Abstract

Background Severe asthma significantly impacts a minority of children with asthma, leading to frequent symptoms, hospitalisations and potential long-term health consequences. However, accurate global data on severe asthma epidemiology is lacking. This study aims to address this gap, providing data on severe asthma epidemiology, regional differences and associated comorbidities.

Methods We conducted a rigorous systematic review and meta-analysis following a registered protocol (PROSPERO CRD42023472845). We searched PubMed, Scopus and Web of Science for cohort or cross-sectional studies published since 2003, evaluating severe asthma incidence and prevalence in children. Study quality and risk of bias were assessed using STROBE guidelines.

Results Nine studies investigating European children with asthma (aged 5–18 years) were included in the meta-analysis. No significant publication bias was found. The overall severe asthma prevalence in children with asthma was 3% (95% CI 1–6; I2=99.9%; p<0.001), with no significant difference between males and females. Prevalence estimates varied significantly depending on the diagnostic criteria used (Global Initiative for Asthma: 6%; European Respiratory Society/American Thoracic Society: 1%; other: 3%). Because none of the examined studies were prospectively designed, incidence rates could not be determined.

Conclusions This systematic review and meta-analysis provide the first robust assessment of severe asthma prevalence among European children. Our findings underscore the need for comprehensive research to address knowledge gaps in severe asthma, including determining incidence rates, standardising definitions, investigating regional differences and evaluating comorbidities and treatment strategies.

Shareable abstract

Studies suggest severe asthma affects approximately 3% of children with asthma in Europe, leading to frequent symptoms, hospitalisations and decreased quality of life. https://bit.ly/3zKi0ZJ

Introduction

According to the World Health Organization (WHO), asthma is a significant noncommunicable disease that causes disability and healthcare burden worldwide. An estimated 262 million people worldwide suffer from asthma, with a significant proportion being children. This chronic respiratory illness significantly impacts the lives of millions of children, causing recurrent symptoms, limitations in physical activity, and school absenteeism [1].

While asthma presents challenges across the spectrum of disease severity, understanding severe asthma (SA) is critical. Studies suggest that SA affects a minority of children with asthma, with prevalence estimates ranging from 2.1 to 10% [2]. Though less common than milder forms, SA represents a significant subset of childhood asthma, disproportionately driving morbidity and healthcare resource utilisation [2, 3]. In children with SA, the impact extends beyond frequent respiratory symptoms. They are at increased risk of severe exacerbations leading to hospitalisations or even emergency room visits, negatively affecting their school attendance, social activities and overall well-being [4–7]. Moreover, research suggests that severe, uncontrolled asthma, especially if experienced during childhood, could contribute to long-term lung function impairments, potentially affecting respiratory health throughout the lifespan [8].

Despite the significant impact of SA on children, accurate epidemiological data on its global prevalence is limited. This knowledge gap hinders our ability to optimise healthcare responses, allocate resources effectively and advocate for the needs of affected children. This systematic review and meta-analysis aim to address critical knowledge gaps in severe paediatric asthma epidemiology, providing robust data on prevalence, geographic variations and associated comorbidities.

Methods

The protocol of our systematic review and meta-analysis was registered and published with the international prospective register of systematic reviews before the study (www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023472845; register number CRD42023472845). We reported our results using the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) [9] and Meta-Analysis of Observational Studies in Epidemiology (MOOSE) [10] guidelines.  

Outcome

Primary outcomes were the prevalence and incidence of SA in children. Secondary outcomes focused on the evaluation of diagnostic criteria and the description of the therapeutic approach. Further analysis included prevalence estimates according to sex, country, age group, age at diagnosis, symptoms onset, comorbidities and SA endotypes.

Search strategy

A highly sensitive and extensive search strategy was designed to retrieve all articles combining the terms “severe asthma”, “therapy-resistant asthma”, “prevalence”, “incidence” and “child” from the major electronic bibliographic databases (PubMed, Scopus and Web of Science). The bibliographic search focused on the last 20 years (2003–2023). We included only articles in the English language (table S1).

Additionally, according to PRISMA guidelines [9], we extended our search using other methods, such as searching for relevant articles via organisations, websites and citations.

Search results were compiled using the citation management software Rayaan. According to quality standards for reporting meta-analysis of observational studies, two researchers (A. Licari and P. Magri) independently screened the eligible articles. Full texts of deemed eligible articles were retrieved and assessed for inclusion by the same investigators. Any discrepancies were resolved by discussion and consensus.

