Swimming training for asthma in children and adolescents aged 18 years and under

Summary of findings for the main comparison. Swimming training for asthma in children and adolescents aged 18 years and under

Swimming training for asthma in children and adolescents aged 18 years and under
Patient or population: children and adolescents aged 18 years and under studies with asthma Settings: Recruited from asthma clinics. Asthma diagnosis by recognised criteria. Intervention: Swimming training programme‐ meeting minimum intensity criteria (> Weekly, > 20 minutes, > 4 weeks)
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Swimming training
Quality of life¹ Paediatric Asthma Quality of Life Questionnaire (PAQLQ). Scale from: 1 (worse) to 7 (better). Follow‐up: mean 9 weeks	The mean change in quality of life in the control group was ‐1.87	The mean change in quality of life in the intervention group was 0.26 (1.05 lower to 1.58 higher)		50 (1 study¹)	⊕⊕⊝⊝ low^2,3
Asthma symptoms Different scales in different studies (lower scores mean fewer symptoms) Follow‐up: 6‐9 weeks	The mean change in asthma symptoms ranged across control groups from 0 to ‐2.14 standard deviations	The mean asthma symptoms in the intervention groups was 0.06 standard deviations less (0.58 lower to 0.47 higher) see comment		58 (2 studies)	⊕⊕⊝⊝ low^3,4,5	The difference of 0.06 standard deviations would equate to a small difference on Living with Asthma Questionnaire (LWAQ) or a composite 12‐point scale of < 0.5 units. The effect size is < 0.2 representing a small effect.
Exacerbations requiring hospital admission	see comment	see comment	see comment	see comment	‐	Outcome not reported
Exacerbations requiring a course of oral corticosteroids	see comment	see comment	see comment	see comment	‐	Outcome not reported
Urgent asthma physician visits¹ Number of times the child visited a physician's office/clinic for an asthma flare up Follow‐up: mean 2 months	The mean urgent asthma physician visits in the control group was 0.17 visits in 2 months	The mean urgent asthma physician visits in the intervention groups was 0.08 higher (0.25 lower to 0.42 higher)		44 (1 study¹)	⊕⊕⊕⊝ moderate³
Resting lung function Forced expiratory volume (FEV1) in 1 second (litres) Follow‐up: 6‐12 weeks	The mean change in resting FEV1 ranged across control groups from 0.05‐0.15 litres	The mean difference in FEV1 in the intervention groups was 0.10 L higher (0 to 0.2 higher)		113 (4 studies)	⊕⊕⊕⊝ moderate³	The mean difference is comparable to the difference in FEV1 in children with asthma (N = 4, n = 719 ⁷) comparing low dose fluticasone propionate (100 mcg) daily with placebo mean difference (MD) 0.1 L [0.15, 0.36] (Adams 2008)
Fitness⁶ Maximal oxygen consumption (VO2 max) Follow‐up: mean 12 weeks	The mean VO₂ max in usual care control groups was 39 mL/kg/min	The mean fitness in the swimming intervention groups was 9.67 mL/kg/min higher (5.84 to 13.51 higher)		32 (2 studies)	⊕⊕⊕⊕ high	The 25% difference for swimming compared to control in VO₂ max is clinically meaningful. It is larger than the differences seen in physical activity studies in children without asthma, range 5% to 15% (Armstrong 2011) and that seen in children with asthma undertaking physical training 9% (Counil 2003).
The basis for the assumed risk* is provided in the table. The corresponding risk (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; FEV1 forced expiratory volume in one second; L: litres;
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Control group: golf ² High risk of attrition bias assessed in study ³ The confidence interval does not rule out a null effect or harm ⁴ Comparison groups differed; usual care or golf ⁵ Effect size < 0.2 represents small effect ⁶ Pooled studies with non‐active usual care control group only ⁷ N= number of studies; n= number of participant

Background

Description of the condition

Originating from the Greek word 'panting', asthma is a disease associated with chronic inflammation of the airways accompanied by hyper‐reactive responses of the bronchi (Ward 2002). These heightened responses result in the obstruction of airflow, which manifests as symptoms such as coughing, wheezing, chest tightness and shortness of breath (Beilby 2006). Asthma is the most common chronic medical condition among children and is one of the most common causes of hospitalisation and medical visits in the same age group (World Health Organization Media Centre 2011). It is estimated that 235 million people have asthma, while approximately a quarter of the 40 million Americans with asthma are children under the age of 17 (American Lung Association 2010). Multiple epidemiological studies from around the world indicate that the prevalence of asthma among children and adolescents is rising (Wong 2008).

Physical activity can lead to increased airway resistance in many people with asthma, precipitating an episode of exercise‐induced asthma. Fear of such episodes may lead to decreased participation in physical activity, as suggested by numerous studies that have reported that children with asthma have lower cardiorespiratory fitness than their peers (Clark 1988; Lang 2004; Welsh 2005).

Description of the intervention

Several studies in children with asthma have demonstrated that physical exercise does improve aerobic fitness as well as reduce episodes of wheeze, hospitalisations, school absenteeism and, to a lesser degree, medication usage (Welsh 2005). As a subtype of physical training, swimming is often suggested as the ideal form of physical activity for individuals with asthma. Swimming training is a structured regular exercise programme through supervised aquatic activities, which aims to increase cardiorespiratory fitness.

How the intervention might work

There are a number of postulated reasons as to why swimming training may be superior to other forms of physical training for children with asthma. These include, the air above the pool being warmer and humidified, low pollen count exposure, hydrostatic pressure on the chest wall reducing expiratory effort and work, relative hypoventilation due to controlled breathing leading to increased carbon dioxide and horizontal posture (Bar‐Or 1992; Bernard 2010; Downing 2011; Inbar 1991; Wardell 2000). These are on top of the effects of any physical training, namely increased self‐esteem, self‐confidence and improved cardio‐pulmonary fitness. Of note however, are the concerns raised over the past decade in relation to the potential pro‐asthmatic effect of chlorine by‐products in pools (Nickmilder 2007; Uyan 2009).

Why it is important to do this review

Asthma is an increasingly common chronic disease in many parts of the world and it has become important to identify safe and potentially beneficial exercises for the condition. Swimming is a form of exercise that has been commonly lauded as 'healthy' for people with asthma, often being implemented in guidelines without an explicit evidence base. Several reviews have been done and many emphasise the need for further research and analysis of the existing literature on the subject (Chandratilleke 2012; Fisk 2010; Goodman 2008; Ram 2000; Ram 2005). There has been no shortage of studies on the effect of swimming, often with different results and study designs. Thus it is crucial to provide a systematic analysis and critical appraisal of the current research finding and identify whether swimming is safe and beneficial for children and adolescents with asthma.

It is also important to establish the safety of swimming training for people with asthma, especially in light of the concerns about chlorinated pools (Bernard 2003; Bernard 2006a).

Objectives

To determine the effectiveness and safety of swimming training as an intervention for asthma in children and adolescents aged 18 years and under.

Methods

Criteria for considering studies for this review

Types of studies

We included all available randomised controlled trials (RCTs) and quasi‐RCTs (i.e. using a quasi‐random allocation method such as allocation by date of birth or day of the week) of children undergoing swimming training. We identified and included studies reported in abstract form and requested data from trialists where no full publication was found. We did not impose any restrictions on language, year of publication and type of publication of a study.

Types of participants

We included studies of children and adolescents aged 18 years and under, with physician‐diagnosed asthma or based on objective criteria as stated in study methods, such as bronchodilator response, or both. We included studies with participants having any severity of asthma.

Types of interventions

We included studies with swimming training, defined as a formal swimming programme of at least one session per week, with each session lasting at least 20 minutes and running over a minimum of four weeks. Studies could have a comparison group receiving usual care without any intervention, or undertake a non‐physical activity or physical activity other than swimming.

Types of outcome measures

We assessed the effects of interventions in these categories of outcomes where available: patient‐related, health economic and objective measures of lung function, airway reactivity and inflammation

Primary outcomes

Quality of life measured by disease specific or generic questionnaires (e.g. Paediatric Asthma Quality of Life Questionnaire (PAQLQ) (Juniper 1996).
Asthma control measured by questionnaires/symptom diaries.
Exacerbations of asthma requiring attendance at hospital.
Systemic steroid use for exacerbations of asthma.

Secondary outcomes

Bronchodilator use.
Use of preventer medication (e.g. inhaled corticosteroids [ICS]).
Lung function (including peak expiratory flow (PEF), forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC)).
Exercise capacity (cardio‐pulmonary fitness).
Bronchial hyper‐responsiveness (determined by a formal direct or indirect challenge test).
Time‐off required from employment or education.
Utilisation of healthcare services.

