Exercise for the management of cancer‐related fatigue in adults

Fiona Cramp; James Byron-Daniel

doi:10.1002/14651858.CD006145.pub3

. 2012 Nov 14;2012(11):CD006145. doi: 10.1002/14651858.CD006145.pub3

Exercise for the management of cancer‐related fatigue in adults

Fiona Cramp ^1,^✉, James Byron-Daniel ²

Editor: Cochrane Pain, Palliative and Supportive Care Group

PMCID: PMC8480137 PMID: 23152233

Abstract

Background

Cancer‐related fatigue is recognised as an important symptom associated with cancer and its treatment. A number of studies have investigated the effects of physical activity in reducing cancer‐related fatigue. This is an updated version of the original Cochrane review published in The Cochrane Library (2008, Issue 1). The original review identified some benefits of physical activity on fatigue in cancer both during and after adjuvant treatment. We identified a number of limitations in the evidence, providing clear justification for an updated review.

Objectives

To evaluate the effect of exercise on cancer‐related fatigue both during and after cancer treatment.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (Issue 1, 2011), MEDLINE (1966 to March 2011), EMBASE (1980 to March 2011), CINAHL (1982 to March 2011), British Nursing Index (January 1984 to March 2011), AMED (1985 to March 2011), SIGLE (1980 to March 2011) and Dissertation Abstracts International (1861 to March 2011) using key words. We also searched reference lists off all studies identified for inclusion and relevant reviews. In addition, we handsearched relevant journals and contacted experts in the field of cancer‐related fatigue.

Selection criteria

We sought and included randomised controlled trials (RCTs) that investigated the effect of exercise on cancer‐related fatigue in adults.

Data collection and analysis

Two review authors independently assessed the risk of bias of studies and extracted data based upon predefined criteria. Where data were available we performed meta‐analyses for fatigue using a random‐effects model.

Main results

For this update we identified a total of 56 studies (4068 participants) for inclusion (28 from the original search and 28 from the updated search), with the majority carried out in participants with breast cancer (28 studies). A meta‐analysis of all fatigue data, incorporating 38 comparisons, provided data for 1461 participants who received an exercise intervention and 1187 control participants. At the end of the intervention period exercise was seen to be statistically more effective than the control intervention (standardised mean difference (SMD) ‐0.27, 95% confidence interval (CI) ‐0.37 to ‐0.17). Benefits of exercise on fatigue were observed for interventions delivered during or post‐adjuvant cancer therapy. In relation to diagnosis, we identified benefits of exercise on fatigue for breast and prostate cancer but not for those with haematological malignancies. Finally, aerobic exercise significantly reduced fatigue but resistance training and alternative forms of exercise failed to reach significance.

Authors' conclusions

The findings of the updated review have enabled a more precise conclusion to be made in that aerobic exercise can be regarded as beneficial for individuals with cancer‐related fatigue during and post‐cancer therapy, specifically those with solid tumours. Further research is required to determine the optimal type, intensity and timing of an exercise intervention.

Plain language summary

The effect of exercise on fatigue associated with cancer

Fatigue, or tiredness, is recognised as a side effect of cancer and its treatment. In the past people with cancer were encouraged to rest if they felt fatigued. It is important that individuals with cancer receive appropriate support and advice to help them cope with any side effects of the treatment or disease. Physical exercise has been suggested as helpful in reducing the fatigue that is associated with cancer. A number of studies have been carried out to investigate the effects of exercise both during and after treatment. The current review was carried out to evaluate the effect of physical exercise on fatigue related to cancer. Fifty‐six studies, involving a total of 4068 participants, were included in this review. Results suggest that physical exercise such as aerobic walking and aerobic cycling can help to reduce fatigue both during and after treatment for cancer. The benefits of exercise on fatigue were observed specifically for people with breast cancer and prostate cancer.

Background

This review is an update of a review previously published in The Cochrane Library (2008, Issue 1) on exercise for the management of cancer‐related fatigue (Cramp 2008). As a result of improved therapy, people with cancer are surviving longer and having to deal with the long‐term consequences of the disease and its therapy. Consequently there has been an increasing number of individuals who need supportive care to enhance their quality of life (Lucia 2003). This has led to an increasing recognition of the symptoms associated with cancer and cancer management with relief of these symptoms emerging as an important dimension of cancer patient care.

Description of the condition

Cancer‐related fatigue is an abstract, multidimensional subjective experience, affecting 70% to 100% of the cancer patient population (Mock 2001b). It has a profound effect on the whole person, physically, emotionally and mentally (Ahlberg 2003), and can persist for months or even years following completion of treatment. It can have a phenomenal impact on a patient's life, interfering with daily activities (Curt 2000) and may also potentially have devastating social and economic consequences (Fletchner 2002). It can hinder a patient's chance of remission or even cure, owing to the effect it can have on the desire to continue with treatment (Morrow 2001).

In spite of the prevalence and impact of cancer‐related fatigue there are limited data available with regards to the precise aetiology, pattern over time and exacerbating and relieving factors (Fletchner 2002), thus complicating the development of effective management interventions (Dimeo 2002). The aetiology of cancer‐related fatigue remains to be fully established and a number of causes have been suggested, such as the effect of tumour and cancer treatment, comorbid medical conditions including anaemia, hypothyroidism, cytokines and sleep problems, psychological factors such as anxiety and depression, and loss of functional status (Lucia 2003; Wagner 2004; Mustian 2007). The cause of cancer‐related fatigue may also differ between individuals as well as according to the phase of the disease and the type of treatment received (Ryan 2007).

Description of the intervention

The National Comprehensive Cancer Network (NCCN 2012) has developed guidelines for the management of cancer‐related fatigue. Initially any treatable factors that may cause fatigue should be identified and treated. The panel identified several factors that are treatable namely: pain, emotional distress, sleep disturbance, anaemia, nutrition, activity level, medication side effects, alcohol/substance abuse and comorbidities. If the patient does not have any treatable contributing factors or cancer‐related fatigue persists, then additional treatment is recommended depending on the patient's clinical status. This incorporates education and counselling, general strategies for the management of fatigue, and pharmacological and non‐pharmacological interventions. In line with these guidelines the role of non‐pharmacological interventions in the management of cancer‐related fatigue is supported by Mustian 2007 and colleagues who have identified psychosocial therapies, physical exercise and a range of other interventions as potentially beneficial. Activity enhancement is also recommended by the NCCN as one of the non‐pharmacological interventions at all stages of the disease process: during active treatment, in disease‐free patients on long‐term follow‐up and at the end of life (NCCN 2012).

The physical dimension of cancer‐related fatigue is likely to have an organic aetiology (Dimeo 2001). The effect of treatment and a reduction in physical activity can lead to a reduction in physical performance (NCCN 2012). Thus, the patient requires an increased effort to accomplish normal everyday activities, which leads to the perception of fatigue. This is further exacerbated by impairment of skeletal muscle function intensified by a lack of activity (Lucia 2003).

How the intervention might work

Exercise has been demonstrated to be effective in reducing fatigue and improving the exercise tolerance of healthy individuals as well as those with long‐term conditions (Mock 2005). It has been suggested that changes brought about by physical activity may counteract the negative effects the tumour and toxic therapy have on the capacity for physical performance (Dimeo 2002). Activity could reduce cancer‐related fatigue by improving functional capacity, resulting in a reduced effort and improvement in the perception of fatigue (NCCN 2012). Rest, the preferred recommendation for cancer‐related fatigue in the past, is likely to be counterproductive as inactivity leads to muscle wasting and a loss of cardiorespiratory fitness, leading to increased fatigue (Dimeo 2001). Winningham 1992 developed a theoretical framework for cancer‐related fatigue (Winningham's Psychobiological‐Entropy model) which proposes that a balance between rest and activity can reduce fatigue, whereas an imbalance can lead to deterioration suggesting that too little or too much exercise may exacerbate it. More recently a biobehavioural model has been suggested to explain the benefits that exercise may have upon cancer‐related fatigue (Al‐Majid 2009). The model encompasses biological, psycho‐behavioral and functional variables that are implicated in the induction of cancer‐related fatigue.

The benefit of exercise may not be limited to the improvement of the physical dimension of fatigue. It could also relieve the emotional and mental dimensions. Exercise has been demonstrated to improve mood and reduce anxiety and fear in patients (Dimeo 2001).

Why it is important to do this review

There are two complementary Cochrane systematic reviews focusing on the treatment of cancer‐related fatigue at all stages of cancer. These complementary reviews have examined the role of drugs (Minton 2010) and psychosocial interventions (Goedendorp 2009) in modifying cancer‐related fatigue. However, there is currently no definitive stand on exercise for cancer‐related fatigue from the American College of Sports Medicine, with minimal detail given on exercise prescription in those that do recommend it (Lucia 2003; Mustian 2007). Furthermore there is limited information on specific groups of cancer patients where caution may be required (Mustian 2007). This review update was conducted to analyse the increasing number of trials and where possible to identify the optimal exercise for reducing cancer‐related fatigue.

Objectives

To evaluate the effect of exercise on cancer‐related fatigue both during and after cancer treatment.

A secondary objective, subject to available data, was to explore the effect of exercise in different types of cancer populations. Groupings were determined based on tumour type, type of cancer treatment received and stage of cancer treatment, that is, either during or after treatment. Based upon the findings of the previous review it was also proposed to explore the effects of different modes of exercise.

Methods

Criteria for considering studies for this review

Types of studies

We considered only randomised controlled trials (RCT) for inclusion.

Types of participants

We included studies that evaluated the effect of exercise on cancer‐related fatigue in adults of any age, regardless of gender, tumour type, tumour stage and type of cancer treatment. Participants may have been actively receiving treatment, be in long‐term follow‐up or receiving palliative care.

Types of interventions

Included studies needed to evaluate and report the effect of physical exercise on cancer‐related fatigue. The studies should compare exercise with no exercise, a usual care group (i.e. no specific exercise programme prescribed) or an alternative treatment or exercise regime for fatigue associated with cancer. The intervention could take place in any setting and be delivered to a group or individual participant. All types of physical exercise were considered for inclusion, including aerobic exercise, strength training and flexibility exercises. Studies that investigated an exercise programme accompanied by attempts to promote participant engagement were included. In contrast, studies that explored multi‐dimensional programmes in which the effects of exercise alone could not be determined were excluded. Studies were also excluded if a specific exercise programme was not described and participants were only given advice or education about the potential benefits of exercise.

