Background: Nosocomial infection have a relative high prevalence, which is necessary to reduce in order to improve the quality of patient care. The aims of this work is to study the behaviour of Serratia marcescens in our hospital as between 1987 and 1995.
Methods: Between February 1987 and March 1995 we detected 679 isolates of Serratia marcescens in 504 patients. We used serotype as first marker and phagotype as second, and evaluation of PGFE as a discriminator of doubtful strains was done.
Results: 35.8% of the strains were from the respiratory tract, 37.2% from urinary tract; and 11.7% were isolated in blood culture, among them 23.3% came form children younger than 7 years. We noticed a tendency to decrease the number of isolates along the studied period. The most frequent serotypes were O:6, O:3 and O:2; representing the 36.0% of the total. Serotypes O:6;14, O:4, O:5, O:11 and polyagglutinables accounted for 37% of the total. The frequency was variable from year to year, and the predominant serotypes were different in every one of the hospitals in which the center is divided. A 60% of the patients were hospitalized in the General Hospital Vall d'Hebron building, in ICU (20%) and in chirurgical services (25%). Ninety-six patients had more than one isolate, 91 of them (94.8%) can be classified by phenotypic test. The PFGE is discriminatory in three of the five unclassificable isolates. In more than 35% of the patients the strains isolated along the time are different. The 66.7% of the patients acquired Serratia strains in the same admission, and in some cases with few days of difference. We detected 17 cross infections, predominantly in ICUs. With PFGE we could discriminate isolates which produced cross infections between patient who are not in the same hospital.
Conclusions: Although the prevalence of Serratia marcescens is diminishing, it is able to produce crossed infections that, in general, affected few patients. The serotype and phagotype discriminate 94.8% of isolates. The PFGE is high discriminatory and reproducible, only 6.8% of the 44 strains tested by this method can not be typed.