Predicting the Need for Hospitalization in Children with Acute Asthma

doi:10.1378/chest.98.6.1355

Chest

Volume 98, Issue 6, December 1990, Pages 1355-1361

https://doi.org/10.1378/chest.98.6.1355 Get rights and content

In an attempt to identify factors which influence the decision of physicians to admit patients with acute asthma to the hospital, we studied prospectively 200 children (age 5.6±3.1 years, mean±SD) presenting to our emergency room with acute asthma. The children were assessed on arrival, and on disposition from the Emergency Room by one of the investigators. After obtaining historic data, a clinical score was assigned, and oxygen saturation and pulmonary function were measured. Of the 134 (67 percent) children who were discharged home from the Emergency Room, five returned within seven days and one was subsequently admitted. The clinical score on disposition was the sole variable found to best predict the decision for hospitalization (sensitivity 73 percent, specificity 95 percent). Of the variables obtained at presentation, the resulting decision tree found the clinical score to predict the decision for hospitalization (sensitivity 79 percent, specificity 75 percent). When the individual components of the clinical score were analyzed, the degree of dyspnea, as assessed by the investigator, was chosen as the rule to predict the hospitalization decision (sensitivity 88 percent, specificity 71 percent). We conclude that the decision with respect to the need for hospitalization in acute childhood asthma, is in practice based mainly on careful clinical evaluation. Pulmonary function and SaO₂ measurements, although helpful adjuncts in the assessment of acute asthma, do not appear to contribute to the identification of patients who need hospital admission.

Section snippets

Methods

Our study population was comprised of 200 children who presented to the ER of the Hospital for Sick Children, Toronto, between September and December 1988, for the treatment of acute asthma when one of the investigators was available. Children with acute bronchiolitis or with complicating pulmonary or cardiac disorders were excluded from the study.

On arrival to the ER, the following historic data were obtained from the parents of each subject: age, sex, duration of present episode, number of

Results

Of the 200 patients in the study, 66 (33 percent) were hospitalized, and 134 (67 percent) were discharged home. Five of the discharged patients returned to the ER within seven days, and one was subsequently hospitalized. All patients were treated with nebulized albuterol, and in addition, 8 percent received intravenous theophylline, and 8 percent received systemic corticosteroids. The mean age of all patients was 5.6±3.1 years (range 0.4 to 16 years). No statistically significant age difference

Discussion

The objective of the present study was to determine the value of clinical measurements and common, noninvasive laboratory variables both alone and in various combinations in predicting physicians’ decisions to admit children with acute asthma to the hospital. Although the hospitalized and the discharged groups of patients had statistically significant differences in most of the variables measured, the clinical score, based on standard physical examination, was found to be the most effective in

ACKNOWLEDGMENT

The authors thank J. Chay for secretarial assistance.

References (31)

S Lulla et al.
Emergency management of asthma in children
J Pediatr
(1980)
AF Arnold et al.
Acute severe asthma: factors that influence hospital referral by the general practitioner and self-referral by the patients
Br J Dis Chest
(1983)
SG Kelsen et al.
Emergency room assessment and treatment of patients with acute asthma: adequacy of the conventional approach
Am J Med
(1978)
AS Rebuck et al.
Assessment and management of severe asthma
Am J Med
(1971)
ER McFadden
Clinical physiologic correlates in asthma
J Allergy Clin Immunol
(1986)
EA Mitchell
International trends in hospital admissions rates for asthma
Arch Dis Child
(1985)
HR Anderson
Increase in hospitalization in childhood asthma
Arch Dis Child
(1978)
RT Jackson et al.
Mortality from asthma: a new epidemic in New Zealand
Br Med J
(1982)
RM Sly
Increases in deaths from asthma
Ann Allergy
(1984)
D Stableforth
Death from asthma
Thorax
(1983)

MR Sears et al.

Deaths from asthma in New Zealand

Arch Dis Child

(1986)

DR Ownby et al.

Attempting to predict hospital admission in acute asthma

Am J Dis Child

(1984)

MD Baker

Pitfalls in the use of clinical asthma scoring

Am J Dis Child

(1988)

MA Fischl et al.

