To determine whether hand washing would increase with sustained feedback based on measurements of soap and paper towel consumption.

Prospective trial with a nonequivalent control group.

Open multibed rooms in the Omaha Veterans Affairs Medical Center's Surgical Intensive Care Unit (SICU) and Medical Intensive Care Unit (MICU).

Unit staff.

Every weekday from May 26 through December 8,1998, we recorded daytime soap and paper towel consumption, nurse staffing, and occupied beds in the SICU (intervention unit) and the MICU (control unit) and used these data to calculate estimated hand washing episodes (EHWEs), EHWEs per occupied bed per hour, and patient-to-nurse ratios. In addition, from May 26 through June 26 (baseline period) and from November 2 through December 8 (follow-up period), live observers stationed daily for random 4-hour intervals in the MICU and the SICU counted actual hand washing episodes (CHWEs). The intervention consisted of posting in the SICU, but not in the MICU, a graph showing the weekly EHWEs per occupied bed per hour for the preceding 5 weeks.

Directly counted hand washing fell in the SICU from a baseline of 2.68 ± 0.72 (mean ± standard deviation) episodes per occupied bed per hour to 1.92 ± 1.35 in the follow-up period. In the MICU, episodes fell from 2.58 ± 0.95 (baseline) to 1.74 ± 0.69. In the MICU, the withdrawal of live observers was associated with a decrease in estimated episodes from 1.36 ± 0.49 at baseline to 1.01 ± 0.36, with a return to 1.16 ± 0.50 when the observers returned. In the SICU, a similar decrease did not persist throughout a period of feedback. Estimated hand washing correlated negatively with the patient-to-nurse ratio (r = -0.35 for the MICU, r = -0.46 for the SICU).

Sustained feedback on hand washing failed to produce a sustained improvement. Live observers were associated with increased hand washing, even when they did not offer feedback. Hand washing decreased when the patient-to-nurse ratio increased.

To determine the percentage of cross-transmissions in an intensive care unit (ICU) with high nosocomial infection (NI) rates according to the data of the German Nosocomial Infection Surveillance System.

A 14-bed surgical ICU of a 1,300-bed, tertiary-care teaching hospital.

Prospective surveillance of NIs during a period of 9 months. If an NI was present, the isolates of the following indicator pathogens were stored and typed by species: Staphylococcus aureus, Enterococcus species, Escherichia coli, Pseudomonas aeruginosa, Acinetobacter baumannii, and Enterobacter species. Pulsed-field gel electrophoresis was performed for typing of S. aureus strains and arbitrarily primed polymerase chain reaction was applied for the other pathogens. The presence of two indistinguishable strains in two patients was considered as one episode of cross-transmission.

Two hundred sixty-two patients were observed during a period of 2,444 patient-days; 96 NIs were identified in 59 patients and the overall incidence density of NI was 39.3 per 1,000 patient-days. For 104 isolates, it was possible to consider typing results. Altogether, 36 cross-transmissions have lead to NIs in other patients. That means at least 37.5% of all NIs identified were due to cross-transmissions.

Because of the method of this study, the percentage of NIs due to cross-transmission identified for this ICU is an “at least number”. In reality, the number of cross-transmissions, and thus the number of avoidable infections, may have been even higher. However, it is difficult to assess whether the percentage of NIs due to cross-transmission determined for this ICU may be the crucial explanation for the relatively high infection rate in comparison to other surgical ICUs.

To examine the relationship between nurse staffing levels and the rate of nosocomial viral gastrointestinal infections (NVGIs) in a general pediatrics population.

Retrospective descriptive study.

A general pediatrics ward at The Hospital for Sick Children in Toronto, Ontario, Canada, a 320-bed, tertiary-care pediatric institution.

