Scolaris Content Display Scolaris Content Display

Addition of long‐acting beta2‐agonists to inhaled corticosteroids for chronic asthma in children

Collapse all Expand all

Abstract

Background

Long‐acting beta2‐agonists (LABA) in combination with inhaled corticosteroids (ICS) are increasingly prescribed for children with asthma.

Objectives

To assess the safety and efficacy of adding a LABA to an ICS in children and adolescents with asthma. To determine whether the benefit of LABA was influenced by baseline severity of airway obstruction, the dose of ICS to which it was added or with which it was compared, the type of LABA used, the number of devices used to deliver combination therapy and trial duration.

Search methods

We searched the Cochrane Airways Group Asthma Trials Register until January 2015.

Selection criteria

We included randomised controlled trials testing the combination of LABA and ICS versus the same, or an increased, dose of ICS for at least four weeks in children and adolescents with asthma. The main outcome was the rate of exacerbations requiring rescue oral steroids. Secondary outcomes included markers of exacerbation, pulmonary function, symptoms, quality of life, adverse events and withdrawals.

Data collection and analysis

Two review authors assessed studies independently for methodological quality and extracted data. We obtained confirmation from trialists when possible.

Main results

We included in this review a total of 33 trials representing 39 control‐intervention comparisons and randomly assigning 6381 children. Most participants were inadequately controlled on their current ICS dose. We assessed the addition of LABA to ICS (1) versus the same dose of ICS, and (2) versus an increased dose of ICS.

LABA added to ICS was compared with the same dose of ICS in 28 studies. Mean age of participants was 11 years, and males accounted for 59% of the study population. Mean forced expiratory volume in one second (FEV1) at baseline was ≥ 80% of predicted in 18 studies, 61% to 79% of predicted in six studies and unreported in the remaining studies. Participants were inadequately controlled before randomisation in all but four studies.

There was no significant group difference in exacerbations requiring oral steroids (risk ratio (RR) 0.95, 95% confidence interval (CI) 0.70 to 1.28, 12 studies, 1669 children; moderate‐quality evidence) with addition of LABA to ICS compared with ICS alone. There was no statistically significant group difference in hospital admissions (RR 1.74, 95% CI 0.90 to 3.36, seven studies, 1292 children; moderate‐quality evidence)nor in serious adverse events (RR 1.17, 95% CI 0.75 to 1.85, 17 studies, N = 4021; moderate‐quality evidence). Withdrawals occurred significantly less frequently with the addition of LABA (23 studies, 471 children, RR 0.80, 95% CI 0.67 to 0.94; low‐quality evidence). Compared with ICS alone, addition of LABA led to significantly greater improvement in FEV1 (nine studies, 1942 children, inverse variance (IV) 0.08 L, 95% CI 0.06 to 0.10; mean difference (MD) 2.99%, 95% CI 0.86 to 5.11, seven studies, 534 children; low‐quality evidence), morning peak expiratory flow (PEF) (16 studies, 3934 children, IV 10.20 L/min, 95% CI 8.14 to 12.26), reduction in use of daytime rescue inhalations (MD ‐0.07 puffs/d, 95% CI ‐0.11 to ‐0.02, seven studies; 1798 children) and reduction in use of nighttime rescue inhalations (MD ‐0.08 puffs/d, 95% CI ‐0.13 to ‐0.03, three studies, 672 children). No significant group difference was noted in exercise‐induced % fall in FEV1, symptom‐free days, asthma symptom score, quality of life, use of reliever medication and adverse events.

A total of 11 studies assessed the addition of LABA to ICS therapy versus an increased dose of ICS with random assignment of 1628 children. Mean age of participants was 10 years, and 64% were male. Baseline mean FEV1 was ≥ 80% of predicted. All trials enrolled participants who were inadequately controlled on a baseline inhaled steroid dose equivalent to 400 µg/d of beclomethasone equivalent or less.

There was no significant group differences in risk of exacerbation requiring oral steroids with the combination of LABA and ICS versus a double dose of ICS (RR 1.69, 95% CI 0.85 to 3.32, three studies, 581 children; moderate‐quality evidence) nor in risk of hospital admission (RR 1.90, 95% CI 0.65 to 5.54, four studies, 1008 children; moderate‐quality evidence).

No statistical significant group difference was noted in serious adverse events (RR 1.54, 95% CI 0.81 to 2.94, seven studies, N = 1343; moderate‐quality evidence) and no statistically significant differences in overall risk of all‐cause withdrawals (RR 0.96, 95% CI 0.67 to 1.37, eight studies, 1491 children; moderate‐quality evidence). Compared with double the dose of ICS, use of LABA was associated with significantly greater improvement in morning PEF (MD 8.73 L/min, 95% CI 5.15 to 12.31, five studies, 1283 children; moderate‐quality evidence), but data were insufficient to aggregate on other markers of asthma symptoms, rescue medication use and nighttime awakening. There was no group difference in risk of overall adverse effects, A significant group difference was observed in linear growth over 12 months, clearly indicating lower growth velocity in the higher ICS dose group (two studies: MD 1.21 cm/y, 95% CI 0.72 to 1.70).

Authors' conclusions

In children with persistent asthma, the addition of LABA to ICS was not associated with a significant reduction in the rate of exacerbations requiring systemic steroids, but it was superior for improving lung function compared with the same or higher doses of ICS. No differences in adverse effects were apparent, with the exception of greater growth with the use of ICS and LABA compared with a higher ICS dose. The trend towards increased risk of hospital admission with LABA, irrespective of the dose of ICS, is a matter of concern and requires further monitoring.

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Plain language summary

Addition of long‐acting beta2‐agonists to inhaled corticosteroids for chronic asthma in children

Background

Most consensus statements recommend use of long‐acting beta2‐agonists (LABA) as adjunct therapy to inhaled corticosteroids (ICS) for poorly controlled asthma, despite the use of low‐dose ICS.

Review question

What are the benefits and safety of the combination of LABA and ICS in children with persistent asthma when compared with the same dose or a higher dose of ICS alone?

What evidence did we find?

From available evidence until January 2015, we found 39 eligible studies evaluating the combination of LABA and ICS in children with persistent asthma. Of these, 28 studies compared LABA with the same dose of ICS, and the remaining studies compared LABA with a larger dose of ICS.

The number of people who had an exacerbation (worsening of symptoms) that required treatment with oral steroids was not significantly different. However, lung function improved in people taking LABA and steroids compared with the same dose of steroids only or larger doses of steroids. No evidence suggested increased serious adverse events or adverse events (also known as side effects) with the addition of LABA.

Compared with the same dose of ICS, people used less of their rescue/relief bronchodilator treatment. There was no benefit for control of asthma symptoms when LABA added to ICS was compared with higher doses of ICS. The higher dose of ICS was associated with 1.2 cm per year lower growth than was observed with the combination of LABA and a lower dose of ICS.

Conclusion

In children with persistent asthma, the combination of LABA and ICS did not reduce the risk of exacerbations requiring steroid treatment but did improve lung function when compared with the same, or a higher, dose of ICS. No differences in adverse effects were apparent, with the exception of better growth with use of ICS and LABA compared with a higher ICS dose. The trend towards increasing chances of hospital admission indicates the need for continuous monitoring and additional trials in children.

Quality of the evidence

Overall, we judged the quality of evidence to be moderate. Most outcomes showed wide confidence intervals, which led to downgrading of the quality of evidence to moderate. In a few outcomes for which open‐label studies contributed data, we further downgraded evidence quality to low.

Authors' conclusions

Implications for practice

Evidence is insufficient at present to firmly support use of LABA as an adjunct therapy to ICS as a step 3 strategy to reduce risk of asthma exacerbations requiring steroids, as compared with using the usual dose of ICS (step 2) or an increased dose of ICS (step 3). The wide confidence intervals do not rule out a superior effect of either treatment. Stepping up therapy with the addition of LABA to the usual dose of ICS improves lung function beyond that observed when remaining on ICS as step 2 strategy, but with no apparent benefits of asthma symptom control and use of rescue SABA. Similarly, significant improvements in morning PEF observed with the combination of LABA and usual ICS dose versus an increased dose of ICS have not been associated with improvement in other indicators of lung function and asthma control. The apparent reduction in growth associated with use of 400 to 800 μg/d of BDP‐equivalent raises concern when high‐dose beclomethasone or budesonide is considered as increased ICS (step 3 therapy). Of note, the trend towards increased hospital admission with LABA, irrespective of the dose of ICS, and toward serious adverse health events compared with an increased dose of ICS is a matter of some concern and calls for larger, longer‐term trials in children with substantial morbidity, to clarify this issue.

Implications for research

Future trials should have the following characteristics.

Population

A large study is urgently needed in children with moderate and severe airway obstruction and with higher asthma morbidity (e.g. prior hospital admission, requirement for OCS) at baseline than those recruited to trials aggregated in this review. Stratification according to degree of airway obstruction (i.e. baseline FEV1) and inclusion of younger, preschool‐aged children should feature in the design of such trials. Use of diagnostic criteria for asthma that do not require a positive bronchodilator response for enrolment would allow the study to be more generalisable to the general paediatric asthma population and would reduce the potential overestimation of effect on lung function (by preselecting responders to SABA).

Interventions

Future interventions should test the combination therapy delivered by a single inhaler (combining LABA and ICS) to ensure no use of LABA as monotherapy. Interventions may include head‐to‐head comparisons of salmeterol versus formoterol, combined with low or moderate doses of ICS. The control intervention should focus on increased doses of ICS (step 3), so that two step 3 treatment strategies are compared.

Design

  • Double‐blinding, adequate randomisation and complete reporting of withdrawals and dropouts with an explicit definition of the intention‐to‐treat population analysed.

  • Intervention period of 24 to 52 weeks or longer, to properly assess the impact on exacerbations requiring systemic corticosteroids and those resulting in hospital admission, as well as adverse health events (growth, adrenal function, bone mineralisation, serious adverse health events).

  • Clear reporting of the percentage of (and reasons for) non‐eligibility of approached participants and of those enrolled in the run‐in period is required, as inadequate reporting of the selected population results in difficulty identifying to whom the results can be generalised.

  • Complete reporting of continuous (denominators, mean change and mean standard deviation of change) and dichotomous (denominators and rate) data in the units used in this systematic review would allow aggregation of data.

Outcomes of particular importance to assess include the following.

  • Exacerbations requiring rescue systemic corticosteroids.

  • Asthma‐related hospital admission or acute care visit.

  • Compliance with either intervention both before (for ICS) and after randomisation (for both ICS and combination therapy). The impact of compliance with combination therapy versus placebo and ICS on the magnitude of the effect size should be examined.

  • Cost‐effectiveness of use of combination inhalers as compared with ICS alone.

  • Serious or overall adverse events associated with LABA or ICS especially growth, adrenal function, bone mineralisation.

  • Functional measures including quality of life.

An ongoing large six‐month study on safety and benefit of LABA and ICS (salmeterol and fluticasone) in children four to 11 years of age may provide further evidence to support our preliminary outcomes (NCT01462344).

Summary of findings

Open in table viewer
Summary of findings for the main comparison.

LABA + ICS compared with same dose of ICS for children with chronic asthma

Patient or population: children with chronic asthma

Settings: outpatients

Intervention: LABA + ICS

Comparison: same dose of ICS

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

Number of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Increased dose of ICS

LABA + ICS

Number of participants with exacerbations requiring systemic steroids

86 per 1000

94 per 1000

RR 0.95
(0.70 to 1.28)

1669
(12 studies)

⊕⊕⊕⊝
Moderatea

Number of participants with exacerbations requiring hospitalisation

19 per 1000

33 per 1000

RR 1.74
(0.90 to 3.36)

1292
(6 studies)

⊕⊕⊕⊝
Moderatea

Serious adverse events

16 per 1000

18 per 1000

RR 1.17
(0.75 to 1.85)

4022
(16 studies)

⊕⊕⊕⊝
Moderatea

Total number of withdrawals

127 per 1000

94 per 1000

RR 0.80
(0.67 to 0.94)

4374
(23 studies)

⊕⊕⊝⊝
Lowa,b

Change in FEV1 (L) at endpoint

Baseline mean FEV1 ranged from 1.65 L to 1.9 L (baseline data reported in 4 studies only)

Mean FEV1 change from baseline with LABA + ICS was 0.08 L higher (0.06 to 0.1 higher)

1942 (9 studies)

⊕⊕⊝⊝
Lowa,b

Change in morning PEF (L/min) at endpoint

Illustrative post‐treatment PEFs range from 235 to 290 L/min (data from 3 recent studies)

Mean PEF change from baseline with LABA + ICS was 10.20 L/min higher (8.14 to 12.26 higher)

3934 (16 studies)

⊕⊕⊕⊝
Moderatea

Total number of adverse events

547 per 1000

568 per 1000

RR 1.04 (0.98 to 1.10)

3284
(15 studies)

⊕⊕⊕⊝
Moderateb

*The basis for the assumed risk (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; FEV1: Forced expiratory volume in 1 second;ICS: Inhaled corticosteroids; LABA: Long‐acting beta2‐agonists; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

aLarger sample size may change the outcome.

bOpen‐label study contributed data.

Open in table viewer
Summary of findings 2.

LABA + ICS compared with increased dose of ICS for children with chronic asthma

Patient or population: children with chronic asthma

Settings: outpatients

Intervention: LABA + ICS

Comparison: increased dose of ICS

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

Number of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Increased dose of ICS

LABA + ICS

Number of participants with exacerbations requiring systemic steroids

41 per 1000

69 per 1000

RR 1.69
(0.85 to 3.32)

581 (3 studies)

⊕⊕⊕⊝
Moderatea

Number of participants with exacerbations requiring hospitalisation

8 per 1000

18 per 1000

RR 1.90
(0.65 to 5.54)

1008
(4 studies)

⊕⊕⊕⊝
Moderatea

Serious adverse events

24 per 1000

39 per 1000

RR 1.54
(0.81 to 2.94)

1343
(7 studies)

⊕⊕⊕⊝
Moderatea

Total number of withdrawals

70 per 1000

67 per 1000

RR 0.96
(0.67 to 1.37)

1491
(7 studies)

⊕⊕⊕⊝
Moderatea

Change in FEV1 (L) at endpoint

Mean baseline FEV1 ranged from 1.6 to 1.7 L

Mean FEV1: change from baseline with LABA + ICS was 0.01 L higher (‐0.03 to 0.05 higher)

526 (2 studies)

⊕⊕⊕⊝
Moderatea

Change in morning PEF (L/min) at endpoint

Mean change in end of treatment PEF ranged from 16.7 to 39.2 L/min

Mean PEF: change from baseline with LABA + ICS was 8.73 L/min higher (5.15 to 12.31 higher)

1283 (5 studies)

⊕⊕⊕⊝
Moderatea

Total number of adverse events

569 per 1000

576 per 1000

RR 1.01 (0.92 to 1.10)

1254
(6 studies)

⊕⊕⊕⊕
High

*The basis for the assumed risk (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; FEV1: Forced expiratory volume in 1 second; ICS: Inhaled corticosteroids; LABA: Long‐acting beta2‐agonists; PEF: Peak expiratory flow; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

aLarger sample size may change the outcome.

