To examine the effectiveness of high-intensity aerobic training compared with low-intensity training in terms of energy cost of locomotion, peak oxygen uptake, peak power, and self-reported physical function in children with juvenile idiopathic arthritis (JIA).
Eighty children with JIA, ages 8–16 years, were enrolled in a randomized, single-blind controlled trial. Both groups participated in a 12-week, 3-times–weekly training program consisting of high-intensity aerobics in the experimental group and qigong in the control group. Subjects underwent exercise testing measuring submaximal oxygen uptake at 3 km/hour (VO2submax) as the primary outcome, maximal oxygen uptake, and peak power at the beginning and end of the program. Physical function was measured using the Child Health Assessment Questionnaire (C-HAQ).
The exercise program was well tolerated in both groups. There was no difference in VO2submax or any other exercise testing measures between the groups through the study period and no indication of improvement. Both groups showed significant improvements in C-HAQ with no difference between the groups. Adherence was higher in the control group than the experimental group.
Our findings suggest that activity programs with or without an aerobic training component are safe and may result in an important improvement in physical function. The intensity of aerobic training did not seem to provide any additional benefits, but higher adherence in the qigong program may suggest that less intensive regimens are easier for children with JIA to comply with, and provide a degree of benefit equivalent to more intensive programs.
Juvenile idiopathic arthritis (JIA) affects ∼1 in 1,000 children (1). These children participate in less physical activity and have more sleep hours than their peers (2). A meta-analysis of studies examining peak oxygen consumption (VO2peak) showed a 22% reduction in children with arthritis compared with their peers (3). Other studies have shown a reduction in muscular endurance in children with JIA (4, 5). These factors may lead to a spiral of deconditioning and inactivity resulting in further deconditioning and limited participation, culminating in prolonged disability (6).
Traditionally, physical therapy for JIA has aimed to manage pain, preserve range of motion, and limit the strain on arthritic joints (7, 8). More active forms of therapy have been recently instituted (9), and guidelines have included recommendations for fitness and strengthening exercise to improve function and promote lifetime physical activity (10).
Systematic reviews examining aerobic training in adults with rheumatoid arthritis (RA) found improvements in aerobic capacity, muscle strength, and disease activity, with possible beneficial effects on pain, function, and quality of life (11, 12). Early studies in children suggested that exercise was well tolerated and might improve function (13–15). However, these studies were small, uncontrolled, and relied on field testing rather than laboratory measurements. Subsequently, Oberg et al (16) reported that children with arthritis achieved improvements in muscular strength and endurance after a 3-month training program and that electromyographic abnormalities improved with training. An uncontrolled study of 25 children with polyarticular JIA demonstrated improvements in aerobic capacity and flexibility after an 8-week, low-impact aerobic and resistance program (17). A recent randomized controlled study of aerobic pool exercise involving 54 children suggested (nonsignificant) improvements in quality of life, joint status, and submaximal endurance (18).
Maximal oxygen uptake (VO2max) protocols used to assess fitness in adults are rigorous and have proven difficult to apply in pediatrics. Less intensive peak VO2 (VO2peak) protocols have been used in children (19). However, VO2peak may be an inappropriate outcome measure in groups with significant limitations. Instead, submaximal VO2 (VO2submax) may be used as a predictor of VO2max or as an independent measure of performance in standardized activities (18, 20–23). Giannini and Protas (24) reported that children with JIA had higher submaximal heart rate (HR) and tended towards higher VO2submax (higher oxygen requirement) as compared with matched controls during a bicycle task. Children with JIA may also have an exaggerated energy cost of locomotion (higher VO2submax) that manifests as early fatigue during normal activity. Thus VO2submax may provide a means of measuring disability in JIA and provide a well-tolerated measure of aerobic fitness and response to therapy.
The goals of this study were to examine the effectiveness of a high-intensity 12-week program in terms of VO2submax in children with inflammatory arthritis and to determine the effectiveness of this program in terms of self-reported physical function, VO2peak, and peak power.
