Role of physical activity and sleep duration in growth and body composition of preschool-aged children

Abstract

Objective

The impact of physical activity patterns and sleep duration on growth and body composition of preschool-aged children remains unresolved. Aims were 1) to delineate cross-sectional associations between physical activity components, sleep, total energy expenditure (TEE) and body size and composition; and 2) to determine whether physical activity components, sleep and TEE predict 1-y changes in body size and composition in healthy preschool-aged children.

Methods

Anthropometry, body composition, accelerometry and TEE by doubly labeled water were measured at baseline; anthropometry and body composition were repeated 1-y later (n=111).

Results

Cross-sectionally, positive associations between sedentary activity (SED) and weight and fat-free mass (FFM) (p=0.009-0.047), and negative association between moderate-vigorous physical activity (MVPA) and percent fat mass (FM) (p=0.015) were observed. TEE and activity energy expenditure (AEE) were positively associated with weight, body mass index (BMI), FFM, and FM (p=0.0001-0.046). Prospectively, TEE, AEE, physical activity level (PAL), MVPA, but not SED, were positively associated with changes in BMI (p=0.0001-0.051) and FFM (p=0.0001-0.037), but not percent FM. Sleep duration inversely predicted changes in FM (p=0.005) and percent FM (p=0.006).

Conclusions

Prospectively, MVPA, TEE, AEE and PAL promote normal growth and accretion of FFM, whereas sleep inversely predicts changes in adiposity in preschool-aged children.

Introduction

Early childhood is a period of complex and integrated physical, cognitive and psychosocial development. In the preschool years, children grow rapidly, becoming taller and leaner. Because of asynchronous linear and ponderal growth, body mass index (BMI) declines after infancy to a nadir around 6 years of age, followed by a period of rapidly rising BMI, often referred to as adiposity rebound (1). Early adiposity rebound has been identified as one of the critical risks for the development of obesity (2). It may occur as early as 3 years of age in children with obesity. The role that physical activity plays in normal growth and physical maturation represented by changes in body composition remains unresolved (3, 4). Moderate-vigorous physical activity (MVPA) has been shown to be inversely related to adiposity in preschool-aged children (5), however, it does not appear to drive the timing of the adiposity rebound (6).

A systematic review of 48 studies examined associations between objectively measured physical activity and adiposity in children and adolescents (7). Consistent evidence of negative associations between physical activity and adiposity was seen in 79% of these predominantly cross-sectional studies. Of the few studies (5/48) that focused on preschool-aged children, three reported an inverse relationship between physical activity and adiposity. A more recent review demonstrated that the inverse association between MVPA and measures of adiposity was independent of sedentary time (8). Prospective data are conflicting but one systematic review concluded that objectively measured physical activity may not be a determinant of fat gain (9).

A systematic review of 17 studies in preschool-aged children in part corroborated the inverse association between physical activity and obesity status; however, the association depended on the outcome measure of adiposity (5). In 60% (3/5) of the studies using percent fat mass, an inverse relationship with physical activity was found compared with 18% (2/11) of the studies that used BMI. The few longitudinal studies in preschoolers substantiate a negative association between physical activity and later adiposity (10-12); however, more research using quantitative, objective devices is needed to understand the role of sedentary activity (SED) and MVPA on the growth and body composition of preschool-age children.

Evidence also is emerging that sleep, as diet and physical activity, may play a critical role in the metabolic and hormonal milieu affecting growth and development of children (13). Short sleep duration has been identified frequently as a risk factor for childhood obesity in cross-sectional (14) and longitudinal (15-18) studies; however, most of these epidemiological studies relied on parent-reported sleep. Objectively measured sleep duration using accelerometry may clarify the nature of the sleep-obesity associations reported in children.

The aims of this study were two-fold: 1) to delineate cross-sectional associations between physical activity components, sleep, total energy expenditure (TEE) and body size and composition; and 2) to determine whether physical activity components, sleep and TEE predict 1-y changes in body size and composition in healthy preschool-aged children.

