Growth Trajectories of Young Children's Objectively Determined Physical Activity,Sedentary Behavior,and Body Mass Index

Abstract

Purpose:

This study examined trajectories in children's physical activity (PA), sedentary behavior, and body mass index (BMI) for both genders, and relationships among these trajectories, from childhood through early adolescence.

Methods:

A total of 261 seconds and third-grade children (135 girls; Mean_age = 7.81; 73% White) from two U.S. elementary schools participated in this study. Their objective PA, sedentary behavior, and BMI were measured yearly from 2012 to 2015. The outcome variables were accelerometer-determined daily moderate-to-vigorous PA (MVPA), sedentary behavior, and BMI—calculated as height divided by weight squared. A series of latent growth curve models (LGCM) were employed to analyze the trajectories in the outcome variables and relationships over time through AMOS version 23 in 2017.

Results:

The models generally fitted the data well. In detail, children's MVPA increased slightly by the end of the first year and then declined during follow-ups (p < 0.05), with boys having more MVPA time. Sedentary hours increased at 2-year follow-ups (p < 0.05), but decreased slightly at year 3 follow-up in both genders. BMI increased gradually for boys (p < 0.05), particularly in year 3. Trajectories in MVPA were negatively related to BMI trajectories (p < 0.05), and trajectories in MVPA and sedentary behavior did not interact to affect BMI trajectories.

Conclusions:

Maintaining or increasing MVPA and limiting sedentary behavior should be components of efforts to prevent excess weight gain during the transition from childhood to early adolescence. Children's MVPA and sedentary behavior are independent determinants of BMI changes.

Introduction

A growing body of evidence has shown that physical activity (PA) and sedentary behaviors independently predict the development of obesity and chronic diseases, such as diabetes, and therefore, should be addressed as separate constructs.^1,2 Given that PA is low and sedentary behaviors are high among children and adolescents,^3,4 efforts to promote physically active lifestyles and limit time spent in sedentary behaviors have been top priorities among health professionals.^4,5 Such initiatives are based upon the notion that PA is beneficial, whereas sedentary behaviors are detrimental for the health of youth, and that these behaviors become habitual and track over time with the potential to significantly impact health during adulthood.^2,6–9 Many transitions (i.e., increased peer influence) from childhood to early adolescence influence individuals' PA and sedentary behaviors,¹⁰ and consequently, the tracking of these behaviors are likely to vary during this period of transition. To inform effective interventions and policies designed to promote PA and reduce sedentary behavior for achieving optimal health, it is critical to fully understand how these behaviors change from childhood to early adolescence.

In general, boys are more physically active and less sedentary than girls from early adolescence, onward.^6,8,11,12 Yet, few studies have targeted children before they progress into early adolescence. Although longitudinal studies have established gender-specific trajectories in PA,^6,8,12 few studies have addressed gender differences in trajectories of PA along with sedentary behaviors across time. Also, European researchers have documented that body mass index (BMI) trajectories from adolescence to adulthood were moderate to strong in girls and moderate in boys¹³—little is known with regard to the longitudinal relationship of BMI trajectories to PA or sedentary behavior trajectories from childhood to early adolescence. Additionally, no longitudinal evidence examining previously identified gender gaps in PA, sedentary behavior, and BMI trajectories are present in the literature. It remains unclear from the existing literature: (1) when gender differences in PA, sedentary behavior, and BMI emerge, (2) whether such differences become larger over time, and (3) whether these change patterns vary by developmental period. Understanding gender differences in behavior changes over time and the relationships of these changes is important in guiding the development of more effective behavioral interventions. In particular, investigating the dynamic relationships among trajectories in PA, sedentary behavior, and BMI will help to better understand the underlying mechanisms of individuals' behavior and may be useful in informing gender-tailored behavioral intervention programs.

Despite increasing research attention over the past decade to the tracking of PA behaviors,^14,15 including some recent literature on PA, sedentary behavior, and BMI, during transition from elementary school to middle school,^16–18 little is known concerning the dynamic of relationships among trajectories in objectively determined PA, sedentary behavior, and BMI on a yearly basis from childhood to early adolescence. Therefore, this study was designed to help fill the knowledge gap in this area of inquiry. Specifically, the primary aim is to examine trajectories of children's PA, sedentary behavior, and BMI over time across gender. The secondary aim is to examine associations of trajectories in PA and sedentary behavior with BMI trajectories over time. To my knowledge, this is the first study that discerns trajectories in objective PA, sedentary behavior, and BMI in both boys and girls from childhood to early adolescence.

