SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Objective:

The purpose of this study is to determine whether time spent in objectively measured physical activity is associated with change in body mass index (BMI) from ages 9 to 15.

Design and Methods:

The participants were enrolled in the National Institute of Child Health and Human Development Study of Early Child Care and Youth Development (n = 938). At ages 9, 11, 12, and 15 the time spent in moderate-to-vigorous physical activity (MVPA) was objectively measured, and BMI was calculated (kg/m2). Longitudinal quantile regression was used to analyze the data. The 10th, 25th, 50th, 75th, and 90th BMI percentiles were modeled as the dependent variables with age and MVPA (h/day) modeled as predictors. Adjustment was also made for gender, race, sleep, healthy eating score, maternal education, and sedentary behavior.

Results:

A negative association between MVPA and change in BMI was observed at the 90th BMI percentile (−3.57, 95% CI −5.15 to −1.99 kg/m2 per hour of MVPA). The negative association between time spent in MVPA and change in BMI was progressively weaker toward the 10th BMI percentile (−0.27, 95% CI −0.62 to 0.07 kg/m2 per hour of MVPA). The associations remained similar after adjusting for the covariates, and when the analyses were stratified by gender.

Conclusion:

Time spent in MVPA was negatively associated with change in BMI from age 9 to 15. The association was strongest at the upper tail of the BMI distribution, and increasing time spent in MVPA could help reduce the prevalence of childhood obesity.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

It is estimated that 17% of children in the US are obese as measured by body mass index (BMI) (1). The high prevalence of childhood obesity has remained relatively constant over the past decade, with no evidence of a decline (1). It is important to lower the prevalence of childhood obesity since child and adult health are compromised by obesity in early life (2-4). One approach that has the potential to address the high prevalence of childhood obesity is to increase the time children spend in physical activity (5).

However, it remains to be confirmed whether more time spent in physical activity leads to reductions in childhood obesity. A review of longitudinal observational studies published between 2000 and 2009 concluded that there was no evidence of an association between time spent in objectively measured physical activity and changes in measures of childhood obesity (6). Since the publication of that review, additional longitudinal studies have been published in this research area (7-9). Riddoch et al. followed over 4,000 children from age 11 to 13, and observed a negative association between objectively measured moderate-to-vigorous physical activity (MVPA) and changes in fat mass and BMI (9). Similarly, a negative association between objectively measured MVPA and change in BMI was reported in a study that followed 280 children from age 8 to 9 years (8). In contrast, Metcalf et al. found a null association between objectively measured MVPA levels and changes in BMI, and fat mass, in a sample of 202 children followed over the ages of 7-10 years (7). Interestingly, Basterfield et al. observed 403 children from age 7 to 9 and found a negative association between objectively MVPA and changes in BMI among boys, but not girls (10).

The mixed results reported in those longitudinal studies do not fully support the general consensus that physical activity can reduce the prevalence of childhood obesity. There is a need for additional longitudinal studies that test the association between objectively measured physical activity and measures of childhood obesity in large samples of children. Most of the previous studies only included two time points, and modeled the mean change in measures of childhood obesity (6, 8-10). A major limitation of modeling the mean is that the tails of a distribution are not specifically investigated (11). When using BMI to study childhood obesity, the upper tail of the BMI distribution is of primary interest. Therefore, the purpose of the present study was to determine whether objectively measured MVPA is associated with changes in the BMI distribution over the ages of 9, 11, 12, and 15.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Participants and study design

The children in the present study participated in the NICHD Study of Early Child Care and Youth Development (12). The children were recruited at birth (1991) from community hospitals affiliated with a university, at ten geographic locations throughout the US (Little Rock, AR; Irvine, CA; Lawrence KA; Boston, MA; Philadelphia, PA; Pittsburgh, PA; Charlottesville, VA; Seattle, WA; Hickory and Morganton, NC; and Madison, WI). After screening, a total of 1,364 children were eligible for participation and these children were enrolled into the study. The baseline demographics of the sample reflected US 1990 census data. At ages 9 (2000), 11 (2002), 12 (2003), and 15 (2006) the children wore accelerometers that allowed for an objective estimate of time spent in physical activity. At the same ages the children were also measured for height and weight that allowed for the calculation of BMI.

Body mass index

At the population level, BMI is used to monitor the prevalence of childhood obesity (1), so it is important to identify factors that associate with BMI. The heights of the participants were measured using a seven foot measuring stick fastened to the wall, and a T-square was rested on the child's head while standing tall and straight; duplicate heights were measured to the nearest 0.32 cm. The weights of the participants were measured using a physician's 2-beam scale; duplicate weights were measured to the nearest 0.1 kg. The height and weight measures were used to calculate BMI (kg/m2) for each participant. BMI was used throughout as studies have shown that BMI is more powerful, and easier to interpret, than BMI z-scores in longitudinal studies (13).