Types of outcome measures

The outcomes of interest were:

patient‐reported fatigue measured using reliable and valid assessment tools;
exercise maintenance on follow‐up;
time spent exercising;
valid and reliable measures of aerobic capacity;
quality of life measures;
anxiety;
depression;
self efficacy (the individual's belief in their own ability to be physically active).

Search methods for identification of studies

Electronic searches

Please see Appendix 1 for the search strategy.

We used the following databases to obtain relevant studies for this review (the original search was conducted up to July 2007):

The Cochrane Central Register of Controlled Trials (CENTRAL) (Issue 1, 2011);
MEDLINE (1966 to March 2011);
EMBASE (1980 to March 2011);
CINAHL (1982 to March 2011);
British Nursing Index (January 1984 to March 2011);
AMED (1985 to March 2011);
SIGLE (1980 to March 2011);
Dissertation Abstracts International (1861 to March 2011).

Searching other resources

We checked the reference lists of all articles obtained for additional studies.
We handsearched the following journals up to April 2011: Cancer, Journal of Clinical Oncology, Psycho‐Oncology, Cancer Practice, Oncology Nursing Forum.
We contacted four experts in the field of cancer‐related fatigue in order to identify any research that may not have been published.
We obtained unpublished literature through searches of conference proceedings up to June 2011.
We attempted to communicate with the study authors to secure information not presented in the studies.
There were no language restrictions.

Data collection and analysis

We retrieved all studies in which the abstract made reference to an exercise trial in a population of cancer participants in full. Where abstracts were not available and the study could not be excluded based upon the title alone we retrieved the full text. For a study to be included it had to include fatigue as an outcome measure and at least one treatment arm had to be exercise. Two independent review authors screened all the retrieved full‐text articles for inclusion criteria. Although there was initially some discrepancy based upon the appropriateness of the outcome measures, following discussion there was 100% agreement.

Two review authors extracted data from the included studies. If there was disagreement we proposed that there would be a meeting with a third independent reviewer to reach a consensus.

For the original review we assessed the methodological quality of each study using the Oxford quality scale, which is concerned with the presence and adequacy of randomisation, the presence and adequacy of blinding, and description of participant withdrawals (Jadad 1996).

In addition, for each study we extracted:

number of participants in each arm;
type of control group;
demographic characteristics, including age and gender;
type of cancer;
type of treatment and stage of treatment, for example during or after treatment;
the exercise undertaken in each intervention group;
duration of intervention, intensity and total number of exercise sessions;
duration of follow‐up;
outcome measures employed, including means and standard deviations;
attrition rates.

Assessment of risk of bias in included studies

For the update of the review we adapted the methods from those described by the Cochrane Pregnancy and Childbirth Group 2012. The authors independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) with any disagreements resolved by discussion. We assessed the following for each study:

Random sequence generation (checking for possible selection bias). We assessed the method used to generate the allocation sequence as: low risk of bias (any truly random process, e.g. random number table, computer random number generator); high risk of bias (any non‐random process, e.g. odd or even date of birth, hospital or clinic record number); unclear risk of bias.
Allocation concealment (checking for possible selection bias). We assessed the method used to conceal allocation to interventions prior to assignment for whether intervention allocation could have been foreseen in advance of, or during, recruitment, or changed after assignment. We assessed the methods as: low risk of bias (e.g. telephone or central randomisation, consecutively numbered sealed opaque envelopes); high risk of bias (open random allocation, unsealed or non‐opaque envelopes, alternation, date of birth); unclear risk of bias.
Blinding of outcome assessment (checking for possible detection bias). We assessed the methods used to blind study participants and outcome assessors from knowledge of which intervention a participant received. We considered studies to be at low risk of bias if they were blinded, or if we judged that the lack of blinding would be unlikely to affect the results. We assessed blinding specifically in relation to the fatigue outcome.
Incomplete outcome data (checking for possible attrition bias due to the amount, nature and handling of incomplete outcome data). We assessed the completeness of fatigue outcome data including attrition and exclusions from the analysis. We assessed methods as: low risk of bias (e.g. no missing outcome data, missing outcome data balanced across groups); high risk of bias (e.g. numbers or reasons for missing data imbalanced across groups, intention‐to‐treat analysis not performed); unclear risk of bias.
Selective reporting (checking for reporting bias). We assessed the methods as: low risk of bias (where it is clear that all of the study’s pre‐specified fatigue outcomes have been reported); high risk of bias (where not all the study’s pre‐specified fatigue outcomes have been reported); unclear risk of bias.

Measures of treatment effect

Fatigue outcomes were likely to be reported in different ways in the identified studies. It was therefore difficult to predict what data would be available to be combined. If the data were available, and it was appropriate to do so, we proposed that the studies would be combined in a meta‐analysis. We proposed to calculate the mean difference in fatigue intensity between exercise and control groups including usual care and alternative treatment groups. Subgroup analysis would also be conducted if the data were available. Separate analysis would be implemented according to tumour type, for example breast cancer participants; treatment received, for example chemotherapy or radiotherapy; and the stage of treatment the participant was at when the exercise programme was administered, that is either during or after cancer treatment. If heterogeneity between studies was suspected, the possibility of utilising a random‐effects model of meta‐analysis would be considered. Further to this and if the information was available, we proposed to calculate the number of participants who experienced at least a 50% reduction in fatigue. This would be used to calculate the number needed to treat to benefit (NNTB) for a 50% reduction in fatigue.

Assessment of heterogeneity

Where appropriate, heterogeneity of the data was formally assessed using the I² statistic (Higgins 2003). A value greater than 50% may represent substantial heterogeneity.

Results

Description of studies

In the original review, following a comprehensive literature search including screening of titles and abstracts (where available), we retrieved 51 full‐text references. From these, we excluded 23 publications and identified 28 (2083 participants) as appropriate for inclusion in the current review. In the updated search we retrieved a further 58 full‐text references following screening of titles and abstracts (where available). From these publications, we excluded 23, a further seven were found to be linked to previous studies and we identified 28 (n = 1985 participants) as appropriate for inclusion in the updated review. The 28 studies from the original review and the 28 new studies provided a total of 56 studies (4068 participants) for inclusion in the review. In addition, through correspondence with study authors, we identified two protocols for appropriate studies; both studies were still in progress and not yet published and are therefore included in the Characteristics of ongoing studies table.

Included studies

The final selection, based on consensus, resulted in 28 studies being identified for inclusion in the original review and a further 28 identified in the update providing 56 studies for inclusion. Trial characteristics and outcomes can be seen in the Characteristics of included studies table. Five of the included studies incorporated two separate exercise groups and are therefore entered twice for the purposes of statistical analysis.

Participants

Participants had various cancer diagnoses although the majority of studies investigated breast cancer only (Mock 1994; Mock 1997; Segal 2001a; Galantino 2003; Courneya 2003b; McKenzie 2003; Pinto 2003; Headley 2004; Campbell 2005; Drouin 2005; Mock 2005; Pinto 2005; Courneya 2007a; Daley 2007; Heim 2007; Moadel 2007 Mutrie 2007; Yuen 2007a; Battaglini 2008; Hwang 2008; Milne 2008; Payne 2008; Carson 2009; Danhauer 2009; Rogers 2009; Chandwani 2010; Lee 2010; Sequeira 2012). Thirty‐eight studies investigated participants with a specific cancer diagnosis, whereas 18 studies included participants with different diagnoses. The time since diagnosis varied widely between studies and in some cases within studies. Stage of treatment also varied between the included studies, with 25 of the studies investigating participants during cancer treatment (Mock 1994; Mock 1997; Dimeo 1999; Segal 2001a; Coleman 2003a; Segal 2003; Headley 2004; Windsor 2004; Campbell 2005; Drouin 2005; Mock 2005; Courneya 2007a; Monga 2007; Mutrie 2007; Battaglini 2008; Chang 2008; Hwang 2008; Payne 2008; Adamsen 2009; Mustian 2009; Rogers 2009; Segal 2009a; Chandwani 2010; Culos‐Reed 2010; Dodd 2010a), 18 following treatment (Burnham 2002; Courneya 2003b; Galantino 2003; McKenzie 2003; Pinto 2003; Dimeo 2004; Pinto 2005; Thorsen 2005; Culos‐Reed 2006; Daley 2007; Heim 2007; McNeely 2008; Milne 2008; Carson 2009; Shelton 2009; Lee 2010; van Weert 2010; Sequeira 2012) and the remaining 13 studies including participants both during and post‐cancer treatment (Courneya 2003a; Courneya 2003c; Cohen 2004; Brown 2006; Moadel 2007; Yuen 2007a; Courneya 2008; Oh 2008; Courneya 2009; Danhauer 2009; Galvão 2010; Oh 2010; Santa Mina 2012).

The mean age of participants ranged from 39 to 70 years, with the majority of studies reporting a mean age that fell within the fifth decade. Twenty‐nine of the studies recruited females only, which is perhaps not surprising given that 28 of the studies investigated breast cancer only. Twenty studies included a mixed sample of males and females with the remaining seven studies recruiting males only. For detailed information on study participants see the Characteristics of included studies table.

