Background Acute respiratory infection remains a common presentation to Emergency Departments. Oxygen saturations (Sao₂) may be useful in determining which febrile infants require chest x-rays (CXR) in investigation for bacterial pneumonia (PNA). This study aimed to determine whether Sao₂ is clinically useful in excluding bacterial PNA in febrile infants <24 months.

Methods A febrile infant registry was instituted at a tertiary care military hospital (55 000 annual patients, 27% children) from December 2002–December 2003. Eligible patients consisted of infants <3 months with temperature ≥38°C or 3–24 months with temperature ≥39°C. Bacterial PNA was defined in this cohort by a CXR revealing a ‘lobar infiltrate’ by a board-certified radiologist. Descriptive statistics are presented on groups who received CXR versus groups who did not, and on infants who had bacterial PNA versus those who did not. Student t tests were used to compare maximum temperature (Tmax), RR, and Sao₂. Logistic regression for PNA was performed using age, sex, Tmax, RR, HR and Sao₂. A Receiver Operator Characteristic (ROC) curve was created to show Sao₂ cut-off points as related to sensitivity and specificity.

Results 985 patients (55% boys; median age: 12 months) met entry criteria. 790 underwent CXR and 82 were diagnosed with bacterial PNA. Sao₂ was lower in infants with bacterial PNA (96.6%±2.5% vs 97.7%±1.8%, p<0.001). Sao₂ was also predictive of bacterial PNA by logistic regression (p=0.017) but the ROC curve yielded a poor sensitivity/specificity profile (area under curve (AUC) of 0.6786).

Conclusions In febrile infants, Sao₂ was not found to be clinically useful for excluding bacterial PNA.