Moderate-to-vigorous physical activity

Children are recommended to spend at least 60 min per day in physical activity, especially MVPA (5). At ages 9, 11, and 12 Actigraph 7164 accelerometers (ActiGraph, Fort Walton Beach, FL, USA) were worn, and at age 15 ActiGraph GT1M accelerometers (ActiGraph, Fort Walton Beach, FL, USA) were worn, at the right hip for 7 days at each age. The accelerometers were removed when bathing, participating in water sports, and when asleep. The participants had to wear the accelerometer for at least 3 days, 10 h per day (60 min of zero counts per minute were considered non-wear time), to be included in the present study (14, 15). The accelerometers were initialized to collect count data in 1-min sampling periods, and higher counts per minute (cpm) indicate more intense and more frequent movement. A cutpoint of ≥ 2,296 was used to define MVPA in the present study. This cutpoint was developed by Evenson et al. and was recently shown to be an optimal estimate of MVPA in a sample of children aged 5-15 (16, 17).

Covariates

Gender, race, socioeconomic status, sleep, healthy eating score, and sedentary behavior were included as covariates. Gender, race (white or other), and socioeconomic status have been associated with BMI (1) and physical activity levels (14). Gender and race were self-reported at study entry, as was maternal education (high school graduate or less, some college, Bachelor's degree, or Graduate/professional degree). The latter variable was used as a marker of socioeconomic status in the present study. Hours of sleep on the previous night was self-reported at age 15 by the participants, and there is evidence that sleep is associated with physical activity levels and BMI (18). Healthy eating was self-reported by the participant at age 15; the frequency of consuming fruit juice, a green salad or other raw vegetables, cooked vegetables, or fruit on the previous day was recalled by the participants (none, 1, 2, or 3+ times per day). Healthy eating score was used as a marker of diet quality in the present study, as those with lower diet quality may have an increased likelihood of having higher BMI (19). Time spent in sedentary behavior was objectively measured using accelerometers (<100 cpm) at ages 9, 11, 12, and 15 (17). Sedentary behavior is distinct from physical activity levels and there is evidence that sedentary behavior is an independent risk factor for childhood obesity (20).

Statistical analysis

For descriptive purposes, the baseline demographics for the children with anthropometric data are presented. The means and standard deviations are given for the anthropometric and MVPA variables, and frequencies and percentages are given for gender, race, and maternal education. For the main analysis, longitudinal quantile regression was used. This statistical approach models the effect of predictors across the distribution of continuous outcome variables (11, 21). Traditional regression models only consider the effect of predictors on the mean of a continuous outcome variable. The main advantage of using quantile regression to model BMI in the context childhood obesity is that the upper tail of the BMI distribution is of most clinical and public health importance (11). The interpretation of the coefficients from quantile regression is the same as with traditional ordinary least squares regression (i.e., the coefficient represents the change in the dependent variable for every unit change in the predictor). All quantile regression analyses in the present study modeled the effect of predictors at the 10th, 25th, 50th, 75th, and 90th BMI percentiles. In model 1, BMI was included as the dependent variable and age and age2 were included as predictors. This was to describe changes in BMI from age 9 to 15, and the higher order age variable (age2) was included to test for curvilinear changes in BMI from age 9 to 15. In model 2, MVPA was additionally included as a time-varying predictor to determine whether time spent in MVPA was associated with changes in BMI from age 9 to 15. In model 3, gender, race, maternal education, hours of sleep, and healthy eating score were also included as predictors, and in model 4, sedentary behavior was included as a time-varying covariate. The purpose of models 3 and 4 was to determine whether any association between MVPA and BMI remained with adjustment for the covariates. The data were arranged in the long format and so each participant had four rows of data to represent each time point. Participants with missing BMI at age 9 were removed from dataset. To account for the correlation between repeated measures a first order autoregressive correlation structure was used, and the 95% confidence intervals were estimated using 500 bootstrap samples (21). All analyses were conducted using Stata 12.0 (StataCorp LP, College Station, TX).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

From the original 1,364 children recruited 80%, 79%, 78%, and 74% were followed up at ages 9, 11, 12, and 15. Anthropometric data were available for 84-86% of the children followed up at ages 9, 11, 12, and 15. Those with missing BMI data were more likely to have been born to mothers with a lower level of education, and boys were more likely to have missing BMI data at ages 9 and 11. The demographic characteristics of the sample at age 9 with anthropometric data are presented in Table 1. The sample was split in terms of gender and over 80% of the children were white (Table 1). On average, the boys spent more time in MVPA compared to the girls at all ages. Decline in average MVPA occurred in both boys and girls from age 9 to 15 (overall average declines of 30.2 min/day and 28.9 min/day, respectively) (Table 1). The average BMI was similar for both boys and girls at all ages, and increased overall by 4.70 kg/m2 from age 9 to 15 (Table 1).

