Criteria for considering studies for this review
Types of studies
Randomised controlled trials (RCTs) will be included, irrespective of language, age of publication or publication status. Cross-over and cluster-RCTs will be included, however we will analyse the data obtained from them separately.
Types of participants
We will consider participants meeting the following criteria.
Types of interventions
We will include studies which compare any type of water-based exercise to another water-based exercise, land-based exercise or usual care. We will document the duration and frequency of sessions and the overall length of the programme. At least one group in each comparison must use water-based exercise.
The following possible comparisons may be included:
Swimming versus hydrogymnastics
Swimming versus hydrotherapy
Water-based exercise versus land-based exercise
Water-based exercise versus usual care
Types of outcome measures
Quality of life (measured using a validated questionnaire, e.g. Asthma Quality of Life Questionnaire (AQLQ))
Exacerbations leading to a course of oral steroids
Exacerbations leading to health centre visits
Exacerbations leading to hospitalisations
Use of rescue medication (e.g. short-acting beta2-agonist (SABA)) or symptom controller (e.g. long-acting beta2-agonist (LABA))
Symptoms (as measured on a validated scale (e.g. Lara Asthma Symptom Scale (LASS))
Lung function by physiological measures (e.g. spirometry, FEV1, FVC)
Reduction in medication usage (e.g. inhaled corticosteroids or leukotriene inhibitors)
Adverse events (any reaction due to the water treatment process (e.g. chlorine), exercise induced-bronchoconstriction (EIB))
Search methods for identification of studies
We will identify trials from the Cochrane Airways Group Specialised Register of trials (CAGR), which is derived from systematic searches of bibliographic databases including the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL, AMED and PsycINFO (see Appendix 1 for further details). We will search the CAGR using the strategy in Appendix 2 .We also will search LILACS (Latin American and Caribbean Health Sciences), PEDro (The Physiotherapy Evidence Database), SIGLE (System for Information on Grey Literature) and Google Scholar. We will handsearch the respiratory journals and meeting abstracts of the American Thoracic Society (ATS), European Respiratory Society (ERS) and British Thoracic Society (BTS).
We will check for ongoing studies in: ClinicalTrials.gov (http://clinicaltrials.gov/); the metaRegister of Controlled Trials (mRCT) (http://www.controlled-trials.com/); International Clinical Trials Registry Platform (ICTRP) (http://www.who.int/ictrp/en/); Pan African Clinical Trials (http://www.pactr.org) and EU Clinical Trials Register (https://www.clinicaltrialsregister.eu/).
We will apply no language or date restrictions. We will search all databases from their inception to the present date.
We will use Endnote X5 to manage the references and exclude duplicate articles found in the different searches.
Searching other resources
We will check the reference lists of all primary studies and review articles for additional references. We will contact the authors of identified trials and ask them to identify other published and unpublished studies. We will also contact exercise equipment manufacturers and experts in the field.
Data collection and analysis
We will perform data collection and analysis according to the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a).
Selection of studies
Two review authors (AJG, VS) will independently screen titles and abstracts of all the potential studies we identify as a result of the search and code them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. We will retrieve the full-text study reports/publications and two review authors (AJG, VS) will independently screen the full text to identify studies for inclusion. We will record the reasons for the exclusion of any ineligible studies. We will resolve any disagreement through discussion or, if required, we will consult a third person (BNGS). We will identify and exclude duplicates and collate multiple reports of the same study so that each study rather than each report is the unit of interest in the review. We will record the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table.
Data extraction and management
Two review authors (AJG, VS) will independently extract the data from the primary studies using a standard form (Google Spreadsheet, e.g. http://goo.gl/pd0KZ) developed for this purpose, which includes methods, participants, intervention and outcomes. One review author will transfer data into Review Manager (RevMan 2012). We will double check that data are entered correctly by comparing the data presented in the systematic review with the study reports. A second review author will spot-check study characteristics for accuracy against the trial report.
Assessment of risk of bias in included studies
Two review authors (AJG, VS) will independently assess the risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). Any disagreement will be resolved by discussion or by involving a third author (MSP). We will assess the risk of bias according to the following domains.