Interventions

Mode, intensity and timing of exercise differed across studies. Nineteen studies investigated home‐based/unsupervised exercise programmes (Mock 1994; Mock 1997; Segal 2001a; Coleman 2003a; Courneya 2003a; Courneya 2003c; Galantino 2003; Headley 2004; Windsor 2004; Drouin 2005; Mock 2005; Pinto 2005; Thorsen 2005; Heim 2007; Yuen 2007a; Payne 2008; Mustian 2009; Culos‐Reed 2010; Dodd 2010a), whereas 37 studies investigated supervised, institution‐based exercise programmes (Dimeo 1999; Segal 2001b; Burnham 2002; Courneya 2003b; McKenzie 2003; Pinto 2003; Segal 2003; Cohen 2004; Dimeo 2004; Campbell 2005; Brown 2006; Culos‐Reed 2006; Courneya 2007a; Daley 2007; Moadel 2007; Monga 2007; Mutrie 2007; Battaglini 2008; Chang 2008; Courneya 2008; Hwang 2008; McNeely 2008; Milne 2008; Oh 2008; Adamsen 2009; Carson 2009; Courneya 2009; Danhauer 2009; Rogers 2009; Segal 2009a; Chandwani 2010; Galvão 2010; Lee 2010; Oh 2010; van Weert 2010; Santa Mina 2012; Sequeira 2012). However, some studies investigating supervised exercise programmes encouraged participants to undertake additional home‐based exercise. The remaining study compared a supervised and unsupervised exercise programme (Shelton 2009).

The mode of aerobic exercise varied between studies with 14 studies prescribing a walking programme (Mock 1994; Mock 1997; Segal 2001a; Galantino 2003; Windsor 2004; Drouin 2005; Mock 2005; Heim 2007; Monga 2007; Chang 2008; Payne 2008; Mustian 2009; Rogers 2009; Culos‐Reed 2010), six prescribing stationary cycling (leg: Courneya 2003b; Dimeo 2004; Courneya 2008; Courneya 2009; arm: Dimeo 1999; McKenzie 2003) and a further 23 studies incorporating a range of modalities or allowing the participant to choose their preferred mode of aerobic exercise. One study did not report the mode of exercise carried out by participants (Sequeira 2012). Thirteen studies incorporated resistance training as a component of the exercise programme (Coleman 2003a; McKenzie 2003; Heim 2007; Battaglini 2008; Hwang 2008; Milne 2008; Adamsen 2009; Mustian 2009; Shelton 2009; Culos‐Reed 2010; Galvão 2010; van Weert 2010; Santa Mina 2012) and four studies investigated resistance training in isolation (Segal 2003; Courneya 2007a; McNeely 2008; Lee 2010). Further to this, two studies included an aerobic training arm in addition to a resistance training arm (Yuen 2007a; Segal 2009a). Seven studies included flexibility training as a component of the exercise programme (Courneya 2003a; Courneya 2003c; Heim 2007; Battaglini 2008; Hwang 2008; Culos‐Reed 2010; Lee 2010) although a number of studies incorporated routine stretching as part of the warm‐up or cool‐down, or both. Yoga was investigated in six studies (Cohen 2004; Culos‐Reed 2006; Moadel 2007; Carson 2009; Danhauer 2009; Chandwani 2010), qigong in two studies (Oh 2008; Oh 2010) and seated exercise in one study (Headley 2004). The intensity of exercise varied greatly across studies with comparison complicated by the method used to monitor intensity in each study. This included methods such as heart rate monitoring, predicted oxygen uptake, patient perceived effort using the Borg scale and self paced exercise intensity. The frequency and duration of the exercise sessions also varied greatly between studies with participants encouraged to exercise between two times per week and daily for 10 to 120 minutes per session. In some studies the frequency or duration, or both, was increased incrementally each week whereas in other studies the duration was based upon individual tolerance. The overall time spent exercising for each of the groups was rarely reported.

In the majority of studies (k = 34) the comparison arm was described as a 'no intervention' or 'usual care' control group although in three of the 34 studies the participants received a weekly phone call regarding their symptom experience (Windsor 2004; Pinto 2005; Dodd 2010a). In a further 10 cases the comparison arm was a 'wait list' control (that is, the control group participants remained on a waiting list and were offered the intervention once the study was complete). Alternative comparison groups included relaxation training (Dimeo 2004), general stretching or range of movement exercises (Drouin 2005; Hwang 2008), light weights and stretching (McNeely 2008), Tai Chi (Galantino 2003) and group psychotherapy (Courneya 2003c). In two studies the participants in the control group received usual care but were also encouraged to remain active (Segal 2001a; Coleman 2003a) and in a further study the control arm patients were given written materials relating to general physical activity (Rogers 2009). In addition to the 'usual care' comparison group Daley 2007 also included a placebo group in which participants performed light intensity body conditioning exercises. No details were provided relating to the comparison group in the study by Burnham 2002. In two studies both arms received an exercise intervention, the first of these compared physical activity with physical activity delivered alongside a cognitive behavioural therapy to reduce fatigue (van Weert 2010); the second compared a group‐based intervention to personal training (Santa Mina 2012).

The intervention period varied greatly between studies with a range from three weeks (Dimeo 2004) to one year (Dodd 2010a) and a mode of 12 weeks (k = 17 studies). In some studies duration of the intervention varied between participants, being based upon the duration of cancer treatment. Stratification during randomisation was carried out in the majority of studies in attempt to account for this variation.

For detailed information on interventions see the Characteristics of included studies table.

Outcome measures

Fatigue was assessed using a wide range of outcome measures including the Functional Assessment of Cancer Therapy (FACT) (Courneya 2003a; Courneya 2003b; Courneya 2003c; Segal 2003; Courneya 2007a; Mutrie 2007; Courneya 2008; McNeely 2008; Courneya 2009; Danhauer 2009; Rogers 2009; Segal 2009a; Oh 2010; Santa Mina 2012), a linear analogue self assessment scale (Mock 1994; Mock 1997; Burnham 2002; Pinto 2005; Brown 2006), a numerical rating scale (Carson 2009), the fatigue sub scale of the Profile of Mood States (POMS) (Dimeo 1999; Coleman 2003a; Pinto 2003; Brown 2006; Culos‐Reed 2006), the Piper Fatigue Scale (Mock 1997; Campbell 2005; Drouin 2005; Mock 2005; Daley 2007; Monga 2007; Yuen 2007a; Battaglini 2008; Payne 2008; Dodd 2010a), the Brief Fatigue Inventory (BFI) (Galantino 2003; Cohen 2004; Windsor 2004; Chang 2008; Hwang 2008; Mustian 2009; Shelton 2009; Chandwani 2010), the EORTC‐QLQ‐C30 fatigue sub scale (Dimeo 2004; Thorsen 2005; Culos‐Reed 2006; Oh 2008; Adamsen 2009; Galvão 2010; Lee 2010; Sequeira 2012), the SF‐36 vitality scale (Segal 2001a; McKenzie 2003; Adamsen 2009; Galvão 2010), the Functional Assessment of Chronic Illness Therapy‐fatigue scale (Headley 2004; Moadel 2007; Mustian 2009), the Fatigue Severity Scale (FSS) (Culos‐Reed 2010), the Multidimensional Fatigue Inventory (MFI) (Heim 2007; van Weert 2010) and the Schwartz Cancer Fatigue Scale (Milne 2008). Six studies incorporated more than one fatigue outcome measure (Mock 1997; Brown 2006; Culos‐Reed 2006; Adamsen 2009; Mustian 2009; Galvão 2010).

In addition to measuring fatigue the following outcomes were recorded the most frequently: aerobic capacity/cardiovascular function (k = 36), quality of life (k = 38), body composition (k = 18), physical activity levels (k = 15), general mood (k = 11), depression (k = 20) and anxiety (k = 13). Other outcomes assessed included flexibility, sleep variables, distress, symptom stress, strength, well‐being, happiness, self esteem, satisfaction with life, chemotherapy completion rates, lymphoedema, self perception, psychosocial adjustment, body image and motivational readiness for change.

For detailed information on outcomes measures see the Characteristics of included studies table.

Excluded studies

The 56 publications retrieved and subsequently excluded did not meet the review inclusion criteria for the following reasons: two were reviews, 25 were not randomised controlled clinical trials, 10 did not report fatigue as an outcome measure, eight did not include an appropriate exercise intervention and one study did not report the results based upon the original group allocation. Details of the excluded studies can be found in the Characteristics of excluded studies table.

Risk of bias in included studies

We initially assessed the included studies for quality using the Oxford Quality Score (Jadad 1996) and in the updated review we also assessed studies using the Cochrane 'Risk of bias' tool (Higgins 2011). Following discussion there was 100% agreement in scores between the two review authors. The majority of studies scored two or three on the Oxford Quality Score. All studies lost two points due to the inability to conceal group allocation of study participants to the exercise intervention. It was also noted that observer blinding was rarely reported in the included studies. Studies that scored two lost a further point due to an incomplete description of drop‐outs or an inadequate description of the method of randomisation. In general the Oxford Quality Score was higher in studies that were published more recently. However, it should also be noted that only five of the 56 studies were published prior to 2003. Oxford Quality Scores and 'Risk of bias' tables for each study are presented in the Characteristics of included studies table and a summary of the risk of bias is presented in Figure 1 and Figure 2.

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

'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Approximately half of the included studies reported appropriate random sequence generation. Allocation concealment was less well reported with over half the studies judged to be unclear.

Blinding

Unsurprisingly blinding of the fatigue outcome assessment was not performed adequately in any study. This is due to the self report nature of a subjective outcome such as fatigue combined with the complexity of blinding participants to a physical exercise intervention.

Incomplete outcome data

The majority of studies accounted for all participants and where appropriate performed intention‐to‐treat analysis.

Selective reporting

Fatigue outcomes as reported in the methods for each study were provided for the majority of the included studies.

Other potential sources of bias

A number of the included studies were limited by a small sample size with 24 studies recruiting fewer that 40 participants. Sample size ranged from 10 to 242 with a mean sample size of 69 participants and a median of 52.5. Thirty of the 56 studies either did not carry out or did not report a sample size calculation. The remaining 26 studies carried out a sample size calculation although only 10 of these recruited the required number of participants. Nine studies used fatigue as the basis of the sample size calculation with only four of the nine achieving their recruitment target.

