Glucocorticoids for croup in children

Allison Gates; Michelle Gates; Ben Vandermeer; Cydney Johnson; Lisa Hartling; David W Johnson; Terry P Klassen

doi:10.1002/14651858.CD001955.pub4

. 2018 Aug 22;2018(8):CD001955. doi: 10.1002/14651858.CD001955.pub4

Glucocorticoids for croup in children

Allison Gates ¹, Michelle Gates ¹, Ben Vandermeer ¹, Cydney Johnson ¹, Lisa Hartling ¹, David W Johnson ², Terry P Klassen ^3,^✉

Editor: Cochrane Acute Respiratory Infections Group

PMCID: PMC6513469 PMID: 30133690

Abstract

Background

Glucocorticoids are commonly used for croup in children. This is an update of a Cochrane Review published in 1999 and previously updated in 2004 and 2011.

Objectives

To examine the effects of glucocorticoids for the treatment of croup in children aged 0 to 18 years.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (the Cochrane Library, Issue 2, 2018), which includes the Cochrane Acute Respiratory Infections Group's Specialised Register, Ovid MEDLINE Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations and Ovid MEDLINE (1946 to 3 April 2018), and Embase (Ovid) (1996 to 3 April 2018, week 14), and the trials registers ClinicalTrials.gov (3 April 2018) and the World Health Organization International Clinical Trials Registry Platform (ICTRP, 3 April 2018). We scanned the reference lists of relevant systematic reviews and of the included studies.

Selection criteria

We included randomised controlled trials (RCTs) that investigated children aged 0 to 18 years with croup and measured the effects of glucocorticoids, alone or in combination, compared to placebo or another pharmacologic treatment. The studies needed to report at least one of our primary or secondary outcomes: change in croup score; return visits, (re)admissions or both; length of stay; patient improvement; use of additional treatments; and adverse events.

Data collection and analysis

One author extracted data from each study and another verified the extraction. We entered the data into Review Manager 5 for meta‐analysis. Two review authors independently assessed risk of bias for each study using the Cochrane 'Risk of bias' tool and the certainty of the body of evidence for the primary outcomes using the GRADE approach.

Main results

We added five new RCTs with 330 children. This review now includes 43 RCTs with a total of 4565 children. We assessed most (98%) studies as at high or unclear risk of bias. Compared to placebo, glucocorticoids improved symptoms of croup at two hours (standardised mean difference (SMD) ‐0.65, 95% confidence interval (CI) ‐1.13 to ‐0.18; 7 RCTs; 426 children; moderate‐certainty evidence), and the effect lasted for at least 24 hours (SMD ‐0.86, 95% CI ‐1.40 to ‐0.31; 8 RCTs; 351 children; low‐certainty evidence). Compared to placebo, glucocorticoids reduced the rate of return visits or (re)admissions or both (risk ratio 0.52, 95% CI 0.36 to 0.75; 10 RCTs; 1679 children; moderate‐certainty evidence). Glucocorticoid treatment reduced the length of stay in hospital by about 15 hours (mean difference ‐14.90, 95% CI ‐23.58 to ‐6.22; 8 RCTs; 476 children). Serious adverse events were infrequent. Publication bias was not evident. Uncertainty remains with regard to the optimal type, dose, and mode of administration of glucocorticoids for reducing croup symptoms in children.

Authors' conclusions

Glucocorticoids reduced symptoms of croup at two hours, shortened hospital stays, and reduced the rate of return visits to care. Our conclusions have changed, as the previous version of this review reported that glucocorticoids reduced symptoms of croup within six hours.

Plain language summary

Glucocorticoids for croup in children

Review question

We assessed the effectiveness of glucocorticoids for croup in children to determine if they reduced croup symptoms; minimised return visits to care; shortened length of stay; reduced the need for additional treatments; or had side effects.

Background

Croup causes the throat and windpipe to swell, resulting in hoarseness, a barking cough, and noisy breathing. Glucocorticoids can reduce swelling, making it easier to breathe.

This review was previously published in 1999 and updated in 2004 and 2011.

Search date

We searched for articles published up to 3 April 2018.

Study characteristics

We included 43 studies with 4565 children aged up to 18 years published from 1964 to 2013. The glucocorticoids investigated included beclomethasone, betamethasone, budesonide, dexamethasone, fluticasone, and prednisolone. Most studies (26, 60%) compared any glucocorticoid to placebo. Of these, 15 (58%) tested dexamethasone compared to placebo. Three studies compared 0.6 mg/kg to 0.15 mg/kg dosages of dexamethasone, a common clinical question. Half of the studies (22, 51%) described outpatients who presented to emergency departments or outpatient clinics, and 18 (42%) took place in North America, eight (19%) in Europe, seven (16%) in Asia, and 10 (23%) in Australia. Twenty‐six (60%) studies compared glucocorticoids to fake treatment (placebo); four (10%) compared glucocorticoids to epinephrine; 11 (26%) compared one glucocorticoid to another; three (7%) compared one glucocorticoid to a combination of glucocorticoids; five (12%) compared glucocorticoids given in different ways; and four (9%) compared glucocorticoids given in different amounts.

Study funding sources

Funding sources included government (12%), academic or research institute (7%), industry (19%), or foundations (7%). More than half of studies (56%) did not report funding sources.

Key results

Glucocorticoids improved croup symptoms at two hours (moderate‐certainty evidence), and the effect lasted at least 24 hours (low‐certainty evidence). Glucocorticoids reduced rates of return visits, admissions, and readmissions (moderate‐certainty evidence). When treated with placebo, 204 of every 1000 children will return for medical care. When treated with glucocorticoids, 74 to 153 of every 1000 children will return for medical care. Glucocorticoids reduced length of stay by 15 hours (range 6 to 24 hours), but made no difference in need for additional treatments. Of studies that compared glucocorticoids to placebo, 50% collected data on side effects. Four studies reported rare occurrences of secondary infections (e.g. pneumonia, ear infection). Most other side effects were not severe (e.g. emotional distress, hyperactivity, vomiting). We are not certain which type, amount, and administration mode (oral, inhaled, injected) of glucocorticoids is best for reducing symptoms of croup in children.

Quality of the evidence

Most studies (98%) had methods problems, reporting issues, or both.

Summary of findings

Background

Description of the condition

Croup (laryngotracheobronchitis) is a common cause of upper airway obstruction in children and is characterised by hoarseness, a barking cough, and inspiratory stridor. These symptoms are thought to occur as a result of oedema of the larynx and trachea, which has been triggered by a recent viral infection. Parainfluenza virus type 1 is the pathogen most commonly identified in cases of croup (Rihkanen 2008). Although croup is a self limiting illness, it is a large burden on healthcare systems because of frequent visits to doctors and emergency departments and, when necessary, hospitalisations. Each year, about 3% of children are diagnosed with croup, usually between the ages of six months and three years (Johnson 2009). During a six‐year period in Alberta, Canada, 20,019 infants aged under two years visited emergency departments for 27,355 episodes of croup, of whom 8.0% were admitted to hospital (Rosychuk 2010). Population‐based studies indicate that less than 5% of children with croup are typically admitted to the hospital for treatment (Bjornson 2008).

Description of the intervention

Since the late 1980s it has been recognised that glucocorticoids provide some clinical benefit for children with croup. In 1989, a meta‐analysis by Kairys 1989 showed that treatment with glucocorticoids resulted in significantly greater clinical improvement after 12 hours and a reduced incidence of intubations as compared to placebo. In 2000, a meta‐analysis by Griffin 2000 demonstrated that treatment with glucocorticoids resulted in greater improvements in croup score and fewer hospital admissions compared to placebo. Although racemic adrenaline (epinephrine), or L‐adrenaline, has been shown to provide temporary relief to children with croup, it is not thought to have long‐term benefits (Bjornson 2008; Johnson 2009; Waisman 1992). In the past, the standard management of croup included mist treatment (i.e. treatment with humidified air), although there is little evidence of its effectiveness (Bjornson 2008; Lavine 2001; Moore 2007; Neto 2002; Scolnik 2006).

How the intervention might work

Localised inflammation of the upper airway caused by a virus is believed to cause croup symptoms. The narrowest part of the airway is the subglottic region. Even small amounts of oedema in this region can significantly increase the work of breathing in young children because airway resistance increases dramatically. Poiseuille's law states that airway resistance is related to the radius of the airway to the power of four (Loftis 2006). Glucocorticoids are believed to reduce the degree of inflammation and swelling through their anti‐inflammatory properties, leading to decreased effort in breathing for the child.

Why it is important to do this review

This Cochrane Review was first published in the Cochrane Library in 1999 and included 24 randomised controlled trials (RCTs) that examined the effectiveness of glucocorticoids for children with croup (Ausejo 2000). Since then, there has been continued interest in the use of glucocorticoids to treat young people with croup, and new RCTs have been published. The review was updated in 2004 (Russell 2004) and 2011 (Russell 2011) to include RCTs published since the first version of this review was completed (Ausejo 2000). Areas of uncertainty remained regarding the use of glucocorticoids in children with croup, for example glucocorticoids provided via different modes of administration (oral, nebulised, intramuscular) and the role of glucocorticoids in mild cases of croup. The validity of the results in previous versions of the review was threatened by publication bias. The incorporation of new RCTs may influence the presence of publication bias.

It was important to update this review to incorporate newly published evidence. Furthermore, previous versions of this review did not extract and analyse data for all relevant primary and secondary outcomes, and did not incorporate complete assessments of study‐level risk of bias or outcome‐level certainty of evidence. These elements are now recognised as critical to interpreting the findings of a systematic review. Moreover, data on adverse events, a potentially important outcome, were not extracted. For the first time in this update, we have extracted and analysed data for all primary and secondary outcomes, judged risk of bias for all included studies, and assessed the certainty of the evidence for the primary outcomes using GRADE principles. We have also presented data on adverse events.

Objectives

To examine the effects of glucocorticoids for the treatment of croup in children aged 0 to 18 years.

Methods

Criteria for considering studies for this review

Types of studies

We only included RCTs. Studies that met our inclusion criteria were included regardless of language, publication status, trial conduct and reporting quality, or risk of bias. We excluded all other study designs.

Types of participants

We included studies of children aged 0 to 18 years diagnosed with croup, pseudo croup, or laryngotracheitis. We defined croup as a syndrome consisting of hoarseness, barking cough, and stridor, in which alternative diagnosis of acute stridor was excluded. We included both inpatients and outpatients; we considered children admitted to the emergency department to be outpatients.

Types of interventions

We included studies where the experimental intervention was one or more glucocorticoid, provided via any mode of administration. We placed no restrictions on the type or dose of glucocorticoid administered. We included any study where the comparator was placebo or any other active pharmacologic treatment. Comparisons included: any glucocorticoid compared to placebo; any glucocorticoid compared to epinephrine; one glucocorticoid compared to one or a combination of other glucocorticoids; glucocorticoids given by different modes of administration; glucocorticoids given in different doses. We excluded studies if none of the treatment groups received one or more glucocorticoid.

Types of outcome measures

We included RCTs that measured and reported on one or more of our primary or secondary outcomes, as follows.

Primary outcomes

Change in clinical croup score from baseline to 2, 6, 12, and/or 24 hours.
Return visits or (re)admissions to the hospital or both.

Secondary outcomes

Length of stay in the hospital or emergency department.
Patient improvement at 2, 6, 12, and/or 24 hours (yes or no, as reported in the individual studies).
The use of additional treatments, including: antibiotics, epinephrine, intubation/tracheostomy, mist tent, and/or supplemental glucocorticoids.
Any adverse events.

We excluded studies that did not report on any of the primary or secondary outcomes.

Search methods for identification of studies

Electronic searches

We searched electronic sources using a strategy developed by a research librarian on 3 April 2018 (Appendix 1). We included subject headings and key words for croup and glucocorticoids, and restricted the search to RCTs. We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (the Cochrane Library, Issue 2, 2018), which includes the Cochrane Acute Respiratory Infections Group's Specialised Register, Ovid MEDLINE Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations and Ovid MEDLINE (1946 to 3 April 2018), and Embase (Ovid) (1996 to 3 April 2018, week 14).

An interim update search of the following databases was performed by previous review authors in 2014: CENTRAL (the Cochrane Library 2014, Issue 8), MEDLINE (1966 to September week 1, 2014), MEDLINE In‐Process & Other Non‐Indexed Citations (12 September 2014), and Embase (1974 to September 2014) (Appendix 2). The results for the 2014 searches are included in this update of the review, and previous versions of the review have been used as sources of studies (Ausejo 1999; Ausejo 2000; Russell 2004; Russell 2011).

