Procedure
Patients were recruited from March 2009 to July 2013 in four Dutch university hospitals: VU University Medical Center, Amsterdam; Academic Medical Center, Amsterdam; University Medical Center Utrecht, Utrecht; and Erasmus University Medical Center, Rotterdam. The Medical Ethics Committees of all hospitals approved the study. The trial was registered at the Dutch Trial Registry (NTR1531) and performed according to the 1964 Declaration of Helsinki. Eligible participants were aged 8–18 years, and were currently receiving or within the first year following cancer treatment with chemotherapy and/or radiotherapy [29]. Patients were included when the remaining treatment period included no scheduled hospitalization, and when the clinical condition (according to the treating oncologist) made it possible to exercise.
Exclusion criteria were (previous) treatment with growth hormone, due to its possible influence on bone density; stem-cell transplantation, due to the high impact of the treatment and related seriously low physical condition of the children after this treatment; cardiomyopathy, due to the impact of physical exercise on the heart condition; inability to ride a stationary bike, for testing the primary outcome; and inability to read and write Dutch, to self-reflect or to follow instructions due to learning difficulties (to be able to adequately complete the psychosocial and physical intervention). The treating physician checked the exclusion criteria by viewing medical records, school information and their own clinical impression. Both patients and their parents or legal representatives received spoken and written study information and provided written informed consent prior to participation. After baseline measurements, block randomization was performed by an independent data manager and stratified by age, gender, cancer type (haematological cancer vs. solid tumour), and treatment phase (during vs. after treatment) [29].
Intervention
The 12-week individually performed QLIM intervention included two 45-min physical exercise sessions per week at a local physical therapy practice and one 60-min psychosocial training session once every 2 weeks for the child in the treating pediatric oncology hospital [29]. The parents also received two psychosocial training sessions; a start and evaluation session. Baseline measurements were performed in the treating hospital. The intervention started within 14-days post baseline measurement. In case of seriously low blood counts or fever, the child was instructed to postpone the physical exercise training with a maximum of one-week. As there was often need for a family break after the intense childhood cancer treatment, a one-week holiday was also accepted as a valid reason to postpone a maximum of two physical exercise training-sessions.
The physical exercise training was an individual program performed at a local physical therapy practice. This program was specifically developed for children. It included both aerobic as well as weight-bearing exercises performed in a circuit training-setting with balls, hoops, and running activities. The intensity of the physical exercise training program gradually increased.
In the first eight training sessions (the first month) the training was performed at a peak heart rate (HRpeak) of 66–77%. During these sessions the exercise training included muscle strength training with some elements of aerobic training. In the second month (sessions 9–16) the training impact increased to a HRpeak of 77–90% with a training focusing more on aerobic fitness, supplemented by moderate intensive strength training. In the third month (sessions 17–24) the training included highly intensive aerobic training and strength training at a HRpeak of 90–100%.
All physical therapists received an instruction manual accompanied by verbal explanation. In addition to the exercise training sessions at a local physical therapy center, the intervention also included a homebased program. This home-based program started after the sixth week of the intervention. Children were instructed to perform a number of weight-bearing exercises at a high intensity-level at home at least three times per week, as advised by the physical therapist. One exercise cycle took 11-min to complete and could therefore easily be implemented as part of daily routine. The 11-min exercises were guided by timed-music to increase the joy of participation. Three to six months’ post-intervention, a booster-session was included to reinforce the importance of physical exercise, to show the children’s sport-participation to their parents and peers, and to increase self-esteem in a non-therapeutic sport setting. Children received a tennis-lesson which was preferably provided in a small group of age-matched study participants.
The psychosocial training intervention was an individualized structured program to enhance socio-emotional functioning and coping with disease-related effects. It consisted of psycho-education and cognitive-behavioral techniques including items on expression of feelings, self-perception and coping skills [30]. The intervention aimed to increase general psychosocial functioning as expressed in HrQoL, self-perception, behavior problems and depressive symptoms. The individual sessions were performed parallel to the physical exercise intervention. After each individual session home exercises on the topic of this specific session could be given to the patient if the psychologist considered it necessary. The parent sessions were scheduled at the start and end of the child’s training. The training was performed by a trained paediatric psychologist according to an instruction manual. Details of the psychosocial training, its applicability and evaluation are published elsewhere [30].
