The authors state that they have no conflicts of interest.
This 16-month randomized, controlled school-based study compared change in tibial bone strength between 281 boys and girls participating in a daily program of physical activity (Action Schools! BC) and 129 same-sex controls. The simple, pragmatic intervention increased distal tibia bone strength in prepubertal boys; it had no effect in early pubertal boys or pre or early pubertal girls.
Introduction: Numerous school-based exercise interventions have proven effective for enhancing BMC, but none have used pQCT to evaluate the effects of increased loading on bone strength during growth. Thus, our aim was to determine whether a daily program of physical activity, Action Schools! BC (AS! BC) would improve tibial bone strength in boys and girls who were pre- (Tanner stage 1) or early pubertal (Tanner stage 2 or 3) at baseline.
Materials and Methods: Ten schools were randomized to intervention (INT, 7 schools) or control (CON, 3 schools). The bone-loading component of AS! BC included a daily jumping program (Bounce at the Bell) plus 15 minutes/day of classroom physical activity in addition to regular physical education. We used pQCT to compare 16-month change in bone strength index (BSI, mg2/mm4) at the distal tibia (8% site) and polar strength strain index (SSIp, mm3) at the tibial midshaft (50% site) in 281 boys and girls participating in AS! BC and 129 same-sex controls. We used a linear mixed effects model to analyze our data.
Results: Children were 10.2 ± 0.6 years at baseline. Intervention boys tended to have a greater increase in BSI (+774.6 mg2/mm4; 95% CI: 672.7, 876.4) than CON boys (+650.9 mg2/mm4; 95% CI: 496.4, 805.4), but the difference was only significant in prepubertal boys (p = 0.03 for group × maturity interaction). Intervention boys also tended to have a greater increase in SSIp (+198.6 mm3; 95% CI: 182.9, 214.3) than CON boys (+177.1 mm3; 95% CI: 153.5, 200.7). Change in BSI and SSIp was similar between CON and INT girls.
Conclusions: Our findings suggest that a simple, pragmatic program of daily activity enhances bone strength at the distal tibia in prepubertal boys. The precise exercise prescription needed to elicit a similar response in more mature boys or in girls might be best addressed in a dose–response trial.
Childhood may be a key time to intervene with physical activity to improve bone strength. Furthermore, despite a lack of direct evidence, it is widely believed that early intervention may prevent osteoporosis and related fractures later in life, if activity is maintained.(1) To achieve impact at a population level, physical activity models must be feasible, affordable, and sustainable, and schools may provide the best means to reach large numbers of children from diverse ethnic and socioeconomic backgrounds. Previous school-based interventions showed that increased weight-bearing exercise effectively enhanced BMC and areal BMD as measured by DXA in boys and girls.(2–6) These studies contributed substantially to our understanding of how growing bone responds to exercise; however, all used planar DXA technology, which has inherent limitations.(7)
More recently, pQCT has been used to characterize how bone geometry, (volumetric) density, and bone strength adapt to loading. Results from cross-sectional pQCT studies in children,(8) adults,(9) and controlled studies in animals(10) suggest that weight-bearing exercise or mechanical loading improves bone strength at diaphyseal sites, mainly through changes in cross-sectional geometry rather than through changes in density. In contrast, distal sites seem to adapt to compressive forces through increased BMD.(8,9) To date, only three controlled trials evaluated the effects of high-impact exercise on the preadolescent skeleton using pQCT.(11–13) One trial in pre- and postmenarcheal girls showed that estimated bone strength (density-weighted polar section modulus) at the tibial midshaft changed similarly between intervention and control groups after a 9-month jumping intervention.(11)
To maximize the osteogenic potential of exercise, activities should be high impact and impose dynamic and abnormal strains of varying distribution on the skeleton.(14) In addition, frequent and short bouts of activity separated by a rest period maximize bone's response to loading in growing rats.(15) Recently, we designed a simple school-based program (Bounce at the Bell) based on these principles of bone adaptation and found that it enhanced proximal femur bone mass (by DXA) in a mixed maturity sample of boys and girls.(16) In this study, Bounce at the Bell provided the main bone-loading component within a larger program of daily physical activity, Action Schools! BC (AS! BC).(17,18)
Our primary objective was to determine whether a school-based physical activity model, AS! BC was effective for increasing tibial bone strength at the distal tibia and the tibial midshaft in boys and girls. We hypothesized that, compared with same-sex controls, boys and girls participating in AS! BC would have a greater increase in bone strength at both measured sites after the intervention.
