Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Review - Intervention

Physical activity interventions for people with congenital heart disease

Authors' declarations of interest

Version published: 28 October 2020 Version history

https://doi.org/10.1002/14651858.CD013400.pub2
Collapse all Expand all

Background

Congenital heart disease (ConHD) affects approximately 1% of all live births. People with ConHD are living longer due to improved medical intervention and are at risk of developing non‐communicable diseases. Cardiorespiratory fitness (CRF) is reduced in people with ConHD, who deteriorate faster compared to healthy people. CRF is known to be prognostic of future mortality and morbidity: it is therefore important to assess the evidence base on physical activity interventions in this population to inform decision making.

Objectives

To assess the effectiveness and safety of all types of physical activity interventions versus standard care in individuals with congenital heart disease.

Search methods

We undertook a systematic search on 23 September 2019 of the following databases: CENTRAL, MEDLINE, Embase, CINAHL, AMED, BIOSIS Citation Index, Web of Science Core Collection, LILACS and DARE. We also searched ClinicalTrials.gov and we reviewed the reference lists of relevant systematic reviews.

Selection criteria

We included randomised controlled trials (RCT) that compared any type of physical activity intervention against a 'no physical activity' (usual care) control. We included all individuals with a diagnosis of congenital heart disease, regardless of age or previous medical interventions. 

Data collection and analysis

Two review authors (CAW and CW) independently screened all the identified references for inclusion. We retrieved and read all full papers; and we contacted study authors if we needed any further information. The same two independent reviewers who extracted the data then processed the included papers, assessed their risk of bias using RoB 2 and assessed the certainty of the evidence using the GRADE approach. The primary outcomes were: maximal cardiorespiratory fitness (CRF) assessed by peak oxygen consumption; health‐related quality of life (HRQoL) determined by a validated questionnaire; and device‐worn ‘objective’ measures of physical activity.

Main results

We included 15 RCTs with 924 participants in the review. The median intervention length/follow‐up length was 12 weeks (12 to 26 interquartile range (IQR)). There were five RCTs of children and adolescents (n = 500) and 10 adult RCTs (n = 424). We identified three types of intervention: physical activity promotion; exercise training; and inspiratory muscle training. We assessed the risk of bias of results for CRF as either being of some concern (n = 12) or at a high risk of bias (n = 2), due to a failure to blind intervention staff. One study did not report this outcome. Using the GRADE method, we assessed the certainty of evidence as moderate to very low across measured outcomes.

When we pooled all types of interventions (physical activity promotion, exercise training and inspiratory muscle training), compared to a 'no exercise' control CRF may slightly increase, with a mean difference (MD) of 1.89 mL.kg−1.min−1 (95% CI −0.22 to 3.99; n = 732; moderate‐certainty evidence). The evidence is very uncertain about the effect of physical activity and exercise interventions on HRQoL. There was a standardised mean difference (SMD) of 0.76 (95% CI −0.13 to 1.65; n = 163; very low certainty evidence) in HRQoL. However, we could pool only three studies in a meta‐analysis, due to different ways of reporting. Only one study out of eight showed a positive effect on HRQoL. There may be a small improvement in mean daily physical activity (PA) (SMD 0.38, 95% CI −0.15 to 0.92; n = 328; low‐certainty evidence), which equates to approximately an additional 10 minutes of physical activity daily (95% CI −2.50 to 22.20).

Physical activity and exercise interventions likely result in an increase in submaximal cardiorespiratory fitness (assessed with VO2 mL.kg‐1.min‐1 at the gas exchange threshold; MD 2.05, 95% CI 0.05 to 4.05; n = 179; moderate‐certainty evidence). Physical activity and exercise interventions likely increase muscular strength (measured by maximal voluntary contraction of knee extensions; MD 17.13, 95% CI 3.45 to 30.81; n = 18; moderate‐certainty evidence). Eleven studies (n = 501) reported on the outcome of adverse events (73% of total studies). Of the 11 studies, six studies reported zero adverse events. Five studies reported a total of 11 adverse events; 36% of adverse events were cardiac related (n = 4); there were, however, no serious adverse events related to the interventions or reported fatalities (moderate‐certainty evidence). No studies reported hospital admissions.

Authors' conclusions

This review summarises the latest evidence on CRF, HRQoL and PA. Although there were only small improvements in CRF and PA, and small to no improvements in HRQoL, there were no reported serious adverse events related to the interventions. Although these data are promising, there is currently insufficient evidence to definitively determine the impact of physical activity interventions in ConHD. Further high‐quality randomised controlled trials are therefore needed, utilising a longer duration of follow‐up. 

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Physical activity interventions for people with congenital heart disease

Review question

This review aimed to gather evidence for the use of any physical activity intervention for people with congenital heart disease. We aimed to compare interventions including exercise training, physical activity promotion or lung training with no intervention (usual care).

Background

Congenital heart disease is the term used for a range of birth defects that affect how the heart works. People with congenital heart disease have reduced life expectancy, physical fitness and quality of life. However, due to better prenatal diagnoses, surgical procedures (often performed in the early years of life) and earlier interventions, the survival rate for those born with this disease has improved dramatically, such that most people will now live into adulthood. Exercise training and physical activity interventions are known to improve fitness, physical activity, survival and quality of life in healthy people, but it is not clear how effective these programmes are for people with long‐term medical conditions.

Study characteristics

We searched for studies in September 2019 and identified 15 studies involving 924 participants. The studies used three main types of interventions, including programmes designed to increase physical activity, aerobic fitness and health‐related quality of life and compared physical activity intervention and control interventions in people with congenital heart disease. 

Key results

We included 15 trials with 924 participants. Half of the participants were female. Of the 15 trials, 5 used a total of 500 young people (less than 18 years of age) and 10 trials used a total of 424 adult participants. We found that physical fitness and physical activity may slightly increase but we are very uncertain about quality of life. There is currently no data to say if this small increase in fitness will result in fewer visits to the hospital. But there were no recorded deaths or serious events that were related to participation in physical activity.

Quality of evidence

Using a validated scientific approach (GRADE), the certainty in the evidence base was moderate for fitness, low for physical activity and very low for quality of life. Most outcomes were limited due to small study participant numbers and poor reporting of study details.

Authors' conclusions

Implications for practice

Currently there are no guidelines outlined by the National Institute for Health and Care Excellence (NICE) for physical activity and exercise training in congenital heart disease. Moreover, in the UK there is no provision for cardiac rehabilitation (inclusive of physical activity interventions) for children and adolescents with congenital heart disease and clinical teams are encouraged to develop pathways to increase exercise and physical activity habits. By targeting young people it is suggested that good health and health behaviours will track into adulthood, subsequently reducing hospital admissions, reducing future morbidity and contributing to increasing survival rates. 

Implications for research

This review reports small and modest improvements in maximal and submaximal cardiorespiratory fitness, but there is uncertainty in the prognostic implications of this improvement over a long‐term follow‐up. We require an international effort to produce a large and long‐term randomised multicentre trial of physical activity and exercise interventions with long‐term outcomes of mortality, morbidity, cost effectiveness, cardiorespiratory fitness and health‐related quality of life. Future interventions should classify their patients (and modify the interventions) based on their functional capacity over their lesion‐specific diagnoses—this should help define what types of populations respond to interventions the best (Budts 2013; Budts 2020; Cedars 2020; Moons 2020). A prognostic factors systematic review is also required to assess the current evidence of the prognostic power of cardiorespiratory fitness for patients with congenital heart disease, as it will enable physical activity and exercise interventions to be individualised and evaluated more effectively. 

Summary of findings

Open in table viewer
Summary of findings 1. Summary of findings for the main comparison. Physical activity and exercise interventions compared to usual care for people with congenital heart disease.

Patient or population: people with congenital heart disease.
Setting: hospital‐based and home‐based settings.
Intervention: physical activity promotion, inspiratory muscle training and exercise training interventions. 
Comparison: usual care.

Outcomes 

Length of intervention in weeks (Median, (IQR))

Anticipated absolute effects* (95% CI)

No. of participants (studies)

Certainty of the evidence (GRADE)

Comments

Risk with usual care
 

Risk with all interventions
 

Maximal cardiorespiratory fitness (CRF)

Assessed with: treadmill or cycle ergometry

(12 (12, 26))

Mean 22 to 46

MD 1.89 higher (0.22 lower to 3.99 higher)

732

(14 RCTs (15 training arms))

 ⊕⊕⊕⊝ab
MODERATE

Physical activity and exercise interventions may increase cardiorespiratory fitness slightly. Sensitivity analyses did not change the inference to a clinically important MD.

Health‐related quality of life (HRQoL) Assessed with: Questionnaires

(12 (12, 12))

‐

SMD 0.76 higher (0.13 lower to 1.65 higher)

163 (3 RCTs)

 ⊕⊝⊝⊝cde
VERY LOW

The evidence is very uncertain about the effect of physical activity and exercise interventions on health‐related quality of life. 

Physical activity (PA)

Assessed with: Accelerometer

(19 (12, 46))

Mean 11 to 41

SMD 0.38 
(0.15 lower to 0.92 higher)

328 (4 RCTs)

 ⊕⊕⊝⊝ce
LOW

Physical activity and exercise interventions may increase physical activity slightly.

Submaximal cardiorespiratory fitness

Assessed with: treadmill or cycle ergometry (VO2 mL.kg‐1.min‐1 at the gas exchange threshold).

(12 (12, 15))

Mean 18 to 22

MD 2.05  (0.05 higher to 4.05 higher)

179 (5 RCTs (6 training arms))

 ⊕⊕⊕⊝e
MODERATE

Physical activity and exercise interventions likely results in an increase in submaximal cardiorespiratory fitness.

Muscular strength

(12)

Mean 103

MD 17.13  (3.45 higher to 30.81 higher)
 

18 (1 RCT)

 ⊕⊕⊕⊝e
MODERATE

Physical activity and exercise interventions likely increases muscular strength.

Adverse events

(12 (12, 26))

A total of 11 adverse events were reported in a population of 501 (2.2%), 7 of which were mild (1.4%). Mild adverse events included dizziness, discomfort, musculoskeletal (strains/ sprains) and a minor head injury. The 4 (0.8%) moderate adverse events were cardiac. There were no major adverse events reported.

501 (11 RCTs)

 ⊕⊕⊕⊝f
MODERATE

Physical activity promotion and exercise training interventions did not lead to any serious adverse events.

Hospital admissions 

No data available 

CI: confidence interval; IQR: 25th and 75th quartiles; MD: mean difference; SMD: standardised mean difference; RCT: randomised controlled trial; CRF: Cardiorespiratory fitness; HRQoL: Health‐related quality of life. The mean SMD effect size was interpreted using Cohen effect sizes, i.e. 0.2 represents a small effect, 0.5 a moderate effect, and 0.8 a large effect. 

GRADE Working Group grades of evidence.
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. 

a Statistical heterogeneity was present in CRF (I2 = 75%). But it was explained using a sensitivity analysis by excluding high risk of bias studies (I2 = 18%). Therefore, we chose not to downgrade by 1 level evidence due to inconsistency. 

bThe confidence interval includes both appreciable harm and appreciable benefit (i.e. 95% CI spans 0). Therefore, the certainty of evidence was downgraded by 1 level due to imprecision. 

cInconsistent directions of effect and considerable heterogeneity (HRQoL, I2 = 82%; PA, I2 = 77%). Therefore, the certainty of evidence was downgraded by 1 level due to inconsistency. 

dTotal high risk of bias in all studies included within the analysis, hence bias highly likely. Therefore, the certainty of evidence was downgraded by 2 levels due to the methodological limitations (risk of bias).

eImprecise due to small numbers of events (< 400) (Ryan 2016).  Therefore, the certainty of evidence was downgraded by 1 level due to imprecision. 

fOver 25% of studies did not report data on adverse events. Therefore, the certainty of evidence was downgraded by 1 level due to publication bias.

Background

Description of the condition

Congenital heart disease (ConHD) is a term for a range of developmental abnormalities of the heart or intrathoracic vessels, or both (Mitchell 1971). There are over 18 distinct types of ConHD, ranging in complexity (Rhodes 2008; Sommer 2008a; Sommer 2008b), that can be broadly classified as mild, moderate or severe (Hoffman 2002). During the period 2010 to 2017 the birth prevalence of ConHD is approximately 1% of all live births (9.41 per 1000 births; 95% confidence interval (CI) 8.60 to 10.25), with the mild conditions representing 65% (95% CI 58.7% to 71.7%) of the total ConHD births (Liu 2019). 

Long‐term survival is reduced in ConHD, but due to improved medical care this has improved dramatically over previous decades. The most recent meta‐analysis reported that survival up to the age of 10 years was 81.4% (95% CI 73.80 to 87.90); however, 87.0% of the between‐article variance can be accounted for by the year of the study (Best 2016). Consequently, survival up to the age of 10 years of age has increased from 75% to 90% over the past two decades due to improvements in surgical correction, prenatal diagnosis and earlier interventions (Best 2016). ConHD is therefore a changing chronic medical condition with the highest proportion of deaths now occurring in geriatrics (Khairy 2010).

In both healthy and clinical (coronary heart disease) populations, impaired physical activity (PA) and cardiorespiratory fitness (CRF) are associated with the development of non‐communicable disease, morbidity and mortality (Franklin 2013; Lee 2012; Letnes 2019). People with ConHD have reduced levels of fitness (Amedro 2017); reduced health‐related quality of life (HRQoL) (Amedro 2015); and are less physically active (Brudy 2020; Dua 2007; McCrindle 2007; Sandberg 2016). Fitness has also been heavily associated with future health outcomes in this population (Dimopoulos 2006; Giardini 2009; Müller 2015; Udholm 2018). It is crucial, therefore, that people with ConHD lead an active lifestyle, but there is currently no consensus on how best to improve PA, CRF and HRQoL in people with ConHD.

Description of the intervention

Physical activity consists of any bodily movement involving skeletal muscles that results in increased energy expenditure, whereas exercise training is a planned and structured period of PA with the intention of maintaining or improving physical fitness components (Caspersen 1985). The current guidelines for PA are an average of 60 minutes of moderate to vigorous physical activity (MVPA) a day for young people (< 18 years old) and 150 minutes of MVPA a week in adults (Department of Health 2011). Currently within specialist paediatric cardiac clinics physical activity recommendations are not adequately discussed due to a lack of training, time and knowledge of the current exercise recommendations for people with ConHD (Williams 2017). 

