Authors' objectives: There is no clear preferential option for initial step-up of treatment for children with uncontrolled asthma on inhaled corticosteroid. Evaluate the clinical effectiveness of pharmacological treatments to use in children with uncontrolled asthma on inhaled corticosteroid; identify and evaluate the potential for treatment effect modification to optimise treatment delivery; assess the cost-effectiveness of treatments. In the UK, asthma remains a common medical condition affecting over 1 million children, one of the highest prevalence rates worldwide. Asthma is characterised by wheezing, breathlessness, chest tightness, cough and can affect the child’s quality of life (QoL) by limiting daily activities and causing acute attacks. In the UK, it was estimated that one child is admitted to the hospital every 20 minutes because of an asthma attack. Following a diagnosis of asthma in a child, a stepwise approach to treatment should be taken. The initial level of treatment consists of using a low-dose inhaled corticosteroid (ICS) to prevent symptoms and a short-acting β2-adrenoceptor agonist for relieving symptoms. However, around 10–15% of children have inadequate control of asthma symptoms with a low dose of ICS. The target of asthma management is to control the disease through complete control defined as (a) no daytime symptoms, (b) no night-time awakening due to asthma, (c) no need for rescue medication, (d) no asthma attacks, (e) no exacerbation, (f) no limitations on activity, including exercise, (g) normal lung function, (h) minimal side effects from medication. Hence, when asthma remains uncontrolled, a series of further steps are followed. These consist of a treatment step-up by including add-on preventer therapies, such as long-acting β2-agonists (LABAs) or leukotriene receptor antagonists (LTRAs), an increase of the dose of ICS, or adding sustained-release theophylline. Choosing the best step-up treatment becomes a crucial decision to prevent exacerbation occurrence and avoid poor asthma control, improve the QoL of patients and their families, and optimise the use of NHS resources. No clear preferential option for initial step-up exists. Moreover, there is substantial heterogeneity among individuals in the treatment response. The overall aim of the EINSTEIN study was to identify and synthesise all evidence from randomised controlled trials (RCTs) using individual participant data (IPD) to evaluate the clinical effectiveness of pharmacological treatments to use in children with uncontrolled asthma on ICS. A further aim was to identify and evaluate the potential for treatment effect modification to optimise treatment delivery and maximise patients’ informed treatment choice. We assessed the cost-effectiveness of treatments by developing an economic model to estimate the incremental cost per quality-adjusted life-year (QALY) gained.

