Relationships between physical activity (PA), cardiorespiratory fitness (CRF), and health outcomes are well established in adults (5). It is generally believed that PA and CRF independently affect health outcomes for children and adolescents (29), yet the correlation between PA and CRF is generally low in this population. Although a significant proportion of the recent literature has focused on PA in children and adolescents, CRF remains an important consideration because it reflects many aspects of physiologic function and performance.
It is known that CRF, relative to body mass, declines during adolescence, particularly in girls. Both cross-sectional and longitudinal data indicate that for adolescent girls, CRF declines 2-5% per year (3,11,15). Data from McMurray et al. (19) suggest that African American girls show greater decline in CRF than do white girls ages 8-16. No study, however, has addressed factors related to how CRF may naturally change across time during the transition from middle school through high school with longitudinal analyses, particularly in a biracial sample. Further, although a few studies have used more sophisticated analyses (22), many have investigated the potential influence that correlates from an earlier time point have on predicting fitness at a later time point.
Minimizing the decline over time in CRF is of interest because of the potentially positive impact on future health outcomes. It is important to examine how factors related to CRF interact over time. Identification of factors that potentially modify CRF decline is important for future interventions so that investigators know which variables to target. The purpose of this study was to examine whether physical activity, body mass index, sports participation, and race made a significant contribution to change across time of cardiorespiratory fitness in girls followed from middle school to high school, using growth curve analysis.
Data were collected in two successive cohorts of girls longitudinally during a 4-yr time period as part of a physical activity intervention known as LEAP (Lifestyle Education for Activity Program) (27). Eighth-grade girls participated in baseline measures, and follow-up measures occurred when the girls were in 9th and 12th grades. Thirty-one middle schools were involved in the 8th-grade measures. One middle school (that included 9th grade) and 23 high schools were involved in the 9th-grade measures, and 22 high schools were involved at 12th grade. Between the 9th and 12th grade, two schools merged, causing the loss of one school. School-level percentage of students receiving free or reduced lunch varied from 12 to 64%. Participants from both control and intervention schools were included in this investigation. Trained data collectors used standardized procedures and scripts when administering measures to small groups of girls. All measures were taken in the months of January through May for each year. The study was approved by the University of South Carolina institutional review board. Each participant and parent/guardian (when girls were minors) provided written informed consent.
Participants were adolescent girls who completed a measurement protocol at all three time points (8th, 9th, and 12th grades). A total of 2744 girls (average age = 13.6 ± 0.6 yr, 49% African American) participated in the measurement protocol in 8th grade. The same girls were asked to complete the protocol again in 9th and 12th grades. Girls that had missing data, were absent on one or more days of data collection, had tests that were subject to equipment and/or administrator error during fitness testing, or were outliers were removed from analyses. The data reported in this paper represent 274 girls (13.6 ± 0.6 yr at baseline, 59% African American) who provided complete data on fitness, physical activity, BMI, and sport team participation at all three time points. We tested for differences between the current sample and the remaining members of the sample for whom data were not available at all three time points and found no significant differences for fitness, BMI, MVPA, or VPA. There were significantly more sport participants and African Americans in the current sample.
Cardiorespiratory fitness was assessed using a modified physical working capacity test that predicted the workload at a heart rate (HR) of 170 bpm (PWC170). Each test stage lasted 2 min, with the initial workload for all girls, regardless of body mass, at 0.5 kg. Pedaling frequency throughout the test was 60 rpm, and the girls were instructed to follow a metronome set to 120 bpm to maintain proper cadence. Research assistants kept careful watch to ensure that participants maintained the cadence, and they recorded if cadence was not maintained. At the end of the second minute of each testing stage, the trained research assistant recorded rpm and HR, which was measured via a heart rate monitor (Polar Electro Inc., Lake Success, NY) placed around the chest at the level of the xiphoid process of the sternum. The participant's HR at the end of a stage was used to determine the change in resistance for the next stage. Girls completed three stages of testing to obtain a final stage HR of 160 bpm or higher. In some cases, a fourth stage was necessary to meet the 160-bpm criterion. Girls were verbally encouraged to maintain the pedaling frequency throughout the test. Workloads were expressed as values relative to body weight (weight-relative PWC170; kg·m·min−1·kg−1) and as absolute values (absolute PWC170; kg·m·min−1).