Background

Description of the condition

Inhaled corticosteroids (ICS) are the most effective treatment for long‐term control of asthma in children (Adams 2005; Manning 2008; Adams 2008a; Chauhan 2012). They are recommended as first‐line agents for the management of childhood asthma in all national and international consensus statements (NAEPP 2011; Lougheed 2012; BTS 2014; GINA 2015). When ICS alone are insufficient to achieve asthma control, various options may be considered, such as increasing the dose of ICS (Adams 2008b) or adding a second drug such as a long‐acting beta2‐agonist (LABA) or a leukotriene receptor antagonist (LTRA) (Chauhan 2014).

Description of the intervention

In adults with unsatisfactory asthma control, international guidelines clearly favour the addition of LABA to low or moderate doses of ICS over other options such as increasing the dose of steroids or adding other agents (NAEPP 2011; BTS 2014; GINA 2015). In children five to 12 years of age with insufficient control on ICS, however, recommendations regarding the preferred step 3 strategy and the dose of ICS to which LABA should be added differ markedly across countries. The International Australian and Canadian guidelines recommend increasing the dose of ICS to medium dose (201 to 400 μg beclomethasone equivalent) rather than adding LABA or LTRA to low‐dose ICS in children six to 11 years of age (Lougheed 2012; NAC Guidelines 2014). British Thoracic Society guidelines recommend combination therapy at a low dose (200 to 400 µg/d beclomethasone equivalent) (BTS 2014). Global Initiative for Asthma (GINA) and Australian guidelines recommend increasing the dose of ICS over adding LABA to a higher dose of ICS (400 µg/d) (NAC Guidelines 2014; GINA 2015). American guidelines reveal no clear preference for adding LABA to a low dose of ICS or increasing the dose of ICS for children five to 11 years of age with uncontrolled asthma and taking a low dose of ICS (NAEPP 2011). No formal recommendation is available for their use in preschool‐age children.

How the intervention might work

Data from paediatric clinical trials have been included in few previous meta‐analyses assessing the efficacy and safety of LABA in combination with ICS (Ducharme 2010; Ducharme 2010a). However, Bisgaard 2003 cautioned against routine use of LABA in children, as they did not offer protection against exacerbations and led to increased risk of hospital admission. Other outcomes such as adverse effects, lung function and symptoms were not examined.

Why it is important to do this review

The wide divergence of recommendations likely stems from lack of solid evidence in children to support international asthma guidelines and justifies a systematic review of the topic. In 2010, we published a Cochrane review conducted to compare LABA added to ICS of the same dose, or a higher dose, for adults and children with chronic persistent asthma, which demonstrated that LABA and ICS led to a significant reduction in risk of exacerbations requiring oral steroids (Ducharme 2010; Ducharme 2010a). The Cochrane Collaboration had published a separate paediatric systematic review on the same topic in the year 2009 (Ni Chroinin 2009). Since that time, additional published and unpublished paediatric trials have become available, enabling us to update the review to include newly available evidence and to shed more light on the role of LABA as adjunct therapy to ICS for children with partial control when taking ICS alone.

Objectives

To assess the safety and efficacy of adding a LABA to an ICS in children and adolescents with asthma. To determine whether the benefit of LABA was influenced by baseline severity of airway obstruction, the dose of ICS to which it was added or with which it was compared, the type of LABA used, the number of devices used to deliver combination therapy and trial duration.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials conducted in children for whom a LABA was added to an ICS were eligible.

Types of participants

Children and adolescents two to 18 years of age with persistent asthma who had received daily ICS therapy for at least four weeks before study entry.

Types of interventions

LABA (salmeterol or formoterol) versus placebo administered daily for at least four weeks. LABA added to ICS was compared:

  • with the same ICS dose; or

  • with an increased dose of ICS.

Studies in which maintenance ICS therapy was interrupted for the purposes of run‐in were not eligible for the review. Other co‐interventions such as xanthines, anticholinergics and other anti‐asthmatic medications were permitted, provided that the dose remained unchanged throughout the study. Inhaled short‐acting beta2‐agonists (SABA) and short courses of systemic steroids were allowed as rescue medications.

Types of outcome measures

Primary outcomes

  • Number of asthma exacerbations of moderate intensity, that is, requiring a short course of systemic corticosteroids.

Secondary outcomes

  • Admissions to hospital.

  • Urgent care visits.

  • Pulmonary function tests (morning and evening peak expiratory flow (PEF) or forced expiratory flow rate in one second (FEV1)).

  • Symptoms.

  • Quality of life scores.

  • Use of rescue SABA.

  • Nighttime awakening.

  • Changes in measures of inflammation such as serum eosinophil cationic protein and sputum eosinophils.

  • Rates of clinical and biochemical adverse effects.

  • Any adverse effects including growth suppression, adrenal suppression, bone mineral loss and others. A suite of related Cochrane reviews considered serious adverse effects related to LABA (Cates 2008a; Cates 2009a; Cates 2009b).

Search methods for identification of studies

Electronic searches

We carried out an electronic literature search of the Cochrane Airways Group Specialised Register of asthma trials, which is derived from systematic searches of bibliographic databases including the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE and the Cumulative Index to Nursing and Allied Health Literature (CINAHL), and we handsearched respiratory journals and meeting abstracts (see Appendix 1 for additional details). This Register also includes a variety of studies published in foreign languages. We did not exclude trials on the basis of language. In Appendix 2, we have detailed search methods used in the previous version of this review. For this update, we searched the Register from May 2008 to January 2015, using the strategy presented in Appendix 3.

Searching other resources

We checked reference lists of all included studies and reviews to identify potentially relevant citations.

We searched manufacturers' and clinical trial websites (Glaxo Smith Kline clinical trials website ; AstraZeneca clinical trials website; Novartis clinical trial results website; Clinical Study Results) to identify other published or unpublished study data.

Data collection and analysis

Selection of studies

From the title, abstract or descriptors, two of three review authors (BC, MNC, SM) independently reviewed the literature searches. We excluded all studies that were not randomised controlled trials or that clearly did not fit the inclusion criteria. Two review authors independently reviewed all other citations in full text to assess eligibility.

Data extraction and management

Two of four review authors (BC, MNC, CC, SM) independently extracted data from included trials onto Excel spreadsheets and entered data into the Cochrane software program, Review Manager 5.3 (Review Manager (RevMan)). When necessary, we expanded graphic reproductions and estimations from other data presented in the paper. We contacted primary authors or sponsors to request confirmation of methods and data extraction and to ask for additional information, when needed.

We recorded the following as a 'User defined order'.

  • Mean daily dose of ICS in trials in which both intervention and control groups used the same dose of ICS.

  • Dose difference between groups in studies that compared LABA added to ICS with an increased dose of ICS. Researchers reported both values in chlorofluorocarbon (CFC)‐propelled beclomethasone‐equivalents, where 1 µg of beclomethasone dipropionate equates to 1 µg of budesonide and 0.5 µg of fluticasone propionate (NAEPP 2011), and all doses of inhaled medications as ex‐valve, rather than ex‐inhaler, values.

Assessment of risk of bias in included studies

We assessed risk of bias according to the following domains.

  • Random sequence generation.

  • Allocation concealment.

  • Blinding of participants and personnel.

  • Blinding of outcome assessment.

  • Incomplete outcome data.

  • Selective outcome reporting.

  • Other bias.

For each domain, we judged risk of bias as low, unclear or high, in line with recommendations provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We judged the study to have high methodological quality when reported randomisation procedures and blinding were adequate and low and balanced group attrition was noted, supporting low risk of bias.

Measures of treatment effect

We calculated treatment effects for dichotomous variables as pooled risk ratios (RRs) with 95% confidence intervals (CI). For continuous outcomes, such as pulmonary function tests, we calculated pooled statistics as mean differences (MD) or standardised mean differences (SMD) if results were reported on different scales and were reported with 95% CIs. When standard deviations were not presented but could be estimated from an effect estimate and use of confidence intervals, standard error or P value, we combined the MD with the generic inverse variance function (GIV) in Review Manager.

Unit of analysis issues

We excluded cross‐over studies from contributing data to dichotomous measurements of exacerbations, as we used analyses that assumed measurements were taken from independent samples.

Dealing with missing data

We directly contacted study investigators (or study sponsors when trials had pharmaceutical company sponsorship) to request confirmation of methods and to ask for data missing from the original trial report, if needed.

Assessment of heterogeneity

We examined homogeneity of effect sizes between pooled studies by using the I2 statistic, with 50% or more as the cutoff for exploring possible causes of heterogeneity (Higgins 2003; Higgins 2011). In the absence of heterogeneity, we used the fixed‐effect model; otherwise, we applied the Dersimonian and Laird random‐effects model (DerSimonian 1986) to the summary estimates. We reported results of the fixed‐effect model unless otherwise stated in the text.

Assessment of reporting biases

We planned to use funnel plots to check for indications of possible publication bias and small‐study effects, if we had been able to pool data from 10 or more studies.

Data synthesis

We performed meta‐analyses using the Cochrane statistical package RevMan 5 (Review Manager (RevMan)) and assumed equivalence if the risk ratio estimate and its confidence interval were between 0.9 and 1.1.

We performed the analysis to examine two main comparisons, namely, the combination of LABA and ICS versus:

  • a similar dose of ICS with placebo, representing step 2 of the BTS guidelines; or

  • an increased dose of ICS with placebo, representing step 3 of the BTS guidelines.

When a trial included more than two arms, we considered additional control‐intervention group comparisons for this review. If the same group was used twice as a comparator in a three‐arm study, we halved the number of participants in the group used twice to avoid over‐representation. For event rates, we halved the denominator in the control group (Verberne 1998a; Verberne 1998b; Zimmerman 2004a; Zimmerman 2004b; Pohunek 2006a; Pohunek 2006b; Morice 2008a; Morice 2008b; Eid 2010a; Eid 2010b; SAM40012a; SAM40012b).

Subgroup analysis and investigation of heterogeneity

We planned subgroup analyses to explore possible reasons for heterogeneity and, in the absence of heterogeneity, to identify potential effect modifiers when the magnitude of benefit may vary according to baseline characteristics. We examined the following a priori defined subgroups.

  • Magnitude of airway obstruction at baseline as determined by the mean percent predicted FEV1: classified as mild (80% of predicted or more), moderate (61% to 79% of predicted) or severe (60% of predicted or less) (GINA 2015).

  • Mean dose (ex‐valve) of ICS in comparison 1 when LABA + ICS was compared with placebo + ICS, and the dose difference in comparison 2 when LABA + ICS was compared with increased doses of ICS, both reported in CFC‐propelled beclomethasone‐equivalent doses (µg/d) and portrayed as the user‐defined number.

  • Usual versus higher than usual dose (reported as ex‐valve in µg) of the LABA (salmeterol or formoterol).

  • Type of LABA given (salmeterol vs formoterol).

  • Use of one or two devices to deliver the combination of ICS and LABA.

  • Trial duration with trials ≤ 16 weeks compared with those ≥ 24 weeks.

Sensitivity analysis

We performed sensitivity analyses to assess whether results for our primary outcome were sensitive to blinding, publication status and funding.

Results

Description of studies

Results of the search

See Figure 1 for an overview of the literature search results, their assessment and inclusion of studies in the review. Updated electronic and additional handsearches from May 2008 to January 2015 yielded 283 additional citations. We included eight new trials, resulting in a total of 33 eligible trials representing 39 control‐intervention comparisons.


Study flow diagram.

Study flow diagram.

Please note that some of the study names have changed since the last published version to reflect the Cochrane study naming convention: SAM104926, SFA 100314, SD 039 0719, SD 039 0725a, SD 039 0725b, and SFA100316 and were replaced with De Blic 2009; Pearlman 2009; Berger 2010; Eid 2010a; Eid 2010band Murray 2011, respectively.

Six trials generated an additional control‐intervention comparison. Three trials (representing six control‐intervention comparisons) assessed two forms of additive therapy (ICS and LABA via one inhaler or via two inhaler devices against one dose of ICS): Pohunek 2006a; Pohunek 2006b; Morice 2008a; Morice 2008b; Eid 2010a; Eid 2010b. Three trials (representing six control‐intervention comparisons) assessed one form of ICS and LABA versus two doses of ICS (Verberne 1998a; Verberne 1998b; Zimmerman 2004a; Zimmerman 2004b; SAM40012a; SAM40012b). The review therefore lists 39 control‐intervention comparisons.

Included studies

A total of 33 trials randomly assigned 6381 children. Twenty‐seven trials were available as full‐text publications, and six trials were published as abstracts or were presented in unpublished reports of trials accessed through pharmaceutical company trial registries.

Most (69.7%) studies were funded by producers of both LABA and ICS inhalers: 13 were supported by GSK; nine by AstraZeneca; and one by Allen & Hanburys, a subsidiary of GSK in the United Kingdom (Russell 1995). Two were supported by grant agencies (Lemanske 2010; Lenney 2013), two were independently supported by a charitable organization (Langton Hewer 1995; Stelmach 2007) and five failed to identify the source of funding (Ortega‐Cisneros 1998; Heuck 2000; Zimmerman 2004a; Zimmerman 2004b; Teper 2005; Stelmach 2008).

