Glucocorticoids for acute viral bronchiolitis in infants and young children

Summary of findings for the main comparison. Glucocorticoid versus placebo: summary of findings

Glucocorticoid versus placebo for acute viral bronchiolitis in infants and young children
Patient or population: infants and young children with acute viral bronchiolitis Settings: outpatients and inpatients Intervention: glucocorticoid versus placebo
Outcomes	*Illustrative comparative risks (95% CI)**		Steroid versus placebo	No of participants (studies)	Quality of the evidence (GRADE)
	Assumed risk¹	Corresponding risk
	Placebo	Steroid
Admissions (outpatients) Follow‐up: day 1	Medium risk population		RR 0.92 (0.78 to 1.08)	1762 (8)	high
	162 per 1000	149 per 1000 (126 to 175)
Admissions (outpatients) Follow‐up: day 7	Medium risk population		RR 0.86 (0.7 to 1.06)	1530 (5)	moderate
	250 per 1000	215 per 1000 (175 to 265)
Length of stay (inpatients) days	The mean length of stay ranged across control groups from 0.8 to 6.6 days	The mean length of stay in the intervention groups was 0.18 lower (0.39 lower to 0.04 higher)		633 (8)	high
The basis for the assumed risk* (for example, the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval 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.
¹Assumed risk for admissions was based on the median control group risks across the studies included in the meta‐analysis (medium risk).

Summary of findings 2. Glucocorticoid and epinephrine versus placebo: summary of findings

Glucocorticoid and epinephrine versus placebo for acute viral bronchiolitis in infants and young children
Patient or population: infants and young children with acute viral bronchiolitis Settings: outpatients Intervention: glucocorticoid and epinephrine versus placebo
Outcomes	*Illustrative comparative risks (95% CI)**		Steroid versus placebo	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk¹	Corresponding risk
	Placebo	Steroid
Admissions (outpatients) Follow‐up: day 1	179 per 1000	115 per 1000 (72 to 186)	RR 0.65 (0.4 to 1.05)	400 (1)	Low	NNT: not calculated for non‐significant findings
Admissions (outpatients) Follow‐up: day 7	264 per 1000	169 per 1000 (116 to 251)	RR 0.65 (0.44 to 0.95)	400 (1)	Low	NNT: 11 (95% CI 7 to 76) (based on unadjusted analysis results)
The basis for the assumed risk* (for example, the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval NNT: number needed to treat 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.
¹Assumed risk for admissions was based on the control group risk in the single study included (Plint 2009).

Background

Description of the condition

Acute viral bronchiolitis is the most common acute infection of the lower respiratory tract during the first year of life (Wright 1989). It is diagnosed clinically in infants and young children, based on a history of rhinorrhoea and low‐grade fever that progress to cough and respiratory distress, with findings of tachypnoea, chest retractions and wheeze, crackles, or both, on examination (Bush 2007; Smyth 2006; Taussig 2008). Respiratory syncytial virus (RSV) is responsible for the majority of cases, usually in seasonal epidemics (Smyth 2006; Yusuf 2007). Other viral agents, particularly rhinovirus, human metapneumovirus, bocavirus and adenovirus, may also be involved as single or dual infections (Calvo 2010; Kusel 2006; Mansbach 2008b; Mansbach 2012). Although bronchiolitis is usually a straightforward diagnosis, some variability in its definition exists. This may be due to poor agreement on the identification of early childhood wheezing phenotypes and worldwide differences in disease semantics (Brand 2008; Everard 2009; Mansbach 2008a).

Bronchiolitis is a major cause of clinical morbidity and its financial health burden is substantial. Population‐based studies in developed countries suggest an incidence ratio of approximately 10% within the first year of life, with hospital admissions up to 3% (Koehoorn 2008; Mansbach 2005; Shay 1999; Wright 1989). While mortality is rare, hospitalisations have increased steadily in North America and Europe over the past 10 to 20 years (Langley 2003; Shay 1999; van Woensel 2002), with rising inpatient health care costs (Langley 1997; Paramore 2004; Pelletier 2006). Additionally, a majority of cases with mild illness cared for in the community are responsible for a considerable number of outpatient visits, loss of parental work time and decreased quality of life (Carroll 2008; Mansbach 2007; Robbins 2006). RSV infection, including bronchiolitis, is a major cause of childhood morbidity and mortality at a global level (Nair 2010).

Bronchiolitis involves acute inflammation of the bronchiolar airways initiated by viral infection, regardless of the causative agent. Airway oedema, necrosis and mucous plugging are the hallmark pathological features, and air flow obstruction ensues (Taussig 2008). Factors underlying disease severity are only partially understood, but clinical determinants include lower age, prematurity, chronic lung, heart or neurological disease, immunodeficiency and ethnicity (Damore 2008; Figueras‐Aloy 2008; Meissner 2003; Simoes 2003; Simoes 2008). There is likely a complex interplay between host (i.e. genetic markers), agent (i.e. viral loads, specific agents and co‐infections) and environmental factors (i.e. crowding, tobacco smoke exposure) (Colosia 2012; Collins 2008; DiFranza 2012; Mansbach 2012; Miyairi 2008;Papadopoulos 2002). Basic, translational and clinical research studies are elucidating the association between bronchiolitis, preschool wheezing disorders and later asthma (Martinez 2005; Perez‐Yarza 2007; Singh 2007; Sly 2010).

Description of the intervention

The current treatment for bronchiolitis is controversial. There is substantial variation in its management throughout the world, reflecting the absence of clear evidence for any single treatment approach (Babl 2008; Barben 2003; Brand 2000; Gonzalez 2010; Mansbach 2005; Plint 2004). Many interventions failed to show consistent and relevant effects (Bialy 2011). Recently, both nebulised epinephrine and hypertonic saline have emerged as options for improving relevant outcomes in outpatient and inpatient populations, respectively (Hartling 2011a; Zhang 2011). However, no routine treatment is yet recommended by most evidence‐based clinical practice guidelines worldwide (AAP 2006; Baumer 2007; Turner 2008).

The case of glucocorticoids highlights the uncertainties of research in this field. Trials assessing their use date back to the 1960s, with different potencies, modes of administration, dosages and regimens of these drugs having been recommended (Connolly 1969; Leer 1969). However, results from randomised clinical trials (RCTs) have been heterogeneous, leading to ongoing controversy regarding their use. Differences in participants, care settings and outcomes may account for these conflicting results, and have led to distinct interpretations (Everard 2009; Guilbert 2011; Hall 2007; Weinberger 2003; Weinberger 2007).

How the intervention might work

Glucocorticoid use in bronchiolitis was originally thought to have equivalent benefits to those in acute asthma. Similarities between clinical findings were expected to express equivalent biological and physiological mechanisms attributable to inflammation (Leer 1969). However, evidence suggests there is heterogeneity in inflammatory pathways and mediators activated in different wheezing phenotypes which may underlie bronchiolitis (for example, neutrophil‐ versus eosinophil‐mediated inflammation) (Halfhide 2008). Mechanistic studies have shown that glucocorticoids have limited anti‐inflammatory effects in this condition (Buckingham 2002; Somers 2009) and there is an ongoing debate regarding their efficacy in acute virus‐induced wheezing in preschool children (Bush 2009; Ducharme 2009; Panickar 2009). Further, potential benefits need to be considered in light of possible short‐ and long‐term adverse effects of glucocorticoid use. While the interactive effect of bronchodilators and glucocorticoids has been widely known in asthma, both at a clinical and biological level, its use as a putative treatment option in bronchiolitis has only been explored recently (Plint 2009).

Why it is important to do this review

While guideline implementation has changed prescription patterns, glucocorticoids are still widely used (Barben 2000; Barben 2008; David 2010). The latest version of this review integrated critical results from the two largest multi‐centre studies in this area (Corneli 2007; Plint 2009) and examined the use of combined therapy with bronchodilators or adrenaline. We continue to update the current body of evidence in order to adequately assess the efficacy and safety of glucocorticoids in bronchiolitis.

Objectives

To review the efficacy and safety of systemic and inhaled glucocorticoids in children with acute viral bronchiolitis.

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs irrespective of risk of bias, sample size, publication status or language of publication.

Types of participants

Studies should include infants and young children ≤ 24 months of age with acute viral bronchiolitis. Bronchiolitis was defined clinically as a first episode of acute wheezing, respiratory distress and clinical evidence of a viral infection (cough, coryza, fever). Many bronchiolitis trial reports do not specify clinical findings required for participant inclusion (King 2004); we included all studies if other diagnoses (for example, pneumonia) could be excluded. We did not restrict inclusion based on specific findings on examination (for example, crackles) or viral aetiology.

We excluded studies in which any participant had a history of wheezing or respiratory distress (one or more previous episodes), a formal diagnosis of asthma, or if reporting of these items was unclear. We focused on first time wheezing so results could be directly pertinent to infants with 'typical' viral bronchiolitis, as opposed to children with acute recurrent wheezing. We did not exclude trials based on other reported participant characteristics, including gestational age and co‐morbidities.

We included studies of both inpatients and outpatients (ambulatory care and/or emergency department), and excluded trials in the intensive care setting or with intubated and/or ventilated participants.

Types of interventions

The interventions of interest were short‐term systemic or inhaled glucocorticoids administered for the acute care of bronchiolitis. We considered all types of glucocorticoids, dosages, durations and routes of administration. Glucocorticoids could be administered alone or combined with co‐interventions (for example, bronchodilators), used with or without a fixed protocol. We excluded trials assessing the use of longer courses of glucocorticoids started during the acute phase for the prevention of post‐bronchiolitic wheezing.

Comparators included either placebo or another intervention (for example, bronchodilators, other glucocorticoid). Inhaled isotonic saline is frequently used as a placebo control for inhaled drugs. We excluded studies comparing different doses or regimens of the same glucocorticoid.

Types of outcome measures

We selected primary outcomes based a priori on clinical relevance and patient importance; secondary outcomes assessed other relevant health domains (clinical severity, pulmonary function, healthcare use, patient/parent‐reported symptoms and status, and harms). We included studies if they reported numeric data on at least one primary or secondary outcomes assessed within the first month after acute bronchiolitis. We considered different timings of outcome assessment, based on a priori relevance and available data.

Primary outcomes

Rate of admission by days one and seven for outpatient studies.
Length of stay (LOS) for inpatient studies.

Secondary outcomes

Clinical severity scores.
O₂ saturation, respiratory rate and heart rate.
Hospital re‐admissions (for inpatient studies) and return healthcare visits (for all studies); LOS (for outpatient studies).
Pulmonary function tests.
Symptoms and quality of life.
Short‐ and long‐term adverse events.

We selected the following time points and intervals for clinical scores, O₂ saturation, respiratory and heart rate: 60 and 120 minutes, three to six hours, six to 12 hours, 12 to 24 hours, 24 to 72 hours, and three to 10 days. The time points selected for re‐admissions and return visits were days 1 to 10, and 11 to 30. We also considered data on all other reported outcomes.

Search methods for identification of studies

The previous version of this review used an inclusive search strategy as part of a comprehensive systematic review evaluating the effect of three types of interventions in bronchiolitis (glucocorticoids, epinephrine and other bronchodilators) (Hartling 2011b).

Electronic searches

Previously we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2009, Issue 4), which contains the Cochrane Acute Respiratory Infections Group's Specialised Register, MEDLINE (1950 to November Week 2, 2009), EMBASE (1980 to Week 47, 2009), LILACS (Latin American and Caribbean Center on Health Sciences Information) (1982 to 25 November 2009), Scopus® (1823 to 25 November 2009) and IRAN MedEx (1998 to 26 November 2009).

We developed search strings by scanning search strategies of relevant systematic reviews and examining index terms of potentially relevant studies. We applied and modified a validated RCT filter according to each database (Glanville 2006). We applied no publication or language restrictions. The search strings for each database can be found in Appendix 1 to Appendix 6.

For this 2013 update we searched the Cochrane Central Register of Controlled Trials (CENTRAL) 2012, Issue 12, part of The Cochrane Library, www.thecochranelibrary.com (accessed 21 January 2013), which contains the Cochrane Acute Respiratory Infections Group's Specialised Register, MEDLINE (October 2009 to January week 2, 2013), EMBASE (November 2009 to January 2013), LILACS (Latin American and Caribbean Center on Health Sciences Information) (2009 to January 2013) and Scopus (2009 to January 2013) (Appendix 7).

Searching other resources

To identify unpublished studies and studies in progress we searched the following clinical trials registers on 1 August 2012: ClinicalTrials.gov and ICTRP Search Portal – World Health Organization. We searched the following conference proceedings: Pediatric Academic Societies (2003 to 2012), European Respiratory Society (2003 to 2011), American Thoracic Society (2006 to 2012).

We identified additional published, unpublished or ongoing studies by handsearching reference lists and included or excluded studies of relevant reviews. In addition, we contacted topic specialists.

Data collection and analysis

Selection of studies

Five review authors (AP, LB, LH, NH or RF) independently screened the titles, keywords and abstracts (when available) to determine if an article met the inclusion criteria. These review authors independently assessed the full text of all articles classified as 'include' or unclear' using a standardised form. We resolved disagreements by consensus or by an arbitrator (AP, TK, DJ, or RF).

Data extraction and management

We extracted data using a standardised form in paper or electronic format (available from authors). Seven review authors extracted data (LB, LH, AM, HM, RF, OT or JF) and three review authors (LB, AM or RF) independently checked for accuracy and completeness. We resolved discrepancies by consensus or in consultation with a third review author (TK, AP or DJ). A statistician (BV) checked all quantitative data during analysis. Extracted data included study characteristics, funding, inclusion/exclusion criteria, participant characteristics, interventions, outcomes and results.

Assessment of risk of bias in included studies

We used the Cochrane 'Risk of bias' assessment tool, which includes seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other sources of bias (Higgins 2011). We assessed blinding and incomplete outcome data separately for the following groups of outcomes: healthcare use (rate of admission, LOS, hospital re‐admissions and return healthcare visits); clinical parameters (clinical severity scores, O₂ saturation, respiratory rate and heart rate); pulmonary function; patient/parent‐reported outcomes (symptoms and quality of life measures) and other outcomes such as adverse events. Where trial protocols or trial registers were unavailable, we assessed selective outcome reporting by comparing outcomes reported in the methods and results sections. We summarised risk of bias for each study across outcomes based on individual domain assessments ('high' if one or more domains were high; 'low' if all domains were low; 'unclear' for all other studies).

Three review authors (LB, LH or RF) independently assessed the risk of bias of the included studies; we resolved discrepancies by consensus. One review author (OT) assessed study reports written in Turkish. We pilot tested the risk of bias tool on a sample of five studies and used the results to adapt decision rules (available from authors).

Grading the body of evidence

We used the Evidence‐Based Practice Centers Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, based on the standard GRADE system (GRADE 2009; Owens 2010), to assess domain‐specific and overall strength of evidence on three relevant outcomes: length of stay or admission rate, clinical severity scores and adverse events. Two review authors (LH, RF) independently graded the body of evidence using adapted decision rules.

We examined the following domains: risk of bias, consistency, directness and precision. Risk of bias was considered as low or medium, as we only included RCTs. There is limited evidence regarding clinically significant and patient‐important between‐group differences in this field. We therefore defined a priori thresholds of clinical relevance based on expert opinion and GRADE guidance for the precision domain: risk ratio reduction > 20% for admissions, reduction in LOS > 0.5 days and clinical scale effect sizes based on GRADE guidance (GRADE 2009). We graded overall strength of evidence 'high', 'moderate' or 'low' based on the likelihood of further research changing our confidence in the estimate of effect (when evidence was unavailable or did not permit estimation of an effect, it was considered insufficient).