MATERIALS AND METHODS
Study design and participants
This was a cluster randomized, controlled school-based intervention trial. Schools were the unit of randomization to prevent contamination that would occur if intervention and control children attended the same school. We recruited schools from two school districts in British Columbia, Canada (Vancouver, Richmond). We gave presentations at District principals' meetings, and of 103 schools, the first 20 (19%) schools (Fig. 1) that volunteered to participate were evaluated against our inclusion criteria, which was based on current school physical activity. To identify schools that were not already undertaking physical activity initiatives, we used results from the 2002 BC Ministry of Education Satisfaction Survey,(19) which assessed parent and student satisfaction with current school physical activity on a five-point Likert scale (5 = very satisfied). From the pool of 20 volunteer schools, we invited 11 schools that met our entry criteria (satisfaction scores that ranked 3 or lower) to participate. One principal withdrew his school (before randomization) after determining there was a chance his school could be randomly assigned to the control group.
Thus, 10 schools (3 Richmond schools and 7 Vancouver schools) participated in this study. To ensure both districts were represented in intervention and control groups, we first randomly assigned the Richmond schools to control (CON), level 1 intervention, or level 2 intervention. We stratified the seven Vancouver schools by size (<300 or >300 students) and randomly assigned them to the three groups. Randomization was performed remotely by a third party using random number draw. The intervention arms differed in the amount of facilitation provided to teachers and not in the activity delivered to students. Thus, the two intervention arms were collapsed into one group (INT, seven schools) for this analysis.
All children in grades 4 and 5 attending intervention schools took part in the AS! BC intervention. A total of 514 (47%) children (257 boys and 257 girls) received parental consent to participate in the evaluation. Three girls (one CON and two INT) and two boys (INT) were excluded from this analysis based on conditions that could affect normal physical activity or bone development (osteogenesis imperfecta, type I diabetes, fetal alcohol syndrome, childhood leukemia, brain tumor). All other children were healthy. Ethnicity classification was based on parents' or grandparents' place of birth. The majority of the children were Asian (53%), with both parents or all four grandparents born in Hong Kong or China, India, Philippines, Vietnam, Korea, or Taiwan. The remainder of the sample was white (35%), with parents born in North America or Europe, or were of mixed ethnicity or other ethnic origins (12%). The Clinical Research Ethics Board at the University of British Columbia and the Vancouver and Richmond School Districts approved this study.
Baseline measurements were performed between February and April 2003. The intervention spanned 1.25 school years (phase I: April–June 2003, phase II: October 2003–May 2004), and follow-up measurements were performed between April and June 2004. Thus, the total study period was 16 months, but the median follow-up time was 14.0 months (interquartile range = 13.8–14.1 months).
AS! BC intervention
The AS! BC model is described in detail elsewhere.(17,18) Briefly, AS! BC uses a socio-ecological approach to promote physical activity in elementary schools. The AS! BC model provides a tool for schools and teachers to create individualized Action Plans that increase physical activity opportunities across six Action Zones: (1) School Environment, (2) Physical Education, (3) Classroom Action, (4) Family and Community, (5) Extracurricular, and (6) School Spirit. The AS! BC model provides teachers with training and resources to operationalize their Action Plan with the ultimate goal of providing students with 150 minutes of moderate intensity physical activity per week. This corresponds to the BC Ministry of Education and Canadian curriculum guidelines for physical education.(20,21)
For this study, we introduced a two-component targeted bone-loading program into the Classroom Action Zone of AS! BC. Teachers at INT schools delivered this program in addition to two 40-minute classes (on average) of regular physical education per week. For the first component, teachers at INT schools were required to provide their students with an additional 15 minutes of physical activity for 5 days a week. Teachers chose from a number of different activities including skipping, dancing, playground circuits, and simple resistance exercises with exercise bands. All activities required minimal equipment and could be performed in the classroom, hallway, or on the school playground. Teachers were given a Classroom Action Bin that contained equipment and resources to facilitate these activities.
The second required component was Bounce at the Bell. We asked INT teachers to implement this short activity (∼3 minutes/day) in their classroom three times a day (at the morning, noon, and end of day school bell) for 4 days a week. Teachers instructed students to perform either counter movement jumps (two foot take off, clutch knees, two foot landing) or side to side jumps (one foot take off, opposite foot landing).(16,22) During phase I, students performed 5 two-foot landing jumps (or 10 one-foot landing jumps) at each session. During phase II, teachers were instructed to increase the number of jumps (starting from 5 per session) over each month of the school year, until a maximum of 36 jumps per day was achieved. Children in CON schools participated in their regular program of physical education, which involved two 40-minute physical education classes per week, on average.