We classified studies into one of two comparisons according to the research question addressed. In accordance with therapeutic management as recommended by GINA 2015 and BTS 2014, we considered participants given low‐dose ICS alone as receiving step 2 therapy. We referred to the comparison of LABA versus placebo added to the same ICS dose as step 3 versus step 2 comparison (28 control‐intervention comparisons). Hereafter, we refer to comparisons testing the combination of LABA and ICS versus a double dose of ICS used before randomisation as a step 3 versus step 3 comparison (11 control‐intervention comparisons). Bisgaard 2006 examined LABA added to a lower dose of ICS (BDP 100 µg) than has been advocated by international guidelines as step 3 and included this in the step 3/step 3 comparison.

We describe hereafter characteristics of studies that contributed outcome data to one or more comparisons in the review. For a full description of each eligible study, see Characteristics of included studies.

LABA + ICS versus same dose of ICS as used before randomisation (step 3 vs step 2)

Twenty‐eight control‐intervention comparisons randomly assigned 4753 children to assessLABA versus placebo added to the same dose of ICS in both groups (Langton Hewer 1995; Meijer 1995; Russell 1995; Simons 1997; Verberne 1998a; Akpinarli 1999; Tal 2002; Zimmerman 2004a; Zimmerman 2004b; Malone 2005; Teper 2005; Pohunek 2006a; Pohunek 2006b; Stelmach 2007; Morice 2008a; Morice 2008b; Stelmach 2008; Pearlman 2009; Rutkowski 2009; Berger 2010; Carroll 2010; Eid 2010a; Eid 2010b; Murray 2011; SAM40012a; Lenney 2013; SD 039 0714; SD 039 0718).

Participants

The mean age of participants was 11 years, and males accounted for 59% of study populations. Mean FEV1 % predicted at baseline was ≥ 80% in 19 control‐intervention comparisons: Morice 2008a; Morice 2008b; Langton Hewer 1995; Meijer 1995; Simons 1997; Verberne 1998a; Verberne 1998b; Malone 2005; Teper 2005; Pohunek 2006a; Pohunek 2006b; Stelmach 2007; Stelmach 2008; Pearlman 2009; Berger 2010; Carroll 2010; Murray 2011; Lenney 2013; SD 039 0718); 61% to 79% of predicted in six control‐intervention comparisons (Russell 1995; Akpinarli 1999; Tal 2002; Zimmerman 2004a; Zimmerman 2004b; SD 039 0714) and unreported in the remaining studies. Participants were inadequately controlled before randomisation in all but four studies in which they were described as well controlled (Meijer 1995; Simons 1997; Pohunek 2006a; Pohunek 2006b), or when control was not reported (Teper 2005; Stelmach 2008; Rutkowski 2009; Berger 2010; Eid 2010a; Eid 2010b).

Interventions

Salmeterol was assessed in 12, and formoterol in 16, control‐intervention comparisons. All but one comparison tested the usual recommended dose of the LABA (i.e. salmeterol 50 µg twice daily, formoterol 4.5, 6 or 12 µg twice daily); Langton Hewer 1995 used salmeterol 100 µg twice daily. The dose of ICS (beclomethasone‐equivalent) was 200 μg/d in five studies (Stelmach 2007; Stelmach 2008; Eid 2010a; Eid 2010b; SD 039 0718); 400 µg/d in 13 control‐intervention comparisons (Verberne 1998a; Tal 2002; Malone 2005; Pohunek 2006a; Pohunek 2006b; Morice 2008a; Morice 2008b; Pearlman 2009; Carroll 2010; Murray 2011; SAM40012a; Lenney 2013; SD 039 0714), 500 µg/d in two studies (Meijer 1995; Teper 2005), 800 µg/d in two studies (Rutkowski 2009; Berger 2010) and unspecified or varied in six studies (Langton Hewer 1995; Russell 1995; Simons 1997; Akpinarli 1999; Zimmerman 2004a; Zimmerman 2004b). Eighteen (46%) control‐intervention comparisons assessed the combination of LABA and ICS in a single device; the remainder assessed the efficacy and safety of a LABA administered separately from an ICS.

Trial duration ranged from eight weeks or less in nine studies (Langton Hewer 1995; Simons 1997; Akpinarli 1999; Stelmach 2007; Stelmach 2008; Rutkowski 2009; Pearlman 2009; Carroll 2010; Murray 2011), to 12 to 16 weeks in 14 control‐intervention comparisons (Meijer 1995; Russell 1995; Tal 2002; Zimmerman 2004a; Zimmerman 2004b; Malone 2005; Pohunek 2006a; Pohunek 2006b; Morice 2008a; Morice 2008b; Eid 2010a; Eid 2010b; SD 039 0718; SD 039 0714), to 24 to 26 weeks in two studies (Berger 2010; SAM40012a) to 48 to 52 weeks in three studies (Verberne 1998a; Teper 2005; Lenney 2013).

Although co‐intervention with other prophylactic medications was permitted, trial protocols stipulated that their doses should remain unchanged throughout. The proportion of participants given additional therapy was not consistently reported. Permitted drugs included systemic steroids, anticholinergics and xanthines (Langton Hewer 1995), immunotherapy (Zimmerman 2004a; Zimmerman 2004b) and unspecified agents (Russell 1995). Other preventative medications were not permitted in the other studies except for Teper 2005, in which this was unspecified. Rescue medications such as inhaled SABA and systemic steroids were permitted in all studies.

Outcomes

The primary outcome ‐ the number of children with at least one exacerbation requiring systemic steroids ‐ was reported by 12 studies. When data were not reported, or were described only in a format we could not use directly, we asked study sponsors to provide further information, if possible. Our requests for data on exacerbations requiring rescue oral steroids for Pohunek 2006a; Pohunek 2006b; Stelmach 2007; Morice 2008a; Morice 2008b; SD 039 0718 and SD 039 0714 yielded no response.

Hospital admission data were available for seven studies. Measurement of lung function was reported in most studies. Many studies reported other secondary outcomes. Withdrawals were reported in all but five studies (Meijer 1995; Akpinarli 1999; Teper 2005; Stelmach 2008; Rutkowski 2009). Adverse events were reported in all studies except Stelmach 2008; Rutkowski 2009; Berger 2010; Carroll 2010; Lemanske 2010 and Lenney 2013.

LABA + ICS versus increased dose of ICS (step 3/step3)

A total of 11 studies, representing 1628 children, assessed the addition of LABA versus placebo to ICS therapy with increased dose of ICS in the control group (Ortega‐Cisneros 1998; Verberne 1998b; Heuck 2000; Bisgaard 2006; De Blic 2009; Gappa 2009; Lemanske 2010; Murray 2010; Vaessen‐Verberne 2010; SAM40100; SAM40012b). Three studies did not contribute data (Ortega‐Cisneros 1998; Heuck 2000; Lemanske 2010).

Participants

The mean age of participants was 10 years and 64% were male. Baseline airway obstruction was reported in seven of the 11 studies (Verberne 1998b; Heuck 2000; Bisgaard 2006: Gappa 2009; Lemanske 2010; Murray 2010; Vaessen‐Verberne 2010). Mean FEV1 % predicted at baseline was ≥ 80% in five control‐intervention comparisons (Verberne 1998b; Gappa 2009; Lemanske 2010; Murray 2010; Vaessen‐Verberne 2010).

Interventions

Salmeterol and formoterol were evaluated in eight and three studies, respectively. All comparisons tested the usual recommended dose of the LABA (i.e. salmeterol 50 µg twice daily, formoterol 6 or 12 µg twice daily). Intervention groups in eight studies received BDP equivalent doses of 400 µg/d (Verberne 1998b; De Blic 2009; Gappa 2009; Lemanske 2010; Murray 2010; Vaessen‐Verberne 2010; SAM40100; SAM40012b). BDP at 100 µg/d was used in Bisgaard 2006, and 200 µg/d was used in Heuck 2000. Respective control groups received twice the dose of ICS administered to the intervention group. Four studies assessed LABA and ICS as a single inhaler administration (Bisgaard 2006; De Blic 2009; SAM40100; SAM40012b).

Study duration was six to eight weeks in three studies (Gappa 2009; Murray 2010; SAM40100), 12 to 16 weeks in four studies (Heuck 2000; Bisgaard 2006; De Blic 2009; Lemanske 2010), 26 weeks in two comparisons (Vaessen‐Verberne 2010; SAM40012b), and one year in one study (Verberne 1998b).

All studies recruited children who were taking an ICS at baseline. Rescue medications such as inhaled SABA and systemic steroids were permitted in all trials.

Outcome data (obtaining data from trial authors)

Data on the primary outcome were available from three studies (Verberne 1998b; De Blic 2009; Vaessen‐Verberne 2010). When data were not reported, or were described for an undefined exacerbation or composite of types of exacerbations, we requested study sponsors to provide further information. Our requests for data on exacerbations requiring rescue oral corticosteroids (OCS) from study sponsors for Bisgaard 2006, SAM40012b and SAM40100 have not been successful.

Hospital admission data were available for four studies. Lung function outcomes were available for all studies. Most studies provided data on symptoms, SABA use, adverse events and withdrawals. Two studies provided data on linear growth (Verberne 1998b; Bisgaard 2006).

Excluded studies

Details of 85 excluded studies, for which full‐text articles were examined to judge eligibility, are listed in Characteristics of excluded studies along with reasons for their exclusion (this number is drawn from searches over all years January 2015 across this review).

Risk of bias in included studies

We have provided an overview of judgements on domains related to risk of bias in Figure 2. We have summarised our findings below.


Methodological quality summary: review authors' judgments about each methodological quality item for each included study.

Methodological quality summary: review authors' judgments about each methodological quality item for each included study.

We verified with study authors study details for six control‐intervention comparisons (Russell 1995; Simons 1997; Verberne 1998a; Verberne 1998b; Pohunek 2006a; Pohunek 2006b). Information on the design of GSK‐sponsored studies was provided in correspondence (Appendix 4). A total of 23 comparisons reported the randomisation technique in adequate detail, and we assessed remaining studies as unclear on the basis of inadequate reporting of the randomisation technique.

Allocation

Seventeen studies reported adequate details on allocation concealment of intervention treatment, and 22 provided unclear information.

Blinding

Double‐dummy designs or use of identical inhaler devices in 37 comparisons maintained blinding of the intervention. Two studies had an open‐label design (Ortega‐Cisneros 1998; Berger 2010).

Incomplete outcome data

Information on the definition of intention‐to‐treat principle used across studies was insufficient. Our judgement of this aspect of the studies reflects uncertainty over the reliability of stated methods. However, on the basis of our judgements, we designated 14 comparisons as high‐quality trials reporting complete outcomes, 23 as unclear and two as having high risk of bias (Russell 1995; Eid 2010a; Eid 2010b).

Selective reporting

We did not find major selective reporting bias in included studies. Availability of our prespecified primary outcome ‐ participants with exacerbations requiring rescue systemic steroids ‐ from trial reports was limited. This can be explained in part by the different definitions of exacerbation used by investigators across studies. Despite extensive efforts to obtain data for our primary outcome, we obtained a limited quantity of available data for analysis. We remain uncertain as to whether data for this endpoint were collected in nine studies (Langton Hewer 1995; Meijer 1995; Ortega‐Cisneros 1998; Tal 2002; Pohunek 2006a; Pohunek 2006b; Teper 2005; Stelmach 2007; SAM40100); 21 were at low risk, 14 at unclear risk and four at high risk.

Other potential sources of bias

In all, 23 studies were at low risk and 16 were at unclear risk.

Effects of interventions

See: Summary of findings for the main comparison ; Summary of findings 2

LABA + ICS versus same dose of ICS (step 3 vs step 2)

Primary outcome: participants with at least one exacerbation requiring systemic steroids

Investigators reported no statistically significant differences between treatments in the number of participants with exacerbations requiring systemic corticosteroids (12 studies; RR 0.95, 95% CI 0.70 to 1.28, N = 1669; Analysis 1.1; Figure 3).


Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.1 # participants with exacerbations requiring systemic steroids.

Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.1 # participants with exacerbations requiring systemic steroids.

Subgroup analysis

We performed subgroup analysis based on characteristics of participants and interventions to evaluate their influence on the magnitude of the primary outcome. Airway obstruction as determined by baseline mean FEV1 (Analysis 1.1), dose of ICS (Analysis 3.1), dose of LABA (Analysis 3.2), type of LABA (Analysis 3.3), use of single versus separate inhaler(s) to deliver LABA and ICS (Analysis 3.4) and trial duration (Analysis 3.5) did not influence the magnitude of response.

Sensitivity analysis

We performed a sensitivity analysis on the primary outcome. The primary outcome was robust and was not influenced by the funding source (Analysis 3.6) or by publication status (Analysis 3.7). All studies contributing data to the primary outcome were double‐blinded, thus preventing exclusion of unblinded trials.

Secondary outcomes
Hospital admission, urgent care visit, withdrawal

Researchers found no statistically significant differences in numbers of participants with exacerbations requiring hospital admission (seven studies, RR 1.74, 95% CI 0.90 to 3.36, N = 1292; Analysis 1.2; Figure 4), numbers of participants with exacerbations requiring urgent care visit (one study, RR 0.29, 95% CI 0.09 to 0.96, N = 186; Analysis 1.3), withdrawals due to poor asthma control (14 studies, RR 0.81, 95% CI 0.57 to 1.16, N = 2255; Analysis 1.6), withdrawals due to adverse events (18 studies, RR 0.79, 95% CI 0.52 to 1.21, N = 4117; Analysis 1.7) or withdrawals due to serious non‐respiratory events (two studies, RR 4.66, 95% CI 0.23 to 96.30; N = 318; Analysis 1.8). Withdrawals for any reason were significantly fewer with LABA than with placebo (23 studies, RR 0.80, 95% CI 0.67 to 0.94, N = 4374; Analysis 1.5).


Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.2 # participants with exacerbations requiring hospitalisation.

Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.2 # participants with exacerbations requiring hospitalisation.