All decisions were made explicitly and inter‐rater agreement was calculated (data available from authors). We resolved discrepancies by consensus among two review authors (LH, RF).

Measures of treatment effect

We pooled dichotomous variables using risk ratios (RRs). We derived the number needed to treat to benefit (NNTB) for significant results from primary outcomes. Since the only comparison with significant differences was based on a single trial, the NNTB is shown for that trial's baseline risk.

We analysed measurement scale outcomes as continuous variables. For continuous variables measured on the same scale (for example, respiratory rate), we calculated mean differences (MD) for individual studies and mean differences for the pooled estimates. For those measured on different scales (for example, clinical scores), we calculated MDs for separate studies and standardised MD (SMD) for the pooled estimates. We used changes from baseline for all continuous variables.

Unit of analysis issues

Some of the studies included in this review were multi‐arm or factorial studies in which more than two intervention groups were eligible to contribute several comparisons to a single meta‐analysis. For example, a trial might compare glucocorticoid versus placebo in two arms, and glucocorticoid + bronchodilator versus placebo + bronchodilator in another two arms, with both contributing to the overall glucocorticoid versus placebo comparison. When the comparisons were independent, i.e. with no intervention group in common, we included data from these arms with no transformation and we shown them separately in each forest plot. If needed and feasible, we pooled the active groups to avoid double‐counting of the comparator group when there was more than one active group: for example, two glucocorticoid groups versus placebo. We did not include any treatment groups twice in the same meta‐analysis.

Guidance regarding the analysis of factorial trials mandates caution when results suggest positive interaction/additive effects ('synergism') between study treatments (McAlister 2003; Montgomery 2003). This was the case for a large trial included in this review. We therefore chose to include comparisons separately in meta‐analysis ('within the table analysis'): for example, for the glucocorticoid versus placebo comparison, we included separately glucocorticoid + bronchodilator versus placebo + bronchodilator and glucocorticoid + placebo versus double placebo. We also performed sensitivity analysis pooling all arms ('at the margins analysis').

Dealing with missing data

We extracted information on incomplete outcome data and we classified trials that performed intention‐to‐treat (ITT) analysis as either ITT with all data, ITT with imputation of missing data, ITT with available case analysis, per protocol analysis or treatment‐received analysis (Higgins 2011). We did not impute missing data for drop‐outs. We estimated unreported means from figures or imputed from medians if possible. We computed standard deviations (SDs) from available data (i.e. standard errors, confidence intervals (CI) or P values) when missing. Failing this, we estimated them from ranges and inter‐quartile ranges, or imputed them from a similar study. When standard deviations of change from baseline values were unavailable, we estimated correlation at 0.5 (Follmann 1992; Wiebe 2006). We occasionally encountered clinical score results presented as dichotomous data, for example, using a cut‐off score or time‐to‐event analysis. When methods were feasible and assumptions judged reasonable, we used existing approaches to re‐express odds ratios as standardised mean differences, thus allowing dichotomous and continuous data to be pooled together (Higgins 2011). When data were unavailable for one of the predefined timings of outcome measurement, we used the time point closest or any time point in the range. If there was more than one time point, we chose the one with the largest magnitude of change.

We did not contact trial authors of the individual studies to obtain additional data.

Assessment of heterogeneity

We quantified statistical heterogeneity using the I² statistic. We used the following intervals for interpreting I² statistic values: 0% to 30% low heterogeneity; 30% to 50% moderate heterogeneity; 50% to 75% substantial heterogeneity; and 75% to 100% considerable heterogeneity (Higgins 2011).

Assessment of reporting biases

We assessed reporting biases for the main comparisons and primary outcomes by visual interpretation of funnel plots and testing for funnel plot asymmetry (Egger test) (Higgins 2011).

Data synthesis

We meta‐analysed quantitative results within the different comparisons when studies were consistent on clinical grounds and had available outcome data; we imposed no restrictions based on risk of bias. We performed separate meta‐analyses for studies involving inpatients and outpatients.

We combined results using random‐effects models regardless of heterogeneity, due to expected differences in interventions, outcomes and measurement instruments. We calculated fixed‐effect models in a sensitivity analysis. We conducted meta‐analyses of dichotomous outcomes using Mantel‐Haenszel methods. We used inverse variance methods for continuous outcomes and measurement scales, and combined dichotomous and continuous data into a standardised mean difference whenever needed (Higgins 2011). All results are reported with 95% CI. We used Review Manager software for data management and analysis (RevMan 2012).

Subgroup analysis and investigation of heterogeneity

We planned to investigate heterogeneity by conducting subgroup analyses based on pre‐specified study‐ and participant‐level characteristics. The following subgroups were considered:

Protocolised use of bronchodilators (studies with protocolised use versus no/unclear protocolised use).
RSV status (studies with all participants exclusively RSV‐positive versus some RSV‐negative/unspecified RSV status).
Age of participants (studies with all participants exclusively less than 12 months of age versus some participants older than 12 months/unspecified age).
Atopy (studies with all participants exclusively atopic versus some participants not atopic/unspecified atopic status).
Glucocorticoid: type of glucocorticoids; and daily and overall dose (high versus low).

We explored potential positive or negative (i.e. 'synergistic' or 'antagonistic') interactions between glucocorticoids and bronchodilators by distinguishing trials where bronchodilator use was protocolised (i.e. comparing glucocorticoids + bronchodilator versus placebo + bronchodilator) from studies where use was either at the discretion of the physician or not allowed (Gurusamy 2009). The choice of RSV, age and atopy was based on clinical or biological evidence suggesting possible effect modification of glucocorticoid effects by these parameters. We studied drug type and dose to explore distinct glucocorticoid pharmacokinetic and pharmacodynamic properties; dosing was based on prednisolone equivalents.

We planned to perform subgroup analyses only on the review's primary outcomes. We also collected data from studies that analysed these subgroups at a study level. We assessed subgroup differences comparing changes in effect estimate and CI overlap; statistical tests or meta‐regression techniques were not used.

Sensitivity analysis

We decided a priori to perform sensitivity analyses on primary outcome results of trials with overall low risk of bias. We also checked for differences in the direction and magnitude of primary outcome results when using fixed‐effect models, as well as using pooled data from all factorial trial arms ('at the margins analysis').

Results

Description of studies

Results of the search

The initial 2009 comprehensive search of all electronic databases identified 2249 records, of which 344 were potentially relevant. Handsearching had identified four more studies and overall 348 full‐text articles had been assessed for eligibility. Of 91 studies that used glucocorticoids, 17 trials fulfilled inclusion criteria.

The 2013 search identified 280 further records, of which 13 were assessed for eligibility using full text but all were excluded (flowchart in Figure 1).

Figure 1

Flow of citations through the search and screening procedures of the 2009 review and this 2012 update, studies included in the review and comparisons addressed (GC: glucocorticoids)

Included studies

We included 17 trials with 2596 randomised participants. We considered different comparisons separately between glucocorticoids, alone or with fixed co‐interventions, and either placebo or active controls. Included trials contributed to one or more comparisons, depending on trial arms (Figure 1).

Design, centres and sample sizes

Fifteen trials were parallel‐designed, 14 of which were double‐armed (Bentur 2005; Berger 1998; Cade 2000; Corneli 2007; De Boeck 1997; Goebel 2000; Gomez 2007; Klassen 1997; Mesquita 2009; Richter 1998; Roosevelt 1996; Schuh 2002; Teeratakulpisarn 2007; Zhang 2003) and one was six‐armed (Barlas 1998). Two trials were factorial two‐by‐two (Kuyucu 2004; Plint 2009).

Eleven trials were single‐centred and five included multiple centres (range: 2 to 20) (Cade 2000; Corneli 2007; Goebel 2000; Plint 2009; Teeratakulpisarn 2007); one trial did not clearly report this item (Bentur 2005). All trials were conducted in a single country, either in North, Central or South America, Europe and the Middle East or Asia.

Sample size calculations were reported in 12 trials (Bentur 2005; Berger 1998; Cade 2000; Corneli 2007; Klassen 1997; Mesquita 2009; Plint 2009; Richter 1998; Roosevelt 1996; Schuh 2002; Teeratakulpisarn 2007; Zhang 2003); the outcome used for sample size calculation was the reported primary outcome in all except one trial (Richter 1998). The overall median number of participants per trial was 72 (range 32 to 800), with two large trials counting 600 and 800 (Corneli 2007; Plint 2009, respectively), and all others fewer than 200.

Funding was reported in nine studies, three of which had pharmaceutical industry support (Cade 2000; Richter 1998; Schuh 2002).

Setting and participants

Outpatients were included in eight trials, with 1824 randomised participants and a median of 85 participants per trial (range: 42 to 800) (Barlas 1998; Berger 1998; Corneli 2007; Goebel 2000; Kuyucu 2004; Mesquita 2009; Plint 2009; Schuh 2002). Outpatient settings mostly included paediatric emergency departments. Nine trials included inpatients only, with 772 participants and a median of 61 participants per trial (range: 32 to 179) (Bentur 2005; Cade 2000; De Boeck 1997; Gomez 2007; Klassen 1997; Richter 1998; Roosevelt 1996; Teeratakulpisarn 2007; Zhang 2003). Few details were reported regarding criteria for hospitalisation and the type of admission unit in which patients received care, except for one inpatient trial report (Teeratakulpisarn 2007).

In most trials bronchiolitis was defined by clinical findings; wheezing was always required. Three trials restricted inclusion to bronchodilator responders (Goebel 2000 ‐ outpatients; Teeratakulpisarn 2007 and Zhang 2003 ‐ inpatients). Seven trials only included participants under the age of 12 months, all of which had a mean or median participant age below six months (Bentur 2005; Cade 2000; Corneli 2007; Plint 2009; Richter 1998; Roosevelt 1996; Zhang 2003).

Bronchiolitis severity thresholds were used for inclusion in eight outpatient (Barlas 1998; Berger 1998; Corneli 2007; Goebel 2000; Kuyucu 2004; Mesquita 2009; Plint 2009; Schuh 2002) and two inpatient trials (Gomez 2007; Klassen 1997). Severity was based on clinical scales or respiratory parameters, and thresholds varied. The Respiratory Distress Assessment Instrument (RDAI) baseline score thresholds varied between two and six (less than four usually considered mild bronchiolitis).

Thirteen trials reported testing for RSV at least in a portion of participants, and three trials only included RSV‐positive patients (Bentur 2005; Cade 2000; De Boeck 1997). Prevalence of RSV in the remaining 10 trials varied from 33% to 89% (Barlas 1998; Berger 1998; Corneli 2007; Goebel 2000; Klassen 1997; Mesquita 2009; Plint 2009; Richter 1998; Roosevelt 1996; Schuh 2002).

Atopic status was reported in nine trials (Barlas 1998; Berger 1998; Cade 2000; Plint 2009; Richter 1998; Roosevelt 1996; Schuh 2002; Teeratakulpisarn 2007; Zhang 2003), while one trial reported a family history of wheezing (Corneli 2007). Definitions for atopy and methods of assessment were rarely provided, and when reported were heterogeneous. No trials excluded participants with a history of atopy.

Children with chronic cardiac, pulmonary or neurological conditions or immunodeficiency were frequently excluded. All or some premature infants were explicitly excluded in seven trials (Cade 2000; Corneli 2007; De Boeck 1997; Goebel 2000; Plint 2009; Schuh 2002; Teeratakulpisarn 2007). Other criteria for exclusion were length of illness and glucocorticoid‐related parameters (previous use, history of adverse events, specific contraindications to their use).

Subgroup analyses within studies were reported in five trials (Bentur 2005; Cade 2000; Corneli 2007; Plint 2009; Teeratakulpisarn 2007), two of which being pre‐specified (Corneli 2007; Plint 2009). Subgroups were based on age, RSV status, family or personal history of atopy and eczema, duration and severity of illness, and exposure to smoke and/or dampness.

Interventions

There was heterogeneity regarding the choice of glucocorticoid, its dosage, route of administration and duration of treatment. Dexamethasone was the most frequently tested drug (11 trials). Nine trials used systemic dexamethasone, either oral (Corneli 2007; Klassen 1997; Mesquita 2009; Plint 2009; Schuh 2002), intramuscular (Kuyucu 2004; Roosevelt 1996; Teeratakulpisarn 2007) or intravenous (De Boeck 1997). Single‐day doses were administered for one to five days. Initial dosing was higher (0.5 to 1 mg/kg), with later doses ranging from 0.15 to 0.6 mg/kg. The highest overall dose was seen in Plint 2009 and Schuh 2002 (1 mg/kg followed by 0.6 mg/kg for five days), and the lowest in Mesquita 2009 (single‐dose 0.5 mg/kg). Two trials used inhaled dexamethasone (0.2 mg to 0.25 mg every four to six hours), at least for one day, or until discharge for inpatients (Bentur 2005; Gomez 2007). Systemic prednisone or prednisolone were tested in four trials, three oral (Berger 1998; Goebel 2000; Zhang 2003) and one intravenous (Barlas 1998). Duration varied between one and five days (1 to 2 mg/kg/day, once or twice daily). Three trials used inhaled budesonide (0.5 mg to 1 mg, once or twice daily) for one to six weeks (Barlas 1998; Cade 2000; Richter 1998).

Details on placebos were reported in nine trials. Inhaled placebos included mist (Barlas 1998) and 0.9% saline (Bentur 2005; Richter 1998). Protocolised standard of care was used as a control arm in Zhang 2003.

Eleven trials used protocolised bronchodilators in both glucocorticoid and placebo arms. The choice of bronchodilator, its dose and frequency varied substantially. Seven trials used salbutamol (Barlas 1998; Berger 1998; Goebel 2000; Gomez 2007; Klassen 1997; Kuyucu 2004; Schuh 2002), four used epinephrine (Bentur 2005; Kuyucu 2004; Mesquita 2009; Plint 2009) and one used salbutamol and ipratropium bromide (De Boeck 1997). Nebulised salbutamol was administered during emergency department stay (first two to four hours), or each four to six hours at home or during hospitalisation (1.5 mg to 2.5 mg, or 0.15 mg/kg). Oral administration was also allowed in Goebel 2000. Nebulised epinephrine was administered every six hours to inpatients, or once or twice in the emergency department for outpatients (1 mg to 3 mg). All other trials used bronchodilators at the discretion of the attending physician, often with guidance on the choice of drug and dosage. Additional use of glucocorticoids was often restricted. Supportive measures, i.e. oxygen and intravenous or nasogastric fluids, were usually reported.

Outcomes

Pre‐defined primary outcomes were specified in 12 trials (Cade 2000; Corneli 2007; Goebel 2000; Klassen 1997; Kuyucu 2004; Mesquita 2009; Plint 2009; Richter 1998; Roosevelt 1996; Schuh 2002; Teeratakulpisarn 2007; Zhang 2003), three of which reported more than one primary outcome (Kuyucu 2004; Richter 1998; Teeratakulpisarn 2007). Only the two largest trials used admission as a primary outcome (Corneli 2007; Plint 2009). Other primary outcomes included clinical scales (Goebel 2000; Klassen 1997; Kuyucu 2004; Mesquita 2009; Richter 1998; Schuh 2002), clinical severity parameters or duration of disease (Kuyucu 2004; Roosevelt 1996; Teeratakulpisarn 2007) and symptoms (Cade 2000; Zhang 2003). Timings of primary outcome assessment were reported in 11 trials, six of which used multiple time points. Sample size calculations were either not reported or based on secondary outcomes in Goebel 2000, Kuyucu 2004 and Richter 1998.

