833

Chest Radiographs in the

Evaluation of the Febrile Infant

...,*.. . ,

Richard J. tt2 Chest radiographs are often considered an essential part of the workup of the febrile
George S. Bisset 1111 infant. Anteroposterior and lateral radiographs of the chest are frequently obtained in
Donald R. Kirks1 this group of patients, irrespective of respiratory tract symptoms and/or signs. A total
Arlyce Vanness3 of 226 chIldren (2 years old) with and without signs and symptoms of lower respiratory
tract infections were examined to assess the yield of chest radiographs. The radiograph
was considered positive only if a focal parenchymal infiltrate was present. Hyperinflation
or bronchial thickening was not included these children
as a positive finding because
usually do not receive antibiotics despite the fact that viral illness or reactive airway
disease may be present. In a retrospective study of 105 infants, confidence intervals for
yield were established for children with (95% Cl = 12%, 32%) and without (95% Cl =
0%, 14%) symptoms or signs of lower respiratory tract infection. In a prospective study
of 121 Infants without chest symptoms or signs, confidence levels for positive yield
were better defined (95% Cl = 0%, 3%).
The data suggest that obtaining chest radlographs to look for parenchymal infiltrates
treatable with antibiotics in infants less than 2 years old is necessary only in those
infants who have clinical evidence of lower respiratory tract illness.

AJR 155:833-835, October 1990

The febrile infant poses a diagnostic dilemma for the pediatrician. Frequently,
despite a careful history and physical examination, the source of fever cannot be
found. The chest radiograph is often used to screen such infants for lower
respiratory tract infection (LRTI).
Previous studies have addressed the problem of clinically unsuspected LRTI
discovered with chest radiographs, but results have varied widely [1 -1 2]. In a
series of 330 consecutive children with high fever (temperature >40#{176}C)
younger
than 24 months, McCarthy et al. [1] reported 52 cases of pneumonia, 27 of which
Received January 8, 1990; accepted after revi-
were not clinically suspected. Other authors also have reported radiologic evidence
of pneumonia in young children without clinical signs of pneumonia [4, 7]. However,
sion May 22, 1990.
Heulitt et al. [2] have concluded that in febrile infants younger than 3 months, “a
Recipient of bronze award at the 31st annual
meeting of the Society for Pediatric Radiology, Ce- chest radiograph should be obtained in this group only when signs of respiratory
ronado, CA, April 1988. distress are present.” Lorin [5] had advocated a similar approach for obtaining
1 Department of Radiology, Children’s Hospital chest radiographs in the febrile infant older than 3 months. We undertook this
Medical Center and the Departments of Radiology study to determine the usefulness of routine chest radiographs in febnle infants
and Pediatrics, University of Cincinnati College of
Medicine, Cincinnati, OH 45229-2889. younger than 24 months without an obvious cause of fever.
2 Present address: Department of Radiology,
Minneapolis Children’s Medical Center, 2525 Chi-
cago Ave., S., Minneapolis, MN 55404. Address
reprint requests to A. J. Patterson.
Subjects and Methods
3 Department of Pediatrics, University of Cincin- The total study population was composed of 226 children up to 24 months old who were
nati, College of Medicine, Cincinnati, OH 45267. evaluated for fever in the emergency department at Children’s Hospital Medical Center,
0361-803X/90/1 554-0833 Cincinnati. Fever was defined as a rectal temperature greater than 1 00#{176}Fat presentation.
0 American Aoentgen Aay Society Anteroposterior and lateral chest radiographs were obtained on all patients.
834 PATTERSON ET AL. AJR:155, October 1990