Body mass index.
Height was measured twice to the nearest 1.0 cm with a portable stadiometer, and weight was measured twice to the nearest 0.1 kg with a digital scale; the average of both measures was used for analyses. Both height and weight were measured with participants wearing light clothing and no shoes. Body mass index (BMI) was calculated by dividing weight in kilograms by height in meters squared. The CDC cut points were used to classify girls as normal weight or < 85th percentile, separately at each measurement (16).
Physical activity recall.
The 3-Day Physical Activity Recall (3DPAR) was used to measure participation in physical activity. Scores from the 3DPAR have been validated by concurrent comparison with accelerometer-derived variables for total METs, moderate-to-vigorous physical activity (MVPA), and vigorous physical activity (VPA). Correlations for both 3 d of accelerometry and 7 d of accelerometry were significant for all variables (r = 0.27-0.46 for 3 d and r = 0.35-0.51 for 7 d) in a sample of 70 8th- and 9th-grade girls (54.3% white, 37.1% African American) (25).
The 3DPAR was administered to the subjects in the spring of their 8th-, 9th-, and 12th-grade years. The instrument was always administered on a Wednesday, with girls recalling activities from the immediately preceding Tuesday, Monday, and Sunday. Each day is segmented using a grid format into 30-min time blocks (7:00 a.m. to midnight), which, in turn, are grouped into morning, afternoon and evening time periods. Subjects were asked to report the predominant activity in each of the 30-min blocks. A list of 55 activities, grouped into the categories of eating, work, after school/spare time/hobbies, transportation, sleep/bathing, school, and physical activities and sports was provided, along with a space to write in other. Four more activities (gymnastics/tumbling, kickboxing/Tae Bo, track and field, and trampoline) were added to the list at 12th grade because of the frequency of write-ins at the 8th and 9th grades.
The 3DPAR uses a script, definitions, and graphic figures to explain the intensity of common activities. Light activities are described as requiring little or no movement with slow breathing, moderate activities as requiring some movement and normal breathing, hard activities as requiring moderate movement and increased breathing, and very hard activities as requiring quick movements and hard breathing. Data from each 30-min time block were assigned an MET intensity from the compendium (1) to classify moderate-to-vigorous physical activity (MVPA; ≥ 3 METs) and vigorous physical activity (VPA; ≥ 6 METs). The number of blocks for each intensity (MVPA, VPA) was summed. Then girls were categorized as having on average, two or more blocks of MVPA per day and one or more blocks of VPA per day across the 3 d. Girls who did not meet the criteria of two or more blocks of MVPA per day or one or more blocks of VPA per day were categorized as inactive; girls who met either criterion or both criteria were classified as active.
Sports team participation.
On a separate questionnaire, two items asked about sports team participation during the past year. The questions were adapted from the Youth Risk Behavior Surveillance System survey (7). The first inquired about participation on sports teams run by the school and the second about those run by organizations outside of school. The answers to these two questions were summed, and sports team participants were those girls who reported participating on one or more teams during the previous year.
All analyses were performed using SAS statistical software (version 8.2, SAS Institute, Cary, NC; PROC MIXED), and alpha was set at 0.05 for main effects. To determine outliers, all participants' HR-workload relationships were plotted for all stages of the PWC170. Any participants with R 2 < 0.80 were removed from the sample. Means (SD) were calculated for all variables. Chi-square analyses and t-tests were used to determine whether there were differences for the fitness, BMI, physical activity, and sports team participation variables between African American and white girls at all three time points.
