Pharmacological and surgical interventions for the treatment of gastro‐oesophageal reflux in adults and children with asthma

Summary of findings 1. Medical or surgical intervention, or both, for gastro‐oesophageal reflux disease compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children

Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Medical or surgical intervention, or both, for gastro‐oesophageal reflux disease compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children
Patient or population: treatment for asthma in adults and children with asthma and GORD Setting: outpatient Intervention: medical or surgical intervention, or both, for gastro‐oesophageal reflux disease Comparison: nil intervention, delayed intervention control, or placebo
Outcomes	Risk with nil intervention, delayed intervention control, or placebo	Risk with medical or surgical intervention, or both, for gastro‐oesophageal reflux disease	Relative effect (95% CI)	№. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Number of participants with moderate/severe acute exacerbations Duration of treatment: range 24 weeks to 26 weeks	112 per 1000	63 per 1000 (21 to 171)	OR 0.53 (0.17 to 1.63)	1168 (2 RCTs)	⊕⊕⊝⊝ LOW^a,b	Evidence is uncertain about the effect of medical treatment for GORD on acute exacerbations in adults. There were 60 adults with events in the experimental group (n = 732 individuals) compared to 49 adults in the control group (n = 436 individuals)
Change in FEV₁ (L) Duration of treatment: range 6 weeks to 26 weeks	Mean change in FEV₁ (L) was 0.50	MD 0.1 higher (0.05 higher to 0.15 higher)	‐	1333 (7 RCTs)	⊕⊕⊕⊝ MODERATE^a	There is moderate‐certainty evidence related to the effect of medical treatment for GORD on FEV₁ (L) in adults with asthma. Pooled analysis indicates that the mean difference in treatment was a small, non‐clinically significant improvement of 0.10 L compared to control
Use of "rescue" medications and emergency action plans: B2 use puffs per day Duration of treatment: range 14 weeks to 24 weeks	Mean use of "rescue" medications and emergency action plans: B2 use puffs per day was 2.11	MD 0.71 lower (1.2 lower to 0.22 lower)	‐	239 (2 RCTs)	⊕⊕⊕⊝ MODERATE^a	There is moderate‐certainty evidence related to the effect of medical treatment for GORD on use of rescue medications for people with asthma. A small reduction in average puffs per day (0.71) was detected in the treatment group (n = 115) compared to the control group (n = 124)
Change in quality of life Assessed with AQLQ Scale from 1 to 7; duration of treatment: range 8 weeks to 26 weeks	Mean change in AQLQ was 0.9244	MD 0.21 higher (0.02 lower to 0.44 higher)	‐	1595 (5 RCTs)	⊕⊕⊝⊝ LOW^a,c	Evidence is uncertain related to the effect of medical treatment for GORD on change in AQLQ for people with asthma. Mean difference of 0.21 was smaller than minimum important difference of 0.5 units
Healthcare utilisation ‐ not measured			‐	‐	‐	No studies included in this review reported on this outcome
*The risk in the intervention group (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). AQLQ: Asthma Quality of Life Questionnaire; CI: confidence interval; FEV₁: forced expiratory volume in 1 second; GORD: gastro‐oesophageal reflux disease; MD: mean difference; OR: odds ratio; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aRisk of bias assessment for included studies was unclear or high for important domains of randomisation and allocation concealment (‐1 point). ^bA small number of events resulted in a wide confidence interval (‐1 point). ^cPooled effect estimate is imprecise, with confidence interval including no difference (‐1 point).

Background

This is an update of a Cochrane Review previously published in 2003 (Gibson 2003).

Description of the condition

Gastro‐oesophageal reflux disease (GORD) is the passing of gastric contents through the gastric cardia into the oesophagus. This can be a normal physiological event that occurs mainly after meals during the day in healthy people. Abnormal acid reflux occurs when there is significant exposure (pH < 4.0) to the distal oesophagus for longer than 1.2 hours (cumulative time > 5%) over a 24‐hour period as established by intra‐oesophageal pH monitoring (Johnson 1974; Johnsson 1987). Abnormal acid reflux is commonly associated with upper gastrointestinal (GI) symptoms such as heartburn, dysphagia, regurgitation, and chest pain. There has long been recognition of extra‐oesophageal manifestations including asthma, laryngitis, chronic cough, and rhinosinusitis (Hungin 2005; Mahdavinia 2016; Moore 2010).

The prevalence of GORD in people with asthma is reported to be three times that in the general population (Mauskar 2016). A study from India identified that GORD was present in 40% of people with asthma (Gaude 2016).

Asthma and GORD are common medical conditions that often co‐exist, as shown in different trials. For example, Sontag 1990 studied 104 adults with asthma and found that more than 80% had abnormal GORD on 24‐hour pH monitoring. Compared with healthy controls, people with asthma and reflux had significantly decreased lower oesophageal sphincter pressures, greater oesophageal acid exposure times, more frequent reflux episodes, and longer acid clearance, irrespective of body position and bronchodilator therapy. Studies in children with asthma similarly show a high prevalence of significant GORD (Andze 1991; Martin 1982; Tucci 1993). A systematic review of 28 studies found that symptoms of GORD and abnormal 24‐hour pH monitoring were occurring in 59% and 51% of people with asthma but concluded that data were insufficient to clarify the direction of causality in this association (Havemann 2007). The mechanism underpinning the relationship between GORD and asthma remains unclear, most likely due to the multi‐factorial nature of both conditions (Pacheco 2018).

The role of GORD as a trigger in asthma can be explained by several mechanisms: micro‐aspiration of gastric acid into the respiratory airways (Astarita 2000; Klotz 1971; Mays 1976), vagally mediated reflux inducing bronchoconstriction and airway hyper‐responsiveness (Astarita 2000; Mansfield 1989; Tuchman 1984), and direct acid stimulation of the oesophagus. However, studies in people with asthma have provided conflicting results on the effects of lower oesophageal acidification as a trigger for asthma. Furthermore, the possibility exists that asthma might precipitate GORD (Singh 1983). Thus a temporal association between the two does not conclusively establish that GORD triggers asthma symptoms.

Given recently available technological advancements, impedance‐pH monitoring enables detection of weakly acid, or non‐acid, reflux when pH exposure to the oesophagus is greater than 4 but less than 7 (Ates 2014; Sifrim 2004). It has been suggested that non‐acid reflux may contribute to persistent asthma symptoms, particularly when the individual does not respond to acid suppression treatment (Ates 2014).

Description of the intervention

Narrative reviews have identified the high frequency of GORD in people with asthma, as well as the clinical features and the spectrum of available therapy. Therapy can involve a number of measures, either medical or surgical, to improve symptoms of GORD. Examples of medical treatment include histamine (receptor type 2) antagonists in standard or high dose, proton pump inhibitors, and prokinetics. Presently, proton pump inhibitors are the gold standard treatment for gastro‐oesophageal reflux. Different types of proton pump inhibitors are available, but meta‐analyses fail to show significant differences in efficacy for symptom relief between proton pump inhibitors (Gralnek 2006). Potential surgical options for GORD include Nissen fundoplication and partial posterior semi‐fundoplication (Toupet and Lind techniques). Randomised trials have been conducted on each form of therapy, with conflicting results (Choy 1997; Field 1998; Kahrilas 1996; Simpson 1995; Winter 1997). A systematic review of antireflux surgery in people with asthma found that surgery may improve asthma symptoms but not pulmonary function (Field 1999).

How the intervention might work

Asthma has been proposed as one of the extra‐oesophageal manifestations of GORD. Proton pump inhibitor and histamine (receptor type 2) antagonist drugs aim to suppress acid production in the stomach, thus reducing the risks of acid micro‐aspiration into the airway and vagally mediated bronchoconstriction (Gracie 2016; Harding 2001). Prokinetic therapy with metoclopramide (other prokinetics ‐ domperidone and bethanechol), on the other hand, has been shown to augment gastric emptying and increase lower oesophageal sphincter pressure (Champion 1997). Currently, proton pump inhibitors are recommended as the initial approach to reflux management, and asthma guidelines recommend that, especially in severe cases, the presence of GORD should be investigated and treatment provided when appropriate to improve management of asthma symptoms (King‐Biggs 2019; Moore 2010).

In terms of surgical treatment, fundoplication is one of the most common approaches for treatment of GORD, with success rates of 80% to 90% (Patti 2015). Fundoplication is a procedure that involves wrapping the upper curve of the stomach around the lower portion of the oesophagus, thereby tightening the oesophageal sphincter. This procedure changes the way the gastro‐oesophageal junction functions to reduce the occurrence of transient lower oesophageal sphincter relaxations (Ireland 1993). The best surgical responses are seen in patients with typical symptoms of GORD that demonstrate good response to proton pump inhibitor therapy and in those who have abnormal ambulatory pH studies with good symptom correlation (Oelschlager 2008).

