Articles

Diagnostic value of clinical features at presentation to

identify serious infection in children in developed countries:
a systematic review
Ann Van den Bruel, Tanya Haj-Hassan, Matthew Thompson, Frank Buntinx, David Mant, for the European Research Network on Recognising
Serious Infection investigators*

Summary
Background Our aim was to identify which clinical features have value in conrming or excluding the possibility of
serious infection in children presenting to ambulatory care settings in developed countries.
Methods In this systematic review, we searched electronic databases (Medline, Embase, DARE, CINAHL), reference
lists of relevant studies, and contacted experts to identify articles assessing clinical features of serious infection in
children. 1939 potentially relevant studies were identied. Studies were selected on the basis of six criteria: design
(studies of diagnostic accuracy or prediction rules), participants (otherwise healthy children aged 1 month to 18 years),
setting (ambulatory care), outcome (serious infection), features assessed (assessable in ambulatory care setting), and
sucient data reported. Quality assessment was based on the Quality Assessment of Diagnostic Accuracy Studies
criteria. We calculated likelihood ratios for the presence (positive likelihood ratio) or absence (negative likelihood
ratio) of each clinical feature and pre-test and post-test probabilities of the outcome. Clinical features with a positive
likelihood ratio of more than 50 were deemed red ags (ie, warning signs for serious infection); features with a
negative likelihood ratio of less than 02 were deemed rule-out signs.
Findings 30 studies were included in the analysis. Cyanosis (positive likelihood ratio range 2665220), rapid
breathing (126978), poor peripheral perfusion (2393880), and petechial rash (6188370) were identied as
red ags in several studies. Parental concern (positive likelihood ratio 1440, 95% CI 9302210) and clinician
instinct (positive likelihood ratio 2350, 95 % CI 16803270) were identied as strong red ags in one primary care
study. Temperature of 40C or more has value as a red ag in settings with a low prevalence of serious infection. No
single clinical feature has rule-out value but some combinations can be used to exclude the possibility of serious
infectionfor example, pneumonia is very unlikely (negative likelihood ratio 007, 95% CI 001046) if the child is
not short of breath and there is no parental concern. The Yale Observation Scale had little value in conrming (positive
likelihood ratio range 110670) or excluding (negative likelihood ratio range 016097) the possibility of serious
infection.

Published Online
February 3, 2010
DOI:10.1016/S01406736(09)62000-6
See Online/Comment
DOI:10.1016/S01406736(09)62166-8
*Members listed at end of paper
Department of General
Practice, Katholieke
Universiteit Leuven, Leuven,
Belgium (A Van den Bruel MD,
Prof F Buntinx MD); and
Department of Primary Health
Care, University of Oxford,
Oxford, UK (T Haj-Hassan MSc,
M Thompson MRCGP,
Prof D Mant FRCGP)
Correspondence to:
Dr Ann Van den Bruel,
Department of General Practice,
Katholieke Universiteit Leuven,
Kapucijnenvoer 33 Blok J,
3000 Leuven, Belgium
ann.vandenbruel@med.
kuleuven.be

Interpretation The red ags for serious infection that we identied should be used routinely, but serious illness will
still be missed without eective use of precautionary measures. We now need to identify the level of risk at which
clinical action should be taken.
Funding Health Technology Assessment and National Institute for Health Research National School for Primary Care
Research.

Introduction
Serious infection is an important cause of morbidity and
mortality in children in developed countries. Infections
account for 20% of childhood deaths in England, Wales,
and Northern Ireland, with the greatest number in
children aged 14 years.1 These serious illnesses need to
be distinguished from self-limiting acute illnesses that
are very common in children. A Dutch survey of parents
reported that during a 3-week period, 60% of children
had an acute illness episode and 4% had febrile illness.2
In the UK, acute infections result in 40 consultations
per person-year in children aged less than 1 year, and
13 consultations per person-year in children aged
115 years.3 Additionally, febrile illness accounts for 20%
of all visits to the paediatric emergency department.4

An early and accurate diagnosis of serious infection in

children is essential to reduce morbidity and mortality.
However, diagnosis is not straightforward because of
the low prevalence of serious illness, and even those few
children with serious illness can present at an early
stage when the severity of the illness is not apparent. In
a primary care setting, less than 1% of children assessed
will have a serious illness5 and there is a duty on the
clinician to reassure anxious parents of healthy children
and to diagnose seriously ill children.6 Triage might
need to be done rapidly in a pressured environment or
by telephone, and by sta who might have limited
paediatric experience. Consequently, the diagnosis could
be missed at rst contact,7 sometimes with serious
consequences.8

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Panel: Criteria for selection of studies

Design
Studies that assessed diagnostic accuracy or derived
prediction rules were selected. Narrative reviews, letters,
editorials, comments, and case series of less than 20 patients
were excluded. Systematic reviews and meta-analyses were
used only as a source of references.
Participants
Studies needed to include children aged between 1 month
and 18 years. Studies that included children above or below
this age range were selected if they reported age-stratied
analyses (so that children aged <1 month or >18 years could
be excluded) or if the proportion of children out of range was
less than 50%. Studies in children with pre-existing immune
suppression (such as HIV infection or neutropenia due to
chemotherapy) were excluded.
Setting
Studies in ambulatory care settings (dened as general or
family practice, paediatric outpatient clinics, paediatric
assessment units, or emergency departments) were selected.
Studies done in developing countries were excluded because
of the dierent range of diseases and more advanced stage of
disease at presentation. We used the United Nations list to
dene developed countries, which include Europe, Canada,
the USA, Australia, New Zealand, and Japan.

1560 articles identied

in initial search

1860 articles screened

1684 excluded in initial screen

of title and abstract
1474 wrong design
83 wrong population
7 wrong setting
112 wrong outcome
8 tests not feasible in
ambulatory care
176 articles identied for further
assessment
79 articles identied
from reference lists
of retrieved articles
and NICE guidelines,
and from expert
consultation
255 full-text articles assessed for
eligibility
151 excluded
55 wrong design
45 wrong population
21 wrong setting
25 wrong outcome
5 laboratory tests

Outcome
Studies that assessed serious infection were selected. Serious
infection was dened as sepsis (including bacteraemia),
meningitis, pneumonia, osteomyelitis, cellulitis,
gastroenteritis with dehydration, complicated urinary tract
infection (positive urine culture and systemic eects such as
fever), and viral respiratory tract infections complicated by
hypoxia (eg, bronchiolitis).
Diagnostic features
Studies that assessed possible triage tests in ambulatory care
were selected. Imaging, invasive tests (such as lumbar
puncture or joint aspiration), and microbiological tests were
not considered; studies reporting laboratory tests available for
near-patient testing were selected (although not reported
here).
Data reporting
Studies were selected if reconstruction of the two-by-two
tables was possible.

See Online for webappendix

Our attempts to draw up guidance for clinicians in the

UK, Belgium, and the Netherlands showed scarcity of
evidence on the diagnosis and management of children
presenting with acute illness. WHO has sponsored largescale studies in resource-poor countries9,10 that provide
evidence relevant to those settings; however, in developed
countries the evidence base seems more limited and
fragmented, and the range of diseases is dierent. The
very low prevalence of serious disease also increases the

300 articles identied in

update in 2009

104 eligible articles underwent

assessment of quality
68 excluded
57 spectrum*
1 reference standard
3 duplicate data
7 insucient data

36 articles included in review

30 assessed clinical features
(included in present analysis)
6 assessed laboratory test results
only (analysed elsewhere)

Figure 1: Flow diagram for selection of studies

Only the rst reason for exclusion (as ordered in the panel) is reported.
NICE=National Institute for Health and Clinical Excellence. *Patient population
not representative of patients in clinical practice.

diagnostic challenge. Therefore, we undertook a

systematic review of the evidence from developed
countries to identify which clinical features have value in
conrming or excluding the possibility of serious infection
in children presenting to ambulatory care settings.