Reported outcomes included healthcare use domains and clinical severity parameters (all trials), pulmonary function (De Boeck 1997), patient/parent‐reported symptoms and status (seven trials: Berger 1998; Cade 2000; Plint 2009; Roosevelt 1996; Schuh 2002; Teeratakulpisarn 2007; Zhang 2003) and other outcomes, including adverse events (10 trials: Bentur 2005; Cade 2000; Corneli 2007; Klassen 1997; Kuyucu 2004; Plint 2009; Richter 1998; Roosevelt 1996; Teeratakulpisarn 2007; Zhang 2003). Not all outcome and time point results were reported.

Admission rates were assessed in all eight outpatient trials, both by day 1 (all trials) and day 7 (three trials; Corneli 2007; Plint 2009; Schuh 2002). Kuyucu 2004 and Goebel 2000 reported admissions by days 5 and 6, respectively, and were pooled with day 7 results. LOS was reported in eight of nine inpatient trials (except Roosevelt 1996) and three outpatient trials (Berger 1998; Corneli 2007; Goebel 2000). Criteria for admission or discharge were rarely reported. Considerable variability was found in control group admission rates (from 0% to 44% by day 1, and 0% to 49% by day 7) and mean LOS (0.8 to 6.6 days) (Table 1). Hospital re‐admissions for inpatients and return healthcare visits up to one month were mentioned in six trials, with variable assessment methods (Berger 1998; Klassen 1997; Plint 2009; Roosevelt 1996; Schuh 2002; Teeratakulpisarn 2007).

See: Summary of findings for the main comparison Glucocorticoid versus placebo: summary of findings; Summary of findings 2 Glucocorticoid and epinephrine versus placebo: summary of findings

Table 1. Placebo group risk of admission/length of stay

Study	Placebo group ‐ participants	Placebo group ‐ primary outcomes
OUTPATIENT STUDIES		Risk of admission day 1 (%)	Risk of admission day 7 (%)
Barlas 1998	30	17%	NR
Berger 1998	18	11%	NR
Corneli 2007	295	41%	49%
Goebel 2000	24	8%	21%
Kuyucu 2004	11	0%	0%
Mesquita 2009	32	22%	NR
Plint 2009	201	18%	26%
Schuh 2002	34	44%	47%
INPATIENT STUDIES		Length of stay (mean ± SD days)
Bentur 2005	32	6.3 ± 8.8
Cade 2000	79	2 ± 2.2
De Boeck 1997	15	6.6 ± 1.2
Gomez 2007	25	0.8 ± 0.2
Klassen 1997	32	2 ± 0.7
Richter 1998	19	3 ± 1.6
Teeratakulpisarn 2007	85	2.8 ± 1.7
Zhang 2003	24	5 ± 3.3

NR = not reported
SD = standard deviation

Clinical severity scales were assessed in all except one trial (Zhang 2003), often using more than one scale (Corneli 2007; Plint 2009; Richter 1998; Schuh 2002). Measurement instruments were developed specifically for nine trials (Barlas 1998; Bentur 2005; Berger 1998; Cade 2000; De Boeck 1997; Goebel 2000; Richter 1998; Roosevelt 1996; Teeratakulpisarn 2007), mostly based on previous scales by Schuh 1990, Tal 1983 and Westley 1978. The RDAI was used in eight trials (Corneli 2007; Gomez 2007; Klassen 1997; Kuyucu 2004; Mesquita 2009; Plint 2009; Richter 1998; Schuh 2002). Corneli 2007 and Plint 2009 also used the Respiratory Assessment Change Score (RACS), based on RDAI and respiratory rate (both originally reported by Lowell 1987). All scales included items on wheezing and accessory muscle use; other respiratory items (for example, timing or location of wheezing) or disease domains (for example, general status, nutrition) were less frequently used. Oxygen saturation, respiratory and heart rates were reportedly measured in most trials. Heterogeneity in timings of repeated measurements was found; the two most frequently time points assessed were 60 minutes and three to six hours.

Measurement of patient/parent‐reported symptoms was inconsistent. Five trials reported symptoms data (Cade 2000; Plint 2009; Richter 1998; Roosevelt 1996; Teeratakulpisarn 2007). There were differences in the specific symptoms addressed (for example, respiratory, feeding), the measurement instrument used (i.e. questionnaires, diaries) and the time points of assessment. No trial reported the use of generic or disease‐specific quality of life instruments.

Other reported outcomes included temperature measurements (Corneli 2007; Plint 2009; Roosevelt 1996), time to resolution or length of illness (Roosevelt 1996; Zhang 2003), and duration of oxygen therapy or fluids (Bentur 2005; Richter 1998; Roosevelt 1996; Teeratakulpisarn 2007; Zhang 2003). Data on the use of bronchodilator co‐interventions were often reported as an outcome.

Adverse events were mentioned in six trials (Corneli 2007; Goebel 2000; Klassen 1997; Kuyucu 2004; Plint 2009; Teeratakulpisarn 2007). Five of these studies assessed specific gastrointestinal, endocrine or infectious complications. There was heterogeneity and incomplete reporting regarding which adverse events were pre‐specified, their definitions and measurement methods. All adverse effects were short‐term and no study assessed long‐term harms.

Excluded studies

Eighty‐four out of 361 excluded papers involved glucocorticoids. Motives for exclusion from this subset mostly included inappropriate population (for example, trials including participants with a history of previous wheezing, or > 24 months old), type of publication and non‐RCT study design (Characteristics of excluded studies).

Risk of bias in included studies

We assessed overall risk of bias as 'low' in three trials, as 'high' in seven and 'unclear' in seven. The glucocorticoid and epinephrine versus placebo comparison included one low risk of bias trial. All other comparisons included mostly high risk of bias trials (Figure 2).

Figure 2

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.*

*For multi‐arm studies (Barlas 1998, Kuyucu 2004 and Plint 2009), we included one overall assessment for all trial comparisons, and two assessments for each separate comparison of glucocorticoids versus placebo (with or without protocolised bronchodilator, or with epinephrine or salbutamol).

We found adequate sequence generation and allocation concealment in 10 and 11 trials, respectively (Figure 3). We considered blinding adequate in 10 out of 17 trials for the review primary outcomes and clinical severity parameters. Incomplete reporting explained most 'unclear' assessments. Incomplete outcome data were adequately addressed in 12 out of 17 studies for the review primary outcomes, and 11 out of 17 for clinical severity outcomes; it was unclear or inadequate when there was imbalanced attrition between groups, mostly in longer follow‐up assessments.

Figure 3

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

We considered nine out of 17 studies free from risk of selective outcome reporting. Assessment of this item was challenging given the large number of outcomes reported, the diversity of measurement time points, and the fact that trial protocols were not available. Using trial registry searches, we identified three trial registers and used that data to complete assessments (Corneli 2007; Plint 2009; Teeratakulpisarn 2007).

Regarding publication bias and small study effects, there was no asymmetry in funnel plots for the primary outcomes in the glucocorticoids versus placebo comparison by visual inspection or statistical testing (Egger test for admissions and length of stay, P = 0.98 and P = 0.77, respectively) (Figure 4; Figure 5).

Figure 4

Funnel plot of comparison: 1 Steroid versus placebo, outcome: 1.1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome.

Figure 5

Funnel plot of comparison: 1 Steroid versus placebo, outcome: 1.2 Length of stay (inpatients) ‐ review primary outcome.

Other types of bias assessed as 'unclear' included baseline imbalances, or active arm contamination with other related co‐interventions (Kuyucu 2004 and Schuh 2002, respectively).

Effects of interventions

Results are summarised by comparison, setting and type of outcome. GRADE assessments for the two main comparisons ‐ glucocorticoid versus placebo and glucocorticoid and bronchodilator versus placebo are shown in Table 2 and Table 3. All meta‐analyses used random‐effects models; fixed‐effect models did not modify the direction and magnitude of results unless mentioned.

Table 2. GRADE assessments: glucocorticoid versus placebo

Population	Outcome	Number of studies	Number of participants	GRADE domains				Strength of evidence	Intervention favoured
Population	Outcome	Number of studies	Number of participants	Risk of bias	Consistency	Directness	Precision	Strength of evidence	Intervention favoured
GLUCOCORTICOID versus PLACEBO
Inpatients	Length of stay	8	633	Medium	Consistent	Direct	Precise	High	No difference
	Clinical score : 3 to 6 hours	1	26	Medium	Unknown	Direct	Imprecise	Low	Glucocorticoid
	Clinical score : 6 to 12 hours	3	175	Medium	Consistent	Direct	Imprecise	Moderate	Glucocorticoid
	Clinical score : 12 to 24 hours	3	230	Medium	Consistent	Direct	Imprecise	Moderate	No difference (glucocorticoid favoured)
	Clinical score : 24 to 72 hours	4	113	Medium	Inconsistent	Direct	Imprecise	Low	No difference (glucocorticoid favoured; very close to significant)
Outpatients	Admissions day 1	8	1762	Medium	Consistent	Direct	Precise	High	No difference
	Admissions up to day 7	5	1530	Low	Consistent	Direct	Imprecise	Moderate	No difference
	Clinical score: 60 minutes	4	1006	Low	Consistent	Direct	Precise	High	No difference
	Clinical score: 120 minutes	3	214	Medium	Consistent	Direct	Imprecise	Moderate	No difference
	Clinical score: 3 to 6 hours	2	808	Medium	Inconsistent	Direct	Precise	Moderate	No difference
	Clinical score: 12 to 24 hours	1	69	Medium	Unknown	Direct	Imprecise	Low	No difference
	Clinical score: 3 to 10 days	4	224	Medium	Inconsistent	Direct	Imprecise	Low	No difference
Inpatients/outpatients	Adverse events	5	1123	Low	Consistent	Direct	Precise	Moderate	No difference

Table 3. GRADE assessments: glucocorticoid and epinephrine versus placebo

Population	Outcome	Number of studies	Number of participants	GRADE domains				Strength of evidence	Intervention favoured
Population	Outcome	Number of studies	Number of participants	Risk of bias	Consistency	Directness	Precision	Strength of evidence	Intervention favoured
GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO
Outpatients	Admissions day 1	1	400	Low	Unknown	Direct	Imprecise	Low	Favours epi + dex but NS
	Admissions day 7	1	400	Low	Unknown	Direct	Imprecise	Low	Epi + dex
	Clinical score: 60 minutes	1	400	Low	Unknown	Direct	Precise	Moderate	Epi + dex
	Adverse events	1	400	Low	Unknown	Direct	Imprecise	Low	No difference

dex = dexamethasone
epi = epinephrine
NS = non‐significant

Glucocorticoid versus placebo

Outpatients

Primary outcomes

All eight outpatient studies reported admissions by day 1, and five also reported admissions by day 7. Complete outcome data were available for 1762 participants by day 1 (out of 1824 randomised) and 1530 participants by day 7 (out of 1612 randomised).

The pooled risk ratios (RRs) for admissions by days 1 and 7 were 0.92 (95% confidence interval (CI) 0.78 to 1.08) and 0.86 (95% CI 0.7 to 1.06), respectively, with no significant differences between groups (Analysis 1.1; Figure 6). Heterogeneity was low for day 1 results and moderate for day 7 (I² statistic = 0% and 31%, respectively). There was no relevant change in the magnitude or direction of results when using pooled data from both Plint 2009 arms. Sensitivity analyses for both trials with low overall risk of bias showed comparable results (Analysis 1.22). Overall strength of evidence for these findings was high for day 1 results and moderate for day 7, the latter due to some imprecision in the effect estimate (Table 2; summary of findings Table for the main comparison).

Figure 6

Forest plot of comparison: 1 Steroid versus placebo, outcome: 1.1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome.

Subgroup analysis of studies using protocolised bronchodilator found lower pooled RRs for admissions by both days 1 and 7, but the CIs between subgroups overlapped (Analysis 1.15; Analysis 1.16). For admissions by day 7, the estimate for RR was 0.68 (95% CI 0.44 to 1.05) for protocolised bronchodilator trials (four trials, 581 participants), and 0.95 (95% CI 0.82 to 1.11) for other trials (two trials, 949 participants). Heterogeneity was low in both subgroups.

The two largest outpatient studies only included participants under 12 months of age, while six smaller studies also included older patients (Analysis 1.17; Analysis 1.18). For admissions by day 7, estimates were 0.92 (95% CI 0.80 to 1.06) and 0.67 (95% CI 0.25 to 1.83), for < 12 months (two trials, 1346 participants) and trials including older participants (three trials, 184 participants), respectively. Trials including older participants had a lower effect estimate, but a large CI overlapped with the other subgroup and there was substantial heterogeneity (I² statistic = 60%).

No subgroup analysis according to respiratory syncytial virus (RSV) or atopic status was performed, since no outpatient trial restricted inclusion based on these parameters. Corneli 2007 and Plint 2009 reported pre‐specified subgroup analyses based on atopic status, with no statistically significant differences. Plint 2009 also reported no differences according to RSV status, duration of illness and severity. We chose not to perform analyses based on glucocorticoid type or dose due to heterogeneity in glucocorticoid schemes.

Secondary outcomes

Clinical score data were available for time points/intervals between 60 minutes and 3 to 10 days (Analysis 1.4; Figure 7). Different sets of studies with different scales contributed to each time point, with most data at 60 minutes (four trials, 1006 participants); no trial assessed the period between 24 to 72 hours. There were no significant differences between groups at any time point. Strength of evidence for these findings was high at 60 minutes, with precise and consistent results (standardised mean difference (SMD) ‐0.04; 95% CI ‐0.16 to 0.09; I² statistic = 0%). Evidence was weaker for later results due to imprecision and substantial heterogeneity.

Figure 7

Forest plot of comparison: 1 Steroid versus placebo, outcome: 1.4 Clinical scale scores (outpatients) (change from baseline data).

Six trials reported outcome data on oxygen saturation between 60 minutes and 24 to 72 hours (Analysis 1.6). Data were most frequently reported at 60 minutes (three trials, 936 participants). At three to six hours, results favoured placebo (mean difference (MD) ‐0.43; 95% CI ‐0.84 to ‐0.02; units: %), while for all other time points there were no significant differences between groups.

Respiratory and heart rate data were both reported in six outpatient trials, between 60 minutes and 3 to 10 days (Analysis 1.8; Analysis 1.10). The most frequently assessed time point for both outcomes was 60 minutes; no trial assessed the period between 24 to 72 hours. There were no significant differences between groups for any of these outcomes.

Regarding other health services outcomes, pooled data from three trials (255 participants) reporting length of stay (LOS) of admitted patients did not show significant differences between groups (Analysis 1.3). Return to healthcare visits for bronchiolitis symptoms were only assessed in two trials (863 participants), both showing considerable event rate for a three to four‐week follow‐up period (26% to 53% in all groups; Table 4). Pooled results did not show significant differences between groups (RR 1.04; 95% CI 0.80 to 1.35) (Analysis 1.14).

