The Futility of the Chest Radiograph in the Febrile Infant Without Respiratory Symptoms

(1)

524 PEDIATRICS Vol. 92 No. 4 October 1993

The

Futility

of

the

Chest

Radiograph

in

the

Febrile

Infant

Without

Respiratory

Symptoms

Robert T. Bramson, MD* ; Theo L. Meyer, MD; Martin L. Silbiger, MD*;

Johan G. Blickman, MDX; and Elkan Halpern, PhD9J

ABSTRACT. Objective. Major pediatric textbooks

ad-vocate a chest radiograph as part of the diagnostic evalu-ation for a sepsis workup for febrile infants less than 3 months old. Very few studies have addressed the value of

performing a chest radiograph in this situation. Two

studies previously published lack the numbers to

statis-tically justify a conclusion about the need to perform a chest radiograph in the febrile infant.

Methods. Evaluated were 197 febrile infants 3 months old or less with a history, physical examination, chest radiograph, and other laboratory studies to determine the cause of their fever. This group of infants was combined with the group of infants from two similar studies pub-lished previously in the literature using cumulative

meta-analysis. The combined group resulted in 617

in-fants.

Results. The combined group of infants had 361 in-fants who had no clinical evidence of pulmonary disease on history or physical examination. All 361 infants had normal chest radiograph. These results gave a 95% con-fidence interval that the chance of a positive chest radio-graph in a patient with no pulmonary symptoms would occur less than 1.02% of the time.

Conclusions. The generally advocated policy of ob-taming a chest radiograph as part of the sepsis workup

in febrile infants should be discontinued, and chest

radiographs should be obtained only in febrile infants

who have clinical indications of pulmonary disease.

Pediatrics 1993;92:524-526; chest radiographs, sepsis

workup, infants, fever.

Pediatricians routinely order chest radiographs as

a part of a sepsis workup for febrile infants less than

3 months old, a policy advocated by numerous

au-thors in several major textbooks on pediatrics.

Al-though several studies have correlated chest

radio-graph findings with the presence of clinical signs of

disease, only a few studies have evaluated the utility

of the chest radiograph in the febrile infant.71#{176} This

current study evaluated the efficacy of chest

radio-graphs in febrile infants, and then combined the data

from this study with two very similar studies in the

literature.9’10 The other two studies by themselves

did not have enough patients to statistically justify a

From the ‘Department of Radiology, University of South Florida College of Medicine, Tampa Genera! Hospital, Tampa, FL; the IDivision of Pediatric Imaging, Massachusetts Genera! Hospital, Boston, MA; the §Department of

Pediatrics, University of South Florida College of Medicine, Tampa General Hospital. Tampa, FL; and the 9jCenter for Imaging and Pharmaceutical Research, Massachusetts General Hospital, Boston MA.

Received for publication Mar 8, 1993; accepted Apr 15, 1993.

Acad-emy of Pediatrics.

definite statement concerning the value of a chest

radiograph as part of a sepsis workup in febrile

infants.

The combined data of these three studies was

sub-jected to statistical meta-analysis using methods

de-scribed in the recent medical literature.11’12 The

results achieved by the increased numbers of the

combined study allow a more valid conclusion

con-cerning the generally accepted practice of

perform-ing chest radiographs in febrile infants as part of a

sepsis workup.

METHODS

During a 1-year period every infant, 3 months old or less, who

was admitted to the pediatric emergency room of a university

teaching hospital with a measured rectal temperature of 100.5#{176}For greater was included in the study. One calendar year was used for the study to exclude any unusual variations that might simply be

seasonal in nature.

All patients were seen by pediatric house staff and their

attend-ing pediatricians. A history, physical examination, chest

radio-graph, and laboratory tests deemed clinically appropriate were

performed. Two hundred thirty-eight patients who fit the study criteria were seen. At the start of the study, all examining

physi-cians were asked to carefully record clinical data on each infant,

paying particular attention to the presence of the respiratory symptoms listed in Table 1.

Two radiologists, each experienced in the interpretation of pe-diatric chest radiographs, read the chest radiographs

retrospec-tively without any clinical information about the infants except the

knowledge that they were a part of the study group. The chest

radiographs were classified in four categories, two types of normal

and two types of abnormal. The radiographic criteria for each category are shown in Table 2.