Interventions that aim to improve CRF, PA and HRQoL typically consist of a PA promotion (goal setting, motivational interviews etc.), or exercise training (aerobic/resistance training, sports participation etc.), or a combination of both. PA promotion interventions aim to increase habitual PA behaviours by using psychological conceptual frameworks in order to promote self‐efficacy, goal‐setting and intrinsic motivation. Exercise training interventions, on the other hand, usually prescribe a set 'dose' of exercise either within a hospital, centre or at home. Inspiratory muscle training (IMT) is a new method of intervention, recently gaining popularity in people with chronic cardiorespiratory conditions and aiming to improve ventilatory power and efficiency. IMT is distinctly different from PA promotion and exercise training as it involves training the inspiratory chest muscles against a breathing resistance by using a handheld device. 

How the intervention might work

The 'gold standard' measure of cardiorespiratory fitness is maximal oxygen consumption, which is explained by the Fick equation where oxygen uptake is the product of the cardiac output and the arteriovenous oxygen difference (V̇O2 = Q * a‐V̇O2 diff). Improving cardiorespiratory fitness must target improving oxygen delivery (Q) and/or oxygen extraction at the peripheral sites (a‐V̇O2 diff) of the body, namely the muscles. PA and exercise training are known to improve both cardiac output and oxygen extraction through complex molecular interactions improving myocardial contractility, mitochondrial activity, stem cell proliferation, nitric oxide bioavailability and muscle fibre adaptations (Adams 2017; Gielen 2010). There is evidence that supports inspiratory muscle training to improve ventilatory efficiency and fitness in people with chronic cardiorespiratory pathologies and in healthy people, but the adaptation mechanisms are not well understood (Shei 2018; Wong  2011). It is proposed that the increase in ventilatory efficiency is due to a combination of factors, inclusive of, but not limited to, changing motor recruitment 'diaphragm sparing' and the release of inflammatory cytokines (Shei 2018).

Why it is important to do this review

Cardiorespiratory fitness is lower in people with ConHD and deteriorates faster compared to healthy people (Amedro 2017). This has significant implications as CRF has been associated with future mortality and morbidity in several ConHD conditions (Dimopoulos 2006; Giardini 2009; Müller 2015; Udholm 2018). Currently, there is a dearth of evidence to adequately inform the effectiveness of interventions to improve CRF and consequently long‐term outcomes.

Recent reviews have included both non‐randomised and randomised controlled trials (RCTs), focused on specific types of interventions (e.g. home‐based) or have restricted to particular age groups (Gomes‐Neto 2016; Li 2019; Meyer 2020). Therefore, we present the first Cochrane Review to assess the effectiveness of all types of physical activity interventions and inclusive of all age groups, using only RCT data in people with ConHD. We hope by conducting this review that we can provide clarity on the effectiveness of physical activity interventions, highlight future avenues for research and inform future healthcare policy.

Objectives

To assess the effectiveness and safety of all types of physical activity interventions versus standard care in individuals with congenital heart disease.

Methods

Criteria for considering studies for this review

Types of studies

We planned to include all types of randomised controlled trial (RCT), inclusive of but not limited to parallel, cross‐over and cluster designs. We included only parallel and randomised cross‐over designs: these trials compared all types of physical activity interventions to a 'no physical activity/no exercise' comparator. We included trials irrespective of their duration of follow‐up.

Types of participants

We included all individuals with a diagnosis of ConHD, who were deemed suitable for participation in a physical activity or exercise training intervention. We included all types of congenital heart disease, regardless of previous medical care and categorised them as mild, moderate or severe (Hoffman 2002). We also included paediatric (5 to 18 years old) and adult populations (> 18 years old). 

Types of interventions

We identified and included three types of intervention: physical activity (PA) promotion; exercise training; and inspiratory muscle training (IMT). PA promotion studies incorporated psychological components to promote and educate participants on the benefits of exercise, whereas exercise training studies 'prescribed' exercise at a set dose either in a hospital or in a home‐based setting. IMT was not an anticipated intervention type at the protocol phase; we included it due to its increasing use in clinical care (Pufulete 2019). Furthermore, we included interventions whether they were structured versus unstructured, supervised versus unsupervised, home versus hospital and single versus multicomponent. All interventions were compared to 'no physical activity/physical activity as usual' control; and both the intervention and control group received usual medical care.

Types of outcome measures

Studies should have intended to assess any of the outcomes in both the intervention and the control groups. At the protocol phase we intended to extract outcomes at two time points: at the end of intervention and at long‐term follow‐up (> 12 months). Due to the lack of long‐term follow‐up data, we only extracted the data at the end of the intervention. We sought to report the following primary and secondary outcomes, but they did not form the basis of our inclusion/exclusion criteria.

Primary outcomes

  • Maximal cardiorespiratory fitness (CRF)

  • Health‐related quality of life determined by a validated questionnaire

  • Device‐worn ‘objective’ measures of physical activity

Secondary outcomes

  • Submaximal CRF

  • Validated questionnaire‐based ‘subjective’ measures of physical activity

  • Return to work or full‐time education

  • Hospital admissions

  • Muscular strength determined by:

    • grip strength

    • isokinetic testing

    • muscular endurance capacity

  • Adverse events

We anticipated there would be substantial variability in the reported outcome measures and we approached the primary outcomes as follows.

Cardiorespiratory fitness (CRF)

We pooled peak oxygen consumption (peak V̇O₂) measured in millilitres per kilogram per minute (mL.kg−1.min−1) as our measure of maximal CRF. Peak V̇O2 was assessed by validated cardiopulmonary exercise test protocols, measuring oxygen consumption directly or indirectly (by estimated oxygen consumption), using either a treadmill or cycle ergometer. The submaximal CRF outcome was oxygen consumption per kilogram of body mass (V̇O2 mL.kg−1.min−1) at the gas exchange threshold (GET). 

Health‐related quality of life (HRQoL)

As anticipated there was large variability in the HRQoL scales used; we pooled HRQoL data in a meta‐analysis where appropriate and reported all HRQoL in Table 1 and summarised in text.

Open in table viewer
Table 1. Health‐related quality of life

Author

Sample size

Type of intervention

Questionnaire

Domain

Intervention vs. control at follow up

Mean (Standard deviation) & between group P value

Direction of effect

Overall risk of bias

Duppen 2015
 

73
 

Exercise training
 

SF‐36 (8 domains) +

Physical functioning

94.6 (10.9) 95.0 (8.5) P = 0.71 

 Exercise = control

High

SF‐36 (8 domains) +

Bodily pain

96.6 (8.1) 93.2 (17.3) P = 0.96

 Exercise = control

High

SF‐36 (8 domains) +

General health

68.3 (27.9) 67.5 (19.1) P = 0.94 

 Exercise = control

High

SF‐36 (8 domains) + 

 Vitality

71.7 (18.0) 70 (17.0) P = 0.56 

 Exercise = control

High

SF‐36 (8 domains) +

Role limitations due to physical limitations

100 (0.0) 91.7 (21.2) P = 0.16 

 Exercise = control

High

SF‐36 (8 domains) +

Social functioning

95.8 (10.0) 100.0 (0.0)P = 0.09 

 Exercise = control

High

SF‐36 (8 domains) +

Role limitations due to emotional problems

100 (0.0) 100 (0.0) P = 0.72

 Exercise = control

High

SF‐36 (8 domains) +

Mental health

85.0 (16.8) 81.3 (10.2) P = 0.65

 Exercise = control

High

TACQOL +

Symptoms 

95.6 (7.90) 97.8 (3.7) P = 0.16

 Exercise = control

High

TACQOL +

Impact of cardiac surveillance

85 (6.30) 85.7 (4.80) P = 0.07 

 Exercise = control

High

TACQOL +

Worries

95.7 (8.70) 88.6 (270) P = 0.67 

 Exercise = control

High

TACQOL‐CF +

Pain and physical symptoms

26.3 (6.2) 24 (7.15) P = 0.21

 Exercise = control

High

TACQOL‐CF +

Motor functioning

30.1 (1.9) 29.5 (4.37) P = 0.51

 Exercise = control

High

TACQOL‐CF +

Cognitive functioning

28.6 (6.2) 29.7 (6.9) P = 0.05

 Exercise = control

High

TACQOL‐CF + 

Social functioning

31.5 (1.2) 32 (0) P = 0.45

 Exercise = control

High

TACQOL‐CF +

Positive emotional functional

14.2 (3.5) 14.43 (2.9) P = 0.39

 Exercise = control

High

TACQOL‐CF +

Negative emotional functioning

12.7 (2.12) 14.2 (2.2) P = 0.34

 Exercise = control

High

Madhavi 2011

112

Exercise training

SF‐36 +

SF36 total score

23.3  (13.2) 11.3 (14.3) P < 0.001

 Exercise > control

High

Klausen 2016
 

158

Physical activity promotion
 

Danish Paediatric QoL Inventory +

Generic

NR

 Exercise = control

High

Danish Paediatric QoL Inventory +

Disease specific

NR

 Exercise = control

High

Opotowsky 2018

28

Exercise training

MLHFQ ‐ 

MLHFQ

20.1 (11.4) 27.7 (10.9) P = 0.13

 Exercise = control

High

Sandberg 2018

23

Exercise training

EQ5D VAS +

EQ5D VAS

76.2 (15.2) 76.3 (20.7) P = 0.31

 Exercise = control

High

Novakovic 2018
 

14 

Exercise training (Interval)
 

SF36 +

Physical component

86.7 (40.2) 101.0 (16.6) P > 0.05

 Exercise = control

High

SF36 +

Mental component

80.0 (21.0) 87.3 (21.9) P > 0.05

 Exercise = control

High

Novakovic 2018
 

13 

Exercise training (Continuous)
 

SF36 +

Physical component

103 (5.2) 101.0 (16.6) P > 0.05

 Exercise = control

High

SF36 +

Mental component

89.3 (18.4) 87.3 (21.9)  P > 0.05

 Exercise = control

High

Westhoff-Bleck 2013

40

Exercise training

 KCCQ +

 KCCQ

NR

 Exercise = control

High

Winter 2012
 

 

46
 
 

Exercise training
 

 

SF‐36 +

Mental component

P = 0.17

 Exercise = control

High

SF‐36 +

Physical component

P = 0.20

 Exercise = control

High

ConHD‐TAAQOL +

Symptoms

85 (10.79) 83 (10.47) P = 0.31

 Exercise = control

High

ConHD‐TAAQOL +

Worries

77 (15.4) 84.5 (9.21) P = 0.30

 Exercise = control

High

ConHD‐TAAQOL +

Impact

84 (7.19) 85.25 (6.20) P = 0.91

 Exercise = control

High

Each study's outcome of health related quality of life was individually assessed using risk of bias 2; all studies were judged as a high risk of bias under the domain 'Measurement of the outcome'. '+' =  a higher score represents better health; '‐' =  a lower score represents better health; HRQoL, Health related quality of life; SF‐36,  36‐Item Short Form Health Survey;  MLHFQ, Minnesota living with heart failure questionnaire; TACQOL, TNO/AZL child quality of life questionnaire; ConHD TAAQOL, The congenital heart disease ‐ TNO/AZL adult quality of life questionnaire; EQ5D VAS, EuroQol Vertical Visual Analogue Scale; KCCQ, Kansas City Cardiomyopathy Questionnaire.

Device‐worn measures of physical activity

Physical activity was measured by accelerometry only. We pooled data regardless of device (Actigraph GT3X, Actigraph GT1M), device settings (epoch ranged 5 to 60 seconds) and activity parameter measured (time spent in MVPA per day as a percentage (n = 1) and minutes spent in MVPA per day) as per the protocol. There were no heart rate data to analyse. and activity parameter measured (time spent in MVPA per day as a percentage (n = 1) and minutes spent in MVPA per day) as per the protocol. There were no heart rate data to analyse. 

Search methods for identification of studies

Electronic searches

We undertook a systematic search of the following databases on 23 September 2019.

  • CENTRAL in the Cochrane Library (Issue 9 of 12, 2019)

  • Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, MEDLINE Daily and MEDLINE (Ovid, 1946 to 19 September 2019)

  • Embase (Ovid, 1980 to 2019 week 38)

  • Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCOHost, 1937 to 23 September 2019)

  • Allied and Complementary Medicine Database (AMED) (Ovid, 1985 to September 2019)

  • BIOSIS Citation Index (Clarivate Analytics, 1926 to 23 September 2019)

  • Web of Science Core Collection (Clarivate Analytics, 1900 to 23 September 2019)

  • Latin American and Caribbean Health Sciences Literature (LILACS) (Bireme, 1982 to 23 September 2019)

  • DARE (NIHR Centre for Reviews and Dissemination www.crd.york.ac.uk, from inception to 31 March 2015 (when this database stopped adding records)).

We applied the Cochrane sensitivity‐maximising RCT filter for MEDLINE and for Embase terms as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2011; Higgins 2011). For CINAHL, we used the Cochrane CINAHL RCT filter (Glanville 2019). For all other databases, except CENTRAL, LILACS and DARE, we applied an adaptation of the Cochrane RCT filter. See Appendix 1 for the search strategies.

Searching other resources

We searched ClinicalTrials.gov on 11 September 2020 for ongoing or unpublished trials.

We also searched by hand the reference list of relevant reviews, randomised and non‐randomised studies, and editorials for additional studies. We contacted authors of studies and experts in the field to ask for any missed, unreported or ongoing trials. We also searched for any retraction statements and errata for included studies.

Data collection and analysis

Selection of studies

Two review authors (CAW and CW) independently screened titles and abstracts for inclusion from all the potential trials we identified from the searches. We then sourced full texts and both review authors (CAW and CW) independently read them to confirm eligibility; in the event of exclusion, we documented the reasons. This was facilitated by Covidence systematic review software. Two authors (LL and RST) arbitrated if any disagreements arose that could not be rectified through discussion,. We have recorded the selection process with a PRISMA flow diagram (Figure 1) and Characteristics of excluded studies (Liberati 2009).

Figure 1
Study flow diagram 

Study flow diagram 

Data extraction and management

Two authors (CAW and CW) independently piloted a data collection form and independently extracted outcome data from included studies. One review author (CW) transferred data into RevMan Web 2019; and CAW checked that the data was entered correctly.

We extracted the following study characteristics.

  • Participants: N randomised, N lost to follow‐up, N analysed, mean age (± standard deviation), gender, severity of condition*, inclusion criteria, and exclusion criteria.

  • Methods: study design, total duration of study, study setting, date of study, withdrawals, number of study centres and location.

  • Interventions: intervention description (including the frequency, intensity, duration and modality of the intervention), comparison, and co‐interventions.