Table 4. Hospital re‐admissions and return healthcare visits (in‐ and outpatients)

Study	Population	Duration of follow‐up	Glucocorticoid‐including group	Placebo or comparator group	Notes
GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO: HOSPITAL RE‐ADMISSIONS
Roosevelt 1996	Inpatients	Days 1 to 14	0	0	(No events in either group)
Klassen 1997	Inpatients	Days 1 to 7	4/35 (11%)	1/32 (3%)	P = 0.36
Teeratakulpisarn 2007	Inpatients	Days 1 to 30	3/89 (3%)	7/85 (8%)
GLUCOCORTICOID versus PLACEBO: RETURN HEALTHCARE VISITS*
Plint 2009 (epinephrine ‐ E; dexamethasone ‐ D; placebo ‐ P)	Outpatients	Days 1 to 22	D + E 95/199 (48%)	P + E 93/198 (47%)	Return to the health care provider for bronchiolitis symptoms Difference between dexamethasone + placebo versus placebo + placebo, was significant in the unadjusted analysis (P = 0.04)
	Outpatients	Days 1 to 22	D + P 106/199 (53%)	P + P 86/201 (43%)
Schuh 2002	Outpatients	Days 7 to 28	9/35 (26%)	14/32 (44%)	Medical visits for continuing symptoms; P = 0.069
Klassen 1997	Inpatients	Days 1 to 7	29/35 (83%)	24/32 (75%)	P = 0.77
Roosevelt 1996	Inpatients	Days 1 to 14	16/65 (25%)	5/53 (9%)	P = 0.01; reported on visits made by the physician; 69% were for non‐respiratory difficulties
Teeratakulpisarn 2007	Inpatients	Days 1 to 30	17/89 (19%)	26/85 (31%)	Visit to emergency room or a private clinic because of respiratory symptoms
GLUCOCORTICOID versus EPINEPHRINE: RETURN HEALTHCARE VISITS
Plint 2009 (dexamethasone + placebo versus epinephrine + placebo)	Outpatients	Days 1 to 22	106/199 (53%)	93/198 (47%)	‐
GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO: RETURN HEALTHCARE VISITS
Plint 2009 (dexamethasone + epinephrine versus placebo + placebo)	Outpatients	Days 1 to 22	95/198 (48%)	86/201 (43%)	‐

*Berger 1998: no difference between groups, but did not report quantitative data.
Data presented as n/N (%)

Plint 2009 reported data on parent‐reported symptoms regarding time to return to normal feeding, sleeping, breathing and no coughing (Table 5). There were no statistically significant differences between glucocorticoid and placebo groups. No outpatient trials assessed or reported pulmonary function or quality of life outcomes.

Table 5. Symptoms and quality of life (in‐ and outpatients)*

Study	Population	Duration of follow‐up	Parameter	Glucocorticoid‐including group	Placebo or comparator group	Notes
GLUCOCORTICOID versus PLACEBO
Teeratakulpisarn 2007	Inpatients	Days 1 to 30	Time from treatment to being symptom free ‐ mean ± SD	7.0 ± 5.9	9.0 ± 6.4	P = 0.035
Cade 2000	Inpatients	Days 1 to 28	Time taken for half of infants to become asymptomatic for 48 hours (95% CI) ‐ time to event analysis	10 (10 to 13)	12 (10 to 16)	HR 1.41 (95% CI 0.98 to 2.04), P = 0.07
			Days with coughing or wheezing episodes ‐ mean ± SD	17.0 ± 7.6 days	17.1 ± 8.5	Mean difference: 0.91 days (95% CI ‐2.72 to 2.41), P = 0.91
Roosevelt 1996#	Inpatients	Day 10 to 14	No current difficulty breathing ‐ n/N (%)	45/45 (100)	37/42 (88)	P = 0.07
			Feeding and drinking well ‐ n/N (%)	45/45 (100)	40/42 (95)	P = 0.57
GLUCOCORTICOID versus PLACEBO, GLUCOCORTICOID versus EPINEPHRINE, GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO
Plint 2009 (epinephrine ‐ E; dexamethasone ‐ D; placebo ‐ P)	Outpatients	Days 1 to 22	Time to return to normal feeding ‐ median (IQR)	D + E: 0.6 (0.2 to 1.3) D + P: 0.8 (0.3 to 1.9) P + E: 0.5 (0.2 to 1.2) P + P: 0.9 (0.3 to 2.1)		Time to return to normal feeding ‐ mean symptom duration ratio D + E versus P + P: 0.63 (unadjusted 95% CI 0.5 to 0.8)¶ Time to return to quiet breathing ‐ mean symptom duration ratio D + E versus P + P: 0.83 (unadjusted 95% CI 0.69 to 1.00) No other reported comparison was statistically significant in adjusted analysis
			Time to return to normal sleeping ‐ median (IQR)	D + E: 0.7 (0.2 to 1.7) D + P: 0.8 (0.3 to 1.9) P + E: 0.8 (0.3 to 1.9) P + P: 0.8 (0.3 to 1.8)
			Time to no coughing ‐ median (IQR)	D + E: 12.6 (7.8 to 18.5) D + P: 13.8 (8.5 to 20.2) P + E: 13.2 (8.1 to 19.3) P + P: 13.3 (8.2 to 19.5)
			Time to quiet breathing ‐ median (IQR)	D + E: 3.1 (1.4 to 6.1) D + P: 3.7 (1.6 to 7.1) P + E: 3.6 (1.5 to 6.9) P + P: 3.7 (1.6 to 7.2)

*Units in days unless otherwise stated; no study assessed or reported data from generic or disease‐specific quality of life instruments; Richter 1998 also reported number of symptom‐free days for a 6‐week follow‐up period

#Roosevelt 1996 primary outcome was time to resolution (defined as number of 12 h periods needed to achieve: O₂ saturation > 95% at room air, accessory muscle score = 0, wheeze = 0 or 1, and normal feeding); only association measures were reported: HR 1.3 (95% CI 0.9 to 1.3), P = 0.22

¶time to symptom relief was analysed by means of parametric survival models with Weibull distributions assumed; 95% CI adjusted for multiple analysis in a factorial trial.

CI = confidence interval
HR = hazard ratio
IQR = interquartile range
SD = standard deviation

Inpatients

Primary outcomes

Eight inpatient trials reported data on LOS (633 participants), with no significant mean difference between glucocorticoid and placebo groups (MD ‐0.18 days; 95% CI ‐0.39 to 0.04; I² statistic = 16%) (Analysis 1.2; Figure 8). On a sensitivity analysis using fixed‐effect models and including all studies, the mean difference reached statistical significance favouring glucocorticoids, with a similar magnitude (MD ‐0.14 days; 95% CI ‐0.25 to ‐0.03). We graded the strength of evidence as high given its precision, consistency and 'Risk of bias' assessments for all included trials (Table 2; summary of findings Table for the main comparison).

Figure 8

Forest plot of comparison: 1 Steroid versus placebo, outcome: 1.2 Length of stay (inpatients) ‐ review primary outcome.

Subgroup analyses showed a statistically significant reduction in LOS in trials with protocolised bronchodilator (‐0.12 days; 95% CI ‐0.23 to ‐0.00; four trials, 206 participants), although CIs overlapped between subgroups (Analysis 1.19). Heterogeneity was low in the protocolised group results (I² statistic = 0%) and moderate in the other subgroup (I² statistic = 38%).

In subgroup analyses according to age and RSV status, CIs overlapped between subgroups for both parameters (Analysis 1.20 and Analysis 1.21). Heterogeneity was low in both < 12 months and RSV‐only trial results, and moderate in the other subgroups.

We did not perform subgroup analyses based on atopic status and glucocorticoid type and dose for the reasons mentioned previously.

Secondary outcomes

Clinical score data were only available for intervals between three to six hours and 24 to 72 hours (Analysis 1.5; Figure 9). Glucocorticoids were favoured at earlier time points (three to six hours, one trial, 174 participants: SMD ‐1.03 (95% CI ‐1.87 to ‐0.19); and 6 to 12 hours, three trials, 269 participants: SMD ‐0.62 (95% CI ‐1.00 to ‐0.23). There were no statistically significant differences at later time points. We assessed the overall strength of evidence for these findings as low or moderate, due to imprecision and low or unknown consistency, often with considerable heterogeneity.

Figure 9

Forest plot of comparison: 1 Glucocorticoid versus placebo, outcome: 1.6 Clinical scores (inpatients) (change from baseline data).

Only two trials reported outcomes of oxygen saturation and respiratory rate at time points between 6 to 12 hours and 24 to 72 hours, one of which also reported heart rate at 12 to 24 hours (Analysis 1.7; Analysis 1.9; Analysis 1.11). There were no significant differences between groups for any outcome or time point.

Both hospital re‐admissions and return healthcare visits were reported by three inpatient studies, with distinct durations of follow‐up; no significant differences were found between groups (Table 4; Analysis 1.12; Analysis 1.13).

Three inpatient trials reported data on parent‐reported symptoms (Table 5). Different sets of symptoms were measured at distinct time points, and methods of measurement and analysis varied. In Teeratakulpisarn 2007 time to being symptom free was significantly shorter in the glucocorticoid group, while Cade 2000 used a different analysis and did not shown any statistically significant differences. There were no differences regarding respiratory symptoms and feeding in both Cade 2000 and Roosevelt 1996. No inpatient trials assessed or reported quality of life outcomes.

De Boeck 1997 reported results from pulmonary function tests on day three. No differences were found in minute ventilation, dynamic lung compliance, and inspiratory and expiratory pulmonary resistance, both before and after nebulised bronchodilator.

All patients

Adverse events

Six trials reported adverse events. Five assessed specific glucocorticoid‐related harms including the two largest studies (Table 6). We considered all harms data together regardless of patient setting in order to adequately assess the safety profile of glucocorticoids. Data were available from 600 to 1579 participants for each safety outcome. We did not pool results given the heterogeneity in definitions, methods and timings of assessment. Individual trial analysis did not show significant differences between glucocorticoids and placebo regarding the occurrence of vomiting, gastrointestinal bleeding, hypertension, pneumonia or varicella.

Table 6. Harms ‐ adverse events

Adverse event		Number of participants	Study	Glucocorticoid‐including group, N (%)	Placebo or comparator group, N (%)	Notes
GLUCOCORTICOID versus PLACEBO, GLUCOCORTICOID versus EPINEPHRINE, GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO
Gastrointestinal	Vomiting	1466	Plint 2009*	D + E: 2/199 (1) D + P: 5/199 (2.5)	P + E: 4/198 (2) P + P: 3/201 (1.5)	Observed in the emergency department by research nurse
			Kuyucu 2004*	No events in either group (D + E, D + S, P + E, P + S)		Methods/timings NR
			Corneli 2007	16/305 (5.2)	14/295 (4.7)	Within 20 minutes after administration of the study medication
	Bleeding	1576	Corneli 2007	No events in either group		#
			Teeratakulpisarn 2007	2/90 (2)	1/89 (1)	Occult blood; also assessed diarrhoea separately. Methods/timings NR
			Plint 2009	D + E: 17/199 (8.5) D + P: 12/199 (6)	P + E: 14/198 (7) P+P: 16/201 (8)	Dark stools; reported by families during the 22‐day telephone follow‐up. No patient had more than 1 episode
Endocrine	Hypertension	1397	Plint 2009	D + E: 0/199 (0) D + P: 1/199 (0.5)	P + E: 1/198 (0.5) P + P: 0/201 (0)	Observed in infants admitted to hospital
			Corneli 2007	No events in either group		#
Infectious	Pneumonia	851	Corneli 2007	1/305 (3.3)	2/295 (7)	Also assessed empyema separately
			Teeratakulpisarn 2007	0/90 (0)	3/89 (3.4)	Methods/timings NR
			Klassen 1997	1/35 (3)	1/37 (3)	Methods/timings NR
	Varicella	1397	Corneli 2007	No events in either group		#
			Plint 2009	No events in either group		Reported by families during the 22‐day telephone follow‐up
General	Tremor	866	Kuyucu 2004	No events in either group		Methods/timings NR
			Plint 2009	D + E: 4/199 (2) D + P: 5/199 (2.5)	P + E: 4/198 (2) P + P: 2/201 (1)	Observed in the emergency department by research nurse
	Pallor/flushing	866	Kuyucu 2004	No events in either group		Methods/timings NR
			Plint 2009	D + E: 23/199 (11.5) D + P: 15/199 (7.5)	P + E: 22/198 (11.1) P + P: 16/201 (8)	Observed in the emergency department by research nurse

Additional reported adverse events: Goebel 2000 reported toxicity data: one patient was "jittery"; no evidence of further treatment complications. Plint 2009 also reported hyperkalaemia observed in infants admitted to hospital (only one case was noted in the dexamethasone group).

*epinephrine ‐ E; dexamethasone ‐ D; salbutamol ‐ S; placebo ‐ P

#Corneli 2007: Study clinicians and research assistants monitored the infants for adverse events during observation in the emergency department. Subsequent adverse events were determined at follow‐up. A patient safety committee, made up of people not involved with patient enrolment, tracked all adverse events.

Glucocorticoid and bronchodilator (epinephrine or salbutamol) versus placebo

Both outpatient trials assessing either of these comparisons used different severity thresholds for patient inclusion: Respiratory Distress Assessment Instrument (RDAI) score above four in Plint 2009 (moderate disease), and scores between 4 and 10 using a trial‐specific clinical scale in Barlas 1998 (mild to moderate disease).

Primary outcomes

The factorial trial Plint 2009 included a comparison of oral dexamethasone and nebulised epinephrine against double placebo (399 analysed participants). This was the largest trial included in the review, with low overall risk of bias. The RRs for admissions by days 1 and 7 were 0.65 (95% CI 0.40 to 1.05) and 0.65 (95% CI 0.44 to 0.95), respectively (Analysis 2.1). There was a statistically significant reduction in admissions by day 7, with a relative risk reduction estimate of 35%. Absolute risk reduction was 9% (95% CI 1 to 17), and the number needed to treat to benefit (NNTB) to reduce one admission by day 7 was 11 (95% CI 7 to 76); these results were obtained through unadjusted analysis. However, the factorial trial design requires special methodological considerations, since this was not the study's main comparison, and there was an unanticipated additive/synergistic effect between epinephrine and dexamethasone. Reported analyses adjusted for multiple comparisons were above the threshold for statistical significance (RR 0.65; 95% CI 0.41 to 1.03). We graded the overall strength of evidence as low for these results given their imprecision and the fact that they were obtained from a single trial (Table 3; summary of findings Table 2).

Barlas 1998, a small high risk of bias trial, compared intravenous prednisolone and nebulised salbutamol versus placebo. Admissions by day 1 (30 participants) showed no statistically significant differences between groups (RR 0.67; 95% CI 0.13 to 3.44) (Analysis 3.1).

Secondary outcomes

Clinical score results at 60 minutes favoured glucocorticoid and epinephrine (SMD ‐0.34; 95% CI ‐0.54 to ‐0.14) (Analysis 2.2), while having an increased heart rate (MD 8.44; 95% CI 4.85 to 12.03) (Analysis 2.5). No differences were found between groups regarding oxygen saturation and respiratory rate (Analysis 2.3; Analysis 2.4). There were also no differences regarding return healthcare visits for bronchiolitis symptoms (RR 1.11; 95% CI 0.89 to 1.38) (Table 4; Analysis 2.6). Symptom results showed reduced time to normal feeding and quiet breathing in the glucocorticoid and epinephrine group (mean symptom duration ratios: 0.63; 95% CI 0.5 to 0.8 and 0.83; 95% CI 0.69 to 1.00) (Table 5). No differences were found in time to normal sleeping and time to no coughing.

Results for clinical scores, oxygen saturation and heart rate at 60 minutes, 120 minutes and three to six hours did not show any differences between groups in the single trial comparing glucocorticoid and salbutamol versus placebo (Analysis 3.2; Analysis 3.3; Analysis 3.4). No further secondary outcomes were assessed in this comparison.