Methods
Search strategy and selection criteria
We searched four electronic databases (Medline, Embase,
DARE, and CINAHL). Search terms (webappendix p 1)

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Design

Setting;
country

Number Proportion
of children
of
children with serious
infection (%)

Quality Age range

rating

Inclusion criteria

Exclusion criteria

Serious infections, composite outcome

Andreola et al
(2007)21

Prosp, cx,
consec

ED; Italy

408

230%

<3 years

Fever of uncertain source and

increased risk of SBIie, all infants
aged 7 days to 3 months with rectal
temperature >38C and children
aged 336 months with ill/toxic
appearance or with rectal
temperature >395C

Antibiotics or vaccination in 48 h before enrolment,

known immunodeciencies, any chronic pathology,
fever >5 days

Baker et al
(1990)22

Prosp,
consec

ED; USA

126

294%

2656 days

Temperature (rectal) >382C

NR

Berger et al
(1996)23

Prosp, cx,
consec

ED;
Netherlands

138

239%

2 weeks to 1 year

Temperature (rectal) 380C

measured on the ward

Gestational age <37 weeks; perinatal complications;

antibiotics or vaccination in previous 48 h; known
previous or underlying disease

Bleeker et al
(2007)24

Prosp, cx,
consec

ED;
Netherlands

381

260%

136 months

Not referred by GP; immune deciencies

Referred to ED for fever without
sourceie, temperature 38C for
which no clear focus could be
identied after assessment by the GP
or after history taking by
paediatrician

Galetto-Lacour
et al (2001)25

Prosp, cx

ED;
Switzerland

124

226%

7 days to 36
months

Temperature (rectal) >380C and no

localising signs of infection from
history or physical examination

Fever >7 days, neonates <1 week of age, children

treated with ABs during the preceding 2 days,
children with known immunodeciencies

Galetto-Lacour
et al (2003)26

Prosp, cx

ED;
Switzerland

99

293%

7 days to 36
months

Temperature (rectal) >38C and

without localising signs of infection
in their history or at physical
examination

Fever >7 days, neonates <1 week of age, children

treated with ABs during the preceding 2 days,
children with known immunodeciencies

Grupo de
Prosp, cx,
Trabajo (2001)27 consec

ED; Spain

739

199%

036 months

Temperature (rectal) 38C

ABs or DTP within 48 h or MMR within 10 days;

systemic central nervous condition; concomitant
analytical changes in blood that interfere with
interpretation of CBC; fever duration >72 h; chronic
illness

Hsiao et al
(2006)28

Prosp, cx,
consec

ED; USA

429

103%

57180 days

Temperature (rectal) >379C

NR

McCarthy et al
(1987)29

Prosp, cx,
consec

ED; USA

143

196%

<24 months

Temperature 383C

NR

McCarthy et al
(1982)30

Prosp, cx,
consec

ED; USA

165

158%

<24 months

Temperature 383C

NR

Nademi et al
(2001)31

Prosp, cx,
consec

PAU; UK

141

291%

016 years

Temperature 38C

Temperature <38C

Thayyil et al
(2005)32

Prosp, cx,
consec

PD; UK

72

111%

136 months

Temperature >39C without

localising signs of infection

ABs 72 h before enrolment, immunodeciencies,

fever >7 days

Thompson et al
(2009)33

Prosp, cx,
consec

PAU; UK

700

553%

3 months to
16 years

Suspicion of acute infection

Children with diseases liable to cause repeated

serious bacterial infection and infections resulting
from penetrating trauma

Trautner et al
(2006)34

Prosp, cx

ED; USA

103

194%

<17 years

Temperature (rectal) 411C

None

Van den Bruel

et al (2007)5

Prosp, cx,
consec

GP, APC, ED; 3981

Belgium

<17 years

Acute illness for a maximum of

5 days

Traumatic or neurological illness, intoxication,

psychiatric or behavioural problems without somatic
cause, exacerbation of a chronic condition

078%

(Continues on next page)

included MeSH terms and free text: serious infections,

children, clinical and laboratory tests, and ambulatory
care. No time or language restrictions were placed on
these searches. The rst search was undertaken in
October, 2008, with an update undertaken in June, 2009.
We checked reference lists of all retrieved articles and
relevant guidelines from the National Institute for Health
and Clinical Excellence published before 2008.11,12 The

Medion database was checked for systematic reviews by

use of the signs and symptoms subheading.
Additionally, domain experts (European Research
Network on Recognising Serious Infection investigators)
were asked to review the list of studies identied and to
report any obvious omissions.
Selection was done by two independent reviewers
(AVdB and TH-H), after piloting on a sample of 20 studies.

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

For the Medion database see

http://www.mediondatabase.nl

Articles

Design

Setting;
country

Number Proportion
of children
of
children with serious
infection (%)

Quality Age range

rating

Inclusion criteria

Exclusion criteria

(Continued from previous page)

Bacteraemia
Crocker et al
(1985)35

Prosp, cx,
consec

ED; USA

201

105%

6 months to
2 years

Temperature (rectal) 394C

Viral exanthema, enanthema, croup, vomiting,

diarrhoea, admitted with a diagnosis of meningitis or
sepsis

Haddon et al
(1999)36

Prosp, cx

ED;
Australia

526

34%

336 months

Temperature (tympanic) 39C

Varicella, croup, or herpes gingivostomatitis

Jae et al
(1991)37

Prosp, cx

ED; Canada

955

28%

336 months

Temperature (rectal) 390C

Focal infection needing immediate AB; clinical

appearance necessitating immediate hospital
admission; specic viral infections; known
immunodeciency condition or chronic illness; AB or
DTP within preceding 48 h

Osman et al
(2002)38

Prosp,
consec

ED; UK

1547

25%

014 years

All children with an infectious illness

NR

Teele et al
(1975)39

Prosp, cx,
consec

ED; USA

600

32%

4 weeks to 2 years

Temperature (rectal) 383C

Previous medical assessment or referral from other

clinician or from other clinic

Waskerwitz et al Prosp, cx,

(1981)40
consec

ED; USA

292

58%

<24 months

Temperature (rectal) 395C

Not previously healthy; weight less than third

percentile or known chronic disease

186

334%

1 month to 5 years

Chief complaint of vomiting,

diarrhoea, or poor oral uid intake

Symptoms >5 days; history of cardiac or renal disease

or diabetes; malnutrition or failure to thrive:
treatment within 12 h in other health facility;
hyponatraemia or hypernatraemia; tonsillectomy
within 10 days; no telephone or beeper for follow-up

83

157%

1 month to 5 years

History of diarrhoea (with or without History of cardiovascular or renal disease; judged by

vomiting) for 5 days and who were the triage nurse to need emergent medical
judged by the ED triage nurse to have intervention
some degree of dehydration