Lung function

LABA added to ICS provided significantly greater improvement in lung function from baseline compared with the same dose of ICS. This was true, irrespective of whether group differences were reported for FEV1 as change in litres (inverse variance (IV) 0.08 L, 95% CI 0.06 to 0.10, nine studies, N = 1942; Analysis 1.9), change in % predicted (MD 2.99%, 95% CI 0.86 to 5.11; seven studies, N = 534; Analysis 1.10) or change in morning PEF (MD 10.20 L/min, 95% CI 8.14 to 12.26, 16 studies, N = 3934; Analysis 1.12; and one study, MD 3.77%, 95% CI 1.84 to 5.70, N = 286; Analysis 1.13) or evening PEF (MD 9.30 L/min, 95% CI 6.96 to 11.65, 12 studies, N = 3140; Analysis 1.14; and MD 3.40%, 95% CI 1.54 to 5.26, one study, N = 286; Analysis 1.15). The change in clinic PEF (Analysis 1.16) and the variability in PEF (Analysis 1.17) could not be aggregated, as they were reported by a single trial. Studies contributing data to lung function endpoints recruited children with mild to moderate airway obstruction, with a range of lung function at baseline; they examined the effects of both salmeterol and formoterol, given in conjunction with a range of doses of ICS. The % fall in FEV1 % predicted due to exercise showed no significant group difference (three studies, MD 0.46%, 95% CI ‐1.00 to 1.93, N = 517; Analysis 1.11).

Symptoms, rescue SABA use and quality of life

LABA added to ICS resulted in significant group differences for the following outcomes: change in daytime use of rescue SABA (MD ‐0.07 puffs/d, 95% CI ‐0.11 to ‐0.02, seven studies, N = 1798; Analysis 1.23) and change in nighttime use of rescue SABA (MD ‐0.08 puffs/d, 95% CI ‐0.13 to ‐0.03, three studies, N = 672; Analysis 1.24).

The addition of LABA did not result in significant group differences for the following outcomes: change in mean symptom scores (SMD ‐0.07, 95% CI ‐0.17 to 0.04, six studies, N = 1653; Analysis 1.18), change in nighttime symptom scores (two studies, MD ‐0.03, 95% CI ‐0.07 to 0.02, N = 534; Analysis 1.19), change in % symptom‐free days at endpoint (MD 0.96, 95% CI ‐1.91 to 3.84, seven studies, N = 1831; Analysis 1.20 ), % symptom‐free days (MD ‐0.04, 95% CI ‐0.20 to 0.12, four studies, N = 623; Analysis 1.21), % symptom‐free nights (MD 0.00, 95% CI ‐2.38 to 2.38, one study, N = 82; Analysis 1.22 ), % days without bronchodilator use (MD 2.07, 95% CI ‐1.03 to 5.16, seven studies, N = 1710, random‐effects model; Analysis 1.25), change in nighttime awakening (number of nights) (MD 0.20, 95% CI ‐2.21 to 2.61, one study, N = 286; Analysis 1.26), % nights with awakening (MD ‐1.10, 95% CI ‐3.51 to 1.31, one study, N = 286; Analysis 1.27), % change in awakening‐free nights (MD 0.60, 95% CI ‐1.05 to 2.26, N = 519; Analysis 1.28), change in rescue medication‐free days (two studies, MD ‐2.20, 95% CI ‐12.15 to 7.75, two studies, N = 231; Analysis 1.29), % change in asthma control days (MD 4.30, 95% CI ‐5.56 to 9.16, two studies, N = 519; Analysis 1.30), change in paediatric asthma quality of life (MD ‐0.02, 95% CI ‐0.14 to 0.10, four studies, N = 668; Analysis 1.31), absolute paediatric asthma quality of life (MD 0.03, 95% CI ‐0.04 to 0.11, 10 studies, N = 2333; Analysis 1.32) and change in paediatric asthma caregiver quality of life (MD 0.07, 95% CI ‐0.05 to 0.18; four studies, N = 669; Analysis 1.33).

Adverse events

There was no statistically significant differences in risk of overall adverse effects (RR 1.04, 95% 0.98 to 1.10, 15 studies, N = 3284; Analysis 1.34), reaching our a priori defined criterion for equivalence. However, for specific adverse effects, confidence intervals are wide, so we cannot rule out differences in any of these specific events. Specifically, there was no significant group differences in risk of oral candidiasis (RR 3.41, 95% CI 0.73 to 15.87, six studies, N = 1341; Analysis 1.35), tremor (RR 3.07, 95% CI 0.38 to 25.05, six studies, N = 1467; Analysis 1.36), palpitations (RR 0.44, 95% CI 0.08 to 2.31, six studies, N = 1238; Analysis 1.37), headache (RR 1.10, 95% CI 0.90 to 1.33, 14 studies, N = 2966; Analysis 1.38), vomiting (RR 0.74, 95% CI 0.34 to 1.62, three studies, N = 707; Analysis 1.39), otitis media (RR 0.70, 95% CI 0.30 to 1.63, three studies, N = 707; Analysis 1.40), upper respiratory tract infection (RR 0.86, 95% CI 0.58 to 1.27, five studies, N = 1186; Analysis 1.41), urticaria (RR 0.11, 95% CI 0.01 to 2.04, one study, N = 248; Analysis 1.42), cardiovascular adverse events (RR 0.31, 95% CI 0.01 to 7.49, two studies, N = 148; Analysis 1.43) and serious adverse events (RR 1.17, 95% CI 0.75 to 1.85, 17 studies, N = 4021; Analysis 1.4), Although effects on growth could not be aggregated because only one study documented this outcome (Verberne 1998a), data show no statistically significant group differences in growth velocity over 52 weeks among prepubertal children (mean age 10 to 11 years) when beclomethasone 400 μg with salmeterol was compared with beclomethasone 400 μg alone (5.1 cm vs 4.5 cm, respectively; Analysis 1.47). In three studies that recorded mortality, no deaths were mentioned (Analysis 1.44).

LABA and ICS versus increased dose of ICS (step 3 vs step 3)

Eight studies on 1520 participants contributed data to outcomes under this comparison (Verberne 1998b; Bisgaard 2006; De Blic 2009; Gappa 2009; Vaessen‐Verberne 2010; Murray 2010; SAM40100; SAM40012b).

Primary outcome: participants with at least one exacerbation requiring systemic steroids

Despite correspondence with study sponsors to obtain data on exacerbations requiring rescue oral steroids, we obtained data from only three studies (Verberne 1998b; De Blic 2009; Vaessen‐Verberne 2010). There was no significant group differences in the number of participants with exacerbations requiring OCS (RR 1.69, 95% CI 0.85 to 3.32, three studies, N = 581; Analysis 2.1; Figure 5). All trials contributing data to the primary outcome recruited participants with mild airway obstruction (FEV1 % predicted ≥ 80%).


Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.1 # participants with exacerbations requiring oral steroids.

Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.1 # participants with exacerbations requiring oral steroids.

Subgroup analysis

We performed subgroup analysis to evaluate the potential influence of characteristics of participants and interventions on the magnitude of the primary outcome. Dose of ICS (Analysis 4.1), dose of LABA (Analysis 4.2), type of LABA (Analysis 4.3), use of single versus separate inhaler(s) to deliver LABA and ICS (Analysis 4.4) and trial duration (Analysis 4.5) did not influence the magnitude of response.

Sensitivity analysis

We performed sensitivity analysis by including data from Lemanske 2010, which was a cross‐over study that reported data on the number of participants with exacerbations requiring oral corticosteroids and contributed the greatest weight by including the largest number of participants. There was no significant group difference in numbers of participants with exacerbations requiring OCS (RR 0.93, 95% CI 0.64 to 1.33, four studies, N = 895; Analysis 5.1). We were not able to perform the other sensitivity analysis, as all trials contributing data on the primary outcome were funded by producers of LABA and ICS, were published as full‐text articles and were double‐blinded.

Secondary outcomes
Hospital admission, urgent care visit, withdrawal

There was no significant group difference in the number of participants with exacerbations requiring hospital admission (RR 1.90, 95% CI 0.65 to 5.54, four studies, N = 1008; Analysis 2.2; Figure 6) or an urgent care visit (RR 5.13, 95% CI 0.25 to 105.10, one study, N = 158; Analysis 2.3). Data also showed no statistically significant group differences in overall risk of all‐cause withdrawals (RR 0.96, 95% CI 0.67 to 1.37, eight studies, N = 1491; Analysis 2.5), withdrawals due to poor asthma control (RR 0.34, 95% CI 0.05 to 2.13, four studies, N = 862; Analysis 2.6) and withdrawals due to adverse events (RR 0.76, 95% CI 0.19 to 3.07, five studies, N = 951; Analysis 2.7). Only one trial reported withdrawals due to a serious non‐respiratory event, none of which occurred (Analysis 2.8), thus preventing aggregation.


Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.2 # participants with exacerbations requiring hospitalisation.

Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.2 # participants with exacerbations requiring hospitalisation.

Lung function

LABA added to ICS led to greater improvement from baseline in the change in morning PEF (MD 8.73 L/min, 95% CI 5.15 to 12.31, five studies, N = 1283; Analysis 2.11) and evening PEF (MD 6.5 L/min, 95% CI 2.64 to 10.37, four studies, N = 1163; Analysis 2.12). Similarly, data showed a significant group difference in change in PEF recorded at clinic visit at the end point (MD 8.33 L/min, 95% CI 2.12 to 14.54, two studies, N = 637; Analysis 2.13). Changes from baseline in FEV1 between treatment options were not statistically significant (MD 0.01 L, 95% CI ‐0.03 to 0.05, two trials, N = 526; Analysis 2.9; and MD 0.38%, 95% CI ‐0.39 to 1.15, two trials, N = 682; Analysis 2.10). Data were insufficient for pooling of other lung function data.

Symptoms and SABA

There was no statistically significant group difference in change in daytime asthma symptom score (MD 0.01 L, 95% CI ‐0.20 to 0.23, three studies, N = 329; Analysis 2.17) and change in nighttime asthma symptom score (MD 0.01 L, 95% CI ‐0.20 to 0.23, three studies, N = 329; Analysis 2.18). Studies were insufficient for aggregation of other data related to markers of symptoms and use of rescue SABA.

Adverse events

There was no statistically significant difference in risk of overall adverse effects (RR 1.01, 95% CI 0.92 to 1.10, seven studies, N = 1254; Analysis 2.28), meeting our a priori criteria for equivalence. However, the specific adverse events have wide confidence intervals, so we cannot rule out a difference in any of these specific events. Data show no significant group difference in risk of oral candidiasis (RR 0.76, 95% CI 0.17 to 3.30, three studies, N = 182; Analysis 2.29), headache (RR 1.13, 95% CI 0.85 to 1.50, five studies, N = 1230; Analysis 2.30) and serious adverse events (RR 1.54, 95% CI 0.81 to 2.94, seven studies, N = 1343; Analysis 2.4). Other adverse events including vomiting, cold, upper respiratory tract infection and death could not be aggregated because trials reporting data are lacking. Two studies measured linear growth over one year (Verberne 1998b; Bisgaard 2006); findings favoured LABA treatment for children by an average of 1.21 cm/y (95% CI 0.72 to 1.7; Analysis 2.34).

Discussion

Summary of main results

Exacerbations requiring systemic steroids in school‐aged children with inadequately controlled asthma despite the use of daily low‐dose ICS were not significantly reduced by adding LABA to ICS compared with using the same, or an increased, dose of ICS. A priori defined subgroup analysis indicates that characteristics of participants or of the intervention did not influence outcomes.

Although not statistically significant, a trend towards increased risk of exacerbations requiring hospital admissions was noted in children treated with combination therapy. This trend towards increased risk of hospital admission with the addition of LABA compared with the same dose of ICS (step 2) or an increased dose of ICS (step 3) is a matter of concern, particularly as combination therapy failed to show any benefit in reducing severity markers including the primary outcome, that is, exacerbation requiring systemic steroids.

With regard to secondary outcomes, lung function endpoints consistently favoured the addition of LABA to ICS therapy, whether compared with the same or an increased dose of ICS. A modest reduction in the use of rescue SABA was evident with the addition of LABA to ICS when compared with the same dose of ICS; however, data were insufficient to pool for comparison with an increased ICS dose. In contrast, LABA added to ICS did not result in significantly greater improvement in asthma symptoms compared with the same or an increased dose of ICS. With the exception of growth, data show no statistically significant group differences in reported adverse events; overall adverse events met our definition of equivalence. The combination of LABA and ICS led to greater gain in linear growth than was seen with an increased dose of ICS; this is consistent with recent findings of a dose‐response effect of ICS on growth in children (Pruteanu 2014).

In this paediatric review, LABA added to ICS did not result in improvement in most other clinical indicators of asthma control and future risk of exacerbations. We recognised that absence of a significant group difference in other clinical indicators of asthma control may be due to mild asthma severity among most participants. Yet, aggregation of the best available evidence to date provides little data to support the addition of LABA to ICS in children insufficiently controlled by ICS monotherapy.

Overall completeness and applicability of evidence

Although we identified several unpublished studies, we had only limited success in obtaining useable data for our primary outcome. We successfully obtained data for exacerbations requiring rescue systemic steroids and for hospital admissions for a small number of trials from a recent meta‐analysis of GSK‐sponsored trials (Bateman 2008). Few data in study reports are available as downloads from pharmaceutical company trial results registries.

Quality of the evidence

Overall we judged the quality of evidence to be moderate. Most outcomes showed wide confidence intervals, which led to downgrading of evidence quality to moderate. In a few outcomes for which open‐label studies contributed data, we further downgraded evidence quality to low.

Potential biases in the review process

These findings are consistent with findings of Bisgaard 2003. Of note, about half of the trials in our review were conducted in school‐aged children treated predominantly with ICS and LABA delivered in separate (rather than single) devices. Although adherence to ICS may have been suboptimal, we cannot speculate whether different results would have been obtained if most trials had used a single device to deliver LABA and ICS. Is it possible that ongoing inflammation associated with use of a lower dose of ICS or tachyphylaxis associated with prolonged use of LABA may be associated with more severe exacerbations with combination therapy. In light of the prevailing uncertainty and an FDA mandate, a large six‐month study, to evaluate safety and benefit of LABA and ICS (salmeterol and fluticasone), is ongoing in children of four to 11 years of age. Outcomes of the study will shed more light on the safety and benefits of this combination (NCT01462344).

Agreements and disagreements with other studies or reviews

Our findings contrast with some of the estimates derived from the systematic review of adult trials performing the same comparisons (Ducharme 2010). Indeed, when compared with a similar dose of ICS, LABA added to ICS reduces by 20% the risk for adults with exacerbations requiring systemic steroids (RR 0.77, 95% CI 0.68 to 0.88; Ducharme 2010). This was accompanied by notably greater improvement in lung function (170 mL in FEV1) and symptom‐free days (+ 17%) and a modest reduction in use of rescue SABA (‐0.7 puffs/d). Given the smaller lung volumes in children, the observed 80 mL greater improvement in FEV1 associated with LABA added to ICS in children may be of clinical importance. However, observed improvement in lung function was expected, given that LABA is a bronchodilator, and children were selected primarily on the basis of significant reversibility with SABA to confirm the diagnosis of asthma. This apparent discordance between outcomes may be due to a more rapid effect of LABA on lung function, which is more easily detectable in studies of short duration, whereas a longer period of follow‐up may be required to detect an effect on exacerbations, particularly among children with normal or near normal lung function.