Other comparisons

These included glucocorticoid versus bronchodilator (epinephrine or salbutamol), glucocorticoid and bronchodilator (epinephrine or salbutamol) versus different bronchodilator (epinephrine or salbutamol), and direct comparisons between different types of glucocorticoid (prednisolone versus budesonide). All trials were performed in the outpatient setting, and all except one were small‐sized and had a high risk of bias.

Primary outcomes

The glucocorticoid versus epinephrine comparison included data from two trials (444 participants) for admissions by day 1, and one trial by day 7 (399 participants). Risk of bias was low for one trial and high for the other. There were no significant differences between groups at both time points (Analysis 4.1). Only one small high risk of bias trial included data on day 1 admissions for both glucocorticoid versus salbutamol (45 participants) and glucocorticoid and salbutamol versus epinephrine comparisons (30 participants), with no differences between arms (Analysis 5.1; Analysis 7.1). There were no admissions in another trial including the latter comparison, as well as glucocorticoid and epinephrine versus salbutamol (Analysis 6.1; Analysis 7.1).

Barlas 1998 multi‐arm trial also performed an unblinded comparison between systemic prednisolone and inhaled budesonide, with no statistically significant differences in admissions by day 1 (Analysis 8.1).

Secondary outcomes

When compared to glucocorticoid at 60 minutes, epinephrine use was associated with lower clinical scores (SMD 0.31; 95% CI 0.12 to 0.50) and higher oxygen saturation (MD ‐0.99; 95% CI ‐1.46 to ‐0.52; units: %) (two trials, 442 participants), while heart rate was lower with glucocorticoids (MD ‐7.56 bpm; 95% CI ‐11.34 to ‐3.79) and there were no differences in respiratory rate (Analysis 4.2; Analysis 4.3; Analysis 4.4; Analysis 4.5). There were no differences in the single trial assessing clinical scores and heart rate at later time points.

Salbutamol was also favoured when compared to glucocorticoids, in clinical scores at 60 minutes and three to six hours (SMD 0.65; 95% CI 0.01 to 1.28; and SMD 0.70; 95% CI 0.06 to 1.34, respectively), but not at 120 minutes (Analysis 5.2). Heart rate at 120 minutes was lower in glucocorticoid group (MD ‐7.53 bpm; 95% CI ‐14.28 to ‐0.78) and there were no differences in oxygen saturation at any time point (Analysis 5.3; Analysis 5.4).

At 3 to 10 days, clinical scores and respiratory rate results favoured glucocorticoids and epinephrine as compared to salbutamol (SMD ‐1.22; 95% CI ‐1.98 to ‐0.46, and MD ‐13.70; 95% CI ‐20.56 to ‐6.84, respectively) (Analysis 6.2; Analysis 6.3; Analysis 6.4). There were no other differences at earlier time points and regarding heart rate.

Oxygen saturation at 60 and 120 minutes was higher in the epinephrine group when compared to glucocorticoid and salbutamol (MD ‐1.54; 95% CI ‐2.85 to ‐0.23, and MD ‐1.27; 95% CI ‐2.41 to ‐0.13, respectively) (Analysis 7.2; Analysis 7.3; Analysis 7.4; Analysis 7.5). No other statistically significant differences were found in clinical scores, oxygen saturation or respiratory or heart rate at other time points.

When comparing systemic prednisolone and inhaled budesonide, oxygen saturation results favoured budesonide at 60 minutes and 120 minutes (MD ‐1.46; 95% CI ‐2.74 to ‐0.18, and MD ‐1.73; 95% CI ‐3.06 to ‐0.40, respectively), and heart rate was lower with prednisolone at three to six hours (Analysis 8.3; Analysis 8.4). No differences were found in all other outcomes and time points (Analysis 8.2).

Plint 2009 reported safety assessments comparing glucocorticoid and epinephrine (Table 6). Pallor was observed in 7.5% of participants in the glucocorticoid group, compared to 11.1% in the epinephrine group. There were no significant differences in vomiting, bleeding, hypertension, varicella and tremor between glucocorticoids and epinephrine. No other trial from any of the other comparisons reported adverse events data.

Discussion

Summary of main results

Results from this review do not suggest a clinically relevant stand‐alone effect of systemic or inhaled glucocorticoids in either outpatient and inpatient settings (summary of findings Table for the main comparison). There were no statistically significant differences in outpatient admissions by days 1 and 7, and pooled RR estimates favouring glucocorticoids versus placebo were below commonly used thresholds for clinical relevance. Strength of evidence was moderate to high, indicating our confidence in these effect estimates. There were also no differences in secondary outcomes, particularly clinical scores, oxygen saturation and respiratory symptoms. For inpatient trials, precise and consistent results did not show differences in length of stay (LOS) as compared to placebo. The lower boundary of the pooled estimate confidence interval (CI) was about nine hours, likely excluding a clinically relevant benefit from glucocorticoids. While clinical score results were superior during the first day of treatment, no consistent differences were found at later time points or in any other secondary outcomes. Subgroup analyses according to age and respiratory syncytial virus (RSV) status did not suggest effect modification by these factors; heterogeneity did not allow adequate analysis of atopy and glucocorticoid type or dose.

Exploratory evidence suggests that combined glucocorticoids and bronchodilators may have clinically relevant benefits. A large factorial trial with low risk of bias found that high‐dose dexamethasone with epinephrine reduced admissions by day 7 when compared to placebo, in outpatients with moderately severe bronchiolitis (summary of findings Table 2). The unadjusted risk ratio (RR) reduction estimate was 36%, and 11 children with bronchiolitis had to be treated to reduce one admission given the study's baseline risk. Clinical scores and symptoms results supported this benefit. However, these are the findings of a single study and should be interpreted cautiously. There were methodological issues with trial design and results may have arisen by chance. Further evidence regarding combined therapy is scarce and imprecise, and exploratory subgroup analysis was not conclusive as to an additive/synergistic effect of glucocorticoids combined with bronchodilators.

No relevant differences were found in short‐term general and intervention‐specific adverse effects for these comparisons. However, balancing harms and benefits of glucocorticoids alone or combined was hampered by the lack of long‐term safety data.

Overall completeness and applicability of evidence

The heterogeneous definition of bronchiolitis is often a motive for controversy when interpreting trial and review results (DiTraglia 2004; Weinberger 2003; Weinberger 2007). There is no international consensus due to variation in semantics and clinical findings (for example, in the UK, 'crackles' are often key to diagnosis, as opposed to 'wheeze' in North America) (Everard 2009). A first episode of wheezing may be a manifestation of wheezing phenotypes with heterogeneous biological, genetic, viral or environmental determinants, and distinct prognosis (Brand 2008; Martinez 2005; Sly 2008). However, research is still ongoing to identify simple, valid and universal discriminative and prognostic tools to prospectively distinguish between them (Brand 2008; Schultz 2010; Sly 2008). We used a pragmatic definition and focused on first time wheezing so results could be directly pertinent to infants with 'typical' viral bronchiolitis, as opposed to those with acute recurrent wheezing or asthma.

We found variability in both bronchiolitis severity and glucocorticoids schemes, but this did not affect the consistency of results. Baseline disease in outpatients was often moderate, but the use of different clinical criteria and scales limited the comparison between trials, particularly for inpatients. The wide range of control group admission rates and LOS can be partially explained by differing disease severity, but it also reflects variation in bronchiolitis management, for example, different admission/discharge criteria and standards of care (Babl 2008; Barben 2003; Brand 2000; Christakis 2005; Gonzalez 2010; Mallory 2003; Mansbach 2005). Our findings were consistent in trials performed worldwide, and results likely apply to settings with different resources and management strategies.

Most studies were restricted to healthy infants, often excluding children with chronic conditions and prematurity. Lack of evidence for this subset of patients is problematic, since many are particularly at risk of adverse outcomes (Damore 2008; Figueras‐Aloy 2008; Meissner 2003). Epidemiological studies have highlighted the short‐ and long‐term impact of RSV disease in prematurity (Figueras‐Aloy 2008; Simoes 2008), and underlying changes in respiratory pathophysiology may limit the external validity of our results in these populations.

Results from subgroup analyses did not identify any subset of participants with a different response to glucocorticoids. Older aged and atopic children are at higher risk of recurrent wheezing and asthma (Castro‐Rodriguez 2000), and both factors have been traditionally proposed as markers of underlying glucocorticoid‐responsive phenotypes in first‐time wheezers (Weinberger 2007). We found no conclusive evidence of such effect with age. We were unable to study atopy, but subgroup analyses from individual studies did not identify any significant differences. Specific viruses may also modulate response, as RSV and rhinovirus infections are associated with recurrent wheezing and the latter is a stronger predictor and possibly more responsive to glucocorticoids (Jackson 2008; Korppi 2007; Lehtinen 2007; Stein 1999). We found no differences according to RSV status, while other viral aetiologies were not reported. Accumulating evidence shows that glucocorticoids have reduced effectiveness in later acute recurrent wheezing (Bush 2009; Panickar 2009). Further, each of these factors per se has limited prognostic accuracy in defining stable wheezing phenotypes (Brand 2008; Simpson 2010; Sly 2008). Our results suggest that 'typical' viral bronchiolitis is not glucocorticoid‐responsive. Potential methodological limitations include the use of aggregated data and heterogeneity in definition, ascertainment and reporting of subgroups.

We found promising exploratory results from one large trial using combined dexamethasone with epinephrine for moderately ill outpatients. Although reliance on findings from single precise well‐conducted trials is often reasonable (Glasziou 2010), in this factorial trial the additive interaction between treatments was unanticipated, and this limits the interpretation of its results (McAlister 2003; Montgomery 2003). Our observational and exploratory subgroup analyses of protocolised bronchodilators may indirectly support an additive effect, but findings were not conclusive for both outpatients and inpatients. The latter are often a separate population due to differences in severity, duration of symptoms or non‐response to initial bronchodilators, and these may affect response to therapy. Replication is therefore needed to improve our confidence in the direction, precision and magnitude of the effect estimates for outpatients, and its applicability for inpatients.

Whether results from combination therapy can be generalisable to different glucocorticoid or bronchodilator schemes is also not known. Systemic dexamethasone is favoured in another common viral respiratory disorder, croup (Bjornson 2008). Its long half‐life and stronger potency may account for its effect, but underlying pathological changes are distinct between these two conditions. Plint 2009 used multiple high doses of dexamethasone. A previous dose‐finding trial suggested similar results with a single high dose, although there was no placebo comparator (Schuh 2008); the lowest efficacious dose remains unknown. The choice of bronchodilator is also undecided. A recently updated Cochrane review on epinephrine in bronchiolitis showed a reduction in first day outpatient admissions, as well as other short‐term severity outcomes (Hartling 2011a). This might explain part of the early benefit of combined therapy seen in Plint 2009. Further research is needed to clarify whether combined epinephrine is superior to combined salbutamol, particularly given the variation in bronchodilator choice in practice.

Evidence from basic and translational research may support a synergistic effect of combined therapy, but it is not clear how this reconciles with the limited effect of glucocorticoids alone. Inflammation pathways and mediators involved in bronchiolitis seem to be distinct from those in glucocorticoid‐sensitive asthma. Innate immunity, specific cytokine dysregulation patterns and neutrophilic inflammation may be relevant for some early wheezing phenotypes (Bont 2009; Halfhide 2008), which could explain the limited biological action of glucocorticoids alone (Buckingham 2002; Lehtinen 2007; Somers 2009). Paradoxically, clinical and biological synergism between glucocorticoids and bronchodilators has been a major topic in asthma treatment (Giembycz 2008). Two‐way molecular interactions exist, including beta₂‐agonist‐stimulated glucocorticoid‐mediated gene transcription (Kaur 2008) and glucocorticoid‐induced increase in the transcription of the ß₂‐receptor gene (Black 2009). Epinephrine’s α‐adrenergic vasoconstricting and oedema‐reducing activity could confer an additional short‐term benefit. Whether these mechanisms are involved in acute bronchiolitis therapy, and the role of specific types and doses of bronchodilators and glucocorticoids, is unknown.

These positive results should be balanced against incomplete data on harms. Safety concerns are expected when considering the widespread use of epinephrine and glucocorticoids in young children with viral wheezing, particularly with repeated high glucocorticoid doses (Bush 2009; Frey 2009). Current data from RCTs and observational studies in croup suggest a favourable short‐term safety profile from both dexamethasone and epinephrine (Bjornson 2008;Zhang 2005). Considering all trials, our results do not suggest any serious or frequent short‐term expected or unexpected harms from glucocorticoids in the absence of co‐morbidities. However, the power to detect important differences was limited due to the infrequent occurrence of events, and adverse event detection was heterogeneous. Glucocorticoids also raises long‐term safety issues. Their use in prematurity for neonatal respiratory distress has been associated with effects on adrenal function, cardiovascular responses, somatic and lung growth, and neurodevelopment (Doyle 2010; Karemaker 2008a; Karemaker 2008b; Onland 2008; Wilson‐Costello 2009). Evidence is scarce, however, regarding effects of short‐term use in otherwise healthy term infants, and none of these were studied in included trials. Further pharmacoepidemiologic data are needed to permit adequate short and long‐term risk‐benefit assessments.

Quality of the evidence

Two key factors affected the strength of evidence: potential risk of bias in the included studies, and sparsity of data for many of the outcomes and comparisons, with imprecise estimates and unknown consistency across studies.

A majority of trials had unclear risk of bias, usually due to incomplete or inadequate reporting, and many comparisons only included small trials at high risk of bias. Inadequate allocation concealment and blinding were likely to be relevant given the nature of interventions (for example, inhaled versus systemic administration) and outcome assessments (for example, physician‐based admissions or discharge decisions). Incomplete outcome data were often found, with losses of follow‐up in outpatient trials. However, for the main glucocorticoid versus placebo comparison, sensitivity analyses restricted to low risk of bias trials did not change the direction or magnitude of results for primary outcomes, highlighting their consistency.

Sparsity of data was a result of a large number of comparisons as well as variability in the choice of outcomes and timing of assessments. Within trials, this also led to frequent uncertainties regarding selective outcome reporting. The message around consistency and relevance of outcomes is not new to this field (Flores 1997; King 2004; Klassen 1996). The absence of standardised, validated and patient‐important outcome measures has been a serious threat to bronchiolitis trial validity. Our primary outcomes focused on hospital use, which has clear implications for patients, families and health services. However, there is no guidance supporting the choice of methodologically sound and patient‐important outcomes. Lack of reporting of admission and discharge criteria is also problematic given the wide variation in bronchiolitis management. Additionally, the choice of clinical scales was inconsistent. The Respiratory Distress Assessment Instrument (RDAI) was used in a considerable number of trials, but its clinimetric properties ‐ for example, responsiveness and interpretability ‐ are not well known, which limits the interpretation of findings. This was compounded by the absence of quality of life measures. Further work is needed to define a core set of clinically important efficacy and safety outcome measures and timing of assessments, for trials and systematic reviews in this field.

Potential biases in the review process

Some limitations have already been described, others should also be highlighted. We did not obtain further data from authors of included studies, which might have clarified 'Risk of bias' assessments and further added to reported trial characteristics and secondary outcome results. There is scarce guidance on how to investigate synergism/antagonism at a systematic review level, therefore our approach should be considered exploratory, including our use of factorial trial results. However, we performed sensitivity analyses of different analysis methods and these did not show a change in the direction of results. Our choice of outcome time intervals may have been a source of heterogeneity, although it was limited by the sparsity of reported data. Limitations of subgroup analyses are well known and have been addressed. Grading of evidence was limited by the lack of guidance regarding clinically relevant differences in studied outcomes.

