-effective and is consistent with the practice of universal precautions during an invasive procedure.

To determine the epidemiology of Staphylococcus aureus infections in hemodialysis patients.

S aureus isolates from surveillance cultures and from sites of infection were evaluated by both bacteriophage typing and restriction endonuclease digestion of plasmid DNA.

A hemodialysis unit in Brugge, Belgium.

S aureus isolates from 11 chronic hemodialysis patients who had participated in the placebo arm of a previously reported placebo-mupirocin comparative study.

Of 75 S aureus isolates evaluated, 63 were from cultures of nares and 12 from infections (three arteriovenous fistula infections, four wound infections, and five bacteremias). All isolates were typed by bacteriophages and 56 (75%) had plasmids. Three patients developed 12 infections. Eleven infections were caused by isolates previously identified in surveillance cultures. Only one infection was caused by a strain not identified previously in surveillance cultures.

These results support the hypothesis that S aureus isolates causing infections in hemodialysis patients are of endogenous origin.

To identify and quantify the bacterial and fungal flora present at body sites used for vascular catheterization of infants in a neonatal intensive care unit.

Quantitative skin cultures were obtained from a group of neonatal patients to determine the bacterial flora found on the skin at four sites. Quantitative cultures of the jugular, subclavian, umbilical, and femoral sites were obtained on 50 infants, ranging in age from 2 days to 3 months old.

The neonatal intensive care unit of St. Christopher's Hospital for Children, a university-affiliated tertiary care children's hospital.

Colony counts ranged from 0 to 106 colony-forming units/ 10 cm2. Types of organisms found were consistent with other published studies and included coagulase-negative staphylococci, Staphylococcus aureus, yeast, aerobic gram-negative rods, Enterococcus species, Corynebacterium species, and alpha-hemolytic streptococci. There was a significantly higher mean colony count at the combined jugular/femoral sites versus the subclavian site (P<0.01) and umbilical site (P<0.05). Mean colony counts did not differ significantly between the jugular and femoral site, or between the subclavian and umbilical site. The umbilical site was more likely to be colonized with aerobic gram-negative rods, Enterococcus species, and yeast, while the subclavian had coagulase-negative staphylococci as the predominant organism. The jugular and femoral sites demonstrated a higher colony count of aerobic gram-negative rods, Enterococcus species and yeast than the other sites.

If central venous catheters need to be in place for extended periods of time, placement at a site with lower bacterial densities on the skin may help minimize catheter-associated infections. This study supports the subclavian as the preferred site.

To assess the feasibility of computerized nationwide surveillance of nosocomial infections in Belgium, and to obtain preliminary national and hospital-specific incidence data.

Prospective multicenter cohort study of surgical wound infections (SWI).

All 218 acute care hospitals in Belgium in the period October 14 to December 14, 1991.

Eighty-five of 218 acute care hospitals (39%) succeeded in collecting the required information and in completing this pilot study, although 50% of the participating hospitals had no previous experience in nosocomial infection surveillance. Seventy percent of the small-size hospitals (<200 beds) did not participate, mainly because of shortages of manpower. A lack of collaboration from clinicians was a problem in most participating hospitals.

SWI postdischarge surveillance was most successful when based on information collected by the surgeons at the surgical outpatient clinic; by this method, postdischarge information was obtained on 43.9% of all surgical procedures.

A total of 201 infections were observed among 10,537 operations, with a crude incidence rate of 1.91 per 100 operations or 1.51 per 1,000 person-days of observation. Infection rates by operation type and risk indicators are congruent with those of the literature. Survival analysis showed that the overall cumulative infection risk at 21 days postprocedure attained 8 1.6% of the 30-days risk.

This nationwide network for nosocomial infection surveillance has introduced the practice of computerized surveillance of performance in a large number of hospitals. Still, several aspects of the surveillance demand to be improved: the collaboration of the clinicians, the quality of the data, and the postdischarge surveillance methodology.

To evaluate trends in the occurrence and diagnosis of mycobacterial disease.

Prospective surveillance study using a central tuberculosis registry at a 611-bed teaching community hospital.