Of the 226 children, 105 (63 boys and 42 girls) were evaluated in TABLE 1: Yield of Chest Radiographs in Febrile Infants
a retrospective study. Children both with and without symptoms or
signs of LRTI were included. The children ranged in age from 1 week Symptoms or Signs Radiograph
of LRTI Total
to 22 months, with a mean age of 8.4 months. Criteria for inclusion Positive Negative
were the following: Radiograph folders were randomly selected from
Present 14 54 68
the current patient file in the radiology department. Patients were
Absent 1 36 37
entered in the study if chest radiographs were obtained as part of an
emergency department evaluation for fever between January 1984 Total 15 90 105
and June 1987 and the patients were younger than 24 months old at
Note. LATI = lower respiratory tract infection. Sensitivity = .93, specificity
the time. After review of the clinical record, patients were subse-
= .40, x2= 4.88,p< .05.
quently excluded if they were being treated for a known acute illness
or if they had a preexisting condition such as congenital heart disease
or lung disease of prematurity that might place them at greater than
by wheezing (n 4) and dyspnea (n = 3). The most frequent
=
normal risk for LRTI.
signs of LRTI were rales/rhonchi (n = 20), wheezing (n 1 1), =
The remaining 1 21 children (56 boys and 65 girls) were evaluated
in a prospective study between July 1 987 and January 1988. Febrile and retractions (n = 10). Of the 68 patients with symptoms
children without signs or symptoms of LRTI were studied in an or signs of LRTI, 14 (21 %) had evidence of air-space disease
attempt to better define confidence intervals for the yield of chest on radiographs (95% Cl = 12%, 32%).
radiographs in this clinical setting. Patients’ ages ranged from 1 week Of the remaining 37 patients without symptoms or signs of
to 23 months, with a mean age of 5.8 months. Criteria for inclusion LRTI, one (3%) had evidence on radiographs of air-space
in this group were as follows: Consecutive patients less than 24 disease (95% Cl = 0%, 14%). A 21 -month-old boy who had
months old who had chest radiographs as part of an emergency adventitious sounds on chest auscultation and focal air-space
department evaluation for fever were identified. After review of the
disease on chest radiographs was felt to have “transmitted
clinical record, patients were excluded if not previously healthy or if
upper airway noise” rather than rales or rhonchi on the
any symptoms or signs of LRTI were found and recorded before
pediatrician’s recorded physical examination. As such, he was
radiographs were obtained.
For all children, vital signs and the results of a history and physical assigned to the group with no symptoms or signs of LRTI.
examination were recorded. All chest radiographs were reviewed by
three pediatric radiologists. In order to prevent underreading, the
Prospective Study
chest radiographs of 21 age-matched “control subjects” with clinical
evidence of LRTI were randomly included during review of radio- By design, all patients in this study were free of symptoms
graphs from the prospectively evaluated group of children without
and signs of LRTI. None of the 121 patients (0%) had radio-
symptoms or signs of LRTI. Radiographs of three of these children
graphs showing air-space disease (95% Cl = 0%, 3%).
(1 4%) showed air-space disease. The radiologists knew of the pres-
ence, but not the size, of the control group and were unaware of the
patients’ history and physical findings. Presence of focal parenchymal Discussion
opacification was considered radiologic evidence of LRTI. Peribron-
chial thickening and/or air-trapping in the absence of focal opacities In infants, the presence of fever without localizing signs
were not considered evidence of LRTI for the purposes of this study. has always been a significant concern. The clinician’s fear of
Symptoms of LRTI were considered to be present if coughing, missing an occult infectious process often prompts an expe-
wheezing, or dyspnea were noted at the time of emergency depart- ditious and sometimes costly evaluation. Although certain
ment evaluation. Nasal congestion and rhinorrhea were attributed to
baseline data may be necessary, one must be able to assess
upper respiratory tract infection for the purposes of this study.
the cost/benefit and risk/benefit ratios of each test. In the
Signs of LRTI were considered to be present if one or more of the
following were recorded on physical examination: rales, rhonchi, case of radiologic evaluation, the cost and risk of radiation
wheezing, retractions, grunting, nasal flaring, or focally decreased exposure (albeit small) must be weighed against the diagnos-
breath sounds. tic information provided.
Because of the wide range of normal respiratory rates and the In our retrospective study, we evaluated the positive yield
increase in respiratory rate with fever, tachypnea was not considered from chest radiographs in two groups offebrile children: those
to be either a symptom or sign of LRTI. Wheezing was considered a with clinical signs and symptoms of LRTI and those with no
symptom if reported on the history and a sign if recorded on physical respiratory tract symptoms. The prevalence of positive find-
examination. ings on chest radiographs in the group of children with signs
Exact confidence intervals from the binomial distribution were
and/or symptoms of LRTI was 21 %. This value is obviously
calculated for both retrospective and prospective study groups.
affected by the clinical criteria applied by the examining
pediatrician in deciding whether or not to obtain a chest
radiograph. It is common practice among some pediatricians
Results
at our institution to treat a child for LRTI on the basis of the
Retrospective Study presenting symptoms and physical findings alone. If chest
radiographs were obtained of all of these children, one would
Two groups were categorized according to whether or not assume that some would be positive and some negative.
symptoms or signs of LRTI were present (Table 1). Sixty- Of greater interest to us was the yield of chest radiographs
eight patients had symptoms and/or signs of LRTI. The most in the subset of children without symptoms or signs of LRTI.
frequently reported symptom was coughing (n 55), followed
= In the retrospective study, the confidence interval for positive
AJR:155, October1990 CHEST RADIOGRAPHS OF FEBRILE INFANTS 835

