zyxwvutsrqponm
zyxwvutsrqpo
zyxwvutsr
J. Paediatr. Child Health (1996) 32, 532-535
The prediction of hospital admission in children with
acute asthma
zyxwv
SH TWADDELL,’ RL HENRY,’ JL FRANCIS and PG GIBSON2
‘University of Newcastle and 2Departrnent of Respiratory Medicine, John Hunter Hospital, New South Wales, Australia
Objective: To determine whether a single assessment of children at the time of presentation to the emergency department
would discriminate accurately between those requiring admission and those who could be managed at home and to examine
the appropriateness of these decisions.
Methodology: Fifty-three children were assessed using a table recommended by Australian and New Zealand respiratory
pediatricians, which categorizes children as probably being able to manage at home (group l ) , may need admission to
hospital (group 2) and certainly need admission to hospital (group 3) on the basis of oximetry, presence of wheeze and pulsus
paradoxus.
Results: Nine out of 1 1 children assigned to group 1 were managed at home and 15/17 who were predicted to require
admission were admitted. No individual component of the assessment dominated the decision made. Of the 25 children
allocated to group 2, 18 were admitted.
Conclusions: The method employed was highly predictive of outcome for half of the children who presented with asthma.
However, 25/53 (47%)were assigned by the table to a recommendation for further assessment; this limits its usefulness.
Key words: admission, asthma, oximetry
Asthma is a common reason for children to be admitted to
hospital.’ -3 In particular many children present to their general
practitioner or to an emergency department with an acute attack
of asthma. A decision needs to be made about whether they can
be managed safely at home or whether admission to hospital is
necessary. Inappropriate discharge is associated with a risk of
representation or relapse; unnecessary admission causes distress and disruption to the child and family and is expensive.
Many attempts have been made to help doctors decide which
patients with asthma require admission to hospital. Various
clinical scoring systems designed for both ~ h i l d r e n ~and
-~
adult^^-'^ have been developed; none has been universally
accepted. Some scales have relied on clinical parameters such
as heart rate, respiratory rate, pulsus paradoxus, breathlessness,
chest wall signs and wheezing; others have used measurements
such as oximetry or peak expiratory flow and some have used a
combination. No accepted system that predicts who requires
admission to hospital and who can be discharged has been
published.
The Australian and New Zealand Respiratory Paediatricians
Group have published a set of guidelines for the management of
asthma.” Derived in large measure from the published literature,
the guidelines include a table (Table 1 ) which attempts to clarify
the factors to consider in admission or discharge. The paper
Correspondence: Professor RL Henry, Department of Paediatrics, John
Hunter Hospital, Locked Bag 1, Hunter Region Mail Centre, NSW 2310,
Australia.
SH Twaddell, Dip App Sci, B Med Sci; RL Henry, MB. BS. MD, FRACP.
Dip Clin Epid; JL Francis, BA. M Med Stats. PG Gibson MB, BS, FRACP.
Accepted for publication 22 July 1996.
indicated that the consensus table should not be used in isolation
but that other factors needed to be considered. The aims of the
current study were to determine: (i) whether the guidelines
accurately predicted those children were admitted to hospital
and those sent home; and (ii) whether that decision was
appropriate.
METHODS
Fifty-three children with acute asthma and a mean age of 5.1
years (range 1-1 5) were assessed by one of the authors (SHT)
at the time of presentation to the emergency department at the
John Hunter Hospital, Newcastle. Children were assessed at the
same time as their clinical assessment by the resident medical
officer on duty in the emergency department. All assessments
were conducted prior to the children receiving any therapy for
asthma at the hospital. The clinical assessment performed by
the resident medical officer occurred independently of the study,
as did the decision about whether the child would be admitted
to hospital or managed at home.
The member of our research team (SHT) made the clinical
observations which enabled assessment according to Table 1 .