Search methods for identification of studies

Electronic searches

We identified trials using the Cochrane Airways Group's Specialised Register of trials, which is derived from systematic searches of bibliographic databases including the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library, latest issue), MEDLINE, EMBASE, CINAHL, AMED and PsychINFO and handsearched respiratory journals and meeting abstracts (please see Appendix 1 for further details). The TSC searched all records in the Specialised Register coded as 'asthma' using the following terms: swim* or pool* or ((water* or aquatic*) and exercise*). We conducted additional searches of CENTRAL, PubMed, CINAHL and EMBASE. See Appendix 2 for the search strategies. Searches were conducted in November 2011. A repeat search of Cochrane Central Register of Controlled Trials (CENTRAL) database in July 2012 did not find any new studies.

We conducted a search of national and international trial registers including The Australian New Zealand Clinical Trials Registry, Chinese Clinical Trial Register, ClinicalTrials.gov register, Current Controlled Trials metaRegister of Controlled Trials (mRCT) – active registers, Hong Kong clinical trials register, International Clinical Trials Registry Platform Search Portal, South African National Clinical Trial Register and UK Clinical Trials Gateway. We searched all databases from their inception to the present, with no restriction on language of publication.

Searching other resources

We checked reference lists of all studies assessed to identify other relevant studies. We consulted experts in the field for ongoing or unpublished studies. We contacted the author(s) of any identified abstracts or unpublished studies to ascertain the study design and outcome measures. We included abstracts and unpublished studies if sufficient information on the study design was available, but only included them in the meta‐analysis where we had data on outcome measures.

Data collection and analysis

Selection of studies

At least two review authors (YF, HL, WN) independently identified studies and assessed whether they met the inclusion criteria. We resolved any discrepancies through consultation with a third review author (JW, RWB, or SB). The review authors initially identified studies for inclusion based on citation and abstract but if there was insufficient information to make a determination, we retrieved the full article. We eliminated duplicate studies by comparing authors' names and titles of studies.

Data extraction and management

At least two review authors (JW, WN, HL, YF) independently extracted and entered data onto a standard extraction form.

We extracted the following characteristics.

Methods: study design, location, number of centres, duration of study; methods of analysis.
Participants: recruitment, target participants, N screened, N randomised, N completed, asthma diagnosis criteria, severity of asthma, gender, age, other inclusion criteria, exclusion criteria.
Interventions: setting of intervention: indoor/outdoor, chlorinated/non‐chlorinated; swimming training supervisor; description of intervention: length session/frequency, duration; control; co‐interventions.
Outcomes: pre‐specified, follow‐up period.

Coding for subgroup analysis: adolescents/children; asthma mild/moderate/severe, length swimming session, indoor/outdoor pool, chlorinated/non‐chlorinated.

Any discrepancies were resolved though consultation with a third review author.

Assessment of risk of bias in included studies

Two review authors (YF, HL, WN) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Since it was not possible to blind participants in these studies this criterion was not assessed. Disagreement was resolved by discussion or by involving a third review author (JW). We examined 'Risk of bias' assessment through the synthesis of a 'Risk of bias' table.

We assessed the risk of bias according to the following domains.

Random sequence generation.
Concealment of allocation.
Blinding of assessors.
Incomplete outcome data.
Selective outcome reporting.

We noted any other potential sources of bias. We graded each domain as having 'low', 'high' or 'unclear' risk of bias.

Measures of treatment effect

If studies used the same scale to measure a continuous outcome, for example lung function or asthma medication use, we calculated mean differences (MDs) and 95% confidence intervals (CIs) using change from baseline where data were available. However, if the measurements pre‐ and post intervention were unavailable we used the absolute values in the groups. Where different scales were used to measure a continuous outcome, we calculated a standardised mean difference (SMD) and 95% CI. We determined the minimum threshold for a clinically significant effect for outcomes such as asthma control, via established published standards.

For dichotomous outcomes (exacerbations of asthma or adverse events), we expressed results as Peto odds ratio (Peto OR) with 95% CI.

Dealing with missing data

We assessed categories of missing data under missing outcomes, missing summary data and missing individuals as in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We noted missing or unclear data on data collection forms and contacted trial authors for clarification. We also asked study authors to provide data for unreported outcomes. We assumed that loss of participants occurring prior to performance of baseline measurements had no effect on the eventual outcome data for that study. We addressed missing standard deviations by imputing data, either using studies in the same meta‐analysis or for changes from baseline (Abrams 2005), calculating a correlation coefficient from baseline and final measurements for outcomes available in other studies (see Table 1). We used the correlation coefficients to calculate and impute the standard deviation of change from baseline. We explored the impact of imputation in the overall assessment of results by sensitivity analyses.

Table 1. Correlation coefficients: from studies for imputing standard deviation of mean changes

Weisgerber 2008	Swimming group	Golf group
FEV1 % predicted	0.6491	0.9179
FVC % predicted	0.8722	0.9064
FEF 25% to 75% % predicted	0.3823	0.9234
Coopers test	0.7655	0.8588
Peak heart rate	0.9259	0.3264
Exercise time	0.8200	0.8584
Weisgerber 2003	Swimming group	Usual care group
FEV1 L	0.5983	0.9433
FEV1 % predicted	0.5585	0.8495
FVC (L)	0.7779	0.9781
FVC % predicted	0.9229	0.4247
PEF (L/s)	0.5115	0.7406
PEF % predicted	0.0319	0.9062
FEF 25% to 75% % predicted	0.8634	0.8904

FEF = forced expiratory flow; FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; PEF = peak expiratory flow

Assessment of heterogeneity

We used the I² statistic to measure heterogeneity among the trials in each analysis Higgins 2011. Interpretation of statistical heterogeneity was according to the recommendation of Higgins 2011, as follows:

0% to 40%: might not be important;
30% to 50%: may represent moderate heterogeneity;
50% to 90%: may represent substantial heterogeneity;

Where substantial heterogeneity ( I² > 50%) was identified, we explored it using pre‐specified subgroup analyses where possible.

Assessment of reporting biases

Where reporting bias was indicated (see 'Selective reporting bias' above), we attempted to contact study authors to ask them to provide missing outcome data.

Data synthesis

We combined studies that measured the same outcome in meta‐analyses where data were available. Where there was a high level of heterogeneity we considered three options; not conducting a meta‐analysis, exploring causes of heterogeneity and conducting a random‐effects meta‐analysis.

Subgroup analysis and investigation of heterogeneity

We pre‐specified subgroup analyses between the following groups where sufficient studies could be included.

Purely observed versus non‐physical versus physical (non‐swimming) controls.
People under 12 years of age, and those 12 years and above.
Severity of asthma as defined in the included studies.
Length of swimming session.
Chlorinated versus non‐chlorinated indoor pools.
Indoor versus outdoor pools.

Sensitivity analysis

We identified studies with a high risk bias random sequence generation, concealment of allocation, incomplete outcome data or selective outcome reporting and conducted sensitivity analyses (by excluding these studies). We did not consider blinding as this was not possible due to the nature of the intervention. Pooled results obtained using absolute outcome values were compared with pooled results using change values in sensitivity analyses presented in Table 2 and Table 3. Pooled results using standard deviations for change values were compared when calculated from coefficients from different studies in sensitivity analyses presented in Table 2 and Table 3.

Table 2. Sensitivity analyses

Outcome	N	SD calculation method	Effect size [95% CI]
Quality of life (PAQLQ)	1	Change (no imputation)	MD 0.26 [‐1.05, 1.58]
Quality of life (PAQLQ)	1	Absolute	MD 0.16 [‐1.17, 1.50]
Symptoms (PAQLQ)	1	Change (no imputation)	MD 0.07 [‐1.12, 1.26]
Symptoms (PAQLQ)	1	Absolute	MD 0.12 [‐1.20, 1.43]
Caregiver quality of life (PACQLQ)	1	Change (no imputation)	MD 0.71 [‐0.83, 2.25]
Caregiver quality of life (PACQLQ)	1	Absolute	MD 0.61 [‐0.86, 2.07]
Asthma symptoms (LWAQ index)	1	Change (no imputation)	MD ‐0.10 [‐2.55, 2.36]
Asthma symptoms (LWAQ index)	1	Absolute	MD 0.83 [‐1.44, 3.11]
Exercise Capacity: Any measure (control: usual care)	4	Cycle ergometry (Watts)	MD 1.34 [0.82, 1.86]
Exercise Capacity: Any measure (control: usual care)	4	Swimming ergometry (kp)	MD 1.41 [0.88, 1.94]

LWAQ = Living with Asthma Questionnaire; MD = mean difference; N= number of studies; PACQLQ = Paediatric Asthma Caregiver’s Quality of Life Questionnaire; PAQLQ = Paediatric Asthma Quality of Life Questionnaire; SD = standard deviation