Study selection (inclusion/exclusion criteria)

We included cohort and cross-sectional studies about the epidemiology of SA in children with asthma, reporting prevalence and incidence with a 95% confidence interval. We excluded non-English articles, articles that studied populations aged below 5 years or over 18 years and articles published before 2003. Inclusion criteria are described in table S2.

Risk of bias assessment

Eligible articles were assessed for the risk of bias according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [11, 12]. The risk of bias was assessed using a tool derived from the STROBE statement, which included the following questions: 1) Were all patients (or a relevant proportion >90%) followed up until the end of the study? 2) Were all patients free of the outcome of interest at baseline included? 3) Were diagnosis/exclusion of SA free from outcome misclassification? Possible answers to these items were “yes”, “no” and “unclear”. All three items were used to assess included cohort studies, while the last two were used for included cross-sectional studies.

Data extraction

We extracted information from each eligible study using a standardised data sheet. The information extracted from each article is listed in table S3. If not directly reported, the prevalence rate was calculated. Disagreements regarding data extraction were resolved by discussion and consensus.

Data synthesis and statistical analysis

We estimated prevalence or incidence and their corresponding 95% confidence intervals. Prevalence was estimated by pooling results of cross-sectional or cohort studies at baseline. Incidence was estimated by pooling results of cohort studies, for the cohort studies free from SA at baseline. Heterogeneity between studies was evaluated with a χ2 test and quantified with the I2 statistic. I2 is a measure of the level of heterogeneity, expressed in three categories on the basis of low, moderate and high I2 values (25, 50 and 75%, respectively). For the secondary outcomes, planned subgroup analysis was based on diagnostic criteria of SA (according to Global Initiative for Asthma (GINA) versus European Respiratory Society/American Thoracic Society (ERS/ATS) versus other institutional definitions), sex (males versus females), age of the population, age at diagnosis and age at symptom onset, country (Mediterranean Europe versus Northern Europe), associated allergic disease, and other comorbidities. Prevalence across sex, geographical area and diagnostic criteria subgroups was compared using one-stage individual patient meta-analysis methods. We used the statistical software package Stata 18.0 (StataCorp USA).

Results

A total of 385 articles were found from selected electronic databases. After removing 138 duplicates, 247 articles were reviewed by title and abstract, and of this group, 213 articles were excluded. 34 full-text articles were screened and 24 were excluded (table S4). The reasons for exclusion included populations that were not eligible due to age composition (adults or preschool children) or diagnosis composition (nonasthma patients or pre-selected children with SA). Other exclusions were due to the use of nonapplicable diagnostic criteria (such as SA diagnosed by survey), irrelevance of outcome and two instances of duplicated data. At the end of the full-text screening, 10 articles were assessed for eligibility [13–22].

Moreover, according to PRISMA guidelines, we extended our search via websites and citation searching and identified 10 more articles suitable for retrieval. These publications were evaluated by title and abstract, and then they were screened in full-text. All 10 articles were excluded (table S4) because of noneligibility of population, noneligibility of outcome, data duplication or wrong study design (case–control studies). 10 articles were included in the final review and analysis (figure 1).

FIGURE 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.

Quality assessment

Among the included articles, all the cross-sectional analyses showed a low risk of bias. As all the included cohort studies presented epidemiological data in the form of transversal analysis among the baseline population, after discussion, we agreed to assess bias using the same question tool as the cross-sectional studies, omitting the additional question for follow-up verification proposed for cohort studies. All of them showed a low risk of bias.

Study characteristics

The included studies were based on paediatric asthma populations aged from 5 to 18 years. Though inclusion criteria stated the lower age limit of 6 years old in our initial study design, after discussion, we decided to extend it to 5 years old in order to include significant data from large epidemiological studies [16–18]. In one case, paediatric subgroup data were extracted from the entire population (adults and children examined) [17]. Investigated age class was homogeneous among studies, covering all the interested ages; only two articles focused on a specific age group (10 and 12 years old) [21, 22]. Only one article reported the age at onset and diagnosis [14]. Information about the sex composition of the population was extracted from four articles [17, 18, 20, 22] and SA prevalence according to sex was analysed.

All the included studies, except for one conducted in the USA [15], were conducted in Europe. Six studies [13, 16, 17, 20–22] were conducted in Northern Europe (United Kingdom, Norway, Sweden, Finland, Netherlands), two [14, 19] in Mediterranean Europe (Spain) and one article [18] presented a European multicentre study involving both Northern European (United Kingdom, Denmark and Netherlands) and Mediterranean European countries (Italy and Spain).