The control group received usual care according to local guidelines and preferences.
Data collection and instruments
Measurements of all primary and secondary outcomes took place in the treating hospitals at baseline, after completion of the intervention at 4 months and at long-term (12-months) follow-up. Both the aerobic fitness and the muscle strength tests were performed by blinded assessors. The same test-equipment was used in all centres, with one exception: The Dual-energy-X-ray absorptiometry (DXA)-scanner. I.e. in all but one centres a Hologic DXA scanner with the same software was used to measure bone mineral density (BMD). In the Erasmus university hospital in Rotterdam, a Lunar scanner was used. To correct for the different scan systems, we used a transcription model to compare the data. Visits to physical therapists, sport-centres and psychologists were not structurally monitored in the control group.
Primary outcomes
Cardiorespiratory fitness was assessed by peak oxygen uptake (VO2peak expressed in ml•kg•min− 1) during a cardiopulmonary exercise test using the Godfrey protocol [31]. The test was performed on an electronically braked cycle ergometer with a paddling frequency of 60–80 rpm. During the test, expired air was collected, heart rate was monitored, and ventilator gas exchange data were determined breath-by-breath. The VO2peak was defined as the mean score of the final 30 s of the test. Cardiorespiratory fitness data were included in the analyses for children that achieved a HRpeak of at least 180 beats per minute, and/or a RERpeak of ≥1.0.
Muscle strength was assessed using a hand-held dynamometer (CITEC; C.I.T. Technics, Groningen, the Netherlands) [32]. All children performed three repetitions (both left and right) per muscle group. The highest-score out of six repetitions was used for further analyses. Upper-body muscle strength was calculated by summing the highest-score of the shoulder, elbow and grip strength, and the sum of the highest hip, knee and ankle-dorsiflexion scores was used for lower body muscle strength.
Secondary outcomes
Body composition was determined using percentage of fat mass (%FM) and lumbar spine (L1-L4) BMD as measured by DXA.
Physical activity was measured with an Actical accelerometer (B series, Philips Respironics Actical MiniMitter, Murrysville, PA, USA) by a 15-s time-interval and expressed as mean counts per minute [33, 34]. Mean counts per minute is a physical activity score including horizontal, vertical and depth motion scores in one end-score; higher scores indicate more activity [33, 34]. The accelerometer was attached to an elastic waist belt, and worn on the left hip during daytime at waking-hours (between 6:00 am and 11:59 pm). The accelerometer was worn on four consecutive days: Wednesday to Saturday, in the week following the measurements in the hospital. Assessing 4 days was found to be sufficient and less invasive than monitoring for one-week [35]. After wearing, participants sent the accelerometers back to the research team by postal mail. Per minute data were assessed, excluding large (> 60 min) periods of consecutive zeros to validate wear-time. A mean cpm score over the recorded days and wear-time data was derived and used in the analyses.
Fatigue was assessed with the overall-fatigue score of the child self-report version of the PedsQL™ Multidimensional fatigue scale (acute version) [36, 37]. This instrument is designed to measure both the child’s and the parent’s perception of fatigue in pediatric patients [36]. The module encompasses 3 subscales: general fatigue (6 items), sleep/rest fatigue (6 items), and cognitive fatigue (6 items), and an overall fatigue score (all 18 items). Scores were calculated according to the manual and ranged from 0 to 100 with lower scores indicating higher levels of fatigue [36]. For the present study only the overall-fatigue score was taken into the analyses. The Dutch version (8–18 year) has adequate psychometric properties and normative scores of the Dutch population are available.
Total general-HrQoL was measured with the Dutch self-report version of the PedsQL™ Generic Core Scales for children aged 8–12 and 12–18 years [36, 38]. This consists of 4 multi-item subscales: physical functioning (8 items), emotional functioning (5 items), social functioning (5 items), and school functioning (5 items). A total HrQoL score was calculated according to the PedsQL™ manual; this scale has a range of 0–100, with higher scores reflecting a better HrQoL. The PedsQL™ has proven to be reliable and valid in pediatric patients [36]. The Dutch version has adequate psychometric properties, and normative scores of the Dutch population are available [38].
