Temperature
Greater
Than
or Equal
to 40 C in Children
Less Than
24 Months
of Age: A Prospective
Study
Paul L. McCarthy, M.D., James F. Jekel, M.D., and Thomas F. Dolan, Jr., M.D.
From the Departments ofPediatrics and Epidemiology and Public Health, Yale University School of Medicine and Yale-New Haven Hospital, New Haven, Connecticut
ABSTRACT. In a prospective study, .330 consecutive
chil-dren less than 24 months old coming to the emergency room
of Yale-New Haven Hospital with a temperature 40 C
were evaluated. Nearly all patients had a white blood cell
(WBC) count, erythrocyte sedimentation rate (ESR)
(Win-trobe), blood culture, and chest roentgenogram. Eighty-eight
percent were evaluated 24 to 48 hours later. The mean WBC
count and ESR were significantly elevated in children with
positive blood cultures or pneumonia. The risk of bacteremia
was increased threefold and the risk of pneumonia was
increased twofold in children with a WBC count 15,000/
cii mm or an ESR 30 mm/hr compared to children
without leukocytosis or elevated ESR. Sixty-one percent of
children with bacteremia or pneumonia, 63% of children in
whom these diagnoses were not apparent on physical
exam-ination, and 86% of children with otitis media complicated
by pneumonia or bacteremia had either a WBC count 15,000/cu mm or an ESR 30 mm/hr. A WBC count 15,000/cu mm and an ESR 30 mm/hr were more
effective than a polymorphonuclear leukocyte count
10,000/cu mm and/or a band count 500/cu mm in
screening young children with high fever for bacteremia, pneunlonia, or complicated otitis media. Pediatrics,
59:663-668, 1977, BACTEREMIA, PNEUMONIA, FEVER, WHITE BLOOD
CELL COUNT, ERYTHROCYTE SEDIMENTATION RATE.
than 24 months of age may appear as seemingly trivial illnesses. Samson4 reported that children
less than 16 months of age with bacterial
menin-gitis may have none of the classic physical
findings associated with this disease.
Consequently, there has been a continuing effort to develop criteria which are indicators of
serious illnesses in febrile children. McCarthy and
Dolan5 reported the association between a temperature 41.1 C and bacterial meningitis. Age less than 24 months, temperature 40 C,
and leukocytosis (white blood cell {WBC] count 15,000/cu mm) were correlated with bacter-emia in several reports.’’ Todd’ found that a polymorphonuclear leukocyte (PM N) count
10,000/cu mm and/or neutrophil band count
500/cu mm were excellent indicators of
bacte-rial infections.
We studied consecutive children less than 24 months old who came to the Yale-New Haven Hospital pediatric emergency room with a temperature 40 C to evaluate the use of WBC count and erythrocyte sedimentation rate (ESR)
as indicators of bacteremia, pneumonia, or other
culture-proven bacterial illnesses.
Recent investigators have emphasized the diag-nostic challenge of the young child with fever. McCarthy et al.,’ McGowan et al.,2 and Teele et al.’ have noted that bacteremia in children less
(Received July 13; revision accepted for publication October
13, 1976.)
ADDRESS FOR REPRINTS: (P.L.M.) Department of
Pedi-atrics, Yale-New Haven Hospital, 789 Howard Avenue, New
INITIAL
HISTORY AND PHYSICAL
SUGGESTED STUDIES INITIAL
PNEUMONIA CBC, ESR, BC, CXR
FOLLOW UP
CLINIC OR ADMI SSION INITIAL IMPRESSION
RECOGN I ZA BLE
VIRAL INFECTION
IViral ezonthem -snonthem,
ICroup, Gostrosntsritis, Post L immunization tivir, Other
OTITIS MEDIA
RECOGNIZABLE BACTERIAL INFECTION
ICeIIuIitis, Epiglottitis,
I Meningitis, Sepsis, Other
AT DISCRETION OF
HOUSE OFFICER
CBC1, ESR1, BC, CXR4
CBC, ESR, BC
[OtMrs ot discretion
L of touss officer’
NON-SPECIFIC ILLNESS
1RhInoMo, Ned throot,
I
Non-sp.clfic rash, MildI
vomIting end /sr dlerrheo,I Irritobllity, Other
CBC, ESR, BC, CXN
C Soo test ]
Repeat #{163}zominotion,
Review of FINAL studIes: DIAGNOSIS LP, CXR,
Cultures
1Complets blood count
tErythrocyte ssdimentotion rote 3BIood culture
4Chist roentgsnogrom ‘See test
FIG. 1. Protocol used to evaluate febrile children.