Why it is important to do this review

To date, clinical research trials have described mixed results on the effectiveness of GORD treatment to support management of asthma symptoms. The original publication of this Cochrane Review was unable to show improvement in asthma management‐related outcomes following treatment for GORD and identified that specific subgroups may benefit (Gibson 2003). Since 2003, several trials using different types of medical and surgical treatments for GORD have been published. Therefore, an update of the review is required to examine newly published evidence on the role of GORD treatment for asthma in adults and children.

Objectives

To evaluate the effectiveness of GORD treatment in adults and children with asthma, in terms of its benefits for asthma.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs), both parallel and cross‐over. Cluster‐randomised controlled trials were eligible, provided the data had been or could be adjusted for clustering. We included studies if they were reported as full text or were published as an abstract only, or if unpublished data were available. We included studies for which only a subset of participants met all inclusion criteria, provided disaggregated data could be obtained.

Types of participants

We included adults and children with a diagnosis of both asthma and GORD. Asthma was diagnosed according to international or national guidelines (e.g. GINA 2019), or was diagnosed by a medical practitioner with use of an objective lung function measurement. Similarly, a formal diagnosis of GORD by a medical practitioner was required based on symptoms or objective measurements such as 24‐hour pH studies/manometry or oesophagoscopy with or without biopsy.

Types of interventions

We included pharmacological and surgical interventions for treatment of patients with GORD. Pharmacological interventions consisted of: antacids (e.g. Gaviscon), proton pump inhibitors (e.g. lansoprazole, esomeprazole, omeprazole, rabeprazole, pantoprazole, dexlansoprazole, omeprazole with sodium bicarbonate), histamine 2 receptor antagonists (e.g. ranitidine), and prokinetics (e.g. baclofen, domperidone, bethanechol). We did not specify minimum dosages or duration of intervention for any of these pharmacotherapies. Surgical interventions included Nissen's fundoplication and the Bianchi procedure (i.e. total oesophagogastric disconnection). These pharmacological and surgical interventions could be evaluated on their own as individual therapy, or as a combined pharmacological and surgical package intervention.

Comparator groups included no intervention, delayed intervention control, and placebo.

Types of outcome measures

Primary outcomes

Acute asthma exacerbations (either number of participants with events or rate) as reported by trialists
Hospital admissions (either number of participants with events or rate) and length of stay
Emergency room or unscheduled doctor visits (either number of participants with events or rate)

Secondary outcomes

Lung function: spirometry, measured as forced expiratory volume in 1 second (FEV₁; litres per minute) and peak expiratory flow rate (PEFR) (morning only)
Use of "rescue" medications and emergency action plan as reported by trialists (e.g. self‐report via diary or questionnaire, electronic monitoring, prescription monitoring, pharmacy claims data)
Asthma symptoms score (ideally measured by a validated scale such as the Asthma Control Questionnaire (ACQ); Juniper 1999)
Nocturnal symptoms (e.g. self‐report via a diary or questionnaire)
Quality of life (e.g. Asthma Quality of Life Questionnaire (AQLQ)) with responder analysis (dichotomous analysis of people who achieved the minimal important difference versus those who did not)
Treatment preferences (e.g. determined by self‐report)
Adverse events

Search methods for identification of studies

Electronic searches

We identified studies by searching the following databases and trial registries.

Cochrane Airways Trials Register (Cochrane Airways 2019), via the Cochrane Register of Studies (searched 23 June 2020).
Cochrane Central Register of Controlled Trials (CENTRAL), in the Cochrane Library, via the Cochrane Register of Studies (searched to 23 June 2020).
MEDLINE Ovid SP (ALL) 1946 to 22 June 2020 (searched 23 June 2020).
Embase Ovid SP 1974 to 2020 week 25 (searched 23 June 2020).
US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov) (searched 23 June 2020).
World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch) (searched 23 June 2020).

The database search strategies are listed in Appendix 1.

We searched all databases and trials registries from their inception to the present, with no restriction on language or type of publication. We handsearched conference abstracts and searched grey literature through the Cochrane Airways Trials Register and the CENTRAL database. Searches were conducted by the Information Specialist for Cochrane Airways.

Searching other resources

In addition, we screened reference lists of all available primary studies and review articles to identify potentially relevant citations. We contacted the authors of primary studies regarding other published and unpublished trials known to them.

Data collection and analysis

Selection of studies

We used the Cochrane Screen4Me workflow to assist with assessment of search results by matching records in the search results to records already screened in Cochrane Crowd (http://crowd.cochrane.org) and labelled as an RCT or as not an RCT (Marshall 2018; McDonald 2017; Noel‐Storr 2018; Thomas 2017), and we assessed records using the RCT Classifier, a machine learning model that distinguishes RCTs from non‐RCTs. The Cochrane Airways Group Information Specialist removed studies identified in the previous version of this review (Gibson 2003).

Following these initial assessments, we imported data into EndNote software for removal of duplicate references. Two independent review authors (ZK and KVC) screened titles and abstracts and coded them as 'potentially include: full text review required' or 'exclude'. We resolved disagreements through discussion and consensus, erring on the side of caution, with discrepancies requiring full‐text review. Two independent review authors (ZK and KVC) then reviewed full‐text articles for eligibility and coded them as 'include', 'ongoing', 'exclude but relevant', or 'exclude'. Again, we resolved disagreements through discussion and consensus. We recorded this selection process in the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses flow diagram (PRISMA; Moher 2009), as well as in the Characteristics of excluded studies table.

Data extraction and management

Two review authors independently extracted characteristics, outcome data, and risk of bias data from all included studies onto a pilot‐tested, standardised data extraction template using Microsoft Word (all review authors). We resolved any emerging conflicts by discussion and consensus or with a third review author as required. Data extracted for characteristics included:

methods: country, design, objective/aim, study site, methods of analysis;
participants: eligibility for study, randomisation numbers per group, participants completed per group, age, gender, comorbidities, diagnostic criteria for asthma, diagnostic criteria for GORD, association between asthma and GORD tested, major exclusion criteria, baseline severity of asthma (FEV₁ in litres, FEV₁ per cent predicted, PEFR), baseline severity of GORD, baseline complications of GORD;
interventions: duration of intervention, type of intervention, type of control; and
outcomes: pre‐specified outcomes, follow‐up period, outcomes measured.

Assessment of risk of bias in included studies

Two review authors (all authors) independently assessed risk of bias for each study for random sequence generation, allocation concealment, blinding of participants and outcome assessors, handling of missing data, selective outcome reporting, and other threats to validity in the studies, in line with recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019). We also conducted a retrospective risk of bias assessment for all original studies included in the previous version of this review.

We judged each potential source of bias as high, low, or unclear risk, and we provided a justification for the judgement in the 'risk of bias' tables. A summary of risk of bias judgements across different studies for each of the domains listed is provided in the results. We noted information on risk of bias related to unpublished data or correspondence with trialists in the risk of bias table.

We used the pre‐specified protocol to guide review processes when assessing risk of bias associated with undertaking this systematic review and noted deviations from this protocol in the section titled Differences between protocol and review.

Measures of treatment effect

We combined data from included trials using Review Manager 5.3. We analysed continuous and dichotomous data using a fixed‐effect model for all studies deemed similar enough to be pooled. We considered a random‐effects model in the presence of substantial heterogeneity (≥ 50% based on the I² statistic). We undertook meta‐analysis only when we considered treatments, participants, and underlying clinical questions sufficiently similar to be meaningfully pooled. We described any data that could not be pooled using narrative synthesis in the Results section under Effects of interventions, and we entered any data presented as a scale with a consistent direction of effect.

When multiple trial arms were reported in a single trial, we included only the relevant arms. If two intervention arms were both relevant, when appropriate, we combined them as if they were one study arm. For studies reporting both change from baseline and endpoint scores, we used change from baseline scores. When both per‐protocol and intention‐to‐treat analyses were provided, we used the latter as a preference.

For continuous outcomes, we used mean differences (MDs) with 95% confidence intervals (CIs) or standardised mean differences (SMDs).

For dichotomous outcomes, we calculated Mantel‐Haenzsel odds ratios (ORs) with 95% CIs. In instances of rare events (calculated as ≤ 5% of the population), we used Peto ORs. For data reported as rates (e.g. exacerbations), we planned to analyse on the log scale as hazard ratios and to combine using the random‐effects model and generic inverse variance.

Unit of analysis issues

This review included a mixture of cross‐over and parallel studies, producing the potential for unit of analysis issues. We addressed unit of analysis issues by excluding cross‐over trials from meta‐analyses, unless paired data were available (e.g. paired t‐test). We analysed data from cluster‐randomised controlled trials only if available data had been or could be adjusted for potential clustering effects (Higgins 2019a). We would have made adjustments for clustering by contacting the original study authors to identify intracluster correlation coefficients. However, none of the included studies involved clusters, and ultimately, this was not necessary.