Table 3. Subgroup and sensitivity analyses

Outcome	N	SD calculation method	All studies	Non‐chlorinated / ventilated chlorinated pool		Chlorinated pool		Unknown chlorination status
Outcome	N	SD calculation method	Effect size [95% CI]	N	Effect size [95% CI]	N	Effect size [95% CI]	N	Effect size [95% CI]
MD FEV1 L	3	Change (Correlation from Weisgerber 2003)	MD 0.11 [0.01, 0.22]	2	0.16 [0.02, 0.30]	0		1	0.05 [‐0.12, 0.21]
MD FEV1 L	4	Absolute	MD 0.20 [0.05, 0.35]	2	0.24 [0.04, 0.44]	0		2	0.15 [‐0.08, 0.37]
MD FEV1 % predicted	4	Change (Correlation from Weisgerber 2003)	MD 8.47 [3.60, 13.33]	2	6.40 [0.29, 12.50]	1	11.82 [3.48, 20.16]	1	15.70 [‐15.48, 46.88]
MD FEV1 % predicted	4	Absolute	MD 2.39 [‐3.90, 8.69]	2	‐0.38 [‐8.08, 7.32]	1	10.97 [‐3.35, 25.29]	1	3.80 [‐13.11, 20.71]
MD FVC L	3	Change (Correlation from Weisgerber 2008)	MD 0.11 [‐0.06, 0.28]
	3	Change (Correlation from Weisgerber 2003)	MD 0.09 [‐0.00, 0.19]	2	0.20 [0.07, 0.33]	0		1	‐0.05 [‐0.19, 0.10]
	4	Absolute	0.31 [0.12, 0.50]	2	0.33 [0.07, 0.59]	0		2	0.28 [‐0.00, 0.56]
MD FVC% predicted	5	Change (Correlation from Weisgerber 2008)	MD 3.85 [‐0.58, 8.28]	3	2.00 [‐2.13, 6.14]	1	8.89 [2.65, 15.12]	1	16.80 [‐27.63, 61.23]
	5	Change (Correlation from Weisgerber 2003)	MD 4.53 [0.26, 8.80]	3	2.89 [‐2.00, 7.78]	1	8.89 [2.65, 15.12]	1	16.80 [‐27.63, 61.23]
	5	Absolute	MD 6.76 [‐0.93, 14.46]	3	5.28 [‐6.96, 17.52]	1	9.27 [‐4.34, 22.89]	1	8.20 [‐6.84, 23.24]
MD FEF 25‐75 % predicted	4	Change (Correlation from Weisgerber 2008)	12.63 [2.73, 22.53]	2	8.75 [‐7.61, 25.10]	1	19.02 [3.26, 34.78]	1	18.07 [0.52, 35.62]
	4	Change (Correlation from Weisgerber 2003)	12.47 [1.12, 23.82]	2	8.59 [‐7.76, 24.95]	1	19.02 [3.26, 34.78]	1	18.07 [0.52, 35.62]
	4	Absolute	12.40 [5.30, 19.49]

FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; MD = mean difference; N= number of studies; SD = standard deviation

Results

Description of studies

Results of the search

See Figure 1

Figure 1

PRISMA flow diagram

There were 1520 citations identified from the initial search of the pre‐specified databases but no relevant studies from trial registers. Three additional articles were retrieved following handsearching of reference lists of potentially relevant studies and another two were added from studies included in a Cochrane Systematic Review entitled 'Physical training for asthma' (Chandratilleke 2012)'. There were a total of 1367 citations after duplicates were removed, of which 1242 were removed after title screening by two review authors. The remaining 125 articles were screened by two review authors based on their abstracts. From this, 40 were identified as potentially relevant and full text articles were retrieved. Additional data on eligibility were requested from trial authors if required, following which at least two review authors independently agreed that eight articles fulfilled the study inclusion criteria of the review (Altintas 2003; Matsumoto 1999; Varray 1991; Varray 1995; Wang 2009; Weisgerber 2003; Weisgerber 2008; Wicher 2010). The remaining 32 citations were excluded and reasons for exclusion have been specified. See Characteristics of excluded studies.

Included studies

See Characteristics of included studies and Table 4 and Table 5.

Table 4. Comparison of baseline characteristics of included studies

ID/Location/ full publication	n randomised/ Withdrawals	% male	Mean age (range)	Asthma diagnosis	Asthma severity
Varray 1991 /France/ yes	14/0	86%	11 (9‐13)	Atopy, BDR, clinical	Not known
Varray 1995/ France/ yes	18/0	78%	11 (9‐13)	Atopy, BDR, clinical	Not known
Matsumoto 1999/ Japan/ yes	16/0	88%	11 (8‐12)	ATS criteria + hospital admission	Severe
Altintas 2003/ Turkey/ no	26/0	50%	10 (5‐14)	GINA criteria	moderate
Weisgerber 2003/ USA/ yes	26/16	50%	8(7‐14)	ATS criteria	Moderate persistent
Weisgerber 2008/ USA/ yes	78 recruited /61 commenced intervention/ 16 withdrawals	44%	10 (7‐14)	NAEPP/NHLBI criteria	Mild, moderate, or severe persistent
Wang 2009/ Taiwan/ yes	30/0	67%	10 (7‐12)	ATS criteria	Mild, moderate, or severe persistent
Wicher 2010/ Brazil/ yes	71/10	44%	10 (7‐18)	GINA criteria	moderate persistent

ATS = American Thoracic Society; BDR = bronchodilator response; EIB = exercise induced bronchoconstriction; FEF = forced expiratory flow; FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; GINA = Global Initiative for Asthma; hr = hour, ICS = inhaled corticosteroids; NAEPP = National Asthma Education and Prevention Program; NHLBI = National Heart, Lung, and Blood Institute; PEF = peak expiratory flow; QOL = quality of life; Tx = treatment;

See: Summary of findings for the main comparison Swimming training for asthma in children and adolescents aged 18 years and under

Table 5. Comparision of intervention characteristics of included studies

ID/Location/ full publication	Asthma treatment	Swimming time/frequency per week/n weeks/ pool type. Swimming supervision.	Comparison group	Follow‐up point in weeks/ outcomes in meta‐analysis
Varray 1991 /France/ yes	not known	60 minutes/ 2/12/ indoor. Supervisor: physical education teacher.	usual care	12 / FEV1, FVC, FVC%pr, VO2max, VEmax,
Varray 1995/ France/ yes	regular Tx	60 minutes/2/ 12/ indoor. Supervisor: physical education teacher.	usual care	12/ VO2 max,
Matsumoto 1999/ Japan/ yes	regular Tx/ 50% ICS, 75% theophylline, 38% cromoglycate	30 minutes/6/6/ indoor heated. Supervisor: unknown.	usual care	6/ aerobic capacity, EIB
Altintas 2003/ Turkey/ no	100‐400mcg ICS	90/3/ 6/ indoor non‐chlorinated. Supervisor: unknown.	usual care	6/ physical work capacity, walk test, PEF % pred, FEV1 % pred, FVC % pred.
Weisgerber 2003/ USA/ yes	regular Tx	45 minutes/2/6/ indoor. Supervisor: certified swim instructor unaware of study participation.	usual care	6‐8 / FEV1, FEV1 % pred, FVC, FVC % pred, FEF25‐75%, PEF, Symptoms
Weisgerber 2008/ USA/ yes	Regular Tx, ICS 52% swim, 65% golf	60 minutes/3/9/ indoor, chlorinated. Supervisors: certified swim instructors who had attended 1‐hr training seminar outlining asthma safety.	Golf sessions 60 minutes/ 3/ 9	9 /QOL, symptoms, exercise capacity,healthcare utilisation, FEV1, FEV1% pred, FVC, FVC% pred, FEF 25‐75%,
Wang 2009/ Taiwan/ yes	regular Tx	30 minutes/ 3/ 6/ outdoor non‐chlorinated. supervisor: certified swimming instructors not aware of involvement in the study.	usual care	6/ FEV1, FEV1% pred, FVC, FVC% pred, FEF50%, FEF 25‐75%, PEF,
Wicher 2010/ Brazil/ yes	all ICS 1000mcg/day, LABA BD	60 minutes/ 2/ 12/ indoor, ventilated, chlorinated. Supervisor: unknown.	usual care	12/ FEV1, FEV1% pred, FVC. FEF 25‐75%,

BDR = bronchodilator response; EIB = exercise induced bronchoconstriction; FEF = forced expiratory flow; FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; hr = hour, ICS = inhaled corticosteroids; PEF = peak expiratory flow; QOL = quality of life; Tx = treatment;

Eight studies involving 262 children or adolescents, published between 1991 and 2010, met the inclusion criteria. Sample sizes ranged from 14 to 71 participants. All studies were randomised controlled trials of children or adolescents with asthma, with swimming training in the intervention group. Swimming training programmes varied; in seven studies swimming sessions lasted for between 30 and 90 minutes, occurring two to three times a week with the length of the swimming training programme varying from six to 12 weeks. In Matsumoto 1999 the sessions were 30 minutes long but occurred six times a week for participants with severe asthma who had been recruited after an inpatient exacerbation. Seven pools were located indoors; one pool was chlorinated (Weisgerber 2008), one pool was chlorinated and well‐ventilated (Wicher 2010) and one pool was non‐chlorinated (Altintas 2003). In four studies chlorination status was not unspecified (Matsumoto 1999; Varray 1991; Varray 1995; Weisgerber 2003). The pool was outdoor and non‐chlorinated in Wang 2009.