PATIENTS AND METHODS
Eighty patients with JIA, ages 8–16 years, who were considered stable by their rheumatologist and unlikely to require modification of therapy during the study were recruited from The Hospital for Sick Children. No restrictions were placed on medication use, but every effort was made to maintain stable dosage through the study.
Patients were excluded if they had significant cardiac, pulmonary, or metabolic comorbidity; if they had moderate or severe hip pain while walking (judged by the patient as ≥3 on a 4-point scale) because hip pain was identified as a major limitation to participation in our pilot study (25); if they were engaged in ≥3 hours per week of extracurricular physical activity, excluding physiotherapy pool programs because these children might not show additional gains from fitness training; or if they were unable to cooperate with training or testing.
Block randomization balanced for pubertal stage (less than or equal to Tanner stage 2 versus greater than or equal to Tanner stage 3) and degree of disability as measured by the Childhood Health Assessment Questionnaire (C-HAQ) (26) (score ≥0.125 versus score <0.125) was undertaken, as these factors might independently influence fitness measures. The concealed allocation scheme involved sequential opaque envelopes in blocks of 2–4. Although it was impossible to blind the subjects because the programs were obviously different, study personnel responsible for administering the fitness testing, statistical analysis, and manuscript preparation were blinded to patient allocation.
Groups undertook a 12-week exercise program consisting of 1 supervised session and 2 unsupervised sessions per week. The supervised sessions, held in 6 locations, were led by an exercise therapist with a 1:4 instructor to subject ratio.
Experimental sessions commenced with a 10-minute warm-up mimicking the exercise routine, followed by light stretching. The aerobic training program, drawn from dance and martial arts (cardio-karate), was designed to avoid activities that might be unsafe (details of the exercise programs are available from the authors). The 30-minute sessions increased progressively in intensity from low to moderate/high as tolerated. Sessions concluded with 5–10 minutes of passive stretching. HR was measured either as a manual 15-second count at the carotid artery or by HR monitor (Polar 650i, Polar Instruments, Kempele, Finland). Target HR range was >75% of the maximal HR (MHR) determined from VO2peak testing conducted at enrollment. Instructors encouraged subjects to exercise in this range. Rating of perceived exertion (RPE) using the children's OMNI scale (27) was used to determine exercise intensity.
Control sessions were held at the same locations as the experimental group session, but at different times. The program was based on qigong (28), a gentle relaxation program similar to tai chi. Tai chi has been reported as a safe and feasible activity for adults with RA (29). Sessions were based on an 18-posture program designed to avoid aerobic training or elevating HR. Each posture was repeated 8 times. A cool down of gentle stretching concluded the session. Subjects recorded HR and RPE to ensure that it was below 75% of MHR.
Unsupervised sessions for both groups followed the same program as in-class sessions, but used videotaped instruction. Subjects were asked to complete 2 at-home video sessions per week, and if a class session was missed, the subject was instructed to make up that session at home.
Subjects underwent 3 testing sessions: the first at enrollment, during which subjects were familiarized with the testing equipment and the data collected was used in balancing the randomization process; the second planned 2 weeks later at the time of group allocation and just prior to commencement of the program; and the third within 2 weeks of completion of the training program.
At each session, height (Harpenden Stadiometer, London, UK) and weight (SR555 Stand-on Scale System, SR Instruments, Tonawanda, NY) measurements were obtained to the nearest 0.1 cm and 0.1 kg, respectively, with the subjects wearing light clothes, but not shoes. Percent body fat was calculated from skin-fold measurements of the bicep, tricep, subscapular, and suprailiac crest regions using the average of 3 trials and entered into the Slaughter equation (30).
The VO2submax was measured while treadmill walking at 1.5 km/hour and 3.0 km/hour for 5 minutes (8, 31). Absolute VO2submax at 3 km/hour was the primary outcome measure. During all treadmill testing expired gases were collected continuously, with ventilatory equivalent ratio for oxygen (VE/VO2), ventilatory equivalent ratio for carbon dioxide (VE/VCO2), respiratory exchange ratio (RER), and HR recorded at 20-second intervals (Physiodyne Max-II metabolic cart, Quogue, NY) using a 4-lead electrocardiogram system (GE Case 8000, General Electric Medical Systems, Milwaukee, WI) and RPE was assessed using the Children's OMNI scale (27). Steady state for each stage was recorded as the average of the last minute of VO2 measurements.
