The prediction of hospital admission in children with acute asthma

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