In-school training of teachers at INT schools (N = 48 across phases I and II) was conducted by the AS! BC Support Team.(17) To monitor compliance with program delivery, we asked INT teachers to complete activity logs. Teachers recorded the type, frequency, and duration of each activity undertaken with their class each day of the school week. Intervention teachers also recorded the number of sessions of Bounce at the Bell and the number of jumps per session that their students performed each day. The physical activity level of CON schools was also monitored using a modified version of the activity log.
Descriptive and independent variables
Standing height was measured to the nearest 0.1 cm using a wall-mounted digital stadiometer (Model 242; Seca, Hannover, MD, USA), and body weight was measured to the nearest 0.1 kg using an electronic scale (Model BWB-800P; Tanita, Arlington Heights, IL, USA). Length of the left tibia was measured as the distance from the distal edge of the medial malleolus to the tibial plateau (to the nearest 0.1 cm) using an anthropometric tape. For each variable, the mean of two measures is reported. Muscle cross-sectional area (MCSA, mm2) was assessed at the proximal two-thirds site (66% of total tibial length) of the left tibia with pQCT (XCT-2000; Stratec Medizintechnic, Pforzheim, Germany).
Maturity status was assessed at baseline and follow-up using self-report Tanner staging (breast stage for girls, pubic hair stage for boys).(23) Children were provided with line drawings of the five Tanner stages and were instructed by a research assistant to choose the drawing that best represented their current stage of development. Children completed the form in a private setting away from the other children. Boys and girls were classified as prepubertal if they were Tanner stage 1 at baseline and early pubertal if they were Tanner stage 2 or 3 at baseline. Girls' menarcheal status at baseline and follow-up was determined by self-report questionnaire. To estimate lower limb power, maximal height (cm) for vertical jump was measured using the Vertec device (Fitness Source, Concord, Ontario, Canada), and maximal distance (cm) for standing long jump was assessed according to standard protocol.(24) A modified version of the Physical Activity Questionnaire for Children (PAQ-C)(25,26) was used to assess self-reported physical activity. A general physical activity score (PA Score) was calculated as an average of the nine PAQ-C items in a continuous range between 1 (low activity) and 5 (high activity) and an estimate of time (h/week) spent in common sports and activities designated as loaded (impact > walking, load time) was generated from item 1. A validated food frequency questionnaire (FFQ)(27) was used to determine dietary calcium (mg/day). The PAQ-C and FFQ were administered at baseline and follow-up plus three additional times during the study period (June 2003, September 2003, and January 2004). The average across the five reports is presented for PA Score, load time, and dietary calcium.
Primary and secondary outcomes
To address the primary objective, we used pQCT (XCT-2000) to assess change in bone strength index (BSI, mg2/mm4) at the distal tibia and change in polar strength strain index (SSIp, mm3) at the tibial midshaft. Briefly, a 30-mm planar scout view was acquired over the joint line to locate a standard anatomical reference (distal surface of the tibial plafond). The distal site (8% of total tibial length) and the tibial midshaft (50% of total tibial length) were measured proximally from this reference line (2.3 ± 0.2-mm slice thickness, 0.4-mm voxel size, scan speed 30 mm/s). One trained technician, who was not blinded to group assignment, acquired all scans. A phantom was scanned daily to maintain quality assurance. In our laboratory, short-term precision for 14 subjects (mean age, 12–27 years) ranged from 0.72% (ToA) to 0.94% (CoA) at the midshaft.
All scans were analyzed using Stratec software, Version 5.5,(28) by the same trained technician who acquired the scans. At the distal tibia, contour mode 1 (200 mg/cm3) was used to determine secondary outcomes of total bone cross-sectional area (ToA, mm2) and total BMD (ToD, mg/cm3), which in turn were used to calculate BSI (ToA × ToD2). At the midshaft, separation mode 1 (480 mg/cm3) was used to determine SSIp. The SSIp is calculated as the integrated product of the section modulus and CoD and provides an estimate of torsional bone strength (Fig. 2).(29) Secondary outcomes at the midshaft were total area (ToA, mm2), cortical area (CoA, mm2; separation mode 1, 480 mg/cm3), and cortical density (CoD, mg/cm3; separation mode 1, 711 mg/cm3). The lower threshold of 480 mg/cm3 was chosen for CoA to match the threshold used for the primary outcome (SSIp), whereas 711 mg/cm3 was chosen for CoD to minimize the partial volume effect.
