Preventive Medicine 53 (2011) 268–270

Contents lists available at ScienceDirect

Preventive Medicine

j ourna l homepage: www.e lsev ie r.com/ locate /ypmed

Tracking of children's body-mass index, television viewing and dietary intakeover five-years

Natalie Pearson a,b,⁎, Jo Salmon b, Karen Campbell b, David Crawford b, Anna Timperio b

a School of Sport, Exercise & Health Sciences, Loughborough University, UKb Centre for Physical Activity and Nutrition Research, School of Exercise and Nutrition Sciences, Deakin University, Australia

⁎ Corresponding author at: School of Sport, ExerciseLoughborough University, Epinal Way, Loughborough, L

E-mail addresses: n.l.pearson@lboro.ac.uk (N. Pearso(J. Salmon), Karen.campbell@deakin.edu.au (K. Campbedavid.crawford@deakin.edu.au (D. Crawford), anna.timp(A. Timperio).

0091-7435/$ – see front matter © 2011 Elsevier Inc. Aldoi:10.1016/j.ypmed.2011.07.014

a b s t r a c t

a r t i c l e i n f o

Available online 26 July 2011

Keywords:TrackingChildrenWeightTV viewingDietary intake

Objective. To examine the tracking of children's body mass index, television viewing, and dietary intakeover five-years.

Methods. In 2002/3 (T1) parents of children aged 5–6 years (n=175) and 10–12 years (n=121), fromVictoria, Australia, completed measures assessing their child's frequency of fruit, vegetable, and energy-densesweet and savory snack consumption, and their child's television (TV) viewing. Children's height and weightwere measured by researchers and sex–age adjusted body mass index (BMI) calculated. All measures wererepeated in 2006 (T2) and 2008 (T3). Generalized estimating equations (GEE) (standardized stability

coefficients, β) were used to assess tracking and were interpreted as: β b0.3=low, 0.3–0.6=moderate, andN0.6=high.

Results. High standardized stability coefficients were found for BMI (β=0.74–0.92), TV viewing(β=0.65–0.73), and frequency of fruit consumption (β=0.73–0.89) among younger and older children.Moderate-to-high standardized stability coefficients were found for frequency of vegetable (β=0.52–0.86),energy-dense sweet (β=0.41–0.65), and savory snack consumption (0.40–0.67) among younger and olderchildren.

Conclusions. BMI, TV viewing and dietary intake patterns are moderate-highly stable throughoutchildhood and into adolescence. Further research that identifies and targets high risk groups to preventincreased BMI, reduce TV viewing and promote healthy dietary behaviors may be justified.

© 2011 Elsevier Inc. All rights reserved.

Introduction

Modifiable lifestyle factors (e.g. dietary intake and activity) andadiposity are major risk factors of noncommunicable diseases (NCDs)(Hopkins and Williams, 1986; World Health Organization, 2008).NCDs are the consequence of risks and exposures originating in earlychildhood (Fuemmeler et al., 2009; Harsha et al., 1987) and mod-ifiable risk factors identified in adults are increasingly prevalent inchildren. Understanding the stability of lifestyle risk factors over timeis imperative.

Tracking is defined as 1) the stability of a certain risk factor overtime or 2) the predictability of a measurement of a certain risk factorearly in life for values of the same risk factor later in life (Twisk et al.,1997). The strength of tracking of weight status and adiposity variessubstantially between studies due to differences in length of follow-up

& Health Sciences,eicestershire, LE11 3TU, UK.n), jo.salmon@deakin.edu.aull),erio@deakin.edu.au

l rights reserved.

and anthropometric measures (Freedman et al., 2005; Serdula et al.,1993; Singh et al., 2008), while strength of tracking of young people'sdietary intake patterns shows variations by food type and length offollow-up (Kemper et al., 1999; Li andWang, 2008; Twisk et al., 1997).A recent review suggested that high levels of television viewing areestablished in childhood, with moderate-high levels of trackingthrough childhood and adolescence (Biddle et al., 2010). Few studies,however, have systematically examined how modifiable lifestylefactors track through childhood and from childhood through adoles-cence (Borehamet al., 2004; Kelder et al., 1994; Twisk et al., 1997), andeven fewer consider multiple risk factors for NCDs in the same study,making comparisons regarding the strength of tracking coefficientsdifficult. This study examined tracking of body mass index (BMI),television viewing, and frequency of fruit, vegetable, and energy-densefood consumption among a sample of Australian children in twodistinct age groups.

Methods

This study was approved by the Deakin University Human Research EthicsCommittee, the Victorian Department of Education and Training and the CatholicEducation Office.

269N. Pearson et al. / Preventive Medicine 53 (2011) 268–270

Sample

Baseline data were collected in 2002/03 (T1) and follow-up data in2006 (T2) and2008 (T3). At T1, 39 state and Catholic elementary schoolsin metropolitan Melbourne, Australia were selected from postcodes inthehigh,mid and lowsocioeconomic areas; 24 schools tookpart. In total,1612 families of first year primary school-aged children (5–6 years) and2085 families of grade 5–6 children (10–12 years) were invited toparticipate.

