Importance: Pneumonia is a leading cause of morbidity and mortality in children. It is important to identify the clinical symptoms and physical examination findings associated with pneumonia to improve timely diagnosis, prevent significant morbidity, and limit antibiotic overuse.
Objective: To systematically review the accuracy of symptoms and physical examination findings in identifying children with radiographic pneumonia.
Data sources and study selection: MEDLINE and Embase (1956 to May 2017) were searched, along with reference lists from retrieved articles, to identify diagnostic studies of pediatric pneumonia across a broad age range that had to include children younger than age 5 years (although some studies enrolled children up to age 19 years); 3644 unique articles were identified, of which 23 met inclusion criteria.
Data extraction and synthesis: Two authors independently abstracted raw data and assessed methodological quality. A third author resolved disputes.
Main outcomes and measures: Likelihood ratios (LRs), sensitivity, and specificity were calculated for individual symptoms and physical examination findings for the diagnosis of pneumonia. An infiltrate on chest radiograph was considered the reference standard for the diagnosis of pneumonia.
Results: Twenty-three prospective cohort studies of children (N = 13 833) with possible pneumonia were included (8 from North America), with a range of 78 to 2829 patients per study. The prevalence of radiographic pneumonia in North American studies was 19% (95% CI, 11%-31%) and 37% (95% CI, 26%-50%) outside of North America. No single symptom was strongly associated with pneumonia; however, the presence of chest pain in 2 studies that included adolescents was associated with pneumonia (LR, 1.5-5.5; sensitivity, 8%-14%; specificity, 94%-97%). Vital sign abnormalities such as fever (temperature >37.5°C [LR range, 1.7-1.8]; sensitivity, 80%-92%; specificity, 47%-54%) and tachypnea (respiratory rate >40 breaths/min; LR, 1.5 [95% CI, 1.3-1.7]; sensitivity, 79%; specificity, 51%) were not strongly associated with pneumonia diagnosis. Similarly, auscultatory findings were not associated with pneumonia diagnosis. The presence of moderate hypoxemia (oxygen saturation ≤96%; LR, 2.8 [95% CI, 2.1-3.6]; sensitivity, 64%; specificity, 77%) and increased work of breathing (grunting, flaring, and retractions; positive LR, 2.1 [95% CI, 1.6-2.7]) were signs most associated with pneumonia. The presence of normal oxygenation (oxygen saturation >96%) decreased the likelihood of pneumonia (LR, 0.47 [95% CI, 0.32-0.67]).
Conclusions and relevance: Although no single finding reliably differentiates pneumonia from other causes of childhood respiratory illness, hypoxia and increased work of breathing are more important than tachypnea and auscultatory findings.