Methods

Study Design and Participants

A longitudinal design was used to evaluate the effect of physical activity on growth and body composition of preschool-aged children, defined as ages 3 to 5 years at baseline, living in the Greater Houston Area in 2010-2012. The study design called for measurements at baseline and at 1-year follow-up. Children were eligible if the child had no major illnesses. Exclusion criteria included any medical illness or medication affecting growth or limiting participation in physical activities or sports. Children on prescription drugs or with chronic diseases including metabolic or endocrine disorders, asthma treated with steroids, and sleep apnea were excluded from the study. The Institutional Review Board for Human Subject Research for Baylor College of Medicine and Affiliated Hospitals approved the protocol. All parents/primary caretakers gave written informed consent to participate in this study.

The children were recruited and enrolled into the study with a balance of boys and girls, and 3, 4 and 5 y/o using fliers at local clinics and preschool centers in Houston. A total of 204 families responded to our recruitment efforts and 127 children ((age 4.6±0.9 y), sex (51% M) and race/ethnicity (26/29/37/8% white/black/Hispanic/mixed race-ethnicity)) were consented to the study after full parental disclosure of the study design and procedures. Children and parents were admitted to the Children’s Nutrition Research Center (CNRC) metabolic research unit. At baseline, parents completed questionnaires assessing sociodemographics, child care, and child physical activity participation. Anthropometric and body composition measurements were completed on the child; and accelerometry and heart rate monitoring for activity monitoring and the doubly labeled water (DLW) method for free-living TEE to be conducted at home were initiated. A total of 119 children ((age 4.6±0.9 y), sex (49% M) and race/ethnicity (29/30/33/8% white/black/Hispanic/mixed race-ethnicity)) completed the baseline study. Subsequently, 111 children returned after 1 year for repeated anthropometric and body composition measurements only (Table 1). The children who did not complete the study did not differ from those who did, in terms of age, sex and race/ethnicity (t-test and chi-square).

Table 1.

Child and familial characteristics

	n	Mean ± SD or frequency%
Child age (y)	111	4.6 ± 0.9
Sex (M/F)	58/53	52/48
Race/ethnicity (W/B/H/M)	31/31/40/9	28/28/36/8
Daycare (hrs/wk)	111	22.0 ± 18.2
Mother age (y)	111	34.6 ± 7.4
Mother BMI (kg/m2)	111	28.9 ± 6.6
Household number	111	5.0 ± 1.4
Number of children	111	3.0 ± 1.3
	n	Frequency (%)
Mother education
Less than or High school	12	10.8
Some college/associated degree	44	39.6
College degree or higher	53	47.7
Missing	2	1.8
Household income
< 25k	20	18.0
25k ≤ 50k	32	28.8
50k ≤ 75k	18	16.2
75k ≤ 100k	20	18.0
≥ 100k	19	17.1
Household size
≤ 3 people in household	11	9.9
4 people in household	31	27.9
5 people in household	35	31.5
6 people in household	16	14.4
> 6 people in household	18	16.2

Abbreviations: W/B/H/M, white, black, Hispanic, multiracial.

Parent-reported Physical Activity

Physical Activity Questionnaire for Children (19) modified for preschool-aged children was completed by the parent to assess the participation in 20 activities in the previous 7 days. Seven activities not applicable to preschool-aged children and questions referring to physical education and recess were deleted from the questionnaire.

Child Body Size and Composition

Body weight to the nearest 0.1 kg was measured with a digital balance and height to the nearest 1 mm was measured with a stadiometer in duplicate and repeated if the two measurements differed by >0.2 kg or 0.5 cm, respectively (20). BMI was calculated as weight/height² (kg/m²). Overweight was defined as ≥85^th percentile but <95^th percentile for BMI, and obese was defined as ≥95^th percentile for BMI, according to the Centers for Disease Control and Prevention (21). Body composition including fat-free mass (FFM) and fat mass (FM) was measured using dual energy x-ray absorptiometry (DXA; Delphi-A, software version 12; Hologic, Bedford, MA).

Physical Activity

Accelerometry: ActiGraph GT3X+

The study design called for the measurement of physical activity for seven consecutive days at baseline only. During the CNRC visit, research assistants demonstrated how to wear the monitors to both mother and child, and provided written instructions for proper care. Children were asked to wear the monitors continuously 24 h/d for seven days, except during swimming or bathing.