Random sequence generation (selection bias)
Allocation concealment (selection bias)
Blinding of participants and personnel (performance bias)
Blinding of outcome assessment (detection bias)
Incomplete outcome data (attrition bias)
Selective reporting (reporting bias)
Other bias (other sources of bias related to a particular trial design (cross-over or cluster-randomised) or specific circumstances (interventions mixed))
We will classify the risk of bias as low, high or unclear (Higgins 2011b).
Measures of treatment effect
Considering the proposed objectives of this review, we may use four types of measurements of treatment effect for the primary studies.
Dichotomous data: outcomes measures evaluated as binary responses. We will show the summarised results as risk ratio (RR) or odds ratio (OR) and, if necessary, we will calculate the number needed to treat to benefit (NNTB) from the RR or OR. This type of data may be found for the number of acute exacerbations, for example.
Continuous data: outcome measures evaluated by numerical quantity. We will combine the results using the mean difference (MD) for measures using the same scale or the standardised mean difference (SMD) where different scales are used to evaluate the same outcome. This type of data may be found for studies measuring quality of life, for example.
Count and rates: this measure of effect may mean that exacerbation episodes are counted twice or more in the same person. If a meta-analysis is possible, we may use the risk ratio as it represents the rate of events in the two groups by dividing one by the other.
Time-to-event: we may use the hazard ratio to determine when the event happened and its link with the intervention.
If we find more than one study which analyses the same outcome, we will conduct a meta-analysis. All statistical parameters will use 95% confidence intervals (CI) (Deeks 2011).
Unit of analysis issues
We will consider the individual patient to be the unit of analysis. If we include cluster-randomised trials, the unit of analyses will be the group which was allocated the intervention. In cross-over designs we will only include the phase one data to avoid a carryover effect.
Dealing with missing data
We will contact authors of studies by email if the study reports do not report the outcomes measures of interest, do not describe the process of randomisation, do not describe intention-to-treat analysis or are missing any other necessary data. We will send at least two emails to the corresponding author and, if no answer is obtained, we will present and discuss this in the text.
Assessment of heterogeneity
We will assess the inconsistencies between studies using the I2 statistic and describe the percentage of variability in effect. We will consider heterogeneity to be substantial when I2 > 50% (Deeks 2011). If we find substantial heterogeneity among studies, we will explore this with subgroup analysis of study characteristics: participants, interventions, etc.
Assessment of reporting biases
Reporting biases can affect the likelihood of publication and statistically significant results may be selectively reported. It is important that this is assessed because reporting biases may over or under-estimate the treatment effect. If mismatches are identified, we will contact the authors for further information. We also plan to explore the impact of including such studies by conducting a sensitivity analysis. We will perform a funnel plot asymmetry test if we are able to meta-analyse 10 or more trials.
We will conduct a meta-analyses if the combination of data is possible. We will use a fixed-effect model unless there is substantial heterogeneity (I2 > 50%) where we will choose a random-effects model. We will present the data in forest plot graphics produced by RevMan 5.2. If data combination is not possible, we will describe the individual studies. We also plan to create a 'Summary of findings' table using the methods and recommendations described in Section 8.5 and Section 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b; Schünemann 2011) and using the GRADEpro software (GRADEpro 2008). We plan to include the following outcomes: quality of life, number of acute exacerbations, level of severity of disease, lung function, reduction in medication usage and adverse events.
Subgroup analysis and investigation of heterogeneity
We will conduct subgroup analysis considering participant and intervention characteristics.
Type of exercise (swimming (horizontal), hydrogymnastics (vertical))
Frequency of exercise (how many sessions/week)
Intensity: light (1.6 to 2.9 metabolic equivalents (METs)), moderate (3 to 5.9 METs), vigorous ≥ 6 METs
Water temperature and air humidity
Water treatment process (chlorinated versus non-chlorinated)
The sensitivity analysis will help us to understand whether the findings are dependent on the authors' interpretation. We will test studies with limited information to check whether the robustness of the assumptions influences the review findings. Moreover, we will conduct sensitivity analyses with studies at high and unclear risk of bias. We also plan to conduct sensitivity analyses to look at the effects of missing data.