Intervenciones de entrenamiento con ejercicios físicos para niños y adultos jóvenes durante y después del tratamiento para el cáncer infantil
Resumen
Antecedentes
Se ha informado de una disminución del estado físico de los pacientes y supervivientes de cáncer infantil. En ello influyen los efectos negativos de la enfermedad y el tratamiento del cáncer infantil. Con frecuencia se ha informado de que el entrenamiento con ejercicios para pacientes adultos con cáncer mejora la condición física. En los últimos años, también se ha publicado literatura sobre este tema para niños y adultos jóvenes con cáncer, tanto durante como después del tratamiento. Esta es una actualización de la revisión original que se realizó en 2011.
Objetivos
Evaluar el efecto de una intervención de entrenamiento con ejercicios físicos en la aptitud física (es decir, la capacidad aeróbica, la fuerza muscular o el rendimiento funcional) de los niños con cáncer dentro de los primeros cinco años a partir de su diagnóstico (realizado durante o después del tratamiento del cáncer), en comparación con un grupo de control de niños con cáncer que no recibieron una intervención de ejercicios.
Determinar si el ejercicio físico durante los primeros cinco años del diagnóstico tiene un efecto sobre la fatiga, la ansiedad, la depresión, la autoeficacia y la CVRS y determinar si hay algún efecto adverso de la intervención.
Métodos de búsqueda
Se realizaron búsquedas en las bases de datos electrónicas del Registro Cochrane de Ensayos Controlados (Cochrane Register of Controlled Trials, CENTRAL), MEDLINE, EMBASE, CINAHL y PEDro; en los registros de ensayos en curso y en las actas de congresos el 6 de septiembre de 2011 y el 11 de noviembre de 2014. Además, se realizó una búsqueda manual de listas de referencia.
Criterios de selección
La revisión incluyó ensayos controlados aleatorizados (ECA) y ensayos clínicos controlados (ECC) que compararon los efectos del entrenamiento con ejercicios físicos con ningún entrenamiento en pacientes que se encontraban en los primeros cinco años del diagnóstico de cáncer infantil.
Obtención y análisis de los datos
Dos autores de la revisión identificaron de forma independiente los estudios que cumplían los criterios de inclusión, realizaron la extracción de datos y evaluaron el riesgo de sesgo mediante formularios estandarizados. La calidad de los estudios se calificó con arreglo a los criterios de la Clasificación de la Evaluación de Recomendaciones, Desarrollo y Evaluación (GRADE).
Resultados principales
Además de los cinco estudios de la revisión original, esta actualización incluyó un ECA adicional. En total, el análisis incluyó 171 participantes, todos durante el tratamiento de la leucemia linfoblástica aguda (LLA) infantil.
La duración de las sesiones de entrenamiento varió de 15 a 60 minutos por sesión. Tanto el tipo de intervención como el período de intervención variaron en todos los estudios incluidos. Sin embargo, el grupo de control siempre recibió la atención habitual.
Todos los estudios tenían limitaciones metodológicas, como un número reducido de participantes, métodos de aleatorización poco claros y diseños de estudio a ciegas en el caso de un ECA, y todos los resultados eran de calidad moderada a muy baja (GRADE).
La aptitud cardiorrespiratoria fue evaluada por la prueba de correr‐caminar de 9 minutos, la prueba de subir y bajar escaleras cronometradas, la prueba de subir y bajar cronometradas, y la prueba de correr en el transbordador de 20 metros. Los datos de la prueba de los 9 minutos de marcha y de la prueba de las escaleras de subida y bajada podrían ser agrupados. Los resultados combinados de la prueba de caminata de 9 minutos mostraron diferencias significativas entre el grupo de intervención y el grupo de control, a favor del grupo de intervención (diferencia de medias estandarizada (DME) 0,69; intervalo de confianza (IC) del 95%: 0,02 a 1,35). Los datos agrupados de la prueba cronometrada de las escaleras no mostraron diferencias significativas en la aptitud cardiorrespiratoria (DME ‐0,54; IC del 95%: ‐1,77 a 0,70). Sin embargo, hubo una considerable heterogeneidad (I2 = 84%) entre los dos estudios sobre este resultado. Los otros dos resultados de un solo estudio, la prueba de la carrera de la lanzadera de 20 m y la prueba de tiempo de subida y bajada, también mostraron resultados positivos en cuanto a la aptitud cardiorrespiratoria a favor del grupo de intervención.
Sólo un estudio evaluó el efecto del ejercicio sobre la densidad mineral ósea (cuerpo entero), mostrando un efecto de intervención positiva estadísticamente significativo (DME 1,07; IC del 95%: 0,48 a 1,66). Los datos agrupados sobre el índice de masa corporal no mostraron ninguna diferencia estadísticamente significativa en la puntuación final entre el grupo de intervención y el de control (DME 0,59; IC del 95%: ‐0,23 a 1,41).
Tres estudios evaluaron la flexibilidad. Dos estudios evaluaron la dorsiflexión del tobillo. Un estudio evaluó la dorsiflexión activa del tobillo, mientras que el otro evaluó la dorsiflexión pasiva del tobillo. No hubo diferencias estadísticamente significativas entre el grupo de intervención y el de control con la prueba de dorsiflexión activa del tobillo; sin embargo, a favor del grupo de intervención, se encontraron para la dorsiflexión pasiva del tobillo (DME 0,69; IC del 95%: 0,12 a 1,25). En el tercer estudio se evaluó la flexibilidad del cuerpo mediante la prueba de la distancia de sentarse y alcanzar, pero no se identificó ninguna diferencia estadísticamente significativa entre el grupo de intervención y el de control.
Tres estudios evaluaron la fuerza muscular (rodilla, tobillo, espalda y pierna, y la fuerza muscular inspiratoria). Sólo la puntuación de la combinación de la fuerza de la espalda y las piernas mostró diferencias estadísticamente significativas en la puntuación final de la fuerza muscular entre el grupo de intervención y el de control (DME 1,41; IC del 95%: 0,71 a 2,11).
Aparte de una subescala de la escala de cáncer (Preocupaciones; valor P = 0,03), ninguna de las escalas de calidad de vida relacionada con la salud mostró una diferencia significativa entre ambos grupos de estudio en el resultado final. En cuanto a los demás resultados de la fatiga, el nivel de actividad diaria y los eventos adversos (todos evaluados en un estudio), no hubo diferencias estadísticamente significativas entre el grupo de intervención y el de control.
Ninguno de los estudios incluidos evaluó el gasto energético en actividades, el tiempo dedicado al ejercicio, la ansiedad y la depresión, o la autoeficacia como resultado.
Conclusiones de los autores
Los efectos de las intervenciones de entrenamiento con ejercicios físicos para los participantes en el cáncer infantil todavía no son convincentes. Las posibles razones son el reducido número de participantes y los diseños insuficientes de los estudios, pero también puede ser que este tipo de intervención no sea tan eficaz como en los pacientes adultos con cáncer. Sin embargo, los primeros resultados muestran algunos efectos positivos en la aptitud física del grupo de intervención en comparación con el grupo de control. Hubo efectos positivos de la intervención en la composición corporal, la flexibilidad, la aptitud cardiorrespiratoria, la fuerza muscular y la calidad de vida relacionada con la salud (elementos relacionados con el cáncer). Estos fueron medidos por algunos métodos de evaluación, pero no todos. Sin embargo, la calidad de la evidencia fue baja y no se encontraron estos efectos positivos en los demás resultados evaluados, como la fatiga, el nivel de actividad diaria y los eventos adversos. Se necesitan más estudios con objetivos e intervenciones comparables, que utilicen un mayor número de participantes y que también incluyan diagnósticos distintos de la LLA.
PICO
Resumen en términos sencillos
Intervenciones de entrenamiento con ejercicios físicos para niños y adultos jóvenes durante y después del tratamiento para el cáncer infantil
Antecedentes
El cáncer infantil es menos común que el cáncer de adultos, con una tasa de 144 a 148 casos por cada millón de niños. A menudo se necesita para la curación un tratamiento intensivo que incluye formas combinadas de tratamiento como cirugía, quimioterapia, radioterapia o una combinación. Estas modalidades de tratamiento suelen ir acompañadas de efectos secundarios, como sensación de malestar (náuseas), infecciones graves, daños en los órganos (corazón, pulmón, riñón, hígado), disminución de la densidad mineral ósea (minerales inferiores, como el calcio, en los huesos, lo que los hace más frágiles), pero también disminución de la fuerza muscular y de la aptitud física.
En el pasado, se aconsejaba a los niños que se recuperaran en la cama y que descansaran lo más posible. Actualmente, se considera que demasiada inmovilidad puede dar lugar a una disminución adicional del estado físico y del funcionamiento físico. Estos efectos secundarios podrían prevenirse o reducirse mediante la introducción de un programa de entrenamiento con ejercicios físicos durante el tratamiento del cáncer infantil o poco después de éste.
Características de los estudios
Se buscaron en las bases de datos científicas estudios de comparación de los efectos del entrenamiento con ejercicios físicos dentro de los primeros cinco años después del diagnóstico de cáncer infantil en comparación con ningún entrenamiento. Los participantes tenían menos de 19 años de edad y padecían cualquier tipo de cáncer infantil. La evidencia está actualizada hasta noviembre de 2014.