Agreements and disagreements with other studies or reviews

Two previous non‐Cochrane systematic reviews assessed the use of glucocorticoids in acute bronchiolitis, one of which also performed meta‐analysis (Garrison 2000; King 2004). None of the reviews included data from the two recent large glucocorticoid outpatient trials. There was some discordance in inclusion criteria regarding population and interventions: Garrison 2000 only included inpatient trials and was restricted to systemic glucocorticoids, and no review excluded previous wheezing. The choice of primary outcomes and their definitions, timings and analysis also differed. While Garrison 2000 highlighted a statistically significant reduction in LOS for inpatients, this analysis used a modified outcome definition. When comparing similar analyses for this outcome, quantitative results were comparable between all reviews, including ours, and suggest no relevant benefit from glucocorticoids in inpatients. Outpatient descriptive and quantitative results from King 2004 also found no difference in admissions. No previous review assessed the hypothesis of synergism between glucocorticoids and bronchodilators at an analysis level, while subgroup analyses assessing possible dose‐response and effect modifiers like age and RSV status showed similar negative results.

Figure 1

Flow of citations through the search and screening procedures of the 2009 review and this 2012 update, studies included in the review and comparisons addressed (GC: glucocorticoids)

Figure 2

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.*

Figure 3

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

Figure 4

Funnel plot of comparison: 1 Steroid versus placebo, outcome: 1.1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome.

Figure 5

Funnel plot of comparison: 1 Steroid versus placebo, outcome: 1.2 Length of stay (inpatients) ‐ review primary outcome.

Figure 6

Forest plot of comparison: 1 Steroid versus placebo, outcome: 1.1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome.

Figure 7

Forest plot of comparison: 1 Steroid versus placebo, outcome: 1.4 Clinical scale scores (outpatients) (change from baseline data).

Figure 8

Forest plot of comparison: 1 Steroid versus placebo, outcome: 1.2 Length of stay (inpatients) ‐ review primary outcome.

Figure 9

Forest plot of comparison: 1 Glucocorticoid versus placebo, outcome: 1.6 Clinical scores (inpatients) (change from baseline data).

Analysis 1.1

Comparison 1 Glucocorticoid versus placebo, Outcome 1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome.

Analysis 1.2

Comparison 1 Glucocorticoid versus placebo, Outcome 2 Length of stay (inpatients) ‐ review primary outcome.

Analysis 1.3

Comparison 1 Glucocorticoid versus placebo, Outcome 3 Length of stay (outpatients).

Analysis 1.4

Comparison 1 Glucocorticoid versus placebo, Outcome 4 Clinical scores (outpatients).

Analysis 1.5

Comparison 1 Glucocorticoid versus placebo, Outcome 5 Clinical scores (inpatients).

Comparison 1 Glucocorticoid versus placebo, Outcome 6 O2 saturation (outpatients).

Analysis 1.6

Comparison 1 Glucocorticoid versus placebo, Outcome 6 O₂ saturation (outpatients).

Comparison 1 Glucocorticoid versus placebo, Outcome 7 O2 saturation (inpatients).

Analysis 1.7

Comparison 1 Glucocorticoid versus placebo, Outcome 7 O₂ saturation (inpatients).

Analysis 1.8

Comparison 1 Glucocorticoid versus placebo, Outcome 8 Respiratory rate (outpatients).

Analysis 1.9

Comparison 1 Glucocorticoid versus placebo, Outcome 9 Respiratory rate (inpatients).

Analysis 1.10

Comparison 1 Glucocorticoid versus placebo, Outcome 10 Heart rate (outpatients).

Analysis 1.11

Comparison 1 Glucocorticoid versus placebo, Outcome 11 Heart rate (inpatients).

Analysis 1.12

Comparison 1 Glucocorticoid versus placebo, Outcome 12 Hospital readmissions (inpatients).

Analysis 1.13

Comparison 1 Glucocorticoid versus placebo, Outcome 13 Return healthcare visits (inpatients).

Analysis 1.14

Comparison 1 Glucocorticoid versus placebo, Outcome 14 Return healthcare visits (outpatients).

Analysis 1.15

Comparison 1 Glucocorticoid versus placebo, Outcome 15 Admissions at day 1 (outpatients) ‐ subgroup analysis protocolised use of bronchodilator.

Analysis 1.16

Comparison 1 Glucocorticoid versus placebo, Outcome 16 Admissions within 7 days (outpatients) ‐ subgroup analysis protocolised use of bronchodilator.

Analysis 1.17

Comparison 1 Glucocorticoid versus placebo, Outcome 17 Admissions at day 1 (outpatients) ‐ subgroup analysis age.

Analysis 1.18

Comparison 1 Glucocorticoid versus placebo, Outcome 18 Admissions within 7 days (outpatients) ‐ subgroup analysis age.

Analysis 1.19

Comparison 1 Glucocorticoid versus placebo, Outcome 19 Length of stay (inpatients) ‐ subgroup analysis protocolised use of bronchodilator.

Analysis 1.20

Comparison 1 Glucocorticoid versus placebo, Outcome 20 Length of stay (inpatients) ‐ subgroup analysis age.

Analysis 1.21

Comparison 1 Glucocorticoid versus placebo, Outcome 21 Length of stay (inpatients) ‐ subgroup analysis RSV status.

Analysis 1.22

Comparison 1 Glucocorticoid versus placebo, Outcome 22 Admissions (days 1 and 7) (outpatients) ‐ sensitivity analysis with only low overall RoB.

Analysis 1.23

Comparison 1 Glucocorticoid versus placebo, Outcome 23 Length of stay (inpatients) ‐ sensitivity analysis with only low overall RoB.

Analysis 2.1

Comparison 2 Glucocorticoid and epinephrine versus placebo, Outcome 1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome.

Analysis 2.2

Comparison 2 Glucocorticoid and epinephrine versus placebo, Outcome 2 Clinical scores (outpatients).

Comparison 2 Glucocorticoid and epinephrine versus placebo, Outcome 3 O2 saturation (outpatients).

Analysis 2.3

Comparison 2 Glucocorticoid and epinephrine versus placebo, Outcome 3 O₂ saturation (outpatients).

Analysis 2.4

Comparison 2 Glucocorticoid and epinephrine versus placebo, Outcome 4 Respiratory rate (outpatients).

Analysis 2.5

Comparison 2 Glucocorticoid and epinephrine versus placebo, Outcome 5 Heart rate (outpatients).

Analysis 2.6

Comparison 2 Glucocorticoid and epinephrine versus placebo, Outcome 6 Return healthcare visits (outpatients).

Analysis 3.1

Comparison 3 Glucocorticoid and salbutamol versus placebo, Outcome 1 Admissions (day 1) (outpatients) ‐ review primary outcome.

Analysis 3.2

Comparison 3 Glucocorticoid and salbutamol versus placebo, Outcome 2 Clinical scores (outpatients).

Comparison 3 Glucocorticoid and salbutamol versus placebo, Outcome 3 O2 saturation (outpatients).

Analysis 3.3

Comparison 3 Glucocorticoid and salbutamol versus placebo, Outcome 3 O₂ saturation (outpatients).

Analysis 3.4

Comparison 3 Glucocorticoid and salbutamol versus placebo, Outcome 4 Heart rate (outpatients).

Analysis 4.1

Comparison 4 Glucocorticoid versus epinephrine, Outcome 1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome.

Analysis 4.2

Comparison 4 Glucocorticoid versus epinephrine, Outcome 2 Clinical scores (outpatients).

Comparison 4 Glucocorticoid versus epinephrine, Outcome 3 O2 saturation (outpatients).

Analysis 4.3

Comparison 4 Glucocorticoid versus epinephrine, Outcome 3 O₂ saturation (outpatients).

Analysis 4.4

Comparison 4 Glucocorticoid versus epinephrine, Outcome 4 Respiratory rate (outpatients).

Analysis 4.5

Comparison 4 Glucocorticoid versus epinephrine, Outcome 5 Heart rate (outpatients).

Analysis 5.1

Comparison 5 Glucocorticoid versus salbutamol, Outcome 1 Admissions (day 1) (outpatients) ‐ review primary outcome.

Analysis 5.2

Comparison 5 Glucocorticoid versus salbutamol, Outcome 2 Clinical scores (outpatients).

Comparison 5 Glucocorticoid versus salbutamol, Outcome 3 O2 saturation (outpatients).

Analysis 5.3

Comparison 5 Glucocorticoid versus salbutamol, Outcome 3 O₂ saturation (outpatients).

Analysis 5.4

Comparison 5 Glucocorticoid versus salbutamol, Outcome 4 Heart rate (outpatients).

Analysis 6.1

Comparison 6 Glucocorticoid and epinephrine versus salbutamol, Outcome 1 Admissions (day 1) (outpatients) ‐ review primary outcome.

Analysis 6.2

Comparison 6 Glucocorticoid and epinephrine versus salbutamol, Outcome 2 Clinical scores (outpatients).

Analysis 6.3

Comparison 6 Glucocorticoid and epinephrine versus salbutamol, Outcome 3 Respiratory rate (outpatients).

Analysis 6.4

Comparison 6 Glucocorticoid and epinephrine versus salbutamol, Outcome 4 Heart rate (outpatients).

Analysis 7.1

Comparison 7 Glucocorticoid and salbutamol versus epinephrine, Outcome 1 Admissions (day 1) (outpatients) ‐ review primary outcome.

Analysis 7.2

Comparison 7 Glucocorticoid and salbutamol versus epinephrine, Outcome 2 Clinical scores (outpatients).

Comparison 7 Glucocorticoid and salbutamol versus epinephrine, Outcome 3 O2 saturation (outpatients).

Analysis 7.3

Comparison 7 Glucocorticoid and salbutamol versus epinephrine, Outcome 3 O₂ saturation (outpatients).

Analysis 7.4

Comparison 7 Glucocorticoid and salbutamol versus epinephrine, Outcome 4 Respiratory rate (outpatients).

Analysis 7.5

Comparison 7 Glucocorticoid and salbutamol versus epinephrine, Outcome 5 Heart rate (outpatients).

Analysis 8.1

Comparison 8 Glucocorticoid versus glucocorticoid (prednisolone versus budesonide), Outcome 1 Admissions (day 1) (outpatients) ‐ review primary outcome.

Analysis 8.2

Comparison 8 Glucocorticoid versus glucocorticoid (prednisolone versus budesonide), Outcome 2 Clinical scores (outpatients).

Comparison 8 Glucocorticoid versus glucocorticoid (prednisolone versus budesonide), Outcome 3 O2 saturation (outpatients).

Analysis 8.3

Comparison 8 Glucocorticoid versus glucocorticoid (prednisolone versus budesonide), Outcome 3 O₂ saturation (outpatients).

Analysis 8.4

Comparison 8 Glucocorticoid versus glucocorticoid (prednisolone versus budesonide), Outcome 4 Heart rate (outpatients).

Summary of findings for the main comparison. Glucocorticoid versus placebo: summary of findings

Glucocorticoid versus placebo for acute viral bronchiolitis in infants and young children
Patient or population: infants and young children with acute viral bronchiolitis Settings: outpatients and inpatients Intervention: glucocorticoid versus placebo
Outcomes	*Illustrative comparative risks (95% CI)**		Steroid versus placebo	No of participants (studies)	Quality of the evidence (GRADE)
	Assumed risk¹	Corresponding risk
	Placebo	Steroid
Admissions (outpatients) Follow‐up: day 1	Medium risk population		RR 0.92 (0.78 to 1.08)	1762 (8)	high
	162 per 1000	149 per 1000 (126 to 175)
Admissions (outpatients) Follow‐up: day 7	Medium risk population		RR 0.86 (0.7 to 1.06)	1530 (5)	moderate
	250 per 1000	215 per 1000 (175 to 265)
Length of stay (inpatients) days	The mean length of stay ranged across control groups from 0.8 to 6.6 days	The mean length of stay in the intervention groups was 0.18 lower (0.39 lower to 0.04 higher)		633 (8)	high
The basis for the assumed risk* (for example, the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval 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.
¹Assumed risk for admissions was based on the median control group risks across the studies included in the meta‐analysis (medium risk).

Summary of findings for the main comparison. Glucocorticoid versus placebo: summary of findings

Summary of findings 2. Glucocorticoid and epinephrine versus placebo: summary of findings

Glucocorticoid and epinephrine versus placebo for acute viral bronchiolitis in infants and young children
Patient or population: infants and young children with acute viral bronchiolitis Settings: outpatients Intervention: glucocorticoid and epinephrine versus placebo
Outcomes	*Illustrative comparative risks (95% CI)**		Steroid versus placebo	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk¹	Corresponding risk
	Placebo	Steroid
Admissions (outpatients) Follow‐up: day 1	179 per 1000	115 per 1000 (72 to 186)	RR 0.65 (0.4 to 1.05)	400 (1)	Low	NNT: not calculated for non‐significant findings
Admissions (outpatients) Follow‐up: day 7	264 per 1000	169 per 1000 (116 to 251)	RR 0.65 (0.44 to 0.95)	400 (1)	Low	NNT: 11 (95% CI 7 to 76) (based on unadjusted analysis results)
The basis for the assumed risk* (for example, the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) CI: confidence interval NNT: number needed to treat 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.
¹Assumed risk for admissions was based on the control group risk in the single study included (Plint 2009).