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

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

*As ConHD is an incredibly varied and complex disease we have classified the severity of the condition using the Hoffman 2002 criteria as ‘mild’, ‘moderate’ or ‘severe’ (see Appendix 2 for further information). We have chosen the Hoffman classification as it is very inclusive and does not bias against individual intra‐diagnosis differences; it has since been adopted in the most recent guidelines from the US Task Force for adult congenital heart disease (Warnes 2008). We have used these criteria to describe the study data (to aid the reader); we have not used it for subgroup analyses. 

Assessment of risk of bias in included studies

Two review authors (CAW and CW) independently assessed risk of bias for each study using the recently revised 'Risk of bias in randomised trials (RoB 2)' tool (accessed: 28 January 2020) (Higgins 2019). 

We assessed risk of bias for each study outcome using the following Cochrane RoB 2 criteria (Higgins 2019).

  • Bias arising from the randomisation process

  • Bias due to deviations from intended interventions

  • Bias due to missing outcome data

  • Bias in measurement of the outcome

  • Bias in selection of the reported result

For each domain a series of signalling questions with the answers (yes, probably yes, no information, probably no, no) determine the risk of bias (low risk, some concerns and high risk). We included text alongside the judgements to provide supporting information for our decisions (see Risk of bias in included studies). We decided the risk of bias for an outcome (e.g. health‐related quality of life (HRQoL)) by its performance in each domain: if we judged one domain 'some concerns' or 'high risk' this judgement was taken for the whole outcome. We assessed the risk of bias of maximal and submaximal cardiorespiratory fitness (CRF), HRQoL, physical activity (PA) and muscular strength at follow‐up. The effect of assignment or 'intention to treat' was our effect of interest and we have summarised the risk of bias in traffic lights on the forest plots, in Table 1 for HRQoL and in text.

Measures of treatment effect

We analysed continuous data as mean difference with 95% confidence intervals (CIs). Where an outcome was measured and reported in more than one way, we report a standardised mean difference (SMD) with 95% CIs. We interpreted the SMD using the two approaches recommended in the Handbook (Schünemann 2017). First, we interpreted the mean SMD effect size using the following rule of thumb based on Cohen effect sizes (i.e. 0.2 represents a small effect, 0.5 a moderate effect, and 0.8 a large effect) (Faraone 2008). In addition, for physical activity data we have converted the SMD back to the original scale units (minutes of moderate to vigorous activity) by multiplying the pooled mean SMD by an among‐person standard deviation for a particular trial (Opotowsky 2018). We have corrected for any differences in the direction of the scales for HRQoL (i.e. when some scales have a lower score for a better QoL, a reduction in score would indicate an improvement, whereas a scale that awards higher scores for better QoL would see an increase in score indicating a positive outcome). We included HRQoL data in the meta‐analysis only if it was the overall or total HRQoL score. For the outcome 'adverse events' where there was a dichotomous variable (event or no event), we analysed this using count data and summarised in text. Where data were skewed and reported as medians and interquartile ranges, we converted them to means and standard deviations using validated equations (Wan 2014). 

Unit of analysis issues

We only identified studies with individual randomisation. Only one study presented long‐term follow‐up data (Winter 2012); in this instance we extracted data at the end of the intervention to keep consistency with all the other trials. One trial contained three arm: continuous exercise training; interval exercise training; and a control group (Novakovic 2018). In this case we divided the number randomised to the control group in half to obtain the denominator for data analysis; the means and standard deviation for the control group remained unchanged for both comparisons. One study had a randomised cross‐over design, but only contributed data prior to the cross‐over (Fritz 2020); it was therefore not necessary to consider a washout period. 

Dealing with missing data

We contacted multiple authors to verify key study characteristics (such as randomisation), clarify data queries and obtain missing numerical outcome data. Where data were presented graphically and we were unable to obtain numerical data from the authors, we used WebPlotDigitizer to extract this information.

Assessment of heterogeneity

We explored heterogeneity amongst included studies qualitatively (through visual inspection of forest plots and by comparing the characteristics of included studies), and quantitatively (using the Chi² test of heterogeneity and the I² statistic). We used a threshold of I² greater than 50% to represent substantial heterogeneity for continuous outcomes (Deeks 2017). 

Assessment of reporting biases

We were able to pool more than 10 studies in our primary outcome 'maximal cardiorespiratory fitness'. We subsequently created and examined a funnel plot and used the Egger test to explore possible small‐study biases for this primary outcome (Egger 1997).

Data synthesis

We performed meta‐analyses with 95% CIs including all available studies where appropriate (i.e. when treatments, participants, and the underlying clinical question were similar enough for pooling to be appropriate). We used random‐effects meta‐analyses for all analyses due to the qualitative (types of interventions and severities of ConHD) and quantitative (statistical) heterogeneity present. Evidence of substantial heterogeneity was confirmed using the I2 statistic of more than 50%, giving further justification for a random‐effects analysis. Random effects provides a more conservative statistical approach, as the confidence interval around a random‐effects estimate is wider than a confidence interval around a fixed‐effect estimate (Heran 2008a; Heran 2008b). Where an outcome was measured and reported in more than one way, we report a standardised mean difference (SMD) with 95% CI. 

We processed data in accordance with guidance in the Handbook (Higgins 2011). We completed data synthesis and analyses using RevMan Web software (RevMan Web 2019); and we conducted meta‐regression analysis using the “metareg” command in Stata version 14.2 (Stata 2015 [Computer program]). We created additional figures using GraphPad (GraphPad Prism).

We could not pool some HRQoL. In this instance we adopted a modified version of a vote counting table, allowing us to summarise descriptive data, risk of bias and the direction of effect. Whilst this synthesis without meta‐analysis (SWiM) method has significant limitations, we believe it to be the only SWiM method that allows us to communicate the results in a transparent and concise format (Campbell 2020).

Subgroup analysis and investigation of heterogeneity

We split the outcome 'maximal cardiorespiratory fitness' into two subgroup analyses: Analysis 1.1 reports the effect of the type of physical activity intervention (i.e. PA promotion, exercise training and inspiratory muscle training (IMT); and Analysis 1.6 reports the effect of the intervention in each group of ConHD (i.e. single ventricle, tetralogy of Fallot and other/mixed ConHD).

 We used meta‐regression to assess the potential treatment effect modifiers from all interventions (PA promotion, exercise training and IMT) on maximal cardiorespiratory fitness. Due to the limited number of studies to co‐variate ratio we limited the meta‐regression to univariate analysis only (Higgins 2011). The meta‐regression included the following co‐variates.

  • Type of intervention (PA promotion or exercise training or IMT (categorical variable)).

  • ‘Dose’ of exercise intervention (dose = number of weeks of exercise training × average number of sessions/week × average duration of session in minutes) (continuous variable).

  • Length of intervention/follow‐up period (continuous variable).

  • Sample size (continuous variable).

  • Setting (home‐ or centre‐based) (categorical variable).

  • Study location (North America or Europe or Asia) (categorical variable).

  • Age of participants (paediatrics or adults) (categorical variable).

  • Percentage of male participants (continuous variable).

  • Baseline cardiorespiratory fitness  (continuous variable).

  • Risk of bias (categorical variable).

Due to the lack of data, we pooled all individual ConHD lesions (Table 2). This has limitations due to the within‐condition and between‐condition heterogeneity in clinical status (Amedro 2017; Kempny 2012). We therefore used pre‐intervention (baseline) cardiorespiratory fitness as a meta‐regression covariate to account for the heterogeneity. 

Open in table viewer
Table 2. Individual ConHD lesions pooled into the meta‐analyses

Study

Age

Type of ConHD and (sample size, %)

Classification

Avila 2016

35 ± 11

Repaired ToF (17, 100%)

Severe

Duppen 2015

15 ± 3

Fontan circulation (43, 48%)

[Intra‐atrial lateral tunnel 47%; extracardiac conduit 44%; other 9%]

Severe

Repaired ToF (47, 52%)
 

Fritz 2020

30 ± 9

Fontan circulations (42, 100%)

[Atrioventricular Anastomosis 19%; Atriopulmonary Anastomosis 21%; Total Cavopulmonary Connection 60%]

Severe

Klausen 2016

15 ± 1

Coarctation of the aorta (52, 33%)

Severe

TGA (35, 22%)

Steno–Fallot tetralogy (21, 13%)

Double outlet right ventricle (7, 4%)

Truncus arteriosus (4, 3%)

Atrioventricular septal defect (9, 6%)

TCPC (6, 4%)

Other (24, 15%)

Madhavi 2011

29 ± 11

Acyanotic Congenital Heart Disease (112, 100%)

Mild

Moalla 2006

13 ± 1

Fontan (4, 22%)

Mild to  Severe

TGA (Senning/Mustard procedure) (5, 28%)

Repaired ToF (5, 28%)

Repaired ASD (4, 22%)

Morrison 2013

15 ± 2

Minor CHD (no surgical intervention) (39, 27%)

Mild to Severe

Acyanotic corrected ConHD (61, 43%)

Cyanotic corrected (30, 21%)

Cyanotic palliated (13, 9%)

Novakovic 2018

38 ± 8

Repaired ToF (30, 100%)

Severe

Opotowsky 2018

41 ±12

ToF with pulmonary stenosis or atresia or, DORV (13, 46%)

Severe

TGA with a systemic RV, (9, 32%)

Fontan (2, 7%)

Pulmonary atresia with intact ventricular septum with biventricular repair (2, 7%)

Truncus arteriosus (1, 4%)

Ebstein anomaly (1, 4%)

Sandberg 2018

30  ±  11

ToF (5, 22%)

Severe

ccTGA & d‐TGA (8, 35%)

TCPC (5, 22%)

PA (2, 9%)

Complete AV‐septal defect (1, 4%)

Ebstein anomaly (1, 4%)

Miscellaneous (1, 4%)

Therrien 2003

35 ± 9

Repaired ToF (18, 100%)

Severe

Westhoff‐Bleck 2013

29 ± 3

Systemic right ventricle ‐ TGA [Mustard] (48, 100%)

Severe

Winter 2012

32 + 11

Systemic right ventricle ‐ TGA and ccTGA (46, 100%)

Severe

van Dissel 2019

 

40 ± 9

 

ToF (12, 30%)

Severe

 

TGA (13, 33%)

Fontan circulation (9, 22%)

Pulmonary atresia (3, 7.5%)

Other (3, 7.5%)

ToF, tetralogy of Fallot; TGA, transposition of the great arteries; ccTGA, congenitally corrected transposition of the great arteries; d‐TGA, dextro‐transposition of the great arteries; ASD, atrial septal defect; TCPC, Total cavopulmonary connection; DORV, double‐outlet right ventricle; RV, right ventricle.

Sensitivity analysis

We performed the following sensitivity analyses for the outcome of maximal cardiorespiratory fitness: removal of high risk of bias studies; direct versus indirect methods of measuring/estimating peak V̇O2; the use of a fixed‐effect model; insertion of all available change scores; and the removal of computed outcome scores (converting medians and interquartile ranges to means ± standard deviations). We did not perform sensitivity analyses for the other outcomes within the review, due to the lack of studies included in the respective outcomes and the similarity of the studies pooled. 

Summary of findings and assessment of the certainty of the evidence

We created the ‘summary of findings Table 1' using RevMan Web 2019 and reported the following outcomes: maximal cardiorespiratory fitness (CRF), health‐related quality of life (HRQoL), device‐worn ‘objective’ measures of physical activity (PA), submaximal CRF, muscular strength, and adverse events.

Two reviewers (CAW & CW) independently conducted GRADE analysis using GRADEpro GDT. Where disagreements arose, we asked co‐authors (LL & RST) to arbitrate. We used GRADE to assess the certainty of the available evidence, helping to inform decisions based on this evidence (Schünemann 2017). We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of the body of evidence, as it relates to the studies that contribute data to the meta‐analyses for the prespecified outcomes. We justified all decisions to downgrade the quality of studies using footnotes.

We used methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions using GRADEpro software (Higgins 2011). Long‐term follow‐up (> 12 months) post intervention was our follow‐up period of most interest. However, as only one study reported long‐term follow‐up, we only report short‐term follow‐up (immediately post intervention) in the 'Summary of findings' table.

Results

Description of studies

See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of ongoing studies; Characteristics of studies awaiting classification.

Results of the search

We identified 3806 references through our electronic and manual searches. After de‐duplication and title and abstract screening,  we retrieved 124 references. After screening the full text,  we identified 15 RCTs from 39 references (see Figure 1). Searching of the reference lists of eligible publications did not reveal additional publications for inclusion.

We contacted 18 corresponding authors for further information regarding study inclusion. When we could not reach the authors, we included these studies (n = 5) in the Studies awaiting classification table.

Included studies

Population

We included 15 RCTs with 924 participants (50% ± 12% male) in the review. There were five paediatric RCTs and 10 adult RCTs with 500 participants and 424 participants respectively. All paediatric RCTs were based in Europe, whereas adult trials were based in Europe (n = 6), North America (n = 3) and Asia (n = 1). There were 11 RCTs that included severe classification participants (n = 559); three RCTs that pooled mild, moderate and severe classifications (n = 254); and one RCT that included mild classification participants only (n = 111). Table 2 reports the individual ConHD lesions that we pooled into the meta‐analyses. 

Intervention

We identified three distinct types of interventions: exercise training (n = 11); physical activity promotion (n = 3); and inspiratory muscle training (IMT) (n = 1). See Table 3 for the characteristics of exercise training trials. Physical activity promotion aims were varied: Morrison 2013 and Klausen 2016 used motivational techniques (interviewing and goal setting vs. text 'e‐based' encouragement) to improve physical activity and fitness in children and adolescents; whereas another intervention used a family‐based psychological intervention with a subcomponent of physical activity promotion with the aim of improving HRQoL, time/behaviour in school and sports enjoyment in young children (van der Mheen 2019). The only IMT study included within the review aimed to assess the efficacy of IMT in adults with severe ConHD (Fontan circulations). The intervention was a randomised cross‐over design using a commercially available inspiratory muscle trainer. The participants completing three sets of 10 to 30 repetitions every day for six months, the intensity could be adjusted from 10 cm H2O to 90 cm H2O and was individualised for every training session to maintain an optimal training effect (Fritz 2020). 