Resultados clave
Esta revisión incluyó cinco ensayos controlados aleatorizados (estudios clínicos en los que las personas son asignadas al azar a uno de dos o más grupos de tratamiento) y un ensayo clínico controlado (estudios clínicos en los que las personas son asignadas a uno de dos o más grupos de tratamiento, pero esto no se hace de manera aleatoria) que evaluó los efectos de un programa de entrenamiento de ejercicios físicos en niños durante el tratamiento del cáncer. La leucemia linfoblástica aguda infantil (LLA) es un cáncer de los glóbulos blancos y es el tipo más común de cáncer infantil. Por esa razón, los investigadores se centran a menudo en este tipo de cáncer, ya que proporcionará el mayor número de pacientes en el menor tiempo posible. En total, el análisis incluyó 171 participantes con LLA. Los resultados de la revisión mostraron que había algunos beneficios pequeños del entrenamiento con ejercicios físicos sobre la composición corporal (porcentaje de masa grasa, músculos y huesos), la flexibilidad, la aptitud cardiorrespiratoria (cuán efectivos son el corazón y los pulmones en la entrega de oxígeno al cuerpo), la fuerza muscular y la calidad de vida, pero la evidencia fue limitada. Esto puede estar relacionado con un programa inadecuado para niños con cáncer, o debido a estudios mal diseñados. Se necesitan más estudios que evalúen los efectos del ejercicio en diversas poblaciones con cáncer infantil. Además, los resultados actuales no aportan evidencia suficiente para identificar el programa óptimo de entrenamiento con ejercicios físicos para los niños con cáncer, ni proporcionan información sobre las características de los pacientes que se beneficiarán o no de tal programa. Todavía deben aclararse estos aspectos importantes.
Authors' conclusions
Summary of findings
| Physical exercise training compared to usual care for children and young adults during and after treatment for childhood cancer | |||||
| Patient or population: children and young adults during and after treatment for childhood cancer Settings: hospital and non‐hospital Intervention: physical exercise training Comparison: usual care | |||||
| Outcomes | Illustrative comparative risks* (95% CI) | No of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | ||||
| Usual care group | Exercise group | ||||
| Cardiorespiratory outcomes | |||||
| 9‐minute run‐walk test | The mean 9‐minute run‐walk test in the control group was | The mean difference between the study groups on 9‐minute run‐walk test was 681 feet (95% CI 132 to 1230) in favour of the intervention group | 68 | ⊕⊕⊝⊝ | SMD 0.69 (95% CI 0.02 to 1.35) |
| Timed up‐and‐down stairs test | The mean time used for timed up‐and‐down stairs test in the control groups was 9.0 sec and 8.6 sec in the 2 studies | The mean difference between the study groups on timed up‐and‐down stairs was ‐0.94 sec (95% CI ‐2.94 to 1.06) in favour of the intervention group | 68 | ⊕⊝⊝⊝ | SMD ‐0.54 (95% CI ‐1.77 to 0.70) |
| Timed up‐and‐go time test stopwatch Follow‐up: mean 3 months | The mean timed up‐and‐go time test in the control group was 8.3 sec | The mean timed up‐and‐go time test in the intervention group was 6.6 sec (1.3 SD); showing a mean difference of ‐1.8 (95% CI ‐2.7 to ‐0.8) | 40 (1 study) | ⊕⊕⊝⊝ | SMD ‐1.15 (95% CI ‐1.83 to ‐0.48) |
| Body composition outcomes | |||||
| Bone mineral density (total body) | The control group had a mean SDS on (total body) bone mineral density of ‐1.1 (95% CI ‐0.8 to ‐1.4) | After the 24 months' intervention the mean SDS of the intervention group on total body bone mineral density was 0.3 better than in the control group (‐0.8 SDS (95% CI ‐0.6 to ‐1.2) | 51 | ⊕⊕⊕⊝ | SMD 1.07 (95% CI 0.48 to 1.66) |
| BMI Quetelet Index | The 2 studies reported a mean change in Quetelet index of the control groups of 1.0 and 0.6 in the 2 studies | The intervention group had a mean change in Quetelet index of 1.2 and 0.7 in the 2 studies The mean difference between the study groups on BMI was 0.18 points (95% CI 0.07 to 0.30) in favour of the intervention group | 64 | ⊕⊕⊝⊝ | SMD 0.59 (95% CI ‐0.23 to 1.41) (BMI increased in the intervention group) |
| Muscle endurance/strength outcomes | |||||
| Ankle dorsiflexion strength | The mean ankle dorsiflexion strength in the control group was 0.22 kg (normalized for participant's weight) | The mean ankle dorsiflexion strength in the intervention groups was 0.25 kg; showing a mean difference of 0.03 (95% CI ‐0.04 to 0.1) | 28 | ⊕⊕⊝⊝ | SMD 0.29 (95% CI ‐0.46 to 1.04) |
| Health‐related quality of life | |||||
| Health‐related quality of life | The mean health‐related quality of life in the control group was | The mean health‐related quality of life in the intervention group was | 28 | ⊕⊕⊝⊝ | SMD ‐0.23 (95% CI ‐0.98 to 0.51) |
| Fatigue | |||||
| General fatigue (range: 0‐100) | The mean general fatigue in the control group was | The mean general fatigue in the intervention group was 3.3 points. The mean difference between the study groups was ‐0.15 (95% CI ‐3.2 to 2.9) | 22 | ⊕⊝⊝⊝ | SMD ‐0.04 (95% CI ‐0.88 to 0.8) |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the table. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group. | |||||
| GRADE Working Group grades of evidence | |||||
| 1. Rated down for imprecision (sample size less than threshold rule‐of‐thumb n = 400, Schünemann 2009) and for high risk of bias in the study of Tanir 2013. 2 Rated down for imprecision (small sample size and confidence intervals include both appreciable benefit and appreciable harm, defined by SMD = ‐0.5 and SMD = 0.5, Schünemann 2009), for inconsistency, and for high risk of bias in the study of Tanir 2013. 3 Rated down for imprecision (small sample size). 4 Rated down by 1 level for imprecision (small sample size) and by 1 level for unclear risk of bias for sequence generation and outcome assessor blinding in study of Moyer‐Mileur 2009. 5 Rated down by 2 levels for imprecision (small sample size and confidence intervals include both appreciable benefit and appreciable harm, defined by SMD = ‐0.5 and SMD = 0.5, Schünemann 2009). 6 Rated down by a total of 3 levels for imprecision (small sample size and confidence intervals include both appreciable benefit and appreciable harm, defined by SMD = ‐0.5 and SMD = 0.5, Schünemann 2009), and for high risk of (selection) bias in the study of Yeh 2011. | |||||
Background
Description of the condition
Only a small percentage of the total population experience childhood cancer; approximately 144 to 148 cases per million children (American Cancer Society 2012; Cancer Research UK 2011). However, the impact of childhood cancer is significant. Many studies report a decreased physical fitness (aerobic capacity and muscle strength), in patients and survivors of acute lymphoblastic leukaemia (ALL), which is the most common type of childhood cancer (Aznar 2006; Hartman 2009; Hovi 1993; Marchese 2004; Moyer‐Mileur 2009; San Juan 2008; Warner 1998; Warner 2008; Wright 1998; Wright 2005), and also in children with cancer in general (Arroyave 2008; Cox 2008; De Caro 2006; Hartman 2008; Ness 2005; Ness 2009; Winter 2009). Reduced daily energy expenditure and lower levels of physical activity have been described as the most important cause of this reduced state of physical fitness in children with cancer (Warner 2008). In addition, a considerable number of survivors of childhood cancer experience motor function disability (Geenen 2007; Van Brussel 2006), which is mostly related to negative motor signs, such as insufficient muscle activity or muscle weakness (Hartman 2008; Wright 2005).
Positive effects of exercise training on physical fitness have been reported in studies with adult with cancer (Cramp 2008; Oldervoll 2004; Schmitz 2005; Watson 2004). It is hypothesized that similar results are possible in children with cancer, or survivors of childhood cancer (Moyer‐Mileur 2009).
Description of the intervention
The intervention under consideration was a physical exercise training programme, introduced within the first five years following the diagnosis of childhood cancer. The exercise training should aim to increase physical fitness by aerobic, anaerobic, strength, or mixed fitness training.
How the intervention might work
Cancer and cancer treatment induce lean tissue degeneration and can, therefore, potentially cause abnormalities in the cardiac and skeletal muscle (Schneider 2007). A decline in protein synthesis and protein degeneration by cancer and its treatment can reduce muscle mass. This can result in a decreased oxidative enzyme activity and a decrease in the number of proteins necessary for metabolism (Schneider 2007). People with cancer often experience muscle weakness, a decreased functional capacity, decreased flexibility, reduced mobility, and diminished health‐related quality of life (HRQoL) (Hartman 2008; Schneider 2007). In addition, a decreased psychosocial functioning and HRQoL as a result of cancer has impact on a person's motivational drive and can result in a poorer self perception of one's ability to perform physical activity (Warner 2008; Wright 1998).
Physical activity can prevent or diminish the negative effects of a sedentary life‐style, such as obesity, poor skeletal health, fatigue, and poor mental health, thereby increasing a person's HRQoL. Increasing physical activity is possible by adopting a less inactive life‐style and increasing sports participation. Beneficial effects of physical activity during or shortly after cancer therapy are an increase in muscle mass and plasma volume, improved lung ventilation and lung perfusion, and an increased cardiac reserve.
Dimeo (Dimeo 2001) suggests such beneficial effects of resistance exercise training on muscle mass in patients with cancer who are treated with glucocorticoids, as was seen in adult patients with other diseases treated with glucocorticoids and resistance exercise.
Why it is important to do this review
Despite the positive results of exercise interventions on fatigue and physical fitness in adults with cancer, the evidence for benefits in children with cancer is limited. Studies within the population of children with cancer and survivors of childhood cancer are emerging and the first data have been published. However, the number of participants in the various publications is small and the variety in type of cancer limited, making it difficult to draw more generalized conclusions. In making healthcare management decisions, participants and clinicians must weigh the benefits and drawbacks of supportive care. Pooled data can help in this decision‐making process.