Gastroenteritis causing dehydration

Gorelick et al
(1997)41

Prosp, cx

ED; USA

Shavit et al
(2006)42

Prosp

ED; Canada

Joe et al
(1983)43

Retro

ED; USA

241

54%

6 months to
6 years

First episode of fever and seizures

Did not undergo lumbar puncture and nal outcome

was not available; children with a predisposition to
meningitis

Oringa et al
(1992)44

Retro,
Consec

ED;
Netherlands

309

74%

3 months to
6 years

First episode of fever and seizures

NR

Oostenbrink
et al (2001)45

Retro

ED;
Netherlands

256

387%

1 month to
15 years

Signs of meningeal irritation

Patients with a history of severe neurological disease

or ventricular drainage, and those referred from
other hospitals

MahabeeGittens et al
(2005)46

Prosp, cx

ED; USA

510

86%

259 months

Cough and at least one of laboured,

rapid, or noisy breathing, chest or
abdominal pain, or fever

Currently taking ABs, smoke inhalation, foreign body

aspiration, or chest trauma; known diagnoses of
asthma, bronchiolitis, sickle cell disease, cystic
brosis, chronic cardiopulmonary disease

Taylor et al
(1995)47

Prosp, cx,
consec

ED; USA

572

73%

<2 years

Temperature 380C

Acute wheezing or stridor, history of chronic

pulmonary disease, chest radiograph interpreted as
indeterminate by both radiologists (n=2), clinical
diagnosis of pneumonia with no radiograph
obtained (n=2)

Meningitis

Pneumonia

Meningococcal infection
Nielsen et al
(2001)48

Prosp, cx,
consec

PD;
Denmark

208

188%

>1 month to
<16 years

Haemorrhages in the skin, detected Second or more inclusion in the study

at admission or during hospital stay
plus rectal temperature >38C within
24 h of admission

Wells et al
(2001)49

Prosp, cx,
consec

ED; UK

218

110%

15 years

Non-blanching rash

NR

Prosp=prospective. Cx=cross-sectional. Consec=consecutive. Retro=retrospective. ED=emergency department. SBI=serious bacterial infection. NR=not reported. GP=general practitioner. ABs=antibiotics.
DTP=diphtheria, tetanus, and pertussis vaccine. MMR=measles, mumps, and rubella vaccine. CBC=complete blood count. PAU=paediatric assessment unit. PD=paediatric department. GP=general practice.
APC=ambulatory paediatric care. See text for denitions of quality rating.

Table 1: Characteristics of included studies

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Discrepancies between the reviewers were resolved by a

third independent reviewer (MT).
The studies were selected in two rounds, rst on title
and abstract and second on full text, against the six
criteria shown in the panel.

09

Temperature threshold
used in study
385 to 389C
39 or 395C
40C

33

04

24
49

Quality assessment

Data extraction and analysis

Data were extracted by one reviewer (AVdB) and checked
by a second reviewer (TH-H). Any identied errors were
discussed and corrected; two-by-two tables were
reconstructed on the basis of information in the study or
information retrieved from the study investigators. In
the case of an empty cell, 05 was added to each cell. We
calculated the likelihood ratios for the presence (positive
likelihood ratio) or absence (negative likelihood ratio) of
each clinical feature and pre-test and post-test probabilities
of the outcome. Condence intervals were calculated on
the basis of the standard error of a proportion by use of
STATA version 9.2. The post-test values for temperature
were plotted against pre-test prevalence on a log scale by
use of R software. All other clinical features were
categorised on the basis of their diagnostic value as either
red ags (ie, warning signs for serious infection) or as
rule-out signs for serious infection. Clinical features were

02
Post-test probability

Quality of selected studies and assessment of potential

bias was assessed by the Quality Assessment of
Diagnostic Accuracy Studies (QUADAS) instrument,
including additional items as recommended by the
Cochrane Collaboration.13 Quality assessment was
completed by one reviewer (AVdB) and checked by a
second reviewer (TH-H). Any disagreements were
resolved by discussion involving all researchers when
appropriate. In cases where doubt remained, study
investigators were contacted for clarication. The rst
two QUADAS items (spectrum bias and reference
standard validity) were used as exclusion criteria.
Spectrum bias was judged present in case-control studies
with healthy controls, or in studies in which participants
were selected on the basis of the performance of the
reference standard. The validity of the reference standard
was judged by a clinical review committee consisting of a
minimum of three researchers.
Studies selected for analysis were given an A, B, C, or
D rating. If insucient data were given to be condent
that a criterion had been met, it was assessed as not
being met. Studies fullling all QUADAS criteria were
rated A. Studies without total verication with the
reference standard or with interpretation of the index
feature unblinded to the results of the reference standard
were rated D. Studies without an independent reference
standard, with interpretation of the reference standard
unblinded to the results of the index feature, or with an
unduly long period between recording of the index
feature and outcome were rated C. All other studies
were rated B.

23

45

31

47

01

5
005

38
37 39

001
001

005

01
Pre-test probability

02

04

09

Figure 2: Probability of serious illness in children, by temperature threshold

Reference numbers of studies are shown. Temperature above (closed symbols) or below (open symbols) threshold
in eight studies in health-care settings with dierent pre-test probabilities of serious infection. The distance of the
symbol from the diagonal line indicates the diagnostic value of temperature measurement in the study (applying
the specied threshold). The height above the diagonal line gives the rule-in value; the height below the line the
rule-out value. The gure is plotted on a log scale to achieve visual separation of the studies done in settings with a
low prevalence of infection (in reference 45, the estimated post-test probability was 0% if the temperature was
below 385C, which cannot be plotted on a log scale, so there is no lower symbol). For studies that reported more
than one cuto point, only the value with the highest positive likelihood ratio is shown; for example, the study in a
low prevalence setting also reported cuto points of 38C (positive likelihood ratio 150 and negative likelihood
ratio 040) and 39C (positive likelihood ratio 230 and negative likelihood ratio 060).5 Two of the studies in high
prevalence settings also reported data for temperature cuto points of 385C, 390C, and 400C, all of which
were associated with likelihood ratios of approximately 100.24,31

deemed red ags if, when positive, they substantially

raised the probability of illnessie, positive likelihood
ratio of more than 50. Clinical features were deemed
rule-out signs if, when negative, they substantially
lowered the probability of illnessie, negative likelihood
ratio of less than 02.14 When a study reported more than
one result on the same clinical feature with dierent cuto points, the result with the highest positive likelihood
ratio or lowest negative likelihood ratio was reported.
Features were included in the gures if at least one study
reported a positive likelihood ratio of more than 50 or
negative likelihood ratio of less than 02.
We categorised studies according to setting, with
prevalence of serious infection as a proxy: less than 5% was
dened as low prevalence, 520% as intermediate, and
more than 20% as high prevalence setting. We report both
the pre-test and post-test probabilities of serious infection
for each study in dumbbell plots. Meta-analysis was done
with the bivariate method in STATA version 9.2 when at
least four studies on that clinical feature were available.