Sixteen of the identified studies either provided no information regarding adherence to the prescribed intervention or did not report it in a meaningful way. The remaining 40 studies provided data on adherence to the exercise intervention with some of the studies also reporting exercise levels for the comparison group. Methods for reporting adherence varied between studies, for example, some studies reported the percentage of supervised sessions attended whereas others were based upon subjective reporting of exercise. Comparison between studies was complicated by the different methods used for reporting, however, for the studies that reported the percentage of supervised sessions attended adherence varied between 61% (Milne 2008) and 98.4% (Courneya 2003b). A few of the studies monitored activity in the comparison group and acknowledged that the level of exercise may have resulted in some participants in the control group carrying out similar levels of activity to individuals in the intervention group. Three studies compared time spent exercising between the intervention and comparison group; all three reported that the intervention group performed exercise for a statistically longer period during the intervention period (Courneya 2003c; Pinto 2005; Mutrie 2007). For analysis purposes all studies considered participants in the group to which they had been assigned, regardless of adherence, hence the estimated benefit of exercise did not take into consideration whether or not the participants adhered to the prescribed activity.

Only 11 of the 56 studies failed to provide adequate information regarding participants who withdrew or dropped out of the study. In the majority of cases participants who dropped out or withdrew were not included in the analysis although this was not entirely clear in all studies. Definition of intention‐to‐treat analysis varied between studies with some studies using this term to describe the inclusion of non‐adherent participants in the analysis, whereas others used the term to describe the inclusion of participants who failed to complete the outcome measures or lost contact with the investigators, or both.

Follow‐up assessment of long‐term outcomes was poor with 35 of the 56 studies failing to assess outcomes beyond the end of the intervention period. From the remaining studies that did include a follow‐up assessment, four did not present the follow‐up data in the original publication (Headley 2004; Pinto 2005; Courneya 2007a; Culos‐Reed 2010). The length of follow‐up for the remaining studies varied from four weeks (Windsor 2004) to one year (Dodd 2010a) following the end of the intervention period.

The primary outcome was not identified in all studies, with the majority of studies employing multiple outcome measures. Fatigue reduction alone was the main purpose of the intervention in only eight studies (Dimeo 1999; Dimeo 2004; Mock 2005; Heim 2007; Yuen 2007a; Chang 2008; Adamsen 2009; Dodd 2010a). In the remaining studies fatigue was either one of two or more main outcomes, a secondary outcome or reported as a sub scale of a quality of life measure. In 11 of the studies improvement in quality of life was identified as the single main purpose of the intervention (Courneya 2003a; Courneya 2003c; Campbell 2005; Brown 2006; Courneya 2007a; Daley 2007; Moadel 2007; Mutrie 2007; Milne 2008; Chandwani 2010; Oh 2010). Thirteen studies identified two primary outcomes (Mock 1994; Mock 1997; Segal 2001a; Coleman 2003a; Courneya 2003b; Galantino 2003; McKenzie 2003; Segal 2003; Headley 2004; Windsor 2004; Monga 2007; McNeely 2008; Santa Mina 2012) and 13 studies identified three or more outcomes without identifying the primary purpose of the intervention (Pinto 2003; Drouin 2005; Pinto 2005; Thorsen 2005; Hwang 2008; Oh 2008; Payne 2008; Danhauer 2009; Mustian 2009; Segal 2009a; Culos‐Reed 2010; Galvão 2010; van Weert 2010). The tools used to assess each of the outcomes of interest varied greatly between studies making comparisons difficult. Some of the included studies did not report the results in full and often only where statistically significant differences were identified.

Effects of interventions

Twenty‐eight studies were identified for inclusion in the original review and an additional 28 in the updated review, providing 56 studies for inclusion (n = 4068 participants). For the purposes of meta‐analysis, where data were not reported in full, we made an attempt to contact the corresponding author (n = 30). Twelve of the 30 authors subsequently provided post‐test mean data +/‐ standard deviation (SD) for the fatigue outcome employed. The remaining authors either did not respond, could not be located or reported that the data were not available. In the original review we carried out separate meta‐analyses on fatigue data using post‐test means and change data. The two comparisons provided very similar results and we subsequently decided to base the remaining comparisons on the post‐test means as the data were more readily available. For the purposes of the updated review all comparisons were based upon post‐test means.

Five studies were not appropriate for inclusion in the meta‐analyses as they did not incorporate an acceptable control arm. The five studies excluded compared specific exercise with a general exercise control arm (Lee 2010), individually supervised versus a group exercise control (Santa Mina 2012), supervised exercise versus an unsupervised control (Shelton 2009), resistance exercise versus aerobic exercise (van Weert 2010) and finally exercise versus exercise and education (Sequeira 2012).

Fatigue

All studies: exercise versus control, post‐test means

We used a meta‐analysis to combine the post‐test results of the 51 remaining studies providing 56 possible comparisons due to the inclusion of two intervention arms in five studies. However, post‐test means +/‐ SD were not available for 18 of the 56 comparisons. The remaining 38 comparisons provided data for 1461 participants who received an exercise intervention and 1187 participants in the control arm. At the end of the intervention period exercise was statistically more effective than the control intervention (standardised mean difference (SMD) ‐0.27, 95% confidence interval (CI) ‐0.37 to ‐0.17) with moderate statistical heterogeneity identified (P = 0.03; I² = 33.0%) (Analysis 1.1).

1.1 — Comparison 1: Fatigue: all data, Outcome 1: Exercise versus no exercise control; post‐test means

Intervention timing

Twenty‐five studies provided separate data for participants receiving an exercise intervention during cancer treatment. We used a meta‐analysis to combine the post‐test results of the 25 studies with 28 comparisons possible due to the inclusion of two intervention groups in three studies. However, post‐test means +/‐ SD were not available for 10 of the 28 comparisons. The remaining 18 comparisons provided data for 824 participants who received an exercise intervention and 642 participants in the control arm. At the end of the intervention period exercise was statistically more effective than the control intervention (SMD ‐0.23, 95% CI ‐0.33 to ‐0.12) with no evidence of statistical heterogeneity (P = 0.94; I² = 0%) (Analysis 2.1).

2.1 — Comparison 2: Fatigue: during anti‐cancer therapy, Outcome 1: Exercise versus no exercise control; post‐test means

Nineteen studies investigated participants post‐cancer treatment, however, four of these studies were not appropriate for inclusion in a meta‐analysis due to the lack of an appropriate control arm (Shelton 2009; Lee 2010; van Weert 2010; Sequeira 2012). We also used a meta‐analysis to combine the post‐test results of the remaining 15 studies in which participants received an exercise intervention following cancer treatment with 15 comparisons possible. However, post‐test means +/‐ SD were not available for five of the 15 comparisons. The remaining 10 comparisons provided data for 272 participants who received an exercise intervention and 267 participants in the control arm. At the end of the intervention period exercise was statistically more effective than the control intervention (SMD ‐0.44, 95% CI ‐0.79 to ‐0.09) with a high level of statistical heterogeneity detected (P < 0.0002; I² = 72.0%) (Analysis 3.1).

3.1 — Comparison 3: Fatigue: post anti‐cancer therapy, Outcome 1: Exercise versus no exercise control

The remaining studies recruited a mixed sample of participants during and post‐treatment for cancer and did not provide separate results.

Disease‐specific comparisons

The majority of studies were carried out in breast cancer populations (k = 28 studies that only recruited breast cancer participants; n = 1671 participants). Two of these studies were not appropriate for inclusion in a meta‐analysis due to the lack of an appropriate control arm (Lee 2010; Sequeira 2012). We used a meta‐analysis to combine the post‐test results of the remaining 26 studies, with 29 comparisons possible due to the inclusion of two intervention groups in three studies. However, post‐test means +/‐ SD were not available for 11 of the 29 comparisons. The remaining 18 comparisons provided data for 672 participants who received an exercise intervention and 511 participants in the control arm. At the end of the intervention period exercise was statistically more effective than the control intervention (SMD ‐0.35, 95% CI ‐0.51 to ‐0.19) with a moderate level of statistical heterogeneity detected (P = 0.06; I² = 36.0%) (Analysis 4.1).

4.1 — Comparison 4: Fatigue: breast cancer, Outcome 1: Exercise versus no intervention control; post‐test means

Seven studies were carried out in prostate cancer populations with an overall total of 491 participants. One study was not appropriate for inclusion in the meta‐analysis due to the lack of an appropriate control arm (Santa Mina 2012). We used a meta‐analysis to combine the post‐test results of the remaining six studies, with seven comparisons possible due to the inclusion of two intervention groups in one study. However, post‐test means +/‐ SD were not available for one of the seven comparisons. The remaining six comparisons provided data for 239 participants who received an exercise intervention and 176 participants in the control arm. At the end of the intervention period exercise was statistically more effective than the control intervention (SMD ‐0.45, 95% CI ‐0.78 to ‐0.11) although a high level of statistical heterogeneity was detected (P = 0.03; I² = 59.0%) (Analysis 5.1).

5.1 — Comparison 5: Fatigue: prostate cancer, Outcome 1: Exercise versus no exercise control; post‐test means

Five studies were carried out in populations with haematological malignancies (lymphoma, myeloma and leukaemia) with an overall total of 260 participants. One study was not appropriate for inclusion in the meta‐analysis due to the lack of an appropriate control arm (Shelton 2009). We used a meta‐analysis to combine the post‐test results of the remaining four studies providing four comparisons. The comparisons provided data for 114 participants who received an exercise intervention and 106 participants in the control arm. At the end of the intervention period there was no statistically significant difference between the exercise and the control arm (SMD ‐0.15, 95% CI ‐0.42 to 0.11) with no evidence of statistical heterogeneity detected (P = 0.91; I² = 0%) (Analysis 6.1).

6.1 — Comparison 6: Fatigue: haematological malignancies, Outcome 1: Exercise versus no exercise control; post‐test means

Two further studies recruited disease‐specific cohorts, specifically a colorectal cancer population (Courneya 2003a) and a head and neck cancer population (McNeely 2008). The remaining studies (k = 14) recruited participants from more than one diagnostic group.

Mode of exercise

The majority of studies incorporated an aerobic exercise arm (k = 30), although one of the studies was not appropriate for inclusion in a meta‐analysis due to the lack of an appropriate control arm (van Weert 2010). The remaining 29 studies provided 31 possible comparisons due to the inclusion of two intervention groups in two studies. However, post‐test means +/‐ SD were not available for nine of the 31 comparisons. The remaining 22 comparisons provided data for 832 participants who received an aerobic exercise intervention and 701 participants in the control arm. At the end of the intervention period aerobic exercise was statistically more effective than the control intervention (SMD ‐0.22, 95% CI ‐0.34 to ‐0.10) with a moderate level of statistical heterogeneity detected (P = 0.20; I² = 20.0%) (Analysis 7.1).

