Objectives Research supports the use of exercise to improve quality of life and reduce the side effects of breast cancer treatment, such as fatigue and decreased aerobic capacity. Previously published reviews have focused on reporting the outcomes of exercise interventions, but have not critically examined the exercise prescriptions. The purpose of this review is to evaluate the application of the principles of exercise training in the exercise prescriptions reported in intervention studies for breast cancer survivors.
Methods Databases were searched for randomised controlled trials of exercise in women diagnosed with breast cancer. Data were extracted to evaluate the application of the principles of exercise training, the reporting of the components of the exercise prescription and the reporting of adherence to the exercise prescription.
Results Of the 29 papers included, none applied all principles of exercise training. Specificity was applied by 64%, progression by 41%, overload by 31%, initial values by 62% and diminishing returns and reversibility by 7% of trials. No study reported all components of the exercise prescription.
Conclusion The application of the principles of exercise training varied greatly, and reporting of the exercise prescribed and completed was incomplete. When principles of exercise training are applied to the development of exercise protocols, there is greater confidence that non-significant findings reflect lack of efficacy of exercise rather than deficiencies in the prescription. Incomplete reporting of the exercise prescription and adherence to the prescription limits the reproducibility of the intervention, and the ability to determine the dose of exercise received by participants.
Statistics from Altmetric.com
Over 1.38 million women are diagnosed with breast cancer annually and more than 458 000 will die from the disease worldwide.1 Early detection and advancements in treatment options have led to a steady improvement in 5-year relative survival rate in developed countries, such as Canada, where the 5-year relative survival rate is currently 87%.2 This trend has resulted in a growing population of breast cancer survivors who are potentially faced with a number of long-term side effects of cancer and its treatment. Some of these include decreased aerobic capacity and strength, weight gain, fatigue and reduced quality of life.3,–,7
Exercise has been proposed as an intervention that can be used to reduce side effects related to cancer treatment. The terms physical activity and exercise are sometimes used interchangeably, however, for the purpose of this review we will differentiate between the two. Physical activity is a general term for any body movement via skeletal muscles that results in an increase in energy expenditure, whereas exercise is a specific type of physical activity that is repetitive and planned with the objective to improve or maintain physical fitness.8 Evidence from systematic reviews and meta-analyses of the literature supports the use of exercise as a way to increase aerobic fitness and muscular strength, reduce body weight and improve body composition, increase quality of life and decrease fatigue in breast cancer survivors.4 ,9,–,12 The cumulative evidence supporting the benefits of exercise for breast cancer survivors is encouraging and has led to several published exercise recommendations.13,–,15 Although it is important that exercise be promoted for breast cancer survivors, it may be premature to outline specific exercise prescriptions until published exercise interventions are evaluated more rigorously. Previous reviews have examined the methodological quality of exercise intervention studies in breast cancer survivors, with most attention paid towards study design (eg, single group vs randomised controlled trials (RCTs)).4 ,5 ,10 ,16 ,17 These reviews have summarised the exercise prescription, but have not critically evaluated whether or not the exercise prescribed was the most appropriate given the desired outcome (eg, improved body composition or increase in aerobic capacity) and target population (eg, women undergoing chemotherapy or women reporting cancer-related fatigue). Therefore, it is currently unclear how to tailor the exercise prescriptions at different times throughout the cancer continuum (ie, during or after adjuvant treatment) or to develop specific prescriptions to achieve a particular outcome (ie, to increase aerobic fitness or to prevent decline, to reduce cancer-related fatigue or improve quality of life).
The main principles of exercise training are well established in the exercise field; these are specificity, overload, progression, initial values, reversibility and diminishing returns (table 1).18 Rigorous application of these principles of exercise training in the design of an exercise prescription ensures that the most appropriate type and dose of exercise is employed to achieve the desired outcome, therefore maximising the potential benefits. When inappropriate interventions are prescribed for the stated outcome goals, studies may fail to find significant results and draw erroneous conclusions about the efficacy of exercise. Furthermore, it is important that the exercise prescription is reported in sufficient detail to allow for replication of the intervention, and adherence of participants is reported to allow the reader to understand the dose of exercise that was delivered during the intervention.