yield was 0-i 4%. By evaluating a larger group of similar prospective study of children without symptoms or signs of
children in our prospective study, we were able to narrow the LRTI, we have established a 95% confidence interval of 0-
confidence interval for positive yield to 0-3%. Although the 3% for positive yield of chest radiographs in this group. The
same age and fever criteria were applied to both retrospective data suggest that chest radiographs are necessary only in
and prospective study groups, the mean age of the infants in those infants younger than 2 years old who have clinical
the retrospective study was higher (8.4 months vs 5.8 evidence of lower respiratory tract illness.
months). This may represent an evolution in criteria used by
pediatric house officers for obtaining a chest radiograph in
infants without an obvious source of fever. As patients in the ACKNOWLEDGMENTS
retrospective study were evaluated from 1984 to i 987 and
We thank Frank Schleuter for technical assistance, Bruce Lindgren
patients in the prospective study from June 1987 to January
for statistical analysis, and Cathi Churchillfor manuscript preparation.
1 988, the age discrepancy between the groups could be
explained if progressively more restrictive criteria were used
on children over the age of, for example, 3 months. As a REFERENCES
result of this selection bias, relatively fewer older infants would
be induded in the more recent group (prospective study 1. McCarthy PL, Jekel JF, Dolan TF. Temperature greater than or equal to
group), thus lowering the mean age. An informal survey 40#{176}C
in children less than 24 months of age: a prospective study. Pediat-
rics 1977;59:663-668
involving pediatric house officers lends support to this trend. 2. Heulitt MJ, Ablow AC, Santos CC, O’Shea TM, Hilfer CL. Febrile infants
Similarly to the study by Heulitt et al. [2], we chose focal less than 3 months old: value of chest radiography. Radiology 1988;167:
parenchymal opacification as the criterion for a “positive” 135-1 37
chest radiograph. This finding was chosen because focal 3. Alario AJ, McCarthy PL Markowitz A, Komguth P. Aosenfield N, Leventhal
JM. Usefulness of chest radiographs in children with acute lower respira-
opacification prompts the pediatricians at our institution to
tory tract disease. J Pediatr 1987;1 1 1 : 187-1 93
start antibiotic therapy in the appropriate clinical setting. 4. McCarthy PL, Jekel JF, Stashiwick CA, Spiesel SZ, Dolan TF. History and
Hypennflation and/or peribronchial thickening is usually re- observation variable in assessing febrile children. Pediatrics 1980;65:
garded by our dinicians as evidence for viral or reactive 1090-1095
airways disease, which may not require antibiotic therapy. 5. Lorin Ml. Fever without localizing signs in children. Postgrad Med 1983;
73:295-300
Heulitt et al. used only physical signs in categorizing pa- 6. Pantell RH, Naber M, Lamar A, Dias JK. Fever in the first six months of
tients. By comparison, we chose to categorize patients in our life. Clin Pediatr 1980;19:77-82
retrospective study according to presence or absence of 7. Tomlinson WA. High fever. Am J Dis Child 1975:129:693-696
symptoms or signs of LRTI. In our study, symptoms of LRTI, 8. Roberts KB, Borzy MS. Fever in the first eight weeks oflife. Johns Hopkins
MedJ 1977:141:9-13
rather than signs, were a more sensitive predictor of air-space
9. McGowan JE, Bratton L, Klein JO, Finland M. Bacteremia in febrile children
disease appearing on radiographs (symptom sensitivity = .87, seen in a “walk-in’ pediatric clinic. N EngIJ Med 1973:288:1309-1312
sign sensitivity = .47). The combination of symptoms and 10. Dechovitz AB, Moffet HL. Classification of acute febrile illnesses in child-
signs of LRTI was fairly specificfor radiologic airspace disease hood. Clin Pediatr 1968:7:649-653
(specificity = .83) but not very sensitive (sensitivity = .40). 1 1 . McGravey A, Wise PH. Evaluation of the febrile child under 2 years of age.
J Emerg Med 1984:1:299-305
The central issue is whether omission of the chest radio- 12. Teele DW, Pelton SI, Grant MJA, et aJ. Bacteremia in febrile children under
graph in the febnle infant (without clinical evidence of LRTI) 2 years of age: results of cultures of blood of 600 consecutive febrile
will subject the child to substantial risk. As a result of our children seen in a walk-in” clinic. J Pediatr 1975:87:227-230