Classification of the chest radiographs into the four arbitrary groups chosen should be explained. Most pediatric radiologists agree that inflammatory lung disease in infants causes peribron-chial inflammation and, therefore, leads to air trapping.

Differen-tiation between normal chest radiographs and those with only air

trapping and peribronchial thickening may be subjective, but the method used has precedent.3”3’4 Differences in the radiologists’ interpretations were handled in the following manner. All films on

which the two radiologists did not agree initially were re-read,

without clinical information by both radiologists. If disagreement on classification persisted, then the films were interpreted a third time and discussed jointly at a conference between the two

radi-ologists, still without clinical information. If no agreement resulted, then both scores were recorded. For the purpose of calculations in this paper, the more abnormal chest radiograph classification was used if there was an unresolved difference in the

interpretation between the two radiologists. The chest radio-graph classification was then matched with each infant’s clinical information.

Of the initial 238 infants, 41 were eliminated from the study because their chest radiographs revealed other pathology such as bronchopulmonary dysplasia or congenital heart disease. The re-maining 197 radiographic studies were grouped as described, and

they were evaluated for the presence or absence of each clinical

symptom.

at Viet Nam:AAP Sponsored on September 1, 2020

www.aappublications.org/news

(2)

ARTICLES 525 TABLE 1. Clinical Findings

Evaluated in This Study

of Potential Pulmonary Disease

Rales Coryza

Rhonchi Grunting

Retractions Stnidor

Wheezes Nasal flaring

Tachypnea (>50/mm) Cough

TABLE 2. Classification Criteria for Types of Chest

Radiographs

Normal chest radiographs

Type 0, normal, with none of the findings seen in types I through 3.

Type 1, hyperinflation, determined by the presence of ei-ther flat diaphragms on the lateral projection or by the highest level of the diaphragm below the sixth anterior rib on the frontal projection. These chest radiographs were considered normal for this study. Documented hy-perinflation without peribronchial thickening was felt to be normal.

Abnormal chest radiographs

Type 2, hyperinflation plus penibronchial thickening, which was defined as the walls of the bronchi being thicker than their usual “paper-thin” appearance.

Type 3, hyperinflation, peribronchial thickening, and pul-monary infiltrates.

The data from this study were then combined with similar data from the studies of Heulitt et a19 and Cram et al.1#{176}This enlarged, combined data pool made it possible to make statistically valid conclusions attempting to establish the role of the chest radio-graph in the evaluation of the febrile infant.

RESULTS

In this study 40 infants (20%) had a normal

respi-ratory tract history and physical examination (Table

3). Thirty-six of these patients had normal, type 0,

chest radiographs. Four children with no clinical

res-piratory tract findings had type I chest radiographs,

although one radiologist classified one of these as a

normal (type 0) chest. The significance of

hyperinfla-tion on these four radiographs is questionable, and it

was considered safe to identify these as normal

chests. Initially the radiologists agreed on the

classi-fication of all 197 chest radiographs except eight. The

differences in these eight films were as follows: one

infant was given a type 0 by one radiologist, and a

type I by the other (0/ 1), two infants were given a

type I by one radiologist, and a type 2 by the other

(I /2), five infants were given a type 2 by one

radi-ologist, and a type 3 by the other (2/3). After the

re-interpretation conferences, the unresolved

differ-ences between the two radiologists occurred in only

four cases, an interobserver agreement of 97.97%.

TABLE 3. Clinical Findings for Types of Chest Radio-graphs

40 infants who had no clinical findings 36 infants had a normal chest radiograph

4 infants had a normal type I chest radiograph 0 infants had an abnormal type 2 chest radiograph 0 infants had an abnormal type 3 chest radiograph 157 infants who had at least one positive clinical sign

60 infants had a normal chest radiograph 44 infants had a normal type I chest radiograph

28 infants had an abnormal type 2 chest radiograph 25 infants had an abnormal type 3 chest radiograph

Based on analysis of the 40 patients with no clinical

signs of pulmonary disease, the probability of a

nor-mal chest roentgenogram was 91 .2% or greater (95%

confidence interval).