MATERIALS AND METHODS
Froni Ju1 15, 1974, to July 1, 1975, children
less than 24 months old with a rectal temperature
40 C who were seen in the pediatric
emer-genc room at Yale-New Haven Hospital were
evaluated according to a fever protocol (Fig. 1).
Each child was seen by the pediatric resident
on call. After an initial history and physical
examination, and prior to obtaining laboratory
data, the house officer recorded an initial
ilnpres-sion. Prior to obtaining a chest roentgenogram or performing a lumbar puncture, the resident recorded findings suggestive of pneumonia or
meningitis.
If a child had a recognizable viral infection,
further laboratory studies were done only at the
discretion of the house officer, e.g., a chest
roentgenogram for a child with rubeola. If a
diagnosis of a recognizable bacterial infection was
made, studies which were appropriate for the
illness were Ol)tained in addition to those
suggested by the protocol, e.g., a wound culture
for a child with cellulitis. No mvringotomies or
l)t111ionarY taps were performed on children with
otitis niedia or pneumonia, respectively. Lumbar
puncture was suggested for children with
nonspe-cific signs and symptoms who were less than 6 months of age and/or had a WBC count
20,O()O/cu 111111. Otherwise, a lumbar puncture
was left to the discretion of the house officer. A
urine culture was done if a child had a nonspecific
illness without upper respiratory tract symptoms.
All children entered on the fever protocol with first febrile seizures had a lumbar puncture
performed.’ Children with a temperature 41.1
C and those less than 3 months old with a
temperature 40 C were admitted for
observa-tion and/or therapy after a WBC count, ESR
determination, lumbar puncture, chest
roentgen-ogram, and cultures of blood and urine were done. ‘#{149}
The resident listed an initial diagnosis after
WBC count, ESR, lumbar puncture, and chest
roentgenogram results were known. An attending pediatrician (P.L.M. or T.F.D.) was then consulted about the patient. Children were then admitted or sent home with a referral to a fever follow-up clinic staffed by the authors (P.L.M. and T.F.D.) and the house officer who saw the child initially. At the fever follow-up visit, which took place 24 to 48 hours after the initial exami-nation, a repeat examination and a review of bacteriology cultures were done.
Results of blood cultures were interpreted by the bacteriology laboratory and nonpathogenic contaminant organisms were excluded: Staph
ijlo-coccus epidermidis (two) and Serratia species
TABLE I
SELECTED DIAGNOSTIC METHODS AND RESULTS
RESULTS
#{176}Seetext.
with positive blood cultures were admitted for
evaluation (including lumbar puncture, chest
roentgenogram, and repeat blood culture) of soft
tissue complications, especially meningitis,
pneu-monia, septic arthritis, or cellulitis. Children with an initially positive urine culture
(
10,000
organisms per cubic millimeter) obtained by
clean-catch bag method had a repeat urine
culture done by suprapubic aspiration. The ESR
was measured by the Wintrobe method.
In the 12-month study period, 334 children less than 24 months old with a temperature 40 C were seen and 330 were studied according to the fever protocol. Table I shows the laboratory and
other diagnostic studies performed on children
with high fever. Ninety instances of bacteremia, pneumonia, or other culture-proven bacterial
illnesses occurred in 82 patients. Of 329 blood
cultures, 7.3% yielded a pathogen: Diplococcus
pnellnloniae (15), Heniophilus influenzae type b
(
5), Salmonella species (2), and Neisseriameningi-tidis (2). All children with septic meningitis had
positive blood cultures (H. influenzae {3}, N.
nieningitidis [1]). Two children with Salmonella
bacteremia yielded the same organism in the
stool. One child had shigellosis. One child with an
Lschericlua coli urinary tract infection had
pneu-mococcal bacteremia. Of 52 children with
pneu-monia, one had bacteremia (H. influenzae type
b).
Five patients in the study were less than 3
months old, and one had a positive blood culture.