An index predicting relapse and need for hospitalization in patients with acute bronchial asthma

N Engl J Med

(1981)

CC Rose et al.

Performance of an index predicting the response of patients with acute bronchial asthma to intensive emergency department treatment

N Engl J Med

(1984)

Cited by (77)

Predicting hospitalization of pediatric asthma patients in emergency departments using machine learning
2021, International Journal of Medical Informatics
Citation Excerpt :
As hospitalization contributes the largest proportion (∼65 %) of asthma care costs, early identification of the need for hospitalization is important [8,9], as it may reduce ED boarding time, facilitate more effective use of ED resources, or enable provision of life-threatening care to patients with a high risk for hospitalization. Most of the previous studies on developing hospitalization decision models have focused on the prediction of safe discharge from hospitals, for which they have usually used clinical scores or a conventional statistical model, such as logistic regression [10–18]. Despite the ease of calculation and implementation in real-world setting, these conventional approaches (e.g., scoring model, conventional statistical model, etc.) typically lack the strength to provide statistical accuracy [19,20].
The timely identification of patients for hospitalization in emergency departments (EDs) can facilitate efficient use of hospital resources. Machine learning can help the early prediction of ED disposition; however, application of machine learning models requires both computer science skills and domain knowledge. This presents a barrier for those who want to apply machine learning to real-world settings.
The objective of this study was to construct a competitive predictive model with a minimal amount of human effort to facilitate decisions regarding hospitalization of patients.
This study used the electronic health record data from five EDs in a single healthcare system, including an academic urban children’s hospital ED, from January 2009 to December 2013. We constructed two machine learning models by using automated machine learning algorithm (autoML) which allows non-experts to use machine learning model: one with data only available at ED triage, the other adding information available one hour into the ED visit. Random forest and logistic regression were employed as bench-marking models. The ratio of the training dataset to the test dataset was 8:2, and the area under the receiver operating characteristic curve (AUC), accuracy, and F1 were calculated to assess the quality of the models.
Of the 9,069 ED visits analyzed, the study population was made up of males (62.7 %), median (IQR) age was 6 (4, 10) years, and public insurance coverage (66.0 %). The majority had an Emergency Severity Index score of 3 (52.9 %). The prevalence of hospitalization was 22.5 %. The AUCs were 0.914 and 0.942. AUCs were 0.831 and 0.886 for random forests, and 0.795 and 0.823 for logistic regression. Among the predictors, an outcome at a prior visit, ESI level, time to first medication, and time to triage were the most important features for the prediction of the need for hospitalization.
In comparison with the conventional approaches, the use of autoML improved the predictive ability for the need for hospitalization. The findings can optimize ED management, hospital-level resource utilization and improve quality. Furthermore, this approach can support the design of a more effective patient ED flow for pediatric asthma care.
Repeated dyspnea score and percent FEV<inf>1</inf> are modest predictors of hospitalization/relapse in patients with acute asthma exacerbation
2014, Respiratory Medicine
Citation Excerpt :
In several studies in which peak flow or %FEV1 was performed as part of a study protocol, there was no significant difference in either initial or subsequent performance between those that were discharged successfully, and those hospitalized or relapsed [5,28,29]. There are fewer published studies on the use of dyspnea scores in predicting outcome in asthma exacerbation [9]. The Modified Borg Dyspnea Scale has its proponents [18–21] and detractors [30,31].
(1) Compare ideal cut-off points for DS and %FEV₁ at 1 and 3 h to predict hospitalization/relapse in subjects with moderate to severe asthma exacerbation (2) Develop a multivariate regression model using DS, %FEV₁, demographic, and clinical variables to predict hospitalization/relapse.
