Reducing Hospitalizations for Childhood Asthma: Insights from Systematic Reviews

Asthma is the most common chronic pediatric disease in industrialized countries, affecting 8-10% of children1,2 and is a leading cause of pediatric hospitalizations with around 20% of children readmitted within 1 year.3 ,4

A hospital admission for an asthma exacerbation suggests suboptimal control and is a strong independent predictor of future hospital admissions, with a small associated increase in asthma mortality.5 In addition to previous hospital admissions, children with repeat presentations to the emergency department emergency, uncontrolled asthma, and those from lower socioeconomic families are also at risk of future hospitalizations for asthma.3

Hospitalizations present a psychosocial burden for the child and family, and an economic burden for the health system. The 2017 Lancet Asthma Commission calls for treating each asthma exacerbation as a "lung attack," similar to a "heart attack," and for a holistic assessment of a child’s asthma control, including comorbidities, adherence, environmental changes and psychosocial factors, to improve asthma control and prevent unplanned hospital admission in the future.6 

To date, systematic reviews for the primary prevention of hospitalization in (low-risk) children with asthma support the use of home educational interventions,7 corticosteroids8, and self-management action plans.9

A meta-analysis examining the effect of inpatient interventions such as standardized clinical pathways, electronic decision support, family education, self-care plans, reserved follow-up, and available medications on subsequent hospital use,10 found that no Hospitalization affected admissions within 30 days of the index hospitalization, but multimodal interventions reduced admissions over 12 months.10

Multimodal intervention, including hospital and home interventions, resulted in an 80% reduction in admissions within 30 days of the index hospitalization, suggesting that interventions isolated to hospital care are insufficient to reduce hospitalizations .eleven 

There is a lack of systematic reviews of interventions designed to prevent asthma hospitalization for children at risk (i.e., those with prior hospitalizations, multiple emergency presentations, uncontrolled asthma, and/or of lower socioeconomic status) across settings. care, which is home, primary care and hospital.

Knowledge of which interventions are most effective and from which care setting can inform efforts to reduce hospitalizations. Therefore, we aim to summarize the evidence regarding the clinical and cost effectiveness of interventions to reduce asthma hospitalizations and improve asthma outcomes (use of rescue oral corticosteroids, school absenteeism, quality of life) in children at risk, considering the areas of care.

> Search and Study Eligibility Criteria
In collaboration with a physician librarian, a strategic search was developed using search terms asthma, readmission, hospitalization, and children. Medline, Embase, Pubmed and Cochrane Library were searched (January 1, 2002 to April 9, 2020) and a protocol was registered in PROSERPO (CRD42019122760).

We limited the search to abstracts available in English and articles published after 2002 because this was when the Global Asthma Initiative emphasized asthma as a chronic inflammatory airway disorder, highlighting the importance of chronic disease management (i.e. education, self-sufficiency and control).6

The inclusion criteria were: 

1. randomized controlled trial study design; 
2. children ≤18 years old; 
3. at least 50% of participants had been previously hospitalized; 
4.  outcome of reported hospitalizations; 
5. any intervention aimed at improving the management of asthma symptoms, such as: educational programs, home care, telemonitoring, multimodal interventions, any type of medication; 
6. the control group may include treatment as usual, wait list, placebo or other intervention; 
7. any environment; and 
8. minimum of 10 per group.

Because the authors wanted to identify interventions that would help break the cycle of hospitalizations, only study populations where at least half of the children had been hospitalized were included.

If a lower percentage was selected, the intervention would be aimed at primary prevention and if a higher percentage was chosen, the interventions tended to focus primarily on hospital care. Studies were excluded if children only visited emergency departments or had not been previously admitted.

> Study selection and data extraction
Two reviewers (LS and KC) divided all eligible abstracts and screened them by title and abstract to exclude irrelevant articles. The full texts of all potentially relevant articles were retrieved and verified for final inclusion.

The two reviewers collated a 10% random sample of each other’s first and second screenings to estimate interrater reliability. Cohen’s kappa statistic was calculated as 0.41 and 1.0, respectively. Discrepancies were resolved through review and consensus.

The reviewers independently extracted study characteristics and outcome data using a standardized data collection form. Outcomes included hospitalizations, quality of life, cost-effectiveness, rescue oral corticosteroid use, and school attendance.

The quality of each study was assessed using a Cochrane risk of bias tool that considered: 

1. the sequence of random result generation, 
2. allocation concealment, 
3. blinding, 
4. handling of missing data, 
5. .the selection of reported results, and (vi) other risks of bias.12

> Synthesis of results
The objective was to synthesize the results through a meta-analysis. However, due to the heterogeneity of interventions in the included studies, duration of follow-up and reporting of outcomes (mean number of hospitalizations versus hospitalization as a dichotomous outcome), a meta-analysis of outcomes was not performed.