With regard to the second comparison ‐ LABA and ICS versus a higher dose of ICS ‐ findings also differ from a Cochrane Review of studies in adults (Ducharme 2010a), which demonstrates a significant reduction associated with LABA in the risk of patients with exacerbations requiring rescue systemic steroids (RR 0.88, 95% CI 0.78 to 0.98). Despite identification of 11 studies, only four provided data for the primary outcome. A modest improvement (< 9 L/min) in morning and evening PEF, but not in FEV1, was associated with use of LABA compared with a higher ICS dose. Insufficient reporting prevented aggregation of most outcomes. However, the trend towards a higher proportion of exacerbations requiring hospital admission and serious adverse events in children using LABA in combination with ICS, compared with a high dose of ICS, is a matter of concern. Findings are consistent with an overview of Cochrane Reviews evaluating the safety of formoterol or salmeterol in children with asthma (Cates 2012), in which review authors reported an additional three children per 1000 who suffered a non‐fatal serious adverse event with combination therapy in comparison with ICS over three months. Meanwhile, available data are also insufficient to allow firm recommendations regarding the preference of increasing the ICS dose versus adding LABA to ICS as a step 3 strategy. One must weigh the greater linear growth (reported in only two trials with beclomethasone and budesonide ‐ molecules known to be associated with growth suppression) (Skoner 2000; CAMP Research Group 2012) and the improvement in PEF against the possible, but unproven, increased risk of greater severity of exacerbations associated with combination therapy.

Data show no group differences in adverse effects or withdrawals due to adverse effects when the combination of LABA and ICS was compared with the step 2 or step 3 strategy. Of note, side effects were scarcely reported in short‐term trials, and long‐term studies were lacking. Moreover, although an increased dose of ICS calls for assessment of growth, adrenal function and bone mineralisation in children, no trial reported data on adrenal function and bone mineralisation that could be aggregated. Only two studies reporting the addition of LABA to 400 versus 800 μg of beclomethasone (Verberne 1998b) and to 100 versus 400 μg of budesonide (Bisgaard 2006) examined growth, for a differential of 300 to 400 μg of BDP‐equivalent. The observed reduction in growth averaging 1.2 cm/y is consistent with the documented decrease in linear growth associated with 400 μg/d of BDP (Sharek 1999; Pruteanu 2014) and the documented dose‐response relationship between growth impairment and ICS dose (Pruteanu 2014). Any apparent benefit of doubling the dose of ICS should be weighed against the possible impact on growth compared with other therapeutic regimens; it deserves careful evaluation (Pruteanu 2014).

Study flow diagram.
Figures and Tables -
Figure 1

Study flow diagram.

Methodological quality summary: review authors' judgments about each methodological quality item for each included study.
Figures and Tables -
Figure 2

Methodological quality summary: review authors' judgments about each methodological quality item for each included study.

Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.1 # participants with exacerbations requiring systemic steroids.
Figures and Tables -
Figure 3

Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.1 # participants with exacerbations requiring systemic steroids.

Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.2 # participants with exacerbations requiring hospitalisation.
Figures and Tables -
Figure 4

Forest plot of comparison: 1 LABA versus placebo: both groups receiving similar dose ICS, outcome: 1.2 # participants with exacerbations requiring hospitalisation.

Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.1 # participants with exacerbations requiring oral steroids.
Figures and Tables -
Figure 5

Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.1 # participants with exacerbations requiring oral steroids.

Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.2 # participants with exacerbations requiring hospitalisation.
Figures and Tables -
Figure 6

Forest plot of comparison: 2 LABA + ICS versus placebo + higher dose of ICS, outcome: 2.2 # participants with exacerbations requiring hospitalisation.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 1 # participants with exacerbations requiring systemic steroids.
Figures and Tables -
Analysis 1.1

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 1 # participants with exacerbations requiring systemic steroids.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 2 # participants with exacerbations requiring hospitalisation.
Figures and Tables -
Analysis 1.2

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 2 # participants with exacerbations requiring hospitalisation.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 3 # participants with exacerbations requiring urgent care visit.
Figures and Tables -
Analysis 1.3

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 3 # participants with exacerbations requiring urgent care visit.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 4 Serious adverse events.
Figures and Tables -
Analysis 1.4

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 4 Serious adverse events.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 5 Total # withdrawals.
Figures and Tables -
Analysis 1.5

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 5 Total # withdrawals.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 6 # withdrawals due to poor asthma control or exacerbation.
Figures and Tables -
Analysis 1.6

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 6 # withdrawals due to poor asthma control or exacerbation.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 7 # withdrawals due to adverse events.
Figures and Tables -
Analysis 1.7

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 7 # withdrawals due to adverse events.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 8 # withdrawals due to serious non‐respiratory event.
Figures and Tables -
Analysis 1.8

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 8 # withdrawals due to serious non‐respiratory event.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 9 Change in FEV1 (L) at endpoint.
Figures and Tables -
Analysis 1.9

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 9 Change in FEV1 (L) at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 10 Change in FEV1 at endpoint (% predicted) stratifying on baseline FEV1.
Figures and Tables -
Analysis 1.10

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 10 Change in FEV1 at endpoint (% predicted) stratifying on baseline FEV1.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 11 % fall in FEV1 % predicted due to exercise.
Figures and Tables -
Analysis 1.11

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 11 % fall in FEV1 % predicted due to exercise.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 12 Change in morning PEF (L/min) at endpoint.
Figures and Tables -
Analysis 1.12

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 12 Change in morning PEF (L/min) at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 13 Change in morning PEF (% predicted).
Figures and Tables -
Analysis 1.13

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 13 Change in morning PEF (% predicted).

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 14 Change in evening PEF (L/min) at endpoint.
Figures and Tables -
Analysis 1.14

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 14 Change in evening PEF (L/min) at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 15 Change in evening PEF (% of predicted).
Figures and Tables -
Analysis 1.15

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 15 Change in evening PEF (% of predicted).

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 16 Change in clinic PEF (L/min).
Figures and Tables -
Analysis 1.16

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 16 Change in clinic PEF (L/min).

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 17 Change in PEF variability at endpoint.
Figures and Tables -
Analysis 1.17

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 17 Change in PEF variability at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 18 Mean change in asthma symptom score.
Figures and Tables -
Analysis 1.18

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 18 Mean change in asthma symptom score.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 19 Change in nighttime symptom score.
Figures and Tables -
Analysis 1.19

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 19 Change in nighttime symptom score.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 20 Change in % symptom‐free days at endpoint.
Figures and Tables -
Analysis 1.20

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 20 Change in % symptom‐free days at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 21 % symptom‐free days.
Figures and Tables -
Analysis 1.21

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 21 % symptom‐free days.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 22 % symptom‐free nights at 52 ± 4 weeks.
Figures and Tables -
Analysis 1.22

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 22 % symptom‐free nights at 52 ± 4 weeks.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 23 Change in # daytime rescue inhalations (puffs per day) at endpoint.
Figures and Tables -
Analysis 1.23

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 23 Change in # daytime rescue inhalations (puffs per day) at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 24 Change in # nighttime rescue inhalations at endpoint.
Figures and Tables -
Analysis 1.24

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 24 Change in # nighttime rescue inhalations at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 25 % days without bronchodilator usage.
Figures and Tables -
Analysis 1.25

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 25 % days without bronchodilator usage.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 26 Change in nighttime awakening (number of nights) at endpoint.
Figures and Tables -
Analysis 1.26

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 26 Change in nighttime awakening (number of nights) at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 27 % nights with awakening.
Figures and Tables -
Analysis 1.27

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 27 % nights with awakening.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 28 % change in awakening‐free nights.
Figures and Tables -
Analysis 1.28

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 28 % change in awakening‐free nights.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 29 Change in rescue‐free days (%).
Figures and Tables -
Analysis 1.29

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 29 Change in rescue‐free days (%).

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 30 Change in % asthma‐control days at endpoint.
Figures and Tables -
Analysis 1.30

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 30 Change in % asthma‐control days at endpoint.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 31 Change in quality of life (P‐AQLQ).
Figures and Tables -
Analysis 1.31

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 31 Change in quality of life (P‐AQLQ).

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 32 Quality of life (P‐AQLQ).
Figures and Tables -
Analysis 1.32

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 32 Quality of life (P‐AQLQ).

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 33 Change in paediatric asthma caregiver quality of life (P‐AQLQ).
Figures and Tables -
Analysis 1.33

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 33 Change in paediatric asthma caregiver quality of life (P‐AQLQ).

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 34 Total # adverse events.
Figures and Tables -
Analysis 1.34

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 34 Total # adverse events.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 35 # participants with oral candidiasis.
Figures and Tables -
Analysis 1.35

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 35 # participants with oral candidiasis.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 36 # participants with tremor.
Figures and Tables -
Analysis 1.36

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 36 # participants with tremor.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 37 # participants with tachycardia or palpitations.
Figures and Tables -
Analysis 1.37

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 37 # participants with tachycardia or palpitations.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 38 # participants with headache.
Figures and Tables -
Analysis 1.38

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 38 # participants with headache.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 39 # participants with vomiting.
Figures and Tables -
Analysis 1.39

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 39 # participants with vomiting.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 40 # participants with otitis media.
Figures and Tables -
Analysis 1.40

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 40 # participants with otitis media.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 41 # participants with upper respiratory tract infection.
Figures and Tables -
Analysis 1.41

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 41 # participants with upper respiratory tract infection.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 42 # participants with urticaria.
Figures and Tables -
Analysis 1.42

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 42 # participants with urticaria.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 43 # participants with adverse cardiovascular events.
Figures and Tables -
Analysis 1.43

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 43 # participants with adverse cardiovascular events.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 44 Deaths.
Figures and Tables -
Analysis 1.44

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 44 Deaths.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 45 # participants with exacerbations requiring hospitalisation.
Figures and Tables -
Analysis 1.45

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 45 # participants with exacerbations requiring hospitalisation.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 46 Change in height (cm) as SD scores at 24 ± 4 weeks.
Figures and Tables -
Analysis 1.46

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 46 Change in height (cm) as SD scores at 24 ± 4 weeks.

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 47 Change in height at 1 year.
Figures and Tables -
Analysis 1.47

Comparison 1 LABA versus placebo: both groups receiving similar dose of ICS, Outcome 47 Change in height at 1 year.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 1 # participants with exacerbations requiring oral steroids.
Figures and Tables -
Analysis 2.1

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 1 # participants with exacerbations requiring oral steroids.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 2 # participants with exacerbations requiring hospitalisation.
Figures and Tables -
Analysis 2.2

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 2 # participants with exacerbations requiring hospitalisation.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 3 # participants with exacerbations requiring urgent care visit.
Figures and Tables -
Analysis 2.3

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 3 # participants with exacerbations requiring urgent care visit.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 4 Serious adverse events.
Figures and Tables -
Analysis 2.4

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 4 Serious adverse events.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 5 Total # withdrawals.
Figures and Tables -
Analysis 2.5

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 5 Total # withdrawals.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 6 # withdrawals due to poor asthma control or exacerbation.
Figures and Tables -
Analysis 2.6

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 6 # withdrawals due to poor asthma control or exacerbation.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 7 # withdrawals due to adverse events.
Figures and Tables -
Analysis 2.7

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 7 # withdrawals due to adverse events.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 8 # withdrawals due to serious non‐respiratory event.
Figures and Tables -
Analysis 2.8

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 8 # withdrawals due to serious non‐respiratory event.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 9 Change in FEV1 (L) at endpoint.
Figures and Tables -
Analysis 2.9

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 9 Change in FEV1 (L) at endpoint.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 10 Change in FEV1 % predicted at endpoint.
Figures and Tables -
Analysis 2.10

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 10 Change in FEV1 % predicted at endpoint.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 11 Change in morning PEF (L/min) at endpoint.
Figures and Tables -
Analysis 2.11

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 11 Change in morning PEF (L/min) at endpoint.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 12 Change in evening PEF (L/min) at endpoint.
Figures and Tables -
Analysis 2.12

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 12 Change in evening PEF (L/min) at endpoint.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 13 Change in clinic PEF (L/min).
Figures and Tables -
Analysis 2.13

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 13 Change in clinic PEF (L/min).

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 14 Change in morning PEF (% predicted) at endpoint.
Figures and Tables -
Analysis 2.14

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 14 Change in morning PEF (% predicted) at endpoint.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 15 Change in evening PEF (% predicted) at endpoint.
Figures and Tables -
Analysis 2.15

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 15 Change in evening PEF (% predicted) at endpoint.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 16 Change in % of days with a peak flow variability ≥ 20%.
Figures and Tables -
Analysis 2.16

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 16 Change in % of days with a peak flow variability ≥ 20%.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 17 Change in daytime asthma symptom score (mean over study period).
Figures and Tables -
Analysis 2.17

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 17 Change in daytime asthma symptom score (mean over study period).

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 18 Change in nighttime asthma symptom score (mean over study period).
Figures and Tables -
Analysis 2.18

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 18 Change in nighttime asthma symptom score (mean over study period).

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 19 Change in % of days without asthma symptoms.
Figures and Tables -
Analysis 2.19

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 19 Change in % of days without asthma symptoms.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 20 # daytime rescue inhalations (puffs per day; mean over study period).
Figures and Tables -
Analysis 2.20

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 20 # daytime rescue inhalations (puffs per day; mean over study period).

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 21 # nighttime rescue inhalations (puffs per day; mean over study period).
Figures and Tables -
Analysis 2.21

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 21 # nighttime rescue inhalations (puffs per day; mean over study period).

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 22 # daytime rescue inhalations at endpoint.
Figures and Tables -
Analysis 2.22

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 22 # daytime rescue inhalations at endpoint.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 23 Change in daytime rescue inhalations (puffs per day).
Figures and Tables -
Analysis 2.23

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 23 Change in daytime rescue inhalations (puffs per day).

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 24 Change in nighttime rescue inhalations (puffs per day).
Figures and Tables -
Analysis 2.24

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 24 Change in nighttime rescue inhalations (puffs per day).