Six articles presented a cross-sectional analysis [14, 15, 19–22], while the remaining four were cohort studies [13, 16–18]. Epidemiological data of SA in children with asthma were extracted: prevalence data were reported in all the included articles in cross-sectional analyses and in cohort studies at baseline. On the contrary, data on incidence were not reported.

All the included studies referred to institutional diagnostic criteria when describing SA. GINA criteria [23] were adopted in four studies [13, 14, 16, 18] and ERS/ATS [24] criteria were cited in three articles [15, 20, 22]. WHO criteria [25] and other definitions provided by national institutions (British Thoracic Society (BTS) [26] and Spanish Asthma Management Guidelines (GEMA) [27]) were also reported [17, 19, 21]. Although diagnostic criteria were properly described, only a few references were found to diagnostic processes (such as spirometry test, fractional exhaled nitric oxide test, skin prick test, peripheral eosinophilic count and levels of allergen-specific immunoglobulin E (IgE)), pathology characterisation (including allergic or other comorbidities and asthma endotypes) and treatment strategies. These data were rare, not uniform and not applicable for statistical analysis.

Prevalence and incidence of SA in children

All the epidemiological data were calculated from the nine European studies [13, 14, 16–22]. No data on incidence were available. The obtained data on SA prevalence in children with asthma are summarised in table 1. In the retrieved studies, the overall prevalence of SA was 3% (95% CI 1–6; I2=99.9%; p<0.001) (figure 2).

TABLE 1

Overall prevalence of severe asthma (SA) and prevalence by subgroup (sex, region, diagnostic criteria) in the analysed studies

FIGURE 2

The overall severe asthma prevalence in children with asthma from studies included in the systematic review and meta-analysis. Random effects and restricted maximum-likelihood model.

Moreover, we investigated SA prevalence among included studies that focused on sex subgroups. The prevalence of SA was 1% (95% CI 0–4; I2=99.88%) in males and 1% (95% CI 0–3; I2=99.83%) in females. The test of group differences resulted in p=0.96 (figure 3).

FIGURE 3

The prevalence of severe asthma in different subgroups: a) according to sex; b) according to geographical variations. Random effects and restricted maximum-likelihood model.

Based on the countries where included studies were conducted, we investigated SA prevalence in Northern and Mediterranean Europe. In Northern Europe, the prevalence was 3% (95% CI 1–5; I2=99.88%), while in Mediterranean Europe it was 8% (95% CI 3–15; I2=99.82%). The test of group differences resulted in p=0.07 (figure 3).

Finally, we analysed data about the prevalence of SA according to adopted diagnostic criteria. We considered the GINA and the ERS/ATS definitions of SA and aggregated other definitions (WHO, BTS and GEMA). SA prevalence was 6% (95% CI 3–9; I2=99.50%) according to GINA criteria, 1% (95% CI 0–1; I2=99.22%) according to ERS/ATS criteria and 3% (95% CI 0–9; I2=99.82%) according to other reported diagnostic criteria. The test of group differences resulted in p<0.001 (figure 4).

FIGURE 4

The prevalence of severe asthma in children according to adopted diagnostic criteria. Random effects and restricted maximum-likelihood model.

The single study [15] that we excluded from the statistical analysis investigated patients affected by asthma but specifically described as difficult-to-treat; because of this, its results could not be aggregated and are here reported separately. SA prevalence among this population was 41%.

Description of clinical features, comorbidities, diagnostic tools and criteria, and therapies in included studies

A clinical description of the pathology was not provided in the included articles.

Few studies reported associated comorbidities among patients with SA (table S6). The most reported comorbidities were a history of atopy in general, atopic dermatitis and allergic rhinitis. Moreover, nonallergic diseases were also found in three studies that focused on rhinosinusitis, nasal polyposis and gastro-oesophageal reflux disease [17, 18, 20]. However, the limited numbers and the lack of uniformity of these results did not allow us to analyse the prevalence of these pathologies in affected patients.

Although the diagnosis of SA generally involves the use of high doses of anti-asthmatic medications, which may vary based on the criteria adopted, the treatment of SA was rarely detailed in the included articles. The anti-asthmatic drugs reported in SA patients included high doses of inhaled corticosteroids (ICS), combined with long-acting beta-agonists (LABAs) or long-acting muscarinic antagonists (LAMAs) and/or anti-leukotrienes drugs. The use of oral corticosteroids was rarely described. Reported values of high-dose ICS varied among studies. One study [17] reported a daily dose of beclomethasone or budesonide >800 µg for children over 12 years old and >400 µg for children under 12 years old, and reported a daily dose of fluticasone >400–500 µg·day−1. Three studies [20, 21, 22] reported higher values of a daily dose of beclomethasone or budesonide, >1600 µg·day−1 for children over 12 years old and >800 µg·day−1 for children under 12 years old. Treatment according to GINA guidelines was reported in three articles [13, 14, 17], which described steps 3, 4 and 5 for children with SA under 11 years old and steps 4 and 5 for those over 11 years old.

