[Septicemia associated with central venous catheterization in a children's hospital. A multivariate study].

BACKGROUND: The blood stream infections (BSI) are the principal nosocomial infection in the child hospitals. In this study we estimate the incidence of BSI associated with central venous catheterization, and estimate different risk and protective factors, through a multivariate study.

MATERIAL AND METHODS: The study have followed in a prospective way during 6 months all the children with central venous catheterization (489 catheters), from the moment of insertion until withdrawal, collecting various data previous to the development of the infection: place of insert, type of catheter, duration, clinic information, microbiology, and the treatments administered through the catheter. In was accomplished an multivariate analysis with logistic regression, for two principal effect variables, the catheter colonization and the catheter related BSI.

RESULTS: The incidence of catheter related BSI was 5.5% and for local infection 11.2%. The density of incidence was 3.15 and 6.42 for each 1,000 catheters-day, respectively. The logistic regression model included: colonization of the skin in the insertion point > 15 colonies, days with antibiotics through catheter, use of lipidic parenteral solutions and fever, previous to the infection. The area under the ROC curve was 0.72.

CONCLUSIONS: In children with septicemias associated with central catheterization the predictors or sentry criterion for the decision on when to withdraw a catheter are colonization (> 15 colonies) of the insert point, together with the use of lipidic parenteral solutions or extended antibiotic treatment.