Abstract
INTRODUCTION: Severe acute asthma in children is associated with substantial morbidity and may require pediatric ICU (PICU) admission. The aim of the study was to determine risk factors for PICU admission.
METHODS: The study used a retrospective multicenter case-control design. The cases included children admitted to the PICU because of severe acute asthma and a history of out-patient treatment by pediatricians or pediatric pulmonologists. Controls were children with asthma without a PICU admission for severe acute asthma. The children were matched for sex, age, hospital, and time elapsed since the diagnosis of asthma. Fourteen possible risk factors were analyzed.
RESULTS: Sixty-six cases were matched to 164 controls. In univariate analysis, all but one of the analyzed variables were significantly associated with PICU-hospitalization. After multivariate conditional logistic regression analysis, 4 risk factors remained significant. These included active or passive smoking, allergies, earlier hospitalization for asthma, and non-sanitized home.
CONCLUSIONS: Physicians and parents should be aware of these risk factors, and efforts should be made to counteract them.
Introduction
Asthma is the most prevalent chronic pediatric disorder, affecting approximately 6–12% of the Dutch pediatric population.1 Severe acute asthma (SAA) may be life threatening, as some children fail to respond to conventional asthma treatment (ie, administration of oral/parenteral steroids and inhaled β2 agonists). Severe complications may occur, such as pneumothorax, ventilator-associated pneumonia, rhabdomyolysis, circulatory failure, and even death.2,3 Case series have suggested that certain characteristics, such as having divorced parents, lower socioeconomic class, poor adherence to treatment, and pre-existent severe asthma, may enhance the risk of SAA.3–11 Most of these studies stem from outside Europe, and were published before 2005. Recent European studies on this topic are lacking. This is the first case-control study to assess risk factors for pediatric ICU (PICU) admission for pediatric SAA patients in the Netherlands. We aimed to identify characteristics that could predict the need for PICU admission for SAA among pediatric out-patients at Dutch hospitals. Early recognition could help to develop preventive strategies and improve efficacy of treatment for these children with asthma.
QUICK LOOK
Current knowledge
Severe acute asthma in children is associated with substantial morbidity and mortality. The most severe cases require intensive care admission. Risk factors for ICU admission include environmental exposures and previous hospitalization.
What this paper contributes to our knowledge
In a case-control study, risk factors for ICU admission in children with severe asthma included allergies, exposure to passive smoking, active smoking, unsanitized home(s), and a previous hospitalization for asthma. Admission for severe asthma occurred in 16% of patient without a previous asthma diagnosis.
Methods
Study Population
In a retrospective case-control study, we obtained data on asthmatic children who did and did not develop SAA warranting PICU admission. Data were collected in a standardized manner in 4 out-patient pediatric clinics in the southwestern region of the Netherlands (Amphia Hospital, Breda; Maasstad Hospital, Rotterdam; Erasmus Medical Center-Sophia Children's Hospital, Rotterdam; and Reinier de Graaf Hospital, Delft). As a retrospective case-control study, the medical ethics committee waived the need for informed consent. Children in this region experiencing SAA and who require intensive care treatment are transferred to the PICU of the Erasmus Medical Center-Sophia Children's Hospital. The admission criteria to the PICU were clinically predefined and did not change during the study period. They included respiratory failure and imminent exhaustion, no clinical benefit after continuous nebulization of bronchodilator drugs, and the need for intravenous administration of salbutamol (which is an indication for PICU admission according to Dutch guidelines) and/or the need for mechanical ventilation. All children with doctor-diagnosed asthma treated as out-patients by pediatricians or pediatric pulmonologists before PICU admission because of SAA from January 1994 to October 2006 were eligible as cases in this study. Children who had not been diagnosed with asthma prior to their PICU admission and presented with their first asthmatic attack were excluded from the study, because defining controls would not be feasible. Hence, this study focused on the population of children diagnosed with asthma and being treated by pediatricians or pediatric pulmonologists in out-patient clinics from 3 general hospitals and one university hospital. Controls were patients from the same participating referring hospitals, diagnosed with asthma and treated on an out-patient basis by pediatricians or pediatric pulmonologists, and who never needed PICU admission for any reason.
