Predicting bacteremia in hospitalized patients. A prospectively validated model
D W Bates1, E F Cook, L Goldman
1Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts.
Annals of Internal Medicine
|October 1, 1990
Summary
This study developed a model to predict bacteremia in hospitalized patients, identifying low-risk groups where positive blood cultures are less reliable. The model helps stratify patient risk for better clinical decisions.
Area of Science:
- Infectious Diseases
- Clinical Microbiology
- Epidemiology
Background:
- Bacteremia poses a significant risk to hospitalized patients.
- Accurate prediction of bacteremia is crucial for timely and appropriate treatment.
- Current methods for diagnosing bacteremia can have limitations in predictive accuracy.
Purpose of the Study:
- To develop and validate a predictive model for bacteremia in hospitalized individuals.
- To identify patient subgroups with a low probability of bacteremia.
- To assess the positive predictive value of blood cultures in specific patient populations.
Main Methods:
- Prospective cohort study involving 1516 blood culture episodes.
- Data collected from a tertiary care hospital.
- Multivariate analysis to identify predictors of true bacteremia.
Main Results:
- Key predictors of bacteremia included high temperature, fatal disease, shaking chills, IV drug use, acute abdomen, and major comorbidity.
- A low-risk group showed a 1% misclassification rate and low positive predictive value (14%) for positive blood cultures.
- A high-risk subset had a true bacteremia rate of 16%.
Conclusions:
- The developed model effectively stratifies hospitalized patients by bacteremia risk.
- This model can aid decisions regarding blood culture ordering and antibiotic initiation.
- Further validation in diverse settings is recommended for broader clinical application.