To examine the change over time in CRF, growth curve analysis (hierarchical linear modeling) was used. This type of analysis provides an advantage over traditional repeated-measures analysis because it simultaneously takes into account means and covariances of repeatedly measured variables. It examines both change at the individual level (within-person change over time) and at the group level (between-person change over time) and involves two levels of analysis. The first level shows growth (change) over time by fitting slopes at the individual level. The unit of measurement is each subject's initial intercept and slope, with slope being the unit of analysis (it represents the expected change each participant will experience over time). A second level of analysis involves relating predictors to interindividual differences in change. Thus, the first step is to establish a baseline (unconditional) model that does not include predictor variables and that serves as a basis for comparison. The next step in the analysis was to establish models for single predictors of CRF (race, BMI, VPA, MVPA, sport participation) to determine significant predictors. The final step of the analysis involves fitting a model that controls for all variables at all time points. Time was coded as 0 (8th grade), 1 (9th grade; 9 minus 8), or 4 (12th grade; 12 minus 8) years to account for the variably spaced measurement occasions. The fitness variable and BMI were continuous, and race, MVPA, VPA, and sports participation were categorical. To assist with interpretation of the models, BMI was centered. To center the data, we subtracted the age-and sex-specific 85th percentile for BMI value (using the Centers for Disease Control criteria (16)) from each girl's BMI. Thus, positive values indicate BMI above the 85th percentile, and negative values indicate BMI below the 85th percentile. All models were run controlling for intervention status, which was not significant in any model.
Six separate final models were constructed; four models were run for weight-relative fitness, with two models using MVPA as the physical activity variable and two models using VPA as the physical activity variable. MVPA and VPA correlate too highly with each other to be placed into one model together. Each model for MVPA and VPA was run with and without BMI in the model. Two other models were run for absolute fitness, with one model for MVPA and the other for VPA. Both absolute models included BMI as an independent variable.
Because the mean fitness of the girls increased from 8th to 9th grade and decreased from 9th to 12th grade, a quadratic term (time squared) was tested and used in all models except for the unconditional means model. The quadratic term was treated as a fixed effect, and the time variable was treated as both a fixed effect and random effect (the random effect is created by the slope of the time variable). Race was treated as a time-invariant variable, whereas BMI, MVPA, VPA, and sports participation were time-varying covariates (i.e., BMI, MVPA, VPA, and sports participation were considered at each of the measurement times-8th, 9th, and 12th grade).
For the multivariate model, race, BMI, sport participation, and either MVPA or VPA were entered into the model. Interaction terms with each of the variables (MVPA, VPA, BMI, and sport participation) with time, time squared, and race were included. A backward elimination process was used to eliminate interaction terms that were not significant (P > 0.10) and main effects if they were not involved with an interaction and were not significant (P > 0.05). To determine the best model fit, we examined goodness-of-fit statistics, deviance statistics (compares log-likelihood), Alkaike's information criterion (AIC), and Bayesian information criterion (BIC). Lower values indicate better model fit.
Means and standard deviations for weight-relative and absolute fitness (PWC170), BMI, sport participation, and percentage of girls meeting MVPA (two or more blocks per day) and VPA (one or more blocks per day) guidelines are presented in Table 1. White girls had significantly higher fitness in 9th (P < 0.05) and 12th grades (P < 0.01) than did African American girls. A higher proportion of white girls than African American girls achieved two or more blocks of MVPA at 8th grade (P < 0.01) and one or more blocks of VPA at 8th grade (P < 0.01). Also, a higher proportion of white girls than African American girls participated in sport (P < 0.05) at each of the three time points. Age at baseline of the African American girls was higher than white girls (African American = 13.6 ± 0.6, white = 13.4 ± 0.5, P = 0.01), and fewer of the parents of African American girls had higher than a high school education than did the white girls (African American = 58.7%, white = 79.6%, P = 0.001). The BMI was stable across time, with no significant race difference at any time point (baseline BMI = 22.9 ± 5.5).