One hundred fifty-seven children had one or more

of the clinical findings listed in Table I. Fifty-three of

these patients had an abnormal chest radiograph. One hundred four children had at least one clinical feature of respiratory disease, but had a normal chest

radiograph. The probability of an abnormal chest

radiograph in our series increased to 60% (35/58) for

those infants with any one of the clinical findings of lower respiratory tract infection such as rales,

rhon-chi, wheezes, and retractions. Table 4 shows a

corre-lation between the chest radiographic classification

of infants and selected clinical findings in this study.

Combining the 40 infants with normal respiratory

tract history and physical examinations in this study

with the 148 normal patients in the study by Cram et al1#{176}and the 173 patients with no findings of

pulmo-nary disease in the study by Heulitt et a!9 results in

361 infants with no clinical evidence of pulmonary

disease, all whose chest radiographs were normal.

The 95% confidence interval based on all 361 infants

implies that the probability of a normal chest

roent-genogram in an infant with no clinical evidence of

pulmonary disease is 98.98% or greater.

Two hundred fifty-six infants had some clinical

findings of pulmonary disease. Of these children, one

third (85) had positive chest radiographs.

DISCUSSION

Many pediatricians advocate chest radiographs to

assist in the diagnosis of the cause of unexplained

fever in infants. Studies in older age populations

suggest that chest roentgenograms in the absence of

pulmonary symptoms have no diagnostic value.7’8

Although similar studies in older pediatric

popula-tions suggest that asymptomatic children, like

adults, may not need a chest radiograph, a review

of the literature reveals only two studies that predict

the need for a chest roentgenogram in the febrile

infant.9’10 Both of these studies and the current study

arrive at the same conclusion. Independently each

study lacks the number of patients necessary to make

a definite statement that might alter the current

prac-tice of routinely obtaining a chest radiograph to ex-plain the cause of fever in a young infant. Together the three studies make a strong statement that

statis-tically justifies the recommendation that a chest

ra-diograph does not add useful information to the

TABLE 4. Correlation of Some Selected Clinical Findings With Chest Radiograph Classification

58 infants had either rales, retractions, or wheezes 7 infants had a normal type 0 chest radiograph 16 infants had a normal type I chest radiograph

14 infants had an abnormal type 2 chest radiograph 21 infants had an abnormal type 3 chest radiograph 13 infants had only coryza

7 infants had a normal type 0 chest radiograph 4 infants had a normal type I chest radiograph

2 infants had a type 2 chest radiograph 0 infants had a type 3 chest radiograph

(3)

Bramson Heulitt9 Crain’#{176}

I. Kline, M, Lorin, M. Fever without localizing signs. In: Oski F, DeAngelis

C, Feigin R, Warshaw JK, eds. Principles Practice of Pediatrics. Philadel-phia: Lippincott; 1990: 1023

2. Siegel J. Bacterial and viral infections of the newborn. In: Oski F, DeAn-gelis C, Feigin R, Warshaw JK, eds. Principles Practice of Pediatrics. Phila-delphia: Lippincott; 1990: 474

3. Boyer K. Nonbacterial pneumonia. In: Oski F, DeAngelis C, Feigin R, Warshaw, J, mis. Principles Practice of Pediatrics. Philadelphia: Lippincott;

1990: 1336

4. Jaffe D, Davis A. Fever. In: Schwartz M, Charney E, Curry 1, Ludwig 5, eds. Pediatric Primary Care: A Problem-Oriented Approach. 2nd ed. Chi-cago: Year Book Medical Publishers; 1990: 435

5. Sprinkle R. Fever infirst month oflife. In: Rakel R, ed. Textbook of Family Practice. 4th ed. Philadelphia: W. B. Saunders; 1990: 659

6. Overall J. Pneumonia. In: Behrman R, Vaughan V, Nelson W, eds.

Nelson Textbook of Pediatrics. 13th ed. Philadelphia: W. B. Saunders; 1987:

427-428

7. Saga! 5, Evens R, Forrest J, Bramson R. Efficacy of routine screening lateral chest radiographs in a hospital based population. N Engi IMed.