Methods

Research Design and Participants

This study employed a single-group repeated measures design and was part of a larger parent trial, which was completed in 2015.¹⁹ Data of this study were never published. Participants were 261 seconds and third graders (127 boys, 134 girls; Mean age = 8.27 years) from two suburban elementary schools in Western Texas. Both schools were similar in terms of teacher quality, physical education curricula, and socioeconomic status. In detail, 125 minutes of weekly PA programs and 20 minutes of daily recess were required at both schools. Inclusion criteria for this study were: (1) enrolled in a public elementary school; (2) 7–9 years of age; and (3) without a diagnosed physical or mental disability according to school records. Inclusion criteria were verified through the school records and a self-reported demographic survey. The study was approved by the Institutional Review Board of Texas Tech University and University of Minnesota. Both parental consent and child assent were obtained through parental consent forms and child assent forms before participation. Children's privacy and confidential data were ensured before and through conducting the study.

Children's accelerometer-determined MVPA, sedentary behavior, and BMI were measured four times from 2012 to 2015 (September 2012, May 2013, May 2014, and May 2015). Procedures for accelerometer data collection have been described in detail in Gao et al.¹⁹

Measures

Children's weekly MVPA and sedentary behavior were measured using ActiGraph GTX3 accelerometers (Pensacola, FL)—a valid and reliable device used to assess PA and sedentary behavior among various populations in free-living conditions.²⁰ Accelerometers were worn at the right hip using a neoprene. In the present study, activity counts were set at 15 seconds epoch. All children wore the accelerometer at least 14 hours a day, but only the time slots of 8:00am–10:00pm were utilized for data analysis in this study.²¹ Raw activity counts were interpreted using empirically established cut points that defined different PA intensities (sedentary: 0-100; light PA: 101-2295; and MVPA: 2296 and above).²² The outcomes were children's daily time spent in MVPA and sedentary behavior.

Demographic information, including gender, age, grade, and race/ethnicity, was self-reported by the children at baseline. Children's body height was assessed using a Seca stadiometer for height, rounded in centimeters, and weight was measured with a Detecto digital weight scale (Detecto; Web City, MO) rounded up in kilograms in private rooms to calculate BMI (a ratio of weight and height [kg/m²]) for this study. Height and weight were measured at each time point.

Data Analysis

Data from each accelerometer were downloaded to ActiLife 6.12 for data sorting/processing and were analyzed in 2017. Descriptive analysis was used to first describe demographic and behavioral characteristics of the sample. Trajectory and association analyses were conducted for the targeted outcomes (MVPA, sedentary behavior, and BMI) with analyses being weighted for nonresponse bias. To analyze the change in the outcomes over time and how these changes correlated, a latent growth curve model (LGCM) was adopted. AMOS version 23 was used to conduct these analyses. To address the first aim, data from the four time points were used to construct a normative growth trajectory for MVPA for the entire sample, as well as two models for both genders, separately.²³ Similarly, six more models for the two genders and overall sample's sedentary behavior and BMI were constructed.

In addition, exploratory variables were used to construct an additional multivariate latent conditional model through the addition of time-varying predictors (e.g., MVPA and sedentary behavior) for the overall sample. Given the limited number of participants in the present study, the analysis for each gender was not conducted. A freed-factor loading approach was used to address the second research question, with the latent intercept and latent slope factor estimated for BMI in addition to MVPA and sedentary behavior. BMI latent growth factors were subsequently regressed on MVPA in addition to sedentary behavior trajectories, allowing for the observation of the variability in the intercept and slope of BMI trajectories (Fig. 1). To ensure LGCM estimated model fit, several tests were used, including a χ ² test statistic, Tucker–Lewis Index (TLI), comparative fit index (CFI), normed fit index (NFI), and root-mean-squared error of approximation (RMSEA).²⁴ Generally, nonsignificant results—as seen when the χ ² statistic is p > 0.05, values of 0.90 or higher are seen for the CFI, TLI, and NIF, and the RMSEA is 0.06 or lower—indicate excellent model fit.²⁵

Figure 1.