Table 1. Demographic, anthropometric, and physical activity characteristics of the children
 Total sampleBoysGirls
  1. BMI, body mass index; MVPA, moderate-to-vigorous physical activity. For the demographic variables, the frequencies and percentages are those at age 9, for the children with anthropometric data.

Sample size, N938461477
Race, N (%)   
 White766 (81.7)374 (81.1)392 (82.2)
 Other172 (18.3)87 (18.9)85 (17.8)
Maternal education, N (%)   
 High school or less259 (27.6)140 (30.4)119 (24.9)
 Some college321 (34.2)148 (32.1)173 (36.3)
 Bachelor's degree210 (22.4)99 (21.5)111 (23.3)
 Graduate/professional degree148 (15.8)74 (16.1)74 (15.5)
Height, mean (SD), cm   
 Age 9 (n = 938)135.1 (6.29)135.4 (6.26)134.8 (6.31)
 Age 11 (n = 844)147.2 (7.41)146.7 (7.15)147.8 (7.55)
 Age 12 (n = 835)153.1 (7.71)152.2 (7.93)154.0 (7.41)
 Age 15 (n = 759)168.5 (8.35)173.3 (7.61)164.0 (6.25)
Weight mean (SD), kg   
 Age 9 (n = 938)33.9 (8.94)34.5 (9.45)33.4 (8.38)
 Age 11 (n = 844)44.1 (13.0)44.3 (13.7)43.9 (12.3)
 Age 12 (n = 835)49.3 (14.3)49.2 (15.2)49.5 (13.4)
 Age 15 (n = 759)65.8 (17.1)70.3 (18.6)61.6 (14.3)
BMI, mean (SD), kg/m2   
 Age 9 (n = 938)18.4 (3.74)18.6 (3.92)18.2 (3.55)
 Age 11 (n = 844)20.1 (4.72)20.3 (5.03)19.9 (4.39)
 Age 12 (n = 835)20.8 (4.95)21.0 (5.28)20.7 (4.62)
 Age 15 (n = 759)23.1 (5.23)23.3 (5.52)22.9 (4.94)
MVPA, mean (SD), min/day   
 Age 9 (n = 741)54.2 (25.5)62.3 (28.4)46.7 (19.7)
 Age 11 (n = 626)43.1 (23.5)50.9 (26.0)35.8 (18.0)
 Age 12 (n = 513)35.7 (23.8)42.5 (28.9)29.3 (15.1)
 Age 15 (n = 379)25.6 (17.6)31.1 (19.1)17.8 (12.8)

Changes in BMI are further described using quantile regression in Table 2 (model 1). The median (50th percentile) BMI increased at linear rate of 0.71 kg/m2 per year from age 9 to 15 (predicted overall increase of 4.26 kg/m2, which is similar to the mean increase in BMI). A lower rate of increase in BMI was observed below the median BMI (predicted overall increase of 3.42 kg/m2 at the 10th percentile), and a greater rate of increase in BMI was observed above the median BMI (predicted overall increase of 6.48 kg/m2 at the 90th percentile) (Table 2, model 1). The changes in BMI were curvilinear at the 10th and 25th BMI percentiles, where the rate of change accelerated with increasing age (Table 2, model 1); and were curvilinear at the 75th and 90th BMI percentiles, where the rate of change decelerated with increasing age (Table 2, model 1).

Table 2. Changes in BMI from ages 9 to 15 and the influence of MVPA
 Body mass index (BMI): Boys and girls
 10th percentile25th percentile50th percentile75th percentile90th percentile
  1. MVPA, moderate-to-vigorous physical activity. N/A, higher order age variable not applicable. Data are coefficients (95% CI), and each coefficient represents the predicted change in BMI per unit change in age, age2, or MVPA. The MVPA units are hours per day. The age units are 0, 2, 3 and 6 and the age2 units are 0, 4, 9, and 36 to represent ages 9, 11, 12, and 15 respectively. Model 3 is adjusted for gender, race, maternal education, healthy eating score and hours of sleep. Model 4 is adjusted for covariates in model 3 plus sedentary behavior (hr/d). The number of children included in the analyses for models 1 through 4 are n = 938, n = 866, n = 813, and n = 813, respectively.