The purpose of this review is to specifically evaluate the exercise interventions in RCTs for breast cancer survivors in relation to (1) the application of the principles of exercise training in developing the exercise prescription, (2) the reporting of the components of the exercise prescription in both the methods and the results and (3) the adherence of participants to the prescribed intervention in order to outline how this affects interpretation of the success or failure of interventions.
The following databases were searched to May 2010: Medline, CINAHL, Sport Discus and EMBASE. Search terms related to breast cancer (eg, breast cancer, breast neoplasm, breast carcinoma) and exercise (exercise, aerobic exercise, resistance exercise, physical activity, walking) were used and modified based on suggestions from individual search engines. The search was limited to English-language papers published in peer-reviewed journals between 1990 and May 2010. Exercise studies that included participants with other cancer types were excluded. Only RCTs with one or more treatment arms involving aerobic and/or resistance exercise and a control arm were eligible for inclusion. Behavioural interventions in which the only outcome measures reported were physical activity levels or psychosocial outcomes were excluded. Alternative exercise interventions such as yoga, Pilates or Tai Chi, as well as specific therapeutic interventions (ie, shoulder rehabilitation) were excluded.
Two reviewers (SEN and KLC) inspected the title and abstract of each study identified in the search to determine eligibility. The two lists were compared and the eligibility of any paper in question was resolved by consensus. Identified papers were then obtained in full and further reviewed. Information on the study design, protocol and outcomes were extracted from each publication. For each eligible study, details of the intervention protocols were abstracted using defined criteria. Each study was assigned a rating for each of the principles of exercise training based on application of the principle. Application of the specific principle was assigned a '+', whereas 'NR' (not reported) was assigned if the principle was not used in the prescription. A '?' was assigned if it was unclear whether or not the principle was used, or if the principle was reportedly used but inconsistently applied (eg, in a mixed intervention, using progression for the resistance component but not the aerobic component) or was unclear. The prescribed training programme for each study was abstracted according to the 'FITT' format (frequency (number of sessions per week), intensity (prescribed intensity of the activity), time (duration of each exercise bout) and type (aerobic and/or resistance activity)). Adherence of the study sample to the exercise prescription was abstracted using the FITT format to determine the dose of exercise achieved in the study. The percentage of studies meeting each criterion (application of the principles of exercise training, reporting of the components of the prescription using the FITT format and reporting of adherence using the FITT format) was calculated, but no other statistical techniques were utilised.
The process for the selection of the papers to be included in this review is outlined (figure 1) and relevant data extracted on study design and outcomes from each paper are described (table 2). Fourteen studies (48%) used an aerobic intervention,6 ,19,–,31 3 (10%) used a resistance intervention32,–,34 and 12 (41%) used a combined aerobic and resistance intervention ('mixed').11 ,35,–,45 Most studies had a single intervention versus control group, except for three (10%), three-arm trials (ie, an aerobic-only, resistance-only and control),11 ,44 ,45 which are also referred to as 'mixed' studies. Fifteen studies (52%) were conducted during adjuvant treatment,6 ,11 ,21 ,23 ,25,–,27 ,31 ,35,–,38 ,42,–,44 whereas 14 studies (48%) were conducted after the completion of adjuvant treatments (ie, chemotherapy and/or radiation).19 ,20 ,22 ,24 ,28,–,30 ,32,–,34 ,39,–,41 ,45 The length of the interventions ranged from 7 weeks21 to 24 months.34 Six studies (21%) reported follow-up measures after the intervention was completed, ranging from 12 to 156 weeks.29 ,34 ,38 ,42
Application of the principles of exercise training
The application of the principles of exercise training is detailed and subsequently evaluated in table 3. The principle of specificity was applied appropriately by 9 of 14 aerobic studies (64%),6 ,19,–,21 ,23 ,25 ,28 ,29 ,31 all three resistance exercise studies (100%)32,–,34 and 9 of 12 mixed studies (75%).11 ,35 ,37,–,41 ,43 ,44 It was not reported by two aerobic studies (14%),22 ,26 and incorrectly applied by three aerobic studies (21%)24 ,27 ,30 and three mixed studies (25%).36 ,42 ,45 The principle of progression was applied appropriately by 41% of studies: 7 of 14 aerobic studies (50%),19 ,22 ,24 ,26,–,29 2 of 3 resistance studies (67%)32 ,33 and 3 of 12 mixed interventions (25%).11 ,39 ,44 Of studies reviewed, 28% did not apply this principle (two aerobic (14%),20 ,21 one resistance (33%)34 and five mixed interventions (42%)36,–,38 ,42 ,45), whereas this principle was incorrectly applied or unclear in 31% of studies (five aerobic (36%)6 ,23 ,25 ,30 ,31 and four mixed (33%)35 ,40 ,41 ,43). The principle of overload was applied appropriately by 31% of studies: 4 of 14 aerobic (28%),19 ,21 ,23 ,28 2 of 3 resistance (67%)32 ,33 and 3 of 12 mixed interventions (25%).11 ,35 ,39 Overload was not applied by 45% of studies: five aerobic (35%),20 ,24 ,25 ,27 ,29 one resistance (33%)34 and seven mixed (58%).36,–,38 ,41 ,42 ,44 ,45 For the remaining seven studies (24%), the appropriate application of this principle was unclear. The principle of initial values was applied appropriately by 62% of studies: 7 of 14 aerobic (50%),20,–,23 ,28 ,29 ,31 all 3 resistance (100%)32,–,34 and 8 of 12 mixed interventions (67%),11 ,37,–,42 ,45 whereas 21% of studies (3 aerobic (21%)19 ,25 ,27 and 3 mixed (25%)35 ,36 ,43) did not apply this principle. For 17% of studies, the appropriate application of this principle was unclear (four aerobic (29%)6 ,24 ,26 ,30 and one mixed (8%)44). Follow-up after an intervention has been completed allows for evaluation of the principles of diminishing returns and reversibility. Of the studies reviewed, 21% of studies (two aerobic (14%),20 ,29 two resistance (67%)32 ,34 and two mixed (17%)38 ,42) reported results from follow-up after the intervention was complete; however, the majority did not repeat assessments of all outcome measures. One additional study did follow-up after the intervention was complete, but only on psychosocial variables.41 Of the studies that included a follow-up assessment, only one study (3%) appropriately applied the principle of reversibility20 and only one study (3%) appropriately applied the principle of diminishing returns.32
Reporting of the components of the exercise prescription
Reporting of the components of the exercise prescription used, as well as reporting of adherence to the exercise intervention is summarised in figure 2A,B. Ten studies (34%) failed to report all components of the exercise prescription (FITT) in their methods.6 ,11 ,25 ,30 ,31 ,37 ,38 ,40 ,41 ,44 A prescribed frequency was reported by all but one study (97%).40 A prescribed intensity was reported in 23 of 29 studies (79%).11 ,19,–,24 ,26,–,29 ,31,–,36 ,39 ,40 ,42,–,45 A prescribed duration (time) was reported in 23 of 29 studies (79%).6 ,19,–,30 ,32,–,36 ,39 ,40 ,42 ,43 ,45 Duration was not reported in 1 of 14 aerobic studies (7%)31 and 1 of 12 mixed intervention studies (42%),11 ,37 ,38 ,41 ,44 but was consistently reported in all of the resistance training studies (100%).32,–,34 A prescribed type of exercise was reported in all studies, but two of the aerobic studies (14%)20 ,29 and three of the mixed interventions (17%)36 ,40 ,44 were assigned a '?' because they prescribed 'aerobic exercise' without describing the specific modes of activity included.
Adherence of participants to the prescribed intervention
No study reported complete details about adherence of participants to each component of the original exercise prescription. Two studies (7%) did not report adherence to any of the FITT components in their results.29 ,42 Fourteen studies (48%) reported only one component,6 ,28 ,31,–,36 ,38 ,39 ,41 ,43,–,45 six studies (21%) reported two components11 ,19 ,21 ,24 ,27 ,30 ,40 and six (21%) reported three components.20 ,22 ,23 ,25 ,26 ,37 Frequency of exercise (percentage of exercise sessions attended) was reported in 24 of 29 studies (83%),11 ,19,–,28 ,30 ,32,–,41 ,43 ,45 whereas intensity was reported in 3 of 29 studies (10%).11 ,20 ,23 Information on the duration of exercise bout was reported in 9 of 29 (31%) studies20 ,22 ,23 ,25,–,27 ,37 ,38 ,40 and the type of exercise completed was reported in 11 of 29 studies (48%).6 ,19 ,21 ,22 ,24,–,26 ,30 ,31 ,39 ,44
The number of studies examining exercise benefits for breast cancer survivors has increased tremendously in recent years.4 ,9,–,12 These studies have provided a wealth of evidence for the safety and feasibility of undertaking exercise for breast cancer survivors both during and after adjuvant treatment.15 However, there has been no consistent attention paid to the application of the basic principles of exercise training in developing research exercise prescriptions or rigour in the reporting of the prescriptions. The principles of exercise training are a central tenet that must be followed in order to thoroughly evaluate the efficacy of prescribed exercise interventions to achieve specific outcomes as well as to develop new prescriptions for clinical populations. Although a meta-analysis can improve statistical power to determine efficacy of exercise, it completely ignores the quality of the exercise prescription and thus the reader must assume that the prescription was appropriate and sound. Based on our review, this assumption may not be reliable. Adherence to the principles of exercise training will maximise the potential benefits of exercise. Lack of attention to the principles of exercise training may underestimate the impact that exercise can have.