The purpose of this Cochrane review was to summarize the existing literature on the effectiveness of physical exercise training interventions in children with cancer, implemented within the first five years from diagnosis and to provide a best‐evidence synthesis or meta‐analysis of the reported results. This is an update of the original review that was performed in 2011 (Braam 2013).
Objectives
Primary objective
To evaluate the effect of a physical exercise training intervention on the physical fitness (i.e. aerobic capacity, muscle strength, or functional performance) of children with cancer within the first five years from their diagnosis (performed either during or after cancer treatment), compared to a control group of children with cancer who did not receive an exercise intervention.
Secondary objectives
To determine whether physical exercise within the first five years of diagnosis has an effect on fatigue, anxiety, depression, self efficacy, and HRQoL and to determine whether there are any adverse effects of the intervention.
Methods
Criteria for considering studies for this review
Types of studies
We included randomized controlled trials (RCTs) and controlled clinical trials (CCTs) comparing the effects of physical exercise training within the first five years following the diagnosis of childhood cancer with no training.
We included CCTs in the review when the studies included a well‐defined and comparable control group. Factors that were taken into account regarding comparability were: being children with cancer or survivors of childhood cancer, age, sex, and country of origin.
We included cluster‐randomized trials when the intervention and control groups were comparable in each aspect except for the location of cancer treatment and study recruitment.
We included cross‐over trials when the study results were available for each separate intervention period. We then used the data of the first randomization period.
We did not include reviews but used them to check for relevant references. In addition, we excluded observational studies (including case reports, case‐control studies) and surveys from this review.
Types of participants
Study participants were under 19 years of age at diagnosis of any type of childhood cancer. Participants in the physical exercise training programme needed to be no more than five years from diagnosis. We only included studies that also included adults with cancer when the results of the childhood and adult study populations were reported separately.
Types of interventions
Studies that were included compared a physical exercise training intervention for childhood cancer patients or survivors with a control group receiving care as usual. Care as usual was defined as care when needed, but no specific exercise programme or alternative intervention prescribed to increase physical fitness, HRQoL, self perception, or a combination of these, or to decrease adverse events, fatigue, anxiety, depression, or a combination of these in childhood cancer patients.
The physical exercise training interventions that were offered included different types of training or exercise programmes. For instance, muscle strength or stretching exercises; aerobic exercises; or sports such as gymnastics, swimming, running, or bicycling.
The exercise training intervention could have been additional care during therapy or could have been offered after the standard cancer therapy in a form of rehabilitation. The goals of this exercise training intervention were preventing motor disabilities and a decline in physical fitness, or treating motor function problems that developed during childhood cancer therapy.
The exercise training intervention could have taken place in any setting or location: at home, at a physical therapy centre, in a hospital, or elsewhere. It could either have been a group intervention, or an individual programme.
The duration of the exercise training intervention needed to be at least four weeks in order to be able to report on exercise training effects. The upper limit of the training duration was not fixed for this review. In addition, the duration of physical activities (daily time spent on activities or sports) could differ per protocol.
Types of outcome measures
We included studies evaluating the effect of physical exercise training interventions on physical fitness, HRQoL, fatigue, self efficacy, anxiety, and depression. Furthermore, we studied adverse effects of the intervention programme.
Primary outcomes
-
The primary outcome of this review was physical fitness measured by:
-
-
cardiorespiratory fitness (e.g. peak oxygen uptake (VO2peak), peak work rate (Wmax), endurance time): aerobic or anaerobic exercise capacity tested by ergometry on a cycle ergometer or treadmill, the Wingate anaerobic test, the steep‐ramp‐test, maximal anaerobic running/cycling test, the Cooper test, or another valid instrument;
-
muscle endurance/strength: assessed with a hand‐held dynamometer, the Biodex, the spring scale, the lateral step‐up test, the sit‐to‐stand test, 10 repetitions maximum, the up‐and‐down stairs test, the minimum chair height test, the muscle power sprint test, a 10 x 5‐m sprint test, the six‐minute walk test, the incremental shuttle walking test, or another valid instrument;
-
body composition: using body mass index (BMI), skin‐fold measurement, a dual energy x‐ray absorptiometry (DXA) scan, waist circumference, or the waist‐to‐hip‐ratio;
-
flexibility: conducted with a goniometer, flexometer or with the sit‐and‐reach test, V‐sit test, shoulder or trunk rotation test, straight leg raise, the passive and active ankle dorsiflexion test, or another valid instrument;
-
activity energy expenditure: for example, by using an accelerometer;
-
level of daily activity: assessed by an exercise diary, questionnaire, or by accelerometry;
-
time spent exercising (more than daily activity): assessed by an exercise diary, questionnaire, or by accelerometry.
-
Secondary outcomes
-
Secondary outcomes of the review were:
-
-
HRQoL: measured by the Pediatric Quality of Life Inventory (PedsQL), Child Health Questionnaire (CHQ), and DISABKIDS;
-
fatigue: assessed by the PedsQL Multidimensional Fatigue Scale, Childhood Cancer Fatigue Scale (CCFS), or the Fatigue Scale for a child (FS‐C), the same scale for adolescents (FS‐A), and for parents (FS‐P);
-
anxiety and depression: measured by the Childhood Depression Inventory (CDI) and the Center of Epidemiological Studies Depression Scale (CES‐D);
-
self efficacy: assessed using the Confidence Scale, the Self‐Efficacy Questionnaire for Children (SEQ‐C), or the Children's Self‐Efficacy Scale;
-
adverse effects during the study period by collecting information on the occurrence of sport injuries, infections, fractures, heart failure, the recurrence of cancer, and fever.
-
Search methods for identification of studies
Electronic searches
For this review we searched the following electronic databases of the Cochrane Register of Controlled Trials (CENTRAL) (2014, Issue 3), MEDLINE/PubMed (from 1945 to 11 November 2014), EMBASE/Ovid (from 1980 to 11 November 2014), CINAHL (from 1982 to 11 November 2014), and Physiotherapy Evidence Database (PEDro; from 1929 to 11 November 2014) (www.pedro.org.au/).
The search strategies for the different electronic databases (using a combination of controlled vocabulary and text words) are stated in the appendices (Appendix 1; Appendix 2; Appendix 3; Appendix 4; Appendix 5).
Searching other resources
We located information about trials not registered in CENTRAL, MEDLINE, EMBASE, CINAHL, and PEDro, either published or unpublished, by searching the reference lists of relevant articles and reviews. We scanned the conference proceedings of the International Society for Pediatric Oncology (SIOP), the American College of Sports Medicine (ACSM), the International Congress on Physical Activity and Public Health (ICPAPH), and the American Physical Therapy Association (APTA) electronically, or otherwise by handsearching from 2005 to 2014.
We performed a search in the ISRCTN register (www.controlled‐trials.com), and ClinicalTrial.gov database (www.clinicaltrials.gov) for ongoing trials on 11 November 2014. We did not impose language restrictions and will update the searches every two years.
The search included "children", "childhood cancer", "cancer", "exercise training therapy", and "outcome" or any related word combination.
Data collection and analysis
Selection of studies
After employing the search strategy, two review authors (KB, PT) independently identified studies meeting the inclusion criteria. We obtained in full any study that seemed to meet the inclusion criteria on title and abstract, for closer inspection. We noted reasons for exclusion on a separate form. The review authors resolved discrepancies by reaching consensus. In one case, a third party arbitrator (TT) was needed: we required another opinion on the study of de Macedo 2010. This discussion resulted in inclusion of that study because the training corresponded with the described criteria of the protocol.
Data extraction and management
Two review authors (KB, PT) independently performed data extraction using standardized forms. For each study, we collected information on the study design, participant baseline characteristics, settings, sample size, number of participants in each study arm, type of intervention(s), duration of intervention, randomization and blinding procedure, type of control group, type and duration of cancer treatment, and stage of cancer treatment (e.g. during or after treatment), and duration of participant follow‐up.
The extracted outcome measures included: changes in cardiorespiratory fitness, muscle strength/endurance, body composition, body flexibility, daily energy expenditure per time period (e.g. day, week, or month), and changes in the level of daily activity and time spent exercising. In addition, we used a separate form to collect information on psychosocial outcomes such as HRQoL, fatigue, anxiety and depression, and the child's self efficacy. To collect data regarding any other adverse effect of the intervention, we collected all information reported on adverse events during the intervention period in the included studies. We contacted authors of the studies of which only an abstract was available for additional study information.
In the process of data extraction, we reached consensus on all items.
Assessment of risk of bias in included studies
Two review authors (KB, PT) independently assessed the risk of bias in the included RCTs and CCT. This was done according to the following criteria: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessor (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias, such as significant baseline imbalance between study groups in pre‐score or baseline outcome data. We also looked at differential diagnostic activity to observe differences in study protocol for the intervention group and the control group.
For all 'Risk of bias' items of the included studies, we used the definitions as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We included a 'Risk of bias' summary figure. This figure shows whether a study had a high, low, or unclear risk of bias; a green plus symbol corresponds with a low risk of bias, a red minus symbol corresponds with a high risk of bias, and the yellow question mark symbol corresponds with lack of information or uncertainty over the potential for bias.