There were significant differences found for fitness across the independent variables that were tested in the subsequent analyses of the growth curve models (Table 2). Girls who averaged two or more blocks of MVPA per day across 3 d had higher absolute fitness scores than did those who did not at all three time points (P < 0.05), but higher relative scores at the 9th-grade measurement only (P < 0.01). At all three time points measured, girls who averaged one or more blocks of VPA per day across 3 d (P < 0.05) had higher fitness levels than did girls who were not vigorously active. Girls who were normal weight (P < 0.001) had higher weight-relative fitness levels than did girls who were at risk for overweight/overweight at all three time points. However, at the 9th and 12th grades, girls who were at risk for overweight/overweight had higher absolute fitness levels (P < 0.05) than did girls who were normal weight. At all three measurement time points, sport team participants had higher weight-relative and absolute fitness levels than did nonparticipants (P < 0.05), with the exception of absolute fitness at 12th grade.
Results of fitting two of the baseline models indicated that the overall mean for weight-relative fitness across all occasions and individuals was 11.8 kg·m·min−1·kg−1 (standard error of estimate = 0.19), with an average decrease over time of 0.16 kg·m·min−1·kg−1·yr−1; these numbers represent a linear model that does not take quadratic terms into account, as the full multivariate model does. Although the linear model showed an average decrease over time, fitness scores increased from 8th to 9th grade and decreased from 9th to 12th grade. This trend was seen for the overall sample, as well as for each race and for the normal-weight and overweight groups.
Results of the baseline (unconditional) model indicated that the overall mean for absolute fitness across all occasions and individuals was 699.6 kg·m·min−1 (standard error of estimate = 9.6), with an average increase over time of 10.3 kg·m·min−1·yr−1; again, these numbers represent a linear model. Overall, absolute fitness scores showed a pattern similar to that of the weight-relative scores, with an increase from 8th to 9th grade and a decrease from 9th to 12th grade. This trend was seen for the overall sample, as well as for each race (Table 1). For the normal-weight and overweight groups, a similar pattern was shown, but the values for the overweight group were higher than the values for the normal-weight group at all three time points (Table 2).
After deletion of nonsignificant interactions, the final multivariate growth curve models were constructed. The results of the analyses are presented in Table 3. Six growth models were constructed, with four generated for weight-relative fitness, two each for MVPA and VPA, and two generated for absolute fitness. Each weight-relative MVPA and VPA model was run with and without BMI included in the model. The significant variables in the MVPA and VPA models without BMI were exactly the same (physical activity, time, time2, race × sport interaction, and time × race interaction). Similarly, the significant variables in the models with BMI were almost identical (BMI, physical activity, time, time2, time × BMI interaction, and time × race interaction), with one difference. Whereas sport was significant in the MVPA with BMI model, it was not significant in the VPA with BMI model. Instead, the race × sport interaction was significant in the VPA with BMI model (and not in the MVPA with BMI model). The two models with the best-fit criteria (one with MVPA and the other with VPA) for weight-relative fitness were the models that included BMI:
Specifically, when adjusting for all variables at all time points in the weight-relative, MVPA multivariate model, girls with lower BMI (i.e., girls in the normal-weight category) have higher scores by 0.37 kg·m·min−1·kg−1, girls with two or more blocks per day of MVPA have higher scores by 0.44, and sport participants have higher scores by 0.47. The BMI data were centered for all analyses, meaning that we subtracted the age-and sex-specific 85th percentile for BMI value from each girl's BMI. Thus, positive values indicate BMI above the 85th percentile (overweight category), and negative values indicate BMI below the 85th percentile (normal-weight category). For the weight-relative VPA multivariate model, normal-weight girls have higher scores by 0.36 kg·m·min−1·kg−1, and vigorously active girls have a higher score by 0.51. Most of the significant interactions occurred with time as one of the interaction terms. However, the race-by-sport interaction was significant in both VPA models and the MVPA model without BMI; white sport participants consistently exhibited the highest CRF values over time.