Searching other resources

We searched the US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov) and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (apps.who.int/trialsearch) on 3 April 2018 (Appendix 1). We scanned the reference lists of relevant systematic reviews identified during screening and of the included studies to identify additional relevant primary studies. A separate author group searched the same trial registries as part of the interim update search on 16 September 2014.

Data collection and analysis

Selection of studies

The unique records identified via the search following the removal of duplicates were stored in EndNote reference management software (EndNote). These records were transferred to a Microsoft Excel workbook for screening (Microsoft Excel). Two review authors (AG, CJ) independently screened the records, first by title and abstract (when available), and then by full text. During title and abstract screening, the review authors marked each record as 'include', 'exclude', or 'unsure'. All records marked 'include' or 'unsure' by either review author were selected for full‐text screening. During full‐text screening, the review authors again marked each candidate record as 'include', 'exclude', or 'unsure'. The review authors then convened to reach consensus on which studies should be included. In the case of uncertainty, a third review author (MG) or an independent third party provided arbitration. Members of another author group screened the records identified in the 2014 interim update search following a similar process.

Data extraction and management

Two review authors (AG, MG) extracted data from the English and French language studies using a Microsoft Excel workbook designed by the review author team (Microsoft Excel). Data from the Spanish and Danish language studies were extracted by native or second‐language speakers external to the author team. From each included study, we extracted characteristics of the children (inpatient or outpatient; croup severity), croup score used (if relevant), and experimental and control interventions (type of drug, route of administration, dosage). We also collected data on the primary and secondary outcome measures when available. We extracted croup‐related return visits or (re)admissions or both if specified in the publication. If the reason for a return visit and/or (re)admission was not specified, we extracted all return visits, admissions, and readmissions. In some studies, admissions, return visits, and readmissions were reported. In this case, we extracted the data as follows: admissions when reported with or without return visits and/or readmissions; return visits if reported with or without readmissions, and admissions was not reported; and readmissions if neither admissions nor return visits was reported. One review author (AG, MG, or CJ) verified the data extracted from each individual study to identify and correct errors or omissions. No additional data were obtained from trial authors for this update.

We extracted data from graphs using Plot Digitizer open source software from 12 studies (Amir 2006; Cetinkaya 2004; Duman 2005; Eboriadou 2010; Fifoot 2007; Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 2005; Husby 1993; Kuusela 1988; Roberts 1999). Plot Digitizer and similar tools enable data to be extracted from graphs with greater efficiency and interrater reliability compared to manual extraction (Burda 2017; Kadic 2016). Two review authors (AG, MG) independently extracted data from the available graphs and convened to reach consensus on the values obtained. When the extracted values were very similar, we took the mean of both authors' estimates. When the values did not coincide, we re‐extracted the data to resolve errors in estimation.

Assessment of risk of bias in included studies

We used the Cochrane 'Risk of bias' tool (2010 version) to assess risk of bias for all included studies (Higgins 2011b). Following Cochrane standards, we judged the risk of bias for each record as low, unclear, or high among seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective reporting, and other bias. We determined overall risk of bias as follows: low when all domains were judged as low risk; unclear when one or more domains were judged as unclear risk and no domains were judged as high risk; and high when one or more domains were judged as high risk. Two review authors (AG, MG) independently assessed risk of bias for each included study. For the Danish and Spanish language studies, native or second‐language speakers extracted relevant text snippets from the studies and translated these to English to facilitate the 'Risk of bias' assessments. In the case of disagreement, the review authors discussed the judgements for each domain and overall until consensus was reached. When the review authors could not reach consensus, a third party external to the review team was consulted to facilitate a decision.

When available, we used trial registers to complement the assessment of risk of bias. We searched for trials registered on ClinicalTrials.gov, the WHO ICTRP, the ISRCTN registry, and via Google. We identified trial register records for two (5%) of 43 included studies (Fifoot 2007; Garbutt 2013).

Measures of treatment effect

We entered all search results into Review Manager 5 for analysis (Review Manager 2014). Croup scores were reported as the Westley score (Westley 1978), the telephone outpatient (TOP) score (Bjornson 2016), the Downes and Raphaelly score (Downes 1975), or various author‐created scales. We therefore used standardised mean differences (SMDs) to combine the outcome for any croup score. A treatment effect (difference between treatment means) divided by its measurement variation (e.g. a pooled standard deviation) gives the SMD. We did not find effect estimates to be significantly different between Westley and other croup scores, so we included studies that reported any croup score in the subgroup analyses. Of note, a decrease in Westley score of one point from baseline is thought to be a clinically important change.

We expressed length of stay as mean differences (MDs) and calculated an overall MD. We calculated risk ratios (RRs) for binary data (i.e. return visits and/or (re)admissions, patient improvement, use of additional treatments). We calculated risk differences (RDs) where outcomes had zero events in both groups. For return visits and/or (re)admissions, we calculated the number needed to treat for an additional beneficial outcome (NNTB) for significant results. Because there was substantial variation in control group event rates between studies, we reported the NNTB for the mean control group rate, as well as for the smallest and largest control group rate observed.

We reported data on adverse events narratively.

Unit of analysis issues

We calculated the change from baseline croup score in 28 (65%) studies where the change from baseline measures was not reported directly (Alshehr 2005; Amir 2006; Cetinkaya 2004; Chub‐Uppakarn 2007; Dobrovoljac 2012; Duman 2005; Eboriadou 2010; Fifoot 2007; Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 2005; Godden 1997; Husby 1993; Johnson 1996; Klassen 1994; Klassen 1998; Kuusela 1988; Leipzig 1979; Martinez Fernandez 1993; Massicotte 1973; Rittichier 2000; Roberts 1999; Roorda 1998; Super 1989; Vad Pedersen 1998; Von Mühlendahl 1982).

We pooled counts, means, and variances using standard formulae for six (14%) studies that contained more than one experimental treatment group (Cetinkaya 2004; Eboriadou 2010; Fifoot 2007; Geelhoed 1995c; Johnson 1998; Luria 2001). One study by Geelhoed (Geelhoed 1995a; Geelhoed 1995b), and another by Skowron (Skowron 1966a; Skowron 1966a&b; Skowron 1966b), presented the results of two individual trials in one publication. We treated these as separate comparisons in the analyses and used pooled counts only when they were reported as such in the publications.

Dealing with missing data

When they were not directly reported, we estimated the variances for continuous data in accordance with the work of Abrams 2005 and Follmann 1992. Using standard formulae, we imputed standard deviations from standard errors in three (7%) studies (Alshehr 2005; Johnson 1998; Von Mühlendahl 1982), ranges in three (7%) studies (Alshehr 2005; Roorda 1998; Super 1989), 95% confidence intervals (CIs) in two (5%) studies (Fitzgerald 1996; Klassen 1998), and interquartile ranges (IQRs) in three (7%) studies (Johnson 1996; Klassen 1994; Klassen 1998). For the change in croup score from baseline, when not directly reported (n = 14, 33%), we derived the variance of the change assuming a correlation of 0.5 between pre‐ and post‐treatment scores (Alshehr 2005; Amir 2006; Chub‐Uppakarn 2007; Fitzgerald 1996; Johnson 1996; Klassen 1994; Klassen 1998; Kuusela 1988; Leipzig 1979; Martinez Fernandez 1993; Roorda 1998; Super 1989; Vad Pedersen 1998; Von Mühlendahl 1982).

For 11 (26%) studies where there were inadequate data from which to impute variances for change in croup score or length of stay, we substituted average variances from other studies in the main analysis (Cetinkaya 2004; Dobrovoljac 2012; Eboriadou 2010; Geelhoed 1995c; Geelhoed 2005; Godden 1997; Husby 1993; Kuusela 1988; Massicotte 1973; Roberts 1999; Skowron 1966a; Skowron 1966b). When the number of studies with imputed data within a meta‐analysis is relatively small, Furukawa and colleagues assert that variance data can be safely borrowed from other studies and still provide accurate results (Furukawa 2006). For certain outcomes only one study was included in the comparison, and that study did not report a variance estimate. In this case, we did not calculate a point estimate of effect (Cetinkaya 2004; Duman 2005; Fifoot 2007; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 2005; Rittichier 2000).

We substituted medians for means in eight (19%) studies (Alshehr 2005; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Johnson 1996; Klassen 1994; Klassen 1998; Super 1989; Von Mühlendahl 1982). When data for our prespecified time points (2, 6, 12, and 24 hours from baseline) were not reported, we used time points close to these if available. We substituted one hour for two hours in one study (Dobrovoljac 2012); four hours for six hours in 12 (28%) studies (Alshehr 2005; Amir 2006; Fifoot 2007; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 2005; Godden 1997; Johnson 1996; Klassen 1994; Klassen 1996; Klassen 1998; Massicotte 1973); five hours or discharge for six hours in one study (Johnson 1998); and 14 hours for 12 hours in one study (Massicotte 1973).

Assessment of heterogeneity

We assessed heterogeneity quantitatively with the Chi² test for heterogeneity and the I² statistic (Higgins 2002). The I² statistic indicates the per cent variability due to between‐study (or interstudy) variability as opposed to within‐study (or intrastudy) variability. We considered an I² of less than 40% to be low (potentially unimportant), 30% to 60% to be moderate, 50% to 90% to be substantial, and 75% to 100% to be considerable (see Higgins 2011a, Section 9.5.2).

Assessment of reporting biases

In addition to visually inspecting the funnel plots, we used the rank correlation test and weighted regression for the detection of publication bias (Begg 1994; Egger 1997; Light 1984). We used more than one method because the relative merits of the methods are not well established.

Data synthesis

We used random‐effects models to combine treatment effects regardless of quantified heterogeneity for the analyses of all outcomes.

GRADE and 'Summary of findings' tables

We created 'Summary of findings' tables for our two main comparisons (any glucocorticoid compared to placebo, and any glucocorticoid compared to epinephrine) for the primary outcomes: change in croup score at 2, 6, 12, and 24 hours from baseline, and return visits or (re)admissions or both. We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of the body of evidence as it relates to the studies that contributed data to the meta‐analyses (Atkins 2004). We used methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), employing GRADEpro GDT 2015 software. We justified all decisions to downgrade the certainty of the evidence using footnotes, and made comments to aid the readers' understanding where necessary. Of note, we downgraded single‐study analyses for inconsistency, because there was no evidence of consistency. We also created 'Summary of findings' tables for the remaining comparisons. To not detract from the two main comparisons, these are included in the Additional tables section.

Subgroup analysis and investigation of heterogeneity

We explored heterogeneity between studies using subgroup analyses for the primary outcomes of change in croup score from baseline to 2, 6, 12, and 24 hours, and return visits or (re)admissions or both, using the Chi² test for subgroup differences in meta‐analysis. We explored heterogeneity by croup score, by inpatient or outpatient status, and by glucocorticoid.

Sensitivity analysis

In some analyses, we imputed variance data for most of the included RCTs (e.g. any glucocorticoid compared to placebo, change in croup score after two hours). We undertook sensitivity analyses for these and all other analyses containing imputed variance data using the largest, smallest, and average variances from the other included RCTs. As per the protocol for the review, we did not undertake any additional sensitivity analyses.

Results

Description of studies

Results of the search

An interim search in 2014 identified 30 unique studies. The previous authors screened the studies, and added one new included study, Dobrovoljac 2012, and one new excluded study, Faghihinia 2007. The previous authors classified three trials as awaiting classification (Eboriadou 2010; Garbutt 2013; Mohammadzadeh 2014).

Our 2018 searches identified 79 records. This included 69 records retrieved via the database searches; five records identified via the trial register searches; three trials awaiting classification from the 2014 search (Eboriadou 2010; Garbutt 2013; Mohammadzadeh 2014); and two trials identified via other means (Chub‐Uppakarn 2007; Soleimani 2013). We identified Soleimani 2013 via a search of Google.ca. This study was not indexed in any of the databases that we otherwise searched. We identified Chub‐Uppakarn 2007 following a scan of the list of excluded studies from the previous version of this review, and it appeared to meet our inclusion criteria.

After removing duplicates, we screened 55 records by title and abstract and excluded 44 records. We screened the remaining 11 records by full text and included four studies (Chub‐Uppakarn 2007; Eboriadou 2010; Garbutt 2013; Soleimani 2013). We also included one new ongoing study (ACTRN12609000290291). We added five excluded studies: Gill 2017 and Roked 2015 were not randomised trials; the intervention was not a glucocorticoid in Eghbali 2016 and Mohammadzadeh 2014; and Faraji‐Goodarzi 2018 did not report any relevant outcomes. We excluded one new ongoing study as it reported no relevant outcomes and was terminated due to inadequate recruitment (NCT01748162).