There is some subjectivity in interpretation for a number of these
categories. Prior to the study the observer (SHT) received training
about implementation of the table; in particular he observed
children with asthma together with another of the team (RLH) to
ensure calibration and consistency. Level of consciousness
was assessed by observation of alertness, as evident by response to surroundings, speech or pain. A conscious but tired
child was regarded as having no alteration in consciousness; if
the child was difficult to rouse, altered consciousness was said
zyx
Hospital admission in children with asthma
zyxwvuts
533
to be present. Physical exhaustion was assessed by observing
the general movements of the child with assignment to this
category allocated if the child was having extreme difficulty
producing a respiratory effort and appearing unable to maintain
ventilation. Talking in sentences, phrases or words was assessed
by talking directly to the child and observing his or her interaction with parents and others; the categorization was how
many words could be spoken before the child had to pause for
breath. In children too young to speak or too young to speak full
sentences, this element of the assessment was not scored.
The presence or absence of pulsus paradoxus, central
cyanosis and wheeze on auscultation were easily defined. The
absence of wheeze in the presence of very poor air entry was
classified as a silent chest.
Chest wall signs were quantified by the extent of tracheal tug,
intercostal recession and subcostal recession present when the
child was at rest and breathing undisturbed. Moderate chest
wall signs were scored when intercostal or subcostal recession
were present in the absence of tracheal tug. Marked chest wall
signs were registered when there was tracheal tug, intercostal
and subcostal recession. The table refers to these as ‘accessory
muscle use’.
Sternal retraction was defined as a change in the position of
sternum in relation to the ribs with respiration. Moderate was
defined as a change of up to 1 cm in the position of the sternum
in relation to the ribs between inspiration and expiration.
Retraction greater than 1cm was considered marked sternal
retraction.
Peak expiratory flow (PEF) was based on the best of three
maximum expiratory efforts through a Wright PEF (Airmed Ltd)
meter and was expressed as a percentage of predicted. This
measurement was only used in children whose technique
appeared adequate (usually over the age of 5). Peripheral arterial
oxygen saturation was recorded using the PROPAQ 106 pulse
oximeter (Protocol Systems Inc.). The probe was placed on the
index finger with the child settled and not crying. The mode
measurement recorded over a 30s interval was taken as the
saturation and the reading was recorded only when there was a
stable pulsewave display.
Each child was assigned to one of three categories for each
of the 10 items in Table 1. The overall rating was allocated to the
most severe rating assigned for any of the 10 variables; therefore
if one item suggested that the child certainly needed admission
to hospital, that was the overall rating.
Information was collected from parents about the control of
asthma (symptoms in the week prior to presentation), background severity of the disease (mild episodic, frequent episodic
or persistent symptoms in previous year), home management
skills, triggers to attacks, medications, and cultural and socioeconomic background. Management skills included assessments of the adequacy of parental knowledge, inhaler technique,
compliance, symptom recognition and whether there had been
a written action plan. These were collated to a composite rating
of adequate or inadequate management skills. Cultural and
socio-economic factors included language barriers, use of
public transport to the hospital, distance from hospital and
parental smoking.
For children whom the resident medical officer discharged
from the emergency department, telephone contact was made
with the parents 1 week later by SHT. A scripted questionnaire
sought information regarding the child’s condition in the period
after discharge from the emergency department. In particular,
we sought information about the frequency of medication,
recovery time and how much respiratory embarrassment the
child had experienced at home. For those children who were
admitted to hospital, data were collected about the use of
supplemental oxygen, intravenous medications, and the frequency of the use of nebulized beta agonists. On the basis of
these retrospective assessments, and without regard to the
child’s score on assessment by the table, the appropriateness
or otherwise of the resident medical officer’s decision to admit
or discharge the child was made by review (RLH) of the child’s
subsequent clinical course. In particular, admission was considered warranted if the child received supplemental oxygen,
intravenousfluids or if the frequency of a nebulized bronchodilator
in hospital was more frequent; discharge was considered
appropriate if the frequency of nebulized therapy at home was
less often than once every 3 h and that there was no evidence of
severe respiratory problems such as cyanosis.