Predictors of initial level and rate of change in BMI for the total sample. MVPA, moderate-to-vigorous physical activity; Sedentary, sedentary behavior; BMI, body mass index; the coefficient on the solid arrow line is the standard regression weight. *Indicates a significant path.

Results

Descriptive information about children's PA, sedentary behavior and BMI at the four time-points are displayed in Table 1—overall and separately, by gender.

Table 1.

Descriptive Analyses of Moderate-to-Vigorous Physical Activity, Sedentary Behavior, and Body Mass Index by Gender

	Wave-I			Wave-II			Wave-III			Wave-IV
	N	Mean	SD	N	Mean	SD	N	Mean	SD	N	Mean	SD
MVPA
All	254	43.63	18.87	242	45.23	19.71	186	35.87	19.72	138	29.23	14.86
boy	123	48.67	20.55	115	48.59	19.80	91	38.75	21.02	62	32.80	16.43
girl	131	38.90	15.82	127	42.18	19.20	95	33.11	18.08	76	26.31	12.83
Sedentary
All	253	386.81	79.32	242	414.64	100.43	186	490.38	78.68	138	444.00	114.37
boy	122	373.56	78.69	116	402.71	109.32	91	486.02	78.38	62	437.77	125.25
girl	131	399.15	78.19	126	425.61	90.53	95	494.55	79.15	76	449.08	105.25
BMI
All	261	18.22	3.71	249	18.62	3.89	198	19.24	4.37	131	19.95	4.57
boy	126	18.26	3.90	120	18.64	4.11	96	19.40	4.44	59	21.11	5.18
girl	135	18.18	3.53	129	18.60	3.69	102	19.08	4.33	72	18.99	3.77

Unit for MVPA and sedentary behavior is minutes/day; unit for BMI is the value of weight (kg)/[height (m)].

MVPA, moderate-to-vigorous physical activity; BMI, body mass index.

Trajectories in MVPA, Sedentary Behavior and BMI

Three-year trajectories in MVPA, sedentary behavior, and BMI are presented by gender in Table 1. With fit statistics of CFI = 0.94, TLI = 0.93, NFI = 0.92, and RMSEA = 0.06 (0.05–0.07), the overall sample's MVPA model was observed to fit these data well. As individuals transitioned from childhood to early adolescence, an overall decrease in MVPA was seen, except for a slight increase by the end of first year (mean intercept = 45.94, p < 0.01; mean slope = −14.51, p < 0.01). The covariate estimate between the intercept and slope factors (covariance of slope and intercept = −19.85) proved nonsignificant (p = 0.58) with a standardized correlation of r = −0.15, suggesting children with initially higher MVPA levels had similar decrease rates in MVPA as others over time. Two models for both genders revealed similar trajectory changes in MVPA among boys and girls. That is, gender-specific slopes showed similar slopes in boys and girls (−15.56 and −13.58, respectively) both strongly significant declines (p < 0.05), but not statistically different from each other (p > 0.05). Yet, girls consistently scored lower than boys (boys' intercept = 50.33; girls' intercept = 41.92) over time.

The models for sedentary behavior fit these data well or approached the good fit, with related fit statistics of CFI = 0.90–0.94, TLI = 0.89–0.92, NFI = 0.91–0.95 and RMSEA = 0.052–0.068. In general, children's sedentary behavior increased over 100 minutes in the first 2 years in both genders, but both genders decreased an average of 50 minutes in year 3 (p < 0.05). Boys had relatively lower sedentary behavior time than girls at all four time points (boys' intercept = 378.31; girls' intercept = 405.06; p > 0.5), and the trajectory changes were similar between genders (boy's mean slope = 99.03, girl's mean slope = 76.16, p < 0.05), suggesting there was no significant change in the gender gap over time.

The models for BMI were also observed to fit the model well as indicated by fit statistics of CFI = 0.95–0.97, TLI = 0.95–0.96, NFI = 0.95–0.98, and RMSEA = 0.024–0.058. For the overall sample, children's BMI generally increased (mean slope = 1.47, p < 0.01), yet there was relative intraindividual variability in BMI over time. The covariate estimate between the intercept and slope factors (covariance of slope and intercept = 1.35) proved significant (p = 0.047) with a standardized correlation of r = 0.19, indicating children with initially higher BMI had steeper increases over time. Although differing change patterns were observed for BMI, initial BMIs for boys and girls were quite similar. Table 1 indicates a small decrease in girls' BMI (0.5%) in the last year, with an approximately 4.95% increase in the first 2 years. Conversely, boys demonstrated a gradually increasing BMI trend over 3 years: a 6.24% increase for the first 2 years, with another 8.81% increase during the last year. The gender gap remained consistent in the first 2 years, yet enlarged in year 3 (p < 0.01).