The comparison group was usual care without any intervention in seven studies. One study compared golf sessions for the equivalent time to the swimming group (Weisgerber 2008).

The lowest age of participants was five years (Altintas 2003). Two studies recruited children from seven or eight to 12 years (Matsumoto 1999; Wang 2009), while the upper age of participants was 13 or 14 years in five studies (Altintas 2003; Varray 1991; Varray 1995; Weisgerber 2003; Weisgerber 2008) and 18 years in one trial (Wicher 2010). The proportion of male participants varied between 44% to 88%.

Asthma diagnosis was based on guidelines specified criteria in six studies (Altintas 2003; Matsumoto 1999; Wang 2009; Weisgerber 2003; Weisgerber 2008; Wicher 2010) and in these studies asthma was specified as persistent with severity graded as severe in Matsumoto 1999, moderate in Wicher 2010 and either mild/moderate/severe in Wang 2009; Weisgerber 2003; Weisgerber 2008. Diagnosis of asthma was based on clinical criteria with atopy and bronchodilator responsiveness in two studies (Varray 1991; Varray 1995) which did not specify asthma severity.

Participants' regular medication for asthma was continued during the study in six studies but three did not specify medications used (Varray 1995; Wang 2009; Weisgerber 2003). Participants used inhaled corticosteroids in Matsumoto 1999 (52%), Weisgerber 2008 (50%). All participants in Altintas 2003 were treated with inhaled corticosteroids (100‐400 mcg beclomethasone equivalent/day) and in Wicher 2010, all participants were treated with both inhaled corticosteroids (1000 mcg beclomethasone equivalent/day) and long‐acting beta‐agonists (formoterol 12 mcg two times a day).

Excluded studies

Thirty‐two studies were excluded with reasons provided in Characteristics of excluded studies table. Two studies reported insufficient details about the methods of randomisation, or reported an inadequate method of allocation (Fitch 1976; Huang 1989).

Risk of bias in included studies

Assessment of study quality was limited by incomplete data (Figure 2).

Figure 2

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

Allocation

All eight included studies were randomly allocated but of these only four had detailed descriptions regarding the method of randomisation and were judged to be at low risk of bias (Altintas 2003; Varray 1995; Weisgerber 2003; Weisgerber 2008). Only Weisgerber 2008 and Altintas 2003 supplied details regarding allocation concealment, and the other six studies were classified as unclear risk of bias for allocation.

Blinding

Due to the nature of the study interventions blinding of participants was not possible. All studies were classified as being of high risk for performance bias. No studies specified blinding of outcome assessors although in Matsumoto 1999 it is likely they were aware of group allocation.

Incomplete outcome data

Three studies with a high withdrawal rate were judged to be at high risk of bias (Weisgerber 2003; Weisgerber 2008;Wicher 2010). The remaining four studies were judged to be at low risk of bias.

Selective reporting

Although study protocols were not available, three studies were judged at low risk of bias. Authors of three studies responded to requests for study outcome data; Altintas 2003 supplied group results, and individual patient data were supplied for Weisgerber 2003 and Weisgerber 2008. Five studies were judged at unclear risk of bias.

Other potential sources of bias

No other source of bias were identified.

Effects of interventions

Primary outcomes

Quality of life

Only one study involving 50 participants (Weisgerber 2008) used a validated assessment tool to measure quality of life. This study reported the Pediatric Asthma Quality of Life Questionnaire (PAQLQ) (Juniper 1996); total (23 items, 1 = maximal impairment, 7 = no impairment) measuring overall functional problems for an individual with asthma (physical symptoms, emotional and social). Changes in parental quality of life were measured using the Pediatric Asthma Caregiver’s Quality of Life Questionnaire (Juniper 1996a) (PACQLQ) measuring the problems that are most troublesome to the parents (primary caregivers) of children with asthma (13 items in two domains, activity limitation and emotional function and overall score, seven‐point scale, 1 = maximal impairment, 7 = no impairment).

In Weisgerber 2008, no significant differences were found between swimming and golf groups in quality of life for children PAQLQ total (mean difference (MD) 0.26; 95% confidence interval (CI) ‐1.05 to 1.58), or in the PAQLQ symptom domain (MD 0.07; 95% CI ‐1.12 to 1.26), or for caregivers' quality of life, PACQLQ total MD 0.71; 95% CI ‐0.83 to 2.25 (Analysis 1.1).

Asthma control

Four studies (Varray 1991; Weisgerber 2003; Weisgerber 2008; Wang 2009) measured the impact of swimming on asthma symptoms. Only one study (Weisgerber 2008) used a validated assessment tool. This study reported the Living with Asthma Questionnaire Index (LWAQ index, Hyland 1991) (three items for parental report of asthma symptoms; high score indicating worse effects). In a comparison of swimming and golf exercise groups for this study (n = 50), no significant difference was found for symptoms (MD ‐0.10 LWAQ index: 95% CI ‐2.55 to 2.36) (Analysis 1.2).

Weisgerber 2003 (n = 8) used a non validated questionnaire (score 4 better‐16 worse) that assessed nocturnal coughing, daytime asthma symptoms, effect on activity and use of rescue inhaler. There was no significant difference in scores between swimming and control groups, (MD ‐0.80; 95% CI ‐4.64 to 3.04) (Analysis 1.2). In Wang 2009, asthma severity was assessed using the National Heart, Lung, and Blood Institute (NHLBI) criteria (symptoms: nocturnal awakening, rescue use, activity limitation). They reported a significant improvement in asthma severity post‐intervention in the swimming group compared with the control group but did not show data. Varray 1991 noted symptom status in a qualitative manner with parents reporting participants in the swimming group did not have any decrease in frequency of wheezing attacks or use of regular asthma medication.

We combined the results in a meta‐analysis from Weisgerber 2003 and Weisgerber 2008 (n = 58), in which symptoms were measured on differing scales, and found no statistically significant difference between swimming and control groups, (standardised mean difference (SMD) ‐0.06; 95% CI ‐0.58 to 0.47) (Analysis 1.3; Figure 3).

Figure 3

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.3 Change in asthma symptoms (all).

Exacerbations

No data were available for meta‐analysis of number of exacerbations or exacerbation frequency. Weisgerber 2008 reported that five symptom exacerbations occurred during 700 person‐sessions of the swimming programme (7.1 per 1000 sessions) and one symptom exacerbation occurred during 425 person‐sessions of golf (2.4 per 1000 sessions). All episodes resolved with the use of bronchodilator and none required a clinic visit, emergency department (ED) visit, or hospitalisation. Wicher 2010 reported that no participant was admitted to hospital for asthma attacks during in the "run in" or during the training period in either the swimming or control group.

Corticosteroid use for exacerbations of asthma

None of the eight included studies reported data on oral steroid use during the study period.

Secondary outcomes

No studies reported results for bronchodilator use or time off education/work.

Preventer medication use

In Wicher 2010 all participants used fluticasone 500 mcg daily and authors reported that adherence to treatment with fluticasone and use of rescue salbutamol was similar in the swimming and control groups.

Utilisation of healthcare

Weisgerber 2008 assessed the number of times the child visited the physician's office/clinic or the ED for an asthma flare up in two months preceding the exercise intervention and during the two‐month intervention period. Prior to the study, 41% had an urgent asthma physician visit (n = 44) and 18% an urgent asthma ED visit. There was a statistically significant decrease in urgent asthma physician visits in the pooled single exercise cohort analysis. Participants averaged fewer urgent visits to the clinic for asthma exacerbations during the two months they participated in the intervention compared with the prior two months, mean ‐1.1; SD 3.3 (P = 0.04). In separate group analyses, the decrease in urgent asthma physician visits in the swimming group (n = 27) using the Wilcoxon signed‐rank test was statistically significant (P = 0.03) but not in the golf group (n = 17). However, a comparison of the swimming training and golf groups for urgent asthma physician visits over the two‐month study intervention (Analysis 1.4) did not show a significant difference (MD 0.08; 95% CI ‐0.25 to 0.42), and the likelihood of at least one urgent asthma physician visit was not significantly increased, (Peto odds ratio (OR) 1.64; 95% CI 0.32 to 8.44) (Analysis 1.5). No urgent asthma ED visits occurred in the golf group during the intervention, but the likelihood was not significantly different between swimming training and golf groups (Peto OR 5.77; 95% CI 0.72 to 46.46) (Analysis 1.5).

Lung function

Lung function was assessed at baseline and on completion of the intervention or control period.