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Prevalence*
Study
reference

Global assessment
Parental concern
Clinician instinct that
something wrong
Clinical impression

Age
range

Likelihood ratio (95% CI)

Probability of illness (%)

Positive

Negative

1440 (9302210)
2350 (16803270)

055 (039078)
038 (024060)

Low
Low

<17 years
<17 years

5
36
40
49
42
27
24

Low
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
High

<17 years
336 months
<24 months
15 years
1 month to 5 years
036 months
136 months

830 (6251110)
105 (015748)
275 (156486)
427 (298611)
414 (233735)
220 (178278)
140 (115171)

037 (023062)
100 (090111)
064 (041100)
026 (012056)
028 (010077)
065 (055077)
067 (050088)

Child behaviour
Changed crying pattern 5
24
45
Child drowsy
5
44
45
Child moaning
5
Child inconsolable
5

Low
High
High
Low
Intermediate
High
Low
Low

<17 years
136 months
1 month to 15 years
<17 years
3 months to 6 years
1 month to 15 years
<17 years
<17 years

1050 (4621320)
074 (056096)
049 (025096)
660 (4171050)
199 (129308)
243 (182326)
590 (1971770)
550 (2661150)

067 (051089)
130 (107160)
116 (103131)
065 (049086)
065 (042100)
037 (025056)
092 (081103)
083 (069099)

Child appears ill

5
5

Before test
+ After test if positive
After test if negative

+
+
+
+

+
+
+

+
+

+
+

10

20

30

40

50

60

70

80

90

100

Figure 3: Potential warning signs for serious illness (positive likelihood ratio >50 in at least one study)global assessment and behavioural features
*Setting: low prevalence of serious infection (<5%); intermediate prevalence of serious infection (520%); high prevalence of serious infection (>20%). Parental concern that the illness is dierent
from previous illness. Meningococcal infection only. Gastroenteritis causing dehydration only. Meningitis only.

Prevalence*
Study
reference

Cyanosis

Poor peripheral
circulation

Crackles
Decreased breathing
sounds
Short of breath

Rapid breathing

Age
range

Likelihood ratio (95% CI)

Probability of illness (%)

Positive

Negative

5
33
45

Low
High
High

<17 years
3 months to 16 years
1 month to 15 years

5220 (1050258)
266 (173410)
5020 (297846)

093 (085103)
087 (082093)
088 (081095)

5
42
42
33
41
45
24
5
46||

Low
Intermediate
Intermediate
High
High
High
High
Low
Intermediate

<17 years
1 month to 5 years
1 month to 5 years
3 months to 16 years
1 month to 5 years
1 month to 15 years
1 36 months
<17 years
259 months

3880 (1120134)
471 (207107)
1050 (500221)
1770 (236132)
1170 (478287)
371 (232593)
239 (150382)
600 (3521010)
151 (081283)

090 (080102)
052 (029094)
004 (003060)
092 (088096)
055 (043069)
056 (044073)
083 (073094)
072 (056091)
092 (079107)

5
46||
5
46||
24
5
47||
33

Low
Intermediate
Low
Intermediate
High
Low
Intermediate
High

<17 years
259 months
<17 years
259 months
136 months
<17 years
<2 years
3 months to 16 years

930 (4421970)
221 (089550)
930 (5831480)
111 (070174)
360 (206628)
978 (5711670)
308 (241394)
126 (107149)

082 (068098)
093 (084104)
064 (048085)
096 (078118)
081 (072091)
070 (055089)
037 (023060)
080 (068094)

+
+

+
+

+
+
+

+
+

+
Before test
+ After test if positive
After test if negative

+
+

10

20

30

40

50

60

70

80

90

100

Figure 4: Potential warning signs for serious illness (positive likelihood ratio >50 in at least one study)circulatory and respiratory features
*Setting: low prevalence of serious infection (<5%); intermediate prevalence of serious infection (520%); high prevalence of serious infection (>20%). Meningitis only. Capillary rell more than 2 s.
Gastroenteritis causing dehydration only. Digitally measured capillary rell. ||Pneumonia only.

Role of the funding source

Results

The sponsor of the study had no role in study design,

data collection, data analysis, data interpretation, writing
of the report, or in the decision to submit the paper for
publication. All authors had full access to all the data in
the study. The corresponding author had nal
responsibility for the decision to submit for
publication.

Figure 1 shows the ow diagram of study selection for the

analysis. We selected 36 studies for nal inclusion in the
review, six of which focused on laboratory tests only and
are not included in the analysis reported here.1520 Full
details of the remaining 30 studies are shown in table 1.
The quality of the included studies was modest
(webappendix p 2). Only four studies explicitly mentioned

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Prevalence*
Study
reference

Meningeal irritation

5
44
48
45
5
Petechial rash
44
48
49
45
32
Seizures
5
43
44
45
5
Unconsciousness
44
Decreased skin elasticity 41||
49
Hypotension**
Any abnormal nding in 29
history or physical
examination

Low
Intermediate
Intermediate
High
Low
Intermediate
Intermediate
Intermediate
High
Intermediate
Low
Intermediate
Intermediate
High
Low
Intermediate
High
Intermediate
Intermediate

Age
range

<17 years
3 months to 6 years
>1 month to 16 years
1 month to 15 years
<17 years
3 months to 6 years
>1 month to 6 years
15 years
1 month to 15 years
136 months
<17 years
6 months to 6 years
3 months to 6 years
1 month to 15 years
<17 years
3 months to 6 years
1 month to 5 years
15 years
<24 months

Likelihood ratio (95% CI)

Probability of illness (%)

Positive

Negative

2570 (309213)
275 (16704526)
1390 (5413560)
257 (216306)
1250 (165949)
8370 (4501475)
900 (526153)
700 (4601070)
618 (2681430)
890 (2633040)
2070 (4838860)
590 (1791900)
350 (169717)
168 (066427)
1980 (6176350)
155 (9032677)
1070 (387298)
940 (1994470)
442 (287680)

097 (091103)
052 (035076)
061 (047079)
001 (000015)
097 (091103)
086 (073101)
028 (016048)
019 (008046)
081 (073091)
075 (063091)
094 (086103)
080 (059108)
076 (058100)
096 (090104)
091 (081102)
073 (057093)
067 (056081)
074 (056099)
018 (071044)

+
+

+
+

+
+

+
+

Before test
+ After test if positive
After test if negative

10

20

30

40

50

60

70

80

90

100

Figure 5: Potential warning signs for serious illness (positive likelihood ratio >50 in at least one study)miscellaneous
*Setting: low prevalence of serious infection (<5%); intermediate prevalence of serious infection (520%); high prevalence of serious infection (>20%). Meningitis only. Meningococcal infection.
Diameter more than 2 mm. During examination. ||Gastroenteritis causing dehydration only. **Hypotension dened as 2 SD or more below the mean for age.

masked reading of the reference standard; this item was

scored as unclear in 18 studies. Only seven studies reported
indeterminate or intermediate results. Most studies were
undertaken in emergency departments, with four studies
done in paediatric departments (two paediatric assessment
units),3133,48 and only one done in general practice (which
also recruited non-referred patients from ambulatory
paediatric care and the emergency department).5 Median
prevalence of serious infection was 154% (IQR 80232).
15 studies used a composite outcome of serious infections
consisting of sepsis, bacteraemia, meningitis, pneumonia,
and urinary tract infection (and in some cases additional
infections such as cellulitis, osteomyelitis, and abscess). A
further six studies reported specically on bacteraemia,
three on meningitis, and two each on pneumonia,
meningococcal infection, and gastroenteritis causing
dehydration (table 1).
Figure 2 shows the value of temperature measurement
for diagnosis of serious infection, with dierent cut-o
points and in settings with dierent prevalences of
serious infection. The highest rule-in value was obtained
in the setting with the lowest prevalence, where a
temperature of 40C or more increased the likelihood of
disease from 08% to 50%.5 By contrast, the absence of
high temperature (<385C to 389C) had greatest ruleout value in a study with prevalence of serious infection
of 291%.31 However, this rule-out potential was not seen
in any of the other ve studies with prevalence more than
20% and temperature had no rule-in value in these high
prevalence studies.
Figures 35 show potential red ags for serious
infection (ie, features with a positive likelihood ratio of
more than 50 in at least one study). Both parental