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 25 Change in number of weeks with successful asthma control.
Figures and Tables -
Analysis 2.25

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 25 Change in number of weeks with successful asthma control.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 26 Change in % of days without salbutamol.
Figures and Tables -
Analysis 2.26

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 26 Change in % of days without salbutamol.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 27 Number of nighttime awakenings.
Figures and Tables -
Analysis 2.27

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 27 Number of nighttime awakenings.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 28 Total # adverse events.
Figures and Tables -
Analysis 2.28

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 28 Total # adverse events.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 29 # participants with oral candidiasis.
Figures and Tables -
Analysis 2.29

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 29 # participants with oral candidiasis.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 30 # participants with headache.
Figures and Tables -
Analysis 2.30

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 30 # participants with headache.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 31 # participants with vomiting.
Figures and Tables -
Analysis 2.31

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 31 # participants with vomiting.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 32 # participants with cold.
Figures and Tables -
Analysis 2.32

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 32 # participants with cold.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 33 # participants with upper respiratory tract infection.
Figures and Tables -
Analysis 2.33

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 33 # participants with upper respiratory tract infection.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 34 Linear growth.
Figures and Tables -
Analysis 2.34

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 34 Linear growth.

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 35 Deaths.
Figures and Tables -
Analysis 2.35

Comparison 2 LABA + ICS versus placebo + higher dose of ICS, Outcome 35 Deaths.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 1 # participants with exacerbations requiring oral steroids by dose of ICS in both groups.
Figures and Tables -
Analysis 3.1

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 1 # participants with exacerbations requiring oral steroids by dose of ICS in both groups.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 2 # participants with exacerbations requiring oral steroids by whether LABA dose is usual or higher than usual.
Figures and Tables -
Analysis 3.2

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 2 # participants with exacerbations requiring oral steroids by whether LABA dose is usual or higher than usual.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 3 # participants with exacerbations requiring oral steroids by type of LABA.
Figures and Tables -
Analysis 3.3

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 3 # participants with exacerbations requiring oral steroids by type of LABA.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 4 # participants with exacerbations requiring oral steroids by single inhaler or separate inhalers for LABA and ICS.
Figures and Tables -
Analysis 3.4

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 4 # participants with exacerbations requiring oral steroids by single inhaler or separate inhalers for LABA and ICS.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 5 # participants with exacerbations requiring oral steroids by trial duration.
Figures and Tables -
Analysis 3.5

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 5 # participants with exacerbations requiring oral steroids by trial duration.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 6 # participants with exacerbations requiring oral steroids by whether funded by producers of LABA.
Figures and Tables -
Analysis 3.6

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 6 # participants with exacerbations requiring oral steroids by whether funded by producers of LABA.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 7 # participants with exacerbations requiring systemic steroids by publication status.
Figures and Tables -
Analysis 3.7

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 7 # participants with exacerbations requiring systemic steroids by publication status.

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 8 # participants with exacerbations requiring systemic steroids by blinding of study.
Figures and Tables -
Analysis 3.8

Comparison 3 Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS), Outcome 8 # participants with exacerbations requiring systemic steroids by blinding of study.

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 1 # participants with exacerbations requiring oral steroids by dose of ICS in control groups.
Figures and Tables -
Analysis 4.1

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 1 # participants with exacerbations requiring oral steroids by dose of ICS in control groups.

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 2 # participants with exacerbations requiring oral steroids by whether LABA dose is usual or higher than usual.
Figures and Tables -
Analysis 4.2

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 2 # participants with exacerbations requiring oral steroids by whether LABA dose is usual or higher than usual.

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 3 # participants with exacerbations requiring oral steroids by type of LABA.
Figures and Tables -
Analysis 4.3

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 3 # participants with exacerbations requiring oral steroids by type of LABA.

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 4 # participants with exacerbations requiring oral steroids by single inhaler or separate inhalers for LABA and ICS.
Figures and Tables -
Analysis 4.4

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 4 # participants with exacerbations requiring oral steroids by single inhaler or separate inhalers for LABA and ICS.

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 5 # participants with exacerbations requiring oral steroids by trial duration.
Figures and Tables -
Analysis 4.5

Comparison 4 Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS), Outcome 5 # participants with exacerbations requiring oral steroids by trial duration.

Comparison 5 Sensitivity analysis: LABA + ICS versus placebo + higher dose of ICS, Outcome 1 # participants with exacerbations requiring oral steroids.
Figures and Tables -
Analysis 5.1

Comparison 5 Sensitivity analysis: LABA + ICS versus placebo + higher dose of ICS, Outcome 1 # participants with exacerbations requiring oral steroids.

LABA + ICS compared with same dose of ICS for children with chronic asthma

Patient or population: children with chronic asthma

Settings: outpatients

Intervention: LABA + ICS

Comparison: same dose of ICS

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

Number of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Increased dose of ICS

LABA + ICS

Number of participants with exacerbations requiring systemic steroids

86 per 1000

94 per 1000

RR 0.95
(0.70 to 1.28)

1669
(12 studies)

⊕⊕⊕⊝
Moderatea

Number of participants with exacerbations requiring hospitalisation

19 per 1000

33 per 1000

RR 1.74
(0.90 to 3.36)

1292
(6 studies)

⊕⊕⊕⊝
Moderatea

Serious adverse events

16 per 1000

18 per 1000

RR 1.17
(0.75 to 1.85)

4022
(16 studies)

⊕⊕⊕⊝
Moderatea

Total number of withdrawals

127 per 1000

94 per 1000

RR 0.80
(0.67 to 0.94)

4374
(23 studies)

⊕⊕⊝⊝
Lowa,b

Change in FEV1 (L) at endpoint

Baseline mean FEV1 ranged from 1.65 L to 1.9 L (baseline data reported in 4 studies only)

Mean FEV1 change from baseline with LABA + ICS was 0.08 L higher (0.06 to 0.1 higher)

1942 (9 studies)

⊕⊕⊝⊝
Lowa,b

Change in morning PEF (L/min) at endpoint

Illustrative post‐treatment PEFs range from 235 to 290 L/min (data from 3 recent studies)

Mean PEF change from baseline with LABA + ICS was 10.20 L/min higher (8.14 to 12.26 higher)

3934 (16 studies)

⊕⊕⊕⊝
Moderatea

Total number of adverse events

547 per 1000

568 per 1000

RR 1.04 (0.98 to 1.10)

3284
(15 studies)

⊕⊕⊕⊝
Moderateb

*The basis for the assumed risk (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; FEV1: Forced expiratory volume in 1 second;ICS: Inhaled corticosteroids; LABA: Long‐acting beta2‐agonists; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

aLarger sample size may change the outcome.

bOpen‐label study contributed data.

Figures and Tables -

LABA + ICS compared with increased dose of ICS for children with chronic asthma

Patient or population: children with chronic asthma

Settings: outpatients

Intervention: LABA + ICS

Comparison: increased dose of ICS

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

Number of participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Increased dose of ICS

LABA + ICS

Number of participants with exacerbations requiring systemic steroids

41 per 1000

69 per 1000

RR 1.69
(0.85 to 3.32)

581 (3 studies)

⊕⊕⊕⊝
Moderatea

Number of participants with exacerbations requiring hospitalisation

8 per 1000

18 per 1000

RR 1.90
(0.65 to 5.54)

1008
(4 studies)

⊕⊕⊕⊝
Moderatea

Serious adverse events

24 per 1000

39 per 1000

RR 1.54
(0.81 to 2.94)

1343
(7 studies)

⊕⊕⊕⊝
Moderatea

Total number of withdrawals

70 per 1000

67 per 1000

RR 0.96
(0.67 to 1.37)

1491
(7 studies)

⊕⊕⊕⊝
Moderatea

Change in FEV1 (L) at endpoint

Mean baseline FEV1 ranged from 1.6 to 1.7 L

Mean FEV1: change from baseline with LABA + ICS was 0.01 L higher (‐0.03 to 0.05 higher)

526 (2 studies)

⊕⊕⊕⊝
Moderatea

Change in morning PEF (L/min) at endpoint

Mean change in end of treatment PEF ranged from 16.7 to 39.2 L/min

Mean PEF: change from baseline with LABA + ICS was 8.73 L/min higher (5.15 to 12.31 higher)

1283 (5 studies)

⊕⊕⊕⊝
Moderatea

Total number of adverse events

569 per 1000

576 per 1000

RR 1.01 (0.92 to 1.10)

1254
(6 studies)

⊕⊕⊕⊕
High

*The basis for the assumed risk (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; FEV1: Forced expiratory volume in 1 second; ICS: Inhaled corticosteroids; LABA: Long‐acting beta2‐agonists; PEF: Peak expiratory flow; RR: Risk ratio.

GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

aLarger sample size may change the outcome.

Figures and Tables -
Comparison 1. LABA versus placebo: both groups receiving similar dose of ICS

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 # participants with exacerbations requiring systemic steroids Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

1.1 Mean baseline FEV1 ≥ 80% of predicted

9

1399

Risk Ratio (M‐H, Fixed, 95% CI)

0.97 [0.69, 1.37]

1.2 Mean baseline FEV1 61%‐79% of predicted

2

230

Risk Ratio (M‐H, Fixed, 95% CI)

0.89 [0.48, 1.64]

1.3 Mean baseline FEV1 not reported

1

40

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

2 # participants with exacerbations requiring hospitalisation Show forest plot

7

1292

Risk Ratio (M‐H, Fixed, 95% CI)

1.74 [0.90, 3.36]

2.1 Mean baseline FEV1 ≥ 80% of predicted

3

165

Risk Ratio (M‐H, Fixed, 95% CI)

1.50 [0.36, 6.14]

2.2 Mean baseline FEV1 61%‐79% of predicted

3

772

Risk Ratio (M‐H, Fixed, 95% CI)

1.81 [0.86, 3.82]

2.3 Mean baseline FEV1 not reported

1

355

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

3 # participants with exacerbations requiring urgent care visit Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

3.1 Mean baseline FEV1 ≥ 80% of predicted

1

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

4 Serious adverse events Show forest plot

17

4021

Risk Ratio (M‐H, Fixed, 95% CI)

1.17 [0.75, 1.85]

4.1 Mean baseline FEV1 ≥ 80% of predicted

13

2897

Risk Ratio (M‐H, Fixed, 95% CI)

1.01 [0.53, 1.92]

4.2 Mean baseline FEV1 61%‐79% of predicted

3

762

Risk Ratio (M‐H, Fixed, 95% CI)

1.33 [0.68, 2.59]

4.3 Mean baseline FEV1 not reported

1

362

Risk Ratio (M‐H, Fixed, 95% CI)

2.0 [0.18, 21.86]

5 Total # withdrawals Show forest plot

23

4374

Risk Ratio (M‐H, Fixed, 95% CI)

0.80 [0.67, 0.94]

5.1 Mean baseline FEV1 ≥ 80% of predicted

17

3205

Risk Ratio (M‐H, Fixed, 95% CI)

0.79 [0.66, 0.96]

5.2 Mean baseline FEV1 61%‐79% of predicted

4

794

Risk Ratio (M‐H, Fixed, 95% CI)

0.91 [0.63, 1.32]

5.3 Mean baseline FEV1 not reported

2

375

Risk Ratio (M‐H, Fixed, 95% CI)

0.36 [0.12, 1.11]

6 # withdrawals due to poor asthma control or exacerbation Show forest plot

14

2255

Risk Ratio (M‐H, Fixed, 95% CI)

0.81 [0.57, 1.16]

6.1 Mean baseline FEV1 ≥ 80% of predicted

10

1461

Risk Ratio (M‐H, Fixed, 95% CI)

0.82 [0.55, 1.21]

6.2 Mean baseline FEV1 61%‐79% of predicted

4

794

Risk Ratio (M‐H, Fixed, 95% CI)

0.78 [0.34, 1.81]

7 # withdrawals due to adverse events Show forest plot

18

4117

Risk Ratio (M‐H, Fixed, 95% CI)

0.79 [0.52, 1.21]

7.1 Mean baseline FEV1 ≥ 80% of predicted

13

3053

Risk Ratio (M‐H, Fixed, 95% CI)

0.69 [0.42, 1.13]

7.2 Mean baseline FEV1 61%‐79% of predicted

5

1064

Risk Ratio (M‐H, Fixed, 95% CI)

1.12 [0.51, 2.44]

8 # withdrawals due to serious non‐respiratory event Show forest plot

2

318

Risk Ratio (M‐H, Random, 95% CI)

4.66 [0.23, 96.30]

8.1 Mean baseline FEV1 61%‐79% of predicted

2

318

Risk Ratio (M‐H, Random, 95% CI)

4.66 [0.23, 96.30]

9 Change in FEV1 (L) at endpoint Show forest plot

9

1942

Litres (Fixed, 95% CI)

0.08 [0.06, 0.10]

9.1 Mean baseline FEV1 ≥ 80% of predicted

7

1515

Litres (Fixed, 95% CI)

0.08 [0.05, 0.10]

9.2 Mean baseline FEV1 61%‐79% of predicted

2

427

Litres (Fixed, 95% CI)

0.12 [0.04, 0.19]

10 Change in FEV1 at endpoint (% predicted) stratifying on baseline FEV1 Show forest plot

7

534

Mean Difference (IV, Fixed, 95% CI)

2.99 [0.86, 5.11]

10.1 Mean baseline FEV1 ≥ 80% of predicted

4

214

Mean Difference (IV, Fixed, 95% CI)

4.06 [1.32, 6.80]

10.2 Mean baseline FEV1 61%‐79% of predicted

2

238

Mean Difference (IV, Fixed, 95% CI)

3.35 [‐1.50, 8.20]

10.3 Mean baseline FEV1 not reported

1

82

Mean Difference (IV, Fixed, 95% CI)

‐0.40 [‐5.03, 4.23]

11 % fall in FEV1 % predicted due to exercise Show forest plot

3

517

Mean Difference (IV, Fixed, 95% CI)

0.46 [1.00, 1.93]

11.1 Mean baseline FEV1 ≥ 80% of predicted

3

517

Mean Difference (IV, Fixed, 95% CI)

0.46 [1.00, 1.93]

12 Change in morning PEF (L/min) at endpoint Show forest plot

16

3934

L/min (Fixed, 95% CI)

10.20 [8.14, 12.26]