Open in table viewer
Table 3. Characteristics of exercise training interventions

Study

Age group & severity 

Location & supervision

Frequencya (sessions per week)

Intensity

Timeb (minutes)

Type

Durationc (Weeks)

Dose (a*b*c)

 Avila 2016

Adult & severe (ToF)

Hospital & supervised

 1 to 2

70% to 80% of the maximum HR (increased throughout intervention)

60

Combination of resistance and aerobic dynamic (running, rowing etc.) exercise

12

1080

 Duppen 2015

Paediatric & severe (ToF and Fontan)

Hospital & supervised
 

2 to 3

Resting heart rate plus 60% to 70% of the HR reserve

60
 

Aerobic dynamic 

12
 

1800
 

Madhavi 2011

Adult & mild

NR 

NR 

Individualised (NR) 
 

NR
 

NR

12

N/A

Moalla 2006

Paediatric & mild/severe 

Home & semi‐supervised

3
 

HR at the gas exchange threshold

60
 

Cycling

 12

2160
 

Novakovic 2018
 

Adult & severe (ToF)
 

NR
 

2 to 3 

 80% of HRpeak in high intensity exercise

42
 

Cycling or speed walking 

12
 

1260
 
 

70% of HRpeak in continuous intensity exercise 

Opotowsky 2018

 Adult & severe

Hospital & supervised 

2
 

HR at Gas Exchange Threshold 

60
 

Combination of resistance and aerobic dynamic 

12
 

1440
 

Sandberg 2018

Adult & severe

Home  & semi‐supervised

3

HR was calculated according to the Karvonen method and to achieve BORG 15 to 16 

31

Cycling

12

1116

Therrien 2003

Adult & severe (ToF)

Hospital and home (1:2 ratio) & supervised

3

 60% to 85% of pre training peak VO2 

50

Cycling and walking
 

12

1800

Westhoff‐Bleck 2013
 

Adult & severe (Mustard procedure)

Home &  semi‐supervised 

3‐5

HR corresponding to 50% of peak VO2 

10 to 30

Cycling

24

2550d

Winter 2012

Adult & severe  

Home & semi‐supervised 

3

75% to 90% of max heart  rate (increased throughout intervention)

42

Step aerobics

10

1260

van Dissel 2019

Adult & Moderate and Severe 

Home & semi‐supervised 

3
 

80% of the maximum HR 

45
 

Self‐selected

26

3510

ToF, Tetralogy of Fallot; HR, Heart Rate; NR, Not Reported; d See dose calculations in Characteristics of included studies

Comparison

All studies compared to usual care for their region. Only one study had three arms: two intervention arms (interval and continuous training) and a control arm (Novakovic 2018).

Primary Outcomes

Maximal cardiorespiratory fitness (CRF) was measured in 14 out of 15 (93.3%) studies. Health‐related quality of life (HRQoL) was reported in 8 out of 15 (53.3%) studies, using a variety of validated questionnaires summarised in Table 1. Device‐worn measures of physical activity was reported by four (26.6%) studies, using a range of accelerometers, cut points and parameters such as time spent as a percentage in moderate to very vigorous activity, average minutes of moderate to vigorous activity (MVPA) and total minutes per day spent in MVPA assessed using accelerometer cut‐points greater than 2000 counts (Duppen 2015; Klausen 2016; Morrison 2013; Opotowsky 2018). No study used disease‐specific cut points.

Secondary Outcomes

Only one study numerically reported questionnaire‐based physical activity (Duppen 2015). Klausen 2016 used questionnaires in combination with device‐worn measures but did not report the questionnaire data as it reported similar results. No study measured return to work or full‐time education. One study reported episodes off school for one or more days, however (van der Mheen 2019). No study reported on hospital admissions. Submaximal CRF was reported in a variety of ways: the most commonly reported was the oxygen consumption at the gas exchange threshold (GET) scaled to body mass (mL.kg−1.min−1) (n = 5 studies) and the ventilatory equivalents (V̇E) over volume of carbon dioxide production (V̇E/V̇CO2 slope) (n = 4 studies) (Avila 2016; Duppen 2015; Fritz 2020; Moalla 2006; Novakovic 2018; Opotowsky 2018; van Dissel 2019; Westhoff‐Bleck 2013). Absolute oxygen consumption at the GET (mL.min−1), power output in watts at the GET, VE at the GET, heart rate at the GET and the oxygen uptake efficiency slope were all reported once. Muscular strength was only reported by one study using isokinetic testing (Moalla 2006); and adverse events were reported by 11 studies, independent of whether an adverse event actually took place (Avila 2016; Duppen 2015; Fritz 2020; Klausen 2016; Novakovic 2018; Opotowsky 2018; Sandberg 2018; Therrien 2003; van Dissel 2019; Westhoff‐Bleck 2013; Winter 2012).

Excluded studies

We excluded 85 references during the full‐text review, amongst which were 32 due to wrong study design, 15 because they are ongoing trials and 8 because they included the wrong patient population. For more regarding exclusions see Figure 1 and Characteristics of excluded studies. 

Risk of bias in included studies

Risk of bias assessments for each outcome, including all domain judgements and support for judgement, is located in the Risk of bias section (located after the Characteristics of included studies), at the side of all forest plots and in Table 1 for HRQoL. To access further detailed risk of bias assessment data, please use the following link (doi.org/10.24378/exe.2363). 

Risk of bias of outcomes across all studies was similar and predominately of 'some concerns'. Study authors reported poorly
the details of blinding outcome assessors (patient‐facing members of staff conducting the outcome assessments, i.e. the person conducting the exercise test or questionnaire) and pre‐agreed statistical analysis plans with sufficient detail. 

Across most outcomes risk of bias was similar: we judged it as 'some concerns'. The only exception was HRQoL which we judged to be at high risk of bias due to the nature of self‐reported questionnaires, the lack of blinding of the participants and other outcome assessors. 

Effects of interventions

See: Summary of findings 1 Summary of findings for the main comparison. Physical activity and exercise interventions compared to usual care for people with congenital heart disease.

See summary of findings Table 1 and forest plots (Figure 2, Figure 3, Figure 4, Figure 5, Figure 6).

Figure 2
Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at maximal exercise).

Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at maximal exercise).

Figure 3
Exercise training versus no activity (usual care) in people with congenital heart disease. Outcome: Health related quality of life.

Exercise training versus no activity (usual care) in people with congenital heart disease. Outcome: Health related quality of life.

Figure 4
Physical activity promotion and exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Physical activity (device‐worn).

Physical activity promotion and exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Physical activity (device‐worn).

Figure 5
Exercise training interventions versus no activity (usual care) in people with congenital heart disease. . Outcome: Sub‐maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at the gas exchange threshold).

Exercise training interventions versus no activity (usual care) in people with congenital heart disease. . Outcome: Sub‐maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at the gas exchange threshold).

Figure 6
Exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Muscular strength.

Exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Muscular strength.

Maximal cardiorespiratory fitness 

A total of 14 studies (15 training arms, 732 participants) reported maximal CRF using peak oxygen consumption (peak V̇O2) scaled to body mass (mL.kg−1.min−1). One study had a long‐term follow‐up at 36 months post intervention (Winter 2012); all other studies' follow‐up was at the cessation of the intervention (median 12, IQR 12 to 26 weeks). Most studies reported the post‐score mean and standard deviation. However, van Dissel 2019 reported both a post score and a change score from baseline and Opotowsky 2018 only reported change score from baseline. To ensure consistency, change scores were included only when no post score was reported. 

We pooled all available studies into a random‐effects meta‐analysis, with a subgroup analysis comparing the different types of intervention; we did not consider the result of the subgroup analysis to be significant (Chi² = 5.34, df = 2, P = 0.07, I² = 62.5%). In the pooled analysis there was a mean difference (MD) of 1.89 mL.kg−1.min−1 (95% CI −0.22 to 3.99; 14 studies (15 training arms), 732 participants; I2 = 75%). The subgroup exercise training consisted of 11 studies (435 participants) and there was a mean difference of 2.74 mL.kg−1.min−1 (95% CI 0.36 to 5.12; I2 = 73%) versus a mean difference of −1.71 mL.kg−1.min−1 (95% CI −4.64 to 1.22, I2 = 37%) and 0.70 mL.kg−1.min−1 (95% CI −4.83 to 6.23) in physical activity promotion and inspiratory muscle training respectively (Figure 2).

We performed a further subgroup analysis for the type of congenital heart disease, which reported a pooled mean difference of 1.90 mL.kg−1.min−1 (95% CI −0.14 to 3.95; 14 studies (15 training arms), 732 participants; I2 =73%). The test for subgroup differences revealed no differences between subgroups (P = 1.00); single ventricle (MD 2.06, 95% CI −0.25 to 4.38; n = 153), tetralogy of Fallot (MD 1.97, 95% CI −1.11 to 5.05; n = 104) and other or mixed populations (MD 1.98, 95% CI −1.67 to 5.62; n = 474) all had a similar response to a physical activity intervention (Analysis 1.6).

We performed several separate sensitivity analyses removing high risk of bias studies (MD 0.92, 95% CI −0.27 to 2.11; 12 studies (13 training arms), 603 participants; I2 = 18%) (Madhavi 2011; Therrien 2003); and studies that estimated peak V̇O2 using validated protocols (MD 1.07, 95% CI −0.14 to 2.28; 12 studies (13 training arms), 519 participants) (Madhavi 2011; Morrison 2013). We also report the use of fixed‐effect meta‐analyses (MD 2.00, 95% CI 1.09 to 2.91; 14 studies, 732 participants (15 training arms)); the insertion of all available change scores (MD 1.98, 95% CI 0.09 to 3.86; 14 studies (15 training arms), 732 participants) (Sandberg 2018); and the removal of computed outcome scores (converting medians and interquartile ranges to means ± standard deviations from Avila 2016, Fritz 2020, Klausen 2016, Novakovic 2018, Sandberg 2018 and Winter 2012) (MD 2.84, 95% CI −0.21 to 5.88; 8 studies, 423 participants). 

We used univariate meta‐regression to assess individual predictors of peak V̇O2. We regressed 10 predictors and the risk of bias and the intervention length produced significant associations (for regression coefficients and P values see Table 4 and Figure 7). This indicates that the shorter the intervention and the higher the risk of bias then the greater the effect on peak V̇O2. There was no evidence of publication bias (P = 0.268) (Figure 8). Using GRADE, we assessed the evidence to be of moderate certainty because of imprecision. 

Open in table viewer
Table 4. Univariate meta‐regression analysis 

Potential effect modifiers

Regression coefficient and (Standard error)

P value 

Age (adult vs. paediatric)

2.5 (2.2)

0.262

Baseline CRF (peak V̇O2 mL.kg‐1.min‐1)

−0.2 (0.1)

0.186

Dose (intervention length*no. sessions per week*session length)

−0.2 (0.3)

0.614

Follow‐up period/intervention length (weeks)  

−0.18 (0.1)

0.031

Percentage of male 

−0.5 (0.3)

0.083    

Risk of bias

9.1 (2.2)

< 0.01

Sample size 

−0.001 (0.02)

0.963   

Setting (home or hospital based)

−2.3 (1.6)

0.171   

Study location (Continent) 

−3.5 (1.8)

0.070 

Type of intervention (exercise training, PA promotion, IMT)

−2.3 (1.7)

0.208

Bold = statistically significant; CRF, Cardiorespiratory fitness; 

Figure 7
Meta‐regression analyses investigating the effect of the 'overall risk of bias' and the 'length of intervention'. Outcome: Maximal cardiorespiratory fitness (see ).

Meta‐regression analyses investigating the effect of the 'overall risk of bias' and the 'length of intervention'. Outcome: Maximal cardiorespiratory fitness (see Table 4).

Figure 8
Funnel plot investigating publication bias. Outcome: Maximal cardiorespiratory fitness (Egger 1997 test, P=0.268).

Funnel plot investigating publication bias. Outcome: Maximal cardiorespiratory fitness (Egger 1997 test, P=0.268).

Health‐related quality of life (HRQoL)

HRQoL was reported by eight studies using a variety of validated questionnaires and a median follow‐up of 12 weeks (Table 1). The '36‐item short form health survey' (SF‐36) was reported most frequently (n = 5), followed by the 'Congenital heart disease ‐ TNO/AZL adult quality of life questionnaire' (ConHD TAAQoL) which was reported twice. All other questionnaires were reported once. Where possible we pooled HRQoL scores into a random‐effects meta‐analysis; we could enter only three studies into the analyses due to the variety of measurements reported. The result of the analysis was a standardised mean difference of 0.76 (95% CI −0.13 to 1.65; I2 = 82%), which suggests a moderate effect size indicating a possibly beneficial effect of interventions on HRQoL (Figure 3). When we summarised all the evidence on HRQoL presented in the vote count table, however, this is not supported (Table 1). The vote count table aims to summarise all studies and instruments used to report HRQoL. Out of the 12 HRQoL questionnaires reported by the eight studies, only one questionnaire found a significant improvement in HRQoL (Madhavi 2011). Using GRADE, we judged the certainty of the evidence to be 'very low' due to serious to very serious concerns regarding risk of bias, inconsistency and imprecision.  

Device‐worn 'objective' measures of physical activity 

Four studies (328 participants) used device‐worn measures of physical activity and we entered their data into a random‐effects meta‐analysis (Figure 4). The median follow‐up was 19 weeks (IQR 12 to 39 weeks). There is weak evidence of a small effect on physical activity levels with a standardised mean difference of 0.38 (95% CI −0.15 to 0.92; I2 = 78%). The small effect size indicates a possibly beneficial albeit small effect of moderate to vigorous physical activity levels. Re‐expressing these values into the original scales we can report an approximate 10 minute increase per day in moderate to vigorous physical activity (95% CI −2.50 to 22.20). Using GRADE, we downgraded the certainty of evidence by two levels to low, due to concerns over inconsistency and imprecision.  

Validated questionnaire‐based ‘subjective’ measures of physical activity

No study measured physical activity using only questionnaire measures of physical activity; two studies used them in combination with device‐worn measures (Duppen 2015; Klausen 2016). Active leisure time (sports, walking and cycling) was not different after an exercise intervention; passive leisure time (television and computer) reduced significantly in both the intervention and control group, making its attribution to the exercise intervention challenging (Duppen 2015). Klausen 2016 did not report their questionnaire results as it did not differ from their device‐worn measures. 

Return to work or full‐time education

van der Mheen 2019 reported days off school for children participating in a multicomponent (physical activity promotion and psychological) intervention. The intervention group had 11 episodes of one or more days off school versus 13 episodes in one month in the control group, reported by school teachers. Interestingly when this was reported by mothers there was no effect (15 vs. 15) and the direction of effect was the other direction when reported by fathers (13 vs. 11). 