FEV1: Four studies (n = 113) contributed data on FEV1 (Varray 1991; Wang 2009; Weisgerber 2003; Wicher 2010). The change in FEV1 for swimming training compared with control was small and of borderline statistical significance (MD 0.10 L; 95% CI ‐0.00 to 0.20), with low heterogeneity I² = 32% (Analysis 1.6). Results for FEV1 % predicted in Wicher 2010 were not included in the meta‐analyses as they did not appear compatible with FEV1 results. We requested that the trial authors check their accuracy but no response was received so data were not incorporated into the meta‐analysis. In four studies (Altintas 2003; Wang 2009; Weisgerber 2003; Weisgerber 2008), FEV1 % predicted (n = 83) was significantly greater in the swimming group (MD 8.07; 95% CI 3.59 to 12.54), with moderate heterogeneity I² = 38% (Analysis 1.7; Figure 4).

Figure 4

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.7 FEV1 % predicted (change).

FVC: Four studies (n = 113) contributed data on FVC (Varray 1991; Wang 2009; Weisgerber 2003; Wicher 2010). In a random‐effects meta‐analysis, there was a small, statistically non‐significant difference in FVC for swimming training compared with control (MD 0.10 L; 95% CI ‐0.07 to 0.26), with substantial heterogeneity (I² = 57%) (Analysis 1.8). The difference in FVC % predicted in five studies (Altintas 2003; Wang 2009; Weisgerber 2003; Weisgerber 2008; Wicher 2010, n = 144) was not statistically significant (MD 3.85%; 95% CI ‐0.58 to 8.28), with substantial heterogeneity present (I² = 61%) (Analysis 1.9).

FEF 25‐75: Four studies reported FEF 25% to 75% as per cent predicted (Wang 2009; Weisgerber 2003; Weisgerber 2008; Wicher 2010) but only Weisgerber 2003 reported absolute values. Random‐effects pooled results indicated that swimming training had a significant effect on FEF25% to 75% predicted (MD 12.63; 95% CI 2.73 to 22.53; n = 118) with substantial heterogeneity, (I² = 59%) (Analysis 1.10). The difference between groups when measured in volume was not statistically significant, (MD 0.28 L; 95% CI‐0.15 to 0.72; one study, n = 8) (Analysis 1.11).

PEF (L/min): Peak flow was assessed in Wang 2009 and Weisgerber 2003 (n = 38) and the pooled result demonstrated substantial heterogeneity (I² = 59%). The random‐effects meta‐analysis favoured the swimming group compared with control (MD 62.07 L/min; 95% CI 22.84 to 101.30) (Analysis 1.13).

Exercise capacity and fitness

Five studies assessed the effects of swimming on exercise capacity and fitness (Altintas 2003; Matsumoto 1999; Varray 1991; Varray 1995; Weisgerber 2008), but they used a variety of different measures. The accepted gold standard for fitness testing is maximal oxygen consumption in mL/kg/min (VO₂max) during a maximal effort test in the exercise laboratory. Index Physical Work Capacity (PWC)170 is the work load performed on any type of ergometer resulting in a heart rate of 170/min and is highly correlated to VO₂max. Other distance‐based tests have been developed to meet the need for simpler, inexpensive ways to assess aerobic fitness in children, such as the 20‐metre shuttle run, the five‐minute run, the six‐minute run, the 15‐minute run, the one‐mile run, and Cooper 12‐minute walk/run test (CT12).

VO₂max (mL/kg/min)

Varray 1991 and Varray 1995 conducted measurements in children (n = 32). Swimming training compared with usual care had a positive and significant effect on VO₂ max (MD 9.67 mL/kg/min; 95% CI 5.84 to 13.51), with no heterogeneity (I² = 0%) (Analysis 1.14). Weisgerber 2008 compared swimming training and golf and undertook fitness testing for a subset of 19 of 45 participating children and adolescents, with no significant difference (MD ‐7.00 mL/kg/min; 95% CI ‐14.57 to 0.57) (Analysis 1.14). Studies with usual care or golf control groups were not pooled in view of the high heterogeneity, Chi² = 14.82, df = 2 (P = 0.0006); I² = 93% (Figure 5).

Figure 5

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.14 Exercise capacity: VO2 max (mL/kg/min).

Other measures of exercise capacity

Altintas 2003 measured Index PWC170 – the work load in watts performed on an ergometer (treadmill, cyclo ergometer) that will result with a heart rate of 170/min. The swimming group had a significantly greater PWC170 compared with the control group (MD 0.44 watts; 95% CI 0.13 to 0.75) (Analysis 1.15).

Matsumoto 1999 (n = 16) measured aerobic capacity, defined as the work load at lactate threshold and found aerobic capacity significantly increased in all participants in the swimming training group. The difference for the swimming group compared with control for swimming ergometry was MD 0.22 kp; 95% CI 0.10 to 0.34 and for cycle ergometry (MD 6.80 kp; 95% CI 2.03 to 11.57) (Analysis 1.15).

Results from four studies with a usual care control group (n = 74) that measured exercise capacity were pooled; Altintas 2003 (PWC170, Matsumoto 1999 (cycle ergometry), Varray 1991 and Varray 1995 (VO₂max) (Analysis 1.16; Figure 6). A difference of equivalent magnitude to the pooled VO₂max result was found, with a SMD (SMD) 1.34 ; 95% CI 0.82 to 1.86, and no heterogeneity (I² = 0%).

Figure 6

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.16 Exercise Capacity: Any measure (control: usual care).

Field fitness tests

Altintas 2003 (n = 26) found no significant difference between swimming training and usual care control groups in the six‐minute walk test (MD 38.64 metres; 95% CI ‐9.07 to 86.35) (Analysis 1.17). Weisgerber 2008 found no significant difference in the Coopers 12‐minute walk‐run test between swimming training and golf groups (MD ‐112.93; 95% CI ‐643.28 to 417.41) (Analysis 1.17). Pooling both studies using a SMD did not show any significant difference between swimming training compared with control, SMD 0.15; 95% CI ‐0.34 to 0.63, with moderate heterogeneity, I² = 52% (Analysis 1.18).

Adverse effects

Bronchial hyper‐responsiveness

Two individual studies (Matsumoto 1999; Wicher 2010) reported results for formal direct challenge tests but results in publications were incomplete and they could not be included in a meta‐analysis. Matsumoto 1999 reported that the difference between the mean change in provocative concentration of histamine (PC20) causing a 20% fall in FEV1 in the training and control groups was not statistically significant (P = 0.16). Wicher 2010 (n = 61) measured the provocative concentration of methacholine PC20 causing a 20% fall in FEV1. Results for the groups separately were reported as showing a significant change in the swimming group pre‐ post‐training, PC20 0.31 mg/mL (SD 0.25) before to 0.63 mg/mL (SD 0.78), P = 0.008. At the end of the study there was no significant difference between swimming and control groups (MD 0.41 Ln PC20; 95% CI ‐0.70 to1.51) (Analysis 1.19).

Exercise induced bronchoconstriction

Matsumoto 1999 assessed exercise induced bronchoconstriction using swimming and cycle ergometry. The mean maximal percentage fall in FEV1 induced with the swimming and cycle ergometers' work load set to 175% LT on the relative load was measured. Although a smaller decrease in FEV1 was seen in the swimming training group, the difference between swimming and control groups was not statistically significant (MD ‐4.00 %; 95% CI‐14.83 to 6.83) for swimming ergometry and (MD ‐4.99%; 95% CI ‐21.61 to 11.63) for cycle ergometry (Analysis 1.20).

Subgroup analysis

Pre‐specified subgroup analyses on children versus adolescents or observed/non‐physical/physical activity control groups were not conducted due to limited number of studies. Analyses of lung function measures comparing non‐chlorinated pools and ventilated chlorinated pools with chlorinated indoor pools was limited due to the small number of studies and missing outcome data (Table 3).

Sensitivity analysis

Change from baseline values were used in the outcomes, Pediatric Asthma Quality of Life Questionnaire (PAQLQ), Pediatric Asthma Caregiver's Quality of Life Questionnaire (PACQLQ) and Living with Asthma Questionnaire Index (LWAQ index), FEV1, FVC FEF 25% to 75%. Analyses using absolute values are shown for comparison in Table 2 and Table 3. Results did not differ by direction nor greatly in magnitude.

Correlation coefficients between baseline and final measurements were calculated from two studies (Weisgerber 2003 and Weisgerber 2008) and are shown for comparison in Table 1. We used the correlation coefficients from both to calculate and impute the standard deviation of change from baseline for lung function measures for Altintas 2003, Wang 2009 and Wicher 2010. We report effect sizes using values calculated from the larger study Weisgerber 2008 and included those calculated from Weisgerber 2003 in sensitivity analyses in Table 2 and Table 3. Although there were small variations in the effect sizes, the direction of difference did not change for any outcome.

Discussion

Summary of main results

This review set out to determine the effectiveness of swimming training as an intervention for asthma in children and adolescents. There were no studies comparing swimming training with a usual care non‐active control group for the primary outcomes, quality of life, asthma exacerbations or use of corticosteroids for asthma. No significant benefit on asthma symptoms was seen in studies comparing swimming training with any control group (summary of findings Table for the main comparison). When swimming training was compared with an active control group in one study (Weisgerber 2008), there were no differences in quality of life for children and caregivers, asthma control, exacerbations of asthma and asthma‐related healthcare utilisation.