None of the three school-based trials reported to date that used pQCT(11–13) showed a significant intervention effect for BSI or SSIp in boys or girls. Thus, sample size for this study was determined from our previous trial that showed a 4% greater improvement in estimated bone bending strength at the femoral neck (section modulus by hip structure analysis) for intervention girls compared with control girls after a 7-month jumping intervention.(30) Based on a 2:1 randomization (level 1 and 2 intervention schools collapsed), 80% power, a type I error rate of 5% (two-sided), and an SD of 5%, a total of 60 children were required. To allow for within-sex and between-maturity group comparisons and a 10% attrition rate, we needed 264 children (across the 10 schools). However, all children in grades 4 and 5 in each of the 10 schools were invited to participate, and the consent rate (47%) was greater than expected. Thus, a larger number of children (n = 514) were randomized to control and intervention groups.
Our sample size calculation does not account for the clustered study design. However, we did account for clustering within our statistical analysis. We used a linear mixed effects model to compare the change in primary outcomes (BSI, SSIp) and secondary outcomes (distal tibia: ToA, ToD; midshaft: ToA, CoA, CoD) between intervention and control children. Group (intervention or control) was the fixed effect, and school was the random effect. Separate mixed effects models were created for boys and girls because of known differences in the tempo and timing of growth and maturation between sexes.(23) Covariates (independent variables) were chosen based on known biological and biomechanical relationships with the primary and secondary bone outcomes and relationships established in cross-sectional analyses described previously.(31) For boys, covariates in the final mixed effects models were baseline bone value, tibia length change, and MCSA change. For girls, covariates were baseline bone value, tibia length change, MCSA change, and final Tanner stage. Final Tanner stage was included in the models for girls based on significant univariate relationships with the primary and secondary bone outcomes (data not shown).
Based on results from previous studies that identified differences in the bone response to increased loading between maturity groups,(30,32) the effect of the intervention was also evaluated between maturity groups (PRE, EARLY) by including an interaction term (group × maturity category) in each model. Main effects were considered significant at p < 0.05, and interactions were considered significant at p < 0.1. Because of the relatively small sample size of the subgroups, we used a liberal criterion to avoid missing a potentially important interaction. We used standard residual plots to assess normality, linearity, and homoscedascity, and we identified outliers using Cook's distance values. We performed all analyses according to randomization and did not adjust for multiple comparisons. We calculated the intraclass correlation coefficient (ICC) for each bone outcomes (change values) as:
where sc2 equals the variance between clusters, and sw2 equals the variance within clusters.(33)
Participants and compliance with intervention delivery
We provided the flow of participants through the trial as per CONSORT guidelines(34) (Fig. 1). This analysis includes 410 children (281 INT: 145 boys, 136 girls; 129 CON: 64 boys, 65 girls) who had pQCT scans at baseline and follow-up.
Our progressive intervention was introduced in phase II, and we provided teacher compliance during this time frame. Median compliance with completion of Activity Logs was 97% (interquartile range [IQR]: 89–100%) across CON schools and 94% (IQR: 92–100%) across INT schools. Teachers at INT schools delivered ∼60 minutes more physical activity per week than teachers at CON schools (+58.9 minutes/week; 95% CI, 25.4–92.4). Median compliance with Bounce at the Bell was 74% (IQR: 50–89%) across INT schools. We were unable to assess individual student compliance with Bounce at the Bell. However, intervention teachers delivered Bounce at the Bell and other Classroom Action activities to all students in their classrooms, regardless of whether students had volunteered to be evaluated. Student attendance was determined from school records and averaged 96% across all schools during the study.
Descriptive and independent variables
At baseline, the majority of boys were prepubertal (67%), whereas the majority of girls were early pubertal (60%; Table 1). Despite randomization, slight imbalances appeared in descriptive characteristics between groups at baseline. Boys in the CON group tended to be heavier and have larger MCSA than INT boys. Regarding change in descriptive characteristics, INT boys tended to have a greater improvement in long jump (7.6% versus 5.2%) and vertical jump (16.6% versus 9.9%) performance compared with CON boys. Change in Tanner stage was similar between INT and CON boys (Table 2).
Table Table 1.. Baseline and Change (Where Appropriate) in Descriptive Characteristics for Control and Intervention Boys and Girls
Table Table 2.. Tanner Stage at Baseline and Follow-Up for Control and Intervention Boys and Girls
At baseline, INT girls tended to be taller, heavier, and have larger MCSA than CON girls. At follow-up, INT girls tended to have smaller gains in tibial length (4.9% versus 5.5%), weight (16.7% versus 18.5%), and MCSA (12.9% versus 15.6%) and a greater improvement in long jump performance (7.6% versus 3.4%) than CON girls. Change in Tanner stage was similar between INT and CON girls (Table 2), and a similar proportion of INT and CON girls were postmenarcheal at follow-up (Table 1).
