To determine risk factors for mortality in patients with a nosocomial enterococcal primary bloodstream infection (EPBI) and to assess whether vancomycin resistance placed a patient at increased risk of death.

A retrospective cohort study was conducted in four National Nosocomial Infection Surveillance System hospitals.

Of 145 patients identified with EPBIs, 74 (51%) died, and 26 (18%) had a vancomycin-resistant isolate. Upon comparing patients with EPBIs who survived to those who died, no associations were found between mortality and prior invasive device use, procedure history, type or number of prior nosocomial infections, length of hospitalization before infection, or receipt of vancomycin. Independent predictors of mortality were indices of severity of illness (APACHE II score and comorbidity weighted index), age, the use of third-generation cephalosporins or metronidazole during the week prior to infection, and female gender.

Vancomycin resistance was not an independent predictor of death, and its role was difficult to establish, because cohort patients were among the most severely ill of all hospitalized patients. Enterococcal primary bloodstream infections appear to indicate severe, life-threatening disease processes. The pathogenicity of enterococci and the role of vancomycin resistance as a cause of mortality in patients with EPBIs need to be assessed further.

To determine whether empiric isolation of patients with acute respiratory virus infection symptoms could be discontinued when preliminary shell vial cultures were negative, and the impact of this approach on hospital resources.

In 1993, we retrospectively reviewed respiratory virus test results from 1992 to 1993 and extended data collection prospectively through the 1993 to 1994 season. The rapid test and 48-hour shell vial results were compared to a standard of rapid test plus 5-day shell vial culture results to determine the sensitivity and specificity of these “preliminary” results.

A 400-bed tertiary referral research hospital.

Patients from any inpatient unit or clinic with acute respiratory virus infection symptoms who had a specimen submitted for respiratory virus culture. Patients were placed on empiric respiratory isolation pending culture results.

The overall sensitivity of the combined rapid and 48-hour culture results in adults and children was 97%. All 15 pediatric patients with respiratory syncytial virus infection who had specimens submitted on first suspicion of respiratory virus infection were positive by rapid test. Culture results were positive within 48 hours for 100% of patients with influenza A (15 patients), influenza B (6), and parainfluenza (18) viruses. Of 59 pediatric inpatients who were isolated empirically awaiting 5-day culture results, 31 (52%) ultimately were determined to be culture negative.

Empiric isolation of symptomatic children can be discontinued at 48 hours when both the rapid test and the early culture results are negative. Our institution would have saved 93 days of unnecessary isolation over 2 years had such a policy been in place.

In May 1994, 43 persons in a nursing home were reported with gastroenteritis. An outbreak investigation was conducted to determine risk factors for gastroenteritis among residents and staff.

Data were analyzed using contingency tables; relative risks (RR) and statistical significance were determined with Fisher's Exact Test. The chi-squared statistic to perform a goodness of fit test for the binomial distribution was used to determine whether cases occurred randomly and independently of each other. Stools were tested for bacterial enteric pathogens, ova, and parasites and were examined by electron microscopy, Southern hybridization, and reverse transcription-polymerase chain reaction. Paired sera were collected to detect fourfold rises in antibody titer by enzyme immunoassay against Norwalk viruses.

Of 121 residents, 62 (51%) had gastroenteritis, as did 64 (47%) of the 136 staff. The index case was a nurse who became ill at work and continued to work, while symptomatic, for another 2 days. Only residents who had received medications from this nurse between May 17 and May 20 became ill on the first day of the outbreak (13 of 35 versus 0 of 5). Nurses and nurse aides were more likely than employees without direct resident contact to be cases (46 of 68 versus 18 of 58; RR, 2.18;P<.001). Bacterial stool cultures and parasite examinations were negative. Results of electron microscopy, polymerase chain reaction with Southern hybridization, and enzyme immunoassay indicated the causative agent was a small, round, structured virus similar to the Snow Mountain Agent.

To minimize outbreaks in nursing homes, we recommend that ill staff be excluded from work until symptoms resolve.