A flow diagram illustrating the study selection process for the 2018 search is shown in Figure 1. We added five new RCTs with 330 children (Chub‐Uppakarn 2007; Dobrovoljac 2012; Eboriadou 2010; Garbutt 2013; Soleimani 2013). We included 43 studies (involving a total of 4565 children) in this 2018 updated review (see Characteristics of included studies). Two of the included studies reported on the findings of two individual RCTs each (Geelhoed 1995a; Geelhoed 1995b; Skowron 1966a; Skowron 1966a&b; Skowron 1966b). We presented these as separate comparisons in the analyses and only pooled the data when they were presented as such in the publication (Skowron 1966a&b). One of the included studies was published in both English and Danish (Husby 1993). We used the English report to complete the data extraction and 'Risk of bias' appraisal.

Flow diagram of study selection for the 2018 update searches. In 2014, the review authors identified 30 studies in an interim update search. The review authors included one new study, Dobrovoljac 2012, and excluded one new study, Faghihinia 2007.

Included studies

Participant and trial characteristics

Forty studies (93%) were published in English, and one each in French (Massicotte 1973), Spanish (Martinez Fernandez 1993), and Danish (Vad Pedersen 1998). Three studies (7%) included children with mild croup (Bjornson 2004; Geelhoed 1996; Luria 2001). Sample sizes tended to be small with a median of 72 (interquartile range (IQR) 54 to 99) children. Twenty‐two studies (51%) assessed outpatient children (n = 21 emergency department visits; n = 1 physician office visits) (Alshehr 2005; Amir 2006; Bjornson 2004; Cetinkaya 2004; Cruz 1995; Dobrovoljac 2012; Donaldson 2003; Duman 2005; Eboriadou 2010; Fifoot 2007; Garbutt 2013; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1996; Johnson 1996; Johnson 1998; Klassen 1994; Klassen 1996; Klassen 1998; Luria 2001; Rittichier 2000; Soleimani 2013; Sparrow 2006). Twenty‐one studies (49%) assessed hospitalised children (Chub‐Uppakarn 2007; Eden 1964; Eden 1967; Fitzgerald 1996; Geelhoed 1995c; Geelhoed 2005; Godden 1997; Husby 1993; James 1969; Koren 1983; Kuusela 1988; Leipzig 1979; Martinez Fernandez 1993; Massicotte 1973; Roberts 1999; Roorda 1998; Skowron 1966a; Skowron 1966b; Super 1989; Tibballs 1992; Vad Pedersen 1998; Von Mühlendahl 1982).

Thirty‐one studies (72%) were two‐armed trials (Alshehr 2005; Amir 2006; Bjornson 2004; Chub‐Uppakarn 2007; Cruz 1995; Dobrovoljac 2012; Donaldson 2003; Eden 1964; Eden 1967; Fitzgerald 1996; Garbutt 2013; Geelhoed 1996; Geelhoed 2005; Godden 1997; Husby 1993; James 1969; Johnson 1996; Klassen 1994; Klassen 1996; Koren 1983; Leipzig 1979; Massicotte 1973; Rittichier 2000; Roberts 1999; Roorda 1998; Soleimani 2013; Sparrow 2006; Super 1989; Tibballs 1992; Vad Pedersen 1998; Von Mühlendahl 1982); seven studies (16%) were three‐armed trials (Duman 2005; Eboriadou 2010; Fifoot 2007; Geelhoed 1995c; Johnson 1998; Klassen 1998; Luria 2001); and three studies (7%) were four‐armed trials (Cetinkaya 2004; Kuusela 1988; Martinez Fernandez 1993). Two studies (5%) included two individual two‐armed trials each (Geelhoed 1995a; Geelhoed 1995b; Skowron 1966a; Skowron 1966b).

Characteristics of the comparisons

Twenty‐six studies (60%) investigated any glucocorticoid compared to placebo. Of these, 15 (58%) investigated dexamethasone (Bjornson 2004; Cruz 1995; Dobrovoljac 2012; Eden 1967; Geelhoed 1996; James 1969; Johnson 1996; Koren 1983; Kuusela 1988; Leipzig 1979; Luria 2001; Martinez Fernandez 1993; Skowron 1966a&b; Super 1989; Von Mühlendahl 1982); four (15%) investigated budesonide (Godden 1997; Husby 1993; Klassen 1994; Roberts 1999); three (12%) investigated prednisolone (Eden 1964; Massicotte 1973; Tibballs 1992); one (4%) investigated fluticasone (Roorda 1998); and three (12%) investigated both dexamethasone and budesonide (Cetinkaya 2004; Geelhoed 1995c; Johnson 1998). Four studies (10%) investigated any glucocorticoid compared to epinephrine. Of these, one investigated budesonide (Fitzgerald 1996); two investigated dexamethasone (Kuusela 1988; Martinez Fernandez 1993); and one investigated both dexamethasone and beclomethasone (Eboriadou 2010).

Eleven studies (26%) investigated one glucocorticoid compared to another glucocorticoid. Of these, six investigated dexamethasone compared to budesonide (Cetinkaya 2004; Duman 2005; Geelhoed 1995c; Johnson 1998; Klassen 1998; Vad Pedersen 1998); one investigated dexamethasone compared to betamethasone (Amir 2006); one investigated dexamethasone compared to beclomethasone (Eboriadou 2010); and three investigated dexamethasone compared to prednisolone (Fifoot 2007; Garbutt 2013; Sparrow 2006). Three studies investigated one glucocorticoid compared to a combination of glucocorticoids. Of these, one investigated dexamethasone and budesonide compared to a combination of dexamethasone and budesonide (Klassen 1998), and two investigated dexamethasone compared to a combination of dexamethasone and budesonide (Geelhoed 2005; Klassen 1996).

Five studies (12%) investigated dexamethasone using different modes of administration. Of these, four investigated oral compared to intramuscular dexamethasone (Cetinkaya 2004; Donaldson 2003; Rittichier 2000; Soleimani 2013), and one investigated oral compared to nebulised dexamethasone (Luria 2001). Four studies investigated dexamethasone given in different doses. Of these, three investigated 0.60 mg/kg compared to 0.15 mg/kg dexamethasone (Alshehr 2005; Chub‐Uppakarn 2007; Fifoot 2007), and one investigated both 0.60 mg/kg compared to 0.30 mg/kg and 0.30 mg/kg compared to 0.15 mg/kg dexamethasone (Geelhoed 1995a; Geelhoed 1995b).

Reported outcomes: primary outcomes

Fifteen studies (35%) reported a two‐hour change in croup score (Amir 2006; Chub‐Uppakarn 2007; Dobrovoljac 2012; Duman 2005; Eboriadou 2010; Fifoot 2007; Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 2005; Godden 1997; Husby 1993; Johnson 1996; Roberts 1999; Roorda 1998); 20 (47%) reported a six‐hour change in croup score (Alshehr 2005; Amir 2006; Chub‐Uppakarn 2007; Fifoot 2007; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Godden 1997; Johnson 1996; Johnson 1998; Klassen 1994; Klassen 1996; Klassen 1998; Kuusela 1988; Martinez Fernandez 1993; Massicotte 1973; Roberts 1999; Roorda 1998; Vad Pedersen 1998; Von Mühlendahl 1982); 12 (28%) reported a 12‐hour change in croup score (Alshehr 2005; Chub‐Uppakarn 2007; Fitzgerald 1996; Geelhoed 1995c; Godden 1997; Kuusela 1988; Martinez Fernandez 1993; Massicotte 1973; Roberts 1999; Super 1989; Vad Pedersen 1998; Von Mühlendahl 1982); and 11 (26%) reported a 24‐hour change in croup score (Alshehr 2005; Cetinkaya 2004; Fitzgerald 1996; Godden 1997; Kuusela 1988; Leipzig 1979; Martinez Fernandez 1993; Rittichier 2000; Roberts 1999; Roorda 1998; Super 1989). Of the 29 studies (67%) that reported a change in croup score, 17 (59%) used a validated score (the Westley score or a modified Westley score) (Alshehr 2005; Amir 2006; Cetinkaya 2004; Chub‐Uppakarn 2007; Dobrovoljac 2012; Duman 2005; Fifoot 2007; Godden 1997; Husby 1993; Johnson 1996; Johnson 1998; Klassen 1994; Klassen 1996; Klassen 1998; Rittichier 2000; Roorda 1998; Super 1989); 11 (38%) used author‐created scales (Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 2005; Kuusela 1988; Leipzig 1979; Martinez Fernandez 1993; Massicotte 1973; Roberts 1999; Vad Pedersen 1998; Von Mühlendahl 1982); and one used the score by Downes 1975 (Eboriadou 2010). The studies by Bjornson 2004 and Garbutt 2013 used another validated score, the telephone outpatient (TOP) score, to measure clinical improvement. The TOP score is a two‐item, three‐point score used to assess the presence of stridor and barky cough by asking parents about their child's symptoms in the previous 24 hours (Bjornson 2016). Twenty‐six studies (60%) reported return visits or (re)admissions to the hospital or both (Alshehr 2005; Amir 2006; Bjornson 2004; Cruz 1995; Donaldson 2003; Duman 2005; Eboriadou 2010; Fifoot 2007; Fitzgerald 1996; Garbutt 2013; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 1996; Geelhoed 2005; Johnson 1996; Johnson 1998; Klassen 1994; Klassen 1996; Klassen 1998; Luria 2001; Rittichier 2000; Roberts 1999; Skowron 1966a; Skowron 1966a&b; Skowron 1966b; Soleimani 2013; Sparrow 2006; Vad Pedersen 1998).

Reported outcomes: secondary outcomes

A total of 12 studies (28%) reported length of stay in the hospital or emergency department (Alshehr 2005; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 2005; Godden 1997; Klassen 1998; Kuusela 1988; Leipzig 1979; Roorda 1998; Skowron 1966a; Skowron 1966a&b; Skowron 1966b; Sparrow 2006; Super 1989). Twelve studies (28%) reported patient improvement. Of these, one reported improvement after two hours (Roberts 1999); eight reported improvement after six hours (Eden 1964; Eden 1967; Johnson 1996; Klassen 1994; Klassen 1996; Klassen 1998; Massicotte 1973; Roberts 1999); six reported improvement after 12 hours (Eden 1964; Eden 1967; James 1969; Massicotte 1973; Roberts 1999; Super 1989); and seven reported improvement after 24 hours (Cruz 1995; Donaldson 2003; Eden 1964; Eden 1967; James 1969; Roberts 1999; Super 1989). About two‐thirds of the included studies (n = 29) reported the use of additional treatments. Of these, 11 reported intubation/tracheotomies (Chub‐Uppakarn 2007; Eden 1967; Fitzgerald 1996; Geelhoed 1995c; Godden 1997; James 1969; Johnson 1996; Johnson 1998; Leipzig 1979; Roorda 1998; Skowron 1966a; Skowron 1966a&b; Skowron 1966b); four reported the use of antibiotics (Husby 1993; James 1969; Koren 1983; Rittichier 2000); 13 reported the use of supplemental glucocorticoids (Dobrovoljac 2012; Fifoot 2007; Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Johnson 1996; Klassen 1994; Klassen 1996; Klassen 1998; Rittichier 2000; Roorda 1998; Super 1989; Vad Pedersen 1998); 20 reported the use of epinephrine (Amir 2006; Dobrovoljac 2012; Donaldson 2003; Duman 2005; Fifoot 2007; Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 2005; Godden 1997; Johnson 1996; Johnson 1998; Klassen 1994; Klassen 1996; Klassen 1998; Koren 1983; Rittichier 2000; Roberts 1999; Sparrow 2006; Super 1989; Tibballs 1992); and five reported the use of a mist tent (Alshehr 2005; Johnson 1996; Klassen 1996; Rittichier 2000; Super 1989). Twenty‐two studies reported collecting adverse events data. Of these, seven reported serious adverse events following the administration of glucocorticoids (namely secondary bacterial infections, e.g. pneumonia, otitis media) (Alshehr 2005; Bjornson 2004; Johnson 1996; Klassen 1998; Kuusela 1988; Roberts 1999; Super 1989), and 15 reported no serious adverse events (Chub‐Uppakarn 2007; Duman 2005; Eden 1967; Fifoot 2007; Fitzgerald 1996; Garbutt 2013; Husby 1993; James 1969; Johnson 1998; Klassen 1994; Leipzig 1979; Roorda 1998; Sparrow 2006; Tibballs 1992; Vad Pedersen 1998).

Excluded studies

One study was excluded following the 2014 interim update search. Faghihinia 2007 did not report any usable results. We excluded four studies following the searches in 2018 (Figure 1). Roked 2015 and Gill 2017 were not randomised trials; Eghbali 2016 and Mohammadzadeh 2014 were randomised trials that did not investigate glucocorticoids; and Faraji‐Goodarzi 2018 was a randomised trial that did not report any relevant outcomes. There were no relevant outcomes for an ongoing randomised trial (NCT01748162), and the trial was terminated due to inadequate recruitment (see Characteristics of excluded studies).