The principles of exercise training have been used to guide recommendations for exercise prescriptions to improve bone health in the general population46 and to interpret the literature on exercise interventions and bone health in cancer survivors.47 Unfortunately, few other papers were identified that evaluate the clinical exercise literature with respect to principles of training. In a review of studies that prescribed progressive resistance training to breast cancer survivors, Cheema et al48 did address adequate reporting of study design (blinding, drops out reported, etc), reporting of frequency and intensity of resistance training and compliance to the exercise intervention. However, this information was not critically evaluated. We hope that our review may stimulate interest in evaluating these central tenants of exercise prescription for trials in other populations.
Application of the principles of exercise training
The principle of specificity was the principle most commonly included in the papers reviewed, although one-third of the 29 studies did not adequately apply this principle to the development of the exercise prescription. Overall, the aerobic exercise interventions consistently included the principle of specificity by choosing a population vulnerable to decreased cardiovascular fitness (ie, during chemotherapy6 ,25 ,31) or selecting aerobic fitness as a primary outcome.19,–,21 ,23 ,28 ,29 Studies were judged as unclear if they did not justify an appropriate intervention for increasing the stated outcomes. For example, in an aerobic intervention consisting of step aerobics or circuit training (which included jumping rope and skate-jumping), multiple outcomes were measured, including upper body strength, which is unlikely to be affected by the intervention.27 Two studies did not apply the specificity principle because they either attempted to decrease fatigue in a population that was not experiencing fatigue26 or used aerobic activity, rather than resistance exercise to improve lean body mass.22 Resistance-only interventions appropriately used the principle of specificity by matching the intervention to the study objective of increasing lean body mass,32 determining incidence of lymphoedema with weight training33 or increasing muscle strength.34 Of the mixed interventions, 9 of 12 studies justified their combined use of aerobic and resistance training (or comparison of) with their choice of both aerobic fitness and muscle strength outcome measures.11 ,35 ,37,–,41 ,43 ,44 However, in three of the mixed intervention studies, it was unclear how specificity was applied because outcome measures specific to resistance training (eg, muscle strength) were not included.36 ,42 ,45 When an exercise prescription does not match the target outcomes, non-significant findings would not be surprising and may lead to false conclusions regarding exercise efficacy. Specificity can be achieved with consistent and appropriate selection of exercise modes, outcomes and populations.
The principle of progression was reported more often in resistance training interventions than aerobic interventions. Aerobic interventions were considered adherent to the principle if they clearly described the increases in frequency, intensity or duration (time) of the training programme over time. For example, Courneya et al19 prescribed an aerobic exercise programme that progressed from 15 to 35 min/week, by increasing exercise time by 5 min every 3 weeks. Studies that stated the intervention was progressive, but provided no detail on how the programme progressed, provide little useful information for clinical application of a training programme.6 ,25 ,30 ,31 For example, one study was only progressive for the first 3 weeks of the 8-week long intervention, with no justification as to why progression did not continue.23 In the resistance-only interventions, two of the three studies clearly described the progression in terms of number of sets and weight lifted.32 ,33 Of the mixed interventions, four studies described progression for resistance but not for aerobic exercise35 ,40 ,41 ,43 and five studies failed to include progression at all.36,–,38 ,42 ,45 It may be more straightforward to prescribe and report progression for resistance exercise than aerobic. As a participant is able to perform an exercise with a specific load correctly for the prescribed number of repetitions, they are often asked to increase resistance the next session. It may be perceived as more difficult to prescribe increases in training intensity for aerobic exercise because monitoring intensity may be more difficult for patients to do (eg, heart rate monitors, walking pace). Progression of the duration of aerobic exercise is easier to administer and will build endurance, but may not provide enough overload to increase maximal aerobic capacity.49 Building progression into prescribed protocols is necessary for long-term exercise benefits and should be incorporated into all protocols. Similarly, achieving the prescribed progression and/or describing modifications to the training programme is important information for others seeking to replicate a specific training programme in the community setting.
