To investigate and control an outbreak of bloodstream infections (BSIs) caused by Serratia marcescens and to identify risk factors for respiratory colonization or infection with S. marcescens.

Epidemiologic investigation, including review of medical and laboratory records, procedural investigations, pulsed-field gel electrophoresis (PFGE) typing of environmental and patient isolates, statistical study, and recommendation of control measures.

All patients admitted to a 380-bed, secondary-care hospital in Osaka Prefecture, Japan, from July 1999 through June 2000 (study period).

Seventy-one patients were colonized or infected with S. marcescens; 3 patients who developed primary BSIs on the same ward within 5 days in June 2000 had isolates with indistinguishable PFGE patterns and indwelling intravenous catheters for more than 5 days. On multivariate analysis, among 36 case-patients with positive sputum specimens and 95 control-patients, being bedridden (odds ratio [OR], 15.91; 95% confidence interval [CI95], 4.17-60.77), receiving mechanical ventilation (OR, 7.86; CI95, 2.27-27.16), being older than 80 years (OR, 3.12; CI95, 1.05-9.27), and receiving oral cleaning care (OR 3.10; CI95, 1-9.58) were significant risk factors. S. marcescens was isolated from the fluid tanks of three nebulizers and a liquid soap dispenser. The hospital did not have written infection control standards, and many infection control practices were found to be inadequate (eg, respiratory equipment was used without disinfection between patients).

Poor hospital hygiene and the lack of standard infection control measures contributed to infections hospital-wide. Recommendations to the hospital included adoption of written infection control policies.