Authors' results and conclusions: We identified and screened 4708 publications from the search and confirmed 144 randomised controlled trials as eligible. We obtained individual participant data from 29 trials (5381 participants) and extracted limited aggregate data from a further 19 trials. The majority of trials had low risk of bias. The network meta-analysis suggests that medium-dose inhaled corticosteroid + long-acting β2-agonist is the preferred treatment for reducing odds of exacerbation [odds ratio 95% credibility interval: 0.43 (0.20 to 0.92) vs. low-dose inhaled corticosteroid; 40 studies, 8168 patients] and increasing forced expiratory volume in 1 second [mean difference 95% credibility interval: 0.71 (0.35 to 1.06) vs. low-dose inhaled corticosteroid; 23 studies, 2518 patients] while leukotriene receptor antagonist alone is the least preferred. No clear differences were found for asthma control (16 studies, 3027 patients). Limited pairwise analyses suggest some improvement in health-related quality of life for medium-dose inhaled corticosteroid versus inhaled corticosteroid + long-acting β2-agonist [two studies, paediatric asthma quality of life questionnaire, mean difference 95% credibility interval: 0.91 (0.29 to 1.53)]. The rate of hospitalisation due to an asthma attack ranged from 0.5% to 2.7% of patients across five trials. Slightly fewer patients reported neurological disorders (mild/moderate) on inhaled corticosteroid + long-acting β2-agonist versus inhaled corticosteroid + leukotriene receptor antagonist [odds ratio 95% confidence interval: 0.09 (0.01 to 0.82), one study]. There were no deaths recorded. We did not find convincing, consistent evidence to suggest that age, sex, ethnicity, eczema, eosinophilia, asthma severity would be regarded as an effect modifier. The economic analysis indicated that low-dose inhaled corticosteroid was the most cost-effective treatment option while medium-dose inhaled corticosteroid (alone and + long-acting β2-agonist) was associated with the highest number of quality-adjusted life-years, but their incremental cost-effectiveness exceeded the National Institute for Health and Care Excellence threshold. We identified and screened 4708 publications from the search and retrieved 508 full-text articles for eligibility assessment. We confirmed 144 RCTs as meeting the eligibility criteria for this review and attempted to contact the trial owner for all studies. We could not identify a contact for 4 studies; 46 did not reply; 41 refused access to IPD; 24 agreed to provide IPD, but the legal terms of the data-sharing contract could not be agreed upon between the NIHR and the trial sponsors. We obtained IPD from 29 (20%) of the eligible trials contributing data for 5494 eligible participants and were able to extract limited AgD from a further 19 (13%) of the trials without IPD. The majority of trials included in the analyses had a low risk of bias overall. The Bayesian NMA suggests that ICS Medium + LABA reduces the odds of exacerbation [OR 95% credibility interval (CrI): 0.44 (0.19 to 0.90); 40 studies, 8168 patients] and increases FEV1 [MD 95% CrI: 0.71 (0.35 to 1.06); 23 studies, 2518 patients] compared to ICS Low and also compared to ICS Medium [MD 95% CrI: 0.69 (0.33 to 1.05); 23 studies, 2518 patients] and ICS High [MD 95% CrI: 0.54 (0.24 to 0.81); 23 studies, 2518 patients]. LTRA is the least preferred. We could not find clear differences between treatments for asthma control measured by the ACT/ACQ test (16 studies, 3027 patients). Sensitivity analyses did not change conclusions. We could not conduct NMA for any other outcome due to data limitations. Direct pairwise MA of a limited subset of two trials suggests a reduction in the mean QoL score, measured by the paediatric asthma QoL questionnaire (PAQLQ), for ICS + LABA compared to ICS medium [MD 95% confidence interval (CI) −0.91 (−1.53 to −0.29)], although the upper limit of the 95% CI includes values that would not be considered clinically important. There is insufficient evidence to conclude any further differences in QoL amongst the pairs of treatment classes for which we had sufficient data available. The hospitalisation rate due to an asthma attack ranged from 0.5% of patients to 2.7% of patients across five trials, but data were too limited to perform MA. There was considerable heterogeneity in the recording and coding of AEs data across studies. Slightly fewer patients reported neurological disorders (graded as mild or moderate) on ICS + LABA [one patient (4.3%)] compared to ICS + LTRA [seven patients (33.3%)] in one study [OR, 95% CI: 0.09 (0.01 to 0.82)], and a greater number of patients reported neurological disorders for ICS Medium compared to placebo in three studies [OR, 95% CI: 4.8 (1.12 to 20.60)]. There were no other notable differences in AEs, but analyses were limited by data availability and heterogeneity. There were no deaths recorded in any of the trials. We did not find convincing, consistent evidence to suggest that any of the patient characteristics (age, sex, ethnicity, eczema, eosinophilia, asthma severity), which we examined in exploratory NMR analyses, would be regarded as an effect modifier that is adequately supported by robust statistical evidence and clinical rationale. We cannot rule out the possibility of data availability bias but have tried to mitigate this risk by including both IPD and AgD in analyses wherever possible. In the base-case analysis, ICS Low was the cost-effective option; ICS Medium (alone or + LABA) were not cost-effective, with ICERs of £232,500 and £310,000 per QALY gained, respectively. ICS High, ICS + LTRA and LTRA monotherapy were dominated by alternatives which were less costly and associated with more QALYs. Sensitivity analyses indicated that changes in utilities had an inconsequential effect on the ICERs, while varying the transition probabilities associated with the controlled and exacerbation states by ± 50% concurrently for all treatments resulted in ICS Medium becoming cost-effective at £15,102 per QALY gained. When the costs of branded inhalers were reduced individually by 50%, potentially reflecting a generic alternative, the ICER for ICS Medium + LABA reduced considerably but remained non-cost-effective. All the comparators to ICS Low remained non-cost-effective in the analysis of structural uncertainty. At willingness-to-pay thresholds of £20,000 and £30,000 per QALY, ICS Low had 0.45 and 0.44 probabilities of being cost-effective, respectively. ICS Medium had lower probabilities of 0.20 at £20,000 and 0.21 at £30,000 per QALY; while ICS Medium + LABA was the least likely of being cost-effective, with probabilities of 0.07 and 0.09 at the £20,000 and £30,000 per QALY thresholds, respectively. Overall, we conclude that ICS Medium + LABA would be the recommended step-up treatment for children with asthma that is not well controlled on ICS and that LTRA alone should be avoided. There is very little evidence available on which to make conclusions regarding ICS + Theophylline and insufficient evidence at this time to suggest that treatment effect is modified by patient characteristics (age, sex, ethnicity, eczema, eosinophilia, asthma severity), although further research is recommended. The economic analysis indicated that ICS Low was the most cost-effective treatment, while ICS Medium (alone and + LABA) were associated with the highest number of QALYs; however, they did not meet the criteria to be cost-effective.

Authors' methods: Systematic review and individual participant data network meta-analysis. Studies were eligible if they were parallel or crossover randomised controlled trials comparing at least one of the pharmacological treatments of interest in participants aged 

Project Status: Completed

URL for project: https://www.journalslibrary.nihr.ac.uk/programmes/hta/16/110/16

Year Published: 2025

URL for published report: https://www.journalslibrary.nihr.ac.uk/hta/HGWT3617

URL for additional information: English

English language abstract: An English language summary is available

Publication Type: Full HTA

Country: England, United Kingdom

DOI: 10.3310/HGWT3617

Organisation Name: NIHR Health Technology Assessment programme

Contact Address: NIHR Journals Library, National Institute for Health and Care Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK

Contact Name: journals.library@nihr.ac.uk

Contact Email: journals.library@nihr.ac.uk