7.1 — Comparison 7: Fatigue: aerobic training, Outcome 1: Exercise versus no exercise control; post‐test means

Six studies included a resistance exercise arm although one of the studies was not appropriate for inclusion in a meta‐analysis due to the lack of an appropriate control arm (Lee 2010). The remaining five studies provided data for 237 participants who received a resistance exercise intervention and 164 participants in the control arm. At the end of the intervention period resistance exercise was not statistically more effective than the control intervention (SMD ‐0.18, 95% CI ‐0.39 to 0.02) with no evidence of statistical heterogeneity detected (P = 0.91; I² = 0%) (Analysis 8.1).

8.1 — Comparison 8: Fatigue: resistance training, Outcome 1: Exercise versus no exercise control; post‐test means

Eight studies investigated low‐intensity mind‐body exercises, specifically yoga or qigong, providing eight possible comparisons for inclusion in a meta‐analysis. However, post‐test means +/‐ SD were not available for five of the eight comparisons. The remaining three comparisons provided data for 117 participants who received a mind‐body exercise intervention and 77 participants in the control arm. At the end of the intervention period mind‐body exercise was not statistically different to the control (SMD ‐0.10, 95% CI ‐0.39 to 0.19) with no evidence of statistical heterogeneity detected (P = 0.53; I² = 0%) (Analysis 9.1).

9.1 — Comparison 9: Fatigue: mind‐body exercise, Outcome 1: Exercise versus no exercise control; post‐test means

The remaining studies investigated multimodal exercise interventions, usually incorporating aerobic and resistance training.

Long‐term outcomes

The above results were based upon post‐intervention data as only 12 of the included studies recorded and presented follow‐up data. Five of these studies investigated the effects of an exercise programme delivered during cancer therapy (Windsor 2004; Mutrie 2007; Mustian 2009; Chandwani 2010; Dodd 2010a). Four of the five studies reported that the differences in fatigue levels between intervention and control arm were not significant post‐intervention or at follow‐up (Windsor 2004; Mutrie 2007; Chandwani 2010; Dodd 2010a). Duration of follow‐up was one month (Windsor 2004), three months (Chandwani 2010), six months (Mutrie 2007) and one year (Dodd 2010a) following the intervention. In contrast Mustian 2009 reported that statistically significant improvements in fatigue in the intervention group compared to the control arm were maintained at three months.

Four of the studies that included a long‐term follow‐up investigated the effects of an exercise programme delivered post‐cancer therapy (Daley 2007; Heim 2007; Milne 2008; Sequeira 2012). Daley 2007, Milne 2008 and Sequeira 2012 reported no differences between study arms with regard to fatigue immediately post‐intervention or at 24 week/six‐month follow‐up. In contrast Heim 2007 reported that statistically significant improvements in fatigue in the intervention group compared to the control arm were maintained at three months.

The remaining three studies that included a long‐term follow‐up investigated a mixed sample of patients during and post‐cancer therapy (Cohen 2004; Brown 2006; Courneya 2009). Cohen 2004 stated that outcomes were recorded at one and three months after baseline although there is no differentiation between the two time points in the reported results. Brown 2006 implemented a four‐week intervention and participants were followed up post‐intervention, at eight weeks and at 27 weeks. At eight weeks statistically fewer participants in the control arm reported clinically significant fatigue compared to the intervention arm. No statistically significant differences were reported between groups at eight or 27 weeks. The final study reported no significant differences in fatigue between study arms at the end of the intervention or at the six‐month follow‐up (Courneya 2009).

Exercise Maintenance on Follow‐up

Exercise maintenance was rarely reported and the methods employed were mainly self report and only completed by a small number of study participants.

Time Spent Exercising

Time spent exercising was rarely reported and the methods employed were mainly self report and only completed by a small number of study participants.

Aerobic capacity

Thirty‐eight of the 56 studies measured aerobic capacity or physical fitness, or both. The measures used varied between studies with seven studies employing VO₂ max (an indicator of cardio‐respiratory endurance and aerobic fitness) and the remaining studies employing alternative measures of physical fitness or endurance. Fifteen of the 38 studies reported no difference between the intervention and control groups for post‐test aerobic capacity (Segal 2001a; Burnham 2002; Coleman 2003a; Courneya 2003a; Courneya 2003c; Galantino 2003; Mock 2005; Culos‐Reed 2006; Heim 2007; Mustian 2009; Rogers 2009; Shelton 2009; Culos‐Reed 2010; Galvão 2010; Santa Mina 2012) although it should be noted that two of these studies compared exercise interventions rather than a no exercise control arm (Shelton 2009; Santa Mina 2012). Seventeen of the 38 studies demonstrated a statistically significant difference in aerobic capacity or physical fitness between the intervention and control arm in favour of the intervention (Mock 1994; Mock 1997; Courneya 2003b; Windsor 2004; Campbell 2005; Pinto 2005; Thorsen 2005; Courneya 2007a; Daley 2007; Monga 2007; Mutrie 2007; Chang 2008; Courneya 2008; Milne 2008; Adamsen 2009; Courneya 2009; Segal 2009a). Four of the remaining six studies reported improvements over time but no between‐group comparisons were provided (Pinto 2003; Dimeo 2004; Drouin 2005; Yuen 2007a). One study did not provide any statistical analysis of the aerobic capacity data (Dodd 2010a) and the final study did not reassess aerobic capacity post‐intervention (Battaglini 2008).

Five studies investigated the relationship between aerobic capacity and fatigue; three reported no statistically significant correlation between change in aerobic performance and fatigue (Courneya 2003a; Dimeo 2004; Drouin 2005), whereas Mock 2005 and Milne 2008 reported a statistically significant association between improvements in aerobic fitness and subjective fatigue.

Quality of life

Twenty of the 56 studies did not measure quality of life (QoL) and a further study did not report the quality of life outcome data within the results (Galantino 2003). The remaining 35 studies most frequently used the Functional Assessment of Cancer Therapy to assess QoL, although the version used varied between studies, for example, eight studies used the FACT‐Breast (FACT‐B) and eight the FACT‐general (FACT‐G). Other measures used to assess QoL included the Satisfaction with Life Scale (SWLS), SF‐36, the EORTC‐QLQ‐C30, the QoL Index for Cancer and the Functional Assessment of Chronic Illness Therapy‐Fatigue. Six studies used two measures to assess QoL and one study employed three different QoL measures. The findings were not always consistent between measures.

Sixteen studies reported no statistically significant difference in post‐test QoL between the intervention and control arms (Segal 2001a; Dimeo 2004; Thorsen 2005; Courneya 2007a; Moadel 2007; Courneya 2008; McNeely 2008; Oh 2008; Adamsen 2009; Danhauer 2009; Rogers 2009; Segal 2009a; Culos‐Reed 2010; Lee 2010; Santa Mina 2012; Sequeira 2012), whereas 17 studies reported that exercise was statistically more beneficial to QoL than the control intervention (Burnham 2002; Courneya 2003a; Courneya 2003b; Courneya 2003c; Segal 2003; Headley 2004; Culos‐Reed 2006; Daley 2007; Heim 2007; Monga 2007; Hwang 2008; Milne 2008; Courneya 2009; Mustian 2009; Chandwani 2010; Galvão 2010; Oh 2010).

Campbell 2005 assessed QoL using the FACT‐G, FACT‐B and the SWLS although the FACT‐G was identified as the primary outcome of interest. The FACT‐G as the primary outcome demonstrated a statistically significant improvement in QoL pre‐post test in the exercise arm compared to the control arm. However, the FACT‐B and SWLS did not demonstrate statistically significant differences between arms. Mutrie 2007 employed the FACT‐G as the primary measure of QoL in addition to the FACT‐B. Results demonstrated no statistically significant differences between arms in the FACT‐G although the FACT‐B showed a statistically significant difference between arms post‐intervention in favour of the exercise arm.

Anxiety

Twenty‐one of the 56 studies measured anxiety by means of the State Trait Anxiety Inventory (k = 7), the Profile of Mood State (POMS) Tension‐Anxiety sub scale (k = 7), a visual analogue scale (VAS) (k = 3), the Hospital Anxiety and Depression Scale (HADS) (k = 2), or the Distressed Mood Index (k = 1). In addition, one study measured Social Physique Anxiety which assesses an individual's anxiety in regard to real or perceived evaluation of their physique by others (Milne 2008). One of the 21 studies only measured anxiety at baseline (Brown 2006) and a further study only reported a total HADS score with no separate anxiety data (Heim 2007). Thirteen of the remaining 19 studies that measured anxiety reported no statistically significant benefits of exercise compared to a control arm (Mock 1994; Coleman 2003a; Courneya 2003a; Courneya 2003c; Cohen 2004; Drouin 2005; Pinto 2005; Thorsen 2005; Courneya 2007a; Moadel 2007; Chang 2008; Courneya 2009; Chandwani 2010). Further to this, Burnham 2002 reported that anxiety improved over time in the exercise intervention arm but there was no statistically significant difference in anxiety compared to the control arm. Dimeo 1999 also reported a statistically significant improvement over time in the exercise arm but no comparison with the control arm was reported. Mock 1997, Culos‐Reed 2006 and Oh 2010 reported a statistically beneficial effect of exercise on anxiety compared to a control. In relation to social physique anxiety Milne 2008 also reported exercise to be statistically more effective than a control.