concern that the illness is dierent from previous

illnesses (positive likelihood ratio 1440) and the
clinicians instinct that something is wrong (positive
likelihood ratio 2350) are important red ags in a setting
with a low prevalence of serious infection (gure 3). The
strongest red ags for circulatory or respiratory
impairment are cyanosis, rapid breathing, shortness of
breath, and markers of poor peripheral circulation
(gure 4). Changed crying pattern was a potential red ag
in a low prevalence setting but paradoxically reduced the
probability of serious disease in a high prevalence setting.
Meningeal irritation, petechial rash, and unconsciousness
are important red ags in all settings (gure 5).
Figure 6 reports clinical decision rules with the potential
to exclude the possibility of serious infection. The most
widely studied rule, the Yale Observation Scale, did not
provide satisfactory results. Although the negative
likelihood ratio in two studies was less than 02,30,32 in ve
other studies it ranged from 068 to 097,21,22,25,26,28 and was
associated with post-test probabilities ranging from 10% to
28%. The study in which this score was originally described
obtained data in 198081, before immunisation against
Haemophilus inuenzae and pneumococcus, possibly
accounting for its better performance.30 However, this
explanation would not account for the similar results of
the second study,32 which was done in 2003 in a similar
population of patients to the ve other studies.
Meta-analysis of the seven Yale Observation Scale
studies was limited by signicant heterogeneity
(p=0002), which remained (p=0026) after exclusion of
the rst study with a negative likelihood ratio of less than
02,30 but disappeared (p=0093) after exclusion of both
studies.32 The summary sensitivity of the ve remaining

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Prevalence*
Study
reference

Age
range

Likelihood ratio (95% CI)

Positive

All serious infections

Yale Observation Scale

30
28
32
22
21
25
26
29

Yale Observation Scale

or any normal nding
in history or physical
examination
Five-stage decision tree 5
Pneumonia
Short of breath and
5
parent concerned illness
dierent
Short of breath and
5
clinician concerned
something is wrong
Meningitis
Any abnormal
43
neurological nding or
sought care <48 h
Petechiae or nuchal
44
rigidity or coma
Dehydration from gastroenteritis
Any two of:
41
absent tears, dry
mucous membranes,
ill appearance, poor
peripheral circulation

Probability of illness (%)

Negative

Intermediate
Intermediate
Intermediate
High
High
High
High
Intermediate

<24 months
57180 days
136 months
2656 days
<3 years
7 days36 months
7 days36 months
<24 months

670 (4001110)
110 (062198)
270 (172413)
230 (132390)
180 (138235)
160 (066378)
130 (058292)
233 (179304)

016 (013053)
097 (082115)
019 (003117)
068 (050093)
068 (055085)
091 (074112)
093 (074118)
017 (006051)

Low

<17 years

840 (756938)

004 (000026)

+
+

Low

<17 years

113 (9591330)

007 (001046)

Low

<17 years

1380 (11601630)

007 (001045)

Intermediate

6 months to 6 years

332 (265416)

005 (0003077)

Intermediate

3 months to 6 years

High

1 month to 5 years

610 (380980)

395 (24406377)

Before test
+ After test
After test if negative

+
+

031 (017057)

024 (015039)

10

20

30

40

50

60

70

80

90

100

Figure 6: Clinical decision rules with the potential to rule in or rule out serious infection (positive likelihood ratio >50 or negative likelihood ratio <02 in at least one study)
*Setting: low prevalence of serious infection (<5%); intermediate prevalence of serious infection (520%); high prevalence of serious infection (>20%). Cuto point used: reference 34, more than 8;
reference 21, more than 9; references 24, 27, 28, 30 and 32, more than 10. If yes to any of ve sequential questions: (1) clinician instinct that something is wrong, (2) dyspnoea, (3) temperature more
than 395C, (4) diarrhoea, (5) age 1529 months. Sought care within 48 h before seizure.

studies was 325% (95% CI 217455), with a specicity

of 789% (95% CI 739831), theoretically corresponding
to a positive likelihood ratio of 290 and negative
likelihood ratio of 086.
The best clinical decision rule for excluding the
possibility of serious infection (negative likelihood ratio
004), based on one study,5 involved a ve-stage decision
tree (gure 6). The rule decreased the probability of
serious infections to 003% if negative, but classied
10% of children as potentially seriously ill if positive. The
two decision rules for excluding the possibility of
pneumonia (absence of breathlessness combined with
absence of either parent or clinician concern) performed
equally well (negative likelihood ratio 007, post-test
probability <001%).
Table 2 shows clinical features that were less helpful
in either conrming or excluding the possibility of
serious infection. Common signs and symptoms (cough,
headache, tummy ache, vomiting, diarrhoea, poor
feeding, signs of upper respiratory tract infection) have
little diagnostic value. Failure to smile (positive
likelihood ratio 424) and changed breathing pattern
(positive likelihood ratio 443) are just below the
arbitrary cuto point of 50, so might have some value
8

in conrming but not in excluding (negative likelihood

ratio 064 and 067, respectively) the possibility of
serious infection. The presence of a reactive child
(moving, reaching for objects, looking around the
room)23 or one who is not irritable31 have little diagnostic
value in high prevalence settings. Abnormal skin colour,
described variably as cyanotic, pallor, or ushed or
mottled, seems to be unhelpful as a descriptor, despite
the fact that cyanosis and poor peripheral perfusion
(which causes mottling and pallor) are red ags
(gure 4).
In studies of specic infections, the less helpful clinical
features (table 3) in most cases replicate the ndings in
table 2. Respiratory rate was the most reliable clinical
sign in the diagnosis of pneumonia (positive likelihood
ratio 270400 depending on cuto point) but
breathlessness and auscultatory signs (decreased breath
sounds, crackles, wheezing) had less diagnostic value for
pneumonia than for serious infection as a composite
outcome (gure 4).5,46,47 The individual signs of dehydration
from gastroenteritis (low urine output, sunken eyes, dry
mucous membranes, tachycardia, abnormal respiration)
were all associated with modest likelihood ratios (positive
likelihood ratio 182371).

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Discussion
The strongest red ags for serious infection identied in
this systematic review accord with those previously
identied by WHO for resource-poor countries: reduced
consciousness, convulsions, cyanosis, rapid breathing,
and slow capillary rell (table 4).9 Parental concern and
clinician global impression were also identied as
important diagnostic features in developed countries.
Diculty in feeding seems to be a less helpful red ag in
developed countries than it is in developing countries.
Temperature of more than 40C has value as a red ag in
settings with a low prevalence of serious infection. No
single clinical feature has rule-out value but some simple
combinations can be used to exclude the possibility of
serious infectionfor example, pneumonia is very
unlikely if the child is not short of breath and there is no
parental concern.
The main strength of this systematic review is that it
highlights the nature and diculty of the diagnostic task
facing primary care and hospital clinicians responsible
for identifying seriously ill children at initial presentation
in countries where serious childhood illness is now rare.
We systematically reviewed a range of publications from
which less than 2% of potentially relevant studies
provided adequate and relevant data for inclusion. We
devised innovative methods to aggregate and present the
data so that the results make sense to clinicians, with
graphical representation of the change in the pre-test and
post-test likelihood of serious illness associated with each
clinical feature.
The main weaknesses of our report stem from the
limitations of the studies identied. The most obvious
limitation is the paucity of studies from rst-contact care
settings. One potential weakness that is common to all
diagnostic studies assessing symptoms and clinical signs
is reproducibility. The diagnostic value of a symptom
varies depending on whether it is spontaneously reported
or elicited by questioning. The inter-observer agreement
between clinicians on clinical signs such as capillary rell
time is often poor. Additionally, cultural and language
dierences make aggregation of data from dierent
countries dicult.
A major contrast of our results with the Integrated
Management of Childhood Illness recommendations50 is
that diagnosis of serious infection in children in
developed countries is extremely challenging. Even
warning signs associated with a likelihood ratio of 510
(for example, temperature 40C) might not raise the
probability of disease above 5% in a primary care setting.
Referring all children with a 5% risk to hospital would
overwhelm hospital services; however, informed parents
would probably be unhappy to know that their child was
not being referred despite a 1 in 20 risk of serious
infection. Our analysis also highlights the diculty of
excluding the possibility of serious infection on the basis
of individual clinical featuresclinicians in developed
countries might think that this is the most important