Dealing with missing data

We evaluated missing information regarding participants on an available case analysis basis, as described in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2019). When statistics essential for analysis were missing (e.g. when group means and standard deviations for both groups were not reported) and could not be calculated from other data, we attempted to contact study authors for missing data. We assumed loss of participants that occurred before baseline measurements to have no effect on eventual outcome data of the study. We used the intention‐to‐treat approach as the measure for assessment and discussion of losses after baseline measurements. When data were presented only in abstract or protocol form and attempts to contact study authors (on two occasions) were unsuccessful, we classified studies as excluded but relevant.

Assessment of heterogeneity

We used a combination of tests, including visual inspection of data and the I² statistic to assess statistical heterogeneity. We determined thresholds for the I² statistic according to the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2019), as follows:

0% to 40%: might not be important;
30% to 60%: may represent moderate heterogeneity;
50% to 90%: may represent substantial heterogeneity; and
75% to 100%: considerable heterogeneity.

Funnel plots were to be applied as well if 10 or more studies had been included. We considered the Der‐Simonian and Laird method of analysis presented with a P value less than 0.05 as statistically significant. In the presence of significant heterogeneity (as per criteria above), we re‐analysed data using the random‐effects model (Der Simonian 1986).

Assessment of reporting biases

We planned to explore potential reporting biases by using a funnel plot had meta‐analysis of 10 or more studies been available. Instead, reporting biases were extrapolated within the other bias section in the risk of bias tables. Unpublished data were able to be included in the review, with their status highlighted within the references. When available, outcomes reported in protocols of studies were compared against outcomes reported in publications.

Data synthesis

We combined and analysed data from all trials using Review Manager 2014 (RevMan 5.3) software. We used a fixed‐effect model for all analyses. However, in the presence of substantial heterogeneity (I² statistic ≥ 50%), we used a random‐effects model. We performed sensitivity analyses using a fixed‐effect model.

Subgroup analysis and investigation of heterogeneity

We planned to perform subgroup analyses for adults versus children for all outcomes.

Sensitivity analysis

We performed a sensitivity analysis when results from the same studies were reported with conflicting information from different sources (see Differences between protocol and review).

Summary of findings and assessment of the certainty of the evidence

We created a summary of findings table for outcomes considered to be relevant for either clinical care and/or policy makers. We selected the following:

Acute exacerbations
Hospital admissions and length of stay
Emergency room or unscheduled doctor visits
Use of "rescue" medications
Asthma symptoms

We used the five GRADE considerations (risk of bias, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of the body of evidence as it relates to the included studies reporting these outcomes. We then generated the Summary of findings table 1 using GRADEpro software (GRADEpro GDT) and provided all justifications for downgrading of evidence in the table footnotes.

Results

Description of studies

See Table 1 for a brief summary of the main characteristics of the 23 studies included in this review.

See: Summary of findings 1 Medical or surgical intervention, or both, for gastro‐oesophageal reflux disease compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children

Table 1. Summary of characteristics of included studies

Study ID	Country	N	Duration (weeks)	Intervention	Control	Outcomes of interest in this review
Aiguo 1999	China	54	6	a: Jiang Ni decoction twice daily b: 150 mg ranitidine twice daily plus cisapride 10 mg 3 times daily	Placebo	FEV₁; PEFR; asthma symptoms
Dos Santos 2007	Brazil	44	12	Pantoprazole 40 mg once daily	Placebo	FEV₁; mPEFR; ePEFR; asthma symptoms; nocturnal asthma symptoms; quality of life
Ekstrom 1989	Sweden	50	4	Ranitidine 150 mg twice daily	Placebo	FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; adverse events
Fallahi 2008	Iran	36	6	Omeprazole 20 mg twice daily	Placebo	FEV₁; PEFR
Ford 1994	UK	11	4	Omeprazole 20 mg once daily	Placebo	PEF; use of beta₂‐agonists; asthma symptoms
Frison 2002	Brazil	22	8	Lansoprazole 60 mg once daily plus cisapride 30 mg once daily plus behavioural management	Placebo plus behavioural management	FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms
Goodall 1981	UK	20	6	Cimetidine 200 mg 3 times daily plus 2 at night	Placebo	FEV₁; PEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms
Jiang 2003	China	30	6	Omeprazole 20 mg once daily plus domperidone 10 mg 3 times daily	Nil intervention	FEV₁; PEFR
Kiljander 1999	Finland	57	8	Omeprazole 40 mg once daily	Placebo	FEV₁; PEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms
Kiljander 2006	Multi‐national^a	770	16	Esomeprazole 40 mg twice daily	Placebo	Asthma exacerbations; FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; quality of life; adverse events
Kiljander 2010	Multi‐national^b	916	26	a: Esomeprazole 40 mg once daily b: Esomeprazole 40 mg twice daily	Placebo	Asthma exacerbations, FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; quality of life; adverse events
Kjellen 1981	Sweden	62	8	Conservative treatment (head elevation, warm water after meals, avoiding food 3 hours before sleep, refraining from aspirin and anticholinergic drugs, not raising intra‐abdominal pressure)	Nil intervention	FEV₁;use of beta₂‐agonists; asthma symptoms
Larrain 1991	Chile	90	26	a: Cimetidine 300 mg 4 times daily b: Surgery, posterior gastropexy with cardiac calibration	Placebo	FEV₁; asthma symptoms
Levin 1998	USA	11	8	Omeprazole 20 mg once daily	Placebo	FEV₁; mPEFR; ePEFR; quality of life
Littner 2005	USA	207	24	Lansoprazole 30 mg twice daily	Placebo	Asthma exacerbations, FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; quality of life; adverse events
Maev 2002	Russia	64	8	Omeprazole 20 mg twice daily	Placebo	FEV₁; PEF; asthma symptoms
Meier 1994	USA	15	6	Omeprazole 20 mg twice daily	Placebo	FEV₁; PEF; use of beta₂‐agonists; asthma symptoms
Nagel 1988	UK	15	1	Ranitidine 150 mg in the morning and 300 mg at night	Placebo	mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms
Sharma 2007	India	204	16	Omeprazole 20 mg twice daily plus domperidone 10 mg 3 times daily	Placebo	FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; adverse events
Sontag 2003	USA	75	104	a: Ranitidine 150 mg 3 times daily plus conservative treatment b: Surgery, Nissen fundoplication plus conservative treatment	Symptomatic treatment as needed, Mylanta 30 mL	PEF, asthma symptoms
Stordal 2005	Norway	38	12	Omeprazole 20 mg once daily	Placebo	FEV₁; use of beta₂‐agonists; asthma symptoms; quality of life
Susanto 2008	Indonesia	36	14	Esomeprazole 40 mg daily plus conservative treatment (lifestyle modifications and antacid medication if required)	Conservative treatment (lifestyle modifications and antacid medication if required)	mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms
Teichtahl 1996	Australia	25	12	Omeprazole 40 mg once daily	Placebo	FEV₁;mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms

ePEFR: evening peak expiratory flow rate; FEV₁: forced expiratory volume in one second; mPEFR: morning peak expiratory flow rate.

^aKiljander 2006 included participants from centres in the following countries: Argentina, Brazil, Bulgaria, Canada, Czech Republic, Finland, Hungary, Italy, Mexico, Romania, Sweden, and the United States of America.

^bKiljander 2010 included participants from centres in the following countries: Argentina, Bulgaria, Canada, Czech Republic, France, Germany, Hungary, Italy, Mexico, Poland, Portugal, Slovakia, and the United States of America.

For additional details on the 23 included and 77 excluded studies, see Characteristics of included studies and Characteristics of excluded studies.

Results of the search

The previously published version of this review identified 12 RCTs for inclusion (Gibson 2003). For this update, searches were completely re‐run up to 23 June 2020 to ensure that standards reflect current Cochrane Review procedures (Lefebvre 2019). The original 12 studies were re‐screened according to updated criteria for inclusion. Of these, we excluded two as they did not include participants who fulfilled the pre‐defined diagnosis of asthma and GORD as outlined for this review (Boeree 1998; Gustafsson 1992).

Through update searches, we retrieved 3531 records and we identified one additional record through handsearching of reference lists of relevant studies. After all duplicates were removed by the information specialist, Screen4Me workflow, and EndNote software, ZK and KVC screened 1320 records. On the basis of title and abstract review, we excluded 1205 records as they were irrelevant to the review question. We obtained full text for the remaining 115 records and reviewed them for eligibility. At this stage, we excluded 77 studies, leaving 23 studies for inclusion in this review. Of these, 12 were suitable for inclusion in meta‐analyses. For further details of the screening processes, see the study flow diagram (Figure 1).

Figure 1

PRISMA study flow diagram.