Depression

Twenty‐eight of the 56 studies incorporated an outcome measure for depression including the Centre for Epidemiological Studies‐Depression Scale (CES‐D) (k = 10), POMS‐depression sub scale (k = 9), the Beck Depression Inventory (k = 4), a VAS (k = 3) and the HADS (k = 2). Nineteen of the 28 studies that measured depression reported no statistically significant difference between the intervention and control arm post‐test (Mock 1997; Dimeo 1999; Burnham 2002; Coleman 2003a; Courneya 2003a; Courneya 2003c; Pinto 2003; Cohen 2004; Thorsen 2005; Culos‐Reed 2006; Courneya 2007a; Monga 2007; Mutrie 2007; Chang 2008; Payne 2008; Chandwani 2010; Culos‐Reed 2010; Dodd 2010a; Lee 2010). A further two studies did not report the results from the depression outcome measure (Pinto 2005; Brown 2006) and one study only reported a total HADS score (Heim 2007). Three studies reported a significant improvement in depression in the exercise arm compared to the control arm at the end of the intervention (Courneya 2009; Danhauer 2009; Oh 2010). Daley 2007 also reported statistically less depression following exercise compared to the control arm, however, depression was also statistically less in the placebo exercise group compared to the control arm. Drouin 2005 reported a statistically significant improvement in depression pre‐post test with no corresponding improvement in the control arm, however, statistical analysis for arm by time interaction was not reported. Finally, Mock 1994 reported statistically worse depression in the control arm compared to the intervention arm mid‐way through chemotherapy treatment. This difference was not maintained one month following chemotherapy.

Self efficacy

Only one of the included studies measured exercise self efficacy (Pinto 2005) with the results reported in a separate publication. No statistically significant mediating relationship was demonstrated between self efficacy and minutes of moderate‐intensity activity undertaken.

Discussion

This review provides evidence that exercise is beneficial in the management of cancer‐related fatigue. Statistically significant improvements in fatigue were identified following an exercise programme carried out either during cancer therapy or following cancer therapy. Further to this, statistically significant beneficial effects were identified specific to breast cancer and prostate cancer populations but not for those with haematological malignancies. It remains to be determined whether the type of cancer treatment alters the beneficial effect of exercise on cancer‐related fatigue. Statistically beneficial effects were identified following aerobic training but not following resistance training or low‐intensity mind‐body interventions.

Limitations of the review

This review has incorporated a diverse range of studies with small numbers in several of them. There is a considerable degree of clinical heterogeneity between studies in terms of adjuvant therapy, mode and intensity of exercise, and stage and type of cancer. In addition, moderate statistical heterogeneity was present for the total fatigue (P = 0.03; I² = 33.0%), breast cancer fatigue (P = 0.06; I² = 36.0%) and the aerobic exercise (P = 0.20; I² = 20.0%) meta‐analyses. A high level of statistical heterogeneity was present in the meta‐analyses of studies investigating exercise after cancer therapy (P < 0.0002; I² = 72.0%) and in the prostate cancer population (P = 0.03; I² = 59.0%). However, the meta‐analysis of studies carried out during cancer therapy (P = 0.94; I² = 0%), in haematological patients (P = 0.91; I² = 0%), for resistance exercise (P = 0.91; I² = 0%) and for mind‐body exercise (P = 0.53; I² = 0%) provided no evidence of statistical heterogeneity. The lack of statistical heterogeneity does not exclude heterogeneity, given the small numbers.

Despite a comprehensive search strategy all studies included were published in the English language. This may reflect selective publication of English language studies with statistically significant findings.

Data were not available for the purpose of meta‐analysis for all studies retrieved and identified as suitable. The studies which have not reported the data in full tended to be those that have not identified a favourable outcome for exercise, therefore the results of the meta‐analysis may have been biased in the direction of a positive outcome, making exercise appear more effective. Selective reporting of statistically significant outcomes may also have occurred.

For the purpose of inclusion in the review all studies recorded fatigue as an outcome, however, the primary purpose of the exercise interventions varied between studies. The exercise may therefore have been designed for an alternative purpose, resulting in a lack of beneficial effect for fatigue.

The results of the review should not be considered in isolation as there are a range of non‐pharmacological interventions that may be considered beneficial in the management of cancer‐related fatigue. Interventions that may be delivered in conjunction with an exercise programme include, but are not limited to, psychosocial therapies (Jacobsen 2007; Goedendorp 2009), stress management, nutrition therapy and sleep therapy (Mustian 2007).

Limitations of the included studies

The quality of the included studies was variable with the more recent studies generally being of better quality. It is acknowledged that blinding participants to an exercise intervention may not be a feasible option although methods to reduce this risk of bias should be carefully considered in future trials.

Few of the included studies provided information about people who refused participation in the trial. It is possible that the included participants differed statistically to those who refused participation, particularly in relation to their attitudes towards an exercise programme. It is possible that those who declined to participate would not have achieved the same benefits from an exercise programme as those who chose to participate. Some studies did indicate the main reasons for individuals refusing to participate which included a lack of interest in the study, a refusal or inability to travel, being too busy or having other commitments, a refusal to be randomised and not wanting to be reminded of their cancer.

Statistical power was limited by the small number of participants in several studies. Furthermore, in several studies, fatigue was not identified as the primary outcome, thus sample size calculations (where reported) were based upon a different outcome measure.

The majority of studies were carried out with breast cancer populations. Further research in this area needs to include participants with various cancer diagnoses or other specific cancer populations at various stages of disease, including those receiving palliative care.

In some studies the control groups received less attention from healthcare or research staff, or both, in comparison to the intervention groups. The additional attention may have resulted in a systematic bias (Hawthorne effect) in favour of the intervention.

Few studies included long‐term follow‐up measures and these were not always reported. Further to this a number of studies implemented short‐term interventions (approximately three weeks) which may not have been long enough to produce any statistically significant effects in relation to fatigue.

According to the most recent recommendations from the American College of Sports Medicine (ACSM) (Garber 2011) adults should undertake moderate‐intensity aerobic physical activity for at least 30 minutes on five days per week or vigorous intensity aerobic physical activity for at least 20 minutes on three days each week. The majority of exercise interventions reviewed did not reach these recommendations as exercise was only undertaken two to three times per week at moderate intensity. Twelve of the studies included an intervention that met these recommendations although the participants may have been advised to start at a lower level and progress to 30 minutes on five days per week (Mock 1994; Mock 1997; Segal 2001a; Courneya 2003a; Courneya 2003c; Dimeo 2004; Drouin 2005; Mock 2005; Pinto 2005; Adamsen 2009; Mustian 2009; Rogers 2009). This step‐wise progression is in line with the recommendations from the ACSM (Garber 2011). It should be noted that the ACSM guidelines were developed for healthy individuals and may need to be adapted for cancer survivors according to the stage of disease and adjuvant treatment. ACSM have developed guidelines for older adults which highlight the importance of combined aerobic, resistance and flexibility exercises (Chodzko‐Zajko 2009). These guidelines may be more appropriate for cancer survivors but due to insufficient evidence the optimal exercise programme has not yet been established (Chodzko‐Zajko 2009). This is in line with the ACSM Roundtable Consensus Statement that recognised the need to adapt exercise programmes for the individual cancer survivor according to health status, adjuvant treatment and anticipated disease trajectory (Schmitz 2010). The statement does, however, recommend that all cancer survivors should maintain and increase muscle mass (Schmitz 2010). The effect of resistance training upon fatigue is inconclusive as although there was a trend observed in favour of resistance training this failed to reach significance. In agreement with the current review the consensus group concluded that there was insufficient evidence to recommend alternative forms of exercise such as yoga and tai chi (Schmitz 2010).

Outcome measures

A wide range of unidimensional and multidimensional outcome measures were used in the included studies which has prevented direct comparisons between studies. Data regarding fatigue were not always reported in full or adequately for the purposes of combining outcomes in a meta‐analysis. We made attempts to contact authors where data were not available. It is recommended that multidimensional measures should be used to accurately assess fatigue to ensure that the effects of the intervention are captured in full. For example, an exercise intervention may have beneficial effects on mental and emotional fatigue levels but not physical fatigue levels. These effects would not be captured with a unidimensional tool.

The fatigue measures that were used do not have validated cut‐off scores for mild/moderate/severe fatigue. In addition only one of the outcome measures used had validated minimal clinically important differences (MCIDs) (FACT‐F; Cella 2002), hence it was not possible to determine the number needed to treat to benefit (NNTB) from the data obtained. Included studies did not provide estimates of effect size that could be pooled.

Adherence and contamination

Several studies indicated that participants completed exercise diaries although the results of these were frequently not reported. The accuracy of reporting in exercise diaries has also been questioned (Montoye 1996). Group contamination has previously been reported in studies investigating exercise. This may occur when the control participants undertake exercise or the exercise group do not adhere to the programme. Further contamination may occur when participants do not undertake the exercise at the prescribed intensity or for the prescribed duration. Monitoring of exercise is more difficult when the prescribed programme is home‐based or unsupervised, or both. It is possible that implementation of an exercise programme should be accompanied by a behavioural change intervention to ensure that participants are supported. Only one study monitored participants' self efficacy, which may be an important predictor of adherence to the exercise intervention (Pinto 2005). Adherence to an exercise programme may be lower during cancer treatment as a result of associated adverse effects (Oldervoll 2004), hence patients may require additional support to remain active at this time. Recent research has indicated that providing cancer patients with supporting printed material regarding the exercise or with physical activity monitoring devices may increase levels of physical activity (Vallance 2007).

Authors' conclusions

Implications for practice.

Thirty studies provided overall evidence that aerobic exercise is beneficial in the management of fatigue both during and after cancer treatment and should therefore be considered as one component of a management strategy for fatigue that may include a range of other interventions and education. These findings related specifically to those with solid tumours.

Implications for research.

Further work is necessary to determine the most effective parameters of exercise for fatigue management including multi‐modal exercise (combined aerobic and resistance), frequency and duration of each exercise session, and intensity of exercise. Future studies should also incorporate a long‐term follow‐up. Twenty eight of the included studies were carried out in breast cancer populations therefore future research needs to be undertaken with a broad range of diagnoses, including patients with advanced disease and those receiving palliative care.

Consensus is also needed on the most appropriate multi‐dimensional outcome measure to use. Further work is required to determine cut‐off scores and minimal clinically important differences (MCID) for each of the selected measures.