There were statistically significant benefits for swimming training on resting lung function measured at the conclusion of the swimming or control period, for FEV1, FVC and FEF 25% to 75%, expressed as absolute or percentage predicted values. The difference in FEV1 of 100 mL between swimming and control groups is clinically meaningful and comparable to that found in children with asthma treated with inhaled fluticasone propionate 100 mcg (Adams 2008). There were significant benefits for swimming training on cardio‐pulmonary fitness compared to a non‐active control measured by maximum oxygen uptake. The 25% difference for swimming compared to control in VO₂ max is clinically meaningful. It exceeds the differences seen in a review of physical activity studies in children without asthma, range 5% to 15% (Armstrong 2011) and the 9% difference seen in children with asthma undertaking physical training (Counil 2003). When pooling VO₂ max with other measures of exercise capacity, the result was of similar magnitude.

Overall completeness and applicability of evidence

The eight included studies randomised 262 participants who commenced an intervention, while 42 participants withdrew early (Table 4; Table 5). The number of studies on which conclusions are based are relatively few thus limiting available data. Four studies (Altintas 2003; Matsumoto 1999; Varray 1991; Varray 1995) that focused on cardiopulmonary fitness measures reported few withdrawals. Three studies assessed asthma control and lung function (Wang 2009; Weisgerber 2003; Weisgerber 2008) while one study measured lung function and bronchial hyper‐responsiveness (Wicher 2010). Outcomes assessing cardiopulmonary fitness, asthma symptoms and lung function were measured at baseline and at the end of the intervention or control period.

A recently published update of a review comparing the effects of physical training in participants of all ages with asthma (Chandratilleke 2012) found exercise is well‐tolerated and there was no evidence of adverse effects on asthma symptoms. Our review sought to define the benefits, if any, specifically for swimming training. Since the 1970s, it has been suggested from observational studies that swimming training is not detrimental to asthma control (Fitch 1971; Fitch 1976), and benefits on asthma symptoms and control have been seen in pre ‐ post‐observational design studies (Huang 1989; Rothe 1990).

This review was unable to add to the evidence on potential harmful effects of chlorine on children and adolescents with asthma from swimming training in non‐ventilated pools (Bernard 2010) as the chlorination or ventilation status was not known in four studies, thus restricting use of subgroup analysis.

The practicality of the programmes used in the studies varied from a realistic two sessions per week to a programme with six sessions a week that would be much harder to emulate and sustain outside a study environment.

Quality of the evidence

The eight studies included were randomised controlled trials, although limited published information and lack of response to direct request, meant six studies were classified as at unclear risk of selection bias. No studies could be classified as at low risk for detection bias due to non‐blinding of outcome assessors. Attrition bias risk was classified as high in three studies, due to high withdrawal rates (Weisgerber 2003; Weisgerber 2008; Wicher 2010). In Weisgerber 2008 there was a high rate of withdrawal after randomisation but prior to commencing training in both groups (24% swimming, 17% golf). During the programme the withdrawal rate was 17% in the swimming group and 35% in the golf group. This study only assessed some outcomes (lung function and fitness tests) on a subgroup of participants (31% swimming, 31% golf). No studies were assessed as at high risk of reporting bias (Figure 2).

Using the GRADE criteria (summary of findings Table for the main comparison), the strength of the evidence from this review for the primary outcomes, quality of life and asthma control is low and further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. For the secondary outcomes, urgent asthma visits to a physician or lung function, the strength of the evidence is moderate and it is likely that further research will have an important impact on our confidence in the estimates of effect and may change the estimates. For the secondary outcome exercise capacity, the strength of the evidence from this review is graded high and further research is very unlikely to change our confidence in the estimate of effect.

Potential biases in the review process

Individual patient data for Weisgerber 2008 were analysed and verified by authors (JW, YCF). Correlation coefficients were calculated using the method of Abrams 2005 from studies where baseline and final measurements were available (Weisgerber 2003; Weisgerber 2008) Table 1. The coefficients were used to impute a change‐from‐baseline standard deviation where this was not reported (Altintas 2003; Wang 2009; Wicher 2010). Sensitivity analysis was undertaken using coefficients from both studies. The significance or direction of effect sizes did not differ in sensitivity analysis. Study effects reported in this review are based on imputation using coefficients calculated from the large study Weisgerber 2008 where these were available or Weisgerber 2003 for other outcomes.

Agreements and disagreements with other studies or reviews

The conclusions of this review support physical training benefits found in a systematic review comparing all forms of physical training for people with asthma of any age (Chandratilleke 2012). It is also consistent with the observational evidence of Font‐Ribera 2011 in a prospective longitudinal study following 5738 British children from birth until age 10 years, which found that swimming did not increase the risk of asthma and that swimming was associated with improved lung function and fewer respiratory symptoms, particularly among children with asthma.

Figure 1

PRISMA flow diagram

Figure 2

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

Figure 3

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.3 Change in asthma symptoms (all).

Figure 4

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.7 FEV1 % predicted (change).

Figure 5

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.14 Exercise capacity: VO2 max (mL/kg/min).

Figure 6

Forest plot of comparison: 1 Swimming training versus control, outcome: 1.16 Exercise Capacity: Any measure (control: usual care).

Analysis 1.1

Comparison 1 Swimming training versus control, Outcome 1 Quality of life.

Analysis 1.2

Comparison 1 Swimming training versus control, Outcome 2 Symptoms (change).

Analysis 1.3

Comparison 1 Swimming training versus control, Outcome 3 Change in asthma symptoms (all).

Analysis 1.4

Comparison 1 Swimming training versus control, Outcome 4 Urgent asthma physician visits.

Analysis 1.5

Comparison 1 Swimming training versus control, Outcome 5 Asthma consultation (1 or more) during intervention.

Analysis 1.6

Comparison 1 Swimming training versus control, Outcome 6 FEV1 L.

Analysis 1.7

Comparison 1 Swimming training versus control, Outcome 7 FEV1 % predicted (change).

Analysis 1.8

Comparison 1 Swimming training versus control, Outcome 8 FVC L.

Analysis 1.9

Comparison 1 Swimming training versus control, Outcome 9 FVC % predicted (change).

Analysis 1.10

Comparison 1 Swimming training versus control, Outcome 10 FEF 25% to 75 % predicted (change).

Analysis 1.11

Comparison 1 Swimming training versus control, Outcome 11 FEF 25% to 75% L.

Analysis 1.12

Comparison 1 Swimming training versus control, Outcome 12 FEF 50 % predicted.

Analysis 1.13

Comparison 1 Swimming training versus control, Outcome 13 PEF L/min.

Analysis 1.14

Comparison 1 Swimming training versus control, Outcome 14 Exercise capacity: VO2 max (mL/kg/min).

Analysis 1.15

Comparison 1 Swimming training versus control, Outcome 15 Exercise Capacity: other measures (control: usual care).

Analysis 1.16

Comparison 1 Swimming training versus control, Outcome 16 Exercise Capacity: Any measure (control: usual care).

Analysis 1.17

Comparison 1 Swimming training versus control, Outcome 17 Distance fitness tests‐all (m).

Analysis 1.18

Comparison 1 Swimming training versus control, Outcome 18 Distance fitness tests‐all (m).

Analysis 1.19

Comparison 1 Swimming training versus control, Outcome 19 Bronchial hyper‐responsiveness: ln PC20 methacholine.

Analysis 1.20

Comparison 1 Swimming training versus control, Outcome 20 Exercise induced bronchoconstriction (maximum fall in FEV1 (%).

Summary of findings for the main comparison. Swimming training for asthma in children and adolescents aged 18 years and under