The VO2peak was measured through an incremental, continuous walking task on a treadmill with speed or incline increased every minute until volitional fatigue. No preselected increments were used due to low and heterogeneous fitness common in children with chronic disease (8). Increments were selected by an experienced exercise tester according to the child's HR, RER, RPE, and overall appearance. Test duration was targeted at 6–10 minutes. Criteria for VO2peak were attaining RER >1.0 and reaching age-predicted MHR.
After 10–15 minutes of rest, subjects performed a modified Wingate test via 10-second and 30-second all-out cycling tasks on an isokinetic cycle ergometer (Biodex lower body cycle, Biodex Medical Systems, Shirley, NY). The isokinetic cycle fixes the speed of pedaling to measure the force output and is a highly reproducible and safe method of assessing anaerobic power (32, 33). Peak power was measured over 10 seconds at 90 revolutions per minute, and then over 30 seconds (after a 2-minute recovery period) to estimate total work output in watts.
The C-HAQ was used to assess physical function. Items are scored on a 4-point scale with scores of 0 denoting no disability and 3 denoting severe disability. It is validated for use in JIA (26). We chose to use the C-HAQ in preference to the modified C-HAQ38 as the latter instrument has not yet been validated for longitudinal studies (34). Physical activity levels outside the study were measured using the Habitual Activity Estimation Scale (HAES), a validated tool for measuring activity levels in both healthy and chronically ill children (35, 36). It asks children to recall an average weekday and a weekend day with regard to the proportion of time spent inactive (e.g., sleeping), somewhat inactive (e.g., sitting), somewhat active (e.g., walking), and active (e.g., running), and provides a summary score of these activity levels. Overall quality of life (QOL) and health-related quality of life (HRQOL) were measured on 10-cm visual analog scales (VAS) previously validated for children with JIA, with higher scores indicating better QOL or HRQOL (37).
Adherence and safety.
The study coordinator and fitness instructors maintained frequent contact with the subjects to maintain adherence, and the instructors met together on a monthly basis to review each individual's progress. Subjects measured and recorded RPE, HR (pulse point or monitor), and pain on a 10-cm VAS in a diary for each session they completed. HR monitors were loaned intermittently to check adherence. Children were rewarded with stickers for completed sessions and were able to trade these for small gifts to further encourage adherence. These strategies had enhanced adherence in our pilot study (25).
Safety was monitored in diary records of pain and through joint examination at testing sessions. A joint was considered active if it was swollen or had 2 of the following features: limited range of movement, pain on movement/tenderness, and/or warmth. Range of motion was assessed at testing sessions using the Pediatric Escola Paulista de Medicina Range of Motion scale (38).
A sample size of 35 subjects per group was calculated to detect a meaningful difference between the groups in absolute VO2submax at 3 km/hour, with an α (false-positive) error of 0.05, β (false-negative) error of 0.20, and δ (the size of a clinically important difference) of 10% (39). Based on an anticipated 10% dropout rate, the intended recruitment number was 80 subjects.
The change in VO2submax between the 2 groups was compared using a linear mixed-effects model with compound symmetry covariance structure. Potential confounders (sex, pubertal stage, C-HAQ score, and HAES activity level outside the study) were entered into the model. Differences between the groups for secondary outcomes were tested similarly. Analyses were completed using SAS software version 9.1 (SAS Institute, Cary, NC) and SPSS version 12.0 (SPSS, Chicago, IL).
The primary analysis was intent-to-treat, with the subjects analyzed per their original group assignment regardless of adherence. The exercise dosage used in this study was considered a minimum to improve aerobic capacity. However, exercising less often has been shown to have benefit for adults with arthritis (11, 12). In an attempt to see whether adherence significantly influenced the results, we analyzed data from experimental subjects and controls who participated in >70% of sessions and also those experimental group subjects who attended >70% of sessions and also achieved >75% of MHR in >50% of sessions attended compared with controls. Missing primary outcome data was analyzed through a sensitivity analysis using imputation with the mean and median of nearby points replacing missing values. Post hoc subgroup analyses of Tanner stage (≤ Tanner stage 2 versus ≥ Tanner stage 3), sex (boys versus girls), and baseline activity (HAES <2 active weekday hours versus ≥2 active weekday hours) were also completed.
