Predicting bloodstream infection via systemic inflammatory response syndrome or biochemistry

The yield of blood cultures is approximately 10%. This could be caused by inaccurate prediction of patients with bloodstream infection (BSI).

To evaluate the usability of systemic inflammatory response syndrome (SIRS) or biochemical analyses as predictors for positive blood culture.

We conducted a prospective cohort study at a Danish regional hospital from February 1 to April 30, 2010. All adult patients were included on the first time blood cultures were sampled during admission. Data were obtained from medical records, databases on microbiology, biochemistry, and antibiotic treatment. Data included time of admission, date and result of blood culture, results of biochemical analyses, and clinical measurements on the day of blood culture. Prediction of BSI was analyzed according to both individual parameters and parameters combined in different sepsis score groups. Associations were calculated using multiple logistic regression.

Patients with BSI (68 patients) were compared to patients without BSI (828 patients). Respiratory rate, body temperature, and C-reactive protein were strongest associated with BSI, with adjusted odds ratio (OR) 5.42, 95% confidence interval (CI) 1.13-25.9; OR 2.55, 95% CI 1.34-4.87; and OR 6.06, 95% CI 0.82-44.6, respectively. SIRS was associated with BSI, with crude OR 7.25, 95% CI 1.75-30.1. Neutrophil count and p-carbamide were not associated with BSI: adjusted OR 0.88, 95% CI 0.36-2.13 and OR 1.44, 95% CI 0.82-2.52, respectively. Only one of the sepsis score groups was associated with BSI: crude OR 2.13, 95% CI 1.08-4.19.

SIRS is an adequate predictor of BSI. By contrast, biochemical parameters were not useful as predictors of BSI.