12.1 Mean baseline FEV1 ≥ 80% of predicted

12

2870

L/min (Fixed, 95% CI)

10.79 [8.43, 13.16]

12.2 Mean baseline FEV1 61%‐79% of predicted

3

713

L/min (Fixed, 95% CI)

9.26 [4.42, 14.10]

12.3 Mean baseline FEV1 not reported

1

351

L/min (Fixed, 95% CI)

5.7 [‐2.62, 14.02]

13 Change in morning PEF (% predicted) Show forest plot

1

% (Fixed, 95% CI)

Totals not selected

13.1 Mean baseline FEV1 61%‐79% of predicted

1

% (Fixed, 95% CI)

0.0 [0.0, 0.0]

14 Change in evening PEF (L/min) at endpoint Show forest plot

12

3140

L/min (Fixed, 95% CI)

9.30 [6.96, 11.65]

14.1 Mean baseline FEV1 ≥ 80% of predicted

10

2503

L/min (Fixed, 95% CI)

9.31 [6.67, 11.94]

14.2 Mean baseline FEV1 61%‐79% of predicted

1

286

L/min (Fixed, 95% CI)

11.7 [5.29, 18.11]

14.3 Mean baseline FEV1 not reported

1

351

L/min (Fixed, 95% CI)

5.0 [‐3.58, 13.58]

15 Change in evening PEF (% of predicted) Show forest plot

1

% (Fixed, 95% CI)

Totals not selected

15.1 Mean baseline FEV1 61%‐79% of predicted

1

% (Fixed, 95% CI)

0.0 [0.0, 0.0]

16 Change in clinic PEF (L/min) Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

16.1 Mean baseline FEV1 not reported

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

17 Change in PEF variability at endpoint Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

17.1 Mean baseline FEV1 61%‐79% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

18 Mean change in asthma symptom score Show forest plot

6

1653

Std. Mean Difference (IV, Fixed, 95% CI)

‐0.07 [‐0.17, 0.04]

18.1 Mean baseline FEV1 ≥ 80% of predicted

6

1653

Std. Mean Difference (IV, Fixed, 95% CI)

‐0.07 [‐0.17, 0.04]

19 Change in nighttime symptom score Show forest plot

2

534

Mean Difference (IV, Fixed, 95% CI)

‐0.03 [‐0.07, 0.02]

19.1 Mean baseline FEV1 ≥ 80% of predicted

2

534

Mean Difference (IV, Fixed, 95% CI)

‐0.03 [‐0.07, 0.02]

20 Change in % symptom‐free days at endpoint Show forest plot

7

1831

Mean Difference (IV, Fixed, 95% CI)

0.96 [‐1.91, 3.84]

20.1 Mean baseline FEV1 ≥ 80% of predicted

7

1831

Mean Difference (IV, Fixed, 95% CI)

0.96 [‐1.91, 3.84]

21 % symptom‐free days Show forest plot

4

623

Std. Mean Difference (IV, Fixed, 95% CI)

‐0.04 [‐0.20, 0.12]

21.1 Mean baseline FEV1 ≥ 80% of predicted

1

231

Std. Mean Difference (IV, Fixed, 95% CI)

‐0.21 [‐0.47, 0.05]

21.2 Mean baseline FEV1 61%‐79% of predicted

1

286

Std. Mean Difference (IV, Fixed, 95% CI)

0.12 [‐0.11, 0.35]

21.3 Unclear baseline FEV1

2

106

Std. Mean Difference (IV, Fixed, 95% CI)

‐0.12 [‐0.56, 0.31]

22 % symptom‐free nights at 52 ± 4 weeks Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

22.1 Mean baseline FEV1 not reported

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

23 Change in # daytime rescue inhalations (puffs per day) at endpoint Show forest plot

7

1798

Mean Difference (IV, Fixed, 95% CI)

‐0.07 [‐0.11, ‐0.02]

23.1 Mean baseline FEV1 61%‐79% of predicted

7

1798

Mean Difference (IV, Fixed, 95% CI)

‐0.07 [‐0.11, ‐0.02]

24 Change in # nighttime rescue inhalations at endpoint Show forest plot

3

672

Mean Difference (IV, Random, 95% CI)

‐0.08 [‐0.13, ‐0.03]

24.1 Mean baseline FEV1 61%‐79% of predicted

3

672

Mean Difference (IV, Random, 95% CI)

‐0.08 [‐0.13, ‐0.03]

25 % days without bronchodilator usage Show forest plot

7

1710

Mean Difference (IV, Random, 95% CI)

2.07 [‐1.03, 5.16]

25.1 Mean baseline FEV1 ≥ 80% of predicted

6

1628

Mean Difference (IV, Random, 95% CI)

2.87 [‐0.44, 6.18]

25.2 Mean baseline FEV1 not reported

1

82

Mean Difference (IV, Random, 95% CI)

‐1.0 [‐4.04, 2.04]

26 Change in nighttime awakening (number of nights) at endpoint Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

26.1 Mean baseline FEV1 61%‐79% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

27 % nights with awakening Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

27.1 Mean baseline FEV1 61%‐79% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

28 % change in awakening‐free nights Show forest plot

2

519

Mean Difference (IV, Fixed, 95% CI)

0.60 [‐1.05, 2.26]

28.1 Mean baseline FEV1 ≥ 80% of predicted

2

519

Mean Difference (IV, Fixed, 95% CI)

0.60 [‐1.05, 2.26]

29 Change in rescue‐free days (%) Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

29.1 Mean baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

30 Change in % asthma‐control days at endpoint Show forest plot

2

519

Mean Difference (IV, Fixed, 95% CI)

4.30 [‐0.56, 9.16]

30.1 Mean baseline FEV1 ≥ 80% of predicted

2

519

Mean Difference (IV, Fixed, 95% CI)

4.30 [‐0.56, 9.16]

31 Change in quality of life (P‐AQLQ) Show forest plot

4

668

Mean Difference (IV, Fixed, 95% CI)

‐0.02 [‐0.14, 0.10]

31.1 Mean baseline FEV1 ≥ 80% of predicted

4

668

Mean Difference (IV, Fixed, 95% CI)

‐0.02 [‐0.14, 0.10]

32 Quality of life (P‐AQLQ) Show forest plot

10

2333

Mean Difference (IV, Fixed, 95% CI)

0.03 [‐0.04, 0.11]

32.1 Mean baseline FEV1 ≥ 80% of predicted

6

1586

Mean Difference (IV, Fixed, 95% CI)

0.07 [‐0.01, 0.16]

32.2 Mean baseline FEV1 61%‐79% of predicted

4

747

Mean Difference (IV, Fixed, 95% CI)

‐0.12 [‐0.28, 0.03]

33 Change in paediatric asthma caregiver quality of life (P‐AQLQ) Show forest plot

4

669

Mean Difference (IV, Fixed, 95% CI)

0.07 [‐0.05, 0.18]

33.1 Mean baseline FEV1 ≥ 80% of predicted

4

669

Mean Difference (IV, Fixed, 95% CI)

0.07 [‐0.05, 0.18]

34 Total # adverse events Show forest plot

15

3284

Risk Ratio (M‐H, Fixed, 95% CI)

1.04 [0.98, 1.10]

34.1 Mean baseline FEV1 ≥ 80% of predicted

11

2424

Risk Ratio (M‐H, Fixed, 95% CI)

1.08 [1.00, 1.16]

34.2 Mean baseline FEV1 61%‐79% of predicted

2

474

Risk Ratio (M‐H, Fixed, 95% CI)

0.97 [0.86, 1.09]

34.3 Mean baseline FEV1 not reported

2

386

Risk Ratio (M‐H, Fixed, 95% CI)

0.92 [0.78, 1.09]

35 # participants with oral candidiasis Show forest plot

6

1341

Risk Ratio (M‐H, Fixed, 95% CI)

3.41 [0.73, 15.87]

35.1 FEV1 ≥ 80% of predicted

5

1135

Risk Ratio (M‐H, Fixed, 95% CI)

3.46 [0.60, 19.99]

35.2 Mean baseline FEV1 61%‐79% of predicted

1

206

Risk Ratio (M‐H, Fixed, 95% CI)

3.24 [0.13, 78.62]

36 # participants with tremor Show forest plot

6

1467

Risk Ratio (M‐H, Fixed, 95% CI)

3.07 [0.38, 25.05]

36.1 Mean baseline FEV1 ≥ 80% of predicted

3

959

Risk Ratio (M‐H, Fixed, 95% CI)

5.30 [0.26, 109.66]

36.2 Mean baseline FEV1 61%‐79% of predicted

3

508

Risk Ratio (M‐H, Fixed, 95% CI)

1.46 [0.06, 35.18]

37 # participants with tachycardia or palpitations Show forest plot

6

1238

Risk Ratio (M‐H, Fixed, 95% CI)

0.44 [0.08, 2.31]

37.1 Mean baseline FEV1 ≥ 80% of predicted

3

731

Risk Ratio (M‐H, Fixed, 95% CI)

0.41 [0.05, 3.33]

37.2 Mean baseline FEV1 61%‐79% of predicted

3

507

Risk Ratio (M‐H, Fixed, 95% CI)

0.49 [0.03, 7.61]

38 # participants with headache Show forest plot

14

2966

Risk Ratio (M‐H, Fixed, 95% CI)

1.10 [0.90, 1.33]

38.1 Mean baseline FEV1 ≥ 80% of predicted

8

1779

Risk Ratio (M‐H, Fixed, 95% CI)

1.00 [0.79, 1.26]

38.2 Mean baseline FEV1 61%‐79% of predicted

5

825

Risk Ratio (M‐H, Fixed, 95% CI)

1.34 [0.88, 2.04]

38.3 Mean baseline FEV1 not reported

1

362

Risk Ratio (M‐H, Fixed, 95% CI)

1.4 [0.64, 3.07]

39 # participants with vomiting Show forest plot

3

707

Risk Ratio (M‐H, Fixed, 95% CI)

0.74 [0.34, 1.62]

39.1 Mean baseline FEV1 ≥ 80% of predicted

3

707

Risk Ratio (M‐H, Fixed, 95% CI)

0.74 [0.34, 1.62]

40 # participants with otitis media Show forest plot

3

707

Risk Ratio (M‐H, Fixed, 95% CI)

0.70 [0.30, 1.63]

40.1 Mean baseline FEV1 ≥ 80% of predicted

3

707

Risk Ratio (M‐H, Fixed, 95% CI)

0.70 [0.30, 1.63]

41 # participants with upper respiratory tract infection Show forest plot

5

1186

Risk Ratio (M‐H, Fixed, 95% CI)

0.86 [0.58, 1.27]

41.1 Mean baseline FEV1 ≥ 80% of predicted

5

1186

Risk Ratio (M‐H, Fixed, 95% CI)

0.86 [0.58, 1.27]

42 # participants with urticaria Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

42.1 Mean baseline FEV1 ≥ 80% of predicted

1

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

43 # participants with adverse cardiovascular events Show forest plot

2

148

Risk Ratio (M‐H, Fixed, 95% CI)

0.31 [0.01, 7.49]

43.1 Mean baseline FEV1 ≥ 80% of predicted

1

116

Risk Ratio (M‐H, Fixed, 95% CI)

0.31 [0.01, 7.49]

43.2 Mean baseline FEV1 61%‐79% of predicted

1

32

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

44 Deaths Show forest plot

3

690

Risk Difference (M‐H, Fixed, 95% CI)

0.0 [‐0.01, 0.01]

44.1 Baseline FEV1 ≥ 80% of predicted

3

690

Risk Difference (M‐H, Fixed, 95% CI)

0.0 [‐0.01, 0.01]

45 # participants with exacerbations requiring hospitalisation Show forest plot

7

1292

Risk Ratio (M‐H, Fixed, 95% CI)

1.74 [0.90, 3.36]

45.1 Mean baseline FEV1 ≥ 80% of predicted

3

165

Risk Ratio (M‐H, Fixed, 95% CI)

1.50 [0.36, 6.14]

45.2 Mean baseline FEV1 61%‐79% of predicted

3

772

Risk Ratio (M‐H, Fixed, 95% CI)

1.81 [0.86, 3.82]

45.3 Mean baseline FEV1 not reported

1

355

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

46 Change in height (cm) as SD scores at 24 ± 4 weeks Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

46.1 Mean baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

47 Change in height at 1 year Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

Figures and Tables -
Comparison 1. LABA versus placebo: both groups receiving similar dose of ICS
Comparison 2. LABA + ICS versus placebo + higher dose of ICS

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 # participants with exacerbations requiring oral steroids Show forest plot

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

1.1 Baseline FEV1 ≥ 80% of predicted

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

2 # participants with exacerbations requiring hospitalisation Show forest plot

4

1008

Risk Ratio (M‐H, Fixed, 95% CI)

1.90 [0.65, 5.54]

2.1 Baseline FEV1 ≥ 80% of predicted

2

423

Risk Ratio (M‐H, Fixed, 95% CI)

1.68 [0.22, 12.66]

2.2 Baseline FEV1 61%‐79% of predicted

1

225

Risk Ratio (M‐H, Fixed, 95% CI)

3.17 [0.67, 14.95]

2.3 Mean baseline FEV1 not reported

1

360

Risk Ratio (M‐H, Fixed, 95% CI)

0.33 [0.01, 8.13]

3 # participants with exacerbations requiring urgent care visit Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

3.1 Baseline FEV1 ≥ 80% of predicted

1

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

4 Serious adverse events Show forest plot

7

1343

Risk Ratio (M‐H, Random, 95% CI)

1.54 [0.81, 2.94]

4.1 Baseline FEV1 ≥ 80% predicted

4

729

Risk Ratio (M‐H, Random, 95% CI)

1.15 [0.42, 3.16]

4.2 Baseline FEV1 61%‐79% of predicted

1

223

Risk Ratio (M‐H, Random, 95% CI)

2.90 [1.10, 7.64]

4.3 Mean baseline FEV1 not reported

2

391

Risk Ratio (M‐H, Random, 95% CI)

0.51 [0.10, 2.77]

5 Total # withdrawals Show forest plot

8

1491

Risk Ratio (M‐H, Fixed, 95% CI)

0.96 [0.67, 1.37]

5.1 Baseline FEV1 ≥ 80% of predicted

5

888

Risk Ratio (M‐H, Fixed, 95% CI)

1.15 [0.75, 1.78]