Twenty-one of 24 positive blood cultures
occurred in the 266 children between 3 and 18
months of age. Seventeen children had a
temper-ature 41.1 C; of these, four had bacteremia and
two had bacterial meningitis. No other
associa-tions were noted between age, temperature, and
diagnosis.
Six of 24 children with initially positive blood
cultures had persistent bacteremia with D.
pneu-moniae (four), Salmonella species (one), and N.
;neningitidis (one). Only one child with persisting
bacteremia, in this instance caused by Salmonella
species, had been treated with antibiotics (oral
ampicillin for otitis media) prior to knowing the
results of the first blood culture. Of 18 children
without persisting bacteremia, 12 had received
either oral or intravenous antibiotics prior to knowing the results of the first blood culture because of the presence of meningitis, pneumo-nia, septic shock, or otitis media. The six children
Diagnostic method No. No. (%)
Positive
Complete 1)lOOd cell count 327 ..
Erythrocte sedimentation rate 273 ...
Blood culture 329 24#{176}( 7.3)
Chest roentgenogram 289 52#{176}( 18.0)
Lumbar puncture 137 4’ ( 2.9)
Urine culture 95 3* ( 3.2)
Rectal culture 52 3#{176}( 5.8)
Throat culture 31 1 ( 3.2)
Wound culture :3 3 (100.0)
with nonpersisting bacteremia who received no
antibiotics prior to knowing the results of the first blood culture all had pneumococcal bloodstream
invasion. No soft tissue complications, other than
those noted during the initial evaluation, were
identified in children with either persisting or
nonpersisting bacteremia. After bacterial
blood-stream invasion was documented, antibiotics were used as described previously.’
Two hundred ninety-one children had
follow-up evaluation: 57 were admitted to the hospital
from the emergency room, 151 had fever
follow-U visits, 19 were followed up by private
physi-cians, and 64 were followed up by phone or an
outreach worker.
The mean WBC count and ESR were
signifi-cantly elevated (P .05) in children with
positive blood cultures (16,086 ± 8,350, 29.3 ±
15.3) and positive chest roentgenograms
(16,012 ± 7,356, 29.8 ± 15.4) compared to those
with negative blood cultures (12,384 ± 5,669,
22.6 ± 12.9) and negative chest roentgenograms
(11,980 ± 5,440, 21.9 ± 12.3). In seven children
with nonbacteremic but proven bacterial
infec-tions, the mean WBC count and ESR were
11,200/cu mm and 21 mm/hr, respectively.
Table II demonstrates that children had a threefold greater risk of bacteremia and a twofold greater risk of pneumonia if their WBC count was
15,000/cu mm than if < 15,000/cu mm.
Simi-larly, children had a threefold greater risk of
bacteremia and a twofold greater risk of pneu-monia if the ESR was 30 mm/hr than if it was
< 30 mm/hr.
Otitis media, diagnosed in 148 patients, was often complicated by pneumonia (12 patients) or bacteremia (10 patients, one of whom had septic
TABLE II
PREnIcrIvE VALUE OF WBC COUNT AND ESR FOR BACTEREMIA, PNEUMONIA, OTHER CULTURE-PROVEN BACTERIAL ILLNESSES, COMPLICATED OTITIS MEDIA
Patients WBC Count ESR (mm/hr) WBC Count Neither
(per CII A
nun) ‘- -‘
.30 < 30
15,000/cu ;iiiii
and/or ESR
WBC Count
15,O()()/cu mm
,-15,000 < 15,000 30 mm/hr nor ESR
.30iiin/hr
% with positive blood culture 14.5 4.9 15.7 3.9 14.6 2.9
% withpositive chest x-ray film 25.3 12.7 27.0 14.4 25.2 12.9
% with other culture-proven 1.2 2.5 1.4 2.0 1.6 2.5
bacterial illnesses
% with positive blood culture or 38.5 17.6 40.0 16.3 37.4 14.2
positive chest x-ray film
% with otitis and bacteremia or 28.6 9.4 32.4 7. 1 28.8 3.7
pneumonia
children with complicated otitis media was
18,100/cu mm and 33.7 mm/hr, respectively, compared to a mean WBC count of 12,400/cu
mm and a mean ESR of 22.5 mm/hr in children
with uncomplicated otitis media. If otitis media
was present, complications were three times as frequent if the WBC count was 15,000/cu mm
and four times as frequent if the ESH was 30
mm/hr than if it was < 15,000/cu mm and 30
nim/hr, respectively (Table II).