We resolved discrepancies by discussion so consensus was reached. We rated quality of the outcomes by using the Grading of Recommendation Assessment, Development and Evaluation (GRADE) criteria (Guyatt 2008a; Guyatt 2008b). For purposes of systematic reviews, GRADE defines the quality of a body of evidence ('high', 'moderate', 'low', or 'very low') as the extent to which we can be confident that an estimate of effect or association is close to the quantity of specific interest. The GRADE system entails an assessment of the quality of a body of evidence for each individual outcome (Guyatt 2008b). Factors that may decrease the quality of evidence are: study limitations, inconsistency of results, indirectness of evidence, imprecision, and publication bias. Factors that may increase the quality of evidence are: large magnitude of effect; plausible confounding, which would reduce a demonstrated effect; and dose‐response gradient (Guyatt 2008a). Two review authors (KB, PT) performed the grading of the quality of evidence in consultation with each other. In case of disagreement, they discussed even minor aspects to reach consensus on that matter.
Measures of treatment effect
The main outcome differences between study groups and pooled data are described in the summary of findings Table for the main comparison. In this table, we provided the illustrative comparative risks (with 95% confidence interval (CI)) and differences in standardized mean difference (SMD). For the Cohen's SMD, we took data from the post‐training/control period measurement. The results of the review also include effect estimates of the intervention per outcome measure. Across the included studies, different outcome assessing scales were used. However, in case of BMI, 9‐minute run‐walk test and the timed up‐and‐down stairs test, we were able to combine data of two studies.
For the interpretation of the Cohen's SMD, we used the following criteria (Higgins 2011):
-
less than 0.41 represents a small effect;
-
0.40 to 0.70 represents a moderate effect;
-
greater than 0.70 represents a large effect.
Dealing with missing data
We sought relevant missing data by contacting the primary study author or the corresponding study author. To optimize the strategy for dealing with missing data, we used an intention‐to‐treat (ITT) analysis when possible. The ITT analysis includes all participants who did not receive the assigned intervention according to the protocol as well as those participants who were lost to follow‐up. We investigated attrition rates, for example, drop‐outs and withdrawals, to optimize data analyses.
Assessment of heterogeneity
We assessed heterogeneity both by visual inspection of the forest plots and by a formal statistical test for heterogeneity, that is, the I2 statistic. We defined significant heterogeneity as I2 greater than 50% (Higgins 2011). In case of heterogeneity, we assessed the following potential sources of clinical heterogeneity: participant characteristics, intervention setting; and stratification methods within studies. When we found heterogeneity, we assessed potential reasons for the differences by examining the study characteristics.
Assessment of reporting biases
In the protocol, we had planned to perform a funnel plot; however, due to an insufficient number of studies (fewer than 10) included in this review, we were not able to do so (Higgins 2011).
Data synthesis
We entered the data of the included studies into Review Manager 5 software (RevMan 2011). We performed the analyses according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). By using the GRADE criteria, the quality of the included studies was taken into account when interpreting the results for the review. We used the random‐effects model throughout the review. When we were unable to perform meta‐analysis, we provided all available effect information from the articles.
Subgroup analysis and investigation of heterogeneity
We planned to perform subgroup analyses to evaluate whether the outcome was influenced by differences in the age of the participant, the delivered type of physical exercise training intervention, the duration of the exercise training intervention, the exercise training intervention location, type of childhood cancer, and cancer treatment.
On three review outcomes, a meta‐analysis could be performed; that is, on 9‐minute run‐walk test, the timed up‐and‐down stairs test, and BMI. Unfortunately, apart from the intervention and control groups, 9‐minute run‐walk test, the timed up‐and‐down stairs test, and BMI data were not available for other subgroup characteristics (Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013). Therefore, we could not perform any specific subgroup analyses.
Sensitivity analysis
For those studies that assessed similar outcomes and of which data could be pooled, we performed sensitivity analyses. We assessed whether the outcome would have been different when a study with high or unclear risk of bias would have been excluded from the analyses. This method aimed to assess whether the findings were robust to the decisions made in the process of obtaining them.
Results
Description of studies
Results of the search
Original review in 2011
For the original version of the review (Braam 2013), the electronic database searches in CENTRAL, MEDLINE, EMBASE, CINAHL, and PEDro, searches in ongoing trial registries and abstract books from SIOP, ACSM, ICPAPH, and APTA revealed 743 references in 2011.
After removal of duplicates, the search in 2011 resulted in 710 potentially relevant articles. Initial screening of titles and abstracts excluded a further 700 references that did not meet the criteria for inclusion. The 10 remaining references were read in full text. Two out of these 10 studies were ongoing trials, four studies did not meet all eligibility criteria and were thus excluded, and four studies were included.
Reference list tracking led to two additional articles that potentially could be included. One fulfilled the inclusion criteria and was included in the review. Based on the available information in the congress proceeding of the second study, it was not possible to decide if the second study was eligible for inclusion (Elkateb 2007). This study was moved to Characteristics of studies awaiting classification (see Figure 1).
Study flow diagram.
Update in 2014
Running the searches for the update in the aforementioned electronic databases, and searching the abstract books from SIOP, ACSM, ICPAPH, and APTA in November 2014 revealed an additional 704 references.
After removal of duplicates, 643 potentially relevant articles remained. Initial screening of titles and abstracts excluded an additional 620 references that did not meet the inclusion criteria. The remaining 23 references were read in full text. Four of these 23 studies were ongoing trials (see Characteristics of ongoing studies table), 15 did not meet all eligibility criteria, and three out of the 23 studies were congress proceedings (Braam 2014; Sabel 2013; Senn‐Malashonak 2014; see Characteristics of studies awaiting classification table). Braam 2014 combined two congress proceedings from the same study and for that reason were taken together in this review. Only one out of these 23 studies could be included in the update of the original review.
Reference list and trial registry tracking led to four additional articles. Two of these four studies used a study design that did not match the inclusion criteria of this review. For further information, see Excluded studies and Characteristics of excluded studies table (Ruiz 2010; Shore 1999). The other two studies were trial protocols that were moved to the ongoing trial section (Characteristics of ongoing studies table). (See Figure 1).
One of the ongoing trials in the original review of 2011 (that was referred to as 'Braam 2010') reported results in a congress proceeding and, therefore, changed from an ongoing trial to an awaiting classification study and is now referred to as Braam 2014.
In summary, from the original review and the update together we included six studies in this review.
Included studies
Methods
The six studies included in this review were: de Macedo 2010; Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011. Five of these studies were RCTs, and one study used a quasi‐experimental study design, making it a CCT (Yeh 2011). One study performed a power calculation (Hartman 2009). For trial characteristics and outcomes, see the Characteristics of included studies table.
Participants
In total, the analysis included 171 participants. All participants were diagnosed with childhood ALL and studied during chemotherapy (de Macedo 2010; Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011). Of the 171 children, 98 were boys, 70 were girls (de Macedo 2010; Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011); gender was not reported in three children who dropped out. The number of children per study was small. Hartman 2009 included the largest number of children (51 children) in their study, with 26 children in the usual care group and 25 in the intervention group. The 14 children in the study of de Macedo 2010 were divided into nine children who received care as usual and five who received the intervention. Marchese 2004 included 13 children that performed the exercise intervention and 15 who had care as usual. The 13 children analyzed in the study of Moyer‐Mileur 2009 were divided into seven who received care as usual and sic who received the intervention; one child was lost to follow‐up. Tanir 2013 included 41 children, of which one dropped out, resulting in a group distribution of 19 children in the intervention group versus 21 children in the control group. Yeh 2011 included 22 children in the analyses, of which 12 children received the intervention training programme and 10 received care as usual; two children were not taken into analysis because they were lost to follow‐up.
Five studies reported their exclusion criteria; in one study, these data were missing (Moyer‐Mileur 2009). Four studies had cognitive impairment with or without or mental (developmental) impairment an exclusion criterion (Hartman 2009; Marchese 2004; Tanir 2013; Yeh 2011). One study described having difficulties with the national language (Hartman 2009). Four studies excluded children with neurological impairment (de Macedo 2010; Marchese 2004; Tanir 2013; Yeh 2011). Marchese 2004 and Tanir 2013 excluded children with a genetic disorder, as well as children who had received cancer‐related physiotherapy, or children who had participated in a regular exercise programme in the six months before start of the study. Tanir 2013 excluded children with cardiac, pulmonary, renal, or hepatic dysfunction, whereas de Macedo 2010 excluded children with chronic lung disease, neuromuscular disease, or children treated with radiotherapy.
Intervention
The exercise intervention programme of all six studies included at least a home‐based exercise programme with guidance from a therapist of the treating hospital to optimize physical fitness (Hartman 2009; de Macedo 2010; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011). However, the duration of the entire intervention, duration of each training session, timing, and type of interventions differed across studies. The duration of the training sessions ranged from 15 minutes to 60 minutes. The intervention period ranged from 10 weeks (de Macedo 2010; Yeh 2011) to two years (Hartman 2009). Five out of six studies introduced the exercise intervention during the maintenance treatment period (de Macedo 2010; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011), and in one study it started shortly after diagnosis (Hartman 2009). Five studies determined the effects of an exercise intervention to increase muscle strength of all muscles (Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011). The study of de Macedo 2010 investigated the effect of an inspiratory muscle training programme. They studied the effects of inspiratory muscle training, which was performed with a threshold device using a load of 30% of the maximal inspiratory pressure.
For more details, see the information in the Characteristics of included studies table.
Control
The control groups of the six studies received care as usual, which implies no additional exercise‐related care (de Macedo 2010; Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011). We consider that Tanir 2013 probably made a writing mistake as they reported in the same paper that the control group did and did not receive an exercise intervention. Based on the additional information in the paper, we have now concluded that the control group did not receive an exercise intervention.
With the exception of those of the study of de Macedo 2010, all study participants of the control groups were measured at the same time points as the intervention group. The control group in the study of de Macedo 2010 performed the study assessments during the initial evaluation and after 10 weeks, whereas the intervention group performed the measurements at the end of each training week.