Model 1:     
 Intercept15.0 (14.8, 15.1)15.9 (15.8, 16.1)17.4 (17.1, 17.6)19.8 (19.4, 20.2)23.5 (22.9, 24.2)
 Age0.21 (0.10, 0.31)0.29 (0.19, 0.39)0.71 (0.67, 0.75)1.44 (1.22, 1.67)2.10 (1.66, 2.53)
 Age20.06 (0.04, 0.08)0.06 (0.04, 0.08)N/A−0.11 (−0.14, −0.07)−0.17 (−0.25, −0.09)
Model 2:     
 Intercept15.2 (14.8, 15.6)16.7 (16.3, 17.2)18.5 (18.0, 19.1)21.9 (21.0, 22.8)26.5 (24.6, 28.4)
 Age0.18 (0.02, 0.33)0.21 (0.07, 0.35)0.56 (0.49, 0.63)0.94 (0.57, 1.31)1.26 (0.52, 2.00)
 Age20.06 (0.03, 0.09)0.05 (0.03, 0.08)N/A−0.05 (−0.12, 0.01)−0.11 (−0.23, 0.02)
 MVPA−0.27 (−0.62, 0.07)−0.77 (−1.16, −0.37)−1.31 (−1.70, −0.91)−2.51 (−3.16, −1.87)−3.57 (−5.15, −1.99)
Model 3:     
 Intercept18.4 (16.0, 20.8)20.8 (18.5, 23.2)25.5 (22.2, 28.7)32.9 (26.2, 39.6)43.6 (32.6, 54.6)
 Age0.17 (−0.00, 0.34)0.20 (0.08, 0.33)0.52 (0.45, 0.60)0.79 (0.38, 1.20)1.09 (0.57, 1.61)
 Age20.06 (0.04, 0.09)0.06 (0.04, 0.08)N/A−0.03 (−0.09, 0.04)−0.10 (−0.19, −0.00)
 MVPA−0.37 (−0.79, 0.06)−0.77 (−1.16, −0.38)−1.56 (−2.00, −1.11)−2.51 (−3.48, −1.55)−3.49 (−4.95, −2.04)
Model 4:     
 Intercept17.8 (15.4, 20.1)20.7 (18.1, 23.2)24.5 (21.3, 27.8)30.4 (24.5, 36.4)40.4 (28.5, 52.3)
 Age0.19 (0.02, 0.36)0.20 (0.08, 0.32)0.46 (0.38, 0.55)0.66 (0.25, 1.06)1.08 (0.46, 1.70)
 Age20.05 (0.02, 0.08)0.05 (0.03, 0.08)N/A−0.04 (−0.10, 0.02)−0.14 (−0.23, −0.04)
 MVPA−0.31 (−0.72, 0.09)−0.70 (−1.10, −0.30)−1.49 (−1.93, −1.06)−2.30 (−3.17, −1.44)−3.30 (−4.83, −1.78)

At the 50th BMI percentile, time spent in MVPA was negatively associated with changes in BMI from age 9 to 15 [predicted change in BMI of −1.31 (−1.70, −0.91) kg/m2 per hour spent in MVPA] (Table 2, model 2). The association between MVPA and change in BMI was stronger at the upper tail of the BMI distribution [predicted change in BMI of −3.57 (−5.15, −1.99) kg/m2 per hour spent in MVPA at the 90th BMI percentile]; and was weaker at the lower tail of the BMI distribution [predicted change in BMI of −0.27 (−0.62, 0.07) kg/m2 per hour spent in MVPA at the 10th BMI percentile] (Table 2, model 2). The MVPA coefficient corresponding to the 10th BMI percentile was smaller than the 50th BMI percentile MVPA coefficient (−1.04, 95% CI: −1.37, −0.73, P < 0.001); and the MVPA coefficient corresponding to the 90th BMI percentile was larger than the 50th BMI percentile MVPA coefficient (2.26, 95% CI: 0.93, 3.56, P = 0.001). The associations remained similar after adjusting for gender, race, maternal education level, hours of sleep, and healthy eating score (Table 2, model 3), and after adjusting for time spent in sedentary behavior (Table 2, model 4).

Since the time spent in MVPA was lower in the girls compared to the boys, the influence of MVPA on changes in BMI was additionally analyzed by gender (Table 3). MVPA was negatively associated with changes in BMI in both boys and girls with the strength of the associations progressively stronger toward the upper tail of the BMI distribution. The associations remained after adjusting for race, maternal education, hours of sleep, healthy eating score, and sedentary behavior (Table 3).