Scarce data were found regarding targeted therapies for this class of asthma patients involving theophylline and biological therapies. One study [20] reported the use of anti-IgE and anti-interleukin (IL)-5 monoclonal antibodies, respectively, in 26 and 25% of patients affected by SA; while a second study [17], in the Norwegian sub-population, reported no experience of anti-IgE drugs (0%) among patients with SA.

Discussion

This systematic review and meta-analysis represent the first comprehensive investigation into the epidemiology of severe paediatric asthma, providing a thorough and up-to-date synthesis of the available evidence. Our findings estimate an overall SA prevalence of 3% in children within the studied European populations. This value falls within the lower end of the established range (2.1–10%) [2], highlighting that while SA significantly impacts those affected, it represents a minority of children with asthma. However, this translates to a significant number of children experiencing frequent exacerbations, potentially leading to hospitalisations and decreased quality of life. This highlights the need for paediatricians to implement early identification protocols and proactive management strategies.

While our analysis found no overall sex difference in SA prevalence, the variation among individual studies indicates that this is an area worth further investigation. Potential explanations for variability could include hormonal influences (particularly during puberty), sex-specific environmental exposures, differences in healthcare-seeking behaviour or the interplay between sex and other risk factors.

We observed a potential trend toward higher SA prevalence in children from Mediterranean Europe compared to Northern Europe, although this difference did not reach statistical significance. This trend may be primarily due to variations in diagnostic criteria between the regions. Mediterranean countries predominantly used GINA or GINA-derived guidelines, while Northern European countries employed a wider range of criteria, including those outlined by GINA, ERS/ATS and others. However, these regional differences in prevalence also warrant further investigation into region-specific risk factors, such as environmental exposures or healthcare access disparities. Additionally, this finding underscores the importance of investigating other potential contributing factors, including genetic predisposition variations across populations. Identifying the specific factors driving these potential regional differences is crucial for developing targeted interventions and optimising resource allocation for SA management across Europe. Regions with higher prevalence may require increased resources and specialised interventions, while public health campaigns and research priorities should be tailored to address region-specific risk factors. Understanding these disparities is fundamental for developing effective, targeted and equitable healthcare policies and interventions across diverse populations, both in Europe and globally. Furthermore, expanding epidemiological research to include under-represented populations worldwide is essential for developing a comprehensive understanding of childhood SA.

Our results also highlight the inconsistencies in reported SA prevalence among children when different diagnostic criteria are applied. Our findings show that the use of GINA criteria yielded a higher SA prevalence (6%) compared to ERS/ATS (1%) and other aggregated definitions (3%). This discrepancy is likely due to their differing definitions of high-dose ICS thresholds in paediatric populations. The GINA document adopts lower and more inclusive ICS dose thresholds compared to the ERS/ATS criteria, resulting in a higher proportion of children classified as having SA according to the GINA definition. Inconsistencies in SA classification remain a research challenge. Establishing a robust consensus on diagnostic criteria for SA is crucial to advance research and clinical practice. Standardising the definition of SA would ensure consistent identification and classification, reducing misclassification and facilitating direct comparison of results across studies, ultimately leading to more accurate prevalence estimates and improved research capabilities. This harmonisation would inform clinical decision-making and improve patient care by ensuring appropriate treatment for this severe form of asthma.

Despite limited comprehensive data on comorbidities in children with SA, the association with allergic conditions such as atopy and allergic rhinitis warrants attention. Addressing these shared mechanisms could improve the control of allergic conditions and SA symptoms. Therapies targeting specific inflammatory pathways (e.g. anti-IgE and anti-IL5 monoclonal antibodies) may have dual benefits in patients with severe allergic asthma. Beyond the need for thorough allergy assessment and management, large-scale studies should characterise the full range of comorbidities seen in children with SA. This includes nonallergic conditions such as gastro-oesophageal reflux and obstructive sleep apnoea, potentially informing a more holistic approach to treatment.