The results of the analyses for absolute fitness are also presented in Table 3, with one model for MVPA and another for VPA. The significant variables in the absolute models were exactly the same for MVPA and VPA (race, physical activity, sport participation, time, time2, and time × BMI interaction). Although race and sport participation were significant in the absolute models, they were not significant in most of the weight-relative models. On the other hand, BMI was significant in the weight-relative models but not in the absolute models. The time × race interaction was not significant in the absolute models, whereas it was significant in the weight-relative models. The time, time2, and time × BMI were significant in both the weight-relative and absolute models. The two final models for absolute fitness were
Specifically, when adjusting for all variables at all time points in the absolute MVPA multivariate model, white girls have higher scores by 40.3 kg·m·min−1, girls with two or more blocks per day of MVPA have higher scores by 28.4, and sport participants have higher scores by 27.2. For the absolute VPA multivariate model, white girls have higher scores by 40.7 kg·m·min−1, vigorously active girls have higher score by 22.1, and sport participants have higher scores by 28.4. The significant interactions occurred with time as one of the interaction terms-specifically, time by BMI: girls in the overweight category consistently exhibited higher absolute CRF values than did girls in the normal-weight category over time.
Results indicate that physical activity, BMI, and the interaction of race and sports participation were significant contributors to change across time in weight-relative cardiorespiratory fitness (CRF) in middle school girls followed through high school. Physical activity, race, and sport participation were significant contributors to change across time in absolute CRF. The tracking literature suggests a moderate relationship among CRF assessments over time. Malina (17) has reported tracking coefficients for V˙O2max in both sexes of R = 0.55-0.56 from childhood to early adolescence (12 yr old) and R = 0.61 for females 13-16 yr old, with longer time intervals (e.g., 11-18 yr old, 13-21 yr old) displaying lower coefficients. In boys and girls ages 10-12, Pate et al. (26) reported Pearson correlations for PWC170 scores of R = 0.65 for 2 yr (correlation for girls only was R = 0.53). Correlations for CRF over time in the current investigation were similar to previous studies and slightly higher for weight-relative fitness than for absolute fitness (weight relative: R = 0.63 from 8th to 9th grade, R = 0.70 from 9th to 12th grade, and R = 0.64 from 8th to 12th grade; absolute: R = 0.53 from 8th to 9th grade, R = 0.63 from 9th to 12th grade, and R = 0.55 from 8th to 12th grade). Whereas CRF may track reasonably well at the group level, our growth curve analyses showed significant variation in CRF across time among girls. At the group level, our results showed that weight-relative fitness increased from 8th to 9th grade and decreased from 9th to 12th grade. A similar pattern for absolute fitness was reported in a mixed-longitudinal sample of adolescent girls who participated in the CHIC study (20). The effects of timing and tempo of growth and maturation, genetics, and other behavioral factors all play a role in how CRF changes over time; it remains unknown which factors have the most influence regarding change across time in CRF.
Significant variables in the growth curve models from the present study are similar to predictors of CRF found in previous research. Physical activity has been found to be a predictor of V˙O2max in 13- to 27-yr-old males and females (11) and in 14- to 15-yr-old males and females (8). Some investigators (9) have shown that VPA is more highly correlated with CRF than MVPA cross-sectionally (r = 0.45 for VPA and r = 0.30 for MVPA). Correlations between physical activity (both MVPA and VPA) and CRF (both weight-relative and absolute) for each year represented in the current study ranged from R = 0.20-0.31 at 8th grade, R = 0.22-0.25 at 9th grade, and R = 0.11-0.14 at 12th grade. The parameter estimate of the relationship between the linear change in physical activity and the quadratic change in fitness over time could not be statistically modeled because of inadequate degrees of freedom (i.e., the number of parameters to be estimated in the model exceeded the data moments). It is rare to see correlations between fitness and activity much higher than r = 0.32 in the literature, both cross-sectionally (10,23) and longitudinally (4). Regardless, results of the current investigation show that levels of physical activity are cross-sectionally related to CRF throughout high school.