Simple descriptive analyses were performed. A Kappa statistic
was used to compare the table’s predictions with the actual
outcomes. Likelihood ratios were also used to predict the likelihood of admission according to the category assigned by the
table, and pre-test and post-test probabilities were determined.12
For those children assigned by the table to an intermediate
category, the relationship between admission and the possible
confounding external variables of severity of disease, management skills and socio-economic aspects were tested using a x2
zyxwvutsrq
zy
Table 1 Guidelines for the assessment of children with acute asthma
Altered consciousness
Physical exhaustion
Talks in
Pulsus paradoxus
Central cyanosis
Wheeze on auscultation
Use of accessory muscles
Sternal retraction
(in young children)
Initial PEF (% predicted
or % child’s best)
Oximetry on presentation
prior to nebulized
treatment ( S , O ~
Group 1
Probably manage at home
Group 2
May need admission to hospital
Group 3
Certainly need admission and may require ICU
No
No
Sentences
Not palpable
Absent
Present
Absent
No
No
Phrases
May be palpable
Absent
Present
Moderate
Yes
Yes
Words
Palpable
Present
Silent chest
Marked
Absent
Moderate
Marked
> 60%
40-60%
< 40%
>93%
91 -93%
90 and below
534
zyxwvutsrqp
zyxwvutsrq
SH Twaddell eta/.
test. Both the skill and the socio-economic aspects were
analysed as dichotomousvariables and disease severity as mild
episodic, frequent episodic or persistent.
Informed consent was obtained from parents, children who
were 8 years or older and from the resident medical officers in
the emergency department.
RESULTS
Thirty-five of the 53 children who had been recruited were
admitted to hospital. As shown in Table 2, 9/11 children who
were predicted by the table to be able to be managed at home
were and 15/17 children predicted to require admission to
hospital were admitted. There was almost complete concordance
with the medical decision regarding admission or discharge
and the retrospective assessment of appropriate practice. In
49/53 children, there was agreement and in four cases (two
children admitted and two discharged) it was unclear from the
retrospective analysis as to whether an appropriate decision
had been made. In particular, in those four cases the children
were treated with bronchodilators at approximately 3 hourly
Table 2 Comparison of outcomes according to prediction of table and
actual admission or discharge
Prediction by table
Admitted
Discharged
2
18
15
9
Manage at home (Group 1)
Possibly admit (Group 2)
Admit (Group 3)
~~
7
2
~
Kappa statistic on children in groups 1 and 3 = 0.7
Table 3 Distribution of scores for the 10 parameters assessed
Group 1
Manage
at home
Altered consciousness
Physical exhaustion
Talks in
Palpable pulsus paradoxus
Central cyanosis
Wheeze‘
Accessory muscle use
Sternal retraction
Peak flow monitoring
Oximetry
Group 2
Consider
admission
Group 3
Definitely
admit
-
53
52
28
1
4
6
3
-
44
50
51
21
51
4
2
6
5
40
*Two children did not have a wheeze but had good air entry at the time
of initial presentation.
Table 4 Likelihood ratios and post-test probabilitiesof admission based
on predictions of Table 1
Admit
(group 3)
Possibly admit
(group 2)
Manage at home
(group 1)
Likelihood ratio
Post-test
probability
3.86
89
1.32
73
0.11
18
intervals. Of those admitted to hospital a total of seven children
received supplemental oxygen or intravenous therapy, four
received intravenous therapy and five supplemental oxygen (in
addition to the oxygen used as the driving gas to deliver
nebulized bronchodilators).
As shown in Table 3. 5/10 parameters scored showed little
variation. Fifty or more children scored in the most favourable
category for altered conscious state, physical exhaustion, central
cyanosis. wheeze and sternal retraction. Furthermore,peak flow
monitoring was possible in only one-quarter of cases. No single
parameter dominatedthe assignment to the category of ‘definitely
admit’. The use of accessory muscles was the only factor
leading to assignment to the third category in four cases,
palpable pulsus paradoxus and oximetry 590% in two cases,
and cyanosis and talking in words one case each.
Likelihood ratios were calculated (that is the odds that the
prediction of the table would be expected in a child admitted
rather than were they to be managed at home) and these used
to make a comparison of pre-test probability and post-test
probability of admission to hospital (Table 4). In this study twothirds of cases were admitted so that the pre-test probability of
admission was 67%. For both groups 1 and 3, the post-test
probability was altered markedly but there was little difference if
the clinical assessment assigned the child to the ‘possibly
admit’ category.