Subjects with acute exacerbation of asthma (FEV₁ <50% predicted following 30 min of standardized treatment: 5 mg nebulized albuterol; 0.5–1.5 mg nebulized ipratropium; and 50 mg oral prednisone) were eligible. All subjects had %FEV₁ and DS obtained at baseline and hourly for 3 h. Using hospitalization/relapse as the outcome of interest; we compared the area under the receiveroperator curves (AUC) between the 1 and 3 h DS and %FEV₁ measurements, and the AUC for the change in DS and %FEV₁ between baseline and hour-3. We determined ideal cut-points for %FEV₁ and DS to maximize sensitivity and specificity. Sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios (LR) were compared between %FEV₁ and DS. We developed a multivariate regression model examining the association of specific demographic and clinical variables to hospitalization/relapse.
142 patients were included for analysis. The AUC was greatest for the 3-h DS (0.721), followed by the 3-h %FEV₁ (0.669). Optimum cut-off values were a DS of 2, and an FEV₁ of 42%. These were associated with a +LR for the composite outcome of 3.06 and 2.48 respectively. Logistic regression showed baseline DS, 3-h DS, change in DS, and oxygen use at hour 3 were all associated with the composite outcome.
The 3-h score for %FEV₁ and DS performed better than scores at any other time point and better than either parameter over time. The 3-h DS had the greatest association with the composite outcome. Neither test was a strong enough predictor to be used solely for this purpose.
Predicting hospitalization in children with acute asthma
2013, Journal of Emergency Medicine
Acute asthma is one of the most common medical emergencies in children. Appropriate assessment/treatment and early identification of factors that predict hospitalization are critical for the effective utilization of emergency services.
To identify risk factors that predict hospitalization and to compare the concordance of the Modified Pulmonary Index Score (MPIS) with the Global Initiative for Asthma (GINA) guideline criteria in terms of attack severity.
The study population was composed of children aged 5–18 years who presented to the Emergency Departments (ED) of the tertiary reference centers of the country within a period of 3 months. Patients were evaluated at the initial presentation and the 1^st and 4^th hours.
Of the 304 patients (median age: 8.0 years [interquartile range: 6.5–9.7]), 51.3% and 19.4% required oral corticosteroids (OCS) and hospitalization, respectively. Attack severity and MPIS were found as predicting factors for hospitalization, but none of the demographic characteristics collected predicted OCS use or hospitalization. Hospitalization status at the 1^st hour with moderate/severe attack severity showed a sensitivity of 44.1%, specificity of 82.9%, positive predictive value of 38.2%, and negative predictive value of 86.0%; for MPIS ≥ 5, these values were 42.4%, 85.3%, 41.0%, and 86.0%, respectively. Concordance in prediction of hospitalization between the MPIS and the GINA guideline was found to be moderate at the 1^st hour (κ = 0.577).
Attack severity is a predictive factor for hospitalization in children with acute asthma. Determining attack severity with MPIS and a cut-off value ≥ 5 at the 1^st hour may help physicians in EDs. Having fewer variables and the ability to calculate a numeric value with MPIS makes it an easy and useful tool in clinical practice.
Capnographic Monitoring in Respiratory Emergencies
2009, Clinical Pediatric Emergency Medicine
Children with respiratory diseases present commonly to the emergency department. As a continuous, dynamic measure of the ventilatory status, capnography can provide valuable information in the assessment and management of these patients. After a review of the relevant physiology and technology of carbon dioxide monitoring, clinical applications for the use of capnography in patients with respiratory illnesses are discussed. Characteristic waveforms are provided, and their interpretation and clinical significance are discussed. A focus on the current literature investigating the noninvasive monitoring of patients with obstructive lung disease is included. Practical tips for successfully using capnography are also presented.
Continuous noninvasive measurement of pulsus paradoxus complements medical decision making in assessment of acute asthma severity
2006, Chest
Pulsus paradoxus (PP) is a pathophysiologic parameter that is indicative of asthma severity. The ability of PP to categorize acutely asthmatic patients in accordance with the earlier National Asthma Education and Prevention Program (NAEPP) expert panel report 1 guidelines was determined.
An arterial tonometric BP monitor, which was interfaced to an analog-digital converter, executed a periodic amplitude analysis algorithm, which computed PP in real time. The PP measurement was compared to the criterion standard of emergency physicians in determining the hospital admission vs hospital discharge disposition following the NAEPP standardized treatment. Receiver operating characteristics (ROCs) were calculated, and the PP threshold, which maximized sensitivity and specificity, was identified. In a separate laboratory investigation, PP was induced in a healthy volunteer by inspiration through a fixed resistance. Plethysmographic waveform changes, induced by PP, were measured by a second analog-to-digital converter that was connected to a pulse oximeter.