ActiGraph GT3X+ (ActiGraph, Pensacola, FL) was used to measure the amount and frequency of movement of the children (22, 23). The monitor is compact and lightweight. Its output includes activity counts (vertical X, horizontal Y and diagonal Z axes), vector magnitude, which is equal to the square root of ((amplitude X)² + (amplitude Y)² +(amplitude Z)²), and number of steps taken. Vector magnitude activity counts were used in the present analysis. The ActiGraph monitors were affixed above the iliac crest of the right hip with an adjustable elastic belt. In addition, heart rate was measured by Actiheart (CamNtech Ltd, Cambridge, UK) during the 7-d study period. Actiheart was attached to the chest using two electrodes (Skintact Premier, Leonhard Lang GmbH, Innsbruck, Austria). The main sensor was attached left of the sternum and secured with the adhesive tab on the electrode. The lead was attached parallel along the mid-clavicular line at the level of the third intercostal space (upper position) or just below the left breast (lower position). The electrodes were checked and replaced if there was poor adhesion.

Vector magnitude activity counts from ActiGraph GT3X+ and heart rate from Actiheart were downloaded and collapsed into 60-second intervals. To minimize inter-interpreter variation, one investigator conducted the data processing and analysis. Non-wear time was defined as 20 minutes or more of consecutive zero counts, if the interval was not identified as nighttime sleep, nap time or device removal for bathing or aquatic activities in the records completed by the parents. Visual inspection of the accelerometer counts and heart rate data also was used to assess non-wear time, as well as nighttime sleep, nap time and awake times. A plot of activity counts and heart rate per minute for each 24-hour period was used to identify the time of sleep onset and termination. Sleep onset was identified by inactivity (VM counts usually zero) and a gradual decline in heart rate. Sleep termination was identified by abrupt increases in activity and heart rate. In-between sleep onset and termination, there is a period of quiescence with occasional excursions due to body movement. Sleep duration was equal to the sum of nighttime sleep and nap time. The amount of awake time spent in SED and MVPA was computed based on the cut-points of 820 and 3,908 cpm for the ActiGraph vector magnitude, respectively (22). A valid day required a minimum of 1,000 minutes per day of wear time. The sufficient number of days of wearing time was defined as a minimum of 4 valid days including at least one weekend day.

Doubly Labeled Water Method

TEE was measured over a 7-d period using the DLW method. After collection of the baseline urine samples, each participant received by mouth, 0.086 g/kg body weight of ²H₂O at 99.9 atom %²H and 1.38 g/kg body weight of H₂¹⁸O at 10 atom %¹⁸O (Isotec, Miamisburg, OH). Seven postdose urine samples (1 mL each) were collected at home on days 1-7. The isotopic enrichment (²H and ¹⁸O) of the urine samples was analyzed by Gas-Isotope-Ratio Mass Spectrometry. The analytical method and calculation of TEE are presented fully elsewhere (24). Physical activity level (PAL) was calculated as the ratio of TEE over basal metabolic rate computed according to Schofield (25). Activity energy expenditure (AEE) was calculated as the difference between TEE and basal metabolic rate plus thermic effect of food, which was assumed to be equal to 10% of TEE.

Statistics

Numerical methods (i.e. skewness, kurtosis, as well as Shapiro-Wilk test) and graphical methods (i.e. histograms and QQ plots) were used to test for normality. If the distribution of the data were not normal, a log-transformation was performed in order to meet the assumptions of a parametric test.

Mixed-effects linear regression models were developed to investigate cross-sectional associations at baseline between child anthropometry, body composition, TEE and physical activity. Exploratory analysis was done initially for selection of potential confounders including socioeconomic factors, parent and child characteristics. The final mixed models adjusted for child characteristics (age, sex, race/ethnicity, daycare hours/wk, awake time), maternal characteristics (age, BMI, education), and household factors (income, family size).

To test for the impact of TEE and physical activity patterns on 1-y changes in body size and composition, mixed-effects linear regression models were conducted using the changes (follow-up – baseline) in the anthropometric and body composition measures as the dependent variables, TEE and physical activity pattern as the exposures. The final covariates included child characteristics (age, sex, race/ethnicity, daycare hours/wk, awake time, height velocity), maternal characteristics (age, BMI, education), and household factors (income, size). SED and MVPA also were mutually adjusted for one another to determine their independent effects on anthropometry and body composition. Statistical analyses were performed in SAS (version 9.4; SAS Institute Inc., Cary, NC) and STATA (ver13; StataCorp, College Station, TX).