Only 2.9% of children with neither a WBC
count 15,000/cu mm or an ESR 30 mm/hr
had bacteremia (Table II).
Neutrophil band counts 500/cu mm were seen in 91% of the patients who had a complete blood cell count. There was no significant differ-ence in mean band count between children with
and without bacteremia or pneumonia.
Table III shows that one half of the children
with bacteremia and 40% of those with pneu-monia had a WBC count 15,000/cu mm, which was similar to the sensitivity of the ESR
30 mm/hr for these diagnoses. Combining leukocytosis and/or elevated ESR identified three fourths of children with bacteremia, 60% of children with pneumonia, and nearly all children with complicated otitis media.
There were 46 children with pneumonia (27) or bacteremia (19) whose signs and symptoms on initial physical examination did not indicate the
diagnosis, that is, the initial impression was other
than pneumonia or bacteremia. Forty-six percent of these 46 children had a WBC count 15,000/
cu mm, 39% had an ESR 30 mm/hr, and 63%
had one or the other elevation.
A PMN count 10,000/cu mm identified 37%
of children with bacteremia, 25% with pneumo-nia, and 41% with complicated otitis media.
DISCUSSION
Our data demonstrate a clinically useful
asso-ciation between a WBC count 15,000/cu mm
and an ESR 30 mm/hr and pneumonia or bacteremia in children less than 24 months of age with a temperature 40 C. If these selected elevations of WBC count and ESR are docu-mented in a young child with a high fever, the work-up should include a blood culture and a chest roentgenogram. Our data in no way minimize the importance of evaluating this group of patients for other serious illnesses, such as meningitis, cellulitis, or urinary tract infections.
An ESR 30 mm/hr was equally predictive and sensitive for pneumonia or bacteremia as a
WBC count 15,000/cu mm. Combining an
ESR 30 mm/hr and/or a WBC count
15,000/cu mm, while not decreasing the
predictive value (Table II), increased the
sensi-tivity (Table III) of these selected elevations to
61% for children with pneumonia or bacteremia,
63% for children in whom these diseases were not
apparent on physical examination, and 86% for patients with complicated otitis media.
Todd6 showed that a segmented neutrophil
count 10,000/cu mm and/or a neutrophil band count 500/cu mm had 80% specificity and 75%
sensitivity for bacterial infections. In our study,
only one third of the children with bacteremia
TABLE III
SEN5ITIvrn OF WBC COUNT AND ESR FOR BACTEREMIA, PNEUMONIA, COMPLICATED OTITIS,
OTHER CULTURE-PROVEN BACTERIAL ILLNESSES
Patients Positive Blood Culture Positive Chest X-ray Film Other Bacterial Illnesses Positive Blood Culture and/or Chest X-ray Film Goniplicated Otitis Media
% with WBC count 15,000/ 50 40 14 43 55
cu mm
% with ESR 30 mm/hr 58 40 20 44 72
% with WBC count 15,000/
cu mm and/or ESR 30
mm/hr
75 58 29 61 86
-% with neither WBC count
15,000/cu mm nor ESR
30 mm/hr
25 42 71 39 14
PMN count 10,000/cu mm. A.neutrophil band
count 500/cu mm was extremely nonspecific
because nearly all patients had this number of
bands. We recommend a WBC count 15,000/
cu mm as a more effective means of screening young children with high fever for bacteremia or pneumonia.
Portnoy et a!.” studied 246 nonconsecutive, hospitalized children of all ages with pneumonia, bronchiolitis, or laryngotracheobronchitis. No fever data were given. The mean WBC count was
16,725/cu mm in 92 children with radiologic pulmonary infiltrates; in children with bronchio-litis or laryngotracheobronchitis it was 13,550/cu mm and 13,000/cu mm, respectively. Fifty-two percent of children with pneumonia had serologic and/or culture evidence of viral infection.
Our data also demonstrate an association between pneumonia and leukocytosis and delin-eate the predictive value and sensitivity of a WBC count 15,000/cu mm for pulmonary infiltrates in the group of young children with high fever that we evaluated. Only one child (with H. influenzae type b bacteremia) had an
etiology established for the pneumonia.