We edited the excluded studies list to remove legacy excluded studies that evidently did not meet the inclusion criteria (e.g. letters, commentaries, summaries, case studies). We made this change to comply with current Cochrane standards for methods and reporting. We excluded 33 studies in this 2018 updated review.

Ongoing studies

We identified one ongoing study (ACTRN12609000290291). We will assess this study for inclusion in a future update.

Risk of bias in included studies

We judged overall risk of bias to be low in Garbutt 2013, unclear in 30 studies (Alshehr 2005; Bjornson 2004; Chub‐Uppakarn 2007; Cruz 1995; Donaldson 2003; Eden 1964; Eden 1967; Fifoot 2007; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1996; Geelhoed 2005; Godden 1997; Husby 1993; James 1969; Johnson 1996; Johnson 1998; Klassen 1996; Klassen 1998; Koren 1983; Kuusela 1988; Leipzig 1979; Luria 2001; Martinez Fernandez 1993; Massicotte 1973; Roorda 1998; Skowron 1966a&b; Sparrow 2006; Super 1989; Tibballs 1992; Von Mühlendahl 1982), and high in 12 studies (Amir 2006; Cetinkaya 2004; Dobrovoljac 2012; Duman 2005; Eboriadou 2010; Fitzgerald 1996; Geelhoed 1995c; Klassen 1994; Rittichier 2000; Roberts 1999; Soleimani 2013; Vad Pedersen 1998). The 'Risk of bias' tables in Characteristics of included studies show the rationales for our 'Risk of bias' decisions. Figure 2 and Figure 3 show the 'Risk of bias' judgements for the included studies.

Risk of bias graph: Review authors' judgements of risk of bias for each domain and overall presented as percentages across all included studies.

Risk of bias summary: Review authors' judgements of risk of bias among seven domains and overall for each included study.

Allocation

We judged risk of bias for random sequence generation to be low in 26 studies (60%) and unclear in 17 studies (40%). The 16 studies at unclear risk of bias were described as randomised, however the method for generating the randomisation sequence was not clear or not reported (Cetinkaya 2004; Cruz 1995; Dobrovoljac 2012; Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 1996; Godden 1997; Husby 1993; James 1969; Koren 1983; Kuusela 1988; Martinez Fernandez 1993; Roorda 1998; Skowron 1966a&b; Soleimani 2013; Von Mühlendahl 1982). Randomisation was adequately described in the remaining 26 studies. We judged risk of bias for allocation concealment to be low in 18 studies (42%) and unclear in 25 studies (58%). For the 25 studies at unclear risk of bias, there was insufficient information reported in the publication to determine whether or not the groups to which the children were allocated could have been foreseen (Amir 2006; Cetinkaya 2004; Cruz 1995; Donaldson 2003; Duman 2005; Eboriadou 2010; Eden 1964; Eden 1967; Fifoot 2007; Fitzgerald 1996; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 1996; Geelhoed 2005; Godden 1997; Husby 1993; Koren 1983; Leipzig 1979; Rittichier 2000; Roorda 1998; Skowron 1966a&b; Soleimani 2013; Sparrow 2006; Vad Pedersen 1998; Von Mühlendahl 1982). Allocation concealment was adequate in the remaining 18 studies.

Blinding

We judged risk of bias for blinding of participants and personnel to be low in 25 (58%), unclear in 10 (23%), and high in eight (19%) studies. Of the eight studies at high risk of bias, four appeared to be open‐label (Amir 2006; Duman 2005; Rittichier 2000; Vad Pedersen 1998). Cetinkaya 2004 did not explicitly describe any measures taken to blind the participants and personnel from treatment assignment, and any blinding could have been broken. The personnel were not blinded in the study by Fitzgerald 1996. In Eboriadou 2010, the treatments were clearly distinguishable, and the method for blinding was not described even though the study was termed "double‐blind". In Soleimani 2013, only the outcome assessor was blinded. Of the 10 studies assessed as at unclear risk of bias, six were described as double‐blind without any further details about who was blinded and how (Eden 1964; Geelhoed 1996; Husby 1993; Leipzig 1979; Roorda 1998; Von Mühlendahl 1982). In Donaldson 2003, Geelhoed 1995a, Geelhoed 1995c, and Johnson 1998, blinding was attempted, but we judged that the blinding could have been broken; however, it was unclear how often this may have occurred. The remaining studies included satisfactory descriptions of how participants and personnel were blinded.

We judged risk of bias for blinding of outcome assessment to be low in 26 (60%), unclear in 12 (28%), and high in five (12%) studies. For 21 studies (49%), there was no mention of a third‐party outcome assessor, so the judgement for outcome assessment was carried over from blinding of participants and personnel (Cetinkaya 2004; Chub‐Uppakarn 2007; Cruz 1995; Dobrovoljac 2012; Duman 2005; Eboriadou 2010; Eden 1964; Eden 1967; Geelhoed 1995a; Geelhoed 1995b; Geelhoed 1995c; Geelhoed 1996; Geelhoed 2005; Godden 1997; Husby 1993; Koren 1983; Kuusela 1988; Luria 2001; Martinez Fernandez 1993; Massicotte 1973; Sparrow 2006; Tibballs 1992). Of the remaining studies, we judged two as at high risk of bias because the outcome assessors were not blinded (Amir 2006; Vad Pedersen 1998). We judged six studies as at unclear risk of bias. In Donaldson 2003, Johnson 1998, and Rittichier 2000, blinding of the outcome assessors was attempted, but we judged that the blinding could have been broken, although it was unclear how often this may have occurred. The studies by Leipzig 1979, Roorda 1998, and Von Mühlendahl 1982 were described as double‐blind, but it was unclear if the outcome assessors were blinded. In Soleimani 2013, the outcome assessor was described as blinded, but it was unclear how or if the blinding could have been broken. The remaining studies included satisfactory descriptions of how the outcome assessors were blinded.

Incomplete outcome data

We judged risk of bias for incomplete outcome data to be low in 27 (63%), unclear in 12 (28%), and high in four (9%) studies. The four studies at high risk of bias reported large losses to follow‐up that were imbalanced between groups (Dobrovoljac 2012; Geelhoed 1995c; Klassen 1994; Roberts 1999). Dobrovoljac 2012 and Roberts 1999 used the last observation carried forward method to estimate endpoint outcome values. For the studies at unclear risk of bias, in one study the number of children analysed was not reported (Amir 2006), and in six studies it was either unclear to which group the children who were lost to follow‐up had been allocated, or whether or not the losses to follow‐up were balanced between groups (Cruz 1995; Eden 1964; Johnson 1996; Kuusela 1988; Rittichier 2000; Soleimani 2013; Von Mühlendahl 1982). In four studies, losses to follow‐up ranged from 13% to 17% (Fifoot 2007; Luria 2001; Soleimani 2013; Super 1989). In Fitzgerald 1996, loss to follow‐up was 5%, and the last value observation forward method was used to estimate endpoint outcome values. We judged risk of bias due to incomplete outcome data not to be a concern for the remainder of the studies.

Selective reporting

We judged risk of bias for selective reporting to be low in two (5%) and unclear in 41 (95%) studies. For the two studies at low risk of bias (Fifoot 2007; Garbutt 2013), the outcomes in the trial registers matched those reported in the publications. For the remaining 41 studies, no protocol or trial registry was cited in the publication or located via online searches. In all cases, the outcomes reported in the methods matched those reported in the results section of the publications.

Other potential sources of bias

We judged risk of bias from other sources to be low in 34 (79%), unclear in seven (16%), and high in two (5%) studies. For the two studies at high risk of bias, there was a baseline imbalance in croup score (Amir 2006; Vad Pedersen 1998). For six of the studies at unclear risk of bias, there was potential for bias in participant selection because some children were not enrolled due to manpower constraints, failure of the emergency department to contact the research team, or because the emergency department was busy (Geelhoed 1995a; Geelhoed 1995b; Godden 1997; Klassen 1994; Klassen 1996; Klassen 1998; Sparrow 2006). For one study at unclear risk of bias, baseline data were not presented, therefore it was not possible to estimate whether or not baseline imbalances existed between the groups (Skowron 1966a&b).

Effects of interventions

See: Table 1; Table 2

Summary of findings for the main comparison. Any glucocorticoid compared to placebo for croup.

Any glucocorticoid compared to placebo for croup
Patient or population: children with croup  Intervention: any glucocorticoid  Comparison: placebo
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)	Comments**
	Placebo	Any glucocorticoid
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 2 hours)	The mean change in croup score was ‐1.50 to ‐0.81.	The mean change in croup score was 0.65 standard deviations more (1.13 more to 0.18 more).	‐	426  (7 RCTs)	⊕⊕⊕⊝  MODERATE^a	A standard deviation of 0.65 represents a moderate difference between groups.
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 6 hours)	The mean change in croup score was ‐3.23 to ‐0.65.	The mean change in croup score was 0.76 standard deviations more (1.12 more to 0.40 more).	‐	959  (11 RCTs)	⊕⊕⊕⊝  MODERATE^b	A standard deviation of 0.76 represents a large difference between groups.
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 12 hours)	The mean change in croup score was ‐7.62 to ‐1.00.	The mean change in croup score was 1.03 standard deviations more (1.53 more to 0.53 more).	‐	571  (8 RCTs)	⊕⊕⊕⊝  MODERATE^c	A standard deviation of 1.03 represents a large difference between groups.
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 24 hours)	The mean change in croup score was ‐2.56 to ‐1.05.	The mean change in croup score was 0.86 standard deviations more (1.40 more to 0.31 more).	‐	351  (8 RCTs)	⊕⊕⊝⊝  LOW^d	A standard deviation of 0.86 represents a large difference between groups.
Return visits or (re)admissions or both	Study population		RR 0.52  (0.36 to 0.75)	1679  (10 RCTs)	⊕⊕⊕⊝  MODERATE^e
	204 per 1000	106 per 1000  (74 to 153)
Adverse events	13/26 (50%) studies reported collecting adverse events data, and 8/13 (62%) reported no serious adverse events. Bjornson 2004 reported 7 instances of pneumonia (3/359, 0.83% in the dexamethasone group and 4/361, 1.11% in the placebo group). Johnson 1996 reported 1 child with neutropenia consistent with bacterial tracheitis in the dexamethasone group (1/28, 3.57%). Kuusela 1988 reported 7 secondary bacterial infections (pneumonia, sinusitis, otitis media) requiring antibiotic therapy: 5/35, 14% in the dexamethasone group and 2/16, 12.5% in the placebo group. Super 1989 reported 1 child with pneumonitis in the placebo group (1/13, 7.7%) and 2 children with pneumonia in the dexamethasone group (2/16, 12.5%). Roberts 1999 reported 1 instance of exacerbated symptoms, 5 children with emotional distress, 2 with vomiting, and 1 instance of eye irritation in the budesonide group (9/42, 21.4%) and 3 instances of exacerbated symptoms, 6 children with emotional distress, 3 with vomiting, 2 rashes, and 1 instance each of eye irritation and tongue irritation in the placebo group (16/40, 40%).
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  We used Cohen's interpretation of effect sizes to determine the magnitude of the difference between groups (0.2 represents a small effect, 0.5 represents a medium effect, 0.8 represents a large effect). CI: confidence interval; RCT: randomised controlled trial; RR:** risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Any glucocorticoid compared to epinephrine for croup
Patient or population: children with croup  Intervention: any glucocorticoid  Comparison: epinephrine
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)	Comments**
	Epinephrine	Any glucocorticoid
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 2 hours)	The mean change in croup score was ‐4.24 to ‐3.74.	The mean change in croup score was 0.77 standard deviations less (0.24 more to 1.77 less).	‐	130  (2 RCTs)	⊕⊕⊝⊝  LOW^{a, b}	A standard deviation of 0.77 represents a large difference between groups.
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 6 hours)	The mean change in croup score was ‐1.25 to ‐1.10.	The mean change in croup score was 0.10 standard deviations more (1.18 more to 0.97 less).	‐	63  (2 RCTs)	⊕⊝⊝⊝  VERY LOW^{c, d}	A standard deviation of 0.10 represents a minimal difference between groups.
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 12 hours)	The mean change in croup score was ‐3.86 to ‐1.45.	The mean change in croup score was 0.07 standard deviations more (0.57 more to 0.43 less).	‐	129  (3 RCTs)	⊕⊕⊕⊝  MODERATE^e	A standard deviation of 0.07 represents a minimal difference between groups.
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 24 hours)	The mean change in croup score was ‐4.40 to ‐2.01.	The mean change in croup score was 0.17 standard deviations less (0.18 more to 0.51 less).	‐	129  (3 RCTs)	⊕⊕⊕⊝  MODERATE^f	A standard deviation of 0.17 represents a small difference between groups.
Return visits or (re)admissions or both	Study population		RD 0.00  (‐0.04 to 0.04)	130  (2 RCTs)	⊕⊕⊕⊝  MODERATE^g
	0 per 1000	0 per 1000  (0 to 0)
Adverse events	3/4 (75%) studies reported collecting adverse events data. Fitzgerald 1996 reported no serious adverse events. Kuusela 1988 reported 5 cases of secondary bacterial infections (pneumonia, sinusitis, otitis media) requiring antibiotic therapy in the dexamethasone group (5/16, 31.3%). Eboriadou 2010 reported 4 cases of tremor and tachycardia (4/25, 16%) in the epinephrine group.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). We used Cohen's interpretation of effect sizes to determine the magnitude of the difference between groups (0.2 represents a small effect, 0.5 represents a medium effect, 0.8 represents a large effect).    CI: confidence interval; RCT: randomised controlled trial; RD:** risk difference
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Baseline rate (%)	NNTB (95% CI)
Mean baseline rate
30.62	7 (5 to 12)
Smallest baseline rate
2.06	102 (78 to 179)
Largest baseline rate
72.00	3 (2 to 5)