Table 3. Changes in BMI from age 9 to 15 by gender and the influence of MVPA
 Body Mass Index (BMI): Boys
 10th percentile25th percentile50th percentile75th percentile90th percentile
Model 1:     
 Intercept15.1 (14.9, 15.4)16.1 (15.9, 16.3)17.4 (17.1, 17.7)20.0 (19.4, 20.6)24.8 (23.2, 26.3)
 Age0.16 (0.01, 0.30)0.24 (0.11, 0.37)0.71 (0.64, 0.78)1.40 (1.04, 1.77)1.20 (0.96, 1.45)
 Age20.06 (0.03, 0.08)0.06 (0.03, 0.09)N/A−0.09 (−0.15, −0.03)N/A
Model 2:     
 Intercept15.8 (15.2, 16.5)17.2 (16.6, 17.8)19.2 (18.4, 20.0)23.0 (21.8, 24.3)28.5 (25.5, 31.5)
 Age−0.00 (−0.23, 0.22)0.07 (−0.13, 0.28)0.48 (0.37, 0.58)0.69 (0.15, 1.23)0.78 (0.37, 1.19)
 Age20.07 (0.04, 0.11)0.06 (0.03, 0.10)N/A−0.03 (−0.12, 0.07)N/A
 MVPA−0.50 (−1.02, 0.01)−0.99 (−1.42, −0.56)−1.66 (−2.21, −1.11)−2.97 (−3.83, −2.11)−4.50 (−6.73, −2.27)
Model 3:     
 Intercept19.9 (16.5, 23.4)20.1 (16.9, 23.3)23.9 (19.8, 28.0)30.1 (20.7, 39.5)43.8 (27.5, 60.1)
 Age−0.04 (−0.28, 0.20)0.15 (−0.03, 0.33)0.48 (0.38, 0.58)0.83 (0.24, 1.42)0.67 (0.26, 1.08)
 Age20.08 (0.04, 0.12)0.06 (0.02, 0.09)N/A−0.06 (−0.15, 0.03)N/A
 MVPA−0.52 (−1.06, 0.02)−0.78 (−1.25, −0.32)−1.75 (−2.35, −1.15)−2.75 (−4.12, −1.37)−3.24 (−5.56, −0.91)
Model 4:     
 Intercept16.7 (16.4, 23.0)19.6 (16.2, 22.9)23.4 (19.5, 27.3)25.9 (18.1, 33.6)40.5 (23.2, 57.8)
 Age−0.04 (−0.27, 0.20)0.17 (−0.02, 0.37)0.41 (0.28, 0.54)0.42 (−0.17, 1.01)0.59 (0.12, 1.06)
 Age20.08 (0.04, 0.11)0.05 (0.01, 0.09)N/A−0.03 (−0.12, 0.06)N/A
 MVPA−0.52 (−1.05, 0.04)−0.70 (−1.16, −0.23)−1.69 (−2.28, −1.11)−2.49 (−3.72, −1.26)−2.59 (−5.79, −0.39)
 Body mass index (BMI): Girls
 10th percentile25th percentile50th percentile75th percentile90th percentile
  1. MVPA, moderate-to-vigorous physical activity. N/A, higher order age variable not applicable. Data are coefficients (95% CI), and each coefficient represents the predicted change in BMI per unit change in age, age2, or MVPA. The MVPA units are hours per day. The age units are 0, 2, 3 and 6, and the age2 units are 0, 4, 9, and 36 to represent ages 9, 11, 12 and 15 respectively. Model 3 adjusted for race, maternal education, healthy eating score and hours of sleep. Model 4 is adjusted for the covariates in model 3 plus sedentary behavior (hr/day). The number of boys/girls included in the analyses for models 1 through 4 are n = 461/477, n = 420/446, n = 392/421 and n = 392/421, respectively.

Model 1:     
 Intercept14.9 (14.6, 15.1)15.8 (15.6, 16.1)17.2 (16.9, 17.6)19.6 (19.1, 20.1)23.2 (22.5, 23.9)
 Age0.24 (0.09, 0.39)0.35 (0.21, 0.49)0.73 (0.67, 0.79)1.39 (1.09, 1.68)1.81 (1.19, 2.42)
 Age20.06 (0.03, 0.09)0.05 (0.03, 0.08)N/A−0.10 (−0.15, −0.05)−0.14 (−0.25, −0.04)
Model 2:     
 Intercept15.0 (14.5, 15.5)16.1 (15.5, 16.8)18.3 (17.5, 19.2)21.3 (19.9, 22.6)26.5 (24.2, 28.8)
 Age0.19 (−0.04, 0.42)0.34 (0.14, 0.54)0.60 (0.49, 0.71)1.23 (0.69, 1.77)1.18 (0.44, 1.92)
 Age20.07 (0.03, 0.11)0.05 (0.02, 0.08)N/A−0.09 (−0.18, −0.00)−0.11 (−0.23, 0.01)
 MVPA−0.14 (−0.72, 0.43)−0.40 (−1.05, 0.24)−1.31 (−2.15, −0.46)−2.43 (−3.58, −1.28)−4.22 (−6.25, −2.21)
Model 3:     
 Intercept17.6 (14.6, 20.7)19.6 (15.5, 23.7)25.6 (20.8, 30.5)33.3 (26.3, 40.2)39.7 (26.0, 53.3)
 Age0.19 (−0.01, 0.40)0.30 (0.11, 0.49)0.54 (0.42, 0.65)0.91 (0.43, 1.39)1.27 (0.65, 1.89)
 Age20.07 (0.03, 0.11)0.06 (0.02, 0.09)N/A−0.04 (−0.12, 0.04)−0.15 (−0.25, −0.06)
 MVPA−0.41 (−1.01, 0.19)−0.65 (−1.38, 0.08)−1.69 (−2.56, −0.82)−2.92 (−4.37, −1.47)−4.84 (−6.71, −2.97)
Model 4:     
 Intercept17.3 (14.3, 20.3)19.2 (15.0, 23.5)24.4 (18.8, 30.1)30.5 (23.0, 38.1)37.9 (25.5, 50.2)
 Age0.21 (−0.02, 0.44)0.31 (0.11, 0.51)0.47 (0.33, 0.60)0.71 (0.24, 1.17)1.00 (0.29, 1.70)
 Age20.06 (0.02, 0.10)0.05 (0.01, 0.08)N/A−0.04 (−0.12, 0.04)−0.15 (−0.25, −0.04)
 MVPA−0.41 (−0.98, 0.15)−0.46 (−1.16, 0.24)−1.48 (−2.40, −0.57)−2.63 (−3.97, −1.28)−4.60 (−6.55, −2.65)