Hospital admissions

No study reported this outcome.

Submaximal cardiorespiratory fitness

A total of nine studies (10 training arms) reported a measure of submaximal CRF, with a median follow‐up of 12 weeks. As previously described in the Characteristics of included studies there was a large variety of submaximal CRF parameters. Oxygen consumption scaled to body mass (mL.kg−1.min−1) at the gas exchange threshold (GET) was reported most often and was subsequently entered into a random‐effects meta‐analysis, showing a likely increase in favour of the intervention with a mean difference of 2.05 (95% CI 0.05 to 4.05; 5 studies (6 training arms), 179 participants; I2 = 33%) mL.kg−1.min−1 (Figure 5). All of the studies that contributed data to this meta‐analysis were exercise training interventions (i.e. not PA promotion or IMT). Using GRADE, we judged the certainty of evidence as moderate—we downgraded the certainty of the evidence one level due to concerns over imprecision (< 200 participants). 

Muscular strength

One study (18 participants) reported muscular strength measured by maximal voluntary contraction (N·m) of knee extensions in paediatrics with congenital heart disease. At the end of the exercise (cycling) intervention (12 weeks) there was a mean difference of 17.13 (95% CI 3.45 to 30.81) N·m in favour of exercise training (Figure 6). Using GRADE, we downgraded the certainty of evidence one level to moderate due to imprecision (only 18 participants).

Adverse events (AEs)

Eleven studies (501 participants) reported on the outcome of adverse events, over a median follow‐up period of 12 weeks (IQR 12 to 26 weeks) (Avila 2016; Duppen 2015; Fritz 2020; Klausen 2016; Novakovic 2018; Opotowsky 2018; Sandberg 2018; Therrien 2003; van Dissel 2019; Westhoff‐Bleck 2013; Winter 2012). Of the eleven studies, six studies reported zero adverse events and five studies reported a total of eleven adverse events. Of the 11 AEs, seven were non‐cardiac (63%), characterised by dizziness, discomfort, minor musculoskeletal and minor head injuries. The remaining four cardiac AEs were inclusive of one suspected arrhythmia, one self‐limiting supraventricular arrhythmia (beta‐blocker administered), one episode of ventricular premature complexes (managed conservatively) and one episode of non‐sustained atrial tachycardia that could be related to exercise. There were no reported serious adverse events or fatality. 

Eight studies (377 participants) reported no adverse myocardial changes; seven studies reported no adverse changes to cardiac biomarker B‐type natriuretic peptide (NT‐proBNP), with a further four studies reporting no structural or functional cardiac effects using medical imaging (cardiac magnetic resonance and echocardiography) post intervention. There were no major adverse events reported. Our judgement of the certainty of evidence using the GRADE approach was moderate due to concerns over inconsistency. 

Discussion

Summary of main results

We identified 15 studies (with 924 participants) that were eligible for inclusion in this review. This review shows that based on moderate to very low certainty of evidence that all types of physical activity interventions (physical activity promotion, exercise training and inspiratory muscle training) when compared to usual care may have a small effect on cardiorespiratory fitness and physical activity level but little or no effect on HRQoL. It should be noted that there was high statistical heterogeneity amongst studies assessing cardiorespiratory fitness, physical activity and HRQoL. Seventy‐three per cent of studies reported adverse events (six studies reported zero adverse events and five studies reported a total of eleven adverse events), of which seven of 11 events were of a non‐cardiac nature, and there were no reported serious adverse events or fatalities related to the physical activity interventions. We were unable to find any data related to secondary outcomes on return to work or hospital admissions. The risk of bias under the outcomes of cardiorespiratory fitness and physical activity was predominantly of 'some concerns', for the outcome of health‐related quality of life it was judged to be a high risk of bias.

Overall completeness and applicability of evidence

The generalisability of previous systematic reviews was either limited to only adults (Li 2019), to a specific type of intervention (Meyer 2020), or to a specific population (Scheffers 2020). This review is the first to include only randomised controlled trial data, of all age groups, types of ConHD and types of physical activity intervention. The findings of this review have potentially better external and ecological validity. Many studies have small sample sizes and all studies were published in the last 17 years. We also report 15 ongoing studies, which indicates there is continuing interest in this area. The quality of the evidence was moderate to very low for all outcomes, indicating further research is very likely to have an important impact on our confidence in the estimate of effect.

Quality of the evidence

Overall, there was a general lack of reporting details of the actual intervention. Using GRADE we assessed the quality of evidence to range from moderate to very low across all outcomes.

We downgraded the certainty of evidence for cardiorespiratory fitness to moderate using GRADE, as the confidence interval includes both appreciable harm and appreciable benefit (i.e. 95% CI spans 0). Therefore, we downgraded the certainty of evidence by one level due to imprecision. 

We downgraded the certainty of evidence for health‐related quality of life included in the meta‐analysis to very low using GRADE. This was due to an inconsistent directions of effect (i.e. 95% CI spans 0), considerable heterogeneity (HRQoL, I2 = 82%), a high risk of bias across all studies and impression due to the low numbers of participants (< 400). We therefore downgraded the certainty of evidence by three levels due to inconsistency, methodological limitations (risk of bias) and imprecision.

We downgraded the certainty of evidence for physical activity to low using GRADE. This was due to an inconsistent direction of effect and considerable heterogeneity (i.e. 95% CI spans 0; I2 = 77%); there was also a low number of participants (< 400). We therefore downgraded the certainty of evidence by two levels due to inconsistency and imprecision. 

We downgraded the certainty of evidence for submaximal cardiorespiratory fitness and muscular strength to moderate using GRADE. This was due to the small numbers of events/participants (< 400). We therefore downgraded the certainty of evidence by one level due to imprecision. 

The certainty of evidence for adverse events was downgraded to moderate using GRADE. This is due to over 25% of studies not reporting data on adverse events. We therefore downgraded the certainty of evidence by one level due to publication bias.

Potential biases in the review process

We have documented and justified alterations to our methods from the published protocol in the Differences between protocol and review section (Williams 2019). 

We believe this is the most comprehensive systematic review to date of RCTs in people with ConHD. However, it has some limitations as the overall risk of bias for the included studies was predominately of 'some concerns'. Specifically, blinding of outcome assessors and statistical analysis plans were poorly reported. It is impossible to blind a physical activity/exercise intervention; there were, however, very few reported attempts to blind trial staff to the allocation of participants during randomisation, assessing the outcomes and statistical analysis of the outcomes.

All included studies reported a 'no formal exercise training' intervention comparator. However, there were three active types of intervention and the amount of data is unequally distributed between these types (PA promotion n = 3; exercise training n = 11; and IMT n = 1). This reduces the certainty of evidence in the less well represented types of interventions; they may have a significant potential for improving primary and secondary outcomes, but it could not be assessed with the limited data.

A limitation of the current data is that studies group patients using their individual ConHD lesion (diagnosis) or group multiple different types of ConHD together in a single cohort. Previous studies have reported large variations in fitness and health status between patients who have the same condition. Future studies should adopt a function‐based assessments/interventions approach, which will enable scientists to observe which types of patients respond better to interventions, improving the evidence base for individualising physical activity interventions (Budts 2013; Budts 2020; Cedars 2020; Moons 2020). 

Agreements and disagreements with other studies or reviews

Cardiorespiratory fitness

Both maximal and submaximal measures of CRF have been shown to be prognostic of future mortality and morbidity in congenital heart disease (Dimopoulos 2006; Giardini 2009; Müller 2015; Udholm 2018). In the current review maximal cardiorespiratory fitness increased by a mean difference of 1.89 mL.kg−1.min−1 (95% CI −0.22 to 3.99). In a healthy population an increase of 3.5 mL.kg−1.min−1 (one MET) reduces the chance of cardiovascular diagnosis or event by approximately 15% (Letnes 2019); and in patients with cardiovascular disease a one MET increase is associated with a 8% to 35% (median 16%) reduction in mortality (Franklin 2013). Currently, in ConHD there is no consensus regarding what the prognostic implication is of an increase of 1.89 mL.kg−1.min−1. A recent systematic review in exercise training in patients with Fontan circulations reported a similar estimate of effect to the current study of 1.73 mL.kg−1.min−1 although this was not conducted using a meta‐analysis (Scheffers 2020).

Our ConHD subgroup analysis reported no difference in the response to the intervention between single ventricle, tetralogy of Fallot and other/mixed ConHD populations (P = 1.0). All subgroups responded similarly to the intervention; this may suggest that a functional‐based classification (over the traditional diagnosis/lesion‐based approach), may help to identify groups who respond better to interventions.

This was the first systematic review and meta‐analysis that assessed submaximal fitness parameters. The oxygen consumption at the gas exchange threshold (GET) improved modestly (MD 2.05, 95% CI 0.05 to 4.05); this has also been accompanied with an increase in power output (watts) at the GET (Moalla 2006; Westhoff‐Bleck 2013). Participants therefore had a greater period of time where they could operate in a predominantly aerobic state, which is an indicator of improved fitness.

Health‐related quality of life

Health‐related quality of life was reported in a variety of ways making pooling difficult: we pooled only three studies and there was a standardised mean difference indicating a moderate effect size (SMD 0.76, 95% CI −0.13 to 1.65), which we judged as very low certainty of evidence. However, using a modified vote‐counting table (Table 1), only one study out of eight showed a significant and positive effect on health‐related quality of life (Madhavi 2011). Gratz 2009 and Amedro 2015 reported that people with ConHD had a significantly poorer health‐related quality of life in the domains of physical functioning/physical well‐being and general health. Gratz 2009 also stated that the ConHD population dangerously overestimate their exercise capacity and this could explain the small to no increase in HRQoL within this review.

Physical activity

Re‐calculating the effect estimate into the original scales (minutes of moderate to vigorous physical activity (MVPA)), we can report an approximate 10‐minute increase per day in MVPA (95% CI −2.50 to 22.20). Whilst this is a small increase of MVPA, accumulatively over the course of a week more participants will be achieving the physical activity guidelines. To our knowledge this is the first review to quantitatively analyse the effects of physical activity interventions on physical activity in people with ConHD. However, were unable to perform a meta‐regression on this outcome, due to the lack of studies contributing to the analyses. 

This review summarises the latest evidence on CRF, HRQoL and PA. Although there were only small improvements in CRF and PA and small to no improvements in HRQoL, there were no serious adverse events related to the interventions or adverse cardiac remodelling. These observations support the proposition that physical activity and exercise is safe and the benefits outweigh the potential risks (Koyak 2012). Although these data are promising, there is currently insufficient evidence to definitively determine the impact of physical activity interventions in ConHD. Therefore, further high‐quality randomised control trials are needed utilising a longer duration of follow‐up. 

Study flow diagram 

Figures and Tables -
Figure 1

Study flow diagram 

Navigate to figure in ReviewOpen in new tab

Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at maximal exercise).

Figures and Tables -
Figure 2

Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at maximal exercise).

Navigate to figure in ReviewOpen in new tab

Exercise training versus no activity (usual care) in people with congenital heart disease. Outcome: Health related quality of life.

Figures and Tables -
Figure 3

Exercise training versus no activity (usual care) in people with congenital heart disease. Outcome: Health related quality of life.

Navigate to figure in ReviewOpen in new tab

Physical activity promotion and exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Physical activity (device‐worn).

Figures and Tables -
Figure 4

Physical activity promotion and exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Physical activity (device‐worn).

Navigate to figure in ReviewOpen in new tab

Exercise training interventions versus no activity (usual care) in people with congenital heart disease. . Outcome: Sub‐maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at the gas exchange threshold).

Figures and Tables -
Figure 5

Exercise training interventions versus no activity (usual care) in people with congenital heart disease. . Outcome: Sub‐maximal cardiorespiratory fitness (V̇O2 mL.kg‐1.min‐1 at the gas exchange threshold).

Navigate to figure in ReviewOpen in new tab

Exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Muscular strength.

Figures and Tables -
Figure 6

Exercise training interventions versus no activity (usual care) in people with congenital heart disease. Outcome: Muscular strength.

Navigate to figure in ReviewOpen in new tab

Meta‐regression analyses investigating the effect of the 'overall risk of bias' and the 'length of intervention'. Outcome: Maximal cardiorespiratory fitness (see ).
Figures and Tables -
Figure 7

Meta‐regression analyses investigating the effect of the 'overall risk of bias' and the 'length of intervention'. Outcome: Maximal cardiorespiratory fitness (see Table 4).

Navigate to figure in ReviewOpen in new tab

Funnel plot investigating publication bias. Outcome: Maximal cardiorespiratory fitness (Egger 1997 test, P=0.268).

Figures and Tables -
Figure 8

Funnel plot investigating publication bias. Outcome: Maximal cardiorespiratory fitness (Egger 1997 test, P=0.268).

Navigate to figure in ReviewOpen in new tab

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 1: Maximal cardiorespiratory fitness

Figures and Tables -
Analysis 1.1

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 1: Maximal cardiorespiratory fitness

Navigate to figure in ReviewOpen in new tab

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 2: Health‐related quality of life

Figures and Tables -
Analysis 1.2

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 2: Health‐related quality of life

Navigate to figure in ReviewOpen in new tab

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 3: Physical activity (device‐worn)

Figures and Tables -
Analysis 1.3

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 3: Physical activity (device‐worn)

Navigate to figure in ReviewOpen in new tab

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 4: Submaximal cardiorespiratory fitness (gas exchange threshold)

Figures and Tables -
Analysis 1.4

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 4: Submaximal cardiorespiratory fitness (gas exchange threshold)

Navigate to figure in ReviewOpen in new tab

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 5: Muscular strength

Figures and Tables -
Analysis 1.5

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 5: Muscular strength

Navigate to figure in ReviewOpen in new tab

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 6: Maximal cardiorespiratory fitness (type of ConHD subgroup analysis)

Figures and Tables -
Analysis 1.6

Comparison 1: Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease, Outcome 6: Maximal cardiorespiratory fitness (type of ConHD subgroup analysis)

Navigate to figure in ReviewOpen in new tab
Summary of findings 1. Summary of findings for the main comparison. Physical activity and exercise interventions compared to usual care for people with congenital heart disease.

Patient or population: people with congenital heart disease.
Setting: hospital‐based and home‐based settings.
Intervention: physical activity promotion, inspiratory muscle training and exercise training interventions. 
Comparison: usual care.