Associations among Trajectories in MVPA and in Sedentary Behavior with Trend in BMI

Time-dependent predictors (e.g., MVPA, sedentary behavior) for the overall sample were placed in the multivariate LGCM, which allowed the modeling of observed variability in the intercept and slope factors of BMI trajectory. The recommended goodness-of-fit values were achieved for this model's CFI = 0.91, TLI = 0.90, and NFI = 0.92 and RMSEA = 0.06, with a 90% confidence interval of 0.04–0.07. Moderately high standardized regression coefficients for sedentary behavior and MVPA trajectories were seen from their observed variables with r² values ranging from 0.18 to 0.36, indicating that 18% to 36% of the variances in the latent variables are explained by the variances in the observed variables. As shown in Figure 1, no significant interaction effects were seen for MVPA or sedentary behavior trajectories for BMI (β = −0.06, p = 0.94). In detail, initial BMI trajectories were positively related to sedentary behavior trajectories (β = 0.24, p < 0.05) and negatively, yet nonsignificantly, associated with MVPA trajectories (β = −0.10, p > 0.05). Trajectories in BMI slope were only negatively related to trajectories in MVPA (standardized β = −0.28, p < 0.01). Results suggest that higher BMI was seen at baseline among children with greater sedentary behavior. Furthermore, more gradual increases in BMI over time were seen in those with lower MVPA. Overall, 6% of the variance in BMI intercept and 3.6% of the variance in BMI slope were attributable to the time-dependent predictors.

Discussion

This study examined trajectories in MVPA, sedentary behavior, and BMI as well as the dynamic relationships among these trajectories, as children progress from childhood through early adolescence. Over the study's 3-year duration, MVPA declined for both genders. This finding is in accordance with many previous studies with older populations,^2,8,26 indicating a decline in MVPA from early adolescence to young adulthood.^2,4,8,26 Research evidence suggests that boys are more active and engaged in more MVPA than girls from adolescence to young adulthood.^2,8,26 The same trajectories of MVPA hold true when comparing younger boys to girls in this study. Yet, in the present study, the gender gap for MVPA did not show significant changes over time, decreasing in the first year and remaining consistent in the next 2 years. These patterns may be due to gender differences in sport participation or PA preferences during this transition period, whereby girls may be more likely to select light-to-moderate intensity sports or PA (e.g., dance), whereas boys often prefer moderate-to-vigorous intensity team sports (e.g., soccer).²⁷

Contrary to the consistent decline in MVPA, children's sedentary behavior demonstrated different change patterns over the course of the study. The increase in sedentary time in the first 2 years may be due partially to advancement and popularity of recent technology during this period of time. For example, Nelson et al. suggested that increase in sedentary behavior among boys were due to increases in leisure time computer use.⁵ Sedentary behavior decreased markedly in the third year of this study. This decrease is a desired outcome, as intervention approaches to decrease emerging obesity and prevent chronic diseases often aim to decrease sedentary behavior on a daily basis. When studying U.K. adolescents, Mitchell et al.²⁸ indicated greater increases in sedentary time among girls compared with boys. Yet, no significant changes in sedentary behavior were found for either gender from adolescence to young adulthood in Ortega et al's study.² The finding of this study echoed the former study with adolescents, whereby girls spent more time in sedentary behavior than boys from childhood to early adolescence. Similar to changes in MVPA, the gender gap decreased slightly and then remained similar in the follow-ups. Collectively, it is apparent that the gender differences seen in sedentary time do not follow a consistent pattern as compared with the adolescent samples, suggesting that these might be age and culture specific.

Boys and girls showed different change patterns in BMI, with boys increasing gradually across time, whereas girls experienced increases in the first 2 years after which slight decrease was observed. Of note, the gender gap for BMI enlarged vastly in year 3. Although a previous study observed that boys were more likely to become overweight than girls over a 5-year follow-up period,² it is truly difficult to offer any clear reason for the gender difference in such changes in the present study. To draw valid conclusions on this issue, more research is needed regarding correlates of BMI across time in both genders.