This study was approved by the Research Ethics Boards at The Hospital for Sick Children and Bloorview Kids Rehab. All subjects were enrolled only after fully-informed written consent was obtained from the parent and the child.
Patients and adverse events.
Of 453 potential subjects, 80 were recruited (Figure 1). Ten patients dropped out after randomization and 1 patient was ineligible (6 from the experimental group and 5 from the control group). All dropouts reported lack of time as the reason.
Patient characteristics at enrollment are shown in Table 1. Differences are evident in the distribution of JIA subgroups between the groups. The blocking factors, Tanner stage, and C-HAQ used in randomization were equally distributed.
No adverse events were reported during testing or training sessions. There was no worsening of active joint count, Pediatric Escola Paulista de Medicina Range of Motion scale, C-HAQ, QOL, or HRQOL in either group during the study (Table 2). Low levels of pain were recorded on a 10-cm VAS during training sessions. These were not different between the 2 groups (median 0; range 0–10 in both groups; P = 0.09 Mann-Whitney U test).
Table 2. Exercise testing and questionnaire results at enrollment and completion of the trial for patients that completed the trial*
Values are the mean ± SD unless otherwise indicated. VO2submax = submaximal oxygen consumption; VO2peak = peak oxygen uptake; EPM = Pediatric Escola Paulista de Medicina Range of Motion scale; C-HAQ = Childhood Health Assessment Questionnaire; QOL = quality of life; HRQOL = health-related quality of life.
Estimates and P values are presented for the β-term for the interaction of (group allocation × testing session) in the repeated measures, auto-regressive model, and represent the difference of the paired differences in response seen in each of the 2 groups.
VO2submax 1.5 km/hour
0.4 ± 0.1
0.4 ± 0.1
0.4 ± 0.1
0.4 ± 0.1
9.4 ± 1.6
9.3 ± 1.8
9.5 ± 1.7
8.7 ± 1.4
VO2submax 3.0 km/hour
0.5 ± 0.1
0.5 ± 0.1
0.5 ± 0.1
0.5 ± 0.1
11.5 ± 1.9
11.6 ± 1.8
11.4 ± 1.9
11.0 ± 1.5
1.5 ± 0.5
1.6 ± 0.5
1.4 ± 0.4
1.5 ± 0.5
35.7 ± 7.8
36.2 ± 8.0
33.3 ± 6.8
34.8 ± 8.8
Peak power, watts
211 ± 136
216 ± 137
204 ± 120
233 ± 125
218 ± 133
225 ± 134
209 ± 103
236 ± 114
Active joints, mean ± SD (range)
2.5 ± 5.1 (0–21)
2.1 ± 5.1 (0–21)
3.5 ± 6.8 (0–28)
2.2 ± 6.5 (0–15)
0 ± 0.1
0.1 ± 0.4
0.1 ± 0.1
0.1 ± 0.2
Body fat percentage
23.1 ± 7.5
23.3 ± 8.1
24.0 ± 8.8
22.9 ± 7.3
0.32 ± 0.45
0.21 ± 0.35
0.34 ± 0.49
0.22 ± 0.37
8.5 ± 1.6
8.7 ± 1.4
7.9 ± 1.8
8.4 ± 2.0
8.3 ± 1.9
8.5 ± 1.7
7.7 ± 1.8
7.8 ± 1.9
Patient adherence is outlined in Table 3. Completion of training sessions was 78% in the control arm and 56% in the experimental arm. An average of 2 sessions per week were completed by the experimental subjects and 1.7 sessions per week by the control subjects. The difference was most apparent in the number of home-based sessions. As expected, RPE was significantly different between interventions with a median (range) of 5.0 (0.0–10.0) and 1.0 (0.0–9.0), respectively, for the experimental and control groups (P < 0.0001). Only 51% of the completed experimental sessions succeeded in achieving the target HR. The RPE correlated moderately with HR in the experimental group (r = 0.41; P < 0.0001) and less so in the control group (r = 0.32; P < 0.0001).