Baseline, follow-up, and adjusted difference for change in primary and secondary pQCT outcomes for boys and girls are presented (Table 3). We excluded two boys (one CON and one INT) from the distal tibia analysis and one INT boy from the tibial midshaft analysis. We based our decision to exclude these scans on the fact that Cook's distance values for these boys were 2.5 times greater than the next closest value, and exclusion of their values resulted in a substantial increase in the estimated intervention effect (∼30%) and the SE of the intervention effect (∼15%). There were no discernible movement artefacts or errors in pQCT analysis for these scans.
Table Table 3.. Baseline, Follow-Up, and Adjusted Difference in Change in Distal Tibia and Tibial Midshaft pQCT Outcomes for Control (CON) and Intervention (INT) Boys and Girls
At baseline, bone strength indices (unadjusted) were 2–5% greater in CON boys than INT boys (Table 3). At the distal tibia, adjusted change in BSI tended to be greater for INT boys (+774.6 mg2/mm4; 95% CI: 672.7, 876.4) than CON boys (+650.9 mg2/mm4; 95% CI: 496.4, 805.4), but this difference was not statistically significant. There was, however, a significant group × maturity interaction for change in BSI (Table 4; Fig. 3). Prepubertal INT boys had a significantly greater increase in BSI compared with prepubertal CON boys, whereas there was no significant difference in change between early pubertal INT and CON boys.
Table Table 4.. Adjusted Difference in Change (Intervention – Control) in pQCT Outcomes at the Distal Tibia (8% Site) and Tibial Midshaft (50% Site) for Pre- (PRE) and Early-Pubertal (EARLY) Boys
At the tibial midshaft, adjusted 16-month change in SSIp tended to be greater for INT boys (+198.6 mm3; 95% CI: 153.5, 200.7) than CON boys (+177.1 mm3; 95% CI: 182.9, 214.3), but the difference in change between groups was not significant. The group × maturity interaction was not significant for SSIp at the tibial midshaft; however, there was a tendency for prepubertal INT boys to have a greater gain in SSIp compared with prepubertal CON boys (Table 4).
Baseline values for bone strength indices (unadjusted) were 4–5% greater in INT girls compared with CON girls. At the distal tibia, adjusted change in BSI tended to be greater in INT girls (+918.4 mg2/mm4; 95% CI: 844.2, 992.6) than CON girls (+829.2 mg2/mm4; 95% CI: 721.5, 936.9); however, this difference was not statistically significant (Table 3). At the tibial midshaft, there was no difference in change between INT and CON girls for SSIp. Furthermore, there were no significant group × maturity interactions for change in bone strength indices at either site (data not shown).
At baseline, distal tibia ToA (unadjusted) was similar between CON and INT boys, whereas ToD was ∼3% greater in CON boys than INT boys. Control boys tended to have a greater change in ToA than INT boys, whereas the opposite was true for ToD (Table 3). Similar to results for BSI, a significant group × maturity interaction was found for ToD (Table 4). Prepubertal INT boys had a significantly greater increase in ToD (+3.2 mg/cm3; 95% CI: −0.03, 6.5) compared with CON boys, who showed a decrease in ToD (−3.3 mg/cm3; 95% CI: −8.2, 1.6). There was no significant group × maturity interaction for change in ToA.
At the tibial midshaft, baseline values for ToA, CoA, and CoD were similar between CON and INT boys. There was a tendency for INT boys to have a greater gain in CoA; however, the difference in change between groups was not significant (Table 3). There were no significant group × maturity interactions for change in ToA, CoA, or CoD (Table 4), although prepubertal INT boys tended to have a greater increase in CoA (+29.5 mm2; 95% CI: 26.8, 32.2) than prepubertal CON boys (+25.8 mm2; 95% CI: 22.0, 29.6).
Baseline values for distal tibia ToA and ToD were 1–2% greater in INT girls than CON girls. Similar to boys, adjusted change in ToA tended to be greater in CON girls (+62.0 mm2; 95% CI: 53.5, 70.6) than INT girls (+53.8 mm2; 95% CI: 48.0, 59.5), and change in ToD tended to be greater in INT girls (+12.5 mg/cm3; 95% CI: 9.2, 15.7) than CON girls (+8.3 mg/cm3; 95% CI; 3.4, 13.2), but the differences were not statistically significant (Table 3). There were no significant group × maturity interactions for change in ToA or ToD.
At the tibial midshaft, baseline values for ToA and CoA were 3% greater for INT than CON girls, whereas CoD was similar between groups. There was no significant difference in change between groups for ToA, CoA, or CoD (Table 3), nor were there any significant group × maturity interactions for change in ToA, CoA, or CoD.