BMI was a significant predictor of CRF when it was added to both the MVPA and VPA models for weight-relative fitness. Mota et al. (24) found that 8- to 10-yr-old girls who were overweight or obese were more likely to be unfit compared with normal-weight girls. Similarly, children ages 5-14 yr with a BMI above the 80th percentile were found to pass fewer fitness tests than were their lower-BMI counterparts (12). In a longitudinal study, Koutedakis et al. (14) report that fitness was a significant contributor to adiposity over time in 12- to 14-yr-olds using a similar type of longitudinal modeling analysis (GEE) to the present study. It is important to note that the results of the previously mentioned investigations are based on weight-bearing tests of CRF (shuttle run), whereas the fitness test in the present study is a non-weight-bearing test.
Issues associated with determining CRF using a weight-bearing test versus a non-weight-bearing test, and using weight-relative values versus absolute values, are complex, and they present a conundrum for investigators examining the relationship between fitness and body composition parameters. On the one hand, use of a weight-bearing test and expression of fitness relative to body weight potentially penalize the heavier child/adolescent by making values seem lower than those of their lighter counterparts. On the other hand, use of a non-weight-bearing test can potentially inflate estimation of fitness because of the work that can be done by extra muscle mass, which may or may not translate into "actual" cardiorespiratory fitness. Similarly, absolute expressions of fitness can be higher in larger children/adolescents simply because of size alone (e.g., absolute fitness was higher in girls who were at risk for overweight/overweight at 9th and 12th grades in the current study). One way to address this issue is to examine fitness relative to fat-free mass, which is not possible in the current investigation. Another way to address this issue is to use techniques such as allometric scaling. Although several investigators have proposed the use of allometric scaling (30), there is no standardized procedure available to perform the scaling. This results in different scaling coefficients being developed for different populations, making comparisons across groups difficult.
In the present study, BMI was a significant predictor of weight-relative fitness, but not absolute fitness; however, the interaction of BMI by time was significant in both weight-relative and absolute models. Normal-weight girls seemed to have better fitness than at-risk/overweight girls when expressed relative to body weight, but normal-weight girls seemed to have lower fitness than at-risk/overweight girls when expressed in absolute terms. Race may play a role in these relationships, because it was significant in the absolute models, and the interaction of time and race was significant in the relative models.
Clearly, a complex interaction between fitness, activity, race, and body composition exists, and researchers should continue to determine how to tease out the effects of these different variables over time. Several investigations have indicated that CRF is lower in African American adolescent girls than in white adolescent girls, both cross-sectionally (28) and longitudinally (21). Other investigators have reported that MVPA is positively related to physical fitness among adolescent girls, independently of age and body fatness (8,10), and that the linear relation between declining physical activity and increasing BMI observed among U.S. girls during adolescence is stronger in African American girls than in white girls (13). The causal direction of those relations is not known, but it is plausible that socioenvironmental factors or their interaction with genetic factors contribute to both lower energy expenditure and higher energy intake among African American girls, independently of fitness. Whereas socioeconomic status could be an important variable, it is very closely tied to race, making it difficult to tease out the independent effects of race versus socioeconomic status. It is also possible that low BMI facilitates, and high BMI impedes, VPA, independently of weight-relative fitness. Our findings are consistent with these possibilities, which will require experimental manipulations to confirm or refute.
Sports participation was a significant predictor of fitness in the weight-relative MVPA model and both absolute models; the interaction between race and sports participation was a significant predictor of fitness in both weight-relative VPA models and the weight-relative MVPA model without BMI. Andersen (2) found no relationship between sports participation and fitness in high school boys and girls. There was a dose-response relationship between sports participation (number of sports teams) and time to complete a 1.6-km run/walk in both boys and girls, and there was a dose-response relationship between number of sports teams and PWC170 scores in girls only for a cross-sectional sample of 9- to 15-yr-olds in another study (6). However, a greater increase in submaximal power output was seen in sport participant boys from three longitudinal samples than in nonsport participants (18). In the current investigation, white sport participants consistently had the highest weight-relative fitness scores over time. Again, this highlights the complexity of interactions between fitness, physical activity, race, and body composition over time, particularly for adolescent girls.