Values are the mean ± SD unless otherwise indicated. HR = heart rate; MHR = maximal heart rate.
Training sessions completed
Supervised in class, 12 possible sessions
7.8 ± 2.8
7.3 ± 2.9
Home-based sessions, 24 possible sessions
20.4 ± 7.1
14.9 ± 7.1
Total sessions completed of 36 possible session, n (%)
HR during sessions
95 ± 26
136 ± 33
Sessions in which HR >75% MHR, %
Total minutes of training per session
32.2 ± 8.2
34.3 ± 10.0
Exercise testing outcomes.
Subjects' baseline fitness measured by VO2peak and compared with published age and body surface area matched norms (40) demonstrated a significantly lower VO2peak in our patients (Table 4). Results of exercise testing and questionnaires are also shown in Table 2.
Table 4. Comparison of maximum oxygen consumption values according to body surface area, between subjects at study start and published values for healthy untrained children*
No difference was found in the change in the absolute VO2submax at 3 km/hour between the groups from baseline to completion of the study (Figure 2). Nor were any significant differences seen in the change in other fitness parameters between the groups. When missing data was accounted for through imputation analysis, there was still no difference in the primary outcome between the groups.
Analysis of those with >70% completion of training sessions showed no significant difference in any of the fitness parameters between the 2 groups over time. We examined subjects in the experimental arm who attended >70% of training sessions and achieved HRs >75% of MHR in >50% of sessions. This subgroup contained only 8 patients and, when compared with the control group, showed no significant benefit in VO2submax.
Subgroup analyses for sex, pubertal stage, and baseline activity did not show any difference between the subgroups in VO2submax in response to training. The amount of improvement in self-reported physical function as measured by the C-HAQ was similar between groups (Figure 2), although the within-group change was statistically significant (mean difference −0.12, P < 0.0001) and clinically meaningful in magnitude (41).
This study is the first well-powered, randomized controlled trial of land-based aerobic training in children with JIA. We found that participating in a 12-week exercise program did result in improved physical function as measured by the C-HAQ, but did not result in improved economy of locomotion, and that there was no extra improvement conferred by high-intensity aerobic training. The benefits observed in the C-HAQ might be a result of being involved in a trial and the added attention received from trainers and study personnel (Hawthorne effect), or it may reflect a true benefit of training.
When considering the role of vigorous physical training in childhood arthritis, we must consider that the failure to demonstrate improvements in this study may have been a result of patient selection, the training program instituted, adherence with the program, or a general poor response of fitness training in young children.
It is possible that subjects were too mildly affected as measured by their C-HAQ scores to have benefited from the training program. Previous studies have shown that children with arthritis have reduced fitness when compared with their healthy peers (3), and while many of our subjects had mild disease, they were deconditioned when compared with available normal values for healthy untrained children (40). In adults, treadmill protocols attain 6–10% higher VO2max than bicycle protocols (42, 43). Thus, as the comparison data from Washington et al (40) were obtained using a cycle ergometer and ours were obtained from treadmill testing, it is likely that subjects in this study were even more deconditioned than these data suggest. Nonetheless, even though it might be expected that the deconditioned subjects included in this study would be more likely to benefit from an exercise program, no benefit in fitness parameters was seen.
Submaximal exercise testing has been reported as a safe, simple, and well-tolerated estimate of VO2max and functional ability in children with chronic disease (20, 22–25). Giannini and Protas (24) showed that children with arthritis had higher submaximal HR and trended towards higher VO2submax than healthy peers, suggesting significantly lower submaximal efficiency. The failure of our study to show significant changes after training may be because disease activity of patients in the Giannini and Protas study (24) was significantly greater than in our study, with a mean joint count of 13 compared with <3 in our study. However, we feel that the subjects included represent a realistic cross section of JIA patients in our clinic and that the results are generalizable to clinic populations in most developed countries.