Outcomes 

Length of intervention in weeks (Median, (IQR))

Anticipated absolute effects* (95% CI)

No. of participants (studies)

Certainty of the evidence (GRADE)

Comments

Risk with usual care
 

Risk with all interventions
 

Maximal cardiorespiratory fitness (CRF)

Assessed with: treadmill or cycle ergometry

(12 (12, 26))

Mean 22 to 46

MD 1.89 higher (0.22 lower to 3.99 higher)

732

(14 RCTs (15 training arms))

 ⊕⊕⊕⊝ab
MODERATE

Physical activity and exercise interventions may increase cardiorespiratory fitness slightly. Sensitivity analyses did not change the inference to a clinically important MD.

Health‐related quality of life (HRQoL) Assessed with: Questionnaires

(12 (12, 12))

‐

SMD 0.76 higher (0.13 lower to 1.65 higher)

163 (3 RCTs)

 ⊕⊝⊝⊝cde
VERY LOW

The evidence is very uncertain about the effect of physical activity and exercise interventions on health‐related quality of life. 

Physical activity (PA)

Assessed with: Accelerometer

(19 (12, 46))

Mean 11 to 41

SMD 0.38 
(0.15 lower to 0.92 higher)

328 (4 RCTs)

 ⊕⊕⊝⊝ce
LOW

Physical activity and exercise interventions may increase physical activity slightly.

Submaximal cardiorespiratory fitness

Assessed with: treadmill or cycle ergometry (VO2 mL.kg‐1.min‐1 at the gas exchange threshold).

(12 (12, 15))

Mean 18 to 22

MD 2.05  (0.05 higher to 4.05 higher)

179 (5 RCTs (6 training arms))

 ⊕⊕⊕⊝e
MODERATE

Physical activity and exercise interventions likely results in an increase in submaximal cardiorespiratory fitness.

Muscular strength

(12)

Mean 103

MD 17.13  (3.45 higher to 30.81 higher)
 

18 (1 RCT)

 ⊕⊕⊕⊝e
MODERATE

Physical activity and exercise interventions likely increases muscular strength.

Adverse events

(12 (12, 26))

A total of 11 adverse events were reported in a population of 501 (2.2%), 7 of which were mild (1.4%). Mild adverse events included dizziness, discomfort, musculoskeletal (strains/ sprains) and a minor head injury. The 4 (0.8%) moderate adverse events were cardiac. There were no major adverse events reported.

501 (11 RCTs)

 ⊕⊕⊕⊝f
MODERATE

Physical activity promotion and exercise training interventions did not lead to any serious adverse events.

Hospital admissions 

No data available 

CI: confidence interval; IQR: 25th and 75th quartiles; MD: mean difference; SMD: standardised mean difference; RCT: randomised controlled trial; CRF: Cardiorespiratory fitness; HRQoL: Health‐related quality of life. The mean SMD effect size was interpreted using Cohen effect sizes, i.e. 0.2 represents a small effect, 0.5 a moderate effect, and 0.8 a large effect. 

GRADE Working Group grades of evidence.
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. 

a Statistical heterogeneity was present in CRF (I2 = 75%). But it was explained using a sensitivity analysis by excluding high risk of bias studies (I2 = 18%). Therefore, we chose not to downgrade by 1 level evidence due to inconsistency. 

bThe confidence interval includes both appreciable harm and appreciable benefit (i.e. 95% CI spans 0). Therefore, the certainty of evidence was downgraded by 1 level due to imprecision. 

cInconsistent directions of effect and considerable heterogeneity (HRQoL, I2 = 82%; PA, I2 = 77%). Therefore, the certainty of evidence was downgraded by 1 level due to inconsistency. 

dTotal high risk of bias in all studies included within the analysis, hence bias highly likely. Therefore, the certainty of evidence was downgraded by 2 levels due to the methodological limitations (risk of bias).

eImprecise due to small numbers of events (< 400) (Ryan 2016).  Therefore, the certainty of evidence was downgraded by 1 level due to imprecision. 

fOver 25% of studies did not report data on adverse events. Therefore, the certainty of evidence was downgraded by 1 level due to publication bias.

Figures and Tables -
Summary of findings 1. Summary of findings for the main comparison. Physical activity and exercise interventions compared to usual care for people with congenital heart disease.
Navigate to table in Review
Table 1. Health‐related quality of life

Author

Sample size

Type of intervention

Questionnaire

Domain

Intervention vs. control at follow up

Mean (Standard deviation) & between group P value

Direction of effect

Overall risk of bias

Duppen 2015
 

73
 

Exercise training
 

SF‐36 (8 domains) +

Physical functioning

94.6 (10.9) 95.0 (8.5) P = 0.71 

 Exercise = control

High

SF‐36 (8 domains) +

Bodily pain

96.6 (8.1) 93.2 (17.3) P = 0.96

 Exercise = control

High

SF‐36 (8 domains) +

General health

68.3 (27.9) 67.5 (19.1) P = 0.94 

 Exercise = control

High

SF‐36 (8 domains) + 

 Vitality

71.7 (18.0) 70 (17.0) P = 0.56 

 Exercise = control

High

SF‐36 (8 domains) +

Role limitations due to physical limitations

100 (0.0) 91.7 (21.2) P = 0.16 

 Exercise = control

High

SF‐36 (8 domains) +

Social functioning

95.8 (10.0) 100.0 (0.0)P = 0.09 

 Exercise = control

High

SF‐36 (8 domains) +

Role limitations due to emotional problems

100 (0.0) 100 (0.0) P = 0.72

 Exercise = control

High

SF‐36 (8 domains) +

Mental health

85.0 (16.8) 81.3 (10.2) P = 0.65

 Exercise = control

High

TACQOL +

Symptoms 

95.6 (7.90) 97.8 (3.7) P = 0.16

 Exercise = control

High

TACQOL +

Impact of cardiac surveillance

85 (6.30) 85.7 (4.80) P = 0.07 

 Exercise = control

High

TACQOL +

Worries

95.7 (8.70) 88.6 (270) P = 0.67 

 Exercise = control

High

TACQOL‐CF +

Pain and physical symptoms

26.3 (6.2) 24 (7.15) P = 0.21

 Exercise = control

High

TACQOL‐CF +

Motor functioning

30.1 (1.9) 29.5 (4.37) P = 0.51

 Exercise = control

High

TACQOL‐CF +

Cognitive functioning

28.6 (6.2) 29.7 (6.9) P = 0.05

 Exercise = control

High

TACQOL‐CF + 

Social functioning

31.5 (1.2) 32 (0) P = 0.45

 Exercise = control

High

TACQOL‐CF +

Positive emotional functional

14.2 (3.5) 14.43 (2.9) P = 0.39

 Exercise = control

High

TACQOL‐CF +

Negative emotional functioning

12.7 (2.12) 14.2 (2.2) P = 0.34

 Exercise = control

High

Madhavi 2011

112

Exercise training

SF‐36 +

SF36 total score

23.3  (13.2) 11.3 (14.3) P < 0.001

 Exercise > control

High

Klausen 2016
 

158

Physical activity promotion
 

Danish Paediatric QoL Inventory +

Generic

NR

 Exercise = control

High

Danish Paediatric QoL Inventory +

Disease specific

NR

 Exercise = control

High

Opotowsky 2018

28

Exercise training

MLHFQ ‐ 

MLHFQ

20.1 (11.4) 27.7 (10.9) P = 0.13

 Exercise = control

High

Sandberg 2018

23

Exercise training

EQ5D VAS +

EQ5D VAS

76.2 (15.2) 76.3 (20.7) P = 0.31

 Exercise = control

High

Novakovic 2018
 

14 

Exercise training (Interval)
 

SF36 +

Physical component

86.7 (40.2) 101.0 (16.6) P > 0.05

 Exercise = control

High

SF36 +

Mental component

80.0 (21.0) 87.3 (21.9) P > 0.05

 Exercise = control

High

Novakovic 2018
 

13 

Exercise training (Continuous)
 

SF36 +

Physical component

103 (5.2) 101.0 (16.6) P > 0.05

 Exercise = control

High

SF36 +

Mental component

89.3 (18.4) 87.3 (21.9)  P > 0.05

 Exercise = control

High

Westhoff-Bleck 2013

40

Exercise training

 KCCQ +

 KCCQ

NR

 Exercise = control

High

Winter 2012
 

 

46
 
 

Exercise training
 

 

SF‐36 +

Mental component

P = 0.17

 Exercise = control

High

SF‐36 +

Physical component

P = 0.20

 Exercise = control

High

ConHD‐TAAQOL +

Symptoms

85 (10.79) 83 (10.47) P = 0.31

 Exercise = control

High

ConHD‐TAAQOL +

Worries

77 (15.4) 84.5 (9.21) P = 0.30

 Exercise = control

High

ConHD‐TAAQOL +

Impact

84 (7.19) 85.25 (6.20) P = 0.91

 Exercise = control

High

Each study's outcome of health related quality of life was individually assessed using risk of bias 2; all studies were judged as a high risk of bias under the domain 'Measurement of the outcome'. '+' =  a higher score represents better health; '‐' =  a lower score represents better health; HRQoL, Health related quality of life; SF‐36,  36‐Item Short Form Health Survey;  MLHFQ, Minnesota living with heart failure questionnaire; TACQOL, TNO/AZL child quality of life questionnaire; ConHD TAAQOL, The congenital heart disease ‐ TNO/AZL adult quality of life questionnaire; EQ5D VAS, EuroQol Vertical Visual Analogue Scale; KCCQ, Kansas City Cardiomyopathy Questionnaire.

Figures and Tables -
Table 1. Health‐related quality of life
Navigate to table in Review
Table 2. Individual ConHD lesions pooled into the meta‐analyses

Study

Age

Type of ConHD and (sample size, %)

Classification

Avila 2016

35 ± 11

Repaired ToF (17, 100%)

Severe

Duppen 2015

15 ± 3

Fontan circulation (43, 48%)

[Intra‐atrial lateral tunnel 47%; extracardiac conduit 44%; other 9%]

Severe

Repaired ToF (47, 52%)
 

Fritz 2020

30 ± 9

Fontan circulations (42, 100%)

[Atrioventricular Anastomosis 19%; Atriopulmonary Anastomosis 21%; Total Cavopulmonary Connection 60%]

Severe

Klausen 2016

15 ± 1

Coarctation of the aorta (52, 33%)

Severe

TGA (35, 22%)

Steno–Fallot tetralogy (21, 13%)

Double outlet right ventricle (7, 4%)

Truncus arteriosus (4, 3%)

Atrioventricular septal defect (9, 6%)

TCPC (6, 4%)

Other (24, 15%)

Madhavi 2011

29 ± 11

Acyanotic Congenital Heart Disease (112, 100%)

Mild

Moalla 2006

13 ± 1

Fontan (4, 22%)

Mild to  Severe

TGA (Senning/Mustard procedure) (5, 28%)

Repaired ToF (5, 28%)

Repaired ASD (4, 22%)

Morrison 2013

15 ± 2

Minor CHD (no surgical intervention) (39, 27%)

Mild to Severe

Acyanotic corrected ConHD (61, 43%)

Cyanotic corrected (30, 21%)

Cyanotic palliated (13, 9%)

Novakovic 2018

38 ± 8

Repaired ToF (30, 100%)

Severe

Opotowsky 2018

41 ±12

ToF with pulmonary stenosis or atresia or, DORV (13, 46%)

Severe

TGA with a systemic RV, (9, 32%)

Fontan (2, 7%)

Pulmonary atresia with intact ventricular septum with biventricular repair (2, 7%)

Truncus arteriosus (1, 4%)

Ebstein anomaly (1, 4%)

Sandberg 2018

30  ±  11

ToF (5, 22%)

Severe

ccTGA & d‐TGA (8, 35%)

TCPC (5, 22%)

PA (2, 9%)

Complete AV‐septal defect (1, 4%)

Ebstein anomaly (1, 4%)

Miscellaneous (1, 4%)

Therrien 2003

35 ± 9

Repaired ToF (18, 100%)

Severe

Westhoff‐Bleck 2013

29 ± 3

Systemic right ventricle ‐ TGA [Mustard] (48, 100%)

Severe

Winter 2012

32 + 11

Systemic right ventricle ‐ TGA and ccTGA (46, 100%)

Severe

van Dissel 2019

 

40 ± 9

 

ToF (12, 30%)

Severe

 

TGA (13, 33%)

Fontan circulation (9, 22%)

Pulmonary atresia (3, 7.5%)

Other (3, 7.5%)

ToF, tetralogy of Fallot; TGA, transposition of the great arteries; ccTGA, congenitally corrected transposition of the great arteries; d‐TGA, dextro‐transposition of the great arteries; ASD, atrial septal defect; TCPC, Total cavopulmonary connection; DORV, double‐outlet right ventricle; RV, right ventricle.