Swimming training for asthma in children and adolescents aged 18 years and under
Patient or population: children and adolescents aged 18 years and under studies with asthma Settings: Recruited from asthma clinics. Asthma diagnosis by recognised criteria. Intervention: Swimming training programme‐ meeting minimum intensity criteria (> Weekly, > 20 minutes, > 4 weeks)
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Swimming training
Quality of life¹ Paediatric Asthma Quality of Life Questionnaire (PAQLQ). Scale from: 1 (worse) to 7 (better). Follow‐up: mean 9 weeks	The mean change in quality of life in the control group was ‐1.87	The mean change in quality of life in the intervention group was 0.26 (1.05 lower to 1.58 higher)		50 (1 study¹)	⊕⊕⊝⊝ low^2,3
Asthma symptoms Different scales in different studies (lower scores mean fewer symptoms) Follow‐up: 6‐9 weeks	The mean change in asthma symptoms ranged across control groups from 0 to ‐2.14 standard deviations	The mean asthma symptoms in the intervention groups was 0.06 standard deviations less (0.58 lower to 0.47 higher) see comment		58 (2 studies)	⊕⊕⊝⊝ low^3,4,5	The difference of 0.06 standard deviations would equate to a small difference on Living with Asthma Questionnaire (LWAQ) or a composite 12‐point scale of < 0.5 units. The effect size is < 0.2 representing a small effect.
Exacerbations requiring hospital admission	see comment	see comment	see comment	see comment	‐	Outcome not reported
Exacerbations requiring a course of oral corticosteroids	see comment	see comment	see comment	see comment	‐	Outcome not reported
Urgent asthma physician visits¹ Number of times the child visited a physician's office/clinic for an asthma flare up Follow‐up: mean 2 months	The mean urgent asthma physician visits in the control group was 0.17 visits in 2 months	The mean urgent asthma physician visits in the intervention groups was 0.08 higher (0.25 lower to 0.42 higher)		44 (1 study¹)	⊕⊕⊕⊝ moderate³
Resting lung function Forced expiratory volume (FEV1) in 1 second (litres) Follow‐up: 6‐12 weeks	The mean change in resting FEV1 ranged across control groups from 0.05‐0.15 litres	The mean difference in FEV1 in the intervention groups was 0.10 L higher (0 to 0.2 higher)		113 (4 studies)	⊕⊕⊕⊝ moderate³	The mean difference is comparable to the difference in FEV1 in children with asthma (N = 4, n = 719 ⁷) comparing low dose fluticasone propionate (100 mcg) daily with placebo mean difference (MD) 0.1 L [0.15, 0.36] (Adams 2008)
Fitness⁶ Maximal oxygen consumption (VO2 max) Follow‐up: mean 12 weeks	The mean VO₂ max in usual care control groups was 39 mL/kg/min	The mean fitness in the swimming intervention groups was 9.67 mL/kg/min higher (5.84 to 13.51 higher)		32 (2 studies)	⊕⊕⊕⊕ high	The 25% difference for swimming compared to control in VO₂ max is clinically meaningful. It is larger than the differences seen in physical activity studies in children without asthma, range 5% to 15% (Armstrong 2011) and that seen in children with asthma undertaking physical training 9% (Counil 2003).
The basis for the assumed risk* is provided in the table. The corresponding risk (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; FEV1 forced expiratory volume in one second; L: litres;
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Control group: golf ² High risk of attrition bias assessed in study ³ The confidence interval does not rule out a null effect or harm ⁴ Comparison groups differed; usual care or golf ⁵ Effect size < 0.2 represents small effect ⁶ Pooled studies with non‐active usual care control group only ⁷ N= number of studies; n= number of participant

Summary of findings for the main comparison. Swimming training for asthma in children and adolescents aged 18 years and under

Table 1. Correlation coefficients: from studies for imputing standard deviation of mean changes

Weisgerber 2008	Swimming group	Golf group
FEV1 % predicted	0.6491	0.9179
FVC % predicted	0.8722	0.9064
FEF 25% to 75% % predicted	0.3823	0.9234
Coopers test	0.7655	0.8588
Peak heart rate	0.9259	0.3264
Exercise time	0.8200	0.8584
Weisgerber 2003	Swimming group	Usual care group
FEV1 L	0.5983	0.9433
FEV1 % predicted	0.5585	0.8495
FVC (L)	0.7779	0.9781
FVC % predicted	0.9229	0.4247
PEF (L/s)	0.5115	0.7406
PEF % predicted	0.0319	0.9062
FEF 25% to 75% % predicted	0.8634	0.8904
FEF = forced expiratory flow; FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; PEF = peak expiratory flow

Table 1. Correlation coefficients: from studies for imputing standard deviation of mean changes

Table 2. Sensitivity analyses

Outcome	N	SD calculation method	Effect size [95% CI]
Quality of life (PAQLQ)	1	Change (no imputation)	MD 0.26 [‐1.05, 1.58]
Quality of life (PAQLQ)	1	Absolute	MD 0.16 [‐1.17, 1.50]
Symptoms (PAQLQ)	1	Change (no imputation)	MD 0.07 [‐1.12, 1.26]
Symptoms (PAQLQ)	1	Absolute	MD 0.12 [‐1.20, 1.43]
Caregiver quality of life (PACQLQ)	1	Change (no imputation)	MD 0.71 [‐0.83, 2.25]
Caregiver quality of life (PACQLQ)	1	Absolute	MD 0.61 [‐0.86, 2.07]
Asthma symptoms (LWAQ index)	1	Change (no imputation)	MD ‐0.10 [‐2.55, 2.36]
Asthma symptoms (LWAQ index)	1	Absolute	MD 0.83 [‐1.44, 3.11]
Exercise Capacity: Any measure (control: usual care)	4	Cycle ergometry (Watts)	MD 1.34 [0.82, 1.86]
Exercise Capacity: Any measure (control: usual care)	4	Swimming ergometry (kp)	MD 1.41 [0.88, 1.94]
LWAQ = Living with Asthma Questionnaire; MD = mean difference; N= number of studies; PACQLQ = Paediatric Asthma Caregiver’s Quality of Life Questionnaire; PAQLQ = Paediatric Asthma Quality of Life Questionnaire; SD = standard deviation

Table 2. Sensitivity analyses

Table 3. Subgroup and sensitivity analyses

Outcome	N	SD calculation method	All studies	Non‐chlorinated / ventilated chlorinated pool		Chlorinated pool		Unknown chlorination status
Outcome	N	SD calculation method	Effect size [95% CI]	N	Effect size [95% CI]	N	Effect size [95% CI]	N	Effect size [95% CI]
MD FEV1 L	3	Change (Correlation from Weisgerber 2003)	MD 0.11 [0.01, 0.22]	2	0.16 [0.02, 0.30]	0		1	0.05 [‐0.12, 0.21]
MD FEV1 L	4	Absolute	MD 0.20 [0.05, 0.35]	2	0.24 [0.04, 0.44]	0		2	0.15 [‐0.08, 0.37]
MD FEV1 % predicted	4	Change (Correlation from Weisgerber 2003)	MD 8.47 [3.60, 13.33]	2	6.40 [0.29, 12.50]	1	11.82 [3.48, 20.16]	1	15.70 [‐15.48, 46.88]
MD FEV1 % predicted	4	Absolute	MD 2.39 [‐3.90, 8.69]	2	‐0.38 [‐8.08, 7.32]	1	10.97 [‐3.35, 25.29]	1	3.80 [‐13.11, 20.71]
MD FVC L	3	Change (Correlation from Weisgerber 2008)	MD 0.11 [‐0.06, 0.28]
	3	Change (Correlation from Weisgerber 2003)	MD 0.09 [‐0.00, 0.19]	2	0.20 [0.07, 0.33]	0		1	‐0.05 [‐0.19, 0.10]
	4	Absolute	0.31 [0.12, 0.50]	2	0.33 [0.07, 0.59]	0		2	0.28 [‐0.00, 0.56]
MD FVC% predicted	5	Change (Correlation from Weisgerber 2008)	MD 3.85 [‐0.58, 8.28]	3	2.00 [‐2.13, 6.14]	1	8.89 [2.65, 15.12]	1	16.80 [‐27.63, 61.23]
	5	Change (Correlation from Weisgerber 2003)	MD 4.53 [0.26, 8.80]	3	2.89 [‐2.00, 7.78]	1	8.89 [2.65, 15.12]	1	16.80 [‐27.63, 61.23]
	5	Absolute	MD 6.76 [‐0.93, 14.46]	3	5.28 [‐6.96, 17.52]	1	9.27 [‐4.34, 22.89]	1	8.20 [‐6.84, 23.24]
MD FEF 25‐75 % predicted	4	Change (Correlation from Weisgerber 2008)	12.63 [2.73, 22.53]	2	8.75 [‐7.61, 25.10]	1	19.02 [3.26, 34.78]	1	18.07 [0.52, 35.62]
	4	Change (Correlation from Weisgerber 2003)	12.47 [1.12, 23.82]	2	8.59 [‐7.76, 24.95]	1	19.02 [3.26, 34.78]	1	18.07 [0.52, 35.62]
	4	Absolute	12.40 [5.30, 19.49]
FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; MD = mean difference; N= number of studies; SD = standard deviation

Table 3. Subgroup and sensitivity analyses

Table 4. Comparison of baseline characteristics of included studies

ID/Location/ full publication	n randomised/ Withdrawals	% male	Mean age (range)	Asthma diagnosis	Asthma severity
Varray 1991 /France/ yes	14/0	86%	11 (9‐13)	Atopy, BDR, clinical	Not known
Varray 1995/ France/ yes	18/0	78%	11 (9‐13)	Atopy, BDR, clinical	Not known
Matsumoto 1999/ Japan/ yes	16/0	88%	11 (8‐12)	ATS criteria + hospital admission	Severe
Altintas 2003/ Turkey/ no	26/0	50%	10 (5‐14)	GINA criteria	moderate
Weisgerber 2003/ USA/ yes	26/16	50%	8(7‐14)	ATS criteria	Moderate persistent
Weisgerber 2008/ USA/ yes	78 recruited /61 commenced intervention/ 16 withdrawals	44%	10 (7‐14)	NAEPP/NHLBI criteria	Mild, moderate, or severe persistent
Wang 2009/ Taiwan/ yes	30/0	67%	10 (7‐12)	ATS criteria	Mild, moderate, or severe persistent
Wicher 2010/ Brazil/ yes	71/10	44%	10 (7‐18)	GINA criteria	moderate persistent
ATS = American Thoracic Society; BDR = bronchodilator response; EIB = exercise induced bronchoconstriction; FEF = forced expiratory flow; FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; GINA = Global Initiative for Asthma; hr = hour, ICS = inhaled corticosteroids; NAEPP = National Asthma Education and Prevention Program; NHLBI = National Heart, Lung, and Blood Institute; PEF = peak expiratory flow; QOL = quality of life; Tx = treatment;