Training programs should involve an exercise prescription of adequate frequency, intensity, and duration to achieve the desired results. A minimum of twice-weekly training is required to achieve significant improvements in physical fitness (12, 44). An 8-week study by Klepper (17) involved training twice weekly in class format and once weekly at home. Studies of aquatic training (18, 45) and the pilot study undertaken prior to this trial (25) used once-weekly class-based training for 15, 20, and 12 weeks, respectively. Feedback from our pilot study highlighted difficulties in attending frequent formal training sessions and suggested that a partially home-based program would help adherence (25). The program used in our present study was designed to provide adequate frequency (3 sessions per week), intensity (≥75% MHR), and duration (30 minutes) of aerobic training based on existing recommendations and the findings of earlier studies.
Failure to achieve target HR may have been due to the training program not being strenuous enough to achieve the desired range. While RPE measured in training sessions was significantly higher in the experimental group than in the control group, which suggests at least moderate exertion in the experimental group, HR and RPE were only moderately correlated. It is possible that the more complicated exercise maneuvers in the experimental program, compared with treadmill walking or running, added complexity which translated as a higher RPE without concurrent increase in HR. Also, the novelty of the vigorous program may have resulted in earlier exertion in the subjects with a tailing off towards the end of the session when RPE was measured. Furthermore, the OMNI scale was validated using treadmill or cycling protocols that were continuous in nature, but our program was intermittent, possibly resulting in a lower correlation (27).
Differences in the modality of training and testing may have affected our results. Patients were trained in floor-based programs but tested on treadmill and cycle ergometers. In the pilot study, subjects were trained in a circuit format incorporating treadmill and stationary cycle and then tested on these 2 apparatus (25). Thus there may be an issue with the specificity of the testing method to detect changes resulting from the training modality.
Adherence in this study was higher in the control group than in the experimental group. Takken et al (18) showed good adherence with once-weekly aquatic training, whereas our own pilot study achieved only fair adherence with a once-weekly program (25). A home exercise intervention in cystic fibrosis from Schneiderman-Walker et al (46) reported good long-term adherence with a 3-times–weekly program out to 3 years, while Greenan-Fowler et al (47) were unable to maintain adherence in children with hemophilia in home exercises. Munneke et al (48) found that adult patients with RA maintained good adherence in a 2-year, twice-weekly training program. Our results highlight the difficulties in obtaining adequate and sustained adherence with intensive exercise programs from children with arthritis.
The training HR was only achieved in half of all sessions in the experimental arm, reducing the effective training frequency to once a week when adherence with sessions was taken into account. This may have significantly reduced the training effect, and may explain the lack of efficacy. Post hoc analyses of compliant subjects showed no differences, but these subanalyses may not have been adequately powered to show a difference if one did exist.
Our results are not consistent with studies of fitness training in adults with RA that have shown significant benefits in fitness and functional outcomes (11, 12). They are, however, consistent with existing studies of exercise interventions in children with arthritis that have shown slight or no significant improvement in fitness or functional outcomes (15, 18, 25, 45).
Studies of exercise training in healthy children suggest that prepubertal children experience only modest improvements (<11%) in VO2max after training (49). Pubertal children are capable of achieving 20–25% increases in VO2max with training, which is similar to adults (49). Furthermore, it has been suggested that aerobic training of healthy prepubescent children may require training intensities of up to 90% MHR (50), which are significantly higher than those used in this study. Half of the subjects in this study were postpubertal, and post hoc analysis failed to show a response in VO2max or VO2submax in either the prepubertal or postpubertal groups, which does not support the idea that a differential training effect in these 2 groups could explain our results.
Our findings suggest that activity programs with or without aerobic training are safe and may result in improved physical function. This is supported by other studies of exercise programs in childhood arthritis (17, 18, 25, 45). A higher intensity of aerobic training did not provide any additional benefits in this study to functional outcome. This may be a result of the experimental program not being intensive enough to show a response; however, it would require additional studies of different training protocols to determine whether measurable benefits are achievable in this population. A high adherence rate of 80% in the qigong program may suggest that less intensive regimens are easier for children with JIA to comply with and seem to produce an equivalent degree of benefit as more intensive programs.
Dr. Feldman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.