Risk factors and a clinical index for diagnosis of Pseudomonas aeruginosa bacteremia.

OBJECTIVE: To define risk factors significantly and independently associated with Pseudomonas aeruginosa bacteremia and to combine them in a diagnostic index which will define groups of septic patients at low or high risk of bloodstream infection caused by P. aeruginosa. METHODS: Logistic regression analysis was used to identify risk factors associated with pseudomonal bacteremia among all patients with bacteremia at St Thomas' Hospital, London, from 1969 to 1989. The coefficients of the final logistic model were used to compile a linear model for the diagnosis of pseudomonal bacteremia. The index was validated in another set of patients with bacteremia detected at Beilinson Medical Center, Petah Tiqva, Israel, from 1988 to 1991. RESULTS: Seven factors were significantly and independently predictive of pseudomonal bacteremia: 1) neutropenia; 2) previous or current treatment with antibiotics; 3) cytotoxic or corticosteroid treatment; 4) hospital acquisition of infection; 5) detection in the intensive care unit; 6) male gender; and 7) focus of infection. High-risk foci were the urinary tract with catheter or post-instrumentation, or unknown source. Low-risk foci were bone, joint, meninges, female genital tract and upper respiratory tract. The index score divided patients into three groups with increasing likelihood of P. aeruginosa bacteremia: 1%, 7% and 19%, respectively (p=0.0001). In the validation set, the percentage of P. aeruginosa bacteremia in the three respective groups defined by the index were 5%, 18% and 22% (p=0.0001). CONCLUSIONS: The use of simple clinical and laboratory data known within hours of detection of an infectious episode can define patients at high and low risk for P. aeruginosa bacteremia.