To help illustrate how time spent in MVPA could affect changes in BMI, predicted BMI distributions at ages 9 and 15 are presented in Figure 1. If all children accumulated 60 min of MVPA per day at ages 9 and 15, as opposed to decreasing their time spent in MVPA, then this could reduce the number of children at the upper tail of the BMI distribution (Figure 1).

thumbnail image

Figure 1. Leftward shift at the upper tail of the BMI distribution if all children spent more time spent in MVPA. The solid gray line represents 60 min/day at age 9 and 15. The black dashed line represents the average time spent in MVPA at age 9 (54.2 min/day) and at age 15 (25.6 min/day). The proportion of children that accumulates ≥60 min of MVPA per day was 35% and 5% at ages 9 and 15, respectively. The vertical reference lines correpsond to age-specific BMIs that project to 30 kg/m2 at age 18 (22). Kernel density estimation was used to plot the BMI distributions (Epanechnikov, bandwidth 3).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We found that more time spent in MVPA was negatively associated with changes in BMI from age 9 to 15 in both boys and girls. Importantly, the strength of the negative association was strongest at the upper tail of the BMI distribution. This observation has important clinical and public health implications since it is the upper tail of the BMI distribution that is of concern in the context of childhood obesity. Our longitudinal data lend support to the contention that increasing MVPA levels in all children could lead to a reduction in the number of children classified as obese at the population level.

A variety of analytical methods have been used in previous longitudinal studies that found no consensus on the association between physical activity and changes in measures of childhood obesity (7-10). However, the majority of those studies modeled the mean changes of the obesity variables and therefore did not take into the account the tails of the distributions (6-10). Those studies may have underestimated the strength of the association between physical activity and childhood obesity, or inadvertently reported a null association between physical activity and childhood obesity, by not taking into account the tails of the distributions. To the best of our knowledge no previous studies have used quantile regression to model the association between physical activity and changes in BMI in children. Cross-sectional studies have used quantile regression to investigate the association between obesogenic exposures, other than physical activity, and BMI in children (23-26). In those studies, the obesogenic exposures tended to exert stronger associations at the upper tail of the BMI distribution compared to the lower tail of the BMI distribution (23-26). Those cross-sectional findings relate well to the longitudinal findings reported in the present study.

A common mechanism used to explain childhood obesity is energy balance, which posits that too much energy intake relative to energy expenditure, or too little energy expenditure relative to energy intake, induces a positive energy balance and consequently childhood obesity (27). Under this mechanism, increases in energy expenditure through physical activity have the potential to correct for a positive energy balance and shift the BMI distribution to the left. However, our data suggest that for a given amount of time spent in MVPA, the leftward shift would not be uniform across the BMI distribution. It is possible that children at the upper tail of the BMI distribution expend more energy per time spent in MVPA, compared to children at the lower tail of the BMI distribution (i.e., children with larger BMIs require more energy to move). Therefore, this could explain the stronger association we observed between MVPA and BMI at the upper tail of the BMI distribution.

Alternatively, gene-environment interactions may explain, in part, the non-uniform association we observed. Several obesity susceptibility loci have been identified in children, and those at the upper tail of the BMI distribution are more likely to carry risk alleles at such loci, compared to children at the lower tail of the BMI distribution (28). If children are exposed to low levels of MVPA, those at the upper tail of the BMI distribution could experience greater increases in BMI, compared to children at the lower tail of the distribution, on account of their genetic susceptibility. We do not have genetic data to test this hypothesis; however, there is evidence that more time spent in MVPA attenuates the association between an obesity susceptibility locus (FTO) and BMI in children (29, 30).

Given our observations it is important to identify approaches that can increase childhood physical activity levels, especially among children in the upper half of the BMI distribution. Intervention approaches to increase physical activity levels in children have had limited success (31). It has been suggested that children have an inherent physical activity level (activitystat) and interventions to increase physical activity levels fail as a consequence of children having a biological set point for physical activity (32). However, there is strong evidence that physical activity levels decline with age (33, 34), indicating that children do not have a set point for physical activity and that physical activity levels are amenable to change. There are several social and physical environmental factors that correlate with physical activity levels and it is likely that these can be targeted to prevent declines in physical activity levels during childhood (35). Delivering interventions that successfully target modifiable elements in children's physical and social environments remain a research priority.