Outcomes
The studied primary outcomes were: cardiorespiratory fitness, muscle endurance/strength, body composition, flexibility, and level of daily activity. Secondary outcomes of this review that were mentioned in the studies were: HRQoL, fatigue, and adverse events. The studies did not address the other secondary outcomes (anxiety, depression, and self efficacy).
Because of the different aims and study methods of the six included studies, there was little overlap in used methods and assessed outcomes. Only two studies performed both the 9‐minute run‐walk test and the timed up‐and‐down stairs test to assess cardiorespiratory fitness (Marchese 2004; Tanir 2013), and, another two studies assessed changes in BMI, as part of changes in body composition (Hartman 2009; Moyer‐Mileur 2009). For further information, see the Characteristics of included studies table and the Data and analyses tables.
Excluded studies
We subsequently excluded 21 publications that had been retrieved. There were four studies that included an adult cancer population instead of a paediatric population (Jarden 2013; Oldervoll 2011; Rief 2011; Steel 2011). We excluded six studies based on the used design; one was a case‐control study (Rosenhagen 2011), one used healthy volunteers as a control population (Shore 1999), one used a cross‐over randomized trial design without presenting data after the first intervention period (before cross‐over) (Speyer 2010), and three were uncontrolled studies (Gohar 2011; Jarden 2013; Ruiz 2010). In another three studies the intervention did not match with the intervention of interest for this review (Geyer 2011; Huang 2014; Kurt 2011), and in one study the aim was to increase motor and process function; this outcome did not correspond with any of the primary or secondary outcomes of this review (Emanuelson 2014).
Another eight studies assessed the effects of a training intervention with duration of less than four weeks (Chamorro‐Vina 2010; Chung 2014; Herbinet 2014; Hinds 2007; Speyer 2010; Speyer 2011; William Li 2013; Winter 2013). Furthermore, there was duplication of information; within these eight excluded studies (on intervention duration), two studies were described in multiple reports: the first study was reported in Chung 2014 and William Li 2013, the second study results were presented in Speyer 2010 and Speyer 2011. The final excluded study was a conference proceeding, presenting data of a pilot study (te Winkel 2008). The full study data were presented separately by Hartman 2009, a study that is included in the review.
The exclusion of two studies was based on two exclusion criteria and therefore mentioned twice in the section above (Jarden 2013; Speyer 2010). Information concerning the excluded studies can be found in the Characteristics of excluded studies table.
Risk of bias in included studies
See the 'Risk of bias' section of the Characteristics of included studies table for the exact scores per study and the support for the judgements made.
Allocation
Two out of the six studies generated random sequence generation adequately (Figure 2; Hartman 2009; Marchese 2004). These two studies used block randomization with sealed envelopes (Hartman 2009; Marchese 2004). Both de Macedo 2010 and Tanir 2013 reported that selection and allocation were random; however, it remained unclear how the randomization procedure was carried out in both studies. A non‐randomised design was used in the study of Yeh 2011, leading to a high risk of selection bias. No information on random sequence generation was available for the fifth study (Moyer‐Mileur 2009). None of the studies described the quality of the envelopes, how the envelopes were sealed, or whether they were coded. Therefore, we judged five out of six studies to have an unclear risk of bias for allocation concealment (de Macedo 2010; Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013). One study did not use a randomization method and, therefore, had no allocation concealment (Yeh 2011). In summary, five studies had an unclear risk of selection bias and, due to the absence of a randomization procedure, one study had a high risk of selection bias.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Blinding
Blinding of participants and personnel (performance bias)
Due to the nature of the interventions, blinding was virtually impossible: that is, when the participants needed to perform an exercise intervention and the children and their parents were well informed about the study purpose, participants could not be blinded for the study randomization. This could be a potential performance bias in all studies (Higgins 2011). Therefore, we considered all included studies of this review to have a high risk for performance bias.
Blinding of outcome assessors (detection bias)
It is possible to minimize detection bias with blinding the outcome assessor for the randomization. Two studies used outcome assessors who were blinded for study groups (Figure 2; Hartman 2009; Marchese 2004). In the other four studies, the risk was unclear.
Incomplete outcome data
All studies reported withdrawals and drop‐outs during the intervention period. However, only one study used an ITT analysis to deal with missing data and thus had a low risk of attrition bias (Yeh 2011).
In the study of Marchese 2004, the authors reported missing data for daily logs of activity and heart monitor. Yet, no information was reported on methods used for data imputation. For two other studies, it also remained unclear whether they used a method for missing data imputation (de Macedo 2010; Moyer‐Mileur 2009). Therefore, in these three studies, the risk of attrition bias was unclear.
In the study of Hartman 2009, there was a high risk of attrition bias. The authors used a simple imputation technic to include data for those children who dropped out the study. Yet, they included the data from prior to the elimination. This method is very simple and, therefore, increases the risk for bias due to incomplete outcome data.
The study of Tanir 2013 provided no information about missing data. The authors reported that one child died in the intervention group. However, they did not present data on this child. Therefore, in this study the risk of attrition bias was high.
Selective reporting
In two studies, serious selective reporting was detected (Tanir 2013; Yeh 2011). In the first study, results on general quality of life was reported as difference between sexes, instead of difference between the two study groups (Tanir 2013). In the second study, 'adherence' was mentioned to be an extra, or a secondary outcome (Yeh 2011). Yet, in the results, the authors focused on this item as if it was a primary outcome. In the four other studies, the risk of reporting bias was low.
Other potential sources of bias
This review assessed two other biases: baseline outcome data and diagnostic activity.
First, three studies reported the absence of significant differences in baseline outcome data (de Macedo 2010; Hartman 2009; Moyer‐Mileur 2009). One study reported baseline differences on sex, treatment anxiety, and goniometer results were reported (Tanir 2013). In the final two studies, it remained unclear whether all baseline test scores were significantly different between the two study groups (Marchese 2004; Yeh 2011).
Second, the study outcomes were measured at different time points for the intervention and the control group in the study of de Macedo 2010. In the control group, the outcomes were assessed during the initial evaluation and after 10 weeks, whereas in the intervention group measurements were performed at the end of each training week. This could have led to differential diagnostic activity. We judged this study to be of high risk for this other type of bias. The other studies used the same number of measurements, and they were free of 'differential diagnostic activity' (Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011).
In summary, two out of the six studies showed unclear the risk of these 'other biases' (Marchese 2004; Yeh 2011), in another two studies the risk was considered high (de Macedo 2010; Tanir 2013), and in the last two studies the 'other bias' risk was low (Hartman 2009; Moyer‐Mileur 2009).
Effects of interventions
Because of the different aims and study methods of the six included studies there was little to no overlap in assessed outcomes. We could only pool three outcomes: the timed up‐and‐down stairs test, the 9‐minute run‐walk test (cardiorespiratory fitness), and BMI (body composition).
Cardiorespiratory fitness
In this review, we defined cardiorespiratory fitness as: VO2 peak, Wmax, or endurance time. The included studies assessed physical fitness by the 9‐minute run‐walk test (Marchese 2004; Tanir 2013), timed up‐and‐down stairs test (Marchese 2004; Tanir 2013), timed up‐and‐go test (Tanir 2013), and 20‐m shuttle run test (Moyer‐Mileur 2009).
The 9‐minute run‐walk test showed a significant effect in favour of the intervention (32 children) compared to usual care (36 children) (SMD 0.69 feet; 95% CI 0.02 to 1.35; P value = 0.04). Analysis showed moderate heterogeneity for this item between the studies (I2 = 44%) (Analysis 1.1).
Two studies assessed the timed up‐and‐down stairs test (Marchese 2004; Tanir 2013). The test results were not significantly different between the intervention (33 children) and the control group (36 children) (SMD ‐0.54 seconds; 95% CI ‐1.77 to 0.70; P value = 0.40). There was considerable heterogeneity for this test between the studies (I2 = 84%) (Analysis 1.2). No ITT analysis could be performed due to missing information on two children who dropped‐out the studies.
The timed up‐and‐go test showed a significant positive intervention effect (SMD ‐1.15 seconds; 95% CI ‐1.83 to ‐0.48) (Tanir 2013); children of the intervention group were faster in performing the test.
The 20‐m shuttle run test results showed that children who performed home‐based exercises during their maintenance chemotherapy for ALL (six children) were able to reach higher end‐scores than those in the control group (seven children) (P value = 0.05) (no Review Manager data available). ITT analysis was not performed (Moyer‐Mileur 2009).
Body composition
BMD (Hartman 2009) and BMI (Hartman 2009; Moyer‐Mileur 2009) were assessed as part of the outcome body composition.
The study of Hartman 2009 used a DXA scan to determine BMD (lumbar spine and total body) changes in children with childhood ALL. The assessments were performed at diagnosis, during chemotherapy, and one year after the end of treatment. Analysis showed a significant SMD of 1.07 for total body BMD (95% CI 0.48 to 1.66; P value < 0.001) after the intervention of 24 months (Analysis 2.1). These results revealed a large and significant positive intervention effect on the total body BMD for the intervention group (25 children) compared to the control group (26 children). This analysis was performed according to the principles of the ITT analysis.
Two trials studied the differences in BMI between the intervention group and the control group (Hartman 2009; Moyer‐Mileur 2009). The study of Moyer‐Mileur 2009 found no intervention effect on BMI (SMD 0.02; 95% CI ‐1.07 to 1.11). This study compared six children who received a combined nutrition and exercise programme, with seven children who received usual care. Data of one child that dropped out were not reported and we, therefore, could not perform an ITT analysis on BMI. The study of Hartman 2009 showed a statistically significant difference on BMI in favour of the exercise group (25 children) compared to the control group (26 children) (SMD 0.90; 95% CI 0.32 to 1.48). These BMI analyses were performed according to ITT analysis principles (Hartman 2009). Pooled data analysis for BMI showed a non‐significant intervention effect with an SMD of 0.59 on the Quetelet Index (95% CI ‐0.23 to 1.41; P value = 0.16) (Analysis 2.2) in favour of the intervention group. In addition, analysis also showed no substantial heterogeneity (I2 = 48%) for this item between the studies (Analysis 2.2).
