1742-6723/asset/EMM_left.gif?v=1&s=534581e83b0d76db78894af6dc20181b36b346f5)
1742-6723/asset/EMM_centre.gif?v=1&s=6ff954646038c2b7dc71c36387f999d5af0c44e5)
1742-6723/asset/EMM_right.gif?v=1&s=49e2c6a01efaa239b09b2204155edc68d9a9e82e)
Prediction tool for bacteraemia in children aged 3–36 months
Article first published online: 25 JUN 2007
DOI: 10.1111/j.1742-6723.2007.00981.x
Additional Information
How to Cite
Stathakis, T., Acworth, J. P. and Barnett, A. G. (2007), Prediction tool for bacteraemia in children aged 3–36 months. Emergency Medicine Australasia, 19: 353–358. doi: 10.1111/j.1742-6723.2007.00981.x
Publication History
- Issue published online: 22 JUL 2007
- Article first published online: 25 JUN 2007
- Accepted 30 April 2007
- Abstract
- Article
- References
- Cited By
Keywords:
- bacteraemia;
- fever without source;
- fever;
- prediction tool;
- risk of bacteraemia
Abstract
Purpose: The present study aimed to determine which parameter is the most reliable predictor of bacteraemia in children aged 3–36 months and to develop a simple tool to assess risk of bacteraemia.
Method: A retrospective review was performed on patients aged 3–36 months who presented to a paediatric ED between July 1999 and April 2004. Children with a febrile illness who had a blood culture and full blood count performed were included in the study. Data were collected from the pathology database (AUSLAB) and Emergency Department Information System (EDIS). Variables examined were age, sex, temperature at presentation, white cell count, neutrophil count and blood culture result. Multiple regression analysis was used to determine the independent predictors of bacteraemia. Non-linear regression analysis was applied to explore alternative patterns of bacteraemia risk.
Results: Of the 1488 patients in the dataset, 43 were bacteraemic (2.9%). The most common organism was pneumococcus (74.4%). Sex was evenly distributed (male 52.4%, female 47.6%). Mean temperature at presentation was 38.8°C (95% confidence interval 35.5–41.4°C). The optimal logistic regression model identified neutrophil count as the variable most predictive of bacteraemia, with the odds of bacteraemia increasing by 1.11 for each one-unit increase in neutrophil count.
Conclusions: Neutrophil count is the strongest predictor of bacteraemia in febrile children aged 3–36 months. Based on this, a simple prediction tool can be used to risk stratify this population, and assist in clinical decision making.