Data pertaining to 715 patients were entered into the registry between 1976 and 1991 on the basis of positive smears or cultures for acid-fast bacilli (AFB). Over time, the ratio of isolates of Mycobacterium tuberculosis to isolates of nontuberculous mycobacteria reversed (from 3.2 to 1 between 1976 and 1981 to 1 to 1.6 between 1986 and 1991). The sensitivity of the sputum AFB smear for diagnosis of pulmonary tuberculosis increased (from 0.73 to 0.96), but the positive predictive value of the sputum AFB smear declined (from 0.72 to 0.59). The acquired immunodeficiency syndrome epidemic accounted for these changes only to a small extent.

The increasing occurrence of non-tuberculous mycobacterial disease and the declining positive predictive value of the sputum AFB smear at this hospital imply that respiratory isolation and antituberculous therapy will be used inappropriately for many patients. The development and use of newer tests for early isolation and specific identification of M tuberculosis should be accelerated.

To assess the prevalence of high-level gentamicin-resistant enterococcus (HGRE) colonization, transmission patterns, and spectrum of illness among residents of a long-term care facility.

Monthly surveillance for HGRE colonization of wounds, rectum, and perineum over a 1-year period.

A Veterans Affairs long-term care facility attached to an acute-care facility.

All 341 patients in the facility during the observation period.

Over the 1-year period, 120 patients (35.2%) were colonized with HGRE at least once, with an overall monthly colonization rate of 20± 1.5%. HGRE were isolated from rectum (12.8%), wounds (11.7%), and perineum (9.3%). Patients with the poorest functional status had the highest rate of colonization (P<0.0005). HGRE-colonized patients were more likely to be colonized with methicillin-resistant Staphylococcus aureus (51% versus 25%; P<0.0005). Seventy-four patients (21.7%) were colonized at admission or at the start of the study. Another 46 patients (13.5%) acquired HGRE during the study, including 36 who acquired HGRE while in the long-term care facility and 10 who were positive when transferred back from the acute-care hospital. Based on plasmid profiles, only two patients appeared to have isolates similar to those of current or previous roommates. Carriage of HGRE was transient in most cases. Only 20 patients were colonized for 4 or more months, and those patients usually carried different strains intermittently. Infections were infrequent, occurring in only 4.1% of total patients.

In our long-term care facility, HGRE were endemic, and new acquisition of HGRE occurred frequently. However, only two patients had evidence of acquisition from a roommate, suggesting that cross-infection from a roommate was not a major route of spread of HGRE.

A cluster of bacterial contamination of platelets occurred at a university hospital in a one-month period. This unusual clustering allowed us to examine the likely mechanism of contamination and clinical sequelae.

We reviewed medical records of patients receiving random donor platelet transfusions to determine numbers of platelets transfused, reactions reported, and episodes of bacterial contamination. We also reviewed procedures at the collecting blood agencies and the hospital blood bank.

Four patients received bacterially contaminated platelets during June and July 1991. The rates of reported platelet transfusion reactions increased significantly (P<0.001) from September 1989 through July 1991 (study period); in addition, the rate of contamination of platelets during June and July 1991 was 23-fold higher than during the previous 21 months (P< 0.00 1). Surveillance methodology changed dramatically during the study period, contributing to the recognition of the current cluster. Pathogens isolated from the contaminated platelet pools were Bacillus cereus, Staphylococcus epidermidis, or Pseudomonas aeruginosa in titers ranging from 106 to 108 colony forming units/ml. Four constituent individual platelet units identified as the probable cause of the outbreak (including one postepidemic episode) were significantly older (mean age, 4.8 days) than 106 randomly selected individual platelet units (mean age, 3.7 days; P = 0.04). Platelet pools were transfused an average of 2.5 hours after pooling. Review of blood collection and platelet preparation practices did not identify breaks in procedure or technique that could have caused contamination.

Increased awareness of platelet transfusion reactions by clinical staff and routine culturing of all platelets associated with transfusion reactions will identify contaminated platelets. Identification of contaminated platelets is necessary to treat affected patients appropriately and to determine the prevalence of and risk factors for contaminated platelets.