Clarifica-tion of the etiology of pneumonia in young children with high fever would have required more extensive evaluation, including pulmonary taps9” and viral studies.’#{176}’2 Such evaluation is certainly not indicated routinely because children with pulmonary infiltrates, fever (temperature
40 C), and/or a WBC count 15,000/cu mm
have a rapid symptomatic response to antibiotic therapy.
Because only seven children with
culture-proven bacterial illnesses (other than bacteremia)
were seen in this study, we hesitate to comment on the sensitivity of a WBC count 15,000/cu
mm and/or an ESR 30 mm/hr for this group of patients.
Antimicrobial therapy for pneumonia or other culture-proven bacterial illnesses was not evalu-ated critically in this study. Rather, the duration and route of treatment and the antibiotic selected for children with bacteremia or other culture-proven bacterial illnesses depended on the organism isolated and the extent, if any, of soft tissue involvement.1 Patients with leukocytosis and/or an elevated ESR but without pneumonia or identifiable bacterial illness at the initial visit were not treated with antibiotics but were followed up as inpatients or outpatients as described here.
CONCLUSIONS
1. In children less than 24 months of age with a temperature 40 C, the mean WBC count and
ESR were significantly elevated in those with positive blood cultures or chest roentgenograms compared with those with negative blood cultures or chest roentgenograms.
2. The risk of bacteremia was threefold and the
risk of pneumonia was twofold in children with a
WBC count 15,000/cu mm or an ESR 30
mm/hr compared with children without
leukocy-tosis or elevated ESR.
3. Combining a WBC count 15,000/cu mm
and/or an ESR 30 mm/hr, while not changing the predictive value, increased the sensitivity of
examina-tion, and 86% for children with complicated otitis media.
4. A WBC count 15,000/cu mm or an ESR 30 mm/hr was more sensitive for pneumonia
or bacteremia than a PMN count 10,000/cu
mm. A band count 500/cu mm was seen in
91% of children and, therefore, was nonspecific.
We recommend WBC counts and ESRs as a
useful means of screening young children with
high fever for bacteremia or pneumonia.
REFERENCES
I. McCarthy PL, Crundy CL, Spiesel SZ, Dolan TF:
Bacteremia in children: An outpatient review.
Pedi-atnes 57:861, 1976.
2 McGowan JE, Bratton L, Klein JO, Finland NI:
Bactere-mia in febrile children seen in a “walk in” pediatric
clinic. N EngI J Med 288:1309, 1973.
3. Teele DW, Pelton SI, Mvles JAG, et al: Bacteremia in febrile children under 2 years of age: Results of cultures of 1)100(1 of 6(X) consecutive febrile children
seen in a “walk in” clinic. J Pediatr 87:227, 1975. 4. Samson Jil: Febrile seizures and purulent nieningitis.
JAM.. 210:1918, 1969.
5. McCarthy PL, Dolan TF: Hperpyrexia in children:
Eight ‘ear emergencY room experience. Am J Dis
Child 130:849, 1976.
6. Todd JK: Childhood infections: Diagnostic value of peripheral white blood count and differential cell
counts. Am J Dis Child 127:810, 1974.
7. McCarthy PL, Dolan TF: Temperature 40#{176}C. in children less than three months. Clin Pediatr
15:700, 1976.
8. Portnoy B, Hanes B, Salvatore M, Eckert H: The
peripheral white blood count in respiro virus
infec-tion.J Pediatr 68:181, 1966.
9. Escobar JA, Dover AS, Duenas EL, et al: Etiology of respiratory tract infections in children in Cali, Colombia. Pediatrics 57:123, 1976.
10. Mimica I, Donoso E, Howard J,et al: Lung puncture in the etiologic diagnosis of pneumonia. Am J Dis
Child 122:278, 1971.
11. Glezen WP, Denny F’sV: Epidemiology of acute lower
respiratory disease in children. N EngI J Med 288:498, 1973.
12. McCarthy PL, Frank A, Masters S, et al: Prospective
study of children with pneunionia. Read before
the Ambulatory Pediatric Association, St. Louis, 1976.
1:3. Shuttleworth DB, Charney E: Leukocvte count in
childhood pneumonia. Am J Dis Child 122:393, 1971.
ACKNOWLEDGMENT
\Ve thank Dr. Howard Pearson for review of the