There are several strengths of the current study. We used an objective measure of physical activity, and included four time points from age 9 to 15; previous related studies tended to have smaller sample sizes and/or included two time points (6-10). Longitudinal quantile regression was used to model the association between time spent in MVPA and changes in BMI, allowing for the investigation of MVPA at the tails of the BMI distribution (11, 21). This approach has clinical and public health relevance since it is the upper tail of the BMI distribution that is of concern in the context of childhood obesity. There are also limitations of the current study. We adjusted for key covariates, but residual confounding may remain due to missing covariates or because of measurement error in the covariates included. Replication of our study using more direct measures of adipose tissue would advance this area of research, although BMI is used to monitor population trends in childhood obesity (1). Our data are interpreted at the population level and no inferences can be made with regard to change in an individual's BMI. Our sample was similar to the demographics of the US population in 1990 and so the majority of children are white. It would be worthwhile replicating our study in more diverse populations of children. A common concern of longitudinal studies is attrition, and it is a limitation of the present study that there are missing anthropometric data and accelerometry data. There were no differences in demographic characteristics between those with and without accelerometry data, but those with missing accelerometry data at ages 12 and 15 had slightly higher BMIs on average. The accelerometer model changed at age 15 (ActiGraph 7164 to ActiGraph GT1M) and it is a limitation that the same accelerometer model was not used across all study years. However, there are data showing that the count data from these two accelerometers are comparable (36). Further, our results were similar when only the first three study years were analyzed (data not shown).