Prevalence*

Likelihood ratio
Positive

Negative

Global assessement
No obvious source of fever28

Intermediate

304

087

Decision rule24

High

207

038

NICE trac light system33

High

120

050

Manchester triage system33

High

135

043

Decision rule33

High

131

052

Child no longer smiles5

Low

424

064

Child is irritable5,31

Low and high

133234

057086

Child is somnolent5

Low

225

081

Child is reactive23

High

133197

056079

Changed breathing pattern5

Low

443

067

Cough5

Low

130

073

Signs of URTI5,34

Low and intermediate

046099

101221

Diarrhoea5,23,34

Low, intermediate, and high

099291

069100

Vomiting5,24,31,34

Low, intermediate, and high

083160

069110

Signs of dehydration||5,24

Low and high

107249

098

Poor feeding5,31

Low and high

137154

051083

Age23,28,34

Intermediate and high

098249

077101

Underlying condition34

Intermediate

242

076

Duration of fever or illness5,21,23,24,34

Low, intermediate, and high

076218

074153

Abnormal skin colour5,23,24

Low and high

159195

061097

Tummy ache5

Low

041

115

Headache5

Low

023

120

Tachycardia**33

High

149205

065085

Child behaviour

Respiratory signs

Gastrointestinal signs

Other signs and symptoms

Clinical features were deemed warning signs if, when positive, they substantially raised the probability of illnessie,
positive likelihood ratio of more than 50. Clinical features were deemed rule-out signs if, when negative, they
substantially lowered the probability of illnessie, negative likelihood ratio of less than 02. NICE=National Institute
for Health and Clinical Excellence. URTI=upper respiratory tract infection. *Setting: low prevalence of serious infection
(<5%); intermediate prevalence of serious infection (520%); high prevalence of serious infection (>20%). Duration of
fever (days), history of vomiting, ill clinical appearance, chest wall retractions with or without rapid breathing, poor
peripheral circulation. One red or amber feature or more. At least one of the following: temperature 39C or more,
oxygen saturation 94% or less, tachycardia, rapid breathing. Moving limbs, reaching for objects, looking around the
room; in isolation or in combination. ||Other than skin inelasticity. **Advanced Pediatric Life Support age-specic cuto points or heart rate more than 90th centile.

Table 2: Clinical features of limited help in conrming or excluding the possibility of any serious infection

nding. Our report shows that infection can only be

ruled out if several clinical features are considered
together. However, the best known clinical decision rule,
the Yale Observation Scale, proved disappointing in
ruling out serious infection.
We excluded some studies that have contributed
substantially to the study of diagnosis of serious illness
in children. Baby Check is a score devised to help parents
and clinicians to detect all serious illness (not simply
infection, hence its exclusion) in children aged less than
6 months.51 In a series of 87 children from UK general
practice (of whom three had serious infection), the score
had a sensitivity of 100% and a specicity of 67% (positive
likelihood ratio 30) at the recommended cuto score of

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

Prevalence*

Likelihood ratio
Postive

WHO Young Infants

Study (2008)9*
(odds ratio [95% CI])

Negative

Bacteraemia

Studies in this
review (range
of odds ratios)

History and behaviour

Child is irritable35

Intermediate 148

061

Convulsions

15 (637)

Child is lethargic35

Intermediate 064

110

Diculty in feeding

10 (715)

23

Functional status40

Intermediate 121257

026055

Reduced consciousness

7 (316)

22212

Age (various cut-os)39

Low

033183

066113

Lethargy

Referral status36

Low

174

079

Sti limbs

Child is irritable45

High

076

105

Circulation and respiration

Vomiting44

Intermediate 253

064

Duration of fever or illness44

Intermediate 143

081
064073
076

Meningitis

Paresis or paralysis44

Intermediate 348

Meningococcal infection
Cough48

Intermediate 041

135

Vomiting48

Intermediate 108

094

Grunting46

Intermediate 056

102

Wheezing46

Intermediate 125

095

Duration46

Intermediate 103

093

Pneumonia

15 (2106)

NR

9 (519)

NR

Cyanosis

14 (16117)

56

Slow capillary rell

11 (522)

9262

Rapid breathing

3 (24)

814

Severe chest wall retraction

9 (421)

NR

Grunting

3 (18)

Hypothermia

Sought care in previous 48 h43,44 Intermediate 228292

4 (27)

522

*The WHO Young Infants Study9 included 3177 children aged less than 2 months
recruited from Bangladesh, Bolivia, Ghana, India, Pakistan, and South Africa; the
odds ratios reported are for children less than 6 days old although the researchers
say that the same clinical predictors can be used for children 759 days old with
similar operating characteristics. For better comparison, the range reported
includes only studies in settings with low and intermediate prevalence.
Temperature less than 355C).

Dehydration from gastroenteritis

Abnormal respiration41

High

310

066

Tachycardia

High

218

068

Abnormal radial pulse41

High

310

066

Sunken eyes41

High

371

047

Dry mucous membranes41

High

362

026

Low urine output

High

182

027

41

41

Clinical features were deemed warning signs if, when positive, they substantially
raised the probability of illnessie, positive likelihood ratio of more than 50. Clinical
features were deemed rule-out signs if, when negative, they substantially lowered
the probability of illnessie, negative likelihood ratio of less than 02. *Setting: low
prevalence of serious infection (<5%); intermediate prevalence of serious infection
(520%); high prevalence of serious infection (>20%). With or without clinician
impression of bacteraemia.

Table 3: Clinical features of limited help in conrming or excluding

possibility of specic infections

less than 8; if used to predict all children needing hospital

admission, sensitivity fell to 80% but with no change in
specicity (positive likelihood ratio 24).51 A key study on
diagnosis of dehydration in children was similarly
excluded from our analysis because it included children
who had been admitted to hospital.52 In the diagnosis of
5% or more dehydration, Mackenzie and colleagues52
reported a positive likelihood ratio of 15 for decreased
skin elasticity and 25 for decreased peripheral
perfusion.
Many of the included studies had only a moderate
QUADAS quality rating. However, several of these
decits are dicult to avoid in this particular clinical
situation. For example, identical verication of all
children for a composite outcome of serious infections is
not feasible, because it would require procedures such as
lumbar puncture, chest radiograph, and blood tests in
10