This is the first study to prospectively study the effects of increased physical activity on estimated tibial bone strength in both boys and girls using pQCT. Although the results of the subgroup analysis (group × maturity interaction) should be interpreted with caution, our findings suggest that AS! BC was effective for increasing bone strength at the distal tibia in prepubertal boys; it was not effective for increasing bone strength at the distal tibia or tibial midshaft in pre- or early-pubertal girls or early-pubertal boys. The novel aspects of this study included (1) the unique and pragmatic intervention and (2) the use of pQCT to estimate bone strength changes in pre- and early-pubertal boys and girls.
AS! BC: a unique intervention
The AS! BC model aimed to increase physical activity opportunities within the classroom setting and provided generalist teachers with ideas to achieve this without the need for special equipment or access to the school gymnasium. Within the broader activity component (Classroom Action), Bounce at the Bell was a simple bone-loading exercise that required only 3 minutes each school day to implement.
The osteogenic potential of physical activity is determined by the magnitude of the external load, the dynamic nature of the load, the rate at which the load is introduced, and the duration of the loading bout.(14) Previous school-based studies have incorporated these principles into the design of physical activity interventions, and some have proven effective for increasing bone mineral accrual(3,4,6,11) and estimated bone strength(2,30,35) in boys and/or girls. However, in these trials, the duration of the activity sessions ranged from 10 to 90 minutes, and the sessions were performed once daily, two to three times weekly. In contrast, animal studies have shown that multiple, short bouts of loading interspersed with recovery periods are equally effective as longer bouts.(15,36) Bounce at the Bell, a program of short bouts of high-impact jumping separated by rest periods, was designed to mimic these findings and, in a pilot study, proved effective for increasing proximal femur bone mass in boys and girls 9–11 years of age.(16)
In this study, although intervention children performed Bounce at the Bell, they also performed a variety of other weight-bearing activities such as skipping and running. Also, both boys and girls in this cohort were already active performing 6 h of weight-bearing physical activity per week, on average, and many of the activities involved running. Thus, we cannot rule out the contribution of these other weight-bearing activities to bone adaptation. Activities such as running and skipping impose greater loads on the skeleton than does walking,(37,38) but the strains may not be sufficiently unusual in magnitude and/or distribution to elicit an osteogenic response at the tibia. In contrast, the maximum ground reaction forces (GRFs) of the two-foot countermovement jumps used in Bounce at the Bell are approximately five times body weight.(22) In addition, data from human strain gauge studies indicate that multidirectional activities similar to the jumps performed in Bounce at the Bell produce higher strains and strain rates and more unusual strain distributions than walking or running.(39) Thus, although it is likely that Bounce at the Bell provided the most consistent and unique bone-loading stimulus in the AS! BC intervention, this stimulus was either not of sufficient magnitude (given the highly active children in this study) or not performed enough times or for a long enough duration to elicit a consistent benefit at all sites in both sexes. A dose–response study that evaluates the effect of different exercise regimens on the growing skeleton would provide more definitive answers as to the type of program required to elicit consistent bone strength benefits for girls and boys.
pQCT: a novel imaging modality
To our knowledge, three intervention studies have used pQCT to evaluate the effects of increased physical activity on the growing skeleton.(11–13) There were a number of distinct differences between these trials that prevent comparison between them. These differences include the age and sex of participants, length and type of the intervention, pQCT acquisition and analysis protocols, bone sites measured, and outcome variables reported. This highlights the need for standardized analysis, acquisition, and reporting protocols for pediatric pQCT studies.
Compared with these previous pQCT interventions,(11–13) this study is unique in that we evaluated change in estimated bone strength at both the distal tibial and tibial midshaft. At the distal tibia, which experiences mainly compressive forces, bone strength is dependent on apparent density of the bone tissue.(40) In addition to a relatively dense trabecular bone structure, a large cross-sectional area would be optimal to resist compressive loads.(9) Thus, we estimated bone strength at the distal tibia using BSI—a measure that incorporates both bone (volumetric) density and cross-sectional area.(9)
Given the strong association between compressive bone strength and apparent density,(40) it is not surprising that the greater gain in BSI in prepubertal INT boys was associated with a greater gain in ToD rather than ToA. Total density reflects the contribution of both trabecular density (TrbD) and CoD; however, CoD cannot be measured accurately at the distal tibia because of limitations associated with the thin cortical shell (<2 mm, on average, data not shown).(41) Prepubertal intervention boys did tend to have a greater change in TrbD than prepubertal control boys (data not shown). These results are in accordance with previous cross-sectional pQCT findings in young gymnasts(8) and racquet-sport players.(9)
In contrast, the tibial shaft experiences mainly bending and torsional forces, and resistance to such forces is determined primarily by the distribution and material stiffness of cortical bone.(42) Therefore, we estimated torsional bone strength at the tibial midshaft using pQCT-derived SSIp.(28,29) At the tibial midshaft, the group × maturity interaction was not statistically significant in boys or girls; however, prepubertal intervention boys tended to have a greater increase in SSIp than prepubertal control boys. In an efficacy subgroup analysis that included only prepubertal intervention boys in classes that were at least 80% compliant with Bounce at the Bell, the intervention effect for SSIp was significant (data not shown). Although this finding must be interpreted with caution because of bias associated with efficacy subgroup analyses and the small number of intervention boys included (n = 35), it does suggest that teacher compliance is crucial to the effectiveness of the intervention.
