To compare a surveillance definition of nosocomial bloodstream infections requiring only microbiology data to the Centers for Disease Control and Prevention's (CDC) current definition.

Six teaching hospitals.

We classified a representative sample of 73 positive blood cultures from six hospitals growing common skin contaminant isolates using a definition for bacteremia requiring only microbiology data and the CDC definition for primary bloodstream infection (National Nosocomial Infections Surveillance [NNIS] System review method). The classifications assigned during routine prospective surveillance also were noted, and the time required to classify isolates by the two methods was compared.

Among 65 blood cultures growing common skin contaminant isolates obtained from adults, the agreement rate between the microbiology data method and the NNIS review method was 91%. Agreement was significantly poorer for the eight blood cultures growing common skin contaminant isolates obtained from pediatric patients. The microbiology data method requires approximately 20 minutes less time per isolate than does routine surveillance.

A definition based on microbiology data alone yields the same result as the CDC's definition in the large majority of instances. It is more resource-efficient than the CDC's current definition

To evaluate the distribution of Bordetella pertussis and respiratory syncytial virus (RSV) in the hospital setting.

Air samples were collected using filters in the hospital rooms of 12 children with pertussis and 27 children with RSV infection. Material eluted from these filters was subjected to RSV- and B pertussis-specific polymerase chain reaction (PCR) amplification.

Patients were hospitalized in private rooms in one of two referral centers, a university teaching hospital and a university-affiliated private children's hospital.

12 children (16 days-3 years of age) with documented pertussis infection and 27 patients (10 days-7 years of age) with documented RSV infection.

B pertussis DNA was detected in 7 (58%) of 12 rooms housing pertussis patients and in 16 (25%) of 63 total samples. B pertussis DNA was detected as far as 4 m away from the patient's bedside. The detection of B pertussis DNA in air samples did not change over the short duration of hospitalization. RSV RNA was detected in 17 (63%) of 27 rooms housing RSV-infected patients and in 32 (22%) of 143 total samples. RSV RNA was detected at distances as far as 7 m from the patient's bedside and for up to 7 days of hospitalization.

Using PCR-based detection methods, B pertussis DNA and RSV RNA both can be detected in air samples from the hospital rooms of infected patients. Both can be detected at large distances from a patient's bedside in a minority of cases. These detection methods are suitable for further studies of control measures used to contain nosocomial infections caused by both B pertussis and RSV.

To determine risk factors for an increase in line-associated bloodstream infections (BSIs) in three pediatric intensive-care units at one hospital that recently had changed brands of needleless access device.

Retrospective case-control studies; review of the units' infection control policies and procedures for accessing and replacing components of needleless access devices.

A community tertiary-care hospital's three pediatric intensive-care units.

Children in one of the three intensive-care units with a central venous catheter in place during January 1, 1995, through May 15, 1996, who developed laboratory-confirmed primary BSI. Children who had central venous catheters in place for >48 hours and who did not develop BSI were chosen randomly as controls.

Eight patients met the case definition; they had 11 episodes of BSI. Multivariate analysis identified duration of catheterization and exposure to the IVAC first-generation needleless device as independent risk factors for BSI. Compared with patients from another pediatric intensive-care unit in which the IVAC device also was used but in which an increased BSI rate did not occur, patients from the unit with an increased BSI rate were more likely to receive intermittent (vs continuous) intravenous therapy through one or more lumens. In both units, the IVAC device valve component was replaced every 6 days, and the endcap used to cover the valve (when connected to an unused lumen) was replaced every 24 hours or after each access. The BSI rate returned to baseline after institution of a policy to replace the entire IVAC device, valve, and endcap every 24 hours.

An increased risk of BSI was associated with use of the IVAC first-generation needleless device when replaced every 6 days. This increased risk may have been more pronounced in one pediatric intensive-care unit, because patients were more likely to receive intermittent intravenous therapy. Intermittent intravenous therapy or central venous catheter flushing practices may be important determinants of BSI risk

To test the hypothesis that use of disposable thermometers would result in lower rates of nosocomial Clostridium difficile diarrhea and of total nosocomial infections, compared with electronic thermometers.

Prospective randomized crossover trial.

A 700-bed university hospital providing primary and tertiary care.

All patients admitted to a group of 20 inpatient nursing units.

20 nursing units were randomized into two groups. One group randomly was assigned exclusive use of single-use disposable thermometers for patient temperature measurement, and the other group was assigned exclusive use of electronic thermometers. After 6 months, the assignments were reversed.

Rates of C difficile infections, total nosocomial diarrheal episodes, and total nosocomial infections were prospectively followed in each study unit over 11 months.

26,350 patients were admitted to the study units and hospitalized for 120,529 patient days. There were 947 nosocomial infections (7.86 per 1,000 patient days). Nosocomial C difficile- associated diarrhea defined by positivity to both toxin B (titer ≥1:10) and toxin A was detected in 32 patients (3.4% of all nosocomial infections). A significantly lower rate of nosocomial C difficile-associated diarrhea was observed with disposable thermometer use (0.16 per 1,000 patient days) compared with electronic thermometer use (0.37 per 1,000 patient days, relative risk [RR]=0.44; 95% confidence interval [CI95], 0.21-0.93, P=.026). There was no difference in overall rates of nosocomial infection between the disposable and electronic groups (8.03 and 7.68 infections per 1,000 patient days, respectively; RR, 1.04; CI95, 0.92-1.19; P=.52) or in the overall rate of nosocomial diarrhea (3.34 and 3.40 per 1,000 patient days, respectively; RR, .98; CI95, 0.81-1.19; P=.87).

The incidence of nosocomial C difficile diarrhea was reduced significantly by using single-use, disposable thermometers as compared with electronic thermometers, but there was no effect on either the overall rate of nosocomial diarrhea or the rate of total nosocomial infections.