For the 25 children for whom the table did not make a
decision about admission (group 2), other elements were considered to possibly contribute to the admission or discharge.
Fourteen of these children had mild episodic asthma (10 were
admitted), seven had frequent episodic asthma (five were
admitted) and four had persistent asthma (three were admitted).
There was no significant correlation among background severity
of disease, asthma management skills or socio-economic factors
and admission or discharge from the emergency department (x2
test, P-values> 0.05).
zy
zyxwvu
DISCUSSION
We have explored whether the use of a table, without consideration of any other factors, accurately predicts whether children
with an acute attack of asthma should be managed at home or
in hospital. In 28/53 children, the table resulted in a definite
decision to admit or to manage at home. The agreement between
what actually happened and the retrospective audit of what
should have happened (based on subsequent clinical course)
was high. The table had utility and was both sensitive and
specific when children were allocated to either groups 1 or 3.
Not unexpectedly, many items were of little value for most
children (for example altered conscious state is highly significant
if present but is usually absent). Peak expiratory flow was of little
use as most of the children assessed could not perform the
manoeuvre. Indeed only 2/17 admissions predicted by the table
included a peak expiratory flow of<40% of predicted. The
parameters which did appear to be of most use were the chest
wall signs of tracheal tug, intercostal and subcostal recession
(which are sometimes called use of accessory muscles), oximetry
and pulsus paradoxus. No single element was as good at
predicting outcome as the composite of the table. In particular,
oximetry could not be used in isolation, a finding supported by
other ~ t u d i e s . ’ ~ - ’ ~
It had been anticipated that factors other than the elements of
the table would be important. Indeed the original article” which
Hospital admission in children with asthma
zyxwvutsr
535
presented the table argued that in children assigned to the
‘possibly admit’ category medical and non-medical factors should
be considered before deciding on admission or discharge, and
identified several factors considered important to assess prior
to making a decision. In the current study our attempts to
identify factors which might predict whether admission or discharge was appropriate in the children assigned to the indeterminate category were unsuccessful. In particular, factors such as
background severity of disease, management skills and socioeconomic factors did not prove to be contributing factors.
However, the small sample size raises the possibility of a Type II
error.
The study design demanded that the observer (SHT) be
present in the emergency department when the child presented
and when the child was examined by the resident. This was a
demanding protocol. A larger sample size would have improved
the precision of the estimates of likelihood ratios but is unlikely
to alter the overall message.
A study such as the current one suffers from the absence of a
gold standard in establishing who should or should not be
admitted to hospital. In this situation we chose two outcome
factors; namely, whether the children were admitted to hospital
and a retrospective review of the frequency of medication use
and the subsequent clinical course. It has to be acknowledged
that these factors are not independent of each other. In particular
one might speculate that residents at John Hunter Hospital have
been taught to assess children with acute asthma by judging
many of the parameters that appear in the consensus table. This
would raise the concern that the instrument under test and the
resident medical officer’s decision were not adequately independent. Many of the doctors working in the emergency department
at the John Hunter Hospital were graduates of the University of
Newcastle and had received tuition about the management of
acute asthma by the Professor of Paediatrics (RLH). Furthermore,
the protocol recommendation by the Australian and New Zealand
Respiratory Paediatricians had been endorsed by the emergency
department. However, at the time of the study the table was not
on display in the department and the residents were unaware
that copies of the table were available.
Obviously one would expect a high level of concordance if
the instrument was basically being tested against itself. The
effects of this bias would be to overestimate the usefulness of
the table. Since the table was found to be of limited benefit, our
results may have overestimated its benefit.
Furthermore, the agreement between the resident’s decision
to admit to hospital and the retrospective decision (by RLH)
showed better agreement than between the resident and the
table. This suggests that the residents’ opinions were in concordance with expert opinion but that the table was inferior.