Flexibility
Two studies measured the ankle dorsiflexion range of motion. However, in one study this was done in a passive way (Hartman 2009), and in the other by active contraction of the ankle (Marchese 2004). Therefore, we could not pool data.
According to the ITT analysis shown in Analysis 3.1, the passive ankle dorsiflexion showed a moderate significant positive effect for the 25 children in the intervention group compared to the 26 children in the control group (SMD 0.69; 95% CI 0.12 to 1.25; P value = 0.02) (Hartman 2009). Analysis of the ankle dorsiflexion range of motion, measured in active contraction, showed a non‐significant moderate effect in the intervention group (13 children) compared to the control group (15 children) (SMD 0.46; 95% CI ‐0.29 to 1.22; P value = 0.23) (Analysis 3.1) (Marchese 2004). Because Marchese 2004 only provided the data of the children who completed all measurements, we performed no ITT analysis.
The study of Moyer‐Mileur 2009 assessed body flexibility with the sit‐and‐reach distance test. In this study, there was no difference in the test results between the six children of the intervention and seven children of the control group. P values and ITT analysis were not stated in the text or provided by the authors.
The study of Tanir 2013 used the goniometer to assess the range of motion. Unfortunately, the baseline scores were significantly different, which may have affected the outcomes. This study showed statistically significant differences between the study groups at the end‐measurement with higher scores in the control group, but no significant increase of the goniometer results over time within the study groups. However, the authors did not report the assessment position of the goniometer on the body. Therefore, it was not clear whether a decrease in the goniometry results over time was a positive or a negative study result.
Muscle endurance/strength
Marchese 2004 assessed the knee and ankle strength changes and Tanir 2013 assessed back and leg strength changes by hand‐held dynamometry. In both studies, the authors found a significant effect in favour of the intervention group. Analysis showed that differences between the end scores of the intervention group and the control group were not significantly different for both knee and ankle strength (Analysis 4.1; Analysis 4.2), but that differences were significant for back and leg strength (SMD 1.41; 95% CI 0.71 to 2.11; P value < 0.001) (Analysis 4.3) (Tanir 2013). The SMD of the knee strength was 0.25 (95% CI ‐0.49 to 1.00; P value = 0.51) and the increase of ankle strength was 0.29 (95% CI ‐0.46 to 1.04; P value = 0.44) (Marchese 2004).
The study of Moyer‐Mileur 2009 reported differences in the number of completed push‐ups (with knees on the ground) and used a peripheral quantitative computed tomography of the tibia to determine the muscle mass of the participants. According to the original study data, there was no significant change in the maximum number of push‐ups or muscle mass, within or between the intervention (six children) and control group (seven children). The report of this study did not include the data of these results; therefore, we could not perform a Review Manager analysis.
de Macedo 2010 assessed respiratory muscle strength by measuring the maximal inspiratory pressure and maximal expiratory pressure with a digital manometer and a nozzle to dissipate additional pressure caused by the facial muscles and the oropharynx. In the intervention group (five children), the authors found a significant improvement over time compared to the control group (nine children). However, the end score differences were not significant different between the study groups; SMD for inspiratory breathing muscle strength was 0.33 (95% CI ‐0.77 to 1.43; P value = 0.56) and for expiratory breathing muscle strength the SMD was 0.00 (95% CI ‐1.09 to 1.09; P value = 1.00) (Analysis 4.4; Analysis 4.5).
Due to invalid methods used for missing data imputation, we could not perform an ITT analysis for these outcomes.
Activity energy expenditure
The included studies did not assess activity energy expenditure.
Level of daily activity
One study assessed daily physical activity (Moyer‐Mileur 2009). This study used both the pedometer steps‐per‐day and an activity questionnaire to examine physical activity behaviour. This study showed that the increase in "reported activity in minutes per day" over time was approximately the same for the six children in the intervention group. In the control group, three out of seven children increased in their reported activity in minutes per day. According to the original analyses, the reported activities at baseline and at six months were not statistically significantly different between the intervention group and the control group (Moyer‐Mileur 2009). At 12 months from baseline, there was a higher number of steps recorded in the intervention group compared with the controls, but this difference was of borderline statistical significance (P value = 0.06) (no Review Manager data available) (Moyer‐Mileur 2009). This analysis was not performed according to the ITT procedure.
Time spent exercising (more than daily activity)
The included studies did not assess time spent exercising (more than daily activity).
Health‐related quality of life
One study assessed general HRQoL using the PedsQL Generic Core Scale (version 3.0) (Marchese 2004).
Marchese 2004 found no significant effect on quality of life by the physical exercise training intervention. Overall, the SMD for PedsQL Generic was ‐0.23 points (95% CI ‐0.98 to 0.51; P value = 0.54) (Analysis 5.1). In addition to the participant‐reported data, parent reports also showed no intervention effect: the SMD on the parent general PedsQL questionnaire was 0.38 points (95% CI ‐0.37 to 1.13; P value = 0.32) (Analysis 5.3).
Two studies assessed cancer‐related HRQoL using the PedsQL Cancer Module 3.0 (Marchese 2004; Tanir 2013). However, Tanir 2013 did not report on the total score of the PedsQL Cancer Module, as Marchese 2004 did. On the contrary, Tanir 2013 reported the results of the eight different sub‐scales, which was not available in the study of Marchese 2004. Without the raw data of the questionnaires it was not possible to count the overall sum score from the sub‐scales. For that reason, we could not pool the data on HRQoL (PedsQL Cancer Module). Based on the sub‐scale data, Tanir 2013 found an increase on the HRQoL (cancer‐related items) in both groups by participant‐report on: pain and hurt, nausea, and procedural anxiety scales; without significant differences on the end‐scores. The only significant different end score (P value = 0.03) between the intervention (19 children) and control group (21 children) was found for the sub‐scale assessing worries (in favour of the intervention group) (Tanir 2013).
According to the sum score as assessed in the study of Marchese 2004, the SMD on the PedsQL Cancer Module was 0.16 (95% CI ‐0.58 to 0.91; P value = 0.66) (Analysis 5.2). The parent‐reported intervention effect on cancer‐related HRQoL again showed no intervention effect: the SMD on the parent cancer PedsQL was 0.04 points (95% CI ‐0.70 to 0.79; P value = 0.91) (Analysis 5.4) (Marchese 2004).
Due to missing data, we could not conduct an ITT analysis.
Fatigue
Yeh 2011 measured the effect of a physical exercise training intervention on fatigue. This study used the PedsQL Multidimensional Fatigue Scale. They compared changes on fatigue between the intervention group (12 children) and the control group (10 children) over eight time points within 10 weeks. There were no significant differences between the intervention and control groups on the sub‐scale general fatigue (SMD ‐0.04; 95% CI ‐0.88 to 0.80; P value = 0.92) (Analysis 6.1). More specifically, there was no intervention effect for sleep and rest (SMD ‐0.01; 95% CI ‐0.85 to 0.83; P value = 0.98) (Analysis 6.2), or for cognitive fatigue (SMD 0.07; 95% CI ‐0.77 to 0.91; P value = 0.86) (Analysis 6.3). Apart from a per‐protocol analysis, the study of Yeh 2011 included an ITT analysis. The ITT analysis revealed no significant intervention effects on fatigue.
Anxiety and depression
The included studies did not assess anxiety and depression.
Self efficacy
The included studies did not assess self efficacy.
Adverse events (due to, or not clearly related to, the intervention)
The study of Marchese 2004 reported that none of the children experienced any negative effects from the exercises or experienced complications attributed to the physical programme. The other studies did not report on adverse events (de Macedo 2010; Hartman 2009; Moyer‐Mileur 2009; Tanir 2013; Yeh 2011).
Sensitivity analysis
We performed sensitivity analyses for those outcomes for which pooling was possible (i.e. 9‐minute walk‐run test, timed up‐and‐down stairs test, and BMI) (Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Tanir 2013). We assessed whether the outcome would have been different when a study with high or unclear risk was excluded in the review analyses.
Two studies performed both the 9‐minute walk‐run test and the timed up‐and‐down stairs test (Marchese 2004; Tanir 2013). In these studies, there were three bias items: random sequence generation (selection bias), blinding of outcome assessment (detection bias), and selective reporting (reporting bias), in which Tanir 2013 showed high or unclear bias compared to low bias in the study of Marchese 2004. For these three items, we performed sensitivity analyses. For all other risk of bias items, the two studies scored the same (i.e. high or unclear risk) or performed a combination of high and unclear risk.
The outcome of the sensitivity analysis for the 9‐minute walk‐run test of Marchese 2004 without the data of Tanir 2013 showed an SMD of 0.33 (95% CI ‐0.42 to 1.07) whereas the results including Tanir 2013 showed a significant intervention effect with an SMD of 0.69 (95% CI 0.02 to 1.35). This analysis showed the analyses were consistent among the trials.
A sensitivity analysis was performed for the timed‐up‐and‐down‐stairs test. When we assessed data of the study by Marchese 2004 without the data of Tanir 2013, data showed a non‐significant SMD of 0.11 (95% CI ‐0.64 to 0.85). There were comparable results when we included the data of both studies (SMD ‐0.54; 95% CI ‐1.77 to 0.70). Therefore, the results of the trials were consistent among the trials.