Matching Criteria
Each case patient was matched to at least one control according to all the following criteria:
Sex
Age: ± 1 year for under 5-year-olds and ± 2 years for older children
The same hospital, if possible; if not, the nearest hospital; however, controls for cases from a university hospital were always sought from the Erasmus Medical Center-Sophia Children's hospital.
Period elapsed since the diagnosis of asthma, distinguished by duration of up to 1 year, 1–4 years, 4–7 years, 7–10 years, or > 10 years
Potential Risk Factors
Table 1 shows the candidate risk factors for PICU admission, derived from previous studies.3–11 All these variables were extracted from the medical records. Data from the period prior to the PICU admission were considered as the baseline values for variables that could potentially change over time (eg, smoking, pets, hay fever). This period was, on average, a few months before the admission, at a time when there were no signs or symptoms of the severe asthma exacerbation. If possible, the last documented out-patient visit before PICU admission was chosen. For variables that were not documented during the last out-patient visit, the most recent documentation prior to the PICU admission was used.
Candidate Risk Factors for Pediatric ICU Admission
The variable “eczema” was an exception: this was scored with “yes” if it had occurred at any point prior to the PICU admission. The variable “sanitization of the home” was also an exception, because a house was scored as “sanitized” if the medical record reported that a specialized asthma nurse had visited and advised the family at any point in time before PICU admission, and specific measures for sanitization (eg, anti-dust-mite blankets applied) had been taken. A house was scored as “non-sanitized” if, for example, there were carpets, curtains, or many cuddly toys in the child's bedroom. The ethnicity of the child was based on the ethnic origins of the parents and classified as white or non-white. Social-economic status was derived from both parents' occupations at baseline. Smoking was scored as “passive” when a person who lived in the same house as the child smoked inside the house. “Quality of the home” was a dichotomous variable, rating the house as either old or new. The variable “allergy” referred to allergies for several antigens, excluding pollen, such as allergies for dust mites, animals, and food.
These data were retrieved from the medical record based on positive radioallergosorbent tests (RAST) and/or skin allergy tests. Hay fever was extracted from the medical record and rated separately from the other allergies. Asthma severity levels were based on the amounts and types of medication at baseline. It was always scored as severe if the child had been readmitted to the PICU for asthma. The variables “non-compliance” and “undertreatment” were scored with “yes” only if the physician had specifically noted this in the medical record. “Co-morbidity” was a dichotomous variable, and included every possible medical condition besides asthma from birth till PICU admission. Asthma medication was based on the prescription at the last regular out-patient visit prior to the PICU-admission. The use of inhaled corticosteroids was determined in a dose-dependent (daily amount) and dose-independent (yes or no) fashion.
Statistical Analysis
Variables for which no more than 30% of data were missing were compared between cases and controls using univariate and multivariate analyses. Accounting for matching, the Mantel-Haenszel statistic was used for the univariate analyses of qualitative categorical variables to identify variables associated with an SAA and the need of PICU hospitalization. For the continuous dose-dependent non-parametric variable “prescribed corticosteroids” we used the Mann-Whitney U test. For all the univariate analyses we used statistics software (SPSS 18.0, SPSS, Chicago, Illinois). The significant risk factors for SAA were identified with conditional logistic regression analysis using statistics software (SAS 9.2, SAS Institute, Cary, North Carolina). A P value of < .05 (2-sided) was considered statistically significant. The indices of the goodness of fit −2 log likelihood, Akaike information criterion, and Schwarz Bayesian criterion were obtained using statistics software (SAS 9.2, SAS Institute, Cary, North Carolina).
Results
Study Population
From January 1994 to October 2006, a total of 139 children were admitted to the PICU because of SAA. This group had a male preponderance (61.9%). The median (range) age at PICU admission was 4.3 (0.6–16.5) years. Seventeen children were ventilated during their PICU admission, 3 of whom developed a pneumothorax.