Included studies

Study design

The included studies were published between 1981 and 2010. Fourteen of the included studies were parallel studies, and eight used a cross‐over design (Ekstrom 1989; Ford 1994; Goodall 1981; Kiljander 1999; Levin 1998; Meier 1994; Nagel 1988; Teichtahl 1996). These 23 studies including international multi‐centred studies randomly assigned a total of 2872 participants. Two studies were multi‐national trials (Kiljander 2006; Kiljander 2010), and one study originated from Australia (Teichtahl 1996), one from Chile (Larrain 1991), one from Finland (Kiljander 1999), one from India (Sharma 2007), one from Iran (Fallahi 2008), one from Norway (Stordal 2005), one from Russia (Maev 2002), two from Brazil (Dos Santos 2007; Frison 2002), two from China (Aiguo 1999; Jiang 2003), two from Sweden (Ekstrom 1989; Kjellen 1981), three from United Kingdom (Ford 1994; Goodall 1981; Nagel 1988), and four from United States of America (Levin 1998; Littner 2005; Meier 1994; Sontag 2003).

Participants

A total of 2872 participants were randomly assigned across the 23 studies; 21 studies were adult studies (n = 2598 participants), and two involved a paediatric demographic (n = 274 participants) (Fallahi 2008; Stordal 2005). The total number of participants randomised across all included studies varied from 11 in Ford 1994 to 916 in Kiljander 2010. The age range for adult studies was 20 to 75 years, with a range for child/adolescent studies of 7 to 20 years. Participants from all included studies had diagnoses of both asthma and GORD based on different criteria.

Diagnosis of asthma

Different criteria had been used for asthma diagnosis; these particularly varied in clinical assessment of forced expiratory volume in 1 second (FEV₁) reversibility. Objective pulmonary function tests were required for diagnosis in 19 studies. Four studies diagnosed asthma based on doctors' assessment using symptoms and clinical grounds (Aiguo 1999; Goodall 1981; Larrain 1991; Stordal 2005). Thirteen studies reported baseline lung function as expressed in FEV₁ (Stordal 2005), peak expiratory flow rate (PEFR) (Ford 1994), or both (Aiguo 1999; Dos Santos 2007; Frison 2002; Jiang 2003; Kiljander 1999; Kiljander 2006; Kiljander 2010; Kjellen 1981; Levin 1998; Littner 2005; Sharma 2007). When baseline data were not reported, the inclusion criteria for the study had been substituted as an indicator of asthma severity (Ekstrom 1989). Five studies did not specify the severity of asthma (Fallahi 2008; Goodall 1981; Meier 1994; Nagel 1988; Teichtahl 1996), but baseline asthma symptom scores were reported by Sontag 2003 as indicators of asthma severity. Medications used in the Larrain 1991 and Maev 2002 studies also indicated the severity of asthma as moderate to severe.

Diagnosis of GORD

Different methods were used including history of symptoms, endoscopy, manometry, acid perfusion test, and 24‐hour pH monitoring. Six of the included studies described use of a dual‐lumen/channel probe (Frison 2002; Kiljander 1999; Levin 1998; Meier 1994; Nagel 1988; Stordal 2005); of these, two provided enough information to discern that impedance was also available (Levin 1998; Stordal 2005) (a dual‐lumen pH probe with impedance is the current gold standard for GORD diagnosis). All studies except three diagnosed GORD based on objective tests (Fallahi 2008; Levin 1998; Littner 2005). GORD symptoms were evident in more than 70% of the population in 17 studies (Aiguo 1999; Dos Santos 2007; Ekstrom 1989; Fallahi 2008; Ford 1994; Frison 2002; Goodall 1981; Kiljander 2006; Kiljander 2010; Larrain 1991; Levin 1998; Littner 2005; Meier 1994; Nagel 1988; Sontag 2003; Stordal 2005; Teichtahl 1996). Association of asthma and GORD was examined in seven studies (Dos Santos 2007; Ekstrom 1989; Frison 2002; Goodall 1981; Kiljander 2006; Meier 1994; Nagel 1988).

Intervention

The length of studies varied with intervention duration and ranged from 1 to 104 weeks. Types of interventions included non‐pharmacological treatment, pharmacological treatment, and surgical intervention.

Non‐pharmacological treatment

Kjellen 1981, a parallel study over eight weeks, was the only study to investigate non‐pharmacological conservative anti‐reflux therapy. Interventions involved elevating the head‐end of the bed, drinking warm water after meals, avoiding late meals three hours before bedtime, refraining from aspirin and anticholinergic medications, and avoiding procedures that were known to increase intra‐abdominal pressure.

Pharmacological treatment

Proton pump inhibitors were the mainstay of study treatment in 12 studies. Different dosages and types of proton pump inhibitors were studied, including esomeprazole (40 mg to 80 mg/d), lansoprazole (60 mg/d), omeprazole (20 to 40 mg/d), and pantoprazole (40 mg/d). H₂ antagonists were also studied in five studies (Ekstrom 1989; Goodall 1981; Larrain 1991; Nagel 1988; Sontag 2003). Four studies investigated the effectiveness of combination therapy including omeprazole and domperidone (Jiang 2003; Sharma 2007); lansoprazole and cisapride (Frison 2002); and ranitidine and cisapride (Aiguo 1999). In addition to pharmacological treatment, the study population in Frison 2002 and Sontag 2003 were provided conservative treatment.

Surgical intervention

Only two included studies used surgical intervention. Larrain 1991 randomised participants to three arms, which included medical treatment (cimetidine 300 mg four times a day), placebo, or surgery (posterior gastropexy). Sontag 2003 also randomised participants to three groups: (1) surgical approach by Nissan fundoplication; (2) medical therapy with ranitidine 150 mg three times a day; and (3) control.

Excluded studies

We excluded 77 studies (81 records) from the updated reference list following full‐text assessment against eligibility criteria (see Characteristics of excluded studies). We also excluded two of the old studies included in the previous version of this Cochrane Review. One of the excluded studies recruited patients with chronic obstructive pulmonary disease (COPD) as well as asthma, and results could not be separated (Boeree 1998), whilst another study did not require GORD as an inclusion criterion (Gustafsson 1992). The most common reason for study exclusion was no GORD diagnosis (n = 20 studies), followed by not being unable to confirm eligibility (n = 19), no relevant intervention (n = 10), inappropriate control (n = 9), no asthma diagnosis (n = 6), no control (n = 4), not randomised (n = 3), not all asthma (n = 3), and not all GORD (n = 3).

Risk of bias in included studies

Full details of our risk of bias judgements can be found under the 'Risk of bias' section at the end of each Characteristics of included studies table and in Figure 2. Many studies did not provide sufficient information to assess the risk of bias, resulting in many “unclear” judgements across domains. For unclear risk of bias, study authors were contacted for additional information. Two independent review authors (ZK and KC, HY and KT, or HY and KH) reached agreement when assessing study quality.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

We considered generation of randomisation sequence to be adequate in five studies (Ekstrom 1989; Frison 2002; Larrain 1991; Levin 1998; Meier 1994), and it was unclear in 17 studies (Aiguo 1999; Dos Santos 2007; Fallahi 2008; Ford 1994; Goodall 1981; Jiang 2003; Kiljander 1999; Kiljander 2006; Kiljander 2010; Littner 2005; Maev 2002; Nagel 1988; Sharma 2007; Sontag 2003; Stordal 2005; Susanto 2008; Teichtahl 1996). Kjellen 1981 was assessed to have high risk of bias for using alternation for allocation of study participants to treatment or control groups.

Similarly, we judged the method of allocation concealment to be adequate in five studies (Ekstrom 1989; Frison 2002; Larrain 1991; Levin 1998; Stordal 2005), and we assigned high risk to Kjellen 1981. We judged the remaining 17 studies as having unclear risk due to lack of information about allocation concealment (Aiguo 1999; Dos Santos 2007; Fallahi 2008; Ford 1994; Goodall 1981; Jiang 2003; Kiljander 1999; Kiljander 2006; Kiljander 2010; Littner 2005; Maev 2002; Meier 1994; Nagel 1988; Sharma 2007; Sontag 2003; Susanto 2008; Teichtahl 1996).

Blinding

Adequate blinding of participants and outcome assessors was reported on the basis of information provided by study authors, such as well‐defined study protocols, identical placebo tablets or intervention, central or third party allocation of participants, and same follow‐up and outcomes measurements. Thirteen studies were deemed to have unclear risk of performance bias due to inadequate information provided (Aiguo 1999; Ekstrom 1989; Fallahi 2008; Ford 1994; Frison 2002; Goodall 1981; Kiljander 2006; Kiljander 2010; Maev 2002; Nagel 1988; Sharma 2007; Susanto 2008; Teichtahl 1996). Larrain 1991 was deemed to have unclear risk of bias as a placebo identical to the study drug was used; however, it was not possible to blind the surgical arm due to the nature of the intervention. Blinding of participants was assessed as adequate in six studies (Dos Santos 2007; Kiljander 1999; Levin 1998; Littner 2005; Meier 1994; Stordal 2005), but risk was considered high in three studies (Kjellen 1981; Jiang 2003; Sontag 2003).