Summary of findings 2. Glucocorticoid and epinephrine versus placebo: summary of findings

Table 1. Placebo group risk of admission/length of stay

Study	Placebo group ‐ participants	Placebo group ‐ primary outcomes
OUTPATIENT STUDIES		Risk of admission day 1 (%)	Risk of admission day 7 (%)
Barlas 1998	30	17%	NR
Berger 1998	18	11%	NR
Corneli 2007	295	41%	49%
Goebel 2000	24	8%	21%
Kuyucu 2004	11	0%	0%
Mesquita 2009	32	22%	NR
Plint 2009	201	18%	26%
Schuh 2002	34	44%	47%
INPATIENT STUDIES		Length of stay (mean ± SD days)
Bentur 2005	32	6.3 ± 8.8
Cade 2000	79	2 ± 2.2
De Boeck 1997	15	6.6 ± 1.2
Gomez 2007	25	0.8 ± 0.2
Klassen 1997	32	2 ± 0.7
Richter 1998	19	3 ± 1.6
Teeratakulpisarn 2007	85	2.8 ± 1.7
Zhang 2003	24	5 ± 3.3
NR = not reported SD = standard deviation

Table 1. Placebo group risk of admission/length of stay

Table 2. GRADE assessments: glucocorticoid versus placebo

Population	Outcome	Number of studies	Number of participants	GRADE domains				Strength of evidence	Intervention favoured
Population	Outcome	Number of studies	Number of participants	Risk of bias	Consistency	Directness	Precision	Strength of evidence	Intervention favoured
GLUCOCORTICOID versus PLACEBO
Inpatients	Length of stay	8	633	Medium	Consistent	Direct	Precise	High	No difference
	Clinical score : 3 to 6 hours	1	26	Medium	Unknown	Direct	Imprecise	Low	Glucocorticoid
	Clinical score : 6 to 12 hours	3	175	Medium	Consistent	Direct	Imprecise	Moderate	Glucocorticoid
	Clinical score : 12 to 24 hours	3	230	Medium	Consistent	Direct	Imprecise	Moderate	No difference (glucocorticoid favoured)
	Clinical score : 24 to 72 hours	4	113	Medium	Inconsistent	Direct	Imprecise	Low	No difference (glucocorticoid favoured; very close to significant)
Outpatients	Admissions day 1	8	1762	Medium	Consistent	Direct	Precise	High	No difference
	Admissions up to day 7	5	1530	Low	Consistent	Direct	Imprecise	Moderate	No difference
	Clinical score: 60 minutes	4	1006	Low	Consistent	Direct	Precise	High	No difference
	Clinical score: 120 minutes	3	214	Medium	Consistent	Direct	Imprecise	Moderate	No difference
	Clinical score: 3 to 6 hours	2	808	Medium	Inconsistent	Direct	Precise	Moderate	No difference
	Clinical score: 12 to 24 hours	1	69	Medium	Unknown	Direct	Imprecise	Low	No difference
	Clinical score: 3 to 10 days	4	224	Medium	Inconsistent	Direct	Imprecise	Low	No difference
Inpatients/outpatients	Adverse events	5	1123	Low	Consistent	Direct	Precise	Moderate	No difference

Table 2. GRADE assessments: glucocorticoid versus placebo

Table 3. GRADE assessments: glucocorticoid and epinephrine versus placebo

Population	Outcome	Number of studies	Number of participants	GRADE domains				Strength of evidence	Intervention favoured
Population	Outcome	Number of studies	Number of participants	Risk of bias	Consistency	Directness	Precision	Strength of evidence	Intervention favoured
GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO
Outpatients	Admissions day 1	1	400	Low	Unknown	Direct	Imprecise	Low	Favours epi + dex but NS
	Admissions day 7	1	400	Low	Unknown	Direct	Imprecise	Low	Epi + dex
	Clinical score: 60 minutes	1	400	Low	Unknown	Direct	Precise	Moderate	Epi + dex
	Adverse events	1	400	Low	Unknown	Direct	Imprecise	Low	No difference
dex = dexamethasone epi = epinephrine NS = non‐significant

Table 3. GRADE assessments: glucocorticoid and epinephrine versus placebo

Table 4. Hospital re‐admissions and return healthcare visits (in‐ and outpatients)

Study	Population	Duration of follow‐up	Glucocorticoid‐including group	Placebo or comparator group	Notes
GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO: HOSPITAL RE‐ADMISSIONS
Roosevelt 1996	Inpatients	Days 1 to 14	0	0	(No events in either group)
Klassen 1997	Inpatients	Days 1 to 7	4/35 (11%)	1/32 (3%)	P = 0.36
Teeratakulpisarn 2007	Inpatients	Days 1 to 30	3/89 (3%)	7/85 (8%)
GLUCOCORTICOID versus PLACEBO: RETURN HEALTHCARE VISITS*
Plint 2009 (epinephrine ‐ E; dexamethasone ‐ D; placebo ‐ P)	Outpatients	Days 1 to 22	D + E 95/199 (48%)	P + E 93/198 (47%)	Return to the health care provider for bronchiolitis symptoms Difference between dexamethasone + placebo versus placebo + placebo, was significant in the unadjusted analysis (P = 0.04)
	Outpatients	Days 1 to 22	D + P 106/199 (53%)	P + P 86/201 (43%)
Schuh 2002	Outpatients	Days 7 to 28	9/35 (26%)	14/32 (44%)	Medical visits for continuing symptoms; P = 0.069
Klassen 1997	Inpatients	Days 1 to 7	29/35 (83%)	24/32 (75%)	P = 0.77
Roosevelt 1996	Inpatients	Days 1 to 14	16/65 (25%)	5/53 (9%)	P = 0.01; reported on visits made by the physician; 69% were for non‐respiratory difficulties
Teeratakulpisarn 2007	Inpatients	Days 1 to 30	17/89 (19%)	26/85 (31%)	Visit to emergency room or a private clinic because of respiratory symptoms
GLUCOCORTICOID versus EPINEPHRINE: RETURN HEALTHCARE VISITS
Plint 2009 (dexamethasone + placebo versus epinephrine + placebo)	Outpatients	Days 1 to 22	106/199 (53%)	93/198 (47%)	‐
GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO: RETURN HEALTHCARE VISITS
Plint 2009 (dexamethasone + epinephrine versus placebo + placebo)	Outpatients	Days 1 to 22	95/198 (48%)	86/201 (43%)	‐
*Berger 1998: no difference between groups, but did not report quantitative data. Data presented as n/N (%)

Table 4. Hospital re‐admissions and return healthcare visits (in‐ and outpatients)

Table 5. Symptoms and quality of life (in‐ and outpatients)*

Study	Population	Duration of follow‐up	Parameter	Glucocorticoid‐including group	Placebo or comparator group	Notes
GLUCOCORTICOID versus PLACEBO
Teeratakulpisarn 2007	Inpatients	Days 1 to 30	Time from treatment to being symptom free ‐ mean ± SD	7.0 ± 5.9	9.0 ± 6.4	P = 0.035
Cade 2000	Inpatients	Days 1 to 28	Time taken for half of infants to become asymptomatic for 48 hours (95% CI) ‐ time to event analysis	10 (10 to 13)	12 (10 to 16)	HR 1.41 (95% CI 0.98 to 2.04), P = 0.07
			Days with coughing or wheezing episodes ‐ mean ± SD	17.0 ± 7.6 days	17.1 ± 8.5	Mean difference: 0.91 days (95% CI ‐2.72 to 2.41), P = 0.91
Roosevelt 1996#	Inpatients	Day 10 to 14	No current difficulty breathing ‐ n/N (%)	45/45 (100)	37/42 (88)	P = 0.07
			Feeding and drinking well ‐ n/N (%)	45/45 (100)	40/42 (95)	P = 0.57
GLUCOCORTICOID versus PLACEBO, GLUCOCORTICOID versus EPINEPHRINE, GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO
Plint 2009 (epinephrine ‐ E; dexamethasone ‐ D; placebo ‐ P)	Outpatients	Days 1 to 22	Time to return to normal feeding ‐ median (IQR)	D + E: 0.6 (0.2 to 1.3) D + P: 0.8 (0.3 to 1.9) P + E: 0.5 (0.2 to 1.2) P + P: 0.9 (0.3 to 2.1)		Time to return to normal feeding ‐ mean symptom duration ratio D + E versus P + P: 0.63 (unadjusted 95% CI 0.5 to 0.8)¶ Time to return to quiet breathing ‐ mean symptom duration ratio D + E versus P + P: 0.83 (unadjusted 95% CI 0.69 to 1.00) No other reported comparison was statistically significant in adjusted analysis
			Time to return to normal sleeping ‐ median (IQR)	D + E: 0.7 (0.2 to 1.7) D + P: 0.8 (0.3 to 1.9) P + E: 0.8 (0.3 to 1.9) P + P: 0.8 (0.3 to 1.8)
			Time to no coughing ‐ median (IQR)	D + E: 12.6 (7.8 to 18.5) D + P: 13.8 (8.5 to 20.2) P + E: 13.2 (8.1 to 19.3) P + P: 13.3 (8.2 to 19.5)
			Time to quiet breathing ‐ median (IQR)	D + E: 3.1 (1.4 to 6.1) D + P: 3.7 (1.6 to 7.1) P + E: 3.6 (1.5 to 6.9) P + P: 3.7 (1.6 to 7.2)
*Units in days unless otherwise stated; no study assessed or reported data from generic or disease‐specific quality of life instruments; Richter 1998 also reported number of symptom‐free days for a 6‐week follow‐up period #Roosevelt 1996 primary outcome was time to resolution (defined as number of 12 h periods needed to achieve: O₂ saturation > 95% at room air, accessory muscle score = 0, wheeze = 0 or 1, and normal feeding); only association measures were reported: HR 1.3 (95% CI 0.9 to 1.3), P = 0.22 ¶time to symptom relief was analysed by means of parametric survival models with Weibull distributions assumed; 95% CI adjusted for multiple analysis in a factorial trial. CI = confidence interval HR = hazard ratio IQR = interquartile range SD = standard deviation

Table 5. Symptoms and quality of life (in‐ and outpatients)*

Table 6. Harms ‐ adverse events

Adverse event		Number of participants	Study	Glucocorticoid‐including group, N (%)	Placebo or comparator group, N (%)	Notes
GLUCOCORTICOID versus PLACEBO, GLUCOCORTICOID versus EPINEPHRINE, GLUCOCORTICOID AND EPINEPHRINE versus PLACEBO
Gastrointestinal	Vomiting	1466	Plint 2009*	D + E: 2/199 (1) D + P: 5/199 (2.5)	P + E: 4/198 (2) P + P: 3/201 (1.5)	Observed in the emergency department by research nurse
			Kuyucu 2004*	No events in either group (D + E, D + S, P + E, P + S)		Methods/timings NR
			Corneli 2007	16/305 (5.2)	14/295 (4.7)	Within 20 minutes after administration of the study medication
	Bleeding	1576	Corneli 2007	No events in either group		#
			Teeratakulpisarn 2007	2/90 (2)	1/89 (1)	Occult blood; also assessed diarrhoea separately. Methods/timings NR
			Plint 2009	D + E: 17/199 (8.5) D + P: 12/199 (6)	P + E: 14/198 (7) P+P: 16/201 (8)	Dark stools; reported by families during the 22‐day telephone follow‐up. No patient had more than 1 episode
Endocrine	Hypertension	1397	Plint 2009	D + E: 0/199 (0) D + P: 1/199 (0.5)	P + E: 1/198 (0.5) P + P: 0/201 (0)	Observed in infants admitted to hospital
			Corneli 2007	No events in either group		#
Infectious	Pneumonia	851	Corneli 2007	1/305 (3.3)	2/295 (7)	Also assessed empyema separately
			Teeratakulpisarn 2007	0/90 (0)	3/89 (3.4)	Methods/timings NR
			Klassen 1997	1/35 (3)	1/37 (3)	Methods/timings NR
	Varicella	1397	Corneli 2007	No events in either group		#
			Plint 2009	No events in either group		Reported by families during the 22‐day telephone follow‐up
General	Tremor	866	Kuyucu 2004	No events in either group		Methods/timings NR
			Plint 2009	D + E: 4/199 (2) D + P: 5/199 (2.5)	P + E: 4/198 (2) P + P: 2/201 (1)	Observed in the emergency department by research nurse
	Pallor/flushing	866	Kuyucu 2004	No events in either group		Methods/timings NR
			Plint 2009	D + E: 23/199 (11.5) D + P: 15/199 (7.5)	P + E: 22/198 (11.1) P + P: 16/201 (8)	Observed in the emergency department by research nurse
Additional reported adverse events: Goebel 2000 reported toxicity data: one patient was "jittery"; no evidence of further treatment complications. Plint 2009 also reported hyperkalaemia observed in infants admitted to hospital (only one case was noted in the dexamethasone group). *epinephrine ‐ E; dexamethasone ‐ D; salbutamol ‐ S; placebo ‐ P #Corneli 2007: Study clinicians and research assistants monitored the infants for adverse events during observation in the emergency department. Subsequent adverse events were determined at follow‐up. A patient safety committee, made up of people not involved with patient enrolment, tracked all adverse events.

Table 6. Harms ‐ adverse events

Comparison 1. Glucocorticoid versus placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome Show forest plot	10		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.1 Admissions day 1	10	1762	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.78, 1.08]
1.2 Admissions day 7	6	1530	Risk Ratio (M‐H, Random, 95% CI)	0.86 [0.70, 1.06]
2 Length of stay (inpatients) ‐ review primary outcome Show forest plot	8	633	Mean Difference (IV, Random, 95% CI)	‐0.18 [‐0.39, 0.04]

3 Length of stay (outpatients) Show forest plot	3	255	Mean Difference (IV, Random, 95% CI)	0.10 [‐0.81, 1.01]

4 Clinical scores (outpatients) Show forest plot	11		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 60 minutes	6	1006	Std. Mean Difference (IV, Random, 95% CI)	‐0.04 [‐0.16, 0.09]
4.2 120 minutes	5	214	Std. Mean Difference (IV, Random, 95% CI)	‐0.17 [‐0.55, 0.21]
4.3 3 to 6 hours	5	808	Std. Mean Difference (IV, Random, 95% CI)	‐0.14 [‐0.50, 0.21]
4.4 12 to 24 hours	2	69	Std. Mean Difference (IV, Random, 95% CI)	0.13 [‐0.51, 0.76]
4.5 3 to 10 days	5	224	Std. Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.61, 0.21]
5 Clinical scores (inpatients) Show forest plot	4		Std. Mean Difference (Random, 95% CI)	Subtotals only

5.1 3 to 6 hours	1	174	Std. Mean Difference (Random, 95% CI)	‐1.03 [‐1.87, ‐0.19]
5.2 6 to 12 hours	3	269	Std. Mean Difference (Random, 95% CI)	‐0.62 [‐1.00, ‐0.23]
5.3 12 to 24 hours	3	264	Std. Mean Difference (Random, 95% CI)	‐0.28 [‐0.66, 0.09]
5.4 24 to 72 hours	4	271	Std. Mean Difference (Random, 95% CI)	‐0.53 [‐1.14, 0.08]
6 O₂ saturation (outpatients) Show forest plot	8		Mean Difference (IV, Random, 95% CI)	Subtotals only

6.1 60 minutes	5	936	Mean Difference (IV, Random, 95% CI)	‐0.27 [‐0.73, 0.19]
6.2 120 minutes	2	75	Mean Difference (IV, Random, 95% CI)	‐0.10 [‐1.56, 1.37]
6.3 3 to 6 hours	5	808	Mean Difference (IV, Random, 95% CI)	‐0.43 [‐0.84, ‐0.02]
6.4 24 to 72 hours	1	38	Mean Difference (IV, Random, 95% CI)	0.20 [‐1.01, 1.41]
7 O₂ saturation (inpatients) Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

7.1 6 to 12 hours	1	67	Mean Difference (IV, Random, 95% CI)	‐0.7 [‐1.98, 0.58]
7.2 12 to 24 hours	2	116	Mean Difference (IV, Random, 95% CI)	‐0.44 [‐2.04, 1.16]
7.3 24 to 72 hours	1	67	Mean Difference (IV, Random, 95% CI)	1.10 [‐0.77, 2.97]
8 Respiratory rate (outpatients) Show forest plot	8		Mean Difference (IV, Random, 95% CI)	Subtotals only

8.1 60 minutes	3	861	Mean Difference (IV, Random, 95% CI)	‐0.24 [‐1.51, 1.03]
8.2 120 minutes	2	69	Mean Difference (IV, Random, 95% CI)	‐1.95 [‐9.30, 5.39]
8.3 3 to 6 hours	3	733	Mean Difference (IV, Random, 95% CI)	‐1.12 [‐3.07, 0.82]
8.4 12 to 24 hours	2	69	Mean Difference (IV, Random, 95% CI)	0.15 [‐7.10, 7.40]
8.5 3 to 10 days	4	174	Mean Difference (IV, Random, 95% CI)	‐1.64 [‐7.89, 4.61]
9 Respiratory rate (inpatients) Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

9.1 6 to 12 hours	1	66	Mean Difference (IV, Random, 95% CI)	‐4.0 [‐11.45, 3.45]
9.2 12 to 24 hours	2	110	Mean Difference (IV, Random, 95% CI)	‐1.22 [‐5.08, 2.64]
9.3 24 to 72 hours	1	28	Mean Difference (IV, Random, 95% CI)	‐1.90 [‐15.37, 11.57]
10 Heart rate (outpatients) Show forest plot	9		Mean Difference (IV, Random, 95% CI)	Subtotals only