Our initial search identified 12,606 unique articles. After excluding duplicates, 9535 abstracts were filtered by titles and abstracts alone. 314 full-text articles were reviewed and of these, 12 randomized controlled trials, providing a total of 2719 participants, met the inclusion criteria. 

> Interventions
Interventions were delivered over a period of 3 to 18 months in children with asthma aged 2 to 16 years, most of whom had previously been admitted to hospital. Eight studies provided an education-based program, either in the home, hospital, or 
primary care clinic, 14–16,18–20,22,23 which often included an additional element such as reduction of environmental allergens in the home. home,14 virtual or in-person follow-up,15,19,22 training of primary care physicians in asthma management16 or remote clinical support.19,24

One study provided feedback to the primary care physician about their patient’s symptoms and included guideline-based recommendations and recommendations for changes in treatment.20 Another introduced multisystemic therapy healthcare, which consists of cognitive behavioral and parenting therapies that address the asthma-related problems problematic behaviors in the child, family, school, and medical team.21

One study provided a nutritional supplement of carnitine that is believed to have an anti-inflammatory effect by inhibiting leukotriene synthesis,13 and one provided high doses of inhaled fluticasone at the onset of respiratory illness.17

All interventions involved parents or caregivers in various ways, whether monitoring their children, recording asthma symptoms, delivering medication, or participating in the education program. Some healthcare professional, often a nursing specialist, 17–19, 21, 22, 24 or non-clinical staff were trained to deliver the intervention.13–16

Three trials used usual care as a comparison group, 14, 23, 24 and one used a placebo. 17 Others controlled interaction with health professionals by providing brochures containing educational materials, non-directly supportive family counseling, or minimal contact with a nurse.15, 18, 21, 22

Canino (2016)16 used the same intervention (a culturally adapted family-based education program) delivered in a previous trial15 as the control arm and offered additional training to health professionals in the intervention arm to explore the impact to train primary care providers in asthma symptom management.

> Risk of bias
Of the 12 studies, three had a low overall risk of bias, 17, 20, 22 and nine had some concerns.13–16,18,19,21,23,24 Blinding of participants is difficult in In behavioral intervention trials, one trial was managed to achieve this;22 however, they compared two different education programs and the participants were simply not aware of the differences between the two programs.

In addition to the two medication interventions, 13,17 participants in all other studies were not blinded to their treatment assignment. To minimize the risk of bias, assessors should be blinded, and this was carried out in five behavioral trials18, 20–23 and one medication trial.17

All studies were randomized, but six failed to describe how the random sequence was generated.14,15,19,21,23,24 Two studies used analysis of available cases despite having 29-30% dropouts,13,14 and another two with 12% dropouts.18,19 Three of the studies used analysis of available cases, but only had a drop from 1% to 2% dropout,15,20,24 one study used multiple imputation to account for high dropouts (17-20%),16 the remaining four used the gold standard of intention-to-treat analysis.17,21–23 

Seven studies measured the primary outcome, hospitalizations, through patient diaries, 13–16, 18, 20, 22 four used hospital records, 17,21,23,24 and one was unclear.19 Hospitalizations were reported as either as dichotomous (number of hospitalized patients ≥1) or continuous (as the mean number of hospitalizations).

The former approach does not capture patients who had multiple hospitalizations and the latter could be biased by a small number of participants who experienced a high number of hospitalizations, unless researchers present results in both formats as Kattan et al. and Ng and collaborators (the former through personal correspondence).20,22 The duration of the study also influences the number of hospitalizations with studies ranging from 3 to 18 months. 

> Primary outcome: Hospitalizations
Significant heterogeneity was detected, which could not be explained by the analysis, the type of intervention, the duration of the intervention or the age of the participants.

Effective interventions were multimodal with an educational approach component.14,15,18,22 In addition to asthma education, these studies reduced environmental allergens in the home14 and provided follow-up.14,15,18,22 Compared to standard asthma education, effective educational programs were tailored to the child’s asthma, culturally sensitive, interactive, and technology-assisted.15,18,22,25

The study that provided L-Carnitine supplementation was also effective in reducing hospitalizations at 6 months.13 The study with the largest sample size (n = 937) provided regular patient follow-up and feedback to primary care providers summarizing the child’s recent symptoms, medications and health services used combined with guideline-based recommendations for changes in treatment.20

However, no differences were found in hospitalizations during the 12-month follow-up, likely because 56% of providers reported obstacles in implementing the recommendations.20

> Secondary outcomes: oral corticosteroid use, medications, quality of life and cost-effectiveness
Few studies examined secondary outcomes. Three studies reported the mean number of courses of rescue oral corticosteroids used after hospitalization and three reported the number of children receiving rescue oral corticosteroids (dichotomous outcome, yes/no). These interventions included the use of fluticasone at the first sign of an upper respiratory tract infection17 and education.18, 24

The trial with two intervention arms of skilled nursing support or interactive voice response system with improved quality of life scores was cost-effective.24 No difference was detected between the intervention and control groups for the outcomes of number of school days lost. Exploring the heterogeneity detected in the mean number of rescue oral corticosteroid courses used and quality of life was problematic because only three studies reported these results and the type of analysis performed did not influence the degree of imprecision.