We judged the risk of detection bias due to inadequate blinding of outcome assessment as high in Kjellen 1981 and low in seven studies (Dos Santos 2007; Kiljander 1999; Larrain 1991; Levin 1998; Littner 2005; Meier 1994; Stordal 2005). Fifteen studies had unclear risk of detection bias (Aiguo 1999; Ekstrom 1989; Fallahi 2008; Ford 1994; Frison 2002; Goodall 1981; Jiang 2003; Kiljander 2006; Kiljander 2010; Maev 2002; Meier 1994; Sharma 2007; Sontag 2003; Susanto 2008; Teichtahl 1996).

Incomplete outcome data

We reported high risk of attrition bias in three studies (Dos Santos 2007; Kiljander 2006; Susanto 2008). We assessed risk of attrition bias as low in 19 studies (Aiguo 1999; Ekstrom 1989; Fallahi 2008; Ford 1994; Frison 2002; Goodall 1981; Jiang 2003; Kiljander 2010; Kjellen 1981; Larrain 1991; Levin 1998; Littner 2005; Maev 2002; Meier 1994; Nagel 1988; Sharma 2007; Sontag 2003; Stordal 2005; Teichtahl 1996), and as unclear in one study (Kiljander 1999).

Selective reporting

Selective reporting (reporting bias) was deemed to be at high risk in two studies (Maev 2002; Teichtahl 1996), and was deemed to be at unclear risk in eight studies (Aiguo 1999; Dos Santos 2007; Ekstrom 1989; Goodall 1981; Larrain 1991; Meier 1994; Nagel 1988; Susanto 2008). The remaining thirteen studies were judged to have low risk for reporting bias (Fallahi 2008; Ford 1994; Frison 2002; Jiang 2003; Kjellen 1981; Kiljander 1999; Kiljander 2006; Kiljander 2010; Levin 1998; Littner 2005; Sharma 2007; Sontag 2003; Stordal 2005).

Other potential sources of bias

Twenty studies were assessed as having low risk for other potential bias (Aiguo 1999; Dos Santos 2007; Ekstrom 1989; Fallahi 2008; Frison 2002; Goodall 1981; Jiang 2003; Kiljander 1999; Kiljander 2010; Kjellen 1981; Larrain 1991; Littner 2005; Maev 2002; Meier 1994; Nagel 1988; Sharma 2007; Sontag 2003; Stordal 2005; Susanto 2008; Teichtahl 1996). Two were assessed as high risk for other potential sources of bias; both were provided with the study drug by a pharmaceutical company. In the case of Ford 1994, the company also provided assistance with statistical analyses, and the study design of Kiljander 2006 was set by the pharmaceutical company. Levin 1998 was assessed as having unclear risk for this domain; per publication, there appeared to be no washout phase between treatment periods, and the potential for contamination was not discussed.

Effects of interventions

See summary of findings Table 1 for the main comparison medical or surgical intervention, or both, for gastro‐oesophageal reflux compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children.

Of the eight cross‐over studies included in this review, only one reported results in a manner that could be considered for inclusion in meta‐analyses (Levin 1998). Outcomes from the other seven cross‐over studies were reported narratively (Ekstrom 1989; Ford 1994; Goodall 1981; Kiljander 1999; Meier 1994; Nagel 1988; Teichtahl 1996).

For ease of reading, results related to effects of interventions are collated and reported as "Pooled results" and "Individual study results".

Pooled results

Primary outcomes

Acute exacerbations

Three of the 23 included studies reported data on exacerbations. Of these, two studies reported on the number of participants who experienced a severe asthma exacerbation and could be pooled into a meta‐analysis using a random‐effects model (Kiljander 2010; Littner 2005). The difference between medical treatment with a proton pump inhibitor compared to placebo was uncertain in terms of numbers of participants experiencing a moderate/severe exacerbation (odds ratio (OR) 0.53, 95% confidence interval (CI) 0.17 to 1.63; 1168 participants, 2 studies; P = 0.27; Analysis 1.1; Figure 3; low‐certainty evidence). A sensitivity analysis was performed, as Kiljander 2010 reported different values for this outcome on the online trial registry compared to the published article. According to the sensitivity analysis, this variation does not change the outcome (Table 2).

Figure 3

Forest plot of comparison: medical intervention for gastro‐oesophageal reflux compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children. Analysis, Outcome: 1.1 number of participants with acute exacerbations.

Table 2. Sensitivity analysis: number of participants with an acute exacerbation

	Odds ratio	P value
	M‐H, Random, 95% CI	P value
Per publication data	0.53 (0.17 to 1.63)	0.27
Per trials website data	0.54 (0.17 to 1.74)	0.30

M‐H: Mantel‐Haenszel.

Hospital admissions and emergency room or unscheduled doctor visits

Use of hospital and emergency services or unscheduled doctor visits were not reported as outcomes in any of the studies included in this review.

Secondary outcomes

Forced expiratory volume in one second (FEV₁)

Of the 19 included studies that reported on FEV₁, seven were suitable for meta‐analysis (Aiguo 1999; Frison 2002; Jiang 2003; Kiljander 2010; Kjellen 1981; Levin 1998; Littner 2005). Improvement in change in absolute FEV₁ was detected in adults with asthma and a GORD diagnosis (mean difference (MD) 0.10 L, 95% CI 0.05 to 0.15; 1333 participants, 7 studies; P < 0.0001; Analysis 1.2; moderate‐certainty evidence).

Morning peak expiratory flow rate (PEFR)

Twelve included studies reported morning PEFR; of these, five were able to be pooled (Dos Santos 2007; Kiljander 2010; Levin 1998; Littner 2005; Susanto 2008). A fixed‐effect model was used to demonstrate improvement of 6 L/min with medical treatment compared to placebo (MD 6.02, 95% CI 0.56 to 11.47; 1262 participants, 5 studies; Analysis 1.3; P = 0.03).

Sensitivity analysis was performed using a random‐effects model (see Table 3 for a comparison with the primary analysis).

Table 3. Sensitivity analysis: morning peak expiratory flow rate (L/min)

	Mean difference	P value
	IV, Random, 95% CI	P value
Random‐effects analysis	10.18 (‐1.92 to 22.28)	0.10
Fixed‐effect analysis	6.02 (0.56 to 11.47)	0.03

IV: inverse variance.

Use of "rescue" medications and emergency action plan

Two out of the 14 studies reporting on use of a beta₂‐agonist as "rescue" medication were pooled (Littner 2005; Susanto 2008). Meta‐analysis indicates a reduction in puffs per day with medical treatment for GORD in adults with asthma compared to adults given a control intervention (MD ‐0.71, 95% CI ‐1.20 to ‐0.22; P = 0.005; 239 participants, 2 studies; Analysis 1.4; moderate‐certainty evidence).

Change in quality of life

Quality of life was reported in seven studies (Dos Santos 2007; Frison 2002; Kiljander 2006; Kiljander 2010; Levin 1998; Littner 2005; Stordal 2005). Data for this outcome were collected using a variety of scales and questionnaires across studies. The difference in change in Asthma Quality of Life Questionnaire (AQLQ) scores between medical treatment for GORD in people with asthma compared to control was able to be meta‐analysed, although evidence of any benefit overall was uncertain (MD 0.21, 95% CI ‐0.02 to 0.44; 1595 participants, 5 studies; P = 0.07; Analysis 1.5). A responder analysis was conducted within a single study (Kiljander 2010), highlighting that more people in the intervention group experienced greater than the minimal clinically important improvement in quality of life, as measured by the AQLQ (OR 1.53, 95% CI 1.15 to 2.03; 864 participants, 1 study; P = 0.003; Analysis 1.6; low‐certainty evidence).

According to the paediatric subgroup in Analysis 1.5, any benefit in AQLQ with medical treatment compared to placebo was also uncertain (MD 0.12, 95% CI ‐0.24 to 0.48; 36 participants; 1 study; P = 0.51).

Individual study results

Primary outcomes

Acute exacerbations

Kiljander 2006 reported that fewer asthma exacerbations (n = 22) were noted in the esomeprazole group compared to the placebo group (n = 24). However, the median time to asthma exacerbation was shorter at 42 days in the treatment group compared to 67 days in the placebo group; study authors state that this comparison was not statistically significant (P = 0.70).

One additional study reported this outcome as an adverse event, noting that a participant experienced an acute asthma exacerbation whilst active in the intervention (cimetidine) group, requiring prolonged inpatient admission, and was withdrawn (Goodall 1981).

Hospital admissions and emergency room or unscheduled doctor visits

Use of hospital and emergency services or unscheduled doctor visits were not reported as outcomes in any of the studies included in this review.

