Types of studies

We planned to include randomised controlled trials (RCTs), reported as full‐text, those published as abstract only, and unpublished data. We excluded cross‐over trials because the effects of corticosteroids can persist for a number of weeks or months (Haahtela 1994) and studies did not employ a sufficient washout period between treatment periods to eliminate cross‐over effects.

Types of participants

We planned to include adults (aged ≥ 18 years) and children of school age or older (i.e. aged ≥ 5 years) with a diagnosis of asthma. Adults and children were to be considered in separate comparisons. Preschool wheeze was not considered by this review and as such we excluded studies of preschool children. The diagnosis of asthma was required to be determined by a clinician according to validated national or international guidelines. We excluded participants with any respiratory comorbidities (e.g. bronchiectasis, chronic obstructive pulmonary disease). If a study contained both adults and children, we contacted the study authors to check if disaggregated data were available; if we were unable to source these data we used the average age (≥ 18 years) of study participants to determine suitability for inclusion. If the average age of study participants was less than 18 years, we planned to perform a sensitivity analysis to examine the effect of including or excluding these studies. Finally, if a study included children of both school and preschool age, we excluded the study if the average age was less than five years old.

Types of interventions

We planned to include studies comparing any patient‐ or parent‐initiated oral corticosteroid (OCS), with either placebo, normal care, an alternative active treatment plan (e.g. doubling the dose of inhaled steroids) or an identical personalised asthma exacerbation management plan without patient‐ or parent‐initiated OCS. OCS (any dose or duration) could be combined with other measures for the management of an exacerbation (e.g. personalised asthma action plan, increased use of reliever inhaler) provided that the measure was not part of the randomised treatment. We planned to perform separate comparisons for each type of comparator (e.g. patient‐initiated steroids versus placebo; patient‐initiated steroids versus normal care, etc); separate comparisons would also be performed for adults and children. We defined normal care as any measure that the patient would usually take to manage an exacerbation (e.g. increase reliever inhaler use, seek medical advice, etc). We also planned to include studies where the comparator group comprised a combination of the above (e.g. personalised asthma action plan plus placebo).

Types of outcome measures

Electronic searches

We identified trials from the Cochrane Airways Group's Specialised Register (CAGR) (searched 18 May 2016), which is maintained by the Information Specialist for the Group. The Register contains trial reports identified through systematic searches of bibliographic databases including the Cochrane Central Register of Controlled Trials (CENTRAL; 2016, Issue 4) in the Cochrane Library; MEDLINE Ovid; Embase Ovid; CINAHL EBSCO (Cumulative Index to Nursing and Allied Health Literature; AMED Ovid (Allied and Complementary Medicine; and PsycINFO Ovid; and handsearching of respiratory journals and meeting abstracts (see Appendix 1 for further details). We searched all records in the CAGR using the search strategy in Appendix 2.

We also conducted a search of the US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; searched 24 August 2016) and the World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch; searched 24 August 2016).

When searching all databases we imposed no restriction on language of publication.

Searching other resources

We checked the reference lists of all primary studies and review articles for additional references. We searched relevant manufacturers' websites for trial information.

We planned to search for errata or retractions from included studies published in full‐text on PubMed (www.ncbi.nlm.nih.gov/pubmed) and to report the date this was done.

Selection of studies

Two review authors (MBG, MG) independently screened titles and abstracts of all the potentially‐relevant studies that we identified as a result of the search and coded them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. We retrieved the full‐text study reports/publication and two review authors (MBG, MG) independently screened the full‐text articles to confirm studies for inclusion, or to identify and record reasons for exclusion of the ineligible studies. We resolved any disagreement through discussion or, if required, we consulted a third review author (MM). We identified and excluded duplicates and collated multiple reports of the same study, so that each study rather than each report is the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table (Moher 2009).

Data extraction and management

We planned to use a data collection form for study characteristics and outcome data, which would be piloted on at least one study in the review. Two review authors (MBG, DE) planned to extract study characteristics from included studies. We planned to extract the following study characteristics.

1. Methods: study design, total duration of study, details of any 'run in' period, number of study centres and location, study setting, withdrawals, and date of study.

2. Participants: number, mean age, age range, gender, severity of condition, diagnostic criteria, baseline lung function, smoking history, inclusion and exclusion criteria.