In conclusion, more time spent in MVPA was negatively associated with changes in BMI from age 9 to 15 in both boys and girls. The strength of the association was strongest at the upper tail of the BMI distribution. Increasing the time children spend in MVPA could be an effective public health approach to reduce the prevalence of childhood obesity.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This study was conducted by the NICHD Early Child Care Research Network supported by NICHD through a cooperative agreement that calls for scientific collaboration between the grantees and the NICHD staff. The authors would like to thank Dr. Matteo Bottai for helping with the quantile regression analyses.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Ogden CL, Carroll MD, Curtin LR, Lamb MM, Flegal KM. Prevalence of high body mass index in US children and adolescents, 2007-2008. JAMA 2010; 303: 242-249.
  • 2
    Baker JL, Olsen LW, Sorensen TI. Childhood body-mass index and the risk of coronary heart disease in adulthood. N Engl J Med 2007; 357: 2329-2337.
  • 3
    Tirosh A, Shai I, Afek A, et al. Adolescent BMI trajectory and risk of diabetes versus coronary disease. N Engl J Med 2011; 364: 1315-1325.
  • 4
    Daniels SR. Complications of obesity in children and adolescents. Int J Obes (Lond) 2009; 33 ( Suppl 1): S60-S65.
  • 5
    U.S. Department of Health and Human Services. 2008 Physical Activity Guidelines for Americans. U.S. Department of Health and Human Services 2008; Available at: URL: http://www.health.gov/paguidelines/default.aspx. Accessed June 10, 2012.
  • 6
    Wilks DC, Besson H, Lindroos AK, Ekelund U. Objectively measured physical activity and obesity prevention in children, adolescents and adults: a systematic review of prospective studies. Obes Rev 2011; 12: e119-e129.
  • 7
    Metcalf BS, Hosking J, Jeffery AN, Voss LD, Henley W, Wilkin TJ. Fatness leads to inactivity, but inactivity does not lead to fatness: a longitudinal study in children (EarlyBird 45). Arch Dis Child 2011; 96: 942-947.
  • 8
    Fisher A, Hill C, Webber L, Purslow L, Wardle J. MVPA is associated with lower weight gain in 8-10 year old children: a prospective study with 1 year follow-up. PLoS One 2011; 6: e18576.
  • 9
    Riddoch CJ, Leary SD, Ness AR, et al. Prospective associations between objective measures of physical activity and fat mass in 12-14 year old children: the Avon Longitudinal Study of Parents and Children (ALSPAC). BMJ 2009; 339: b4544.
  • 10
    Basterfield L, Pearce MS, Adamson AJ, et al. Physical activity, sedentary behavior, and adiposity in English children. Am J Prev Med 2012; 42: 445-451.
  • 11
    Hao L, Naimen, D.Q. Introduction. In: Quantile Regression. Liao TF, editor. Thousand Oaks: SAGE Publications; 2007. p 1-6.
  • 12
    NICHD Early Child Care Research Network. Nonmaternal care and family factors in early development: An overview of the NICHD Study of Early Child Care. Appl Dev Psychol 2001; 22: 457-492.
  • 13
    Berkey CS, Colditz GA. Adiposity in adolescents: change in actual BMI works better than change in BMI z score for longitudinal studies. Ann Epidemiol 2007; 17: 44-50.
  • 14
    Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, Mcdowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc 2008; 40: 181-188.
  • 15
    Treuth MS, Baggett CD, Pratt CA, et al. A longitudinal study of sedentary behavior and overweight in adolescent girls. Obesity (Silver Spring) 2009; 17: 1003-1008.
  • 16
    Evenson KR, Catellier DJ, Gill K, Ondrak KS, McMurray RG. Calibration of two objective measures of physical activity for children. J Sports Sci 2008; 26: 1557-1565.
  • 17
    Trost SG, Loprinzi PD, Moore R, Pfeiffer KA. Comparison of accelerometer cut-points for predicting activity intensity in youth. Med Sci Sports Exerc 2011; 43: 1360-1368.
  • 18
    Carter PJ, Taylor BJ, Williams SM, Taylor RW. Longitudinal analysis of sleep in relation to BMI and body fat in children: the FLAME study. BMJ 2011; 342: d2712.
  • 19
    Wu XY, Ohinmaa A, Veugelers PJ. Diet quality, physical activity, body weight and health-related quality of life among grade 5 students in Canada. Public Health Nutr 2011; 4: 1-7.
  • 20
    Must A, Bandini LG, Tybor DJ, Phillips SM, Naumova EN, Dietz WH. Activity, inactivity, and screen time in relation to weight and fatness over adolescence in girls. Obesity (Silver Spring) 2007; 15: 1774-1781.
  • 21
    Wei Y, Pere A, Koenker R, He X. Quantile regression methods for reference growth charts. Stat Med 2006; 25: 1369-1382.
  • 22
    Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 2000; 320: 1240-1243.
  • 23
    Crume TL, Bahr TM, Mayer-Davis EJ, et al. Selective protection against extremes in childhood body size, abdominal fat deposition, and fat patterning in breastfed children. Arch Pediatr Adolesc Med 2012; 166: 437-443.
  • 24
    Beyerlein A, Toschke AM, von Kries R. Risk factors for childhood overweight: shift of the mean body mass index and shift of the upper percentiles: results from a cross-sectional study. Int J Obesity 2010; 34: 642-648.
  • 25
    Beyerlein A, Toschke AM, von Kries R. Breastfeeding and childhood obesity: shift of the entire BMI distribution or only the upper parts? Obesity (Silver Spring) 2008; 16: 2730-2733.
  • 26
    Beyerlein A, Toschke AM, Rosario AS, von Kries R. Risk factors for obesity: further evidence for stronger effects on overweight children and adolescents compared to normal-weight subjects. PLoS One 2011; 6: e15739.
  • 27
    Hill JO. Understanding and addressing the epidemic of obesity: an energy balance perspective. Endocr Rev 2006; 27: 750-761.
  • 28
    Bradfield JP, Taal HR, Timpson NJ, et al. A genome-wide association meta-analysis identifies new childhood obesity loci. Nat Genet 2012; 44: 526-531.
  • 29
    Ruiz JR, Labayen I, Ortega FB, et al. Attenuation of the effect of the FTO rs9939609 polymorphism on total and central body fat by physical activity in adolescents: the HELENA study. Arch Pediatr Adolesc Med 2010; 164: 328-333.
  • 30
    Xi B, Wang C, Wu L, et al. Influence of physical inactivity on associations between single nucleotide polymorphisms and genetic predisposition to childhood obesity. Am J Epidemiol 2011; 173: 1256-1262.
  • 31
    van Sluijs EM, McMinn AM, Griffin SJ. Effectiveness of interventions to promote physical activity in children and adolescents: systematic review of controlled trials. BMJ 2007; 335: 703.
  • 32
    Fremeaux AE, Mallam KM, Metcalf BS, Hosking J, Voss LD, Wilkin TJ. The impact of school-time activity on total physical activity: the activitystat hypothesis (EarlyBird 46). Int J Obes (Lond) 2011; 35: 1277-1283.
  • 33
    Kimm SY, Glynn NW, Kriska AM, et al. Decline in physical activity in black girls and white girls during adolescence. N Engl J Med 2002; 347: 709-715.
  • 34
    Nader PR, Bradley RH, Houts RM, McRitchie SL, O'Brien M. Moderate-to-vigorous physical activity from ages 9 to 15 years. JAMA 2008; 300: 295-305.
  • 35
    Craggs C, Corder K, van Sluijs EM, Griffin SJ. Determinants of change in physical activity in children and adolescents: a systematic review. Am J Prev Med 2011; 40: 645-658.
  • 36
    Kozey SL, Staudenmayer JW, Troiano RP, Freedson PS. Comparison of the ActiGraph 7164 and the ActiGraph GT1M during self-paced locomotion. Med Sci Sports Exerc 2010; 42: 971-976.