Future research also needs to consider issues of maximising adherence and promoting self efficacy to exercise as well as identifying the barriers and facilitators to exercise in people with a cancer diagnosis. Other limitations of the existing research that need to be addressed include concealment of group allocation and observer blinding. Trials also need to be adequately powered to detect a change in fatigue. It is likely that the large number of studies with small numbers of participants may be due to limited funding available for non‐commercial trials. Funding agencies may therefore like to consider funding large studies to determine optimal exercise parameters for this patient group.

What's new

Date	Event	Description
29 September 2021	Amended	Published Note text updated.
3 July 2015	Review declared as stable	This review will be superseded. See Published notes.

Date	Event	Description
19 April 2012	New search has been performed	We ran update searches for studies in March 2011.
19 April 2012	New citation required and conclusions have changed	The updated review includes an additional 28 studies; a further 22 were identified but excluded. We have also added 'Risk of bias' tables. The conclusions have been further developed to identify the 'type' of exercise (aerobic) necessary to reduce cancer‐related fatigue.
24 September 2010	Amended	Contact details updated.
30 October 2008	Amended	Converted to new review format.

*Study characteristics*
Methods	RCT; 2 group parallel design
Participants	235 participants with 21 different cancer diagnoses, median of 84 days since diagnosis, were receiving chemotherapy
Interventions	Multimodal supervised groups; aerobic, resistance and relaxation/body awareness. 6 weeks with 4 sessions/week (24 total), 9 hours/week. Control: usual care
Outcomes	Fatigue QoL Side effects Aerobic capacity Muscle strength Leisure time physical activity
Notes	Methodological quality score: 3 Main purpose of exercise: fatigue was the primary outcome
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was done by computer"
Allocation concealment (selection bias)	Low risk	"The allocation sequence was executed by the clinical research unit and concealed from the project team."
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All patients were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2‐group parallel design
Participants	20 breast cancer patients due to receive adjuvant therapy
Interventions	Aerobic and resistance training at 40% to 60% maximum exercise capacity and stretching; twice per week for 15 weeks, up to 60 minutes per session Control: usual care
Outcomes	Fatigue Total caloric intake Cardiovascular endurance Dynamic muscular endurance Body composition
Notes	Methodological quality score: 1 Main purpose of exercise: total calorie intake
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Stated to be randomised
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not clear if all participants completed the study
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, stratification by tumour type, age, gender and Eastern Co‐operative Oncology Group Score
Participants	103 cancer patients diagnosed within past 12 months and due to receive radiotherapy
Interventions	Structured multidisciplinary sessions incorporating self paced exercise. 8 sessions over 4 weeks with 20 mins of exercise per session. Control: usual care
Outcomes	Fatigue Mood Anxiety
Notes	Methodological quality score: 1 Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomly assigned" but methods not described
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	High risk	Drop‐outs not accounted for and unclear whether an intention‐to‐treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 3 groups, stratification by aerobic capacity and quality of life
Participants	21 breast or prostate cancer survivors minimum 2 months post‐treatment
Interventions	Moderate‐intensity exercise: 40% to 50% HR reserve rising to 60% by week 10 Low‐intensity exercise: 25% to 35% HR rising to 40% by week 10 Both groups supervised exercise 3 times/week for 10 weeks. Sessions initially 14 minutes increasing to 32 by week 10. Treadmill, stationary bicycle and stair climbing. Control: no exercise prescribed
Outcomes	Aerobic capacity Body composition Flexibility Quality of life Fatigue Anxiety Confusion Depression Energy Anger
Notes	Methodological quality score: 2 Main purpose of exercise: to improve physiological and psychological function
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomly assigned" but no method described
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	High risk	Intention‐to‐treat not applied as one participant from the control arm was excluded as they took part in exercise training.
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, stratification by type of adjuvant cancer therapy
Participants	22 breast cancer patients post‐surgery, receiving therapy. Not exercising vigorously 3 times/week for 20 mins or more
Interventions	Exercise: supervised, 2 times/week for 12 weeks. Warm up, 10 to 20 mins of various exercises, cool down. Approximate 60% to 75% age‐adjusted HRmax. Control: no intervention until week 12
Outcomes	Quality of life Fatigue Physical activity levels
Notes	Methodological quality score: 3 Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"randomly allocated by computer‐generated numbers"
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, wait list control
Participants	37 breast cancer patients, diagnosed at least 2 years previously. Had received a variety of adjuvant treatments.
Interventions	Exercise: yoga; supervised in groups (5 to 10 per group) for 120 minutes once/week for 8 weeks, and encouraged to practise daily with CD and illustrated handbook Control: wait list, received usual care
Outcomes	Fatigue Hot flashes Joint pain Mood Sleep disturbance Bother
Notes	Methodological quality score: 3 Main purpose of exercise: to reduce hot flashes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization...random number table"
Allocation concealment (selection bias)	Low risk	"Assignments were concealed in envelopes that were not opened until all patients had completed their baseline assessment."
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	"The research assistant collecting assessment data was kept blind with regard to patient condition assignments". Fatigue was self report.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All patients were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, parallel design
Participants	22 patients with acute myelogenous leukaemia, due to commence chemotherapy
Interventions	Exercise: walking exercise programme during inpatient CT, 12 minutes of supervised walking on 5 days/week for 3 weeks. Aimed to reach a target intensity of resting heart rate plus 30. Control: attention control, no additional intervention specific to fatigue
Outcomes	Fatigue Walk distance Symptom distress Mood (including anxiety and depression)
Notes	Methodological quality score: 2 Main purpose of exercise: to reduce fatigue
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomized" but method not stated
Allocation concealment (selection bias)	Unclear risk	Allocation not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	CCT; 2 groups
Participants	39 lymphoma patients either during treatment or up to 12 months post‐treatment
Interventions	Exercise: yoga class once per week for 7 weeks. Control: wait list
Outcomes	Distress Anxiety Depression Fatigue Sleep disturbance
Notes	Methodological quality score: 0 Main purpose of exercise: to reduce stress
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Group assignment was conducted sequentially using minimization"
Allocation concealment (selection bias)	Low risk	"The allocation process was concealed from all investigators because all the relevant information was entered into a computer program and group assignment was determined by the program."
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	High risk	1 participant dropped out of the study, no data were available for the individual and intention‐to‐treat analysis was not undertaken
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups
Participants	24 multiple myeloma patients receiving high‐dose chemotherapy and tandem peripheral blood stem cell transplantation
Interventions	Exercise: individualised home‐based programme of strength and aerobic training Control: no individual programme, advised to remain active and walk for 20 mins at least 3 times/week
Outcomes	Fatigue Mood disturbance Sleep variables Body composition Strength Aerobic capacity
Notes	Methodological quality score: 2 Main purpose of exercise: to improve fatigue and sleep
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomly assigned" but method of sequence generation not described
Allocation concealment (selection bias)	Unclear risk	"having a research assistant draw a sealed envelope containing group assignment", envelope not described as 'opaque'
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not all participants accounted for and not clear whether intention‐to‐treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT (wait list control); 2 groups
Participants	102 colorectal cancer survivors following surgery within past 3 months
Interventions	Exercise: personalised home‐based cardiovascular and flexibility programme lasting 16 weeks. Received a weekly phone call. Control: no intervention, requested not to commence a structured exercise programme
Outcomes	Quality of life Fatigue Depression Anxiety Cardiovascular fitness Body composition Flexibility
Notes	Methodological quality score: 3 Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"participants were randomly assigned to an exercise group or wait‐list control group using a random‐numbers table"
Allocation concealment (selection bias)	Unclear risk	Allocation not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, groups stratified for content
Participants	108 cancer survivors with various diagnoses
Interventions	Exercise: group psychotherapy plus individualised home‐based cardiovascular and flexibility exercises. 3 to 5 times/week for 20 to 30 mins at 65 to 75% HRmax. Control: group psychotherapy only Both groups 10 weeks
Outcomes	Quality of life Satisfaction with life Depression Anxiety Fatigue Cardiovascular fitness Body composition Flexibility Physical activity levels
Notes	Methodological quality score: 3 Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"classes were randomized to experimental conditions using a random numbers table"
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	12 participants did not complete the study or provide post‐test data and were not included in intention‐to‐treat analysis
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 3 groups, stratified by treatment centre and chemotherapy protocol
Participants	242 women with breast cancer initiating chemotherapy
Interventions	Aerobic exercise: supervised 3 times/week during chemotherapy. Resistance exercise: supervised 3 times/week during chemotherapy. Control: usual care
Outcomes	Quality of life Fatigue Anxiety Depression Self esteem Physical fitness Body composition Chemotherapy completion rate Lymphoedema
Notes	Methodological quality score: 2 Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"randomly assigned...using a computer generated program"
Allocation concealment (selection bias)	Low risk	"The allocation sequence was...concealed from the project directors at each site who assigned participants to groups"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	see Courneya 2007a
Participants	see Courneya 2007a
Interventions	see Courneya 2007a
Outcomes	see Courneya 2007a
Notes	see Courneya 2007a
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"randomly assigned...using a computer generated program"
Allocation concealment (selection bias)	Low risk	"The allocation sequence was...concealed from the project directors at each site who assigned participants to groups"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 3 groups, stratification by chemotherapy and tamoxifen
Participants	108 women 12 to 36 months post‐treatment for breast cancer
Interventions	Exercise: supervised aerobic exercise 3 times per week for 8 weeks Placebo exercise: light intensity body conditioning/stretching Control: usual care
Outcomes	Quality of life Fatigue Depression Self perception Physical activity Aerobic fitness Body composition
Notes	Methodological quality score: 3 Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"A telephone randomization service was provided by an independent trials unit. Randomization to the three treatment arms was on a 1:1:1 ratio and was performed using stratified random permutated blocks"
Allocation concealment (selection bias)	Low risk	"A telephone randomization service was provided by an independent trials unit. Randomization to the three treatment arms was on a 1:1:1 ratio and was performed using stratified random permutated blocks"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, stratified for tumour location
Participants	72 various cancer diagnoses, post anti‐cancer treatment
Interventions	Exercise: supervised stationary bike, 30 mins/per session, 5 days/week. 80% HRmax or 13 to 14 on Borg scale. Control: relaxation training 45 mins, 3 times/week Both groups lasted 3 weeks
Outcomes	HRQOL (including fatigue) Physical performance
Notes	Methodological quality score: 3 Main purpose of exercise: to reduce fatigue severity
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was carried out using a computer‐generated random number list"
Allocation concealment (selection bias)	Low risk	"randomisation sequence was concealed until assignment of interventions"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	3 participants dropped out of the study; unclear if intention‐to‐treat analysis was performed
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 3 groups
Participants	106 patients, majority with breast cancer but patients with ovarian and colorectal cancer also included. All patients were due to commence chemotherapy
Interventions	Exercise during chemotherapy: home‐based for 1 year, 3 to 5 times/week, 20 to 30‐minute/session at 60% to 80% VO₂ peak or 12 to 14 on Borg scale (somewhat hard). Weekly follow‐up calls to assess progress and adjust intensity. Exercise post‐chemotherapy: home‐based for 6 to 8 months, 3 to 5 times/week, 20 to 30‐minute/session at 60 to 80% VO₂ peak or 12 to 14 on Borg scale (somewhat hard). Weekly follow‐up calls to assess progress and adjust intensity. Control: usual care, no exercise prescription. Telephoned weekly to enquire about health and general response to cancer treatment.
Outcomes	Fatigue Sleep disturbance Depression Pain Physical ability Quality of life (not included in the analysis)
Notes	Methodological quality score: 2 Main purpose of exercise: to reduce fatigue
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomized" but method not stated
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	See Dodd 2010a
Participants	See Dodd 2010a
Interventions	See Dodd 2010a
Outcomes	See Dodd 2010a
Notes	See Dodd 2010a
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomized" but method not stated
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups parallel design
Participants	57 patients with prostate cancer, some were receiving radiotherapy
Interventions	Exercise: combined progressive resistance and aerobic training twice/week for 12 weeks, supervised in small groups (n = 1 to 5). Resistance progressed from 12 to 6 repetition maximum with 2 to 4 sets/exercise. Aerobic training: 15 to 20 minutes of cycling, walking or jogging at 65% to 80% HRmax, perceived exertion 11 to 13 on Borg scale. Control: usual care
Outcomes	Fatigue Quality of life Aerobic capacity Balance and falls self efficacy Functional performance Muscle strength Muscle endurance Body composition Blood profile
Notes	Methodological quality score: 3 Main purpose of exercise: to increase muscle mass and strength, physical function, cardiorespiratory capacity and health status.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"participants were randomly assigned...1:1 using a computer random assignment program"
Allocation concealment (selection bias)	Low risk	"The allocation sequence was concealed from the project coordinator and exercise physiologist"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Some missing data
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 3 groups; intervention groups stratified by operation type, historical control
Participants	39 patients with breast cancer post‐treatment (includes 18 drawn from a historical control cohort)
Interventions	Scapula exercise: supervised shoulder stretching and strengthening for 40 minutes once/week for 8 weeks General exercise: supervised range of motion and body conditioning for 40 minutes once/week for 8 weeks Control: historical group
Outcomes	Fatigue Quality of life Depression Shoulder disability, function and range of movement Pain
Notes	Methodological quality score: 1 Main purpose of exercise: to reduce upper limb dysfunction
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization to the two treatment arms was achieved on a 1:1 ratio"
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