Two studies assessed BMI (Hartman 2009; Moyer‐Mileur 2009). In these studies, there were two bias items: random sequence generation (selection bias) and blinding of outcome assessors (detection bias) on which Moyer‐Mileur 2009 showed unclear bias compared to the study of Hartman 2009. For these items, sensitivity analyses were possible. For all other risk of bias items, the two studies scored the same (i.e. low, high, or unclear risk) or performed a combination of high and unclear risk.
The outcome of the sensitivity analysis showed the BMI data of Hartman 2009 without Moyer‐Mileur 2009 (SMD 0.90; 95% CI 0.32 to 1.48). The results of the pooled data revealed an SMD of 0.59 (95% CI ‐0.23 to 1.41). Thus, the results of the sensitivity analyses were consistent among the trials and did not differ from the overall analyses.
Discussion
Summary of main results
Several studies have investigated the effects of physical exercise training interventions on physical fitness in adults with cancer, showing different benefits. Limited studies investigated the effects of such an intervention in a childhood cancer population. In particular, high‐quality studies with an RCT or CCT design are still lacking in this field of research.
This is an update of the original review that was performed in 2011 (Braam 2013). This updated review include six original studies. All studies investigated the effects of a physical exercise training intervention, with a duration of at least four weeks, in children with cancer. They all aimed to improve physical functioning or psychosocial well‐being, and had enrolled children with ALL. However, the studies had two important limitations. First, the total number of participants included in the six studies was limited, and second, the exercise programmes were not always appropriately designed to meet the study goals.
Cardiorespiratory fitness was studied using the 9‐minute run‐walk test, timed up‐and‐down stairs test, timed up‐and‐go time test, and 20‐m shuttle run test. All but the timed up‐and‐down stairs test showed significant positive intervention effects (P value < 0.05).
Bone mineral density increased significantly higher after a physical exercise training intervention when compared with the study control group. Two studies assessed BMI. One study found a significant intervention effect on BMI. However, these results were not found when the data were pooled with the second study.
Four studies assessed flexibility and each study used different test methods. There were no (statistically significant) differences between the study groups in three studies, whereas in the fourth study, there was a statistically significant difference in favour of the exercise group.
One study assessed back and leg strength (muscle strength) and showed a significant intervention effect. The other three studies assessing muscle strength could not report statistically significant intervention effect. There was no statistically significant effect on knee or ankle muscle strength, which were assessed in two studies, or on lung muscle strength (maximal inspiratory and expiratory pressure), which was the primary outcome of the fourth study.
HRQoL assessed by the PedsQL Cancer Module showed some positive effects in the intervention group in comparison to the control group in one study (Tanir 2013). There were no statistically significant differences between the study groups for the level of daily activity and fatigue. In addition, only one study reported no complications attributed to the physical exercise intervention programme, whereas the other studies did not address this item.
None of the six included studies evaluated the outcomes of activity energy expenditure, time spent exercising, anxiety and depression, or self efficacy.
It should be noted that the exercise interventions were not the same and the quality and quantity of the evidence was limited.
For future research, it is advised to assess the effects of one type of exercise intervention in a larger group of children with cancer, preferably in children with ALL as well as other childhood cancer diagnoses. This can be done in well‐designed studies with large sample sizes.
Overall completeness and applicability of evidence
This review provides evidence of modest positive effects of physical exercise training interventions for children with cancer. These modest effects could be due to small sample sizes, various types of interventions provided, and different outcome measures that were used in the six studies. As a result, we could only pool data for 9‐minute walk‐run test, the timed up‐and‐down stairs test, and BMI; therefore, the results of the analysis were instable and weak. However, the meta‐analysis and sensitivity analysis on these three outcomes showed consistent results. Furthermore, the patient population was unintentionally homogeneous since all of the included children had ALL. Therefore, the results of this review are not applicable for other types of childhood cancer.
The Review Manager analyses results of this review were very different to the analyses performed by the authors of some of the studies, which led to different conclusions. For de Macedo 2010, Hartman 2009, and Marchese 2004, the differences were due to different methods of analysis. In this review, we assessed the final outcome differences between the study groups (Analysis 4.1; Analysis 4.4; Analysis 4.5), and did not assess changes over time.
The included studies all had supervised interventions with a duration and intensity in which it was possible to have a physiological response (de Macedo 2010; Hartman 2009; Marchese 2004; Moyer‐Mileur 2009; Yeh 2011). From literature, it is known that supervised exercise interventions in children are more effective compared to non‐supervised programmes (Faigenbaum 2010). It is also known that a well‐designed exercise programme consists of four parameters: mode (type of exercise), intensity, frequency, and duration (ACSM 2010; Ganley 2011). It would be advisable for new studies to first determine if the planned programme includes all elements of these parameters. This will increase the quality of the trials and also increase the comparability.
Appropriate statistical methods are important. The use of incorrect statistical methods can diminish the likelihood of demonstrating the real effects, also in high‐quality interventions. In this review, only one of the included studies used a power calculation (Hartman 2009). In the included studies, the authors assessed baseline (pre‐score) differences between the study groups using the Chi2 test or the Mann‐Whitney U test (Hartman 2009; Moyer‐Mileur 2009; Tanir 2013), the Kruskal‐Wallis test (Moyer‐Mileur 2009; Tanir 2013), and the paired sample T‐test (de Macedo 2010; Tanir 2013). The baseline scores were reported as group mean (de Macedo 2010; Hartman 2009; Marchese 2004; Tanir 2013; Yeh 2011), but also per study participant (Moyer‐Mileur 2009). These baseline differences might have had a large impact on the results and conclusions of this review. It would have been preferable for all authors to have corrected for baseline differences in their analyses. However, this was not done. To increase the quality of evidence of this review, we hoped to be able to pool all raw data (baseline and end of study data) in one database. This would have given us the possibility to correct for these differences. However, not all researchers responded to our request for additional information.
To investigate changes between participants and changes over time the included studies used the paired sample T‐tests (de Macedo 2010; Hartman 2009; Tanir 2013), Friedman two‐way test (Moyer‐Mileur 2009), mixed‐effects model (Yeh 2011), and repeated measure analyses (Hartman 2009; Marchese 2004). The mixed‐effect model and repeated measure analyses are more specific than comparing group mean changes. Therefore, the results of the studies using the better statistical methods are possibly better than the ones using simple statistical techniques. However, in this review, we were unable to use this information in the outcome.
Quality of the evidence
By grading the evidence according to the GRADE criteria (Guyatt 2008b), the overall quality of the individual outcomes varied between moderate and very low. Due to risk of bias, inconsistency, indirectness, imprecision, possible publication bias, or a combination of these, the qualities of the review outcomes were downgraded. None of the individual outcomes were eligible for upgrading. The quality of the evidence is summarized in summary of findings Table for the main comparison. The small number of participants in the trials was the main reason for the low‐quality scores. This is often the case in studies in a paediatric population, and in case of newly introduced interventions. More and larger well‐controlled studies are needed to improve the quality and quantity of evidence. This also emphasizes the need for a core‐set of outcome measures in exercise‐related research in childhood chronic conditions (Van Brussel 2011).
Between the six studies, there was a considerable degree of heterogeneity on mode and intensity of the exercise interventions. When assessing heterogeneity, the 9‐minute walk‐run test and BMI showed no substantial heterogeneity between the two trials (I2 = 44% for 9‐minute walk‐run test and I2 = 48% for BMI). However, the timed up‐and‐down stairs test showed substantial heterogeneity (I2 = 84%).
Potential biases in the review process
The Cochrane Childhood Cancer Group formulated the search strategies for CENTRAL, MEDLINE/PubMed, and EMBASE/Ovid. In addition, we searched two other databases using of a search strategy that we developed ourselves: CINAHL and PEDro. The PEDro database was difficult to search. Therefore, it is possible that we missed one or two studies from this database. However, due to the great overlap between results of the different databases, it is very unlikely that studies were missed.
Agreements and disagreements with other studies or reviews
In 2010, Winter 2010 published a review on childhood cancer and physical activity. This review included 28 studies, and almost half had an uncontrolled study design. Eight studies used healthy controls. Of the four RCTs included in that review, one study included long‐term childhood cancer survivors (mean 12 years from diagnosis). Another RCT offered a two‐ to four‐day intervention, which, therefore, did not match with the inclusion criteria of this Cochrane review (Hinds 2007). The two remaining RCTs of the review by Winter 2010 are also included in this Cochrane review (Hartman 2009; Marchese 2004).
Huang 2011 performed a second review on exercise interventions for childhood cancer patients. They included many of the same studies, but also the study of Chamorro‐Vina 2010, which again introduced an intervention of less than four weeks. Both reviews concluded that results are promising, but that there is a need for more and larger RCTs. Both reviews stated that only a subgroup of the childhood cancer population was tested, since almost all studies concerned children with ALL. These findings are consistent with our findings.
Wolin 2010 performed the third review on exercise intervention for children with cancer. This review studied exercise intervention for adults or children (or both) with cancer. They included 12 studies on children with cancer or survivors of childhood cancer (or both) who did not received a stem cell transplantation and two studies in children with cancer who did receive a haematopoietic stem cell transplantation. Next to the participants who received transplantation, they included also uncontrolled studies and studies with a short intervention duration.
Baumann 2013 performed the fourth review on this subject. This review included 17 studies, of which seven were uncontrolled. They included three RCTs also in our list, but also two others with a short intervention period. Both Baumann 2013 and Wolin 2010 included the study of Shore 1999, which we excluded for this review because it used healthy control participants.
Despite the different studies included in the reviews, all the conclusions were comparable with ours; although studies have limitations in their methodology the results are promising. However, more research is needed to increase the level of evidence.