Trajectories in children's MVPA were negatively associated with trajectories in BMI, whereas initial sedentary behavior was positively related to children's initial BMI. But the two effects of MVPA and sedentary behavior on BMI trajectories were independently and mainly through trajectories in MVPA. The finding is in line with previous studies indicating that PA and sedentary behavior are independent predictors of chronic diseases and mortality and that they should be considered as separate constructs.^2,4,29–31 That is, individuals who had more sedentary time at baseline tended to display higher BMI, and those who had greater decreased MVPA over time were more likely to have higher BMI and vice versa. Yet, trajectories in sedentary behavior and initial MVPA were unrelated for BMI outcomes. Specific speculation for such phenomenon deserves further investigation.

Practical Implications

The findings have important public health implications, in that maintaining or increasing MVPA and limiting sedentary behavior should be components of efforts to prevent excess weight gain during the transition from childhood to early adolescence. Given that the associations among trajectories in MVPA, sedentary behavior, and BMI differ by gender, behavioral interventions might have a different degree of impact on boys as compared with girls. Specifically, trajectories in MVPA and sedentary behavior among boys were more favorable than girls in our sample. Recent large-scale health promotion interventions targeting only adolescent girls have been implemented.³² Given the findings of the current study, tailored behavioral interventions should be developed for girls at a younger age.

Conclusions

Strengths of the current study lie in its objective assessment of PA and sedentary behavior as well as the longitudinal design with repeated measures. The study design allowed for the examination of trajectories in PA-related behavior and weight changes during a critical 3-year time period from childhood to early adolescence. However, study limitations should be taken into account in interpreting the results and in designing future studies. First, the attrition rate in this study was approximately 50% over 3 years. Better retention strategies should be adopted in the future. Second, despite the diverse sample, the sample was drawn from two schools at one geographic region in the southern United States. Therefore, it is recommended that future research recruit samples from multiple schools and geographic locations. Overall, the findings contribute substantially to our understanding of changes in key weight-related health behaviors (PA and sedentary behavior) and dynamic relations between such changes and BMI across the transitional period from childhood to early adolescence. In particular, behavioral interventions may be developed and tailored for girls during this period to maximize MVPA while reducing sedentary behavior.

Footnotes

Acknowledgments

This study was funded by a grant from the National Institute of Child and Human Development (1R15HD071514-01A1). The author would like to thank David Stodden, Nida Roncesvalles, and the research assistants who helped with data collection and students who cooperated with data collection.

Author Disclosure Statement

The author has no financial disclosures and declares no conflicts of interest.

References

Healy

, Wijndaele

, Dunstan

, et al. Objectively measured sedentary time, physical activity, and metabolic risk: The Australian Diabetes, Obesity and Lifestyle Study (AusDiab). Diabetes Care, 2008; 31:369–371.

Ortega

, Konstabel

, Pasquali

, et al. Objectively measured physical activity and sedentary time during childhood, adolescence and young adulthood: A cohort study. PLoS One, 2013; 8:e60871.

Gordon-Larsen

, Adair

, Popkin

. Ethnic differences in physical activity and inactivity patterns and overweight status. Obes Res, 2002; 10:141–149.

Gordon-Larsen

, Nelson

, Popkin

. Longitudinal physical activity and sedentary behavior trends: Adolescence to adulthood. Am J Prev Med, 2004; 27:277–283.

Nelson

, Neumark-Stzainer

, Hannan

, et al. Longitudinal and secular trends in physical activity and sedentary behavior during adolescence. Pediatrics, 2006; 118:e1627–e1634.

Francis

, Morrissey

, Letuchy

, et al. Ten-year objective physical activity tracking: Iowa Bone Development Study. Med Sci Sports Exerc, 2013; 45:1508–1514.

Raustorp

, Ekroth

. Tracking of pedometer-determined physical activity: A 10-year follow-up study from adolescence to adulthood in Sweden. J Phys Act Health, 2013; 10:1186–1192.

Telama

, Yang

, Leskinen

, et al. Tracking of physical activity from early childhood through youth into adulthood. Med Sci Sports Exerc, 2014; 46:955–962.

Trang

NHHD

, Hong

, van der Ploeg

, et al. Longitudinal sedentary behavior changes in adolescents in Ho Chi Minh City. Am J Prev Med, 2013; 44:223–230.