Dexamethasone compared to budesonide for croup
Patient or population: children with croup  Intervention: dexamethasone  Comparison: budesonide
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)	Comments**
	Budesonide	Dexamethasone
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 6 hours)	The mean change in croup score was ‐2.93 to ‐2.00.	The mean change in croup score was 0.46 standard deviations more (0.79 more to 0.13 more).	‐	326  (4 RCTs)	⊕⊕⊝⊝  LOW^{a, b}	A standard deviation of 0.46 represents a moderate difference between groups.
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 12 hours)	The mean change in croup score was ‐3.07 to ‐2.33.	The mean change in croup score was 0.75 standard deviations more (1.19 more to 0.30 more).	‐	84  (2 RCTs)	⊕⊕⊝⊝  LOW^{c, d}	A standard deviation of 0.75 represents a large difference between groups.
Return visits or (re)admissions or both	Study population		RR 0.69  (0.40 to 1.22)	374  (5 RCTs)	⊕⊕⊕⊝  MODERATE^e
	122 per 1000	84 per 1000  (49 to 149)
Adverse events	4/6 (67%) studies reported collecting adverse events data, and 3/4 (75%) studies reported no serious adverse events (Duman 2005; Johnson 1998; Vad Pedersen 1998). Klassen 1998 reported 1 case of oral thrush in the budesonide group (1/65, 1.5%), and 1 case of hives and 1 case of violent behaviour in the dexamethasone group (2/69, 2.9%).
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  We used Cohen's interpretation of effect sizes to determine the magnitude of the difference between groups (0.2 represents a small effect, 0.5 represents a medium effect, 0.8 represents a large effect). CI: confidence interval; RCT: randomised controlled trial; RR:** risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Dexamethasone compared to beclomethasone for croup
Patient or population: children with croup  Intervention: dexamethasone  Comparison: beclomethasone
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Beclomethasone	Dexamethasone
Return visits or (re)admissions or both	Study population		RD 0.00  (‐0.09 to 0.09)	39  (1 RCT)	⊕⊕⊝⊝  LOW^{a, b}
	0 per 1000	0 per 1000  (0 to 0)
Adverse events	Eboriadou 2010 reported no adverse events related to the glucocorticoids.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; RCT: randomised controlled trial; RD: risk difference
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Dexamethasone compared to betamethasone for croup
Patient or population: children with croup  Intervention: dexamethasone  Comparison: betamethasone
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Betamethasone	Dexamethasone
Change in croup score. Assessed with the Westley croup score. Lower scores mean fewer symptoms. (follow‐up: 2 hours)	The mean change in croup score from 1 study was ‐1.68.	The mean change in croup score was 1.38 units more (2.58 more to 0.18 more).	‐	52  (1 RCT)	⊕⊝⊝⊝  VERY LOW^{a, b, c}
Change in croup score. Assessed with the Westley croup score. Lower scores mean fewer symptoms. (follow‐up: 6 hours)	The mean change in croup score from 1 study was ‐1.89.	The mean change in croup score was 1.53 units more (2.75 more to 0.31 more).	‐	52  (1 RCT)	⊕⊝⊝⊝  VERY LOW^{d, e, f}
Return visits or (re)admissions or both	Study population		RR 0.95  (0.67 to 1.34)	52  (1 RCT)	⊕⊝⊝⊝  VERY LOW^{g, h}
	731 per 1000	694 per 1000  (490 to 979)
Adverse events	Amir 2006 did not report collecting adverse events data.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Dexamethasone compared to prednisolone for croup
Patient or population: children with croup  Intervention: dexamethasone  Comparison: prednisolone
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Prednisolone	Dexamethasone
Change in croup score. Assessed with the Westley croup score. Lower scores mean fewer symptoms. (follow‐up: 6 hours)	The mean change in croup score from 1 study was ‐2.35.	The mean change in croup score was 0.19 units less (0.17 more to 0.55 less).	‐	99  (1 RCT)	⊕⊕⊝⊝  LOW^{a, b}
Return visits or (re)admissions or both	Study population		RR 0.39  (0.19 to 0.79)	306  (3 RCTs)	⊕⊕⊕⊝  MODERATE^c
	200 per 1000	78 per 1000  (38 to 158)
Adverse events	Fifoot 2007, Garbutt 2013, and Sparrow 2006 reported no serious adverse events related to the glucocorticoids.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Budesonide and dexamethasone compared to dexamethasone for croup
Patient or population: children with croup   Intervention: budesonide and dexamethasone  Comparison: dexamethasone
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)	Comments**
	Dexamethasone	Budesonide and dexamethasone
Change in croup score. Assessed with different scores in different studies. Lower scores mean fewer symptoms. (follow‐up: 6 hours)	The mean change in croup score was ‐3.24 to ‐1.80.	The mean change in croup score was 0.05 standard deviations less (0.19 more to 0.30 less).	‐	255  (3 RCTs)	⊕⊕⊕⊝  MODERATE^a	A standard deviation of 0.05 represents a minimal difference between groups.
Return visits or (re)admissions or both	Study population		RR 0.91  (0.45 to 1.83)	254  (3 RCTs)	⊕⊕⊝⊝  LOW^b
	100 per 1000	91 per 1000  (45 to 183)
Adverse events	1/3 (33%) studies reported collecting adverse events data. Klassen 1998 reported no adverse events in the dexamethasone or dexamethasone and budesonide group.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  We used Cohen's interpretation of effect sizes to determine the magnitude of the difference between groups (0.2 represents a small effect, 0.5 represents a medium effect, 0.8 represents a large effect). CI: confidence interval; RCT: randomised controlled trial; RR:** risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Oral dexamethasone compared to intramuscular dexamethasone for croup
Patient or population: children with croup  Intervention: oral dexamethasone  Comparison: intramuscular dexamethasone
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Intramuscular dexamethasone	Oral dexamethasone
Return visits or (re)admissions or both	Study population		RR 0.81  (0.58 to 1.12)	440  (3 RCTs)	⊕⊕⊕⊝  MODERATE^a
	259 per 1000	210 per 1000  (150 to 290)
Adverse events	None of the studies reported collecting adverse events data.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Oral dexamethasone compared to nebulised dexamethasone for croup
Patient or population: children with croup  Intervention: oral dexamethasone  Comparison: nebulised dexamethasone
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Nebulised dexamethasone	Oral dexamethasone
Return visits or (re)admissions or both	Study population		RR 0.39  (0.17 to 0.89)	176  (1 RCT)	⊕⊕⊝⊝  LOW^{a, b}
	209 per 1000	81 per 1000  (35 to 186)
Adverse events	None of the studies reported collecting adverse events data.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Dexamethasone 0.30 mg per kg compared to dexamethasone 0.15 mg per kg for croup
Patient or population: children with croup  Intervention: dexamethasone 0.30 mg per kg  Comparison: dexamethasone 0.15 mg per kg
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Dexamethasone 0.15 mg per kg	Dexamethasone 0.30 mg per kg
Return visits or (re)admissions or both	Study population		RR 0.94  (0.06 to 14.27)	60  (1 RCT)	⊕⊝⊝⊝  VERY LOW^{a, b}
	34 per 1000	32 per 1000  (2 to 492)
Adverse events	Geelhoed 1995b did not report collecting adverse events data.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup score (change baseline ‐ 2 hours) by score	7	426	Std. Mean Difference (IV, Random, 95% CI)	‐0.65 [‐1.13, ‐0.18]
1.1 Westley score	5	264	Std. Mean Difference (IV, Random, 95% CI)	‐0.72 [‐1.44, 0.01]
1.2 Non‐Westley score	2	162	Std. Mean Difference (IV, Random, 95% CI)	‐0.51 [‐0.93, ‐0.10]
2 Croup score (change baseline ‐ 6 hours) by score	11	959	Std. Mean Difference (IV, Random, 95% CI)	‐0.76 [‐1.12, ‐0.40]
2.1 Westley score	5	336	Std. Mean Difference (IV, Random, 95% CI)	‐0.79 [‐1.02, ‐0.56]
2.2 Non‐Westley score	6	623	Std. Mean Difference (IV, Random, 95% CI)	‐0.81 [‐1.43, ‐0.18]
3 Croup score (change baseline ‐ 12 hours) by score	8	571	Std. Mean Difference (IV, Random, 95% CI)	‐1.03 [‐1.53, ‐0.53]
3.1 Westley score	2	113	Std. Mean Difference (IV, Random, 95% CI)	‐1.54 [‐2.56, ‐0.53]
3.2 Non‐Westley score	6	458	Std. Mean Difference (IV, Random, 95% CI)	‐0.87 [‐1.45, ‐0.30]
4 Croup score (change baseline ‐ 24 hours) by score	8	351	Std. Mean Difference (IV, Random, 95% CI)	‐0.86 [‐1.40, ‐0.31]
4.1 Westley score	4	169	Std. Mean Difference (IV, Random, 95% CI)	‐1.05 [‐1.72, ‐0.37]
4.2 Non‐Westley score	4	182	Std. Mean Difference (IV, Random, 95% CI)	‐0.70 [‐1.56, 0.16]
5 Croup score (change baseline ‐ 2 hours) by inpatient/outpatient	7	426	Std. Mean Difference (IV, Random, 95% CI)	‐0.65 [‐1.13, ‐0.18]
5.1 Inpatient	5	301	Std. Mean Difference (IV, Random, 95% CI)	‐0.80 [‐1.44, ‐0.16]
5.2 Outpatient	2	125	Std. Mean Difference (IV, Random, 95% CI)	‐0.32 [‐0.93, 0.29]
6 Croup score (change baseline ‐ 6 hours) by inpatient/outpatient	11	959	Std. Mean Difference (IV, Random, 95% CI)	‐0.76 [‐1.12, ‐0.40]
6.1 Inpatient	8	723	Std. Mean Difference (IV, Random, 95% CI)	‐0.72 [‐1.22, ‐0.23]
6.2 Outpatient	3	236	Std. Mean Difference (IV, Random, 95% CI)	‐0.84 [‐1.11, ‐0.56]
7 Croup score (change baseline ‐ 24 hours) by inpatient/outpatient	8	351	Std. Mean Difference (IV, Random, 95% CI)	‐0.86 [‐1.40, ‐0.31]
7.1 Inpatient	7	291	Std. Mean Difference (IV, Random, 95% CI)	‐0.82 [‐1.46, ‐0.19]
7.2 Outpatient	1	60	Std. Mean Difference (IV, Random, 95% CI)	‐1.09 [‐1.71, ‐0.48]
8 Croup score (change baseline ‐ 2 hours) by glucocorticoid	7	426	Std. Mean Difference (IV, Random, 95% CI)	‐0.66 [‐1.10, ‐0.22]
8.1 Budesonide	4	246	Std. Mean Difference (IV, Random, 95% CI)	‐1.01 [‐1.71, ‐0.30]
8.2 Dexamethasone	3	163	Std. Mean Difference (IV, Random, 95% CI)	‐0.49 [‐1.00, 0.03]
8.3 Fluticasone	1	17	Std. Mean Difference (IV, Random, 95% CI)	0.45 [‐0.52, 1.42]
9 Croup score (change baseline ‐ 6 hours) by glucocorticoid	11	959	Std. Mean Difference (IV, Random, 95% CI)	‐0.74 [‐1.07, ‐0.41]
9.1 Budesonide	5	333	Std. Mean Difference (IV, Random, 95% CI)	‐0.81 [‐1.04, ‐0.58]
9.2 Dexamethasone	6	567	Std. Mean Difference (IV, Random, 95% CI)	‐0.62 [‐1.17, ‐0.08]
9.3 Fluticasone	1	17	Std. Mean Difference (IV, Random, 95% CI)	0.06 [‐0.89, 1.02]
9.4 Prednisolone	1	42	Std. Mean Difference (IV, Random, 95% CI)	‐1.87 [‐2.62, ‐1.13]
10 Croup score (change baseline ‐ 12 hours) by glucocorticoid	8	571	Std. Mean Difference (IV, Random, 95% CI)	‐1.04 [‐1.51, ‐0.56]
10.1 Budesonide	3	209	Std. Mean Difference (IV, Random, 95% CI)	‐0.97 [‐1.26, ‐0.68]
10.2 Dexamethasone	5	323	Std. Mean Difference (IV, Random, 95% CI)	‐0.85 [‐1.55, ‐0.15]
10.3 Prednisolone	1	39	Std. Mean Difference (IV, Random, 95% CI)	‐2.40 [‐3.26, ‐1.55]
11 Croup score (change baseline ‐ 24 hours) by glucocorticoid	8	351	Std. Mean Difference (IV, Random, 95% CI)	‐0.89 [‐1.41, ‐0.37]
11.1 Budesonide	2	89	Std. Mean Difference (IV, Random, 95% CI)	‐1.40 [‐1.88, ‐0.93]
11.2 Dexamethasone	6	245	Std. Mean Difference (IV, Random, 95% CI)	‐0.89 [‐1.55, ‐0.22]
11.3 Fluticasone	1	17	Std. Mean Difference (IV, Random, 95% CI)	0.21 [‐0.75, 1.17]
12 Return visits or (re)admissions or both by inpatient/outpatient	10	1679	Risk Ratio (M‐H, Random, 95% CI)	0.52 [0.36, 0.75]
12.1 Inpatient	3	323	Risk Ratio (M‐H, Random, 95% CI)	0.39 [0.12, 1.30]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup score (change baseline ‐ 2 hours) by inpatient/outpatient	2	130	Std. Mean Difference (IV, Random, 95% CI)	0.77 [‐0.24, 1.77]
1.1 Inpatient	1	66	Std. Mean Difference (IV, Random, 95% CI)	0.26 [‐0.22, 0.75]
1.2 Outpatient	1	64	Std. Mean Difference (IV, Random, 95% CI)	1.29 [0.73, 1.84]
2 Croup score (change baseline ‐ 6 hours) by inpatient/outpatient	2	63	Std. Mean Difference (IV, Random, 95% CI)	‐0.10 [‐1.18, 0.97]
2.1 Inpatient	2	63	Std. Mean Difference (IV, Random, 95% CI)	‐0.10 [‐1.18, 0.97]
3 Croup score (change baseline ‐ 12 hours) by inpatient/outpatient	3	129	Std. Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.57, 0.43]
3.1 Inpatient	3	129	Std. Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.57, 0.43]
4 Croup score (change baseline ‐ 24 hours) by inpatient/outpatient	3	129	Std. Mean Difference (IV, Random, 95% CI)	0.17 [‐0.18, 0.51]
4.1 Inpatient	3	129	Std. Mean Difference (IV, Random, 95% CI)	0.17 [‐0.18, 0.51]
5 Croup score (change baseline ‐ 2 hours) by glucocorticoid	2	130	Std. Mean Difference (IV, Random, 95% CI)	0.88 [0.13, 1.63]
5.1 Budesonide	1	66	Std. Mean Difference (IV, Random, 95% CI)	0.26 [‐0.22, 0.75]
5.2 Dexamethasone	1	31	Std. Mean Difference (IV, Random, 95% CI)	1.13 [0.35, 1.91]
5.3 Beclomethasone	1	33	Std. Mean Difference (IV, Random, 95% CI)	1.41 [0.62, 2.19]
6 Croup score (change baseline ‐ 12 hours) by glucocorticoid	3	129	Std. Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.57, 0.43]
6.1 Budesonide	1	66	Std. Mean Difference (IV, Random, 95% CI)	0.02 [‐0.47, 0.50]
6.2 Dexamethasone	2	63	Std. Mean Difference (IV, Random, 95% CI)	‐0.14 [‐1.09, 0.82]
7 Croup score (change baseline ‐ 24 hours) by glucocorticoid	3	129	Std. Mean Difference (IV, Random, 95% CI)	0.17 [‐0.18, 0.51]
7.1 Budesonide	1	66	Std. Mean Difference (IV, Random, 95% CI)	0.21 [‐0.27, 0.70]
7.2 Dexamethasone	2	63	Std. Mean Difference (IV, Random, 95% CI)	0.12 [‐0.38, 0.61]
8 Return visits or (re)admissions or both by inpatient/outpatient	2	130	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.04, 0.04]
8.1 Inpatient	1	66	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.06, 0.06]
8.2 Outpatient	1	64	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.06, 0.06]
9 Length of stay by inpatient/outpatient	1	32	Mean Difference (IV, Random, 95% CI)	‐10.0 [‐33.89, 13.89]
9.1 Inpatient	1	32	Mean Difference (IV, Random, 95% CI)	‐10.0 [‐33.89, 13.89]
10 Additional treatments: use of epinephrine	1	66	Risk Ratio (M‐H, Random, 95% CI)	0.30 [0.03, 2.69]
11 Additional treatments: intubation/tracheostomy	1	66	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.06, 0.06]
12 Additional treatments: supplemental glucocorticoids	1	66	Risk Ratio (M‐H, Random, 95% CI)	0.83 [0.48, 1.43]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup score (change baseline ‐ 6 hours) by inpatient/outpatient	4	326	Std. Mean Difference (IV, Random, 95% CI)	‐0.46 [‐0.79, ‐0.13]
1.1 Inpatient	2	97	Std. Mean Difference (IV, Random, 95% CI)	‐0.63 [‐1.04, ‐0.22]
1.2 Outpatient	2	229	Std. Mean Difference (IV, Random, 95% CI)	‐0.36 [‐0.90, 0.18]
2 Croup score (change baseline ‐ 12 hours) by inpatient/outpatient	2	84	Std. Mean Difference (IV, Random, 95% CI)	‐0.75 [‐1.19, ‐0.30]
2.1 Inpatient	2	84	Std. Mean Difference (IV, Random, 95% CI)	‐0.75 [‐1.19, ‐0.30]
3 Return visits or (re)admissions or both by inpatient/outpatient	5	374	Risk Ratio (M‐H, Random, 95% CI)	0.69 [0.40, 1.22]
3.1 Inpatient	2	95	Risk Ratio (M‐H, Random, 95% CI)	0.64 [0.14, 2.79]
3.2 Outpatient	3	279	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.38, 1.30]
4 Length of stay by inpatient/outpatient	2	184	Std. Mean Difference (IV, Random, 95% CI)	‐0.29 [‐0.72, 0.14]
4.1 Inpatient	1	50	Std. Mean Difference (IV, Random, 95% CI)	‐0.57 [‐1.14, ‐0.00]
4.2 Outpatient	1	134	Std. Mean Difference (IV, Random, 95% CI)	‐0.12 [‐0.46, 0.22]
5 Improvement (at 6 hours) by inpatient/outpatient	1	134	Risk Ratio (M‐H, Random, 95% CI)	1.12 [0.93, 1.34]
5.1 Outpatient	1	134	Risk Ratio (M‐H, Random, 95% CI)	1.12 [0.93, 1.34]
6 Additional treatments: epinephrine	4	321	Risk Ratio (M‐H, Random, 95% CI)	0.45 [0.21, 0.96]
7 Additional treatments: intubation/tracheostomy	2	145	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.04, 0.04]
8 Additional treatments: supplemental glucocorticoids	3	240	Risk Ratio (M‐H, Random, 95% CI)	0.48 [0.18, 1.32]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Return visits or (re)admissions or both by inpatient/outpatient	1	39	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.09, 0.09]
1.1 Outpatient	1	39	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.09, 0.09]