If exercise is expected to yield improvements in outcome measures, a training programme must adhere to the principle of overload by ensuring the prescription is greater in terms of frequency, duration or intensity, than what the individual is already doing. If an individual does less exercise during the intervention than she was doing prior, she may actually detrain and experience no fitness improvements. This could be erroneously interpreted as failure of the exercise programme itself. Few of the aerobic interventions correctly used the principle of overload by prescribing intensity based on measured initial fitness.19 ,21 ,23 ,28 Use of the principle was unclear in two aerobic intervention studies that based their prescription on estimates of maximum heart rate and the validity of this approach in breast cancer survivors is unknown.22 ,26 In addition, three aerobic studies did not adequately describe how initial fitness levels were determined.6 ,30 ,31 Two of three resistance interventions based their prescription on initial fitness,32 ,33 but one was unclear because it did not justify the target intensity and limited participants from lifting more than 20 pounds for safety precautions.34 Of the 12 mixed interventions, 3 studies prescribed intensity based on measured initial fitness,11 ,35 ,39 2 studies were unclear as they reported overload in the resistance portion only of the intervention40 ,43 and 7 mixed studies did not report the use of overload.36,–,38 ,41 ,42 ,44 ,45 For example, one study conducted 20 min of group-based activity twice a week, although the inclusion criteria required that participants must not have been exercising vigorously for three times a week, thus providing the opportunity for detraining in some participants.36 To ensure that sufficient overload is achieved, future studies should strongly consider prescribing exercise relative to measured baseline fitness. Symptom-limited cardiopulmonary exercise testing has been shown to be safe in the cancer population.50 However, a recent review by Jones et al51 examined the use of exercise testing during supervised exercise interventions focused on increasing aerobic fitness (measured as peak oxygen consumption or VO2peak) in cancer survivors and reported that the majority of studies did not adhere to the fundamental principles of exercise testing, as recommended by national and international guidelines for exercise testing. This finding highlights the utility of exercise testing to determine baseline values to aid in the development of effective exercise prescriptions and also acknowledges that even when baseline testing has been used to determine initial values, there may be limitations to the accuracy and reliability of these results. In terms of assessing muscular strength, Schmitz et al33 have demonstrated the safety and efficacy of using one repetition maximal strength testing for both upper and lower extremities in a weight training intervention for breast cancer survivors with lymphoedema.
Recruiting participants who have lower initial values of the outcome of interest at baseline allows more room for improvement and also maximises the potential that overload is sufficient. This principle was typically applied by excluding individuals already physically active at baseline, or including active individuals who were about to begin chemotherapy because they were likely to experience declines in functional capacity during treatment.11 ,23 ,37 Five studies were unclear about application of the initial values principle. For example, Mock et al26 had a primary aim to reduce fatigue with aerobic exercise during treatment and while they excluded physically active women, they did not recruit a sample with high baseline fatigue levels. The initial values principle is important for establishing realistic expected outcomes for the sample. An individual with low levels of aerobic fitness could see greater improvements in aerobic fitness than more fit persons, but even small improvements in fit persons may be clinically relevant. Furthermore, measuring initial values provides information for developing the most effective exercise prescription. Knowing initial values allows the exercise prescription to be tailored to the individual or exercise group and allows the use of other principles of exercise training, such as overload.