5.2 Baseline FEV1 61%‐79% of predicted

1

223

Risk Ratio (M‐H, Fixed, 95% CI)

0.65 [0.30, 1.39]

5.3 Mean baseline FEV1 not reported

2

380

Risk Ratio (M‐H, Fixed, 95% CI)

0.68 [0.20, 2.36]

6 # withdrawals due to poor asthma control or exacerbation Show forest plot

4

862

Risk Ratio (M‐H, Fixed, 95% CI)

0.34 [0.05, 2.13]

6.1 Baseline FEV1 ≥ 80% of predicted

4

862

Risk Ratio (M‐H, Fixed, 95% CI)

0.34 [0.05, 2.13]

7 # withdrawals due to adverse events Show forest plot

5

951

Risk Ratio (M‐H, Fixed, 95% CI)

0.76 [0.19, 3.07]

7.1 Baseline FEV1 ≥ 80% of predicted

4

728

Risk Ratio (M‐H, Fixed, 95% CI)

0.72 [0.14, 3.63]

7.2 Baseline FEV1 61%‐79% of predicted

1

223

Risk Ratio (M‐H, Fixed, 95% CI)

0.91 [0.06, 14.30]

8 # withdrawals due to serious non‐respiratory event Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected

8.1 Mean baseline FEV1 ≥ 80% of predicted

1

Risk Ratio (M‐H, Random, 95% CI)

0.0 [0.0, 0.0]

9 Change in FEV1 (L) at endpoint Show forest plot

2

526

Mean Difference (Fixed, 95% CI)

0.01 [‐0.03, 0.05]

9.1 Baseline FEV1 ≥ 80% of predicted

1

303

Mean Difference (Fixed, 95% CI)

0.0 [‐0.05, 0.05]

9.2 Baseline FEV1 61%‐79% of predicted

1

223

Mean Difference (Fixed, 95% CI)

0.04 [‐0.06, 0.14]

10 Change in FEV1 % predicted at endpoint Show forest plot

2

% (Random, 95% CI)

0.38 [‐0.39, 1.15]

10.1 Baseline FEV1 ≥ 80% of predicted

2

% (Random, 95% CI)

0.38 [‐0.39, 1.15]

11 Change in morning PEF (L/min) at endpoint Show forest plot

5

1283

Mean Difference (IV, Fixed, 95% CI)

8.73 [5.15, 12.31]

11.1 Baseline FEV1 ≥ 80% of predicted

3

704

Mean Difference (IV, Fixed, 95% CI)

9.50 [5.07, 13.93]

11.2 Baseline FEV1 61%‐79% of predicted

1

223

Mean Difference (IV, Fixed, 95% CI)

9.0 [‐0.07, 18.07]

11.3 Baseline FEV1 not reported

1

356

Mean Difference (IV, Fixed, 95% CI)

5.90 [‐2.28, 14.08]

12 Change in evening PEF (L/min) at endpoint Show forest plot

4

1163

Mean Difference (Fixed, 95% CI)

6.50 [2.64, 10.37]

12.1 Baseline FEV1 predicted ≥ 80%

2

584

Mean Difference (Fixed, 95% CI)

7.08 [2.13, 12.02]

12.2 Baseline FEV1 61%‐79% of predicted

1

223

Mean Difference (Fixed, 95% CI)

7.0 [‐2.07, 16.07]

12.3 Baseline FEV1 not reported

1

356

Mean Difference (Fixed, 95% CI)

4.4 [‐4.05, 12.85]

13 Change in clinic PEF (L/min) Show forest plot

2

637

Mean Difference (IV, Fixed, 95% CI)

8.33 [2.12, 14.54]

13.1 Baseline FEV1 ≥ 80% of predicted

1

281

Mean Difference (IV, Fixed, 95% CI)

11.3 [3.84, 18.76]

13.2 Baseline FEV1 not reported

1

356

Mean Difference (IV, Fixed, 95% CI)

1.60 [‐9.63, 12.83]

14 Change in morning PEF (% predicted) at endpoint Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

14.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

15 Change in evening PEF (% predicted) at endpoint Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

15.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

16 Change in % of days with a peak flow variability ≥ 20% Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

16.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

17 Change in daytime asthma symptom score (mean over study period) Show forest plot

3

329

Std. Mean Difference (IV, Fixed, 95% CI)

0.01 [‐0.20, 0.23]

17.1 Baseline FEV1 ≥ 80% of predicted

2

305

Std. Mean Difference (IV, Fixed, 95% CI)

0.01 [‐0.21, 0.24]

17.2 Baseline FEV1 not reported

1

24

Std. Mean Difference (IV, Fixed, 95% CI)

0.03 [‐0.77, 0.84]

18 Change in nighttime asthma symptom score (mean over study period) Show forest plot

3

329

Std. Mean Difference (IV, Fixed, 95% CI)

0.01 [‐0.20, 0.23]

18.1 Baseline FEV1 ≥ 80% of predicted

2

305

Std. Mean Difference (IV, Fixed, 95% CI)

0.01 [‐0.21, 0.24]

18.2 Baseline FEV1 not reported

1

24

Std. Mean Difference (IV, Fixed, 95% CI)

0.00 [‐0.80, 0.80]

19 Change in % of days without asthma symptoms Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

19.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

20 # daytime rescue inhalations (puffs per day; mean over study period) Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

20.1 Baseline FEV1 not reported

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

21 # nighttime rescue inhalations (puffs per day; mean over study period) Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

21.1 Baseline FEV1 not reported

1

Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

22 # daytime rescue inhalations at endpoint Show forest plot

1

puffs/d (Fixed, 95% CI)

Totals not selected

22.1 Baseline FEV1 61%‐79% of predicted

1

puffs/d (Fixed, 95% CI)

0.0 [0.0, 0.0]

23 Change in daytime rescue inhalations (puffs per day) Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

23.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

24 Change in nighttime rescue inhalations (puffs per day) Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

24.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

25 Change in number of weeks with successful asthma control Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

25.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

26 Change in % of days without salbutamol Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

26.1 Baseline FEV1 ≥ 80% of predicted

1

Mean Difference (IV, Fixed, 95% CI)

0.0 [0.0, 0.0]

27 Number of nighttime awakenings Show forest plot

1

Awakenings/yr (Fixed, 95% CI)

Totals not selected

27.1 Baseline FEV1 61%‐79% of predicted

1

Awakenings/yr (Fixed, 95% CI)

0.0 [0.0, 0.0]

28 Total # adverse events Show forest plot

6

1254

Risk Ratio (M‐H, Fixed, 95% CI)

1.01 [0.92, 1.10]

28.1 Baseline FEV1 ≥ 80% of predicted

4

863

Risk Ratio (M‐H, Fixed, 95% CI)

1.04 [0.94, 1.15]

28.2 Mean baseline FEV1 not reported

2

391

Risk Ratio (M‐H, Fixed, 95% CI)

0.94 [0.79, 1.11]

29 # participants with oral candidiasis Show forest plot

2

182

Risk Ratio (M‐H, Fixed, 95% CI)

1.01 [0.15, 6.85]

29.1 Mean baseline FEV1 ≥ 80% of predicted

2

182

Risk Ratio (M‐H, Fixed, 95% CI)

1.01 [0.15, 6.85]

30 # participants with headache Show forest plot

5

1230

Risk Ratio (M‐H, Fixed, 95% CI)

1.13 [0.85, 1.50]

30.1 Baseline FEV1 ≥ 80% of predicted

4

863

Risk Ratio (M‐H, Fixed, 95% CI)

1.08 [0.80, 1.46]

30.2 Mean baseline FEV1 not reported

1

367

Risk Ratio (M‐H, Fixed, 95% CI)

1.44 [0.66, 3.16]

31 # participants with vomiting Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

31.1 Baseline FEV1 ≥ 80% of predicted

1

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

32 # participants with cold Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

32.1 Mean baseline FEV1 ≥ 80% of predicted

1

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

33 # participants with upper respiratory tract infection Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

33.1 Baseline FEV1 ≥ 80% of predicted

1

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

34 Linear growth Show forest plot

2

Mean Difference (Fixed, 95% CI)

1.21 [0.72, 1.70]

35 Deaths Show forest plot

1

Risk Difference (M‐H, Fixed, 95% CI)

Totals not selected

35.1 Baseline FEV1 ≥ 80% of predicted

1

Risk Difference (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

Figures and Tables -
Comparison 2. LABA + ICS versus placebo + higher dose of ICS
Comparison 3. Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 # participants with exacerbations requiring oral steroids by dose of ICS in both groups Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

1.1 Low dose of ICS (≤ 400 μg/d of BDP‐eq)

8

1376

Risk Ratio (M‐H, Fixed, 95% CI)

0.96 [0.67, 1.37]

1.2 Moderate dose of ICS (401 to 800 μg/d of BDP‐eq)

3

270

Risk Ratio (M‐H, Fixed, 95% CI)

0.89 [0.48, 1.64]

1.3 High dose of ICS (> 800 μg/d of BDP‐eq)

0

0

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

1.4 Unspecified dose of ICS or range of dose only mentioned

1

23

Risk Ratio (M‐H, Fixed, 95% CI)

1.09 [0.28, 4.32]

2 # participants with exacerbations requiring oral steroids by whether LABA dose is usual or higher than usual Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

2.1 LABA at usual dose

11

1646

Risk Ratio (M‐H, Fixed, 95% CI)

0.94 [0.69, 1.28]

2.2 LABA at higher than usual dose

1

23

Risk Ratio (M‐H, Fixed, 95% CI)

1.09 [0.28, 4.32]

3 # participants with exacerbations requiring oral steroids by type of LABA Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

3.1 Formoterol

4

591

Risk Ratio (M‐H, Fixed, 95% CI)

0.89 [0.58, 1.39]

3.2 Salmeterol

8

1078

Risk Ratio (M‐H, Fixed, 95% CI)

1.00 [0.66, 1.51]

4 # participants with exacerbations requiring oral steroids by single inhaler or separate inhalers for LABA and ICS Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

4.1 Single inhaler

6

1227

Risk Ratio (M‐H, Fixed, 95% CI)

0.99 [0.66, 1.47]

4.2 Separate inhaler

6

442

Risk Ratio (M‐H, Fixed, 95% CI)

0.90 [0.57, 1.42]

5 # participants with exacerbations requiring oral steroids by trial duration Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

5.1 ≤ 16 weeks

10

1526

Risk Ratio (M‐H, Fixed, 95% CI)

0.90 [0.65, 1.25]

5.2 ≥ 24 weeks

2

143

Risk Ratio (M‐H, Fixed, 95% CI)

1.22 [0.59, 2.52]

6 # participants with exacerbations requiring oral steroids by whether funded by producers of LABA Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

6.1 Charity/grant agency funded

2

49

Risk Ratio (M‐H, Fixed, 95% CI)

1.83 [0.60, 5.56]

6.2 Funded by manufacturers of LABA

9

1580

Risk Ratio (M‐H, Fixed, 95% CI)

0.90 [0.66, 1.23]

6.3 Unknown funding source

1

40

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

7 # participants with exacerbations requiring systemic steroids by publication status Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

7.1 Published as full‐text papers

10

1439

Risk Ratio (M‐H, Fixed, 95% CI)

0.97 [0.69, 1.37]

7.2 Not published as full‐text papers

2

230

Risk Ratio (M‐H, Fixed, 95% CI)

0.89 [0.48, 1.64]

8 # participants with exacerbations requiring systemic steroids by blinding of study Show forest plot

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

8.1 Double‐blinded studies

12

1669

Risk Ratio (M‐H, Fixed, 95% CI)

0.95 [0.70, 1.28]

8.2 Open‐label studies

0

0

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

Figures and Tables -
Comparison 3. Subgroup analyses (comparison 01: LABA + ICS vs same dose of ICS)
Comparison 4. Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS)

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 # participants with exacerbations requiring oral steroids by dose of ICS in control groups Show forest plot

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

1.1 Low dose of ICS (≤ 400 μg/d of BDP‐eq)

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

1.2 Moderate dose of ICS (401 to 800 μg/d of BDP‐eq)

0

0

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

1.3 High dose of ICS (> 800 μg/d of BDP‐eq)

0

0

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

1.4 Unspecified dose of ICS or range of dose only mentioned

0

0

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

2 # participants with exacerbations requiring oral steroids by whether LABA dose is usual or higher than usual Show forest plot

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

2.1 LABA at usual dose

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

2.2 LABA at higher than usual dose

0

0

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

3 # participants with exacerbations requiring oral steroids by type of LABA Show forest plot

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

3.1 Formoterol

2

461

Risk Ratio (M‐H, Fixed, 95% CI)

2.05 [0.72, 5.82]

3.2 Salmeterol

1

120

Risk Ratio (M‐H, Fixed, 95% CI)

1.43 [0.58, 3.50]

4 # participants with exacerbations requiring oral steroids by single inhaler or separate inhalers for LABA and ICS Show forest plot

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

4.1 Combination inhaler

2

461

Risk Ratio (M‐H, Fixed, 95% CI)

2.05 [0.72, 5.82]

4.2 Separate inhaler

1

120

Risk Ratio (M‐H, Fixed, 95% CI)

1.43 [0.58, 3.50]

4.3 Not reported

0

0

Risk Ratio (M‐H, Fixed, 95% CI)

0.0 [0.0, 0.0]

5 # participants with exacerbations requiring oral steroids by trial duration Show forest plot

3

581

Risk Ratio (M‐H, Fixed, 95% CI)

1.69 [0.85, 3.32]

5.1 ≤ 16 weeks

1

303

Risk Ratio (M‐H, Fixed, 95% CI)

2.04 [0.19, 22.26]

5.2 ≥ 24 weeks

2

278

Risk Ratio (M‐H, Fixed, 95% CI)

1.65 [0.81, 3.36]

Figures and Tables -
Comparison 4. Subgroup analyses (comparison 02: LABA + ICS vs higher dose of ICS)
Comparison 5. Sensitivity analysis: LABA + ICS versus placebo + higher dose of ICS

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 # participants with exacerbations requiring oral steroids Show forest plot

4

Risk Ratio (M‐H, Fixed, 95% CI)

Subtotals only

1.1 Baseline FEV1 ≥ 80% of predicted

4

895

Risk Ratio (M‐H, Fixed, 95% CI)

0.93 [0.64, 1.33]

Figures and Tables -
Comparison 5. Sensitivity analysis: LABA + ICS versus placebo + higher dose of ICS