Results

Preschool-aged children (n=111), aged (mean ± SD) 4.6 ± 0.9 y, were enrolled into the study (Table 1). The cohort consisted of 31 white, 31 black, 40 Hispanic and 9 multiracial children. The children were from families representing a wide range of sociodemographic status. Family income ranged from less than 25K to over 100K for an average household size of 5.0 ± 1.4, with 3.0 ± 1.3 children. The majority of parents was married (71%) and had some college or a college degree or higher (87% mothers, 65% fathers). Mean BMI of the mothers was 28.9 ± 6.6 kg/m², with 27% overweight and 39% obese.

The majority of children attended daycare (66.7%). The average number of hours in daycare was 22.0 ± 18.2 h/wk. Children engaged in a variety of physical activities. Mean PAQ-C score (35 ± 9; range 21-68) did not differ by age, sex or race/ethnicity. The most common activities (>80% participation) reported by the parents were playing on swings and slides etc. at the playground, tag or chase, and dancing. Next, bicycling, running, walking and active video games were cited (>40% participation). Sports including baseball/softball, basketball, gymnastics, soccer and swimming were reported for a minority of children (20-30% participation).

Mean child BMI z-score and percent FM at baseline were 0.18 ± 1.06 and 27.4 ± 6.7%, respectively (Table 2). While there were no significant differences in BMI z-score by race/ethnicity, the black children had lower percent FM than the white and Hispanic children (p=0.004). Boys had lower FM (p=0.009), percent FM (p=0.001) and higher FFM (p=0.001) compared with the girls. The percent of the children who were overweight and obese was 8.1% and 9.9%, respectively. Changes in anthropometric indices and body composition over one year did not differ by age, sex or race/ethnicity, except for height velocity which decreased with age (p<0.003). There was considerable variability in growth parameters; notably, weight gain ranged from 0.9 to 7.2 kg/y.

Table 2.

Anthropometry and body composition of the preschool-aged children

	3 y	4 y	5 y	Total
	Baseline Values
n	36	37	38	111
Age (y)	3.6 ± 0.3	4.5 ± 0.3	5.6 ± 0.3	4.6 ± 0.9
Weight (kg)1	16.0 ± 2.6	18.6 ± 3.5	20.7 ± 4.0	18.5 ± 3.9
Height (cm)1	99.8 ± 4.4	107.9 ± 5.2	113.3 ± 5.8	107.1 ± 7.6
BMI (kg/m2)3	16.1 ± 2.0	15.9 ± 2.0	16.0 ± 2.1	16.0 ± 2.0
BMI for age percentile	50.8 ± 27.7	52.2 ± 31.3	56.1 ± 26.6	53.1 ± 28.4
BMI z-score	0.11 ± 1.03	0.15 ± 1.14	0.28 ± 1.02	0.18 ± 1.06
FFM (kg)1,2	11.3 ± 1.6	13.5 ± 2.0	15.2 ± 2.6	13.4 ± 2.6
FM (kg)2,3	4.7 ± 1.8	5.2 ± 2.2	5.6 ± 2.4	5.2 ± 2.2
Percent FM (%)1,2,3	29.0 ± 6.8	27.1 ± 6.7	26.4 ± 6.7	27.4 ± 6.7
1-y Anthropometric and Body Composition Changes
Weight (kg/y)	2.55 ± 0.72	2.88 ± 1.21	2.91 ± 1.54	2.78 ± 1.21
Height (cm/y)1	7.01 ± 0.93	6.32 ± 1.04	6.05 ± 0.98	6.45 ± 1.06
BMI (kg/m2/y)	0.19 ± 0.53	0.47 ± 0.70	0.44 ± 0.93	0.37 ± 0.75
BMI z-score (SD/y)	0.33 ± 0.37	0.26 ± 0.38	0.07 ± 0.45	0.22 ± 0.42
FFM (kg/y)	2.09 ± 0.54	2.47 ± 0.76	2.28 ± 0.79	2.29 ± 0.72
FM (kg/y)	0.42 ± 0.46	0.42 ± 0.73	0.63 ± 1.10	0.49 ± 0.82
Percent FM (%/y)	−1.72 ± 1.89	−2.05 ± 2.30	−1.03 ± 2.94	−1.59 ± 2.45

* Mean ± SD; Abbreviations: BMI, body mass index; FFM, fat free mass; FM, fat mass.

