In pediatric patients younger than 5 years of age, no single historical feature or finding can differentiate pneumonia from other childhood respiratory illnesses reliably, according to the results of a recent review published in the Annals of Emergency Medicine.
The investigators conducted electronic literature searches between 1956 and May 2017 to identify diagnostic studies of pneumonia in children. A total of 23 studies comprising 13,833 patients were ultimately included in this review, with dates of the selected studies ranging from 1982 to 2017. The studies that were chosen ranged in size from 78 to 2829 pediatric patients.
Of the 23 studies selected, 2 included preverbal children only, 10 included preverbal children 5 years of age or younger, 2 included children 6 years of age or younger, and 9 included a broad range of children from preverbal age to adolescence. Overall, 8 studies were conducted in North America, and the other 15 were conducted in countries outside of North America. The overall rate of pneumonia was 19% in the North American studies, compared with a 37% overall rate in the studies conducted outside of North America.
Data on the following clinical symptoms were reported by the researchers: chest pain, cough, difficulty breathing, poor feeding, symptom duration lasting longer than 3 days, and vomiting or diarrhea. Of these symptoms, none had a significantly predictive positive likelihood ratio (LR+), whereas absence of cough was the only finding with a potentially useful negative likelihood ratio (LR–) of 0.47 (95% CI, 0.24-0.70).
Data on the following clinical signs were also reported in the studies: oxygen saturation, auscultatory findings, and work of breathing. Moderate hypoxemia (oxygen saturation ≤95%) and increased work of breathing (eg, grunting, flaring, or retractions) were the 2 clinical signs that were most associated with pneumonia (LR+ 3.5 and 2.1, respectively). Normal oxygen saturation (>96%) decreased a child’s likelihood of having pneumonia (LR– 0.47).
According to the investigators, no single finding is adequate to diagnose pneumonia or exclude a diagnosis of pneumonia reliably. Although LRs may be statistically significant, the closer an LR value is to 1, the less useful it will be in increasing or decreasing probability of disease. Future studies should be designed to assess the diagnostic value of combinations of the signs and symptoms of pneumonia to help determine which patients would benefit most from undergoing radiography and receiving antibiotics.
Gottlieb M, Heinrich SA. How reliable are signs and symptoms for diagnosing pneumonia in pediatric patients? [published online October 26, 2017]. Ann Emerg Med. doi:10.1016/j.annemergmed.2017.09.028