Forty-three NVGIs were detected in 37 patients of 2,929 admissions (1.3%). The monthly NVGI rate correlated significantly with the monthly night patient-to-nurse ratio (r = 0.56) and the monthly day patient-to-nurse ratio (r = 0.50). The nursing hours per patient-day during the preinfection period (PIP) were significantly lower than those during the nonpreinfection period (NPIP; 12.5 vs 13.0). There was no difference between the PIP and the NPIP day patient-to-nurse ratios (3.31 vs 3.32), but there was a significant difference between the PIP and the NPIP night patient-to-nurse ratios (3.26 vs 3.16). The incidence of NVGIs in the 72-hour period after any day when the nursing hours per patient-day were less than 10.5 was 6.39 infections per 1,000 patient-days, compared with 2.17 infections per 1,000 patient-days in periods with more than 10.5 nursing hours per patient-day (rate ratio, 2.94; 95% confidence interval, 2.16 to 4.01).

Nurse understaffing contributed to an increased NVGI rate in our general pediatrics population, and should be assessed as a risk factor in outbreak investigations.

To assess the healthcare burden, morbidity, and mortality of nosocomial Clostridium difficile-associated diarrhea (N-CDAD) in Canadian hospitals.

Laboratory-based prevalence study.

Nineteen acute-care Canadian hospitals belonging to the Canadian Hospital Epidemiology Committee surveillance program.

Hospitalized patients in the participating centers.

Laboratory-based surveillance was conducted for C. difficile toxin in stool among 19 Canadian hospitals from January to April 1997, for 6 continuous weeks or until 200 consecutive diarrhea stool samples had been tested at each site. Patients with N-CDAD had to fulfill the case definition. Data collected for each case included patient demographics, length of stay, extent of diarrhea, complications of CDAD, CDAD-related medical interventions, patient outcome, and details of death.

We found that 371 (18%) of 2,062 tested patients had stools with positive results for C. difficile toxin, of whom 269 (13%) met the case definition for nosocomial CDAD. Of these, 250 patients (93%) had CDAD during their hospitalization, and 19 (7%) were readmitted because of CDAD (average readmission stay, 13.6 days). Forty-one patients (15.2%) died, of whom 4 (1.5% of the total) were considered to have died directly or indirectly of N-CDAD. The following N-CDAD-related morbidity was noted: dehydration, 3%; hypokalemia, 2%; gastrointestinal hemorrhage requiring transfusion, 1%; bowel perforation, 0.4%; and secondary sepsis, 0.4%. The cost of N-CDAD readmissions alone was estimated to be a minimum of $128,200 (Canadian dollars) per year per facility.

N-CDAD is a common and serious nosocomial infectious complication in Canada, is associated with substantial morbidity and mortality, and imposes an important financial burden on healthcare institutions.

The Centers for Disease Control and Prevention and the American Thoracic Society recommend obtaining cultures of at least three sputum specimens for acid-fast bacilli (AFB) from patients in whom tuberculosis (TB) is suspected. On the basis of this, most hospitals isolate patients with suspected TB for 3 days or more until three smear (not culture) results are negative. Our objective was to evaluate the predictive value and sensitivity of these smears.

Observational study.

Four urban medical centers.

The posttest probability of TB given sequential negative AFB smears from 274 patients isolated for suspected TB and the sensitivity of sequential AFB smears from 209 patients with positive results on culture for pulmonary TB were measured.

The posttest probabilities of having TB given one, two, and three negative AFB smears were low: 1.1% (3 of 265; 95% confidence interval [CI95], 0.23% to 3.27%), 0.4% (1 of 262; CI95, 0% to 2.1%), and 0% (0 of 260; CI95, 0% to 1.4%), respectively. Among the 209 patients with positive results on culture for pulmonary TB, 169 (81%) had an expectorated sputum specimen sent, of which 91 (54%) were positive for AFB. Forty (24%) of the 169 patients had a second expectorated sputum specimen sent after the results of the first specimen were negative; only 6 (15%) of these had positive AFB smears. None of the 10 patients in whom the first two expectorated sputum samples yielded an AFB smear without an organism had a third AFB smear that was positive.

Unless there is high clinical suspicion of pulmonary TB in a specific patient, the use of three AFB smears on expectorated sputa is a rational approach to discontinuing isolation for patients with suspected TB.