¹Main effects for age by ANOVA (p<0.05).

²Main effects for sex by ANOVA (p<0.05).

³Main effects for age race/ethnicity by ANOVA (p<0.05).

High compliance was achieved with the 7-d accelerometry; the mean daily wear time was 1404 ± 30 min/d (Table 3). Sleep duration decreased with age, and was higher in white than black (p=0.001) and Hispanic children (p=0.03). Total counts per day and number of steps per day did not differ by age, sex or race/ethnicity; however, partitioning the awake time into percent SED and MVPA showed differences. SED (min/d, %awake time) was greater in black than the other children (p=0.001) and MVPA (min/d, %awake time) was higher in boys than girls (p<0.01). SED was inversely correlated with MVPA (min/d, r=-0.49; %awake time, r=-0.40; p=0.001).

Table 3.

Physical activity of the preschool-aged children measured by accelerometry and doubly labeled water

	3 y	4 y	5 y	Total
Accelerometry
Wear time (min/d)	1403 ± 31	1400 ± 30	1409 ± 30	1404 ± 30
Sleep (min/d)1,3	606 ± 43	603 ± 39	577 ± 39	595 ± 42
Awake (min/d)1,3	796 ± 43	797 ± 42	840 ± 39	812 ± 46
Activity counts (counts*104/d)	115.2 ± 21.1	119.4 ± 25.1	121.0 ± 22.1	118.6 ± 22.7
Steps (steps*103/d)	8.7 ± 1.7	9.5 ± 2.1	9.5 ± 2.1	9.3 ± 2.0
SED (min/d) 3	366 ± 74	353 ± 56	388 ± 67	370 ± 67
SED (%awake time)3	45.9 ± 8.7	44.4 ± 7.5	46.3 ± 7.5	45.6 ± 7.9
MVPA (min/d)2	25 ± 21	55 ± 29	55 ± 23	54 ± 24
MVPA (%awake time)2	6.6 ± 2.8	6.9 ± 3.6	6.5 ± 2.5	6.7 ± 3.0
Doubly-labeled Water
TEE (kcal/d)1,2	1075 ± 143	1246 ± 179	1287 ± 209	1202 ± 200
AEE (kcal/d)1	132 ± 101	230 ± 112	223 ± 132	194 ± 123
PAL1,3	1.28 ± 0.13	1.39 ± 0.13	1.37 ± 0.15	1.35 ± 0.14

^*Mean ± SD; Abbreviations: SED, sedentary activity; MVPA, moderate-vigorous physical activity; TEE, total energy expenditure; AEE, activity energy expenditure; PAL, physical activity level.

¹Main effects for age by ANOVA (p<0.05).

²Main effects for sex by ANOVA (p<0.05).

³Main effects for race/ethnicity by ANOVA (p<0.05).

Mean 7-d TEE, AEE and PAL generated by the DLW method are summarized for the preschool-aged children in Table 3. TEE increased with age (p=0.001) and was higher in boys than girls (p<0.001). AEE and PAL also increased with age (p=0.001). PAL was higher in white than black children (p=0.02).

Cross-sectional regression models of factors associated with child weight, BMI and body composition

Mixed-effects linear regression models were developed to investigate cross-sectional relationships between child body size and composition, TEE and physical activity, adjusting for child characteristics (age, sex, race/ethnicity, daycare hours/wk, awake time), maternal characteristics (age, BMI, education), and household factors (income, size) at baseline (Table 4). Physical activity pattern measured by accelerometry demonstrated positive associations between SED (min/d) and weight and FFM (p=0.009-0.047). MVPA (min/d) was negatively associated with percent FM (p=0.015). With mutual adjustment for SED and MVPA, the positive associations between SED and weight or FFM, and the negative association between MVPA and percent FM persisted. Positive associations were seen between TEE and AEE and weight, BMI, FFM, and FM (p=0.0001-0.046).

Table 4.

