To investigate an outbreak of Serratia marcescens in a maternity hospital (November 1994 to May 1995).

Retrospective analysis of epidemiological data and prospective study of systematic bacteriological samples from patients and environment, with genotyping of strains by arbitrarily primed polymerase chain reaction.

A private maternity hospital, Saint-Etienne, France.

In the neonatal unit, 1 newborn developed a bacteremia, and 36 were colonized in stools with S marcescens. As the colonization of some newborns was shown to occur only a few hours after delivery, the inquiry was extended to other maternity wards, where 8 babies and 4 mothers were found to be colonized. Environmental sampling led to the isolation of S marcescens from a bottle of enteral feed additive in the neonatal unit and from the transducers of two internal tocographs in the delivery rooms. The genotyping of 27 strains showed two different profiles: a major epidemic profile shared by 22 strains (18 from babies of the neonatal unit, 2 from babies of other units, and 2 from breast milk) and another profile shared by 5 strains (2 from transducers of internal tocographs, 2 from babies, and 1 from a mother). The strain isolated from lipid enteral feeding was not available for typing. Although this source of contamination was removed soon from the neonatal unit, the outbreak stopped only when infection control measures were reinforced in the delivery rooms, including the nonreuse of internal tocographs.

In delivery rooms, the quality of hygiene needs to be as high as in surgery rooms to prevent nosocomial colonization or infection of neonates at birth.

To determine mortality, morbidity, and costs attributable to surgical-site infections (SSIs) in the 1990s.

A matched follow-up study of a cohort of patients with SSI, matched one-to-one with patients without SSI.

A 415-bed community hospital.

255 pairs of patients with and without SSI were matched on age, procedure, National Nosocomial Infection Surveillance System risk index, date of surgery, and surgeon.

Mortality, excess length of hospitalization, and extra direct costs attributable to SSI; relative risk for intensive care unit (ICU) admission and for readmission to the hospital.

Of the 255 pairs, 20 infected patients (7.8%) and 9 uninfected patients (3.5%) died during the postoperative hospitalization (relative risk [RR], 2.2; 95% confidence interval [CI95], 1.14.5). Seventy-four infected patients (29%) and 46 uninfected patients (18%) required ICU admission (RR, 1.6; CI95,1.3-2.0). The median length of hospitalization was 11 days for infected patients and 6 days for uninfected patients. The extra hospital stay attributable to SSI was 6.5 days (CI95, 5-8 days). The median direct costs of hospitalization were $7,531 for infected patients and $3,844 for uninfected patients. The excess direct costs attributable to SSI were $3,089 (CI95, $2,139-$4,163). Among the 229 pairs who survived the initial hospitalization, 94 infected patients (41%) and 17 uninfected patients (7%) required readmission to the hospital within 30 days of discharge (RR, 5.5; CI95, 4.0-7.7). When the second hospitalization was included, the total excess hospitalization and direct costs attributable to SSI were 12 days and $5,038, respectively.

In the 1990s, patients who develop SSI have longer and costlier hospitalizations than patients who do not develop such infections. They are twice as likely to die, 60% more likely to spend time in an ICU, and more than five times more likely to be readmitted to the hospital. Programs that reduce the incidence of SSI can substantially decrease morbidity and mortality and reduce the economic burden for patients and hospitals.

To describe the molecular epidemiology of Legionella pneumophila infections in the University of Iowa Hospitals and Clinics (UIHC).

Molecular epidemiological study using pulsed-field gel electrophoresis (PFGE).

A large university teaching hospital.

All surviving isolates obtained from culture-proven nosocomial L pneumophila infections and all surviving isolates obtained from the University of Iowa Hospital and Clinics' water supply between 1981 and 1993.

Thirty-three isolates from culture-proven nosocomial cases of L pneumophila pneumonia were available for typing. PFGE of genomic DNA from the clinical isolates identified six different strains. However, only strain C (16 cases) and strain D (13 cases) caused more than 1 case. Strain C caused clusters of nosocomial infection in 1981, 1986, and 1993 and also caused 4 sporadic cases. Strain D caused a cluster in 1987 and 1988 plus 4 sporadic cases. Of the six strains causing clinical infections, only strains C and D were identified in water samples. PFGE identified three strains in the water supply, of which strains C and D caused clinical disease and also persisted in the water supply during most of the study period.

Specific strains of L pneumophila can colonize hospital water supplies and cause nosocomial infections over long periods of time.

To investigate a cluster of hepatitis B virus (HBV) infections between December 1995 and May 1996 among chronic hemodialysis patients in one county.

Two dialysis centers (A and B) and a hospital (C) in one county.

Six case-patients who were dialyzed in one of two centers, A and B, and had all been hospitalized between January and February 1996 at hospital C.

Patient 1, usually dialyzed in center A sero-converted to hepatitis B surface antigen (HBsAg) in December 1995 and could have been the source of infection for the others, who seroconverted between March and April 1996. Two cohort studies were conducted: one among patients dialyzed in center A to determine where transmission had occurred, and one among patients dialyzed at hospital C at the time patient 1 was hospitalized, to identify factors associated with infection.

Four (15%) of the 26 susceptible patients dialyzed at center A became infected with HBV. Hospitalization at hospital C when patient 1 was hospitalized was associated with infection (P=.002). A cohort study of the 10 susceptible patients dialyzed at hospital C during the time patient 1 was hospitalized did not identify specific risk factors for infection. However, supplies and multidose vials were shared routinely among patients, providing opportunities for transmission.

When chronic hemodialysis patients require dialysis while hospitalized, their HBsAg status should be reviewed, and no instrument, supplies, or medications should be shared among them.