10.

Kristensen

, Møller

, Korsholm

, et al. Tracking of objectively measured physical activity from childhood to adolescence: The European youth heart study. Scand J Med Sci Sports, 2008; 18:171–178.

11.

Riddoch

, Bo Andersen

, Wedderkopp

, et al. Physical activity levels and patterns of 9- and 15-yr-old European children. Med Sci Sports Exerc, 2004; 36:86–92.

12.

Kimm

. Decline in phyhsical activity in black girls and white girls during adolescence. N Engl J Med, 2002; 347:709–715.

13.

Herman

, Craig

, Gauvin

, Katzmarzyk

. Tracking of obesity and physical activity from childhood to adulthood: The Physical Activity Longitudinal Study. Int J Pediatr Obes, 2009; 4:281–288.

14.

Biddle

SJH

, Pearson

, Ross

, Braithwaite

. Tracking of sedentary behaviours of young people: A systematic review. Prev Med (Baltim), 2010; 51:345–351.

15.

Telama

. Tracking of physical activity from childhood to adulthood: A review. Obes Facts, 2009; 2:187–195.

16.

Mitchell

, Pate

, Beets

, Nader

. Time spent in sedentary behavior and changes in childhood BMI: A longitudinal study from ages 9 to 15 years. Int J Obes, 2013; 37:54–60.

17.

Fulton

, Dai

, Steffen

, et al. Physical Activity, Energy Intake, Sedentary Behavior, and Adiposity in Youth. Am J Prev Med, 2009; 37(1 Suppl.):S40–S49.

18.

Dowda

, Taverno Ross

, McIver

, et al. Physical activity and changes in adiposity in the transition from elementary to middle school. Child Obes, 2017; 13:53–62.

19.

Gao

, Pope

, Lee

, et al. Impact of exergaming on young children's school day energy expenditure and moderate-to-vigorous physical activity levels. J Sport Heal Sci, 2016; 6:11–16.

20.

Trost

, Mciver

, Pate

. Conducting Accelerometer-Based Activity Assessments in Field-Based Research. Med Sci Sport Exerc, 2005; 37(Suppl.):S531–S543.

21.

Lee

, Stodden

, Gao

. Young children's energy expenditure and moderate-to-vigorous physical activity on weekdays and weekends. J Phys Act Heal, 2016; 13:1013–1016.

22.

Evenson

, Catellier

, Gill

, Al

. Calibration of two objective measures of physical activity for children. J Sport Sci, 2008; 26:1557–1565.

23.

Bollen

, Curran

. Latent Curve Models: A Structural Equation Perspective. John Wiley & Sons, Hoboken: NJ, 2006.

24.

Clemens

, Shapiro

, Wu

, et al. Monitoring early first-grade reading progress a comparison of two measures. J Learn Disabil, 2014; 47:254–270.

25.

, Bentler

. Cutoff criteria for fit indexes in covariance structure analysis:Conventional criteria versus new alternatives. Struct Equ Model A Multidiscip J, 1999; 6:1–55.

26.

Telama

, Yang

, Viikari

, Välimäki

. Physical Activity from Childhood to Adulthood. Am J Prev Med, 2005; 28:267–273.

27.

Verschuur

, Kemper

HCG

. Habitual physical activity. Med Sport Sci Exerc, 1985; 20:56–65.

28.

Mitchell

, Pate

, Dowda

, et al. A prospective study of sedentary behavior in a large cohort of youth. Med Sci Sport Exerc, 2012; 44:1081–1087.

29.

Trang

NHHD

, Hong

, Dibley

. Cohort profile: Ho Chi Minh City Youth Cohort—changes in diet, physical activity, sedentary behaviour and relationship with overweight/obesity in adolescents. BMJ Open, 2012; 2:e000362.

30.

Yang

, Telama

, Leskinen

, et al. Testing a model of physical activity and obesity tracking from youth to adulthood: The cardiovascular risk in young Finns study. Int J Obes (Lond), 2007; 31:521–527.

31.

Yang

, Telama

, Viikari

, Raitakari

. Risk of obesity in relation to physical activity tracking from youth to adulthood. Med Sci Sports Exerc, 2006; 38:919–925.

32.

Kuo

, Schmitz

, Evenson

, et al. Physical and social contexts of physical activities among adolescent girls. J Phys Act Heal, 2009; 6:144–152.