To describe the characteristics of and trends in nosocomial infection among human immunodeficiency virus (HIV)-infected patients.

Multicenter prospective cohort study.

HIV-infected patients were enrolled at time of first inpatient admission at five Veterans' Administration Medical Centers (VAMCs).

As of March 1995, 2,541 patients with 6,625 inpatient admissions had been monitored in the five VAMCs. A total of 530 nosocomial infections were detected using standard Centers for Disease Control and Prevention definitions. Overall distribution by infection site was 31% for primary bloodstream infections (BSIs), 28% for urinary tract infections, 15% for pneumonia, and 26% for all other sites. Of BSIs, 63% were central line-associated bloodstream infections (CLABs). The rate of CLABs per 1,000 central line days was 6.5 (range, 2.3-8.3) for all patients from participating hospitals, similar to the median CLAB rate of 6.0 for patients in medical intensive-care units (ICUs) of National Nosocomial Infections Surveillance (NNIS) System hospitals from January 1990 through September 1994. For ICU-specific CLABs, the rate from hospitals reporting at least one ICU CLAB was 12.7 (range, 12.1-13.1), comparable to the 90th percentile of NNIS hospital medical ICUs (13.1). Staphylococcus aureus, associated with 35% of BSIs, was the most common nosocomial BSI pathogen. Our data demonstrated the following: 13 (10%) of 134 patients with CD4 counts ≥200 cells/mm3 had a CLAB, compared with 61 (6%) of 1,011 patients with CD4 counts <200 cells/mm3, P=.08; the per-day risk of CLABs did not change with increased duration of catheterization (P=.4); and the per-day risk of a temporary (ie, short-term) CLAB was greater than that of a permanent CLAB (P<.001).

The data suggest that HIV-infected patients were at higher risk of acquiring a BSI than were patients in the NNIS population; patients with CD4 counts ≥200 cell/mm3 and temporary central lines were at increased risk for BSI, perhaps reflecting widespread prophylaxis with trimethoprim-sulfamethoxazole among patients with CD4 counts <200 cells/mm3; and, in contrast to most studies, S aureus, not coagulase-negative Staphylococcus, was the most common BSI pathogen.

To assess the accuracy of an automated data entry system employing optical scanning technology and to provide an analysis of its costs as compared to manual data entry.

The accuracy and cost of automated data entry of 100 surgical-wound infection surveillance questionnaires was compared to manual entry.

The Surgical Directorate, The Royal Hospitals, Belfast, Northern Ireland.

The use of optical scanning technology greatly improved the speed and accuracy of data entry. The time spent by the keyboard operator on data entry was reduced substantially.

For each surgical-wound infection questionnaire automatically processed, there was a saving in clerical time equivalent to $0.63. The automated data entry process resulted in a 22-fold productivity increase compared to manual data entry with validation. After validation, an error rate of <0.2 errors per 1,000 responses was detected in automatically entered data compared to a rate of 12.4 errors per 1,000 responses for manually entered data. The automated system, including validation, provided a sevenfold productivity increase compared to “quick-and-dirty” manual data entry without validation.

Hospital information technology systems may achieve total integration of data management, but realistically this would appear to be very much in the future. Until then, in view of the accuracy and substantial savings in time and money, we recommend the use of automated data entry technology. This system would be especially useful where data are transported from outlying hospitals to a central receiving center for collation and analysis.

In addition to single-hospital outbreaks, interhospital transmission of extended-spectrum β-lactam-resistant (ESBLR) Klebsiella pneumoniae has been suspected in some reports. However, these studies lacked sufficient epidemiological information to confirm such an occurrence.

We reviewed the surveillance data reported to the National Nosocomial Infections Surveillance (NNIS) System during 1986 to 1993 for K pneumoniae isolates and their susceptibility to either ceftazidime, cefotaxime, ceftriaxone, or aztreonam. Pulsed-field gel electrophoresis (PFGE) was used to study available ESBLR K pneumoniae isolates.

Among 8,319 K pneumoniae isolates associated with nosocomial infections, 727 (8.7%) were resistant or had intermediate-level resistance to at least one of these antibiotics. One hospital (hospital A) accounted for 321 isolates (44.2%) of ESBLR K pneumoniae. During 1986 to 1993, the percentage of K pneumoniae isolates that were ESBLR increased from 0 to 57.7% in hospital A, from 0 to 35.6% in NNIS hospitals 0 to 20 miles from hospital A (area B), and from 1.6 to 7.3% in NNIS hospitals more than 20 miles from hospital A, including hospitals located throughout the United States. Analysis of PFGE restriction profiles showed a genetic relationship between a cluster of isolates from hospital A and some isolates from one hospital in area B, and consecutive admission in these two hospitals was confirmed for two patients from whom isolates were available.

These data provide evidence of interhospital transmission of ESBLR K pneumoniae in one region of the United States and stress the interrelationship between hospitals when trying to control antimicrobial resistance.

To determine the rates and routes of Acinetobacter baumanii colonization and pneumonia among ventilated patients in a surgical intensive-care unit (SICU) before and after architectural modifications.

A nonsequential study comparing two groups of patients. All isolates from systematic and clinical samples were genotyped by pulsed-field gel electrophoresis (PFGE). Records of patients hospitalized during the first and second periods were reviewed and findings were compared. Between the two periods, the SICU was remodeled from enclosed isolation rooms and open rooms to only enclosed isolation rooms with handwashing facilities in each room.

All patients hospitalized and mechanically ventilated for more than 48 hours in the 15-bed SICU of the University Hospital of Besançon (France).

For the first and second periods, the rates of colonization were, respectively, 28.1% and 5.0% of patients (P<10−7; relative risk [RR], 2.23; 95% confidence interval [CI95], 1.8-2.75) and the specific rates of bronchopulmonary (BP) colonization were, respectively, 9.1 and 0.5 per 1,000 days of mechanical ventilation (P<10−5). Seven major PFGE isolate types were identified, 4 of which were isolated from 44 of the 47 colonized or infected patients. Logistic regression analysis showed that colonization was not associated with patient characteristics.

Conversion from open rooms to isolation rooms may help control nosocomial BP tract acquisition of A baumanii in mechanically ventilated patients hospitalized in an SICU.

To determine if varicella vaccination of healthcare workers would result in a net cost savings.

A Markov-based decision analysis.

The analysis was based on a hypothetical population of healthcare workers. Data were obtained from exposure records of a tertiary-care hospital and from the literature. Workers were considered potentially susceptible if they had no past history of varicella.

Vaccination of potentially susceptible workers would result in a net cost savings of $59 per person. Serological testing prior to vaccination resulted in smaller net savings. The results were robust across a wide range of assumptions. Importantly, however, the result was very dependent on infection control policy regarding work restrictions for vaccine recipients. If more than 3% of vaccinees were removed from work due to vaccine-associated rash, vaccination no longer would result in a net cost savings.

Varicella vaccination of potentially susceptible healthcare workers can reduce costs and decrease morbidity. Infection control policy regarding work restrictions for vaccine recipients will play a key role in the feasibility of vaccination.