Of these 139 PICU admissions for SAA over the study period, 78 met the inclusion criteria for our study (Figure). Twenty-two cases (16%) were excluded for not receiving asthma treatment prior to the PICU admission. Thirty-four children (24%) were excluded as they were under the care of a general physician. Three cases were excluded for incomplete data. Finally, 9 cases could not be matched with controls and thus were excluded. This resulted in a study population of 66 cases. This study population also had a male preponderance (68.2%). The median (range) age at PICU admission was 5.2 (0.6–16.5) years. Each of these cases was matched with 1 to 4 controls per case, resulting in a total of 164 controls. The case group (66 patients) did not differ from the total PICU admission group (139 patients) with respect to the patient characteristics variables represented in Table 2, except for the variable “earlier hospitalization for asthma.” A possible explanation is that children who are admitted to the hospital for asthma are subsequently treated more often by a pediatrician or pediatric pulmonologist following their hospital admission.
Inclusion flow-chart. PICU = pediatric ICU. SAA = severe acute asthma.
Outcomes of Univariate Analysis
Outcomes Case-Control Study
Fourteen variables for which < 30% data were missing were analyzed in these 66 cases and 164 controls (see Table 2). The variable “comorbidity” was not further subdivided and was not a significant factor in the analysis.
Multivariate Analysis
All but one of the variables presented in Table 2 were statistically significantly linked to PICU hospitalization. The only exception was prescribed corticosteroids in a dose-independent fashion. The conditional logistic regression analysis included all significant variables except for severe asthma. This was excluded because it was highly correlated to the dosages of inhaled corticosteroids and therefore a dependent variable. This multivariate analysis showed that only (active or passive) smoking, allergies, earlier hospitalization for asthma, and non-sanitized home proved to be significant independent risk factors (Table 3). The chi-square value was 42.09 (P < .001). The indices of the goodness of fit (−2 log likelihood, Akaike information criterion, and Schwarz Bayesian criterion) were without covariates 73.69 for all of them. Including the covariates these values changed into 31.60, 39.60, and 46.83, respectively. In terms of reduction of uncertainty the percentages equal to 57.11%, 46.26%, and 36.45%, respectively, which were considered substantial.
Significant Variables for Pediatric ICU Admission After Conditional Logistic Regression Analysis
Discussion
This study demonstrates in a case-control fashion that non-sanitized homes, allergies, (active or passive) smoking, and earlier hospitalization for asthma were independently associated with an increased risk for PICU admissions for SAA in children (see Table 3). These characteristics have been reported by others as well.3,4,6,9 An important observation was that non-sanitized homes and (active or passive) smoking were not uncommon, in spite of the fact that patients, parents, and physicians may be assumed to be aware that these constitute risk factors for worsening of asthma and for SAA.
A remarkable and unanticipated finding of this study is that 16% of children admitted for SAA had no prior diagnosis of asthma. The other 117 children developed SAA in spite of treatment by general physicians (24%) or by pediatricians (36%) and pediatric pulmonologists (20%) (see the Figure). This indicates that there is a need for the implementation of more effective preventive measures, not only in primary care but also in secondary and tertiary care.
Ethnicity (non-white origin) and a dysfunctional family system have been reported as risk factors for potentially fatal asthma,3 but this was not confirmed in our study. Use of medication other than inhaled corticosteroids has also been significantly associated with PICU admission, according to the literature. This finding has also been reported in the adult literature. The systematic review by Alvarez et al12 found that increased use of β2 agonists via inhaled or nebulized administration was associated with a greater risk of near fatal asthma and fatal asthma. The lack of association in the present study could be explained by the strong correlation between asthma severity and administration of bronchodilators such as β2 agonists and anticholinergic drugs. Indeed, after correction for asthma severity, these variables were no longer significant risk factors. An adult study found that the use of high dosages of inhaled β2 agonists was associated with a greater risk of ICU admission for asthma, even after controlling for asthma severity.13
Most patients (cases and controls) used inhaled corticosteroids, which is to be expected, according to the asthma guidelines. In our study, no difference between the use of inhaled corticosteroids between the 2 groups, measured in a dose-independent fashion, was observed (82% in cases and 81% in controls), whereas the average dose in cases was only slightly higher (100 μg/d) than in controls in the univariate analysis. This small difference in dosage lost significance in the multivariate analysis, and might imply that in some cases the average dose of inhaled corticosteroids was too low and that cases were therefore undertreated, receiving insufficient anti-inflammatory treatment. In the absence of reliable information on actual adherence to treatment in both groups, this explanation remains speculative.