Dexamethasone 0.60 mg per kg compared to dexamethasone 0.30 mg per kg for croup
Patient or population: children with croup  Intervention: dexamethasone 0.60 mg per kg  Comparison: dexamethasone 0.30 mg per kg
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Dexamethasone 0.30 mg per kg	Dexamethasone 0.60 mg per kg
Return visits or (re)admissions or both	Study population		RR 1.40  (0.25 to 7.81)	60  (1 RCT)	⊕⊝⊝⊝  VERY LOW^{a, b}
	69 per 1000	97 per 1000  (17 to 539)
Adverse events	Geelhoed 1995a did not report collecting adverse events data.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence  High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Date	Event	Description
3 April 2018	New search has been performed	New authors joined the team to update the review. We updated the searches and included five new trials. Three were newly identified trials (Dobrovoljac 2012; Garbutt 2013; Soleimani 2013); one was previously excluded (Chub‐Uppakarn 2007); and one was previously awaiting classification (Eboriadou 2010). We excluded six new trials, Eghbali 2016; Faghihinia 2007; Faraji‐Goodarzi 2018; Gill 2017; Mohammadzadeh 2014; Roked 2015, and one ongoing trial (NCT01748162). We included one new ongoing trial (ACTRN12609000290291). In this update we assessed the risk of bias and the certainty of the evidence. We added two new primary outcomes: change in croup score after two hours, and patient improvement after two hours. We added adverse events as a secondary outcome and 'Summary of findings' tables. We added two new comparisons: oral compared to nebulised dexamethasone, and dexamethasone compared to beclomethasone.
3 April 2018	New citation required and conclusions have changed	Our conclusions have changed. The previous version of this review concluded that glucocorticoids, as compared to placebo, reduce croup symptoms within six hours and that the effect lasts 12 hours. We conclude that glucocorticoids, as compared to placebo, reduce croup symptoms within two hours and that the effect lasts at least 24 hours.