Figures and Tables -
Table 2. Individual ConHD lesions pooled into the meta‐analyses
Navigate to table in Review
Table 3. Characteristics of exercise training interventions

Study

Age group & severity 

Location & supervision

Frequencya (sessions per week)

Intensity

Timeb (minutes)

Type

Durationc (Weeks)

Dose (a*b*c)

 Avila 2016

Adult & severe (ToF)

Hospital & supervised

 1 to 2

70% to 80% of the maximum HR (increased throughout intervention)

60

Combination of resistance and aerobic dynamic (running, rowing etc.) exercise

12

1080

 Duppen 2015

Paediatric & severe (ToF and Fontan)

Hospital & supervised
 

2 to 3

Resting heart rate plus 60% to 70% of the HR reserve

60
 

Aerobic dynamic 

12
 

1800
 

Madhavi 2011

Adult & mild

NR 

NR 

Individualised (NR) 
 

NR
 

NR

12

N/A

Moalla 2006

Paediatric & mild/severe 

Home & semi‐supervised

3
 

HR at the gas exchange threshold

60
 

Cycling

 12

2160
 

Novakovic 2018
 

Adult & severe (ToF)
 

NR
 

2 to 3 

 80% of HRpeak in high intensity exercise

42
 

Cycling or speed walking 

12
 

1260
 
 

70% of HRpeak in continuous intensity exercise 

Opotowsky 2018

 Adult & severe

Hospital & supervised 

2
 

HR at Gas Exchange Threshold 

60
 

Combination of resistance and aerobic dynamic 

12
 

1440
 

Sandberg 2018

Adult & severe

Home  & semi‐supervised

3

HR was calculated according to the Karvonen method and to achieve BORG 15 to 16 

31

Cycling

12

1116

Therrien 2003

Adult & severe (ToF)

Hospital and home (1:2 ratio) & supervised

3

 60% to 85% of pre training peak VO2 

50

Cycling and walking
 

12

1800

Westhoff‐Bleck 2013
 

Adult & severe (Mustard procedure)

Home &  semi‐supervised 

3‐5

HR corresponding to 50% of peak VO2 

10 to 30

Cycling

24

2550d

Winter 2012

Adult & severe  

Home & semi‐supervised 

3

75% to 90% of max heart  rate (increased throughout intervention)

42

Step aerobics

10

1260

van Dissel 2019

Adult & Moderate and Severe 

Home & semi‐supervised 

3
 

80% of the maximum HR 

45
 

Self‐selected

26

3510

ToF, Tetralogy of Fallot; HR, Heart Rate; NR, Not Reported; d See dose calculations in Characteristics of included studies

Figures and Tables -
Table 3. Characteristics of exercise training interventions
Navigate to table in Review
Table 4. Univariate meta‐regression analysis 

Potential effect modifiers

Regression coefficient and (Standard error)

P value 

Age (adult vs. paediatric)

2.5 (2.2)

0.262

Baseline CRF (peak V̇O2 mL.kg‐1.min‐1)

−0.2 (0.1)

0.186

Dose (intervention length*no. sessions per week*session length)

−0.2 (0.3)

0.614

Follow‐up period/intervention length (weeks)  

−0.18 (0.1)

0.031

Percentage of male 

−0.5 (0.3)

0.083    

Risk of bias

9.1 (2.2)

< 0.01

Sample size 

−0.001 (0.02)

0.963   

Setting (home or hospital based)

−2.3 (1.6)

0.171   

Study location (Continent) 

−3.5 (1.8)

0.070 

Type of intervention (exercise training, PA promotion, IMT)

−2.3 (1.7)

0.208

Bold = statistically significant; CRF, Cardiorespiratory fitness; 

Figures and Tables -
Table 4. Univariate meta‐regression analysis 
Navigate to table in Review
Comparison 1. Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 Maximal cardiorespiratory fitness Show forest plot

14

732

Mean Difference (IV, Random, 95% CI)

1.89 [‐0.22, 3.99]

1.1.1 Exercise training

11

435

Mean Difference (IV, Random, 95% CI)

2.74 [0.36, 5.12]

1.1.2 Physical activity promotion

2

259

Mean Difference (IV, Random, 95% CI)

‐1.71 [‐4.64, 1.22]

1.1.3 Inspiratory muscle training

1

38

Mean Difference (IV, Random, 95% CI)

0.70 [‐4.83, 6.23]

1.2 Health‐related quality of life Show forest plot

3

163

Std. Mean Difference (IV, Random, 95% CI)

0.76 [‐0.13, 1.65]

1.3 Physical activity (device‐worn) Show forest plot

4

328

Std. Mean Difference (IV, Random, 95% CI)

0.38 [‐0.15, 0.92]

1.4 Submaximal cardiorespiratory fitness (gas exchange threshold) Show forest plot

5

179

Mean Difference (IV, Random, 95% CI)

2.05 [0.05, 4.05]

1.5 Muscular strength Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

1.6 Maximal cardiorespiratory fitness (type of ConHD subgroup analysis) Show forest plot

14

732

Mean Difference (IV, Random, 95% CI)

1.90 [‐0.14, 3.95]

1.6.1 Population with a single ventricle

4

154

Mean Difference (IV, Random, 95% CI)

2.05 [‐0.25, 4.35]

1.6.2 Population with repaired Tetralogy of Fallot

4

104

Mean Difference (IV, Random, 95% CI)

1.97 [‐1.11, 5.05]

1.6.3 Other or mixed ConHD populations

7

474

Mean Difference (IV, Random, 95% CI)

1.98 [‐1.67, 5.62]
Figures and Tables -
Comparison 1. Physical activity promotion, exercise training and inspiratory muscle training interventions versus no activity (usual care) in people with congenital heart disease
Navigate to table in Review
Risk of bias for analysis 1.1 Maximal cardiorespiratory fitness

Bias

Study

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Subgroup 1.1.1 Exercise training

Therrien 2003

 High risk of bias Low risk of bias Low risk of bias Some concerns Some concerns High risk of bias

No information on method of randomisation and significant baseline imbalance between groups. There is a 8 year age gap – the intervention group are younger and age of repair was younger. Right ventricular outflow tract (11 vs 22 mmhg) were half in the intervention group. Daily activity levels were less in the intervention group. This may suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate. 

17 of 18 participants (95%) were included at follow up. One person lost from the control group due to a lack of interest.

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

No registry of protocol available. Common CRF parameter reported. No information whether there were multiple eligible analyses. 

Lack of information on randomisation, baseline imbalances and blinding of outcome assessors and no pre‐registered protocol or statistical plan. 

Moalla 2006

 Some concerns Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

There was no information on method of randomisation, there was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

All outcome data was presented

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Overall due to a lack of detail on randomisation, blinding outcome assessors and lack of information in the pre‐registered methods.

Madhavi 2011

 High risk of bias Some concerns High risk of bias High risk of bias Some concerns High risk of bias

Patients were randomly divided into study and control group by simple randomization using the lottery method. However there were substantial differences in patients fitness (as measured by VO2 peak) between the intervention and control group.

Both participants and those delivering the intervention were aware of intervention received. There was no information to support deviations from the intended intervention.

Loss to follow up not explained. All people lost to follow up were in the intervention group. No analysis to check the effect of the missing data.

YMCA cycle test. This is a predictive test not a direct assessment furthermore no description of protocol. Furthermore, there was a lack of blinding outcome assessors and a high effect size seen.

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Overall high due to differences at baseline, missing outcome data without explanation and a lack of blinding outcome assessors. 

Winter 2012

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

“Randomization was performed using sealed envelops. Each participant chose an opaque envelop from a shuffled stack which contained either ‘yes’, which allocated him to the treatment group or ‘no’ which allocated him to the control group. Randomization was stratified by participating centre.” There were no baseline imbalances that would suggest a problem with randomisation. 

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

15% of participants were lost to follow‐up. Dropping out of the study could be related to the outcome. However, the half of participants who dropped out were for reasons that are unrelated to CRF

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

There was a trial register entry. It provided details of time points, but not of analyses nor of specific means of measuring outcomes (specific scales or methods). There is a risk that outcomes were selected after analysis completed

Overall judged some concerns due to missing outcome data, no information on blinding of outcome assessors and the risk that outcomes were selected after analysis completed.

Westhoff‐Bleck 2013

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

There was no information on randomisation or allocation concealment. There are no baseline differences that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

Overall: 16% of people dropped out. Similar numbers from both groups. Reasons for dropping out for some participants were duration of intervention, but some reasons were unrelated. There was no analysis to assess the effect of missing data 

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

There was a trial register entry.  It provided details of time points, but not of analyses nor of specific means of measuring outcomes (specific scales or methods). There is a risk that analyses were chosen to selectively. 

Overall some concerns due to no information about blinding of outcome assessors, pre published statistical plan and lack of information on randomisation.

Duppen 2015

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized in a 2:1 ratio to the exercise group or control group. Randomization was performed by an independent blinded researcher. Stratification was based on gender, ConHD, and age group (10‐12, 12‐15, 15‐18, and 18‐25 years). There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Most people were followed up >95%

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Avila 2016

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Randomization was performed by sealed envelopes, with a computer‐generated allocation sequence. There were no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

All participants were included in the analysis

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Novakovic 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized into 3 groups (2 interventional and 1 control group) with a 1:1:1 ratio using adaptive (urn) randomisation with sealed envelopes and randomisation concealment from the recruiting investigator. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention.

Only 2 people discontinued intervention and 1 person was lost to follow up, reasons given. 90% of randomised sample accounted for in outcome data

Measurements were similar across all intervention groups but there is no information in whether the outcome assessors were blinded. Outcome assessors, through encouragement during testing, could affect the outcome, therefore we have some concerns about potential bias.

Protocol registered (NCT02643810). Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Novakovic 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized into 3 groups (2 interventional and 1 control group) with a 1:1:1 ratio using adaptive (urn) randomisation with sealed envelopes and randomisation concealment from the recruiting investigator. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention.

Only 2 people discontinued intervention and 1 person was lost to follow up, reasons given. 90% of randomised sample accounted for in outcome data

Measurements were similar across all intervention groups but there is no information in whether the outcome assessors were blinded. Outcome assessors, through encouragement during testing, could affect the outcome, therefore we have some concerns about potential bias.

Protocol registered (NCT02643810). Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Opotowsky 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Generation of randomisation sequence and concealment of allocation were adequate. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

All participants were included in the analysis

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Sandberg 2018

 Low risk of bias Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns

Computer based random allocation with allocation and no baseline imbalances that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate. 

12% of participants were lost to follow up 3/16 intervention and 0/13 control.  There is no analysis to look at the effect of the missing data. A small study so the 3 drop outs looks like a large percentage, reasons documented for patient withdrawal.

“This study was performed at two centers. The center recruiting 5 patients (22%) was not able to keep the investigators performing the exercise tests strictly blinded for group allocation.”

The trial register report was registered before the trial (date 2009) It indicates that CRF would be measured using Vo2 Max. 

Unable to locate a detailed statistical plan

van Dissel 2019

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

Computer based random allocation and no baseline imbalances that would suggest a problem with randomisation, but there was no information about allocation concealment.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

Overall: 15% of people dropped out, similar numbers from both groups. Reasons for dropping out for some participants were duration of intervention, but some reasons were unrelated. There was no analysis to assess the effect of missing data 

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to a lack of detail on randomisation, missing outcome data, blinding outcome assessors and lack of information in the pre‐registered methods.

Subgroup 1.1.2 Physical activity promotion

Morrison 2013

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

Participants were randomised using balanced blocks of four to intervention and control groups. Randomisation was not stratified by any factor. There were some baseline imbalances in BMI and Deprivation, which could be of importance in a physical activity promotion trial. The concerns were due to no information on the randomisation process rather than the baseline imbalances.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Twenty‐three participants did not attend reassessment despite being offered three separate appointments, and a number of patients withdrew at this stage. Four patients who were randomised to the intervention group did not attend the group sessions offered. These patients were analysed with the rest of the intervention group on an intention‐to‐treat basis. There was no analysis to assess the effect of missing data.

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Judged some concerns over issues with randomisation, blinding outcome assessors, indirect measurements of CRF, missing outcome data, and no prior published statistical plan.

Klausen 2016

 Low risk of bias Low risk of bias Some concerns Low risk of bias Low risk of bias Some concerns

The randomization was performed by the Copenhagen Trial Unit ten days after the baseline test. A trial investigator centrally randomized eligible patients 1:1. The computer‐generated allocation sequence used permuted blocks with varying sizes of 6, 8, or 10. The allocation sequence and block size were unknown to the trial investigators. There was no baseline imbalance that would suggest a problem with randomisation.

Deviations from the original protocol were approved by the Regional Ethics Committee trial protocol and applied to the ClinicalTrials.gov identifier. Participants  were aware of intervention received. There were no deviations from intervention. The team was blinded to the group allocation of patients and the analysis was appropriate.

39 lost to follow up (24%).Multiple imputations were used to handle missing data. However, multiple imputation based on 'missing at random' does not provide justification to say that there was evidence that results were robust.

The team was blinded to the group allocation of patients.

Protocol registered, with a pre published protocol and statistical plan. Industry standard CRF data were presented.

Some concerns due to 24% loss to follow up.

Subgroup 1.1.3 Inspiratory muscle training

Fritz 2020

 Some concerns Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

There was no information on method of randomisation, there was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention. 

9.5% dropout rate in this study, however for this type of intervention this is reasonable. Authors accounted for 20% drop out in sample size calculations. 

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Overall due to a lack of detail on randomisation, blinding outcome assessors and lack of information in the pre‐registered methods.

Figures and Tables -
Risk of bias for analysis 1.1 Maximal cardiorespiratory fitness
Navigate to table in Review
Risk of bias for analysis 1.2 Health‐related quality of life

Bias

Study

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Madhavi 2011

 High risk of bias Some concerns High risk of bias High risk of bias Some concerns High risk of bias

Patients were randomly divided into study and control group by simple randomization using the lottery method. However there were substantial differences in patients fitness (as measured by VO2 peak) between the intervention and control group.

Both participants and those delivering the intervention were aware of intervention received. There was no information to determine whether there were deviations from the intended intervention.

Loss to follow up not explained. All people lost to follow up were in the intervention group.No analysis to check the effect of the missing data.

The participant assessed this outcome and was aware of the intervention they received. HRQoL is a patient reported outcome. Therefore, it is not unrealistic it could be effected by knowledge of allocation to intervention.

No pre‐registered method (registry or protocol) available. SF‐36 is usually reported in 2 or 8 domains, they report a 'total score' that is statistically significant. Potentially selective reporting due to multiple outcome scales.

High risk across randomisation, missing outcome data and blinding of the outcome assessor. 

Opotowsky 2018

 Low risk of bias Low risk of bias Low risk of bias High risk of bias Some concerns High risk of bias

Generation of randomisation sequence and concealment of allocation were adequate. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

All participants were included in the analysis

The participant assessed this outcome and was aware of the intervention they received. HRQoL is a patient reported outcome. Therefore, it is not unrealistic it could be effected by knowledge of allocation to intervention

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard HRQoL data were presented.

Overall high due to the nature of self reported questionnaires and patients being aware of their assignment.

Sandberg 2018

 Low risk of bias Low risk of bias Low risk of bias High risk of bias Some concerns High risk of bias

Computer based random allocation with allocation and no baseline imbalances that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate

12% of participants were lost to follow up 3/16 intervention and 0/13 control.  There is no analysis to look at the effect of the missing data. A small study so the 3 drop outs looks like a large percentage, reasons documented for patient withdrawal.

The participant assessed this outcome and was aware of the intervention they received. HRQoL is a patient reported outcome. Therefore, it is not unrealistic it could be effected by knowledge of allocation to intervention

The trial register report was registered before the trial (date 2009) It indicates that HRQoL was to be measured at 12 weeks using both EQ5D and SF‐36. Only EQ‐5D is presented .

Overall high due to the nature of self reported questionnaires and patients being aware of their assignment.

Figures and Tables -
Risk of bias for analysis 1.2 Health‐related quality of life
Navigate to table in Review
Risk of bias for analysis 1.3 Physical activity (device‐worn)

Bias

Study

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Duppen 2015

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized in a 2:1 ratio to the exercise group or control group. Randomization was performed by an independent blinded researcher. Stratification was based on gender, ConHD, and age group (10‐12, 12‐15, 15‐18, and 18‐25 years). There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Most people were followed up >95%

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard physical activity data were presented.