Sex and maturity specificity of the bone response
Although boys and girls in intervention schools received the same program of physical activity, AS! BC was effective for increasing tibial bone strength in prepubertal boys only. There are several possible explanations for the lack of an intervention effect in girls. First, previous school-based interventions have reported challenges in promoting physical activity among girls.(43) Thus, it is possible that girls were less motivated to participate in Bounce at the Bell and other Classroom Action activities than boys. Unfortunately, it was not possible to directly observe and quantify children's participation in Bounce at the Bell and other activities for this study. The trend for intervention girls to have a greater increase in distal tibia bone strength suggests that the program has the potential to elicit an osteogenic response in girls. Direct observation of classroom activities, if possible, may be warranted to clarify possible sex differences in exercise participation.
Second, it has been proposed that, in the female skeleton, estrogen modulates the bone response to loading indirectly through two mechanisms. Estrogen may modify bone structure in girls through increased cortical density and relatively thicker cortices compared with boys. Girls' bone would therefore be more stiff, and the magnitude of deformation for a given load would be diminished.(44,45) In this study, CoD was slightly greater in girls than boys at baseline, and girls tended to have a greater increase in CoD than boys. This difference likely reflects differences in maturation as 67% of girls were early pubertal and 60% of boys were prepubertal. It is possible that the strains associated with the present intervention were not sufficient to elicit an osteogenic response at the tibial midshaft in girls.
Heinonen et al.(11) implemented a more intensive jumping program (20 minutes of drop jump-training, 2 times per week) in their 9-month controlled trial, and similar to this study, did not observe an increase in bone strength at the tibial midshaft in intervention girls. However, based on current evidence from animal models, this exercise program of infrequent and longer loading bouts would not be expected to maximize the adaptive response.(14) It may be necessary to include more multidirectional jumps or zig-zag hopping in Bounce at the Bell. Milgrom et al.(39) found this type of activity was associated with the highest principal compression, tension and shear strains, and compression strain rates compared with activities such as walking, jogging, one- or two-leg vertical jumps, and unidirectional hopping in adults.
In boys, AS! BC was effective for increasing tibial bone strength in prepubertal, but not early pubertal, boys. Of the few exercise interventions that evaluated the bone response to loading in boys,(2,5,35) all have included only boys who were prepubertal at baseline. Results from this study suggest that prepuberty may offer a “window of opportunity” for adaptation at the tibia similar to that described for girls at the femoral neck during early puberty.(30) However, these findings should be interpreted with caution given the relatively small number of early pubertal boys. Importantly, this is the first intervention study of boys and girls to examine tibial bone strength by pQCT. Mean values for bone strength may be used to design future trials that more specifically address the role that maturity plays in the bone response to weight-bearing physical activity.
An alternative explanation for the maturity-specific intervention effect may be related to the boys' physical activity levels. Prepubertal intervention boys tended to report less leisure-time weight-bearing physical activity compared with early pubertal intervention boys (6.1 versus 7.7 h/week). Because early pubertal intervention boys were already participating in >1 h of weight-bearing physical activity per day, it is possible that a bone-loading program that was more intense, more frequent, or a of longer duration than Bounce at the Bell may have been required to elicit an osteogenic response in this more mature group. To further study group × maturity interactions within the clustered study design, a larger number of schools would be required to ensure adequate power for subgroup analyses.
There are several methodological limitations of this study. First, although the cluster design was managed by our statistical approach, sample size was calculated without accounting for the clustered study design. Based on the ICC for change in the primary pQCT outcomes, a conservative estimate of the ICC in this cohort is 0.05. This estimate can be used to determine appropriate sample sizes for future school-based trials; however, it should be noted that, because of the small number of schools, the precision of this estimate is limited.(46) In addition, results from the mixed linear model should be interpreted with caution because of the small number of schools. To achieve 80% power with an ICC of 0.05, ∼12 clusters per group are required according to recent recommendations.(47)
Second, an inclusion criterion for this study was that participating schools could not be undertaking any formal school-based physical activity initiatives (outside of regular school PE) as indicated by results from the province-wide Satisfaction Survey.(19) Selecting these schools may have introduced bias into the study because the schools may have been more, or less, eager to participate in the intervention. In addition, schools were from two urban geographic regions and may not represent the general school population in the province of British Columbia. Thus, the generalizability of our results is limited.