The admission rate of 67% of the cohort was much higher
than e ~ p e c t e d . ’ ~ -Previous
’~
audits at the hospital have indicated that 25% of children presenting to the emergency department with acute asthma are admitted. We tried to obtain a
representative sample of children in this study; however, the
study requirements that children must be seen immediately on
presentation and prior to receiving any bronchodilators at the
hospital meant that patient recruitment was difficult. It seemed
that residents and nursing staff tended to notify the research
team when people with moderately severe asthma presented
rather than those with less severe disease. For the current
sample, the likelihood ratios and post-test probabilities suggested that the table did provide useful information when
children were categorized to the ‘probably manage at home’ or
‘needs admission’ categories. The assignment to ‘may need
admission’ did not help in terms of the difference between preand post-test probability. Likelihood ratios are said to be very
stable with changes in prevalence.12 In a population of children
with a 20-25% risk of admission to hospital (which is the sort of
figure one would anticipate), assignment to the category of ‘may
need admission’ to hospital would result in a useful improvement
in the post- compared with the pre-test probability. Clearly
these arguments are theoretical and need to be tested.
Although the table was highly predictive of an outcome in
acute asthma for half the cases, the remaining half were assigned
by the table to a recommendation for further assessment. Further
studies are required to determine whether the table needs to be
applied in an iterative fashion in order to make it more useful.
Obviously this is a testable hypothesis. It may be possible to use
the table on patient presentation to the emergency department
and then a few hours later in problematic cases.
zyxwvuts
ACKNOWLEDGEMENTS
The authors would like to thank Joan Welsh.
REFERENCES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
zyx
Anderson HR. Increase in hospitaladmissions for childhood asthma:
trends in referral, severity, and readmissions from 1970 to 1985 in a
health region of the United Kingdom. Thorax 1989; 44: 614-19.
Carman PG, Landau LI. Increasedpaediatric admissions with asthma
in Western Australia-a problem of diagnosis? Med. J. Ausf. 1990;
152: 23-6.
Connett GJ, Warde C, Wooler E, Lenney W. Audit strategies to
reduce hospital admissions for acute asthma. Arch. Dis. Child 1993;
69: 202-5.
Becker AB, Nelson NA, Simons FER. The Pulmonary Index: Assessment of a clinical score for asthma. Am. J. Dis. Child 1984; 138:
574-6.
Baker MD. Pitfalls in the use of clinical asthma scoring. Am. J. Dis.
Child 1988; 142: 183-5.
Kerem E, Tibshirani R, Canny G et a/. Predicting the need for
hospitalisation in children with acute asthma. Chest 1990; 98:
1355-61.
Lulla S, Newcomb RW. Emergency management of asthma in
children. J. Pedietr. 1980; 97: 346-50.
Ownby DR, Abarzua J, Anderson JA. Attempting to predict hospital
admission in acute asthma. Am. J. Dis. Child 1984; 138: 1062-6.
Fischl MA, PitchenikA. Gardner LB. An index predicting relapse and
need for hospitalisation in patients with acute bronchial asthma. N.
Engl. J. Med. 1981; 305: 783-9.
Griner PF, Glaser RJ. Misuse of laboratory tests and diagnostic
procedures. N. Engl. J. Med. 1982; 307: 1336-9.
Henry RL, RobertsonCF, Asher I et a/. Managementof acute asthma.
J. Paedietr. Child Health 1993; 29: 101-3.
Anon. Clinical epidemiology rounds. Interpretation of diagnostic
data: 5. How to do it with simple maths. Can. Med. Assoc. J. 1983;
129: 947-54.
Geelhoed GC, Landau LI. Le Souef PN. Predictive value of oxygen
saturation in emergency evaluation of asthmatic children. BMJ1988;
297: 395-6.
Geelhoed GC, Landau LI, Le Souef PN. Oximetry and peak expiratory
flow in assessment of acute childhood asthma. J. Pediatr. 1990; 117:
907-9.
Bishop J. Nolan T. Pulse oximetry in acute asthma. Arch. Dis. Child
1991; 66: 724-5.
Mitchell EA. International trends in hospital admission rates for
asthma. Arch. Dis. Child 1985; 60: 376-8.
Hyndman SJ. Williams DRR. Merrill SL. Lipscombe JM, Palmer CR.
Rates of admission to hospital for asthma. BMJ 1994; 308: 1596-1 600.
zy