Blinding outcome assessors and potential bias in the reported result.

Klausen 2016

 Low risk of bias Low risk of bias Some concerns Low risk of bias Low risk of bias Some concerns

The randomization was performed by the Copenhagen Trial Unit ten days after the baseline test. A trial investigator centrally randomized eligible patients 1:1. The computer‐generated allocation sequence used permuted blocks with varying sizes of 6, 8, or 10. The allocation sequence and block size were unknown to the trial investigators. There was no baseline imbalance that would suggest a problem with randomisation.

Participants  were aware of intervention received. There were no deviations from intervention. The team was blinded to the group allocation of patients and the analysis was appropriate

39 lost to follow up (24%).Multiple imputations were used to handle missing data. However, multiple imputation based on 'missing at random' does not provide justification to say that there was evidence that results were robust.

The team was blinded to the group allocation of patients.

Protocol registered, with a pre‐published protocol and statistical plan. Industry standard activity (MVPA) data were presented with an appropriate statistical approach.

Some concerns due to 24% loss to follow up.

Morrison 2013

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

Participants were randomised using balanced blocks of four to intervention and control groups. Randomisation was not stratified by any factor. There were some baseline imbalances in BMI and Deprivation, which could be of importance in a physical activity promotion trial. The concerns were due to no information on the randomisation process rather than the baseline imbalances.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Twenty‐three participants did not attend reassessment despite being offered three separate appointments, and a number of patients withdrew at this stage. Four patients who were randomised to the intervention group did not attend the group sessions offered. These patients were analysed with the rest of the intervention group on an intention‐to‐treat basis. There was no analysis to assess the effect of missing data.

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard physical activity data were presented.

Only judged low risk in one domain. Missing information on randomisation with baseline imbalances, missing outcome data, no information on whether outcome assessors were aware of the intervention received, no pre registered statistical plan.  

Opotowsky 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Generation of randomisation sequence and concealment of allocation were adequate. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate

All participants were included in the analysis

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard physical activity data were presented.

Some concerns in 2 domains measurement of outcome and selection of the reported result
Figures and Tables -
Risk of bias for analysis 1.3 Physical activity (device‐worn)
Navigate to table in Review
Risk of bias for analysis 1.4 Submaximal cardiorespiratory fitness (gas exchange threshold)

Bias

Study

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Moalla 2006

 Some concerns Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

There was no information on method of randomisation, there was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

All/most data available

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Overall due to a lack of detail on randomisation, blinding outcome assessors and lack of information in the pre‐registered methods.

Westhoff‐Bleck 2013

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

There was no information on randomisation or allocation concealment. There are no baseline differences that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

Overall: 16% of people dropped out. Similar numbers from both groups. Reasons for dropping out for some participants were duration of intervention, but some reasons were unrelated. There was no analysis to assess the effect of missing data 

 There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

There was a trial register entry.  It provided details of time points, but not of analyses nor of specific means of measuring outcomes (specific scales or methods). There is a risk that analyses were chosen to selectively. 

Overall some concerns due to the potential lack of blinding, pre published statistical plan and lack of information on randomisation.

Duppen 2015

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized in a 2:1 ratio to the exercise group or control group. Randomization was performed by an independent blinded researcher. Stratification was based on gender, ConHD, and age group (10‐12, 12‐15, 15‐18, and 18‐25 years). There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Most people were followed up >95%

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Avila 2016

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Randomization was performed by sealed envelopes, with a computer‐generated allocation sequence. There were no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

All participants were included in the analysis

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Novakovic 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized into 3 groups (2 interventional and 1 control group) with a 1:1:1 ratio using adaptive (urn) randomisation with sealed envelopes and randomisation concealment from the recruiting investigator. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Only 2 people discontinued intervention and 1 person was lost to follow up, reasons given. 90% of randomised sample accounted for in outcome data

Measurements were similar across all intervention groups but there is no information in whether the outcome assessors were blinded. Outcome assessors, through encouragement during testing, could affect the outcome, therefore we have some concerns about potential bias.

Protocol registered (NCT02643810). Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented. 

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Novakovic 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized into 3 groups (2 interventional and 1 control group) with a 1:1:1 ratio using adaptive (urn) randomisation with sealed envelopes and randomisation concealment from the recruiting investigator. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Only 2 people discontinued intervention and 1 person was lost to follow up, reasons given. 90% of randomised sample accounted for in outcome data

Measurements were similar across all intervention groups but there is no information in whether the outcome assessors were blinded. Outcome assessors, through encouragement during testing, could affect the outcome, therefore we have some concerns about potential bias.

Protocol registered (NCT02643810). Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented. 

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Figures and Tables -
Risk of bias for analysis 1.4 Submaximal cardiorespiratory fitness (gas exchange threshold)
Navigate to table in Review
Risk of bias for analysis 1.5 Muscular strength

Bias

Study

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Moalla 2006

 Some concerns Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns

There was no information on method of randomisation, there was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

All or most patients were included at follow up

Gold standard measure of strength and unlikely outcome assessors can influence score. 

No pre‐registered method (registry or protocol) available. However, industry standard strength data is available.

Due to lack of information on randomisation and lack of a pre‐registered protocol. 
Figures and Tables -
Risk of bias for analysis 1.5 Muscular strength
Navigate to table in Review
Risk of bias for analysis 1.6 Maximal cardiorespiratory fitness (type of ConHD subgroup analysis)

Bias

Study

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Subgroup 1.6.1 Population with a single ventricle

Winter 2012

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

“Randomization was performed using sealed envelops. Each participant chose an opaque envelop from a shuffled stack which contained either ‘yes’, which allocated him to the treatment group or ‘no’ which allocated him to the control group. Randomization was stratified by participating centre.” There were no baseline imbalances that would suggest a problem with randomisation. 

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

15% of participants were lost to follow‐up. Dropping out of the study could be related to the outcome. However, the half of participants who dropped out were for reasons that are unrelated to CRF

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

There was a trial register entry. It provided details of time points, but not of analyses nor of specific means of measuring outcomes (specific scales or methods). There is a risk that outcomes were selected after analysis completed

Overall judged some concerns due to missing outcome data, no information about blinding of outcome assessors and the risk that outcomes were selected after analysis completed.

Westhoff‐Bleck 2013

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

There was no information on randomisation or allocation concealment. There are no baseline differences that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

Overall: 16% of people dropped out. Similar numbers from both groups. Reasons for dropping out for some participants were duration of intervention, but some reasons were unrelated. There was no analysis to assess the effect of missing data 

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

There was a trial register entry.  It provided details of time points, but not of analyses nor of specific means of measuring outcomes (specific scales or methods). There is a risk that analyses were chosen to selectively. 

Overall some concerns due to no information on blinding of outcome assessors, pre published statistical plan and lack of information on randomisation.

Duppen 2015

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized in a 2:1 ratio to the exercise group or control group. Randomization was performed by an independent blinded researcher. Stratification was based on gender, ConHD, and age group (10‐12, 12‐15, 15‐18, and 18‐25 years). There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Most people were followed up >95%

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Fritz 2020

 Some concerns Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

There was no information on method of randomisation, there was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention. 

9.5% dropout rate in this study, however for this type of intervention this is reasonable. Authors accounted for 20% drop out in sample size calculations. 

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Overall due to a lack of detail on randomisation, blinding outcome assessors and lack of information in the pre‐registered methods.

Subgroup 1.6.2 Population with repaired Tetralogy of Fallot

Therrien 2003

 High risk of bias Low risk of bias Low risk of bias Some concerns Some concerns High risk of bias

No information on method of randomisation and significant baseline imbalance between groups. There is a 8 year age gap – the intervention group are younger and age of repair was younger. Right ventricular outflow tract (11 vs 22 mmhg) were half in the intervention group. Daily activity levels were less in the intervention group. This may suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate. 

17 of 18 participants (95%) were included at follow up. One person lost from the control group due to a lack of interest.

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

No registry of protocol available. Common CRF parameter reported. No information whether there were multiple eligible analyses. 

lack of information on randomisation, baseline imbalances and of blinding outcome assessors and no pre‐registered protocol or statistical plan. 

Duppen 2015

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized in a 2:1 ratio to the exercise group or control group. Randomization was performed by an independent blinded researcher. Stratification was based on gender, ConHD, and age group (10‐12, 12‐15, 15‐18, and 18‐25 years). There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Most people were followed up >95%

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement. 

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Avila 2016

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Randomization was performed by sealed envelopes, with a computer‐generated allocation sequence. There were no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

All participants were included in the analysis

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Novakovic 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized into 3 groups (2 interventional and 1 control group) with a 1:1:1 ratio using adaptive (urn) randomisation with sealed envelopes and randomisation concealment from the recruiting investigator. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention.

Only 2 people discontinued intervention and 1 person was lost to follow up, reasons given. 90% of randomised sample accounted for in outcome data

Measurements were similar across all intervention groups but there is no information in whether the outcome assessors were blinded. Outcome assessors, through encouragement during testing, could affect the outcome, therefore we have some concerns about potential bias.

Protocol registered (NCT02643810). Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Novakovic 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Participants were randomized into 3 groups (2 interventional and 1 control group) with a 1:1:1 ratio using adaptive (urn) randomisation with sealed envelopes and randomisation concealment from the recruiting investigator. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention.

Only 2 people discontinued intervention and 1 person was lost to follow up, reasons given. 90% of randomised sample accounted for in outcome data

Measurements were similar across all intervention groups but there is no information in whether the outcome assessors were blinded. Outcome assessors, through encouragement during testing, could affect the outcome, therefore we have some concerns about potential bias.

Protocol registered (NCT02643810). Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Subgroup 1.6.3 Other or mixed ConHD populations

Moalla 2006

 Some concerns Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

There was no information on method of randomisation, there was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

All outcome data was presented

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Overall due to a lack of detail on randomisation, blinding outcome assessors and lack of information in the pre‐registered methods.

Madhavi 2011

 High risk of bias Some concerns High risk of bias High risk of bias Some concerns High risk of bias

Patients were randomly divided into study and control group by simple randomization using the lottery method. However there were substantial differences in patients fitness (as measured by VO2 peak) between the intervention and control group.

Both participants and those delivering the intervention were aware of intervention received. There was no information to support deviations from the intended intervention.

Loss to follow up not explained. All people lost to follow up were in the intervention group. No analysis to check the effect of the missing data.

YMCA cycle test. This is a predictive test not a direct assessment furthermore no description of protocol. Furthermore, there was a lack of blinding outcome assessors and a high effect size seen.

No pre‐registered method (registry or protocol) available. However, industry standard CRF data is available.

Overall high due to differences at baseline, missing outcome data without explanation and a lack of blinding outcome assessors. 

Morrison 2013

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

Participants were randomised using balanced blocks of four to intervention and control groups. Randomisation was not stratified by any factor. There were some baseline imbalances in BMI and Deprivation, which could be of importance in a physical activity promotion trial. The concerns were due to no information on the randomisation process rather than the baseline imbalances.

Both participants and those delivering the intervention were aware of intervention received. There were no deviations from intervention

Twenty‐three participants did not attend reassessment despite being offered three separate appointments, and a number of patients withdrew at this stage. Four patients who were randomised to the intervention group did not attend the group sessions offered. These patients were analysed with the rest of the intervention group on an intention‐to‐treat basis. There was no analysis to assess the effect of missing data.

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Judged some concerns over issues with randomisation, blinding outcome assessors, indirect measurements of CRF, missing outcome data, and no prior published statistical plan.

Klausen 2016

 Low risk of bias Low risk of bias Some concerns Low risk of bias Low risk of bias Some concerns

The randomization was performed by the Copenhagen Trial Unit ten days after the baseline test. A trial investigator centrally randomized eligible patients 1:1. The computer‐generated allocation sequence used permuted blocks with varying sizes of 6, 8, or 10. The allocation sequence and block size were unknown to the trial investigators. There was no baseline imbalance that would suggest a problem with randomisation.

Deviations from the original protocol were approved by the Regional Ethics Committee trial protocol and applied to the ClinicalTrials.gov identifier. Participants  were aware of intervention received. There were no deviations from intervention. The team was blinded to the group allocation of patients and the analysis was appropriate.

39 lost to follow up (24%).Multiple imputations were used to handle missing data. However, multiple imputation based on 'missing at random' does not provide justification to say that there was evidence that results were robust.

The team was blinded to the group allocation of patients.

Protocol registered, with a pre published protocol and statistical plan. Industry standard CRF data were presented.

Some concerns due to 24% loss to follow up.

Opotowsky 2018

 Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns Some concerns

Generation of randomisation sequence and concealment of allocation were adequate. There was no baseline imbalance that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

All participants were included in the analysis

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to no information on blinding of outcome assessors and potential bias in the selection of the reported result.

Sandberg 2018

 Low risk of bias Low risk of bias Low risk of bias Low risk of bias Some concerns Some concerns

Computer based random allocation with allocation and no baseline imbalances that would suggest a problem with randomisation.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate. 

12% of participants were lost to follow up 3/16 intervention and 0/13 control.  There is no analysis to look at the effect of the missing data. A small study so the 3 drop outs looks like a large percentage, reasons documented for patient withdrawal.

“This study was performed at two centers. The center recruiting 5 patients (22%) was not able to keep the investigators performing the exercise tests strictly blinded for group allocation.”

The trial register report was registered before the trial (date 2009) It indicates that CRF would be measured using Vo2 Max. 

Unable to locate a detailed statistical plan

van Dissel 2019

 Some concerns Low risk of bias Some concerns Some concerns Some concerns Some concerns

Computer based random allocation and no baseline imbalances that would suggest a problem with randomisation, but there was no information about allocation concealment.

Both participants and those delivering the intervention were aware of intervention received but there were no deviations from intended interventions and the analysis was appropriate.

Overall: 15% of people dropped out, similar numbers from both groups. Reasons for dropping out for some participants were duration of intervention, but some reasons were unrelated. There was no analysis to assess the effect of missing data 

There was no information on whether outcome assessors were aware of the intervention received. Knowledge of intervention received could have affected outcome measurement.

Protocol registered. Not enough information in the protocol on a statistical plan and no published statistical plan found. Industry standard CRF data were presented.

Due to a lack of detail on randomisation, missing outcome data, blinding outcome assessors and lack of information in the pre‐registered methods.

Figures and Tables -
Risk of bias for analysis 1.6 Maximal cardiorespiratory fitness (type of ConHD subgroup analysis)
Navigate to table in Review