10.1 60 minutes	5	936	Mean Difference (IV, Random, 95% CI)	0.46 [‐1.62, 2.55]
10.2 120 minutes	4	144	Mean Difference (IV, Random, 95% CI)	‐3.54 [‐8.83, 1.75]
10.3 3 to 6 hours	5	808	Mean Difference (IV, Random, 95% CI)	‐0.65 [‐7.01, 5.71]
10.4 12 to 24 hours	2	69	Mean Difference (IV, Random, 95% CI)	1.85 [‐11.18, 14.88]
10.5 3 to 10 days	3	136	Mean Difference (IV, Random, 95% CI)	0.43 [‐8.32, 9.18]
11 Heart rate (inpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

11.1 12 to 24 hours	1	49	Mean Difference (IV, Random, 95% CI)	‐9.0 [‐18.99, 0.99]
12 Hospital readmissions (inpatients) Show forest plot	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

12.1 2 to 10 days	1	67	Risk Ratio (M‐H, Random, 95% CI)	3.66 [0.43, 31.03]
12.2 10 to 30 days	2	292	Risk Ratio (M‐H, Random, 95% CI)	0.41 [0.11, 1.53]
13 Return healthcare visits (inpatients) Show forest plot	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

13.1 2 to 10 days	1	67	Risk Ratio (M‐H, Random, 95% CI)	1.10 [0.86, 1.42]
13.2 10 to 30 days	2	292	Risk Ratio (M‐H, Random, 95% CI)	1.21 [0.30, 4.96]
14 Return healthcare visits (outpatients) Show forest plot	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

14.1 10 to 30 days	3	863	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.80, 1.35]
15 Admissions at day 1 (outpatients) ‐ subgroup analysis protocolised use of bronchodilator Show forest plot	10	1762	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.78, 1.08]

15.1 Protocolised use of bronchodilator	7	717	Risk Ratio (M‐H, Random, 95% CI)	0.85 [0.56, 1.29]
15.2 No protocolised use of bronchodilator	3	1045	Risk Ratio (M‐H, Random, 95% CI)	0.94 [0.79, 1.13]
16 Admissions within 7 days (outpatients) ‐ subgroup analysis protocolised use of bronchodilator Show forest plot	6	1530	Risk Ratio (M‐H, Random, 95% CI)	0.86 [0.70, 1.06]

16.1 Protocolised use of bronchodilator	4	581	Risk Ratio (M‐H, Random, 95% CI)	0.68 [0.44, 1.05]
16.2 No protocolised use of bronchodilator	2	949	Risk Ratio (M‐H, Random, 95% CI)	0.95 [0.82, 1.11]
17 Admissions at day 1 (outpatients) ‐ subgroup analysis age Show forest plot	10	1762	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.78, 1.08]

17.1 All < 12 months	3	1397	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.79, 1.10]
17.2 Including > 12 months	7	365	Risk Ratio (M‐H, Random, 95% CI)	0.88 [0.51, 1.49]
18 Admissions within 7 days (outpatients) ‐ subgroup analysis age Show forest plot	6	1530	Risk Ratio (M‐H, Random, 95% CI)	0.86 [0.70, 1.06]

18.1 All < 12 months	3	1346	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.80, 1.06]
18.2 Including > 12 months	3	184	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.25, 1.83]
19 Length of stay (inpatients) ‐ subgroup analysis protocolised use of bronchodilator Show forest plot	8	633	Mean Difference (IV, Random, 95% CI)	‐0.18 [‐0.39, 0.04]

19.1 Protocolised use of bronchodilator	4	206	Mean Difference (IV, Random, 95% CI)	‐0.12 [‐0.23, ‐0.00]
19.2 No protocolised use of bronchodilator	4	427	Mean Difference (IV, Random, 95% CI)	‐0.31 [‐0.83, 0.20]
20 Length of stay (inpatients) ‐ subgroup analysis age Show forest plot	8	633	Mean Difference (IV, Random, 95% CI)	‐0.18 [‐0.39, 0.04]

20.1 All < 12 months	4	314	Mean Difference (IV, Random, 95% CI)	‐0.10 [‐0.65, 0.45]
20.2 Including > 12 months	4	319	Mean Difference (IV, Random, 95% CI)	‐0.21 [‐0.53, 0.12]
21 Length of stay (inpatients) ‐ subgroup analysis RSV status Show forest plot	8	633	Mean Difference (IV, Random, 95% CI)	‐0.18 [‐0.39, 0.04]

21.1 All RSV‐positive	3	251	Mean Difference (IV, Random, 95% CI)	‐0.09 [‐0.63, 0.45]
21.2 Not all RSV‐positive, or NR	5	382	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.57, 0.17]
22 Admissions (days 1 and 7) (outpatients) ‐ sensitivity analysis with only low overall RoB Show forest plot	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

22.1 Admissions day 1	3	1397	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.79, 1.10]
22.2 Admissions day 7	3	1346	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.80, 1.06]
23 Length of stay (inpatients) ‐ sensitivity analysis with only low overall RoB Show forest plot	1	67	Mean Difference (IV, Random, 95% CI)	0.38 [‐0.46, 1.21]

Comparison 1. Glucocorticoid versus placebo

Comparison 2. Glucocorticoid and epinephrine versus placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.1 Admissions day 1	1	400	Risk Ratio (M‐H, Random, 95% CI)	0.65 [0.40, 1.05]
1.2 Admissions day 7	1	400	Risk Ratio (M‐H, Random, 95% CI)	0.65 [0.44, 0.95]
2 Clinical scores (outpatients) Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 60 minutes	1	399	Std. Mean Difference (IV, Random, 95% CI)	‐0.34 [‐0.54, ‐0.14]
3 O₂ saturation (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1 60 minutes	1	399	Mean Difference (IV, Random, 95% CI)	0.04 [‐0.53, 0.61]
4 Respiratory rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 60 minutes	1	399	Mean Difference (IV, Random, 95% CI)	‐1.16 [‐3.06, 0.74]
5 Heart rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.1 60 minutes	1	399	Mean Difference (IV, Random, 95% CI)	8.44 [4.85, 12.03]
6 Return healthcare visits (outpatients) Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

6.1 10 to 30 days	1	399	Risk Ratio (M‐H, Random, 95% CI)	1.11 [0.89, 1.38]

Comparison 2. Glucocorticoid and epinephrine versus placebo

Comparison 3. Glucocorticoid and salbutamol versus placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (day 1) (outpatients) ‐ review primary outcome Show forest plot	1	30	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.13, 3.44]

2 Clinical scores (outpatients) Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 60 minutes	1	30	Std. Mean Difference (IV, Random, 95% CI)	‐0.30 [‐1.02, 0.42]
2.2 120 minutes	1	30	Std. Mean Difference (IV, Random, 95% CI)	‐0.22 [‐0.94, 0.50]
2.3 3 to 6 hours	1	30	Std. Mean Difference (IV, Random, 95% CI)	‐0.46 [‐1.18, 0.27]
3 O₂ saturation (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1 60 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐0.34 [‐1.75, 1.07]
3.2 120 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐0.67 [‐2.04, 0.70]
3.3 3 to 6 hours	1	30	Mean Difference (IV, Random, 95% CI)	‐1.08 [‐2.43, 0.27]
4 Heart rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 60 minutes	1	30	Mean Difference (IV, Random, 95% CI)	4.67 [‐1.89, 11.23]
4.2 120 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐1.87 [‐14.10, 10.36]
4.3 3 to 6 hours	1	30	Mean Difference (IV, Random, 95% CI)	4.3 [‐2.38, 10.98]

Comparison 3. Glucocorticoid and salbutamol versus placebo

Comparison 4. Glucocorticoid versus epinephrine

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (days 1 and 7) (outpatients) ‐ review primary outcome Show forest plot	2		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.1 Admissions day 1	2	444	Risk Ratio (M‐H, Random, 95% CI)	1.12 [0.66, 1.88]
1.2 Admissions day 7	1	399	Risk Ratio (M‐H, Random, 95% CI)	1.08 [0.77, 1.52]
2 Clinical scores (outpatients) Show forest plot	2		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 60 minutes	2	442	Std. Mean Difference (IV, Random, 95% CI)	0.31 [0.12, 0.50]
2.2 120 minutes	1	45	Std. Mean Difference (IV, Random, 95% CI)	0.35 [‐0.27, 0.98]
2.3 3 to 6 hours	1	45	Std. Mean Difference (IV, Random, 95% CI)	0.42 [‐0.20, 1.05]
3 O₂ saturation (outpatients) Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1 60 minutes	2	442	Mean Difference (IV, Random, 95% CI)	‐0.99 [‐1.46, ‐0.52]
3.2 120 minutes	1	45	Mean Difference (IV, Random, 95% CI)	‐0.07 [‐1.07, 0.94]
3.3 3 to 6 hours	1	45	Mean Difference (IV, Random, 95% CI)	‐0.58 [‐1.74, 0.57]
4 Respiratory rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 60 minutes	1	397	Mean Difference (IV, Random, 95% CI)	0.38 [‐1.44, 2.20]
5 Heart rate (outpatients) Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.1 60 minutes	2	442	Mean Difference (IV, Random, 95% CI)	‐7.56 [‐11.34, ‐3.79]
5.2 120 minutes	1	45	Mean Difference (IV, Random, 95% CI)	0.44 [‐7.59, 8.47]
5.3 3 to 6 hours	1	45	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐8.09, 7.69]

Comparison 4. Glucocorticoid versus epinephrine

Comparison 5. Glucocorticoid versus salbutamol

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (day 1) (outpatients) ‐ review primary outcome Show forest plot	1	45	Risk Ratio (M‐H, Random, 95% CI)	1.0 [0.21, 4.86]

2 Clinical scores (outpatients) Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 60 minutes	1	45	Std. Mean Difference (IV, Random, 95% CI)	0.65 [0.01, 1.28]
2.2 120 minutes	1	45	Std. Mean Difference (IV, Random, 95% CI)	0.36 [‐0.27, 0.98]
2.3 3 to 6 hours	1	45	Std. Mean Difference (IV, Random, 95% CI)	0.70 [0.06, 1.34]
3 O₂ saturation (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1 60 minutes	1	45	Mean Difference (IV, Random, 95% CI)	0.93 [‐0.71, 2.57]
3.2 120 minutes	1	45	Mean Difference (IV, Random, 95% CI)	0.22 [‐0.88, 1.33]
3.3 3 to 6 hours	1	45	Mean Difference (IV, Random, 95% CI)	‐0.03 [‐0.95, 0.88]
4 Heart rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 60 minutes	1	45	Mean Difference (IV, Random, 95% CI)	‐3.30 [‐9.49, 2.89]
4.2 120 minutes	1	45	Mean Difference (IV, Random, 95% CI)	‐7.53 [‐14.28, ‐0.78]
4.3 3 to 6 hours	1	45	Mean Difference (IV, Random, 95% CI)	‐5.12 [‐12.39, 2.15]

Comparison 5. Glucocorticoid versus salbutamol

Comparison 6. Glucocorticoid and epinephrine versus salbutamol

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (day 1) (outpatients) ‐ review primary outcome Show forest plot	1	35	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]

2 Clinical scores (outpatients) Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 120 minutes	1	35	Std. Mean Difference (IV, Random, 95% CI)	‐0.17 [‐0.87, 0.52]
2.2 12 to 24 hours	1	35	Std. Mean Difference (IV, Random, 95% CI)	0.0 [‐0.70, 0.70]
2.3 3 to 10 days	1	35	Std. Mean Difference (IV, Random, 95% CI)	‐1.22 [‐1.98, ‐0.46]
3 Respiratory rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1 120 minutes	1	35	Mean Difference (IV, Random, 95% CI)	‐3.10 [‐9.51, 3.31]
3.2 12 to 24 hours	1	35	Mean Difference (IV, Random, 95% CI)	‐2.80 [‐9.96, 4.36]
3.3 3 to 10 days	1	35	Mean Difference (IV, Random, 95% CI)	‐13.70 [‐20.56, ‐6.84]
4 Heart rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 120 minutes	1	35	Mean Difference (IV, Random, 95% CI)	‐3.20 [‐12.20, 5.80]
4.2 12 to 24 hours	1	35	Mean Difference (IV, Random, 95% CI)	‐1.40 [‐9.36, 6.56]
4.3 3 to 10 days	1	35	Mean Difference (IV, Random, 95% CI)	‐6.30 [‐14.21, 1.61]

Comparison 6. Glucocorticoid and epinephrine versus salbutamol

Comparison 7. Glucocorticoid and salbutamol versus epinephrine

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (day 1) (outpatients) ‐ review primary outcome Show forest plot	2	64	Risk Ratio (M‐H, Random, 95% CI)	5.00 [0.26, 96.13]

2 Clinical scores (outpatients) Show forest plot	2		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 60 minutes	1	30	Std. Mean Difference (IV, Random, 95% CI)	0.36 [‐0.36, 1.08]
2.2 120 minutes	2	64	Std. Mean Difference (IV, Random, 95% CI)	0.25 [‐0.26, 0.77]
2.3 12 to 24 hours	1	34	Std. Mean Difference (IV, Random, 95% CI)	0.30 [‐0.43, 1.02]
2.4 3 to 10 days	1	34	Std. Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.88, 0.56]
3 O₂ saturation (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1 60 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐1.54 [‐2.85, ‐0.23]
3.2 120 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐1.27 [‐2.41, ‐0.13]
4 Respiratory rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 120 minutes	1	34	Mean Difference (IV, Random, 95% CI)	‐0.60 [‐7.39, 6.19]
4.2 12 to 24 hours	1	34	Mean Difference (IV, Random, 95% CI)	3.20 [‐4.27, 10.67]
4.3 3 to 10 days	1	34	Mean Difference (IV, Random, 95% CI)	‐0.40 [‐6.47, 5.67]
5 Heart rate (outpatients) Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.1 60 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐3.33 [‐12.37, 5.71]
5.2 120 minutes	2	64	Mean Difference (IV, Random, 95% CI)	0.62 [‐5.38, 6.62]
5.3 12 to 24 hrs	1	34	Mean Difference (IV, Random, 95% CI)	5.30 [‐3.28, 13.88]
5.4 3 to 10 days	1	34	Mean Difference (IV, Random, 95% CI)	1.0 [‐6.94, 8.94]

Comparison 7. Glucocorticoid and salbutamol versus epinephrine

Comparison 8. Glucocorticoid versus glucocorticoid (prednisolone versus budesonide)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Admissions (day 1) (outpatients) ‐ review primary outcome Show forest plot	1	30	Risk Ratio (M‐H, Random, 95% CI)	3.0 [0.35, 25.68]

2 Clinical scores (outpatients) Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 60 minutes	1	30	Std. Mean Difference (IV, Random, 95% CI)	0.40 [‐0.33, 1.12]
2.2 120 minutes	1	30	Std. Mean Difference (IV, Random, 95% CI)	0.66 [‐0.08, 1.40]
2.3 3 to 6 hours	1	30	Std. Mean Difference (IV, Random, 95% CI)	0.23 [‐0.49, 0.95]
3 O₂ saturation (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1 60 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐1.46 [‐2.74, ‐0.18]
3.2 120 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐1.73 [‐3.06, ‐0.40]
3.3 3 to 6 hours	1	30	Mean Difference (IV, Random, 95% CI)	‐1.17 [‐2.37, 0.03]
4 Heart rate (outpatients) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 60 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐5.93 [‐13.29, 1.43]
4.2 120 minutes	1	30	Mean Difference (IV, Random, 95% CI)	‐7.39 [‐15.57, 0.79]
4.3 3 to 6 hours	1	30	Mean Difference (IV, Random, 95% CI)	‐10.8 [‐18.71, ‐2.89]

Comparison 8. Glucocorticoid versus glucocorticoid (prednisolone versus budesonide)