PERMALINK

Exercise for the management of cancer‐related fatigue in adults

Fiona Cramp

James Byron-Daniel

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Assessment of risk of bias in included studies

Measures of treatment effect

Assessment of heterogeneity

Results

Description of studies

Included studies

Participants

Interventions

Outcome measures

Excluded studies

Risk of bias in included studies

1.

2.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Fatigue

All studies: exercise versus control, post‐test means

1.1. Analysis.

Intervention timing

2.1. Analysis.

3.1. Analysis.

Disease‐specific comparisons

4.1. Analysis.

5.1. Analysis.

6.1. Analysis.

Mode of exercise

7.1. Analysis.

8.1. Analysis.

9.1. Analysis.

Long‐term outcomes

Exercise Maintenance on Follow‐up

Time Spent Exercising

Aerobic capacity

Quality of life

Anxiety

Depression

Self efficacy

Discussion

Limitations of the review

Limitations of the included studies

Outcome measures

Adherence and contamination

Authors' conclusions

Implications for practice.

Implications for research.

What's new

History

*Study characteristics*
Methods	RCT; 2 group cross‐over design
Participants	58 breast cancer patients within 2 years of completing adjuvant therapy
Interventions	Exercise: 2 times per week supervised group programme including aerobic and resistance training for 12 weeks Control: usual care and phone call every 3 weeks
Outcomes	Quality of life Fatigue Anxiety Aerobic capacity
Notes	Methodological quality score: 3. Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Participants were randomly assigned...using a computer‐generated program"
Allocation concealment (selection bias)	Low risk	"Group assignments were concealed from the project director who recruited participants to the trial"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, wait‐list control, stratification by cancer treatment
Participants	128 breast cancer patients diagnosed between 2 weeks and 5 years previously; 48% received adjuvant therapy during the study
Interventions	Exercise: yoga, 12 weeks of one 1.5 hour group class/week, asked to practice at home daily. Were also permitted to attend additional classes. Control: wait list
Outcomes	Fatigue Quality of life Distress including anxiety Spiritual well‐being
Notes	Methodological quality score: 2 Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Randomly assigned" but method not stated
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	CCT, 2 groups
Participants	46 stage I and II breast cancer patients, post‐surgery, receiving radiotherapy
Interventions	Exercise: individualised brisk 20 to 30‐min incremental walk followed by 5‐min cool down 4 to 5 times/week in own setting Control: no intervention
Outcomes	Physical function (12‐min walk test) Symptoms Fatigue
Notes	Methodological quality score: 2 Main purpose of exercise: to increase physical function and reduce symptom intensity
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Following random assignment of the first subject, subsequent subjects were alternately assigned"
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	4 participants withdrew or dropped out of the study and were not included in the analysis
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, stratified by activity level
Participants	21 patients with prostate cancer due to receive radiotherapy
Interventions	Exercise: supervised aerobic exercise (walking on a treadmill), 3 times/week for 8 weeks. 10‐min warm‐up, 30 aerobic, 5 to 10‐min cool down. Target HR = (0.65) (HRmax) + resting HR Control: usual care
Outcomes	Fatigue Aerobic fitness Quality of life Depression Flexibility Strength
Notes	Methodological quality score: 2 Main purpose of exercise: to prevent fatigue and improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomly assigned" but method not described
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 group, stratified by diagnosis
Participants	38 patients with breast or prostate cancer receiving radiotherapy
Interventions	Exercise: home‐based aerobic and resistance training every day for 4 weeks. Target was to increase steps by 5% to 20% each day up to 10,000 steps at a moderate intensity (60% to 70% HR reserve). Plus 11 resistance exercises, increasing repetitions up to 4 sets of 15 with moderate resistance. Control: usual care, requested not to commence formal exercise programme
Outcomes	Fatigue Quality of life Aerobic capacity Strength Body composition Adherence
Notes	Methodological quality score: 2 Main purpose of exercise: to increase aerobic capacity, strength, muscle mass, fatigue and quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomized, using a randomization scheme with blocks of four", not other details provided
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, stratification for hospital and treatment
Participants	203 breast cancer patients during treatment
Interventions	Exercise: supervised 12‐week group exercise 2 times/week, 45 mins/session at moderate intensity. Participants encouraged to exercise once/week at home. Control: usual care
Outcomes	Quality of life Fatigue Depression Mood Body composition Physical activity 12‐minute walk
Notes	Methodological quality score: 3. Main purpose of exercise: to improve quality of life
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was done by telephone to an interactive voice response system"
Allocation concealment (selection bias)	Low risk	"Randomisation was done by telephone to an interactive voice response system"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 group; stratified by treatment (ongoing or complete)
Participants	18 cancer patients with various diagnoses and at various stages of treatment
Interventions	Exercise: Qigong, supervised groups, 1 to 2 sessions/week for 8 weeks, 90 mins/session and practice at home for 1 hour/day Control: usual care, asked to refrain from joining a Qigong class
Outcomes	Quality of life (including a fatigue sub scale) Symptom experience Inflammation
Notes	Methodological quality score: 2 Main purpose of exercise: to improve quality of life, reduce treatment symptoms and inflammation
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was done by a computer program"
Allocation concealment (selection bias)	High risk	Not done
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 2 groups, stratification for age, cancer stage and adjuvant therapy received
Participants	86 stage 0 to II sedentary breast cancer patients post‐treatment
Interventions	Exercise: moderate‐intensity home‐based activities, 55% to 65% HRmax. 10 mins twice/week increasing to 30 mins at least 5 days/week over 12 weeks. Weekly phone calls during exercise. Control: weekly phone call to monitor symptoms
Outcomes	Body composition Activity levels 1‐mile walk Motivational readiness for exercise Mood Fatigue Body esteem
Notes	Methodological quality score: 2 Main purpose of exercise: to increase physical activity, fitness; improve mood, physical symptoms and body esteem
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"randomized" but method not stated
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	RCT; 3 group, stratified for adjuvant treatment, random numbers table
Participants	123 stage I to II breast cancer, within 2 weeks of commencing adjuvant therapy
Interventions	Supervised exercise: 3 times per week progressive walking, further 2 times unsupervised. Unsupervised exercise: 5 times per week progressive walking at 50% to 60% maximal predicted oxygen uptake. Control: usual care
Outcomes	HRQOL Fatigue Aerobic capacity Body composition
Notes	Methodological quality score: 3 Main purpose of exercise: to improve physical function and HRQOL
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"randomly assigned participants to one of three groups using a random numbers table"
Allocation concealment (selection bias)	High risk	"A study coordinator revealed group assignment after baseline testing"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported

*Study characteristics*
Methods	See Segal 2001a
Participants	See Segal 2001a
Interventions	See Segal 2001a
Outcomes	See Segal 2001a
Notes	See Segal 2001a
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"randomly assigned participants to one of three groups using a random numbers table"
Allocation concealment (selection bias)	High risk	"A study coordinator revealed group assignment after baseline testing"
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding not done
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for
Selective reporting (reporting bias)	Low risk	Fatigue outcome reported