Secondary outcomes

Forced expiratory volume in one second (FEV₁)

Of the studies that were not suitable for meta‐analysis, four parallel studies ‐ Maev 2002; Kiljander 2006; Dos Santos 2007; Sharma 2007 ‐ and five cross‐over studies ‐ Goodall 1981; Ekstrom 1989; Meier 1994; Teichtahl 1996; Kiljander 1999 ‐ involving adults reported on FEV₁ with conflicting results. A detailed description of the findings of these studies can be found in Appendix 2.

Morning peak expiratory flow rate (PEFR)

Of the studies not suitable for inclusion in the meta‐analysis, nine reported specifically on morning PEFR (Ekstrom 1989; Ford 1994; Frison 2002; Goodall 1981; Kiljander 1999; Kiljander 2006; Nagel 1988; Sharma 2007; Teichtahl 1996), and six reported an unspecified PEFR value (Aiguo 1999; Fallahi 2008; Frison 2002; Jiang 2003; Maev 2002; Sontag 2003). Results in both groups were conflicting; a detailed description of the findings of these studies can be found in Appendix 2.

Use of "rescue" medications and emergency action plan

Eleven included studies were not suitable for meta‐analysis (Ekstrom 1989; Ford 1994; Frison 2002; Goodall 1981; Kiljander 1999; Kiljander 2006; Kiljander 2010; Kjellen 1981; Nagel 1988; Sharma 2007; Teichtahl 1996); again, results in adults were conflicting. A detailed description of the findings of these studies can be found in Appendix 2.

Use of an emergency action plan was not reported as an outcome in any of the studies included in this review.

Change in quality of life

Two studies reporting on quality of life were not suitable for meta‐analysis (Dos Santos 2007; Frison 2002); both indicated a clinically significant improvement with treatment. Additional details are presented in Appendix 2.

Asthma symptom scores

A total of 20 studies reported on asthma symptoms (Aiguo 1999; Dos Santos 2007; Ekstrom 1989; Ford 1994; Frison 2002; Goodall 1981; Kiljander 1999; Kiljander 2006; Kiljander 2010; Kjellen 1981; Larrain 1991; Littner 2005; Maev 2002; Meier 1994; Nagel 1988; Sharma 2007; Sontag 2003; Stordal 2005; Susanto 2008; Teichtahl 1996); seven of these were cross‐over studies. Overall, results were conflicting; complete details can be found in Appendix 2.

Nocturnal symptoms

Eleven studies reported nocturnal asthma symptoms (Dos Santos 2007; Ekstrom 1989; Ford 1994; Frison 2002; Goodall 1981; Kiljander 1999; Kiljander 2006; Littner 2005; Sharma 2007; Susanto 2008; Teichtahl 1996). Again, results varied and details can be found in Appendix 2.

Treatment preferences

Treatment preferences were reported in two studies (Ford 1994; Goodall 1981), indicating that participants preferred to receive treatment over placebo. Results are described in greater detail in Appendix 2.

Adverse events

Adverse events were not a pre‐specified outcome for this review. However, for safety, we report available data from included studies related to this. Six studies reported that the incidence of adverse events was similar between intervention and control groups (Ekstrom 1989; Goodall 1981; Kiljander 2006; Kiljander 2010; Littner 2005; Sharma 2007); more detail is provided in Appendix 2.

Discussion

Summary of main results

The body of evidence to underpin treatment for gastro‐oesophageal reflux disease (GORD) in asthma has grown since the original publication of this review (Gibson 2003). This update contains evidence from 23 studies involving 2872 participants; despite this, only limited data are available to illuminate many of the pre‐specified outcomes of interest, and we concluded that certainty was low to moderate. Further, only two studies reported on effects of GORD treatment in the paediatric asthma population (Fallahi 2008; Stordal 2005), and two others reported on potential effects of surgical intervention (Larrain 1991; Sontag 2003). Meta‐analysis revealed that the benefit of medical treatment for reducing the number of people who experience a severe exacerbation of asthma is uncertain. Likewise, the benefit of medical treatment for GORD for people with asthma in relation to quality of life as measured by the Asthma Quality of Life Questionnaire (AQLQ) is uncertain. However, benefit was noted for other clinical outcomes including forced expiratory volume in one second (FEV₁; moderate‐certainty evidence) and use of rescue medication (moderate‐certainty evidence). Morning peak expiratory flow rate (PEFR) was meta‐analysed (but was not included in the summary of findings Table 1), indicating improvement with medical treatment for GORD. It must be noted that the magnitude of the differences noted in pooled analyses is not clinically important.

For outcomes for which meta‐analysis was not possible, there was some uncertainty about the benefit of intervention. It was not possible to definitively conclude that asthma symptoms were improved by medical treatment for GORD; this is mainly attributable to inconsistent reporting and use of both validated and invalidated scores by trialists. Nocturnal asthma symptoms were improved with H₂ receptor agonist treatment according to the two studies included in this review; however, similar agreement was not found on review of studies evaluating the effectiveness of proton pump inhibitors.

Of the two studies that captured data on treatment preferences, there was a clear preference of participants for treatment over placebo. However, a limitation that should be considered when this result is interpreted is that it is likely that the participant‐reported preference reflected improvement in GORD symptoms, rather than improvement in asthma symptoms.

Data related to acute presentations to the healthcare system (i.e. hospital admissions and emergency room or unscheduled doctor visits) were not captured or reported in any of the included studies. Therefore, effects of treatment of GORD on acute episodes of asthma and its effect on the healthcare system could not be determined.

Adverse event reporting was generally under‐performed in the included studies, although evidence available in this review does not indicate risk with GORD treatment for asthma symptom management.

The duration of the study period in the included studies varied from only 1 week to 104 weeks. Studies consistently averaged results over the entire treatment, which might cause bias against finding a treatment effect. Further, response to GORD treatment could be underestimated due to the short duration of treatment, as the effect of treatment could be delayed or could require additional time to manifest.

Given lack of standardisation in dosing protocols, it is difficult to determine the most effective prescription to be used in practice. Likewise, head‐to‐head comparisons between different types of medical treatments and between monotherapy versus combination therapy were lacking.

Overall completeness and applicability of evidence

Overall this review contains a good body of evidence with a total of 2872 participants, despite there being insufficient agreement across studies to ascertain clear benefit of GORD treatment for people with asthma. All trials assessed participants with diagnoses of both asthma and GORD. Based on the information provided, only two included studies used the current gold standard of objective GORD diagnosis (dual‐lumen pH probe with impedance) (Levin 1998; Stordal 2005), Most studies either used subjective means to determine the presence of GORD or did not provide enough information to determine the type of probe used. Most trials assessed participants with asthma across a range of spirometric severities, mainly from "mild" to moderate. Broad international recruitment was seen across these studies, with participants enrolled predominantly from the United states, Europe, United Kingdom, India, Brazil, Chile, Iran, and China. As a result, the outcomes of this review could be generalised to most people with asthma and GORD. A wide range of medical treatment regimens were included in this review of 23 studies. Most studies investigated the interventional effects of proton pump inhibitors, which are currently the gold standard of GORD management. Some included studies are outdated, and as a result, investigated agents that are no longer used in practice (i.e. cimetidine ‐ Goodall 1981; Larrain 1991 ‐ and ranitidine ‐ Aiguo 1999; Ekstrom 1989; Nagel 1988; Sontag 2003). Additionally, given the GORD diagnostic techniques and approaches used in the included studies, we are likely to have captured only individuals with comorbid acid reflux and asthma.

Although the pool of evidence has grown since the original publication of this review (Gibson 2003), there remains a paucity of evidence to underpin treatment with surgery or to guide practice in children. Further, few studies reported on the primary outcome of this review (acute asthma exacerbations), and none reported on clinically relevant, pre‐specified outcomes related to hospital utilisation.

Certainty of the evidence

Certainty of the evidence overall was moderate to low. This was impacted by the inability to pool many of the included studies into meta‐analyses. Further, there were no data in the included studies by which to understand the impact of intervention on several pre‐specified outcomes. Certainty was downgraded in the domain of inconsistency because substantial heterogeneity was noted based on the value of the I² statistic in the meta‐analyses and the lack of clear agreement across studies synthesised narratively. We further downgraded the certainty of evidence for indirectness because surgical intervention was underrepresented in the analysis. Certainty was also downgraded in the domain of imprecision owing to the small numbers of events and studies that could be pooled together.

Other domains assessed, including risk of bias and publication bias, presented no cause for downgrading in our assessment of certainty.