Table 4: Comparison of predictors of serious illness identied by this

analysis of studies in developed countries with those identied by the
WHO Young Infants Study in developing countries

every child. Moreover, clinical features are part of the

denition of sepsis, by which the reference standard
cannot be interpreted in a masked manner from the
index test.
There were few studies from settings with a low
prevalence of serious infection and there was often
substantial heterogeneity between studies. In many
cases, the heterogeneity was explicable in terms of
setting, inclusion criteria, and cuto values used.
Additionally, some studies obtained data before the
introduction of vaccines against H inuenzae or
pneumococcus. The interpretation of the data for
temperature is especially dicult because in some
studies high temperature was a criterion for inclusion.
However, we do not think that the exclusion of apyrexial
children from several studies in high prevalence settings
is sucient to account for the poor performance of
temperature overall in such settings. Finally, although we
excluded studies that focused on neonates, our analysis
included studies that spanned a wide age range. This
weakness is attenuated by the fact that several studies
created dichotomous predictive variables (eg, rapid
breathing) by applying age-specic cuto values.
However, very few studies reported results by age; to
disaggregate age, it would be necessary to undertake an
individual patient data meta-analysis, which we will
attempt in the future.
The study that provided evidence for the importance of
parental concern and instinct of the clinician was done in
primary care with a prevalence of serious infection of

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

Articles

only 078%.5 Although we would like to see the

importance of parental concern and clinician instinct
replicated in other studies, these results are unlikely to
be chance eectsthe study was large (N=3981) and the
importance of ndings from clinical global assessment
was also seen in other studies. The nding is also
consistent with the fact that the clinical features that
alarm parents and clinicians in their overall assessment
(eg, changes in a childs behaviour, changed crying
pattern, and inconsolability) were also identied as red
ags in other studies in developed countries.
Not only do red ags have less diagnostic value in
developed countries than developing countries, they will
also be seen infrequently even in children with serious
infection. For example, parental concern that the illness
was dierent was noted in only 34% of children and in
around half the cases of serious infection (464%) the
parent did not express such concern.5 Similarly,
clinicians reported poor peripheral circulation in 03%
of children but it was a sign at presentation in only 10%
of serious cases. This nding emphasises that
identication of red ags is not enough. Children with
serious illness who do not have red ags at presentation
will be missed if eective safeguards are not put in
place.53
Many of the clinical features that did not reach our
predened threshold individually could nonetheless
provide useful information for clinical practice when
considered in combinationfor example, dry mucous
membranes for the diagnosis of dehydration (positive
likelihood ratio 41)41 or diarrhoea for the composite
outcome of serious infection.5,34 The most eective
clinical decision rule identied was based on following
up two red ags (clinicians instinct that something is
wrong and dyspnoea) by asking three further questions
about borderline red ags:5 temperature, diarrhoea, and
age. But, as the example of the Yale Observation Scale
shows, combining several borderline red ags, such as
child not smiling, reactivity, and skin colour, does not
necessarily produce a useful decision rule, since the Yale
score gives equal and categorical weight to every item.
Additionally, skin colour is a combination of features,
including pallor and mottled skin, which could be of
variable importance and subject to variation in
interpretation between clinicians.
Most of the red ags already recommended by WHO
for use in developing countries can be used in the initial
assessment of children presenting to ambulatory care
settings in developed countries. There should be more
emphasis on parental concern in the diagnostic process.
However, we now need to identify the level of risk at
which clinical action should be taken. Additionally, the
relative inability of any combination of clinical features
to eectively exclude the possibility of serious illness in a
one-o consultation means that parents need to be more
actively involved in monitoring red ags and taking
precautionary measures.

Contributors
AVdB conceived the study, undertook the literature search and analyses,
and drafted the report. TH-H undertook the literature search and
analyses, and co-drafted the report. MT conceived the study, undertook
the literature search and analyses, and commented on the report.
FB conceived the study and commented on the report. DM conceived
the study, undertook the analyses, and co-drafted the report.
ERNIE (European Research Network on Recognising Serious Infection)
investigators
Principal investigators Ann Van den Bruel, Matthew Thompson,
Frank Buntinx, David Mant, Tanya Haj-Hassan, Rianne Oostenbrink,
Henriette Moll, Bert Aertgeerts, Monica Lakhanpaul.
Acknowledgments
This study was funded by the Health Technology Assessment Project
07/37/05 (Systematic review and validation of clinical prediction rules for
identifying children with serious infections in emergency departments
and urgent-access primary care) and the National Institute for Health
Research National School for Primary Care Research. We would like to
thank Paul Glasziou (Centre for Evidence Based Medicine, Oxford, UK)
for advising us on the tables and gures. The Fonds Wetenschappelijk
Onderzoek Vlaanderen (Research Foundation-Flanders) initially funded
the study by Van den Bruel and colleagues5 and granted extension to pay
for some of the work presented in this report.
Conicts of interests
We declare that we have no conicts of interest.
References
1
CEMACH. Why children die: a pilot study 2006; England (South
West, North East and West Midlands), Wales and Northern Ireland.
London: CEMACH, 2008.
2
Bruijnzeels MA, Foets M, van der Wouden JC, van den Heuvel WJ,
Prins A. Everyday symptoms in childhood: occurrence and general
practitioner consultation rates. Br J Gen Pract 1998; 48: 88084.
3
Fleming DM, Smith GE, Charlton JR, Charlton J, Nicoll A. Impact
of infections on primary caregreater than expected.
Commun Dis Public Health 2002; 5: 712.
4
Armon K, Stephenson T, Gabriel V, et al. Determining the common
medical presenting problems to an accident and emergency
department. Arch Dis Child 2001; 84: 39092.
5
Van den Bruel A, Aertgeerts B, Bruyninckx R, Aerts M, Buntinx F.
Signs and symptoms for diagnosis of serious infections in
children: a prospective study in primary care. Br J Gen Pract 2007;
57: 53846.
6
Kai J. What worries parents when their preschool children are
acutely ill, and why: a qualitative study. BMJ 1996; 313: 98386.
7
Riordan FA, Thomson AP, Sills JA, Hart CA. Who spots the spots?
Diagnosis and treatment of early meningococcal disease in
children. BMJ 1996; 313: 125556.
8
Thompson MJ, Ninis N, Perera R, et al. Clinical recognition of
meningococcal disease in children and adolescents. Lancet 2006;
367: 397403.
9
The Young Infants Clinical Signs Study Group. Clinical signs that
predict severe illness in children under age 2 months: a multicentre
study. Lancet 2008; 371: 13542.
10 Duke T, Oa O, Mokela D, Oswyn G, Hwaihwanje I, Hawap J.
The management of sick young infants at primary health centres in
a rural developing country. Arch Dis Child 2005; 90: 20005.
11 NICE. Feverish illness in children: assessment and initial
management in children younger than 5 years. London: National
Institute for Health and Clinical Excellence, 2007.
12 NICE. Urinary tract infection in children: diagnosis, treatment and
long-term management. London: National Institute for Health and
Clinical Excellence, 2007.
13 Whiting P, Rutjes AWS, Reitsma JB, Bossuyt PMM, Kleijnen J.
The development of QUADAS: a tool for the quality assessment of
studies of diagnostic accuracy included in systematic reviews.
BMC Med Res Methodol 2003; 3: 25.
14 Jaeschke R, Guyatt GH, Sackett DL. Users guides to the medical
literature. III. How to use an article about a diagnostic test. B.
What are the results and will they help me in caring for my
patients? The Evidence-Based Medicine Working Group. JAMA
1994; 271: 70307.