3. Interventions: intervention, comparison, concomitant medications, and excluded medications.

4. Outcomes: primary and secondary outcomes specified and collected, and time points reported.

5. Notes: funding for trial, and notable conflicts of interest of trial authors.

Two review authors (MBG, DE) planned to independently extract outcome data from the included studies. We planned to note in the 'Characteristics of included studies' table if outcome data were not reported in a usable way. We planned to resolve disagreements by consensus or by involving a third review author (MM). One review author (DE) was responsible for transferring data into the Review Manager (RevMan 2014) file. We planned to double‐check that data were entered correctly by comparing the data presented in the systematic review with the study reports. A second review author (DE) was responsible for spot‐checking study characteristics for accuracy against the trial report.

Assessment of risk of bias in included studies

Two review authors (MBG, DE) planned to independently assess risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We planned to resolve any disagreements by discussion or by involving another review author (MM). We planned to assess the risk of bias according to the following domains.

1. Random sequence generation.

2. Allocation concealment.

3. Blinding of participants and personnel.

4. Blinding of outcome assessment.

5. Incomplete outcome data.

6. Selective outcome reporting.

7. Other bias.

We planned to grade each potential source of bias as high, low or unclear risk, and provide a quote from the study report together with a justification for our judgement in the 'Risk of bias' table. We planned to summarise the 'Risk of bias' judgements across different studies for each of the domains listed, and to consider blinding separately for different key outcomes where necessary (e.g. for unblinded outcome assessment, risk of bias for all‐cause mortality may be very different than for a patient‐reported pain scale). Where information on risk of bias relates to unpublished data or correspondence with a trialist, we planned to note this in the 'Risk of bias' table.

When considering treatment effects, we would take into account the risk of bias for the studies that contribute to that outcome.

Measures of treatment effect

We planned to analyse dichotomous data as odds ratios and continuous data as mean difference or standardised mean difference. We planned to enter data presented as a scale with a consistent direction of effect.

We planned to undertake meta‐analyses only where this was meaningful (i.e. if the treatments, participants and the underlying clinical question were similar enough for pooling to make sense).

We planned to narratively describe skewed data reported as medians and interquartile ranges.

Where multiple trial arms were reported in a single trial, we planned to include only the relevant arms. If two comparisons (e.g. drug A versus placebo and drug B versus placebo) were combined in the same meta‐analysis, we planned to halve the control group to avoid double‐counting.

Unit of analysis issues

For dichotomous outcomes, we planned to report participants, rather than events, as the unit of analysis. For example, for the secondary outcome 'unscheduled visit to a healthcare provider' we would record the number of participants with an unscheduled visit, rather than the number of unscheduled visits per participant.

Dealing with missing data

We planned to contact investigators or study sponsors in order to verify key study characteristics and obtain missing numerical outcome data where possible (e.g. when a study was identified as abstract only). Where this was not possible, and the missing data were thought to introduce serious bias, we planned to explore the impact of including such studies in the overall assessment of results by conducting a sensitivity analysis. 

Assessment of heterogeneity

We planned to use the I² statistic to measure heterogeneity among the studies in each analysis. If we identified substantial heterogeneity we would report it and explore possible causes by prespecified subgroup analysis. 

Assessment of reporting biases

If we were able to pool more than 10 studies, we planned to create and examine a funnel plot to explore possible small study and publication biases.

Data synthesis

We planned to use a random‐effects model and perform a sensitivity analysis with a fixed‐effect model.

Subgroup analysis and investigation of heterogeneity

We planned to carry out the following subgroup analyses.

1. Baseline asthma severity (stratified by background medication).

We planned to use the following outcomes in subgroup analyses.

1. Hospital admissions for asthma.

2. Asthma symptoms at follow‐up.

3. Serious adverse events.

We planned to use the formal test for subgroup interactions in Review Manager (RevMan 2014).

Sensitivity analysis

If necessary, we planned to carry out the following sensitivity analyses to explore the effect of including/excluding:

1. studies that included both adults and children/adolescents, where the average age of participants was < 18 years;

2. unpublished data (i.e. no peer‐reviewed full‐text paper available);

3. studies at high risk of performance or detection bias;

4. studies at high risk of any other bias;

5. studies with missing data.