Potential biases in the review process

Bias may have been introduced in the conduct of the meta‐analyses for this review. We omitted and reported narratively instead all but one of the included cross‐over studies (Levin 1998). The reason for this is that data were not reported in a way that enabled meaningful pooling with parallel studies. Despite rigorous and systematic literature searches, evidence specifically related to surgical intervention is lacking; therefore, it is possible that effects reported in this review are under‐estimated or over‐estimated. Further, Sharma 2007 reported that all 99 participants treated with antireflux therapy experienced improvement and published highly significant benefits over placebo for all outcomes. The review author team in discussion with Dr Chris Cates (data and analysis advisor) believed these results to be implausible and elected to omit these data from the meta‐analyses to prevent drawing misleading conclusions. Likewise, FEV₁ data published in Fallahi 2008 appear implausibly low, no baseline values were reported, and it is not clear which measure of dispersion was used throughout. Clarification has been sought from Sharma 2007 and Fallahi 2008, with no response at the time of publication; hence, results from these studies have been synthesised only narratively in the text of this review.

We identified a number of additional intervention search terms during the course of the review. The search strategy we used contained terms to identify gastro‐oesophageal reflux; this enabled us to identify a range of interventions. The intervention search terms in our strategy were used to broaden rather than limit the gastro‐oesophageal reflux search. We checked if the new search terms added anything to our results and did not find any additional studies; however, we will add the new intervention terms to the search strategy when the review is next updated.

Additionally, although the pool of included studies has grown since publication of the original review (Gibson 2003), we were unable to report on one of our primary outcomes, as none of the included studies reported on hospital and emergency service utilisation or unscheduled doctor visits.

There is potential that inappropriate exclusion of relevant studies may have occurred in addition to data entry error. However, steps were taken to limit the impact of this by having two independent review authors screen, extract, and check data.

Agreements and disagreements with other studies or reviews

Current clinical guidelines suggest that people with asthma, particularly those with nocturnal symptoms, should be evaluated for GORD and treated when appropriate, even in the absence of obvious manifestations (King‐Biggs 2019; Moore 2010). If GORD is detected in people with asthma, the current recommendation is to treat with proton pump inhibitors (National Heart, Lung, and Blood Institute 2007). This recommendation is presently underpinned by single studies.

The original publication of this review was unable to detect clear benefit of treatment for GORD in people with asthma (Gibson 2003). Our updated review includes a greater number of randomised controlled trials, which enabled us to perform pooled analyses for some outcomes, allowing us to draw conclusions with more confidence. Our analyses indicate medical treatment for GORD could benefit people with asthma, specifically with reduced use of rescue medications and improved FEV₁ and morning PEFR. The latter is corroborated by a meta‐analysis of 11 studies comparing proton pump inhibitors to placebo (Chan 2011). From this publication, a subgroup analysis determined that in people with asthma and a diagnosis of GORD, a clinically and statistically significant improvement in morning PEFR was noted (mean difference (MD) 16.90 L/min, 95% confidence interval (CI) 0.85 to 32.95; 1004 participants, 7 studies; P = 0.006), in line with data presented in this review (MD 18.30 L/min, 95% CI ‐0.37 to 36.97; 1262 participants, 5 studies; Analysis 1.3; P = 0.05). Chan 2011 also assessed the effect of proton pump inhibitor therapy on AQLQ(S), noting there is uncertainty about benefit for this outcome (MD 0.20, 95% CI ‐0.078 to 0.472; 1828 participants, 4 studies; P = 0.16), again, supporting the findings of this review (MD 0.21, 95% CI ‐0.02 to 0.44; 1595 participants, 5 studies; P = 0.07; Analysis 1.5). Although it should be noted that for this outcome, Chan 2011 did not perform a subgroup analysis, and some participants included in the analysis did not have a confirmed diagnosis of GORD. Additionally, this study reviewed only studies investigating effects of proton pump inhibitors (Chan 2011), although this review included other classifications of GORD medical therapy. Evidence syntheses related to prokinetic agents and H₂ receptor antagonists are lacking and are likely reflective of the fact that in practice they are routinely administered as adjunct therapy to proton pump inhibitors (Wang 2013).

Available reviews of the existing literature refer to a single study to underpin the conclusion that medical treatment for GORD has uncertain benefit for a number of exacerbations for adults with a diagnosis of asthma (Bardin 2018 Harding 2019). Littner 2005 is included for analysis in this review, and hence, our findings strengthen this position. A recently published conference abstract reports the outcomes of a secondary analysis of morbidity data from four large asthma trials (Tang 2019). These results indicate that use of proton pump inhibitor therapy in children with comorbid GORD and asthma may increase the number of respiratory tract infection‐related asthma exacerbations; however there was no relationship between these variables in adults (Tang 2019).

This review indicates there is not a high risk of adverse events with GORD treatment in the short term (average study duration, 15 weeks) for people with asthma. It should be noted that some literature indicates potential for increased risk of conditions including vitamin B12 deficiency, fracture, and small‐intestine bacterial overgrowth, with longer‐term use of proton pump inhibitors for general treatment of GORD (Jaynes 2019; Nehra 2018). However, review authors note that the evidence underpinning these conclusions is of low certainty and should not outweigh consideration of the benefits of treatment for individual patients.

Likewise, evidence related to effects of surgical intervention for GORD in people with asthma is sparse. An available systematic review of 24 randomised and non‐randomised studies corroborates the findings of this review update (Field 1999). This review reported that surgical intervention improved asthma symptoms by 79% (Field 1999), in line with the two studies identified for inclusion in this review (Larrain 1991; Sontag 2003). Further, a more recent, retrospective study of 208 people with respiratory manifestations of GORD reported that 75% of participants experienced short‐term improvement in respiratory symptoms following fundoplication; in the long term, seven of these individuals relapsed, and four participants experienced latent improvement (Adaba 2014). Findings related to effects of surgical intervention on pulmonary function outcomes were similarly matched. Field 1999 reports that pulmonary function did not benefit as much as symptoms from surgical intervention, improving by 27%. Available evidence included in this review indicates that FEV₁ may see a small, initial improvement postoperatively that is not sustained (Larrain 1991). Further, PEFR was reported to improve, although the analysis was underpowered (Sontag 2003). The ability to conclude with certainty the benefits of surgical intervention for asthma control is limited by the lack of relevant, high‐quality studies.

A systematic review including randomised and non‐randomised studies reports that when proton pump inhibitor therapy is provided for children with GORD and asthma, benefits of treatment for asthma‐related outcomes are uncertain (Sopo 2009). The only randomised trial solely concerning children included in this systematic review is likewise the only one eligible for inclusion in this Cochrane Review (Stordal 2005). Despite a lack of evidence underpinning GORD medical treatment in children living with asthma, guidelines recommend treatment but advise clinicians not to promise improvement in asthma control (National Asthma Council Australia 2019). Based on the outcomes of this review, this recommendation is unlikely to be changed.

Chan 2011 performed a meta‐regression analysis to determine any effect of treatment duration on the outcome of morning PEFR. These researchers conclude that there is no evidence of a relationship between the variables, which is consistent with the observations noted in this review.

Overall, it appears that the findings of this review are in agreement with those of other available evidence syntheses, although other sources focus on single interventions and tended to include both randomised and non‐randomised studies in their analyses.

Figure 1

PRISMA study flow diagram.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Analysis 1.1

Comparison 1: Medical intervention for gastro‐oesophageal reflux compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children, Outcome 1: Number of participants with any moderate/severe acute exacerbations

Analysis 1.2

Analysis 1.3

Analysis 1.4

Comparison 1: Medical intervention for gastro‐oesophageal reflux compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children, Outcome 4: Use of "rescue" medications and emergency action plans: B2 use puffs per day

Analysis 1.5

Comparison 1: Medical intervention for gastro‐oesophageal reflux compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children, Outcome 5: Change in AQLQ