Study flow diagram.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1 Cardiorespiratory fitness outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 1 9‐minute run‐walk test.
Comparison 1 Cardiorespiratory fitness outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 2 Timed up‐and‐down stairs test.
Comparison 1 Cardiorespiratory fitness outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 3 Timed up‐and‐go test.
Comparison 2 Body composition outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 1 Bone mineral density.
Comparison 2 Body composition outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 2 Body mass index.
Comparison 3 Flexibility outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 1 Flexibility.
Comparison 4 Muscle endurance/strength outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 1 Knee strength.
Comparison 4 Muscle endurance/strength outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 2 Ankle dorsiflexion strength.
Comparison 4 Muscle endurance/strength outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 3 Back and leg dynamometry.
Comparison 4 Muscle endurance/strength outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 4 Inspiratory breathing muscle strength.
Comparison 4 Muscle endurance/strength outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 5 Expiratory breathing muscle strength.
Comparison 5 Health‐related quality of life outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 1 PedsQL ‐ General.
Comparison 5 Health‐related quality of life outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 2 PedsQL ‐ Cancer.
Comparison 5 Health‐related quality of life outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 3 Parents PedsQL ‐ General.
Comparison 5 Health‐related quality of life outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 4 Parents PedsQL ‐ Cancer.
Comparison 6 Fatigue outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 1 PedsQL ‐ General Fatigue.
Comparison 6 Fatigue outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 2 PedsQL ‐ Sleep/Rest Fatigue.
Comparison 6 Fatigue outcomes after physical exercise training intervention for children and adolescents during or after childhood cancer, Outcome 3 PedsQL ‐ Cognitive Fatigue.
| Physical exercise training compared to usual care for children and young adults during and after treatment for childhood cancer | |||||
| Patient or population: children and young adults during and after treatment for childhood cancer Settings: hospital and non‐hospital Intervention: physical exercise training Comparison: usual care | |||||
| Outcomes | Illustrative comparative risks* (95% CI) | No of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | ||||
| Usual care group | Exercise group | ||||
| Cardiorespiratory outcomes | |||||
| 9‐minute run‐walk test | The mean 9‐minute run‐walk test in the control group was | The mean difference between the study groups on 9‐minute run‐walk test was 681 feet (95% CI 132 to 1230) in favour of the intervention group | 68 | ⊕⊕⊝⊝ | SMD 0.69 (95% CI 0.02 to 1.35) |
| Timed up‐and‐down stairs test | The mean time used for timed up‐and‐down stairs test in the control groups was 9.0 sec and 8.6 sec in the 2 studies | The mean difference between the study groups on timed up‐and‐down stairs was ‐0.94 sec (95% CI ‐2.94 to 1.06) in favour of the intervention group | 68 | ⊕⊝⊝⊝ | SMD ‐0.54 (95% CI ‐1.77 to 0.70) |
| Timed up‐and‐go time test stopwatch Follow‐up: mean 3 months | The mean timed up‐and‐go time test in the control group was 8.3 sec | The mean timed up‐and‐go time test in the intervention group was 6.6 sec (1.3 SD); showing a mean difference of ‐1.8 (95% CI ‐2.7 to ‐0.8) | 40 (1 study) | ⊕⊕⊝⊝ | SMD ‐1.15 (95% CI ‐1.83 to ‐0.48) |
| Body composition outcomes | |||||
| Bone mineral density (total body) | The control group had a mean SDS on (total body) bone mineral density of ‐1.1 (95% CI ‐0.8 to ‐1.4) | After the 24 months' intervention the mean SDS of the intervention group on total body bone mineral density was 0.3 better than in the control group (‐0.8 SDS (95% CI ‐0.6 to ‐1.2) | 51 | ⊕⊕⊕⊝ | SMD 1.07 (95% CI 0.48 to 1.66) |
| BMI Quetelet Index | The 2 studies reported a mean change in Quetelet index of the control groups of 1.0 and 0.6 in the 2 studies | The intervention group had a mean change in Quetelet index of 1.2 and 0.7 in the 2 studies The mean difference between the study groups on BMI was 0.18 points (95% CI 0.07 to 0.30) in favour of the intervention group | 64 | ⊕⊕⊝⊝ | SMD 0.59 (95% CI ‐0.23 to 1.41) (BMI increased in the intervention group) |
| Muscle endurance/strength outcomes | |||||
| Ankle dorsiflexion strength | The mean ankle dorsiflexion strength in the control group was 0.22 kg (normalized for participant's weight) | The mean ankle dorsiflexion strength in the intervention groups was 0.25 kg; showing a mean difference of 0.03 (95% CI ‐0.04 to 0.1) | 28 | ⊕⊕⊝⊝ | SMD 0.29 (95% CI ‐0.46 to 1.04) |
| Health‐related quality of life | |||||
| Health‐related quality of life | The mean health‐related quality of life in the control group was | The mean health‐related quality of life in the intervention group was | 28 | ⊕⊕⊝⊝ | SMD ‐0.23 (95% CI ‐0.98 to 0.51) |
| Fatigue | |||||
| General fatigue (range: 0‐100) | The mean general fatigue in the control group was | The mean general fatigue in the intervention group was 3.3 points. The mean difference between the study groups was ‐0.15 (95% CI ‐3.2 to 2.9) | 22 | ⊕⊝⊝⊝ | SMD ‐0.04 (95% CI ‐0.88 to 0.8) |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in the table. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group. | |||||
| GRADE Working Group grades of evidence | |||||
| 1. Rated down for imprecision (sample size less than threshold rule‐of‐thumb n = 400, Schünemann 2009) and for high risk of bias in the study of Tanir 2013. 2 Rated down for imprecision (small sample size and confidence intervals include both appreciable benefit and appreciable harm, defined by SMD = ‐0.5 and SMD = 0.5, Schünemann 2009), for inconsistency, and for high risk of bias in the study of Tanir 2013. 3 Rated down for imprecision (small sample size). 4 Rated down by 1 level for imprecision (small sample size) and by 1 level for unclear risk of bias for sequence generation and outcome assessor blinding in study of Moyer‐Mileur 2009. 5 Rated down by 2 levels for imprecision (small sample size and confidence intervals include both appreciable benefit and appreciable harm, defined by SMD = ‐0.5 and SMD = 0.5, Schünemann 2009). 6 Rated down by a total of 3 levels for imprecision (small sample size and confidence intervals include both appreciable benefit and appreciable harm, defined by SMD = ‐0.5 and SMD = 0.5, Schünemann 2009), and for high risk of (selection) bias in the study of Yeh 2011. | |||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 9‐minute run‐walk test Show forest plot | 2 | 68 | Std. Mean Difference (IV, Random, 95% CI) | 0.69 [0.02, 1.35] |
| 2 Timed up‐and‐down stairs test Show forest plot | 2 | 68 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.54 [‐1.77, 0.70] |
| 3 Timed up‐and‐go test Show forest plot | 1 | 40 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.15 [‐1.83, ‐0.48] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Bone mineral density Show forest plot | 1 | 51 | Std. Mean Difference (IV, Random, 95% CI) | 1.07 [0.48, 1.66] |
| 2 Body mass index Show forest plot | 2 | 64 | Std. Mean Difference (IV, Random, 95% CI) | 0.59 [‐0.23, 1.41] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Flexibility Show forest plot | 2 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 1.1 Active ankle dorsiflexion | 1 | 28 | Std. Mean Difference (IV, Random, 95% CI) | 0.46 [‐0.29, 1.22] |
| 1.2 Passive ankle dorsiflexion | 1 | 51 | Std. Mean Difference (IV, Random, 95% CI) | 0.69 [0.12, 1.25] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Knee strength Show forest plot | 1 | 28 | Std. Mean Difference (IV, Random, 95% CI) | 0.25 [‐0.49, 1.00] |
| 2 Ankle dorsiflexion strength Show forest plot | 1 | 28 | Std. Mean Difference (IV, Random, 95% CI) | 0.29 [‐0.46, 1.04] |
| 3 Back and leg dynamometry Show forest plot | 1 | 40 | Std. Mean Difference (IV, Random, 95% CI) | 1.41 [0.71, 2.11] |
| 4 Inspiratory breathing muscle strength Show forest plot | 1 | 14 | Std. Mean Difference (IV, Random, 95% CI) | 0.33 [‐0.77, 1.43] |
| 5 Expiratory breathing muscle strength Show forest plot | 1 | 14 | Std. Mean Difference (IV, Random, 95% CI) | 0.0 [‐1.09, 1.09] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 PedsQL ‐ General Show forest plot | 1 | 28 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.23 [‐0.98, 0.51] |
| 2 PedsQL ‐ Cancer Show forest plot | 1 | 28 | Std. Mean Difference (IV, Random, 95% CI) | 0.16 [‐0.58, 0.91] |
| 3 Parents PedsQL ‐ General Show forest plot | 1 | 28 | Std. Mean Difference (IV, Random, 95% CI) | 0.38 [‐0.37, 1.13] |
| 4 Parents PedsQL ‐ Cancer Show forest plot | 1 | 28 | Std. Mean Difference (IV, Random, 95% CI) | 0.04 [‐0.70, 0.79] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 PedsQL ‐ General Fatigue Show forest plot | 1 | 22 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.04 [‐0.88, 0.80] |
| 2 PedsQL ‐ Sleep/Rest Fatigue Show forest plot | 1 | 22 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.01 [‐0.85, 0.83] |
| 3 PedsQL ‐ Cognitive Fatigue Show forest plot | 1 | 22 | Std. Mean Difference (IV, Random, 95% CI) | 0.07 [‐0.77, 0.91] |
