No clinical criteria were powerful diagnostic discriminators of the presence or absence of pneumonia in children, though some were fair. When in doubt, a CXR is probably warranted, with the exceptions of obvious bronchiolitis or asthma. Low SpO2 (</= 95 to 96%) or increased work of breathing were the best predictors of radiographic pneumonia in children; auscultatory findings and tachypnea were poor. You don't need a CXR if no cough, no fever, no tachypnea, and normal SpO2.
Why does this matter?
Some work in settings without immediate access to x-ray (i.e. retail clinics or some urgent care clinics). This paper can help you decide which patients need imaging or empiric treatment for pneumonia.
"The lungs should be seen and not heard." Dr. Keith Wrenn
This was a systematic review which found 23 studies of clinical findings as predictors of radiographic pneumonia in pediatric patients. The prevalence of pneumonia in these North American children with respiratory symptoms was 19%. They found that hypoxia, specifically SpO2 96% or less, and increased work of breathing (retractions, grunting, nasal flaring) were the best predictors of pneumonia on CXR. Chest pain, fever, and tachypnea were weakly positive predictors. Lung auscultation did not predict pneumonia. The positive likelihood ratio of crackles (rales) on auscultation was 1.2 (95%CI 0.53-1.8); the negative likelihood ratio was 0.90 (95%CI 0.14-2.5). That means auscultation didn't help at all in making the clinical diagnosis of pneumonia in pediatric patients. Normal oxygen saturation >96% and absence of cough, fever, and tachypnea were negative predictors for pneumonia.
Does This Child Have Pneumonia?: The Rational Clinical Examination Systematic Review. JAMA. 2017 Aug 1;318(5):462-471. doi: 10.1001/jama.2017.9039.
Peer reviewed by Thomas Davis, MD.