Abstract
PURPOSE
Aggressive treatment of bacteremia in patients at high risk of septic shock may prove to be beneficial and cost-effective since these patients account for 50% of the mortality. The purpose of this study was to identify risk factors for septic shock in bacteremic patients.
PATIENTS AND METHODS
We retrospectively collected clinical, biologic, and radiologic data from medical records of patients who experienced one episode of bacteremia between March 1 and December 31, 1990, at the University Hospital of Dijon, France. Statistical analysis was univariate and multivariate (logistic regression).
RESULTS
During the survey, 331 patients experienced one episode of bacteremia. Bacteremia due to gram-negative species was more frequently found in surgical units. Septic shocks were equally attributed to gram-negative and gram-positive bacteria. The do-not-resuscitate order was associated with nearly half of the deaths secondary to underlying diseases or poor general condition. Neither hospital-acquired bacteremia nor underlying diseases were associated with a high risk of septic shock. Multivariate analysis showed the following to be risk factors for septic shock: male gender and age older than 75 years (odds ratio = 2.91), creatinine level greater than 20 mg/L (odds ratio = 4.52), prothrombin time less than 60% (odds ratio = 4.86), and presence of an interstitial pattern on the chest radiograph involving more than half of both lung fields (odds ratio = 4.1). Our logistic model with these selected parameters allowed an overall classification of 77% of the studied patients, with a positive predictive value of 55% and a negative predictive value of 89%.
CONCLUSION
Our results provide a positive predictive value that is similar to that of the clinical scores of severity proposed by other authors in selected populations. Our classification, in conjunction with a clinical classification, would allow improvement in the prognosis of patients with bacteremia by early treatment of those at high risk for developing septic shock. Validation of these results with further studies involving a larger population that includes patients with suspected infections, as well as bacteremic patients, is needed.
Collapse