Table 4. Comparison of baseline characteristics of included studies

Table 5. Comparision of intervention characteristics of included studies

ID/Location/ full publication	Asthma treatment	Swimming time/frequency per week/n weeks/ pool type. Swimming supervision.	Comparison group	Follow‐up point in weeks/ outcomes in meta‐analysis
Varray 1991 /France/ yes	not known	60 minutes/ 2/12/ indoor. Supervisor: physical education teacher.	usual care	12 / FEV1, FVC, FVC%pr, VO2max, VEmax,
Varray 1995/ France/ yes	regular Tx	60 minutes/2/ 12/ indoor. Supervisor: physical education teacher.	usual care	12/ VO2 max,
Matsumoto 1999/ Japan/ yes	regular Tx/ 50% ICS, 75% theophylline, 38% cromoglycate	30 minutes/6/6/ indoor heated. Supervisor: unknown.	usual care	6/ aerobic capacity, EIB
Altintas 2003/ Turkey/ no	100‐400mcg ICS	90/3/ 6/ indoor non‐chlorinated. Supervisor: unknown.	usual care	6/ physical work capacity, walk test, PEF % pred, FEV1 % pred, FVC % pred.
Weisgerber 2003/ USA/ yes	regular Tx	45 minutes/2/6/ indoor. Supervisor: certified swim instructor unaware of study participation.	usual care	6‐8 / FEV1, FEV1 % pred, FVC, FVC % pred, FEF25‐75%, PEF, Symptoms
Weisgerber 2008/ USA/ yes	Regular Tx, ICS 52% swim, 65% golf	60 minutes/3/9/ indoor, chlorinated. Supervisors: certified swim instructors who had attended 1‐hr training seminar outlining asthma safety.	Golf sessions 60 minutes/ 3/ 9	9 /QOL, symptoms, exercise capacity,healthcare utilisation, FEV1, FEV1% pred, FVC, FVC% pred, FEF 25‐75%,
Wang 2009/ Taiwan/ yes	regular Tx	30 minutes/ 3/ 6/ outdoor non‐chlorinated. supervisor: certified swimming instructors not aware of involvement in the study.	usual care	6/ FEV1, FEV1% pred, FVC, FVC% pred, FEF50%, FEF 25‐75%, PEF,
Wicher 2010/ Brazil/ yes	all ICS 1000mcg/day, LABA BD	60 minutes/ 2/ 12/ indoor, ventilated, chlorinated. Supervisor: unknown.	usual care	12/ FEV1, FEV1% pred, FVC. FEF 25‐75%,
BDR = bronchodilator response; EIB = exercise induced bronchoconstriction; FEF = forced expiratory flow; FEV1= forced expiratory volume in 1 second; FVC = forced vital capacity; hr = hour, ICS = inhaled corticosteroids; PEF = peak expiratory flow; QOL = quality of life; Tx = treatment;

Table 5. Comparision of intervention characteristics of included studies

Comparison 1. Swimming training versus control

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Quality of life Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

1.1 PAQLQ child	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 PAQLQ symptom domain	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.3 PACQLQ parent	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Symptoms (change) Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

2.1 LWAQ index (control golf)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 Composite score (control usual care)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Change in asthma symptoms (all) Show forest plot	2	58	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.06 [‐0.58, 0.47]

3.1 Control usual care	1	8	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.29 [‐1.73, 1.16]
3.2 Control golf	1	50	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.02 [‐0.58, 0.54]
4 Urgent asthma physician visits Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

4.1 Control golf	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Asthma consultation (1 or more) during intervention Show forest plot	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	Totals not selected

5.1 Urgent asthma Physician visit (⋧1) during intervention	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 Urgent asthma ED visit (⋧1) during intervention	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 FEV1 L Show forest plot	4	113	Mean Difference (IV, Fixed, 95% CI)	0.10 [‐0.00, 0.20]

6.1 Absolute	1	14	Mean Difference (IV, Fixed, 95% CI)	‐0.08 [‐0.44, 0.28]
6.2 Change	3	99	Mean Difference (IV, Fixed, 95% CI)	0.11 [0.01, 0.22]
7 FEV1 % predicted (change) Show forest plot	4	83	Mean Difference (IV, Fixed, 95% CI)	8.07 [3.59, 12.54]

7.1 Control usual care	3	64	Mean Difference (IV, Fixed, 95% CI)	6.55 [1.24, 11.85]
7.2 Control golf	1	19	Mean Difference (IV, Fixed, 95% CI)	11.82 [3.48, 20.16]
8 FVC L Show forest plot	4	113	Mean Difference (IV, Random, 95% CI)	0.10 [‐0.07, 0.26]

8.1 Absolute	1	14	Mean Difference (IV, Random, 95% CI)	‐0.19 [‐0.95, 0.57]
8.2 Change	3	99	Mean Difference (IV, Random, 95% CI)	0.11 [‐0.06, 0.28]
9 FVC % predicted (change) Show forest plot	5	144	Mean Difference (IV, Random, 95% CI)	3.85 [‐0.58, 8.28]

9.1 Control usual care	4	125	Mean Difference (IV, Random, 95% CI)	2.00 [‐1.84, 5.84]
9.2 Control golf	1	19	Mean Difference (IV, Random, 95% CI)	8.89 [2.65, 15.12]
10 FEF 25% to 75 % predicted (change) Show forest plot	4	118	Mean Difference (IV, Random, 95% CI)	12.63 [2.73, 22.53]

10.1 Control usual care	3	99	Mean Difference (IV, Random, 95% CI)	11.07 [‐1.17, 23.30]
10.2 Control golf	1	19	Mean Difference (IV, Random, 95% CI)	19.02 [3.26, 34.78]
11 FEF 25% to 75% L Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

11.1 Absolute	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
12 FEF 50 % predicted Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

12.1 Absolute	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13 PEF L/min Show forest plot	2	38	Mean Difference (IV, Random, 95% CI)	62.07 [22.84, 101.30]

13.1 Absolute	2	38	Mean Difference (IV, Random, 95% CI)	62.07 [22.84, 101.30]
14 Exercise capacity: VO2 max (mL/kg/min) Show forest plot	3		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

14.1 Control: usual care	2	32	Mean Difference (IV, Fixed, 95% CI)	9.67 [5.84, 13.51]
14.2 Control: golf	1	19	Mean Difference (IV, Fixed, 95% CI)	‐7.00 [‐14.57, 0.57]
15 Exercise Capacity: other measures (control: usual care) Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

15.1 Physical Work Capacity (PWC170 watts)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
15.2 Swimming ergometry at lactic threshold (change kp)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
15.3 Cycle ergometry at lactic threshold (change watts)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
16 Exercise Capacity: Any measure (control: usual care) Show forest plot	4	74	Std. Mean Difference (IV, Fixed, 95% CI)	1.34 [0.82, 1.86]

16.1 VO₂max (mL/kg/min)	2	32	Std. Mean Difference (IV, Fixed, 95% CI)	1.65 [0.82, 2.49]
16.2 Physical Work Capacity (PWC170 watts)	1	26	Std. Mean Difference (IV, Fixed, 95% CI)	1.05 [0.22, 1.88]
16.3 Cycle ergometry at lactic threshold (change watts)	1	16	Std. Mean Difference (IV, Fixed, 95% CI)	1.32 [0.21, 2.44]
17 Distance fitness tests‐all (m) Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

17.1 Coopers 12min walk‐run distance (control usual care)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17.2 6min walk distance (control golf)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
18 Distance fitness tests‐all (m) Show forest plot	2	70	Std. Mean Difference (IV, Fixed, 95% CI)	0.15 [‐0.34, 0.63]

18.1 Coopers 12‐minute walk‐run distance (control usual care)	1	26	Std. Mean Difference (IV, Fixed, 95% CI)	0.60 [‐0.19, 1.39]
18.2 6min walk distance (control golf)	1	44	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.14 [‐0.77, 0.48]
19 Bronchial hyper‐responsiveness: ln PC20 methacholine Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

19.1 Children and adolescents	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
20 Exercise induced bronchoconstriction (maximum fall in FEV1 (%) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

20.1 Swimming ergometry	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
20.2 Cycling ergometry	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]

Comparison 1. Swimming training versus control