This study had several strengths. The sample size was fairly large for a longitudinal analysis, and the analytic technique (growth curve analysis) was sophisticated. One limitation of the study is potential sample bias attributable to the number of sport participants, which was higher than the proportion of female sport participants cited in the 2005 YRBSS at 9th grade (65 vs 56%) and 12th grade (60 vs 41%) (7). Another limitation of the study is that there were other variables we did not measure that may be important factors in change in fitness over time (e.g., growth and maturation, fat-free mass).
In summary, MVPA and VPA were consistently related to change in cardiorespiratory fitness over time. Depending on the expression of CRF (weight bearing vs absolute), BMI, race, and sport participation are also important factors related to change in CRF over time. In addition, CRF declined from the 9th to 12th grades. The results suggest that physical activity and sports participation should be encouraged in girls throughout the high school years to maintain levels of CRF over time. Implications for intervention studies could include the creation of intramural sports programs as an avenue for physical activity in high school girls. Further, the longitudinal interactions among CRF, physical activity, race, BMI, and sports participation are complex and warrant further investigation, as does the interaction between race and socioeconomic status.
We thank the administrators, faculty, staff, and participants at 31 middle schools and 24 high schools across South Carolina for their cooperation and participation. We also thank the data-collection team members, LaVerne Shuler and Janna Borden for their project management skills, and Gaye Christmus for her editorial assistance. This research was supported by NHLBI 1RO1 HL57775.
1. Ainsworth, B. E., W. L. Haskell, M. C. Whitt, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med. Sci. Sports Exerc.
32(9 Suppl.):S498-S504, 2000.
2. Andersen, L. B. Changes in physical activity are reflected in changes in fitness during late adolescence: a 2-year follow-up study. J. Sports Med. Phys. Fitness
3. Armstrong, N., J. Balding, P. Gentle, J. Williams, and B. Kirby. Peak oxygen uptake and physical activity in 11- to 16-year-olds. Pediatr. Exerc. Sci.
4. Baquet, G., J. W. Twisk, H. C. Kemper, P. E. Van, and S. Berthoin. Longitudinal follow-up of fitness during childhood: interaction with physical activity. Am. J. Hum. Biol.
5. Blair, S. N., Y. Cheng, and J. S. Holder. Is physical activity of physical fitness more important in defining health benefits? Med. Sci. Sports Exerc.
6. Cleland, V., A. Venn, J. Fryer, T. Dwyer, and L. Blizzard. Parental exercise in associated with Austrailian children's extracurricular sports participation and cardiorespiratory fitness: Across-sectional study. Int. J. Behav. Nutr. Phys. Act.
7. Eaton, D. K., L. Kann, S. Kinchen, et al. Youth risk behavior surveillance-United States, 2005. MMWR Surveill. Summ.
8. Ekelund, U., E. Poortvliet, A. Nilsson, A. Yngve, A. Holmberg, and M. Sjostrom. Physical activity in relation to aerobic fitness and body fat in 14- to 15-year-old boys and girls. Eur. J. Appl. Physiol.
9. Gutin, B., Z. Yin, M. C. Humphries, and P. Barbeau. Relations of moderate and vigorous physical activity to fitness and fatness in adolescents. Am. J. Clin. Nutr.
10. Katzmarzyk, P. T., R. M. Malina, T. M. Song, and C. Bouchard. Physical activity and health-related fitness in youth: a multivariate analysis. Med. Sci. Sports Exerc.