Currently, there is little known about diminishing returns or reversibility in the exercise oncology literature, as the majority of studies have not included follow-up periods. Only one study met our criteria for evaluating reversibility. Daley et al20 conducted follow-up measures 24 weeks after completion of an 8-week aerobic exercise intervention. Physical activity levels decreased from end of study to follow-up and aerobic fitness declined towards baseline. Although Mutrie et al42 also conducted follow-up at 6 months after their supervised group aerobic and resistance intervention, physical activity levels over the follow-up period were not reported, leaving it unclear if the observed decline in submaximal aerobic capacity is due to declines in physical activity. Only one trial met our criteria for including the principle of diminishing returns. Schmitz et al32 used an immediate exercise group (intervention 0–6 months) and delayed exercise group (intervention 6–12 months) and compared both at 12 months. Those from the immediate exercise group who continued to exercise continued to improve, but had less improvement during the follow-up period than they did during the intervention – a prime example of the ceiling effect that can occur with long-term exercise training. Evaluating effects of long-term exercise is an important future area of research so that the lasting benefits of exercise can be established and a realistic time course of a training programme can be mapped.
Reporting of the components of the exercise prescription and adherence to prescribed exercise
Of the 29 papers included in this review, one-third failed to report a minimum of one component of the FITT exercise prescription. Without information on the basic components of the exercise prescription, it is impossible for others to replicate the exercise programme and difficult to evaluate whether or not the intervention was appropriate to achieve the desired outcome. Exercise intensity was the least reported component of the FITT prescription. Intensity is a key component of the exercise prescription, and the appropriate intensity needed to elicit the most therapeutic results is largely unknown. Only when researchers begin to report the prescribed intensity can we begin to determine the minimum intensity that elicits clinical benefits and the maximum intensity beyond which risk of injury is increased and the prescription provides little additional benefit.
Furthermore, reporting of adherence to the prescribed exercise protocol is also of great importance moving forward. Simply stating attendance at exercise sessions does not reveal the exact intensity and duration of exercise that was completed which have implications in relation to observed changes or lack thereof. Reporting sufficient detail about the amount of exercise actually performed by participants enables a more accurate interpretation of study outcomes and, in turn, appropriate translation of the programme in a non-research setting.
What is already known on this topic
Breast cancer is the most commonly diagnosed cancer in women. Exercise has been recommended as a strategy to manage several negative side effects of adjuvant treatment.
What this study adds
This review is the first to critically examine the quality of the exercise prescription reported in exercise intervention trials in breast cancer survivors according to their application of the principles of exercise training, the reporting of the exercise prescription and the reporting of adherence to the exercise prescription.
Recommendations for developing exercise prescriptions for future studies
Based on the findings of this review, we suggest the following recommendations for future exercise intervention trials in breast cancer survivors. When developing the exercise prescription, consider matching the exercise prescription to the outcome of interest (specificity). For example, if muscular endurance is a key outcome, consider a programme of lighter resistance and higher repetitions to achieve this aim. The use of objective baseline testing of the outcome of interest is also recommended to guide the development of an appropriate exercise prescription that will provide a physiological stimulus (overload) and, therefore, will result in training improvements. Building progression into your prescription and outlining how this progression will be achieved should be included in future trials to ensure that the principle of overload is maintained as participants may otherwise experience a plateau in physiological change. Using objective measures of the outcome of interest to determine initial values and taking initial values into account when planning your programme is also recommended and investigators should be ready to customise the programme across a range of baseline fitness levels. If possible, adding a follow-up visit to the research study or community programme will provide information on exercise maintenance in participants, and provide valuable information on the long-term benefits of physical activity. Finally, when reporting the exercise prescription, and adherence to the exercise prescription, include all components of the FITT principle. For adherence, we suggest this type of detail, 'during the intervention participants completed a mean ± SD: 3.5 ± 1.2 days/week, at 70 ± 5% heart rate reserve, for a total of 165 ± 25 min/week of aerobic exercise, which was primarily brisk walking (on treadmill or outside), elliptical machine or cycle ergometer.'
This review examined the application of the principles of exercise training in the exercise prescriptions used in RCTs of exercise in breast cancer survivors. Overall, studies have not adequately designed the exercise prescription in accordance to the principles of exercise training nor adequately reported the components of the exercise prescription or adherence to the exercise prescription. Therefore, the interpretation of the current literature is limited in this respect and information for developing specific exercise prescriptions to reduce side effects of cancer treatment(s) is lacking. By following the principles of exercise training and thoroughly reporting all components of the exercise prescription and adherence, specific evidence-based exercise prescriptions, administered by healthcare professionals, could be available for breast cancer survivors in the near future.
The authors would like to thank reference librarian, Charlotte Beck, for her assistance with setting up the search strategy for this review.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.