The evaluation of febrile children presents a challenge because of the inherent difficulty of the pediatric examination and variability in physician confidence in assessing the severity of a child's illness. Clearly, the first step is obtaining a careful history and conducting a physical examination, followed by appropriate laboratory tests. Tests are considered valuable if they prompt other tests, change therapy or alter follow-up care. White blood cell (WBC) counts are often used to screen for underlying bacterial infection. Increased WBC counts may lead to additional testing, including blood cultures, urinalysis and chest radiographs, or empiric antibiotic treatment because of the possibility of occult bacteremia. However, many experts discourage routine chest radiographs in the absence of respiratory findings. Bachur and associates evaluated the incidence of occult pneumonias seen on chest radiographs in children with fever, leukocytosis and an absence of focal infection on physician examination.

The study adopted guidelines for obtaining chest radiographs in febrile children five years of age or younger established by the emergency department at Children's Hospital in Boston. Febrile children in this age group seen in the emergency department with leukocytosis (a WBC count of 20,000 per mm³ [20 × 10⁹ per L]) or more, a triage temperature of 39.0°C (102.2°F) or higher and no major focal signs of infection were eligible for a chest radiograph. Physicians responsible for the initial evaluation in the emergency department completed a questionnaire indicating diagnosis, general patient appearance, presence of respiratory signs and symptoms, presence of fever and reasons for and findings of chest radiographs, if obtained. Patients were divided into three study groups based on their clinical presentation. Those with a fever and respiratory signs or symptoms of pneumonia were in the “signs” group. Those with fever but only minor or no bacterial infection were in the “no signs” group. Patients eligible for chest radiographs, but who did not receive them, were in the “no radiograph” group. Pneumonias identified in the no-signs group were labeled occult; those detected in the signs group were labeled nonoccult.

Of 389 patients who met the study criteria, 278 were eligible for a chest radiograph (79 patients in the signs group, 146 in the no-signs group and 53 in the no-radiograph group). Occult pneumonia was identified in 26 percent of patients with a WBC count between 20,000 and 25,000 per mm³ (20 and 25 × 10⁹ per L) and in 31 percent of patients with a count greater than 25,000 per mm³. When these results are combined with data from the no-radiograph group, the minimum estimate of cases of occult pneumonia would be 19 percent if all of the radiographs in the latter group were reported as negative. Among the patients with respiratory findings, no individual sign or group of findings was a sensitive predictor of pneumonia. Signs of respiratory distress or lower respiratory tract signs had high specificity but low sensitivity. This finding reflects the large number of occult pneumonias detected in this population.

The authors conclude that obtaining a chest radiograph is appropriate in the evaluation of febrile children under five years of age who have a WBC count of over 25,000 per mm³ and no other treatable source of infection. Chest radiographs also should be strongly considered for patients with similar presentations whose counts are between 20,000 and 25,000 per mm³.