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6

11

Articles

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

12

Bachur RG, Harper MB. Predictive model for serious bacterial

infections among infants younger than 3 months of age. Pediatrics
2001; 108: 31116.
Bonadio WA, Hagen E, Rucka J, Shallow K, Stommel P, Smith D.
Ecacy of a protocol to distinguish risk of serious bacterial
infection in the outpatient evaluation of febrile young infants.
Clin Pediatr (Phila) 1993; 32: 40104.
Lacour AG, Zamora SA, Gervaix A. A score identifying serious
bacterial infections in children with fever without source.
Pediatr Infect Dis J 2008; 27: 65456.
Garra G, Cunningham SJ, Crain EF. Reappraisal of criteria used to
predict serious bacterial illness in febrile infants less than 8 weeks
of age. Acad Emerg Med 2005; 12: 92125.
Baker MD, Bell LM, Avner JR. Outpatient management without
antibiotics of fever in selected infants. N Engl J Med 1993;
329: 143741.
Baker MD, Bell LM, Avner JR. The ecacy of routine outpatient
management without antibiotics of fever in selected infants.
Pediatrics 1999; 103: 62731.
Andreola B, Bressan S, Callegaro S, Liverani A, Plebani M,
Da Dalt L. Procalcitonin and C-reactive protein as diagnostic
markers of severe bacterial infections in febrile infants and children
in the emergency department. Pediatr Infect Dis J 2007; 26: 67277.
Baker MD, Avner JR, Bell LM. Failure of infant observation scales in
detecting serious illness in febrile, 4- to 8-week-old infants.
Pediatrics 1990; 85: 104043.
Berger RM, Berger MY, van Steensel-Moll HA, Dzoljic-Danilovic G,
Derksen-Lubsen G. A predictive model to estimate the risk of
serious bacterial infections in febrile infants. Eur J Pediatr 1996;
155: 46873.
Bleeker SE, Derksen-Lubsen G, Grobbee DE, Donders AR,
Moons KG, Moll HA. Validating and updating a prediction rule for
serious bacterial infection in patients with fever without source.
Acta Paediatr 2007; 96: 10004.
Galetto-Lacour A, Gervaix A, Zamora SA, et al. Procalcitonin, IL-6,
IL-8, IL-1 receptor antagonist and C-reactive protein as identicators
of serious bacterial infections in children with fever without
localising signs. Eur J Pediatr 2001; 160: 95100.
Galetto-Lacour A, Zamora SA, Gervaix A. Bedside procalcitonin and
C-reactive protein tests in children with fever without localizing signs
of infection seen in a referral center. Pediatrics 2003; 112: 105460.
Grupo de Trabajo sobre el Nino Febril de la Sociedad Espanola de
Urgencias de P. The young febrile child. Results of a multicenter
survey. An Esp Pediatr 2001; 55: 510 (in Spanish).
Hsiao AL, Chen L, Baker MD. Incidence and predictors of serious
bacterial infections among 57- to 180-day-old infants. Pediatrics
2006; 117: 1695701.
McCarthy PL, Lembo RM, Fink HD, Baron MA, Cicchetti DV.
Observation, history, and physical examination in diagnosis of
serious illnesses in febrile children less than or equal to 24 months.
J Pediatr 1987; 110: 2630.
McCarthy PL, Sharpe MR, Spiesel SZ, et al. Observation scales to
identify serious illness in febrile children. Pediatrics 1982;
70: 80209.
Nademi Z, Clark J, Richards CG, Walshaw D, Cant AJ. The causes
of fever in children attending hospital in the north of England.
J Infect 2001; 43: 22125.
Thayyil S, Shenoy M, Hamaluba M, Gupta A, Frater J, Verber IG. Is
procalcitonin useful in early diagnosis of serious bacterial infections
in children? Acta Paediatr 2005; 94: 15558.
Thompson MJ, Coad N, Harnden A, Mayon-White R, Perera R,
Mant D. How well do vital signs identify children with serious
infections in paediatric emergency care? Arch Dis Child 2009;
94: 88893.

34

35

36

37
38

39

40

41

42

43

44

45

46

47

48

49

50

51

52
53

Trautner BW, Caviness AC, Gerlacher GR, Demmler G, Macias CG.

Prospective evaluation of the risk of serious bacterial infection in
children who present to the emergency department with
hyperpyrexia (temperature of 106 degrees F or higher). Pediatrics
2006; 118: 3440.
Crocker PJ, Quick G, McCombs W. Occult bacteremia in the
emergency department: diagnostic criteria for the young febrile
child. Ann Emerg Med 1985; 14: 117277.
Haddon RA, Barnett PL, Grimwood K, Hogg GG. Bacteraemia in
febrile children presenting to a paediatric emergency department.
Med J Aust 1999; 170: 47578.
Jae DM, Fleisher GR. Temperature and total white blood cell count
as indicators of bacteremia. Pediatrics 1991; 87: 67074.
Osman O, Brown D, Beattie T, Midgley P. Management of febrile
children in a paediatric emergency department. Health Bull (Edinb)
2002; 60: 3339.
Teele DW, Pelton SI, Grant MJ, et al. Bacteremia in febrile
children under 2 years of age: results of cultures of blood of
600 consecutive febrile children seen in a walk-in clinic. J Pediatr
1975; 87: 22730.
Waskerwitz S, Berkelhamer JE. Outpatient bacteremia: clinical
ndings in children under two years with initial temperatures of
395 degrees C or higher. J Pediatr 1981; 99: 23133.
Gorelick MH, Shaw KN, Murphy KO. Validity and reliability of
clinical signs in the diagnosis of dehydration in children. Pediatrics
1997; 99: E6.
Shavit I, Brant R, Nijssen-Jordan C, Galbraith R, Johnson DW.
A novel imaging technique to measure capillary-rell time:
improving diagnostic accuracy for dehydration in young children
with gastroenteritis. Pediatrics 2006; 118: 240208.
Joe A, McCormick M, DeAngelis C. Which children with febrile
seizures need lumbar puncture? A decision analysis approach.
Am J Dis Child 1983; 137: 115356.
Oringa M, Beishuizen A, Derksen-Lubsen G, Lubsen J. Seizures
and fever: can we rule out meningitis on clinical grounds alone?
Clin Pediatr (Phila) 1992; 31: 51422.
Oostenbrink R, Moons KG, Donders AR, Grobbee DE, Moll HA.
Prediction of bacterial meningitis in children with meningeal
signs: reduction of lumbar punctures. Acta Paediatr 2001;
90: 61117.
Mahabee-Gittens EM, Grupp-Phelan J, Brody AS, et al. Identifying
children with pneumonia in the emergency department.
Clin Pediatr (Phila) 2005; 44: 42735.
Taylor JA, Del Beccaro M, Done S, Winters W. Establishing
clinically relevant standards for tachypnea in febrile children
younger than 2 years. Arch Pediatr Adolesc Med 1995; 149: 28387.
Nielsen HE, Andersen EA, Andersen J, et al. Diagnostic
assessment of haemorrhagic rash and fever. Arch Dis Child 2001;
85: 16005.
Wells LC, Smith JC, Weston VC, Collier J, Rutter N. The child with
a non-blanching rash: how likely is meningococcal disease?
Arch Dis Child 2001; 85: 21822.
Gove S. Integrated management of childhood illness by outpatient
health workers: technical basis and overview. The WHO Working
Group on guidelines for Integrated Management of the Sick Child.
Bull World Health Organ 1977; 75 (suppl 1): 724.
Morley CJ, Thornton AJ, Green SJ, Cole TJ. Field trials of the Baby
Check score card in general practice. Arch Dis Child 1991;
66: 11114.
Mackenzie A, Barnes G, Shann F. Clinical signs of dehydration in
children. Lancet 1989; 334: 60507.
Almond S, Mant D, Thompson M. Diagnostic safety-netting.
Br J Gen Pract 2009; 59: 87274.

www.thelancet.com Published online February 3, 2010 DOI:10.1016/S0140-6736(09)62000-6