Among other findings, our study examined medications used in childhood SA, focusing on high-dose ICSs. Limited data are available on the specifics of medication combinations (ICS, LABA, LAMA, leukotriene modifiers, biologics) employed in children with SA, including dosages, adherence rates and regional variations in usage. A critical knowledge gap exists regarding detailed treatment patterns for SA and their real-world effectiveness in controlling the condition. We lack robust evidence directly comparing the effectiveness of different treatment options in clinical practice. This hinders our ability to determine which specific medication approaches lead to the best outcomes and symptom control in children with SA. Understanding how individual factors, such as age, comorbidities and asthma phenotypes, influence treatment responses is crucial for precise treatment selection. However, this knowledge gap remains a significant barrier.

Strengths and limitations

This systematic review and meta-analysis offer the first comprehensive assessment of severe paediatric asthma epidemiology, providing a rigorous synthesis of the available evidence. The strengths of our study lie in its comprehensive approach, rigorous assessment of potential biases and informative meta-analysis. By combining results from multiple studies, we were able to provide a robust estimate of the overall prevalence of SA among children. In addition, by exploring variations according to sex, geographic region and diagnostic criteria, our investigation identified potential patterns that can inform future research and the development of more targeted interventions.

Our study presents several limitations. The significant heterogeneity observed in our meta-analysis highlights the variability among included studies. This variation suggests the presence of potential influencing factors, which likely include differences in study populations, diagnostic methodologies, geographic settings and time frames. Unfortunately, none of the selected studies were prospectively designed to evaluate the incidence rate. This gap hinders a comprehensive understanding of how severe paediatric asthma develops over time. Incidence data would provide valuable insights into the rate of new SA cases, potentially uncovering periods of heightened risk or identifying protective factors. Additionally, this information would be invaluable for understanding the dynamics of SA development, predicting future trends and developing targeted prevention strategies.

While our prevalence estimates encompass a range of geographical areas, the predominance of European data in large-scale nationwide studies highlights a potential limitation. Further research across diverse global populations is needed to comprehensively understand SA prevalence and its regional variations. Sparse reporting on clinical features, comorbidities, diagnostic processes and treatment strategies represents another significant limitation. More detailed information is needed to ensure robust conclusions about the specific characteristics of severe paediatric asthma and effective management strategies. Future studies with standardised protocols for collecting and reporting such data are essential. This data would guide the development of more precise diagnostic criteria and enable the analysis of treatment success across diverse patient groups and disease characteristics. Finally, a notable limitation of this study is the exclusion of children under 5 years old due to the limited availability of studies specifically addressing SA in this age group within the timeframe of our search. Future research should prioritise investigating the epidemiology of SA in preschool children, as this age group represents a critical period for disease development and intervention.

Conclusion

In conclusion, this systematic review and meta-analysis provides the first comprehensive assessment of SA prevalence in children. Our findings indicate an overall SA prevalence of 3% among European children with asthma, highlighting the need for targeted interventions. Notably, we observed variability in prevalence estimates linked to diagnostic criteria and potential regional differences, warranting further investigation. The lack of longitudinal studies hinders the determination of SA incidence, while limited clinical and therapeutic data underscores the need for standardised data collection protocols. Our study highlights critical knowledge gaps and identifies priority areas for future research to improve the understanding and management of severe paediatric asthma.

Points for clinical practiceOpen questions for research

Do geographical variations in the prevalence of severe childhood asthma suggest an influence of environmental or genetic risk factors?

How can we achieve a universal definition of severe childhood asthma that will improve consistency across diagnosis, research and patient care?

Supplementary materialSupplementary Material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material ERR-0095-2024.SUPPLEMENT

Footnotes

Provenance: Submitted article, peer reviewed.

Data sharing: All data were extracted from published sources that are publicly available. The data used in this analysis can be provided upon request by contacting the corresponding author (M. De Filippo).

Author contributions: P. Magri, S. Manti and A. Licari developed the idea of the study. Data acquisition was done by P. Magri, M. De Filippo, and M. Votto. Statistical analysis was performed by A. De Silvestri. The manuscript was drafted by P. Magri, S. Manti and A. Licari, who had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. The final draft was supervised by G.L. Marseglia. All authors contributed to the final manuscript and checked the final manuscript for correctness.

Conflict of interest: The authors disclose no relevant financial or personal relationships that could influence this work.

Received April 28, 2024.Accepted July 16, 2024.Copyright ©The authors 2024http://creativecommons.org/licenses/by-nc/4.0/

This version is distributed under the terms of the Creative Commons Attribution Non-Commercial Licence 4.0. For commercial reproduction rights and permissions contact permissionsersnet.org