Cross-sectional association between physical activity and child anthropometry and body composition at baseline

	Weight (kg)		BMI (kg/m2)		FFM (kg)		FM (kg)		Percent FM (%)
Predictors	β ± SE	P	β ± SE	P	β ± SE	P	β ± SE	P	β ± SE	P
Physical Activity
Activity counts  (counts*104/d)	−0.017 ± 0.013	0.187	−0.005 ± 0.007	0.452	−0.011 ± 0.009	0.224	−0.008 ± 0.008	0.314	−0.001 ±0.028	0.978
Sleep (min/d)	−0.003 ± 0.007	0.646	−0.003 ± 0.004	0.458	−0.006 ± 0.005	0.274	−0.001 ± 0.005	0.783	0.003 ± 0.017	0.864
SED (min/d)	0.009 ± 0.004	0.047	0.003 ± 0.003	0.275	0.008 ± 0.003	0.009	0.002 ± 0.003	0.532	−0.011 ± 0.010	0.269
MVPA (min/d)	−0.002 ± 0.012	0.864	−0.005 ± 0.007	0.504	0.011 ± 0.009	0.213	−0.015 ± 0.008	0.061	−0.066 ± 0.026	0.015
TEE (kcal/d)	0.010 ± 0.002	<.0001	0.004 ± 0.001	<.0001	0.006 ± 0.001	<.0001	0.004 ± 0.001	0.001	0.005 ± 0.004	0.199
AEE (kcal/d)	0.007 ± 0.002	0.005	0.003 ± 0.002	0.046	0.004 ± 0.001	0.003	0.003 ± 0.002	0.122	0.001 ± 0.005	0.790
PAL	1.505 ± 2.027	0.460	0.119 ± 1.181	0.920	1.734 ± 1.253	0.170	−0.235 ± 1.377	0.865	−3.386 ± 4.257	0.429

Abbreviations: BMI, body mass index; FFM, fat free mass; FM, fat mass; SED, sedentary activity; MVPA, moderate-vigorous physical activity; TEE, total energy expenditure; AEE, activity energy expenditure; PAL, physical activity level. Mixed-effects models: adjusted for age, sex, race/ethnicity, daycare hours, household size, household income, mother's age, BMI and education; models for SED (min/d) and MVPA (min/d) also adjusted for awake time.

Longitudinal regression models of factors influencing child weight, BMI and body composition

Mixed-effects linear regression models were developed to investigate the impact of baseline TEE and physical activity patterns on 1-y changes in child body size and composition, adjusting for child characteristics (age, sex, race/ethnicity, daycare hours/wk, height velocity, awake time), maternal characteristics (age, BMI, education), and household factors (income, size) (Table 5).

Table 5.

Longitudinal effects of physical activity on 1-year changes in anthropometry and body composition

Predictors	Weight (kg)		BMI (kg/m2)*		FFM (kg)		FM (kg)		Percent FM (%)
	β ± SE	P	β ± SE	P	β ± SE	P	β ± SE	P	β ± SE	P
Activity counts (counts*104/d)2	0.006 ± 0.005	0.268	0.006 ± 0.003	0.074	0.007 ± 0.004	0.065	−0.003 ± 0.004	0.476	−0.018 ± 0.010	0.209
Sleep (min/d)3	−0.005 ± 0.003	0.079	−0.004 ± 0.002	0.057	0.0004 ± 0.002	0.865	−0.006 ± 0.002	0.005	−0.022 ± 0.010	0.006
SED (min/d)2	0.0002 ± 0.002	0.895	−0.001 ± 0.001	0.379	−0.001 ± 0.001	0.442	0.001 ± 0.001	0.324	0.008 ± 0.005	0.099
SED (min/d)4	0.002 ± 0.002	0.444	0.0002 ± 0.001	0.867	0.0004 ± 0.001	0.797	0.001 ± 0.001	0.299	0.009 ± 0.010	0.072
MVPA (min/d)2	0.008 ± 0.005	0.093	0.007 ± 0.003	0.032	0.008 ± 0.003	0.026	0.0004 ± 0.003	0.906	−0.001 ± 0.013	0.932
MVPA (min/d)5	0.010 ± 0.005	0.067	0.007 ± 0.003	0.051	0.008 ± 0.004	0.037	0.001 ± 0.004	0.714	0.010 ± 0.010	0.457
TEE (kcal/d)3	0.003 ± 0.001	<.0001	0.002 ± 0.0004	<.0001	0.002 ± 0.001	<.0001	0.001 ± 0.001	0.002	0.002 ± 0.002	0.236
AEE (kcal/d)3	0.003 ± 0.001	<.0001	0.002 ± 0.001	0.001	0.002 ± 0.001	0.001	0.001 ± 0.001	0.049	0.001 ± 0.002	0.507
PAL3	1.869 ± 0.749	0.014	1.287 ± 0.459	0.006	1.270 ± 0.519	0.016	0.485 ± 0.513	0.347	0.807 ± 1.772	0.650