Date	Event	Description
16 September 2014	New citation required but conclusions have not changed	Review was updated, and conclusions remain unchanged.
16 September 2014	New search has been performed	Searches updated. We included one new trial, Dobrovoljac 2012, and excluded one new trial (Faghihinia 2007). We added a two‐hour croup score and a two‐hour improvement outcome.
1 December 2011	Amended	Grammatical correction made to the Plain language summary.
18 July 2011	Amended	Analysis 5.2 contained an error, as the negative signs for the change in croup scores at six hours were not included. The mean difference remains non‐significant.
23 July 2010	New search has been performed	Searches conducted. We added seven new trials since the 2004 publication (Alshehr 2005; Amir 2006; Cetinkaya 2004; Duman 2005; Fifoot 2007; Geelhoed 2005; Sparrow 2006). We excluded three new trials (Chub‐Uppakarn 2007; Custer 2005; Schooff 2005).
20 May 2010	New citation required but conclusions have not changed	New authors joined the team to update the review. The conclusions remain unchanged.
16 August 2008	Amended	Converted to new review format
3 February 2004	Feedback has been incorporated	Feedback incorporated.
7 April 2003	New search has been performed	Searches conducted.
17 August 1997	New search has been performed	Searches conducted.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup score (change baseline ‐ 2 hours) by inpatient/outpatient	1	52	Mean Difference (IV, Random, 95% CI)	‐1.38 [‐2.58, ‐0.18]
1.1 Outpatient	1	52	Mean Difference (IV, Random, 95% CI)	‐1.38 [‐2.58, ‐0.18]
2 Croup score (change baseline ‐ 6 hours) by inpatient/outpatient	1	52	Mean Difference (IV, Random, 95% CI)	‐1.53 [‐2.75, ‐0.31]
2.1 Outpatient	1	52	Mean Difference (IV, Random, 95% CI)	‐1.53 [‐2.75, ‐0.31]
3 Return visits or (re)admissions or both by inpatient/outpatient	1	52	Risk Ratio (M‐H, Random, 95% CI)	0.95 [0.67, 1.34]
3.1 Outpatient	1	52	Risk Ratio (M‐H, Random, 95% CI)	0.95 [0.67, 1.34]
4 Additional treatments: use of epinephrine	1	52	Risk Ratio (M‐H, Random, 95% CI)	2.11 [1.18, 3.76]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup (change baseline ‐ 6 hours) by inpatient/outpatient	1	99	Mean Difference (IV, Random, 95% CI)	0.19 [‐0.17, 0.55]
1.1 Outpatient	1	99	Mean Difference (IV, Random, 95% CI)	0.19 [‐0.17, 0.55]
2 Return visits or (re)admissions or both by inpatient/outpatient	3	306	Risk Ratio (M‐H, Random, 95% CI)	0.39 [0.19, 0.79]
2.1 Outpatient	3	306	Risk Ratio (M‐H, Random, 95% CI)	0.39 [0.19, 0.79]
3 Length of stay by inpatient/outpatient	1	133	Mean Difference (IV, Random, 95% CI)	0.5 [‐0.55, 1.55]
3.1 Outpatients	1	133	Mean Difference (IV, Random, 95% CI)	0.5 [‐0.55, 1.55]
4 Additional treatments: epinephrine	2	232	Risk Ratio (M‐H, Random, 95% CI)	0.69 [0.26, 1.85]
5 Additional treatments: supplemental glucocorticoids	1	86	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.34, 2.48]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup score (change baseline ‐ 6 hours) by inpatient/outpatient	3	255	Std. Mean Difference (IV, Random, 95% CI)	0.05 [‐0.19, 0.30]
1.1 Inpatient	1	72	Std. Mean Difference (IV, Random, 95% CI)	0.16 [‐0.30, 0.63]
1.2 Outpatient	2	183	Std. Mean Difference (IV, Random, 95% CI)	0.03 [‐0.32, 0.39]
2 Return visits or (re)admissions or both by inpatient/outpatient	3	254	Risk Ratio (M‐H, Random, 95% CI)	0.91 [0.45, 1.83]
2.1 Inpatient	1	71	Risk Ratio (M‐H, Random, 95% CI)	1.03 [0.46, 2.29]
2.2 Outpatient	2	183	Risk Ratio (M‐H, Random, 95% CI)	0.58 [0.13, 2.60]
3 Length of stay by inpatient/outpatient	2	204	Mean Difference (IV, Random, 95% CI)	0.44 [‐0.05, 0.92]
3.1 Inpatient	1	71	Mean Difference (IV, Random, 95% CI)	‐1.30 [‐6.75, 4.15]
3.2 Outpatient	1	133	Mean Difference (IV, Random, 95% CI)	0.45 [‐0.04, 0.94]
4 Improvement (at 6 hours) by inpatient/outpatient	2	183	Risk Ratio (M‐H, Random, 95% CI)	1.11 [0.65, 1.90]
4.1 Outpatient	2	183	Risk Ratio (M‐H, Random, 95% CI)	1.11 [0.65, 1.90]
5 Additional treatments: epinephrine	2	183	Risk Ratio (M‐H, Random, 95% CI)	1.42 [0.27, 7.39]
6 Additional treatments: mist tent	1	50	Risk Ratio (M‐H, Random, 95% CI)	1.07 [0.69, 1.65]
7 Additional treatments: supplemental glucocorticoids	2	182	Risk Ratio (M‐H, Random, 95% CI)	1.10 [0.07, 16.66]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup score (change baseline ‐ 6 hours) by inpatient/outpatient	1	129	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.59, 0.19]
1.1 Outpatient	1	129	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.59, 0.19]
2 Return visits or (re)admissions or both by inpatient/outpatient	1	129	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.03, 0.03]
2.1 Outpatient	1	129	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.03, 0.03]
3 Length of stay by inpatient/outpatient	1	129	Mean Difference (IV, Random, 95% CI)	0.25 [‐0.36, 0.86]
3.1 Outpatient	1	129	Mean Difference (IV, Random, 95% CI)	0.25 [‐0.36, 0.86]
4 Improvement (at 6 hours) by inpatient/outpatient	1	129	Risk Ratio (M‐H, Random, 95% CI)	0.97 [0.79, 1.20]
4.1 Outpatient	1	129	Risk Ratio (M‐H, Random, 95% CI)	0.97 [0.79, 1.20]
5 Additional treatments: epinephrine	1	129	Risk Ratio (M‐H, Random, 95% CI)	1.02 [0.15, 6.99]
6 Additional treatments: supplemental glucocorticoids	1	129	Risk Ratio (M‐H, Random, 95% CI)	1.31 [0.52, 3.29]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Return visits or (re)admissions or both by inpatient/outpatient	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Outpatient	3	440	Risk Ratio (M‐H, Random, 95% CI)	0.81 [0.58, 1.12]
2 Improvement (at 24 hours) by inpatient/outpatient	1	95	Risk Ratio (M‐H, Random, 95% CI)	1.07 [0.95, 1.19]
2.1 Outpatient	1	95	Risk Ratio (M‐H, Random, 95% CI)	1.07 [0.95, 1.19]
3 Additional treatments: antibiotics	1	277	Risk Ratio (M‐H, Random, 95% CI)	0.14 [0.02, 1.15]
4 Additional treatments: epinephrine	2	372	Risk Ratio (M‐H, Random, 95% CI)	0.94 [0.71, 1.24]
5 Additional treatments: mist tent	1	277	Risk Ratio (M‐H, Random, 95% CI)	1.34 [0.31, 5.89]
6 Additional treatments: supplemental glucocorticoids	1	277	Risk Ratio (M‐H, Random, 95% CI)	1.10 [0.50, 2.41]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Return visits or (re)admissions or both by inpatient/outpatient	1	176	Risk Ratio (M‐H, Random, 95% CI)	0.39 [0.17, 0.89]
1.1 Outpatient	1	176	Risk Ratio (M‐H, Random, 95% CI)	0.39 [0.17, 0.89]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Return visits or (re)admissions or both by inpatient/outpatient	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.94 [0.06, 14.27]
1.1 Outpatient	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.94 [0.06, 14.27]
2 Additional treatments: epinephrine	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.43 [0.19, 0.98]
3 Additional treatments: supplemental glucocorticoids	1	60	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.06, 0.06]

PERMALINK

Glucocorticoids for croup in children

Allison Gates

Michelle Gates

Ben Vandermeer

Cydney Johnson

Lisa Hartling

David W Johnson

Terry P Klassen

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

GRADE and 'Summary of findings' tables

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Participant and trial characteristics

Characteristics of the comparisons

Reported outcomes: primary outcomes

Reported outcomes: secondary outcomes

Excluded studies

Ongoing studies

Risk of bias in included studies

2.

3.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Summary of findings for the main comparison. Any glucocorticoid compared to placebo for croup.

Summary of findings 2. Any glucocorticoid compared to epinephrine for croup.

Any glucocorticoid compared to placebo

Primary outcomes

1. Change in clinical croup score

1.1. Analysis.

1.2. Analysis.

1.3. Analysis.

1.4. Analysis.

1.5. Analysis.

1.6. Analysis.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Return visits or (re)admissions or both by inpatient/outpatient	1	60	Risk Ratio (M‐H, Random, 95% CI)	1.40 [0.25, 7.81]
1.1 Outpatient	1	60	Risk Ratio (M‐H, Random, 95% CI)	1.40 [0.25, 7.81]
2 Additional treatments: epinephrine	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.78 [0.27, 2.28]
3 Additional treatments: supplemental glucocorticoids	1	60	Risk Ratio (M‐H, Random, 95% CI)	2.81 [0.12, 66.40]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Croup score (change baseline ‐ 2 hours) by inpatient/outpatient	1	41	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐0.29, ‐0.01]
1.1 Inpatient	1	41	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐0.29, ‐0.01]
2 Croup score (change baseline ‐ 6 hours) by inpatient/outpatient	3	178	Mean Difference (IV, Random, 95% CI)	‐0.33 [‐0.50, ‐0.16]
2.1 Inpatient	1	41	Mean Difference (IV, Random, 95% CI)	‐0.38 [‐0.54, ‐0.22]
2.2 Outpatient	2	137	Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.50, 0.35]
3 Croup score (change baseline ‐ 12 hours) by inpatient/outpatient	2	113	Mean Difference (IV, Random, 95% CI)	0.17 [‐1.45, 1.78]
3.1 Inpatient	1	41	Mean Difference (IV, Random, 95% CI)	‐0.65 [‐0.80, ‐0.50]
3.2 Outpatient	1	72	Mean Difference (IV, Random, 95% CI)	1.0 [0.65, 1.35]
4 Croup score (change baseline ‐ 24 hours) by inpatient/outpatient	1	72	Mean Difference (IV, Random, 95% CI)	0.5 [0.14, 0.86]
4.1 Outpatient	1	72	Mean Difference (IV, Random, 95% CI)	0.5 [0.14, 0.86]
5 Return visits or (re)admissions or both by inpatient/outpatient	2	129	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.54, 1.55]
5.1 Outpatient	2	129	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.54, 1.55]
6 Length of stay by inpatient/outpatient	1	72	Mean Difference (IV, Random, 95% CI)	2.0 [‐2.16, 6.16]
6.1 Outpatient	1	72	Mean Difference (IV, Random, 95% CI)	2.0 [‐2.16, 6.16]
7 Additional treatments: epinephrine	1	65	Risk Ratio (M‐H, Random, 95% CI)	1.10 [0.07, 16.80]
8 Additional treatments: intubation/tracheotomy	1	41	Risk Difference (M‐H, Random, 95% CI)	0.0 [‐0.09, 0.09]
9 Additional treatments: mist tent	1	72	Risk Ratio (M‐H, Random, 95% CI)	1.13 [0.69, 1.84]
10 Additional treatments: supplemental glucocorticoids	1	57	Risk Ratio (M‐H, Random, 95% CI)	0.89 [0.27, 2.97]

Methods	Randomised double‐blind trial
Participants	Study period: September 1998 to December 2002 Setting: emergency rooms and outpatient clinics in 3 medical institutes, Abha City, Saudi Arabia Inclusion criteria: children aged 3 months to 9 years who had been given a diagnosis of croup and had persistent, moderately severe respiratory distress (Westley croup score > 3) Exclusion criteria: symptoms or signs suggesting another cause of stridor; history of chronic pulmonary disease; severe systemic disease; immune dysfunction; stridor or intubation for more than 1 month; glucocorticoid therapy in the last 4 weeks before study entry Baseline characteristics (N = 72): proportion male: treatment: 56%; comparator: 53% mean (SD) age, months: treatment: 16.8 (12); comparator: 17.6 (13) median (range) Westley croup score: treatment: 5.0 (3 to 6); comparator: 4.5 (3 to 6)
Interventions	Treatment (N = 36): single dose 0.15 mg/kg oral dexamethasone Comparator (N = 36): single dose 0.6 mg/kg oral dexamethasone
Outcomes	Change in Westley croup score from baseline to 4, 12, and 24 hours; hospitalisation; length of stay in hospital; use of mist tent
Notes	All children received mist therapy throughout the observation period; funding source: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A blocked randomization code was produced by random‐number generating software"
Allocation concealment (selection bias)	Low risk	Quote: "To make the study drugs indistinguishable from each other, they were packaged in opaque containers and diluted on the same amount of solution." "A blocked randomization code was produced ... and the code was not broken until after the study ended and all decisions regarding data analysis were finalized"
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "double‐blind" "To make the study drugs indistinguishable from each other, they were packaged in opaque containers and diluted on the same amount of solution."
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "double‐blind" "to make the study drugs indistinguishable from each other, they were packaged in opaque containers and diluted on the same amount of solution" "the code was not broken until after the study ended and all the decisions regarding data analysis were finalized"
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: 14% (N = 12) of children recruited were excluded prior to randomisation. All randomised children were followed up.
Selective reporting (reporting bias)	Unclear risk	Comment: no protocol identified. All prespecified outcomes from methods appear in results.
Other bias	Low risk	Comment: no other sources of bias identified.
Overall risk of bias   All outcomes	Unclear risk	Comment: at least 1 domain judged as unclear risk, and no domains judged as high risk.

Methods	Randomised controlled trial
Participants	Study period: November 2002 to March 2003 Setting: emergency department of Schneider Children’s Medical Center, Israel Inclusion criteria: children aged 6 months to 6 years with a clinical picture of mild to moderate acute laryngotracheitis presenting to the emergency department. Mild to moderate croup defined as a Westley croup score of 1 to 11. Exclusion criteria: spasmodic croup; acute epiglottitis; bacterial tracheitis; pneumonia; foreign body aspiration; chronic lung disease; congenital or acquired anatomical airway anomalies; immunosuppressed or immunocompromised; treated before arrival at the emergency department with inhaled bronchodilators or corticosteroids in any form; exposed to varicella in the previous 28 days; contradictions to corticosteroid treatment Baseline demographics (N = 52): proportion male: treatment: 73%; comparator: 27% mean (SD) age in months: treatment: 31.1 (20.4); comparator: 26.7 (16.8) mean (SD) Westley croup score: treatment: 3.6 (2.6); comparator: 2.0 (2.5)
Interventions	Treatment (N = 26): single 0.60 mg/kg (maximum 10 mg) dose of intramuscular dexamethasone Comparator (N = 26): single 0.40 mg/kg dose of oral betamethasone Treatments were of equivalent potency; supplemental oxygen was provided to children in whom air saturation was < 95%.
Outcomes	Change in Westley croup score from baseline to 2 and 4 hours; return visits to the emergency department; use of epinephrine
Notes	Funding source: no external funding
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Study participants were assigned by a random numbers table"
Allocation concealment (selection bias)	Unclear risk	Comment: insufficient information provided to judge
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Comment: no blinding described. Subjective outcomes
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Comment: used a third‐party outcome assessor. No blinding. Subjective outcomes
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Comment: the number of children analysed is not reported.
Selective reporting (reporting bias)	Unclear risk	Comment: no protocol identified. All prespecified outcomes from methods appear in results.
Other bias	High risk	Comment: baseline imbalance in croup score
Overall risk of bias   All outcomes	High risk	Comment: at least 1 domain judged as high risk.