11. Kemper, H. C., and R. Verschuur. Maximal aerobic power in 13-and 14-year-old teenagers in relation to biologic age. Int. J. Sports Med.
12. Kim, J., A. Must, G. M. Fitzmaurice, et al. Relationship of physical fitness to prevalence and incidence of overweight among schoolchildren. Obes. Res.
13. Kimm, S. Y., N. W. Glynn, E. Obarzanek, et al. Relation between the changes in physical activity and body-mass index during adolescence: a multicentre longitudinal study. Lancet
14. Koutedakis, Y., C. Bouziotas, A. D. Flouris, and P. N. Nelson. Longitudinal modeling of adiposity in periadolescent Greek schoolchildren. Med. Sci. Sports Exerc.
15. Krahenbuhl, G. S., J. S. Skinner, and W. M. Kohrt. Developmental aspects of maximal aerobic power in children. Exerc. Sport Sci. Rev.
16. Kuczmarski, R. J., C. L. Ogden, S. S. Guo, et al. CDC growth charts for the United States: methods and development. Vital Health Stat.
17. Malina, R. M. Physical activity and fitness: pathways from childhood to adulthood. Am. J. Hum. Biol.
18. Malina, R. M., G. Beunen, J. Lefevre, and B. Woynarowska. Maturity-associated variation in peak oxygen uptake in active adolescent boys and girls. Ann. Hum. Biol.
19. McMurray, R. G., J. S. Harrell, S. I. Bangdiwala, and J. Hu. Tracking of physical activity and aerobic power from childhood through adolescence. Med. Sci. Sports Exerc.
20. McMurray, R. G., J. S. Harrell, C. B. Bradley, S. Deng, and S.I. Bangdiwala. Predicted maximal aerobic power in youth is related to age, gender, and ethnicity. Med. Sci. Sports Exerc.
21. McMurray, R. G., J. S. Harrell, C. B. Bradley, S. Deng, and S.I. Bangdiwala. Gender and ethnic changes in physical work capacity from childhood through adolescence. Res. Q. Exerc. Sport
22. Minck, M. R., L. M. Ruiter, M. W. Van, H. C. Kemper, and J. W. Twisk. Physical fitness, body fatness, and physical activity: theAmsterdam growth and health study. Am. J. Hum. Biol.
23. Morrow, J. R., and P. S. Freedson. Relationship between habitual physical activity and aerobic fitness in adolescents. Pediatr. Exerc. Sci.
24. Mota, J., L. Flores, L. Flores, J. C. Ribeiro, and M. P. Santos. Relationship of single measures of cardiorespiratory fitness and obesity in young schoolchildren. Am. J. Hum. Biol.
25. Pate, R. R., R. Ross, M. Dowda, S. G. Trost, and J. R. Sirard. Validation of a three-day physical activity recall instrument in female youth. Pediatr. Exerc. Sci.
26. Pate, R. R., S. G. Trost, M. Dowda, et al. Tracking of physical activity, physical inactivity, and health-related physical fitness in rural youth. Pediatr. Exerc. Sci.
27. Pate, R. R., D. S. Ward, R. P. Saunders, G. Felton, R. K. Dishman, and M. Dowda. Promotion of physical activity among high-school girls: a randomized controlled trial. Am. J. Public Health
28. Pivarnik, J. M., M. S. Bray, A. C. Hergenroeder, R. B. Hill, and W. W. Wong. Ethnicity affects aerobic fitness in US adolescent girls. Med. Sci. Sports Exerc.
29. Strong, W. B., R. M. Malina, C. J. Blimkie, et al. Evidence based physical activity for school-age youth. J Pediatr.
30. Welsman, J. R., N. Armstrong, A. M. Nevill, E. M. Winter, and B. J. Kirby. Scaling peak V˙O2
for differences in body size. Med. Sci. Sports Exerc.
Keywords:©2007The American College of Sports Medicine
ADOLESCENT; PHYSICAL ACTIVITY; RACE; BMI; SPORTS PARTICIPATION