¹ Abbreviations: β ± SE, coefficient and standard error; SED, sedentary activity; MVPA, moderate-vigorous physical activity; TEE, total energy expenditure; AEE, activity energy expenditure; PAL, physical activity level.

²Mixed-effects model adjusted for age, sex, race/ethnicity, Δheight, awake time, daycare hours, household size, household income, mother's age, BMI, and education.

³Mixed-effects model adjusted for age, sex, race/ethnicity, Δheight, daycare hours, household size, household income, mother's age, BMI, and education.

⁴Mixed-effects model adjusted for age, sex, race/ethnicity, Δheight, awake time, MVPA (min/d), daycare hours, household size, household income, mother's age, BMI, and education.

⁵Mixed-effects model adjusted for age, sex, race/ethnicity, Δheight, awake time, SED (min/d), daycare hours, household size, household income, mother's age, BMI, and education.

^*BMI models did not included Δheight.

Sleep duration negatively predicted the 1-y changes in FM (p=0.005) and percent FM (p=0.006). SED (min/d) was not associated with the anthropometric and body composition changes. MVPA (min/d), however, positively predicted the 1-y changes in BMI (p=0.032) and FFM (p=0.026), but not FM or percent FM. With mutual adjustment for SED (min/d), MVPA (min/d) still positively predicted the changes in BMI (p=0.051) and FFM (p=0.037).

TEE and AEE were positively associated with the 1-y changes in weight, BMI, FFM and FM (p=0.0001-0.049), but not percent FM. PAL positively predicted the 1-y changes in weight, BMI and FFM (p=0.006-0.016), but not FM or percent FM.

Discussion

In healthy preschool-age children, SED was positively associated with body size (weight and FFM) and MVPA was negatively associated with adiposity (percent FM) in the cross-sectional analysis. Prospectively, MVPA and PAL were positively predictive of 1-y changes in BMI and FFM but not FM or percent FM. Interestingly, sleep duration was inversely associated with 1-y changes in FM and percent FM.

This cohort represents healthy preschool-aged children from diverse ethnic/racial and SES backgrounds. Typical of American youngsters, the majority of children attended some daycare and participated in a variety of physical activities depending on their developmental stage (26).

Physical activity levels were determined using age-specific accelerometer cut-points (22) and overall PAL by DLW. Physical activity cut-points were based on measured AEE corresponding to established heart rate thresholds of 110 bpm for sedentary/light and 140 bpm for light/moderate levels of physical activity in preschoolers. These cut-points equate to 1.5 and 2.8 child-specific metabolic equivalents (MET) reflecting the lower levels of AEE achievable and sustainable in developmentally immature preschoolers (27). On average, these children spent 46% of awake time in SED and 7% in MVPA. Because of different accelerometers and varying approaches to define cut-points, comparison across studies is difficult. However, the physical activity of these preschoolers based on mean step counts (9,145 steps^.d⁻¹) was similar to other preschoolers (28).

Our cross-sectional results recapitulate the inverse association between MVPA and adiposity reported in some studies (5). In a review of 17 preschool studies, the inverse association was demonstrated in 60% (3/5) of studies with body composition measures compared with 18% (2/11) in studies with BMI only. BMI is not always an accurate index of adiposity (29). The relative contribution of FFM and FM to weight changes throughout childhood is not differentiated by BMI. During the preschool period, BMI declines along with a decline in percent FM and a steady increase in FFM.