Taking into account the limitations of study quality, the present review found that interventions, which are multimodal and include individualized education with regular follow-up at home, primary care and hospital, may be effective in reducing the number of hospitalizations in at-risk children with asthma.

There was an unexpected lack of randomized controlled trials of the most commonly used medications in asthma, such as corticosteroids, leukotriene antagonists or biologics.

Most corticosteroids in the identified trials were excluded because participants were recruited from primary care or emergency clinics (and did not meet ’at risk’ criteria) or were not followed for a sufficient period of time to make hospitalization evaluated as an outcome for all participants.26–28

The article by Ducharme et al included in this review showed a reduction in the use of rescue oral corticosteroids, but not hospitalization using a short course of fluticasone at the onset of a respiratory illness.17

Studies exploring the effectiveness of biologics such as Benralizumab, an anti-interleukin-5 receptor antibody, were conducted in older adolescents and adults or used emergency visits as an outcome measure and are therefore not included in this study. this review.29 

There was one drug study that showed a benefit in reducing hospitalization.13 This study provided L-carnitine, which may have anti-inflammatory properties, to Egyptian children over a 6-month period. This study has not been replicated and it is uncertain if the results are translatable due to the small sample size and lack of ethnic diversity in the study.

Asthma education remains the cornerstone of asthma management in hospitals and the community. 30,31 However, the authors’ systematic review indicates that asthma education alone is insufficient in reducing hospitalizations, rescue oral corticosteroid use, or school absences. This suggests that the responsibility for managing asthma cannot fall solely on the child and his or her caregiver.

This review supports active modification of asthma risk factors, such as reducing environmental triggers, 14 treating underlying inflammation, 13,17 and regular follow-up15,18,22,25 in addition to education. 14, 15, 18, 22

Similarly, a meta-analysis of quality improvement interventions for hospitalized pediatric patients with asthma supports multimodal interventions and an active approach, such as hand-held medication delivery at discharge, in addition to education.10 

There is a strong argument for strengthening asthma care in the community to facilitate access, and where environmental and psychosocial vulnerabilities can be better addressed.32 This is supported by studies in this review, with hospitalizations reduced through education in the community. home that was culturally adapted and focused on reducing risk factors.14,15

Future research could examine the clinicalness and cost-effectiveness of such care, for example, through decentralization with hospital nurses or asthma educators in vulnerable communities to provide ongoing, individualized education.

Similarly, there is a need to explore the cost and impact of virtual care solutions, such as those based on home spirometry paired with patient report of symptom control on timely treatment. This specific multimodal approach is especially relevant for children with psychosocial risk factors such as being over-represented in asthma morbidity and mortality with deficiencies in the integration of care between hospital, primary care, education and community services.33 

There are strong points in this review. To our knowledge, this is the first systematic review of randomized controlled trials across the continuum of care for the prevention of asthma hospitalization in at-risk children.

The authors conducted an exhaustive investigation that included manual searching of bibliographies in relevant articles. Limiting articles to randomized controlled trials eliminates the bias associated with before-after or interrupted time series studies more commonly used to study quality improvement interventions.

There are limitations to this systematic review. Few studies have been published addressing the pertinent question of how to reduce asthma hospitalizations in at-risk children, thus highlighting the need for more work in this area. As a result, we had limited power to conduct a meta-analysis and affected the assessment of heterogeneity.

Follow-up time ranged from 3 to 18 months, precluding the identification of hospitalizations in some studies, which may be more closely associated with asthma control. Additionally, there was uncertainty around randomization methods and allocation concealment in some studies.

Most did not report results based on intention-to-treat analysis, and the various methods of reporting the primary endpoint (continuous vs. dichotomous) preclude the ability to combine all results in a meta-analysis. Secondary outcomes were rarely reported, limiting knowledge on the effectiveness of the interventions on oral corticosteroid use, quality of life, school attendance, and cost-effectiveness.

Evidence from clinical interventions to reduce asthma hospitalizations among at-risk children suggests that multimodal interventions, delivered in different care settings, may be effective in preventing future hospitalizations.

Interventions should be evaluated for cost-effectiveness.34 Effective interventions may require a combination of treatment targeting the underlying inflammation, methods to reduce asthma triggers and facilitate adherence, tailored to the child’s individualized phenotype and psychosocial profile.