Athletic competence and global self-worth were assessed with the athletic competence and global self-worth subscales of the ‘Self-Perception Profile’ for children aged 8–11 years and for adolescents aged 12–18 years [39, 40]. The Self-Perception Profile has good reliability and validity when used in children aged ≥8 years [39, 40]. Higher scores (0–100) reflect more positive self-perceptions [39, 40].
Behavioural problems were assessed in children aged ≥11 years using the Youth Self-Report with higher T-scores indicating more behavioural problems [41]. It yields a total score ranging from 0 to 100. For the present study, the total problem behaviour scale was taken into the analyses. YSR is a valid and reliable instrument to assess evaluation of internalizing and externalizing behavioural problems [41].
Depressive symptoms were assessed with the Children’s Depression Inventory [42] for children 8–18 years. The total score ranges from 0 to 100 with higher scores reflecting more depressive symptoms. Good internal consistency and test-retest reliability, and a positive correlation with clinicians’ independent global depression ratings, are reported [42].
Demographic and medical characteristics including age, gender, height, weight and body mass index, type of cancer, treatment, and treatment phase (during vs. post) were obtained from medical records.
Adherence and applicability
Session attendance of the physical exercise and psychosocial training intervention was recorded by the therapists. The physical therapist recorded the performed training intensity (heart rate) and possible adaptations to the program on an evaluation and adaptation form. Researchers rated the applicability as ‘good’ when the program could be performed on an intensity level equal to, or < 10% lower than requested and when only small (material) adaptations were needed. The results of the home-based program were self-reported in a training-logbook for personal evaluation. As a result, no valid applicability data of this home-based intervention program were collected, and no analysis on applicability of the home-based program could be performed. Applicability of the psychosocial intervention was assessed by questionnaires which were completed by the participating psychologists and by the patients (details are published elsewhere [30]).
Sample size calculation
Based on a previous uncontrolled study on physical exercise intervention effects on cardiorespiratory fitness levels in children with cancer (2007), the intervention group was expected to show an at least 20% greater improvement in cardiorespiratory fitness than the control group shortly after the intervention [43]. Therefore, at least 26 patients per group were required to detect an effect size of 0.8 [44] between the intervention and control group with a power of 80% and an alpha of 0.05 [29]. Taking 40% dropout into account, we aimed to include 100 patients [29].
Statistical analysis
Data were analysed using IBM SPSS Statistics for Windows (version 20.0). Data are presented as mean (standard deviation [SD]) or median (interquartile range) for all outcomes.
Generalized estimating equations (GEE) analyses with an exchangeable correlation structure were used to simultaneously assess intervention effects on the outcome variable at short and long-term (between-group differences) [45]. This statistical method adjusts for the non-independence of observations over time. Study group, time and the interaction of study group × time were entered in the regression model as independent variables, adjusting for baseline values. The interaction term between study group and time was included to separate the short-term effects from the long-term effects. We also studied within-group changes over time using the same GEE analyses. Intervention effects were evaluated using an intention-to-treat principle. Regression coefficients with 95% confidence levels were reported for intervention effects (between-group differences) from baseline to short-term, and from baseline to long-term, and for the within-group changes over time.
In a secondary per-protocol analysis, we studied intervention effects in children who had 100% attendance to the intervention (n = 20) and compared them to the control group (n = 33).
To study whether the intervention effect on general HrQoL was mediated by physical fitness, physical activity, fatigue, self-perception, depressive symptoms, athletic competence, global self-worth and behavioural problems we used a series of linear regression analyses according to the products-of-coefficients test [46] (Fig. 1). First, we evaluated the intervention effect on HrQoL at the long-term adjusted for the baseline value of HRQoL (c path). Second, we evaluated the intervention effect on the potential mediator at short-term controlled for the mediator at baseline (a path). Third, the association between the potential mediator at short-term and the outcome variable HrQoL at 12 months was calculated, controlled for the intervention and baseline values of the mediator and outcome variable (b path); this step also provides information on the direct intervention effect on HrQoL at 12 months adjusted for the mediator variable (c’ path). The product of coefficients (axb) was used to estimate the relative strength of the mediation effect. We used bootstrapping techniques with 5000 bootstrap resamples to calculate the bias-corrected and accelerated 95% confidence intervals (CI) around the mediation effect (axb) using the SPSS macro provided by Preacher and Hayes [47].