Analysis 1.6

Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Medical or surgical intervention, or both, for gastro‐oesophageal reflux disease compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children
Patient or population: treatment for asthma in adults and children with asthma and GORD Setting: outpatient Intervention: medical or surgical intervention, or both, for gastro‐oesophageal reflux disease Comparison: nil intervention, delayed intervention control, or placebo
Outcomes	Risk with nil intervention, delayed intervention control, or placebo	Risk with medical or surgical intervention, or both, for gastro‐oesophageal reflux disease	Relative effect (95% CI)	№. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Number of participants with moderate/severe acute exacerbations Duration of treatment: range 24 weeks to 26 weeks	112 per 1000	63 per 1000 (21 to 171)	OR 0.53 (0.17 to 1.63)	1168 (2 RCTs)	⊕⊕⊝⊝ LOW^a,b	Evidence is uncertain about the effect of medical treatment for GORD on acute exacerbations in adults. There were 60 adults with events in the experimental group (n = 732 individuals) compared to 49 adults in the control group (n = 436 individuals)
Change in FEV₁ (L) Duration of treatment: range 6 weeks to 26 weeks	Mean change in FEV₁ (L) was 0.50	MD 0.1 higher (0.05 higher to 0.15 higher)	‐	1333 (7 RCTs)	⊕⊕⊕⊝ MODERATE^a	There is moderate‐certainty evidence related to the effect of medical treatment for GORD on FEV₁ (L) in adults with asthma. Pooled analysis indicates that the mean difference in treatment was a small, non‐clinically significant improvement of 0.10 L compared to control
Use of "rescue" medications and emergency action plans: B2 use puffs per day Duration of treatment: range 14 weeks to 24 weeks	Mean use of "rescue" medications and emergency action plans: B2 use puffs per day was 2.11	MD 0.71 lower (1.2 lower to 0.22 lower)	‐	239 (2 RCTs)	⊕⊕⊕⊝ MODERATE^a	There is moderate‐certainty evidence related to the effect of medical treatment for GORD on use of rescue medications for people with asthma. A small reduction in average puffs per day (0.71) was detected in the treatment group (n = 115) compared to the control group (n = 124)
Change in quality of life Assessed with AQLQ Scale from 1 to 7; duration of treatment: range 8 weeks to 26 weeks	Mean change in AQLQ was 0.9244	MD 0.21 higher (0.02 lower to 0.44 higher)	‐	1595 (5 RCTs)	⊕⊕⊝⊝ LOW^a,c	Evidence is uncertain related to the effect of medical treatment for GORD on change in AQLQ for people with asthma. Mean difference of 0.21 was smaller than minimum important difference of 0.5 units
Healthcare utilisation ‐ not measured			‐	‐	‐	No studies included in this review reported on this outcome
*The risk in the intervention group (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). AQLQ: Asthma Quality of Life Questionnaire; CI: confidence interval; FEV₁: forced expiratory volume in 1 second; GORD: gastro‐oesophageal reflux disease; MD: mean difference; OR: odds ratio; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aRisk of bias assessment for included studies was unclear or high for important domains of randomisation and allocation concealment (‐1 point). ^bA small number of events resulted in a wide confidence interval (‐1 point). ^cPooled effect estimate is imprecise, with confidence interval including no difference (‐1 point).

Table 1. Summary of characteristics of included studies

Study ID	Country	N	Duration (weeks)	Intervention	Control	Outcomes of interest in this review
Aiguo 1999	China	54	6	a: Jiang Ni decoction twice daily b: 150 mg ranitidine twice daily plus cisapride 10 mg 3 times daily	Placebo	FEV₁; PEFR; asthma symptoms
Dos Santos 2007	Brazil	44	12	Pantoprazole 40 mg once daily	Placebo	FEV₁; mPEFR; ePEFR; asthma symptoms; nocturnal asthma symptoms; quality of life
Ekstrom 1989	Sweden	50	4	Ranitidine 150 mg twice daily	Placebo	FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; adverse events
Fallahi 2008	Iran	36	6	Omeprazole 20 mg twice daily	Placebo	FEV₁; PEFR
Ford 1994	UK	11	4	Omeprazole 20 mg once daily	Placebo	PEF; use of beta₂‐agonists; asthma symptoms
Frison 2002	Brazil	22	8	Lansoprazole 60 mg once daily plus cisapride 30 mg once daily plus behavioural management	Placebo plus behavioural management	FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms
Goodall 1981	UK	20	6	Cimetidine 200 mg 3 times daily plus 2 at night	Placebo	FEV₁; PEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms
Jiang 2003	China	30	6	Omeprazole 20 mg once daily plus domperidone 10 mg 3 times daily	Nil intervention	FEV₁; PEFR
Kiljander 1999	Finland	57	8	Omeprazole 40 mg once daily	Placebo	FEV₁; PEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms
Kiljander 2006	Multi‐national^a	770	16	Esomeprazole 40 mg twice daily	Placebo	Asthma exacerbations; FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; quality of life; adverse events
Kiljander 2010	Multi‐national^b	916	26	a: Esomeprazole 40 mg once daily b: Esomeprazole 40 mg twice daily	Placebo	Asthma exacerbations, FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; quality of life; adverse events
Kjellen 1981	Sweden	62	8	Conservative treatment (head elevation, warm water after meals, avoiding food 3 hours before sleep, refraining from aspirin and anticholinergic drugs, not raising intra‐abdominal pressure)	Nil intervention	FEV₁;use of beta₂‐agonists; asthma symptoms
Larrain 1991	Chile	90	26	a: Cimetidine 300 mg 4 times daily b: Surgery, posterior gastropexy with cardiac calibration	Placebo	FEV₁; asthma symptoms
Levin 1998	USA	11	8	Omeprazole 20 mg once daily	Placebo	FEV₁; mPEFR; ePEFR; quality of life
Littner 2005	USA	207	24	Lansoprazole 30 mg twice daily	Placebo	Asthma exacerbations, FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; quality of life; adverse events
Maev 2002	Russia	64	8	Omeprazole 20 mg twice daily	Placebo	FEV₁; PEF; asthma symptoms
Meier 1994	USA	15	6	Omeprazole 20 mg twice daily	Placebo	FEV₁; PEF; use of beta₂‐agonists; asthma symptoms
Nagel 1988	UK	15	1	Ranitidine 150 mg in the morning and 300 mg at night	Placebo	mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms
Sharma 2007	India	204	16	Omeprazole 20 mg twice daily plus domperidone 10 mg 3 times daily	Placebo	FEV₁; mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms; nocturnal asthma symptoms; adverse events
Sontag 2003	USA	75	104	a: Ranitidine 150 mg 3 times daily plus conservative treatment b: Surgery, Nissen fundoplication plus conservative treatment	Symptomatic treatment as needed, Mylanta 30 mL	PEF, asthma symptoms
Stordal 2005	Norway	38	12	Omeprazole 20 mg once daily	Placebo	FEV₁; use of beta₂‐agonists; asthma symptoms; quality of life
Susanto 2008	Indonesia	36	14	Esomeprazole 40 mg daily plus conservative treatment (lifestyle modifications and antacid medication if required)	Conservative treatment (lifestyle modifications and antacid medication if required)	mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms
Teichtahl 1996	Australia	25	12	Omeprazole 40 mg once daily	Placebo	FEV₁;mPEFR; ePEFR; use of beta₂‐agonists; asthma symptoms
ePEFR: evening peak expiratory flow rate; FEV₁: forced expiratory volume in one second; mPEFR: morning peak expiratory flow rate. ^aKiljander 2006 included participants from centres in the following countries: Argentina, Brazil, Bulgaria, Canada, Czech Republic, Finland, Hungary, Italy, Mexico, Romania, Sweden, and the United States of America. ^bKiljander 2010 included participants from centres in the following countries: Argentina, Bulgaria, Canada, Czech Republic, France, Germany, Hungary, Italy, Mexico, Poland, Portugal, Slovakia, and the United States of America.

Table 1. Summary of characteristics of included studies

Table 2. Sensitivity analysis: number of participants with an acute exacerbation

	Odds ratio	P value
	M‐H, Random, 95% CI	P value
Per publication data	0.53 (0.17 to 1.63)	0.27
Per trials website data	0.54 (0.17 to 1.74)	0.30
M‐H: Mantel‐Haenszel.

Table 2. Sensitivity analysis: number of participants with an acute exacerbation

Table 3. Sensitivity analysis: morning peak expiratory flow rate (L/min)

	Mean difference	P value
	IV, Random, 95% CI	P value
Random‐effects analysis	10.18 (‐1.92 to 22.28)	0.10
Fixed‐effect analysis	6.02 (0.56 to 11.47)	0.03
IV: inverse variance.

Table 3. Sensitivity analysis: morning peak expiratory flow rate (L/min)

Comparison 1. Medical intervention for gastro‐oesophageal reflux compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Number of participants with any moderate/severe acute exacerbations Show forest plot	2	1168	Odds Ratio (M‐H, Random, 95% CI)	0.53 [0.17, 1.63]

1.2 Change in FEV₁ (L) Show forest plot	7	1333	Mean Difference (IV, Fixed, 95% CI)	0.10 [0.05, 0.15]

1.3 Morning peak expiratory flow (L/min) Show forest plot	5	1262	Mean Difference (IV, Fixed, 95% CI)	6.02 [0.56, 11.47]

1.4 Use of "rescue" medications and emergency action plans: B2 use puffs per day Show forest plot	2	239	Mean Difference (IV, Fixed, 95% CI)	‐0.71 [‐1.20, ‐0.22]

1.5 Change in AQLQ Show forest plot	5	1595	Mean Difference (IV, Random, 95% CI)	0.21 [‐0.02, 0.44]

1.5.1 Adult	4	1559	Mean Difference (IV, Random, 95% CI)	0.24 [‐0.04, 0.51]
1.5.2 Pediatric	1	36	Mean Difference (IV, Random, 95% CI)	0.12 [‐0.24, 0.48]
1.6 AQLQ responder analysis Show forest plot	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only

Comparison 1. Medical intervention for gastro‐oesophageal reflux compared to nil intervention, delayed intervention control, or placebo for asthma in adults and children