A total of 79 patients were enrolled in the study, of whom 63 met a priori inclusion criteria and had uninterrupted data acquisition. The mean PP for patients who were appropriately discharged from the hospital was 9.1 mm Hg (95% confidence interval [CI], 7.3 to 10.9 mm Hg) and differed from the PP of 17.6 mm Hg (95% CI, 13.5 to 21.8; p < 0.001) for patients admitted to the hospital/relapsed. The sensitivity and specificity for physician disposition were 0.83 and 0.89, respectively, and for PP values were 0.78 and 0.78, respectively. The Wilcoxon area under the ROC curve was 0.82 (95% CI, 0.64 to 0.99) following treatment. The risk ratio was 5.32 for hospital admission among patients with a PP of > 11.3 mm Hg. Changes in the photoplethysmography peak height were correlated to PP from the BP monitor by a regression line with a slope of 0.01V/mm Hg.
Continuous PP can aid in determining disposition among emergency department (ED) patients with acute asthma. ED physicians equipped with a PP monitor would be able to objectify the work of breathing and would more closely adhere to NAEPP guidelines. The possibility that a PP detection algorithm could reside in a pulse oximeter warrants further investigation.
A Modified Pulmonary Index Score with predictive value for pediatric asthma exacerbations
2005, Annals of Allergy, Asthma and Immunology
Several clinical asthma scores have been derived from combinations of physical findings in pediatric asthmatic patients.
To test the reproducibility and validity of one such score, the Modified Pulmonary Index Score (MPIS), and to evaluate its predictive value in children hospitalized for asthma.
In the MPIS, 6 categories are evaluated: oxygen saturation, accessory muscle use, inspiratory to expiratory flow ratio, degree of wheezing, heart rate, and respiratory rate. For each of these 6 measurements or observations, a score of 0 to 3 is assigned. To evaluate the reproducibility of the MPIS, inpatients with status asthmaticus were examined by an attending physician, nurse, and respiratory therapist who were blinded to the other observers' scores. To evaluate the validity of the MPIS as a scale of severity of illness in asthmatic patients, the score at admission was compared with selected outcomes in the same patients.
A total of 30 patients participated in this study (mean ± SD age, 7.6 ± 5.5 years). Our finding revealed that the MPIS is highly reproducible with a high degree of interrater reliability across caregiver groups (physician to nurse: r = 0.98; 95% confidence interval [CI], >0.96; physician to respiratory therapist: r = 0.95; 95% CI, >0.92; nurse to respiratory therapist: r = 0.94; 95% CI, >0.90). The admission MPIS positively correlated with intensive care unit admission (P < .001), days of continuous albuterol therapy (P = .002), days of supplemental oxygen (P = .002), and length of hospital stay (P = .004).
The MPIS is a highly reproducible and valid indicator of severity of illness in children with asthma. To our knowledge, this is the first pediatric clinical asthma score demonstrated to be reproducible across groups of health care professionals who treat pediatric patients with asthma.

View all citing articles on Scopus

Presented in part at the 55th Annual Scientific Assembly, American College of Chest Physicians, Boston, October 29–November 2, 1989.

Manuscript received February 12; revision accepted May 15.

View full text

Chest

Clinical InvestigationsPredicting the Need for Hospitalization in Children with Acute Asthma

Section snippets

Methods

Results

Discussion

ACKNOWLEDGMENT

J Pediatr

Br J Dis Chest

Am J Med

Am J Med

J Allergy Clin Immunol

International trends in hospital admissions rates for asthma

Arch Dis Child

Increase in hospitalization in childhood asthma

Arch Dis Child

Mortality from asthma: a new epidemic in New Zealand

Br Med J

Increases in deaths from asthma

Ann Allergy

Death from asthma

Thorax

Deaths from asthma in New Zealand

Arch Dis Child

Attempting to predict hospital admission in acute asthma

Am J Dis Child

Pitfalls in the use of clinical asthma scoring

Am J Dis Child

An index predicting relapse and need for hospitalization in patients with acute bronchial asthma

N Engl J Med

Performance of an index predicting the response of patients with acute bronchial asthma to intensive emergency department treatment

N Engl J Med

Clinical Investigations
Predicting the Need for Hospitalization in Children with Acute Asthma