Third, this cohort was ethnically diverse, and >50% of the children were of Asian descent. We previously reported differences in tibial bone geometry and density between Asian and white children.(31) However, it was not the aim of this study to investigate possible differences in the bone response to physical activity between ethnic groups. Furthermore, the distribution of ethnicities was similar between intervention and control groups, so it was unlikely that ethnicity had an influence in this study. To assess ethnic differences in skeletal adaptations to a school-based intervention, there is a need to determine the ICC for pQCT outcomes across ethnic groups.
Fourth, in any study of growing children, it is important to evaluate children based on maturational status rather than chronological age. However, accurate classification of children by maturity remains a challenge. We used a self-assessment of Tanner stage that has been shown to correspond relatively well with physician's ratings of maturity.(48–50) We also chose to overcome the challenge of single stage classification by collapsing children in Tanner stages 2 and 3 into one early-pubertal group.
Fifth, teacher compliance with the activity logs was excellent (94%); however, overall compliance with daily bouts of Bounce at the Bell was lower (74%; IQR: 50–89%). Given the competing curricular demands on teacher's class time, we are encouraged by these results. In a separate study, we conducted a process evaluation to determine the barriers to undertaking a school-based model of physical activity and the most often cited limitations were competing curriculum demands, lack of preparation time, and needing a supportive school environment.(17) In addition, although we did not assess each child's individual participation in the study—teachers' reported providing physical activity opportunities to all students—not only those that volunteered for the evaluation component.
Finally, long-term reproducibility of pQCT measurements may be influenced by disproportionate longitudinal growth of the tibia caused by a greater contribution of the proximal growth plate (60%) compared with the distal growth plate (40%).(51) As a result, it is not possible to determine the same exact location along the length of the tibia over time. Given the average increase in tibial length in this study (1.7 cm) and the relative contribution of distal and proximal regions, the measurement site would have shifted ∼0.7 mm, on average, over the 16-month study. This is within the 2.3-mm slice thickness. However, we tried to reduce this small effect by using a fixed anatomical landmark to locate the same relative region along the tibia length at the follow-up measurement.
Implications of the findings
The findings we observed in prepubertal boys are promising, given the simple, accessible, and inexpensive nature of the intervention. Future studies that examine the bone strength response (by pQCT) to different doses of weight-bearing physical activity are overdue as, to our knowledge, this has not as yet been undertaken in any age group. It is not known if the observed gains in tibial bone strength will persist in the absence of continued exercise. However, results from two follow-up studies of school-based interventions(52,53) provide support for the short-term (12 months) maintenance of gains in DXA-derived BMC after cessation of an exercise program. Follow-up of the AS! BC cohort is needed to determine whether the bone strength advantage will be preserved in prepubertal boys.
Elementary schools are currently confronted with declining resources for physical education, the absence of physical education specialists, and increasing curricular demands on generalist teachers. The AS! BC model offers a simple classroom-based option for elementary schools to increase physical activity opportunities for all students regardless of skill level. Teachers were provided with training, resources and the opportunity to design their own Action Plan; nevertheless, there were barriers to daily implementation of the model.(17) Innovative strategies that assist teachers to overcome these barriers would likely enhance the opportunity for teachers to provide school-based physical activity for their students.
This study suggests that the Classroom Action component of AS! BC, which includes Bounce at the Bell, was an effective means to enhance bone strength as estimated with pQCT at the distal tibia in prepubertal boys. The physical activity program was not similarly effective in girls or early pubertal boys. Future studies that evaluate different types, bouts, and intensities of exercise are recommended to determine the type of exercise that is most effective for increasing bone strength in both boys and girls.
The authors thank the support and participation of the students, parents, teachers and administrators from the 10 participating schools. We appreciate the statistical guidance of Dr Penny Brasher of the Centre for Clinical Epidemiology and Evaluation. We also thank the AS! BC Support Team led by Bryna Kopelow and Jennifer Fenton of JWSporta. We are also grateful to the BC Ministry of Health, 2010 Legacies Now, BC Ministry of Tourism, Sport and the Arts, and the Canadian Institutes of Health Research for funding support. We also acknowledge that Dr McKay is a Michael Smith Foundation for Health Research (MSFHR) Senior Scholar, Dr Kontulainen is a MSFHR Post-doctoral fellow, and Dr Khan is a Canadian Institutes of Health Research New Investigator.