In their systematic review, Alvarez et al12 reported that the use of inhaled corticosteroids measured in a dose-independent fashion did not significantly increase nor decrease the risk of near fatal asthma and fatal asthma in adults. However, they found that the use of corticosteroids in a dose-dependent fashion showed a trend toward a protective effect against fatal asthma.12 Another study in adults reported that the use of inhaled corticosteroids is associated with a reduced risk of ICU admission among adults hospitalized with asthma.13 These findings are in contrast to our results. Differences between the routine patient care, respective guidelines for treatment of asthma, age, and other characteristics of the patients complicate the comparisons between these studies.
In the study of Belessis et al,4 the variables that include smoking, pets, hay fever, eczema, and carpets were not significant risk factors for PICU admission in children with SAA. One possible explanation may be that the cases and controls in their study were matched on prevalence of asthma, eczema, or hay fever in first-degree relatives. The study of Lyell et al7 indicated that asthmatic children with near fatal asthma were more likely to have hay fever than asthmatic children admitted to the emergency department on one occasion only; passive smoking did not differ between those 2 groups. Smoking and atopy did not increase the risk of near fatal asthma and fatal asthma in studies in adults.12 Earlier PICU admission due to SAA has been shown to be a risk factor for readmission because of SAA.3 In our study, 8 children from the PICU admission group had been readmitted repeatedly. Because this was an exclusion criterion for controls in our study, it could not be identified as a risk factor, but previous (medium care) hospital admission was indeed recognized as a risk factor. These findings were also observed in adult studies.12 Previous studies have identified male sex as a risk factor for PICU admission after acute asthma exacerbation.7 Our study indeed showed a male preponderance, but this may simply reflect the higher prevalence of asthma among boys. One study suggested that asthmatic children admitted to a PICU were more likely to be older than asthmatic children who had been admitted to the emergency department on one occasion only.7 In the present study the median age at the time of PICU admission was relatively low (5.2 years); as it was a matching criterion, it was not compared between cases and controls.
A limitation of the study is that we used a retrospective study design, with information collected from the medical records. For this reason, several variables could not be analyzed reliably, such as quality of the home, psychosocial problems of the child or the family, non-compliance with therapy, (possible) undertreatment, poor inhalation technique, and instructions by a specialized asthma nurse. Data for these variables were either missing or had not been recorded in the medical record. Nevertheless, the available data provide relevant information about asthmatic children who developed severe asthmatic attacks resulting in PICU admission, in spite of treatment by general pediatricians or pediatric pulmonologists.
Conclusions
Allergies, (active or passive) smoking, earlier hospitalization for asthma, and non-sanitized homes were significant risk factors for PICU admission for pediatric patients with SAA. These risk factors were present in spite of treatment by pediatricians or pediatric pulmonologists. Thus, there is room to improve current strategies to prevent SAA. Physicians, parents, and patients should be aware of these risk factors, which are preventable to some extent. Earlier identification of children at risk for PICU admission may result in better asthma management and thus in a reduction of PICU admission for SAA. Future prospective standardized studies are needed to validate these results and especially to determine the role of non-adherence.
Acknowledgments
We thank Ko Hagoort and Dr Tonya White of the Erasmus Medical Center-Sophia, Rotterdam, the Netherlands, for their editorial assistance. We also thank Dr Hugo Duivenvoorden of the Erasmus Medical Center-Sophia, Rotterdam, the Netherlands, for his assistance in the statistical analyses.
Footnotes
- Correspondence: Gerbrich E van den Bosch MD, Erasmus Medical Center-Sophia, PO Box 2060, 3000 CB Rotterdam, The Netherlands. E-mail: g.vandenbosch{at}erasmusmc.nl.
The authors have disclosed no conflicts of interest.
See the Related Editorial on Page 1526
- Copyright © 2012 by Daedalus Enterprises Inc.