In our cross-sectional analysis, SED was positively associated with weight and FFM, and MVPA was inversely associated with percent FM, consistent with some but not all publications. In children, 4-6 y/o, low levels of vigorous physical activity (VPA) were associated with fatness (30). In ethnically diverse preschoolers, VPA was associated with lower odds of obesity (31). In Latino preschoolers, MVPA was negatively associated with BMI z-score (32). VPA was inversely associated with adiposity, independent of sedentary time (33). In contrast to our results for weight and FFM, SED was not associated with BMI z-score in preschool-age children (33-35).

Prospectively, MVPA and PAL positively predicted the 1-y changes in BMI and FFM, but not FM or percent FM, with or without mutual adjustment of sedentary activity. MVPA and PAL promote normal growth and physical maturation, represented by changes in body composition. SED was not predictive of the changes in anthropometry and body composition. Few longitudinal studies in preschoolers are available for comparison. In 3-4 y-old preschoolers followed for three years, MVPA was positively associated with BMI in the first year, and negatively associated in the subsequent two years (11). Tracking of physical activity from 5.6 to 8.6 to 11 y of age showed that MVPA at age 5 was associated with lower FM at 8 and 11 y in boys only (10, 36). Prospective studies in older children are conflicting. In a systematic review, 6/10 studies showed no association between physical activity and change in FM; one reported a weak positive association and 3 reported weak negative associations (9).

Short sleep duration has been identified frequently as a risk factor for childhood obesity in epidemiological studies. In the meta-analysis by Cappuccio et al. (14), seven of 11cross-sectional studies demonstrated a significant association between sleep duration and obesity in children aged 2-20 y. Three of these studies were in preschool-aged children (37-39) with significant odds ratios between short sleep duration and obesity. In our study, sleep duration was not associated with baseline weight, BMI or adiposity, possibly because our cohort had a relatively low number of children with obesity. However, our results confirm cross-sectional analysis in 3 y/o Danish children in which sleep duration by parent-report and accelerometery was not associated with BMI or FM by DXA (40).

Prospectively, sleep duration was inversely predictive of the 1-y changes in FM and percent FM in our cohort. The preschool-aged children with shorter sleep duration had higher fat gains. Longer sleep duration has been associated with lower risk of later childhood obesity in several longitudinal studies (15-18). These results affirm the importance of adequate sleep hygiene among children.

In conclusion, cross-sectionally, SED, TEE and AEE are positively associated with body size, whereas MVPA is negatively related to adiposity in healthy preschool-age children. Prospectively, MVPA, TEE, AEE and PAL are associated with increases in body size, but not adiposity, whereas sleep duration is inversely related to changes in adiposity. Our findings that MVPA and PAL predict normal growth and physical maturation (here represented by changes in body composition) support the role of physical activity in the health and development of young children. From a policy standpoint, children should be afforded the opportunity to engage in MVPA daily in home and daycare settings. The observation that sleep duration was negatively predictive of fat mass gain reinforces the need to promote adequate sleep in young children. Daily MVPA and adequate sleep duration should be promoted for normal growth and body composition of preschool-aged children.

What is known about this subject?

• Studies in preschool-aged children in part corroborate an inverse association between physical activity and adiposity; however, the association depends on the outcome measure of adiposity and study design

• Short sleep duration is inversely associated with childhood obesity but most studies relied on parent-reported sleep

What does this study add?

• Quantitative methods of accelerometry, heart rate monitoring, dual-energy X-ray absorptiometry, and doubly labeled water used to assess accurately the impact of physical activity, energy expenditure and sleep duration on growth and physical maturation here represented by changes in body composition in preschool-aged children

• Moderate-vigorous physical activity, total energy expenditure (TEE), activity energy expenditure (AEE) and physical activity level (PAL) predict normal growth and accretion of fat-free mass but not 1-y changes in adiposity

• Sleep duration inversely predicts 1-y changes in child adiposity

Abbreviations

AEE

activity energy expenditure

BMI

body mass index

DLW

doubly labeled water

DXA

dual energy x-ray absorptiometry

FFM

fat-free mass

FM

fat mass

MET

metabolic equivalents

MVPA

moderate-vigorous physical activity

PAL

physical activity level

SED

sedentary activity

TEE

total energy expenditure

VPA

vigorous physical activity