Methods	Randomised double‐blind controlled trial
Participants	Study period: September 2001 to April 2002 and September 2002 to February 2003 Setting: 4 paediatric emergency departments (Alberta Children’s Hospital in Calgary, Stollery Children’s Health Centre in Edmonton, Winnipeg Children’s Hospital in Winnipeg, or Children’s Hospital of Eastern Ontario in Ottawa) in Canada Inclusion criteria: children aged 3 months to 9 years with mild croup based on an initial medical evaluation. Mild croup was defined as onset within the past 72 hours of a seal‐like, barking cough and a Westley croup score of ≤ 2. Exclusion criteria: symptoms or signs of another cause of stridor; history of congenital or acquired stridor, asthma, exposure to varicella within the previous 21 days, chronic pulmonary disease, severe systemic disease, or known immune dysfunction; treatment with corticosteroids within the past 2 weeks; treatment of respiratory distress with epinephrine prior to enrolment; those enrolled in another clinical trial in the past 4 weeks; parents unable to speak either English or French; lack of a telephone at home; a prior visit to the emergency department because of croup during this episode of the disease Baseline demographics (N = 720): proportion male: treatment: 61%; control: 61% mean (SD) age in months: treatment: 35 (23); control: 35 (23) Westley croup score: treatment: 38% score of 0, 38% score of 1, 24% score of 2; control: 38% score of 0, 43% score of 1, 19% score of 2
Interventions	Treatment (N = 359): single 0.60 mg/kg (maximum of 20 mg) dose of oral dexamethasone Control (N = 361): single dose of oral placebo (10 mL distilled water) Both treatment and placebo included 50 mL of wild cherry‐flavoured syrup.
Outcomes	Return visits to any healthcare provider within 7 days
Notes	Funding source: Canadian Institutes of Health Research; Alberta Children's Hospital Foundation; Children's Hospital of Eastern Ontario Research Institute; Stollery Children's Hospital Foundation; Cumberland Pharmaceuticals
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "computer‐generated randomization scheme, stratified by centre, used random permuted blocks of 6‐20 children to ensure the comparable assignment of eligible patients."
Allocation concealment (selection bias)	Low risk	Quote: "Codes were secured at each center's pharmacy until enrolment and all decisions regarding data analysis had been finalized." The preparations were "not distinguishable" and "packaged in sequentially numbered, sealed bags."
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "double‐blind" "Parents were unable to determine which preparation their child had received" "Preparations were not distinguishable by appearance, volume, weight, taste, or smell."
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "double blind" "Preparations were not distinguishable by appearance, volume, weight, taste, or smell and were packaged in identical syringes in sequentially numbered, sealed bags" "biostatistician who was not otherwise involved in the study performed the data analysis."
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: missing data imputed using intention‐to‐treat analysis. 5% (N = 37) protocol deviations equally distributed between groups. 2% (N = 13) had incomplete follow‐up.
Selective reporting (reporting bias)	Unclear risk	Comment: no protocol identified. All prespecified outcomes from methods appear in results.
Other bias	Low risk	Comment: no other sources of bias identified.
Overall risk of bias   All outcomes	Unclear risk	Comment: at least 1 domain judged as unclear risk, and no domains judged as high risk.

Methods	Randomised double‐blinded controlled trial
Participants	Study period: January 2000 to December 2001 Setting: emergency department at a hospital in Greece Inclusion criteria: children aged 6 months to 5 years presenting to the emergency department with a clinical diagnosis of viral croup by a history of a short period of viral upper respiratory symptoms followed by hoarseness or barking cough and clinical evidence of hoarseness, barking cough, or stridor in the emergency department; modified Downes and Raphaelly croup score ≥ 2; could be managed as outpatients Exclusion criteria: known structural airways anomalies; acute epiglottitis; bacterial tracheitis; pneumonia; foreign body aspiration or past history of laryngoscopy; history of prior intubation; chronic airway obstruction; received steroids in the past 24 hours; required more than 1 treatment with nebulised L‐epinephrine or hospitalisation Baseline characteristics (N = 64): proportion males: treatment 1: 80%; treatment 2: 58%; treatment 3: 65% mean age in years: treatment 1: 2.6; treatment 2: 3.2; treatment 3: 3.4 mean modified Downes and Raphaelly croup score: treatment 1: 5.13; treatment 2: 5.89; treatment 3: 3.95
Interventions	Treatment 1 (N = 25): single 5 mL (1:1000 mg/mL) dose of nebulised L‐epinephrine via nebuliser with oxygen at a rate of 5 L/minute Treatment 2 (N = 19): single 0.60 mg/kg (maximum 8 mg) dose of intramuscular dexamethasone Treatment 3 (N = 20): single 200 µg dose of inhaled beclomethasone dipropionate delivered via AeroChamber device Supplemental oxygen was used for children with oxygen saturation values < 95%.
Outcomes	Change in modified Downes and Raphaelly croup score from baseline to 2 hours; return visits to the emergency department
Notes	Funding source: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "patients were assigned by a random numbers table to receive..."
Allocation concealment (selection bias)	Unclear risk	Comment: insufficient information provided to judge.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Comment: described as "double‐blind", though the interventions were clearly distinguishable, and the mechanism of blinding was not described
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Comment: no description of a third‐party outcome assessor. Carried over judgement from blinding of participants and personnel
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: no missing outcome data
Selective reporting (reporting bias)	Unclear risk	Comment: no protocol identified. All prespecified outcomes from methods appear in results.
Other bias	Low risk	Comment: no other sources of bias identified.
Overall risk of bias   All outcomes	High risk	Comment: at least 1 domain judged as high risk.

Methods	See Skowron 1966a and Skowron 1966b
Participants	See Skowron 1966a and Skowron 1966b
Interventions	See Skowron 1966a and Skowron 1966b
Outcomes	See Skowron 1966a and Skowron 1966b
Notes	See Skowron 1966a and Skowron 1966b
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	See Skowron 1966a and Skowron 1966b
Allocation concealment (selection bias)	Unclear risk	See Skowron 1966a and Skowron 1966b
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	See Skowron 1966a and Skowron 1966b
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	See Skowron 1966a and Skowron 1966b
Incomplete outcome data (attrition bias)   All outcomes	Low risk	See Skowron 1966a and Skowron 1966b
Selective reporting (reporting bias)	Unclear risk	See Skowron 1966a and Skowron 1966b
Other bias	Unclear risk	See Skowron 1966a and Skowron 1966b
Overall risk of bias   All outcomes	Unclear risk	See Skowron 1966a and Skowron 1966b

Methods	Randomised single‐blind controlled trial
Participants	Study period: January 2009 to March 2010 Setting: emergency department at Ali‐Ebne Abitaleb Hospital, Iran Inclusion criteria: children aged 6 months to 6 years admitted to the emergency department with barking cough, stridor, hoarseness, and respiratory distress Exclusion criteria: chronic pulmonary disease; severe croup (croup score > 7); recurrent croup; allergy to corticosteroids; contraindication of corticosteroid (history of tuberous sclerosis, history of varicella infection during the past 3 weeks); history of corticosteroid administration during the last 4 weeks; foreign body; epiglottitis; bacterial tracheitis; immune deficiency Baseline demographics (N = 68): proportion males: 53% mean (SD) age in months: 26.3 (1.5) mean (SD) croup score: treatment 1: 1.81 (0.59); treatment 2: 2.03 (0.47)
Interventions	Treatment 1 (N = 36): single dose 0.60 mg/kg intramuscular dexamethasone Treatment 2 (N = 32): single dose 0.60 mg/kg oral dexamethasone
Outcomes	Return visits or (re)admissions or both
Notes	Not indexed in the databases searched (located via a search of Google.ca); funding source: Zahedan University of Medical Sciences, Zahedan, Iran
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: described as "randomly divided"; unclear how the randomisation sequence was generated
Allocation concealment (selection bias)	Unclear risk	Comment: insufficient information provided to judge.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Comment: participants and personnel were not blinded. Treatments were clearly distinguishable.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: outcome assessor described as blinded. Unclear how they were blinded and if the blinding could have been broken
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Comment: 15% loss to follow‐up; unclear from which group children were lost. No intention‐to‐treat analysis
Selective reporting (reporting bias)	Unclear risk	Comment: no protocol identified. All prespecified outcomes from methods appear in results.
Other bias	Low risk	Comment: no other sources of bias identified.
Overall risk of bias   All outcomes	High risk	Comment: at least 1 domain judged as high risk.

Study	Reason for exclusion
Anene 1996	Randomised controlled trial; children were not diagnosed with croup
Bollobas 1965	Not a randomised controlled trial; children were not diagnosed with croup
Cichy 1983	Not a randomised controlled trial
Connolly 1969	Randomised controlled trial; children were not diagnosed with croup
Couser 1992	Randomised controlled trial; children were not diagnosed with croup
Eghbali 2016	Randomised controlled trial; the intervention was L‐epinephrine
Faghihinia 2007	Randomised controlled trial; no usable results were presented (not clear how many children were in each group)
Faraji‐Goodarzi 2018	Randomised controlled trial; no relevant outcomes
Flisberg 1973	Not a randomised controlled trial
Freezer 1990	Not a randomised controlled trial; study was a retrospective chart review
Gill 2017	Not a randomised controlled trial
Goddard 1967	Randomised controlled trial; children were not diagnosed with croup
Haque 1981	Not a randomised controlled trial; children in the control group received no treatment
Havaldar 1997	Not a randomised controlled trial; children were not diagnosed with croup
Kelley 1992	Not a randomised controlled trial; study was a retrospective chart review
Kunkel 1996	Not a randomised controlled trial; intervention was epinephrine
Ledwith 1995	Not a randomised controlled trial; intervention was epinephrine
Martensson 1960	Not a randomised controlled trial
McDonogh 1994	Not a randomised controlled trial; study was a retrospective chart review
Mohammadzadeh 2014	Randomised controlled trial; intervention was epinephrine
NCT01748162	Randomised controlled trial; did not report any relevant outcomes
Novik 1960	Not a randomised controlled trial
Osváth 1994	Not a randomised controlled trial
Prendergast 1994	Not a randomised controlled trial; intervention was epinephrine
Rizos 1998	Not a randomised controlled trial
Roked 2015	Not a randomised controlled trial; study was a retrospective chart review
Ross 1969	Not a randomised controlled trial; study was a retrospective chart review
Serra 1997	Not a randomised controlled trial
Sumboonnanonda 1997	Randomised controlled trial; children in the control group received no treatment
Sussman 1964	Randomised controlled trial; children were not diagnosed with croup
Tal 1983	Randomised controlled trial; children were not diagnosed with croup
Tellez 1991	Randomised controlled trial; children were not diagnosed with croup
Wilhelmi 1976	Not a randomised controlled trial

Trial name or title	Oral prednisolone compared to oral dexamethasone (in 2 different doses) in children with croup: randomised clinical trial comparing the improvement in Westley croup score
Methods	Randomised controlled trial (parallel)
Participants	Inclusion criteria: children aged 6 months to 10 years with a clinical diagnosis of croup (laryngotracheitis, laryngotracheobronchitis); weight no more than 20 kg; parents could be contacted by telephone; parents speak English Exclusion criteria: high clinical suspicion of an alternative diagnosis; known allergy to prednisolone or dexamethasone; immunosuppressive disease or treatment; steroid therapy in the past 14 days; enrolment in the present study in the past 14 days (i.e. repeat enrolment)
Interventions	Treatment 1: single 1 mg/kg dose of oral prednisolone Treatment 2: single 0.15 mg/kg dose of oral dexamethasone
Outcomes	Westley croup score; unscheduled medical re‐attendance/readmission; length of stay in hospital; use of additional treatments
Starting date	18 March 2009
Contact information	Dr Colin Parker. Princess Margaret Hospital for Children, Roberts Road Subiaco 6008 Western Australia, Australia. Colin.Parker@health.wa.gov.au
Notes	Trial registration at the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (ACTRN12609000290291); funding source: Princess Margaret Hospital Foundation