1974;291 :1001-1004

---

___

-- 8. Hubbell F, Greenfield 5, Tylerj, Chetty K, Wyle F. The impact of routine 192 242 admission chest x-ray films on patient care. N Engl I Med. 1985;312: <12 <8 209-213

100.5 100.4 9. Heulitt M, Ablow R, Santos C, O’Shea M, Hilfer C. Febrile infants

less than 3 months old. Value of chest radiography. Radiology. 1988;167:

135-137

60 60 10. Cram EF, Bulas D, Bijur PE, Goldman HS. Is a chest radiograph

neces-+ + sary in the evaluation of every febrile infant less than 8 weeks of age? + Pediatrics. 1991;88:821-824

+ + 11. Kassirer J. Clinical trials meta-analysis. What do they do for us? N EngI

+ I Med. 1992327:275-274

+ 12. Lau J,Antman EM, Jimenez-Silva J, Kupelnick B, Mosteller F, Chalmers + T. Cumulative meta-analysis of therapeutic trials for myocardial infarc-+ hon. N Engi IMed. 1992;327:248-254

+ 13. Griscom NT. Pneumonia in children some of its variants. Radiology. 1988;167:297-302

+ + 14. Griscom N, Wohl M, Kirkpatrick J.Lower respiratory infections: how _________________________ infants differ from adults. Radiol Clin North Am. 1978;16:367-387 50 + + + + + + + + +

526 CHEST RADIOGRAPH AND THE FEBRILE INFANT

evaluation of a febrile infant who does not have

clini-cal findings of pulmonary disease.

Because many different pediatric examiners saw

the children in this study in the emergency room,

there were differences in the interpretation of the

clinical terms used, such as rales and rhonchi,

differ-ences in the ability of the examiners to recognize

certain findings depending on their individual skills,

and differences of the examiners in recording all their

findings. This is unavoidable in this type of study.

Nearly one third of the 256 infants with pulmonary

symptoms had a positive chest radiograph.

There-fore, in the symptomatic febrile infant a chest

radio-graph may help identify significant pulmonary

dis-ease and must be performed.

In recent years, the use of cumulative

meta-analy-sis has become an increasingly accepted method of

combining data from several similar studies to draw

conclusions that any one study by itself might not be

able to make with statistical confidence.11’2 It is

al-most impossible to find clinical studies that have

been performed in identical ways. Therefore, it

be-comes extremely important to evaluate the studies to

see that the data truly can be combined into a

meta-analysis study. If the studies can be integrated into a

single meta-analysis, then valid statistics may result

from the combined studies. These three studies have

enough similar features that they can, in our opinion,

be combined into a single data group. These

similari-ties are summarized in Table 5. All three groups have

febrile infants of about the same age. The presence of

a similar group of respiratory symptoms is noted in

all three groups, and the chest radiograph is

evalu-ated in all three. The method of radiographic

inter-pretation of the chest radiograph varies slightly

among the three studies, but all arrive at a

classifi-cation of an abnormal chest radiograph versus a

nor-mal radiograph. The fact that each study arrives at

the same conclusion concerning the efficacy of the

chest radiograph in the child with no symptoms of

TABLE 5. Comparison of the Three Studies Subjected to Meta-analysis

(This Study)

Patients, n 197

Age, wk <12

Temperature, rectal, #{176}F 100.5

Symptoms recorded Respiratory rate Rales Rhonchi Retractions Grunting Cough Rhinonrhea Nasal flaring Wheezes Chest radiograph Abnormal/normal

lung disease, and that each study was performed in

a different geographic location, lends credibility to

the conclusions from the combined data study.

One of these three studies by itself has a 95%

con-fidence interval that the chance a positive chest

ra-diograph will occur in a patient with no pulmonary

symptoms happens less than 8% of the time. By

corn-bining the three studies, the 95% confidence interval

that the chance of a positive chest radiograph will

occur in a patient with no pulmonary symptoms drops to I .02% or less. To achieve a greater degree of confidence would require significant expansion of the numbers of studies. For example, it would

re-quire 1000 examinations that were negative to

achieve a 95% confidence interval of 0.37% or less,

and it would require 18 443 negative examinations in

children with no clinical signs of pulmonary disease to achieve a level of 0.02% or less. These three studies

combined give a value for the 95% confidence

inter-val that approaches the area of diminishing returns.

Rather than continuing the studies, it seems more appropriate to change the practice of routinely

order-ing a chest radiograph in febrile infants 3 months old

or less. Chest radiographs should be ordered only in

febrile infants of this age who manifest some clinical

signs of pulmonary disease.

REFERENCES

(4)

1993;92;524

Pediatrics

Halpern

Robert T. Bramson, Theo L. Meyer, Martin L. Silbiger, Johan G. Blickman and Elkan

Symptoms