Active consent was received for 1562 children. Those whosefamilies agreed to be recontacted for future research (n=825) wereinvited to participate in the T2 follow-up. This process was repeatedfor T3. In total, 392 (25.1% of T1) participated at T2 and 312 (20% ofT1) at T3. The analyses presented here are based on a maximum of296 children with complete data at each time point for all of thevariables assessed in this study. Children from families that werefollowed up had lower BMI, and consumed fruit, savory and sweetenergy-dense foods less frequently than those who weren't followedup.

Measures

Height and weight were measured without shoes in private andbody mass index (BMI, kg/m²) calculated and converted as recom-mended for longitudinal analyses (Cole et al., 2005) by subtracting thesex–age population median (US data) (Kuczmarski et al., 2000) fromraw BMI scores (hereafter termed BMI units).

Parents reported duration/day their child usually watched com-mercial and non-commercial television/DVDs in a 1) typical week(Monday to Friday) and 2) typical weekend (Saturday and Sunday).Average duration of television viewing per day was computed, aspreviously described (Salmon et al., 2006).

Table 1Descriptive statistics for all outcome variables at each time points (T1: 2002/3; T2: 2006; T

Mean (SD) Younger children

Boys G

Age (years)Baseline/time 1 5.96 (0.39) 5

Body Mass Index (BMI units) n=76 nTime 1 1.07 (1.52) 1Time 2 1.62 (2.11) 1Time 3 0.81 (3.13) 1

TV viewing (min/day) n=85 nTime 1 155.9 (81.1) 1Time 2 169.5 (83.7) 1Time 3 185.7 (102.1) 1

Fruit (frequency/day) n=86 nTime 1 2.84 (2.07) 2Time 2 2.73 (1.74) 2Time 3 2.87 (1.89) 2

Vegetables (frequency/day) n=81 nTime 1 3.07 (2.02) 3Time 2 3.19 (2.07) 3Time 3 3.07 (1.64) 3

Energy-dense sweet foods(frequency/day)

n=84 n

Time 1 1.62 (1.47) 1Time 2 1.24 (1.22) 1Time 3 1.19 (1.00) 1

Energy-dense savory foods(frequency/day)

n=84 n

Time 1 1.02 (0.79) 0Time 2 0.79 (0.64) 0Time 3 0.78 (0.71) 0

Height and weight were measured without shoes in private by trained researchers and bodlongitudinal adiposity data (Cole et al., 2005). This involves subtracting the sex–age populatiFor convenience, these BMI units of difference from the sex–age population median are refIndependent T-tests between genders within age groups.⁎ pb0.05.

Parents were asked how often their child ate 14 different fruits, 13vegetables, nine types of energy-dense sweet foods (e.g. biscuits,confectionary), and four types of energy-dense savory foods (e.g. savorycrackers, potato crisps) in the last week. These items, adapted from the1995 Australian National Nutrition Survey (Australian Bureau ofStatistics, 1998), had acceptable test–retest reliability (ICC=0.44–0.96) over 2–3 weeks (n=93). Responses were recorded (scorespresented in parentheses) and summed to compute total frequency offruit, vegetables, energy-dense sweet snack and savory snack foodconsumption/day: 4 or more times/day (4); 3 times/day (3); twice/day(2); once/day (1); 4–6 times/week (0.714); 2–3 times/week (0.357);once/week (0.143); not eaten (0).

Statistical analyses

All analyseswere conductedusing Stata version11 (Statacorp, Texas,USA) and stratified by age group and gender. Descriptive statistics wereused to summarize the data at each time point. Tracking was examinedusing Generalized Estimating Equations (GEE). GEE measure theassociation between an indicator at T1 and the same indicator at allother periods (e.g. T2 and T3). GEE produce a standardized stabilitycoefficient (derived from the β-estimate and its estimated SD) that canbe interpreted as a correlation coefficient ranging from 0 (no relation-ship) to 1 (perfect relationship) (Twisk et al., 1997). Recommendationsfor interpreting these correlation-like coefficients are as follows: lowb0.3; moderate 0.3–0.6; high N0.6 (Malina, 1996; Twisk et al., 1997).

Results

Themean age of parents at baseline (T1)was 38.8 (SD=5.17) years.The majority were female carers (93%), born in Australia (72.2%),married (85%), had a university/tertiary qualification (47%) or had

3: 2008) stratified by age group and gender of Australian children.

Older children

irls Boys Girls

.92 (0.45) 11.11 (0.55) 11.12 (0.59)=79 n=51 n=49.21 (1.75) 1.31 (2.84) 1.40 (2.79).77 (2.68) 1.00 (2.73) 2.18 (3.62).70 (2.92) 2.01 (3.38) 2.10 (3.17)=90 n=62 n=5945.9 (72.3) 178.6 (100.2) 170.9 (84.4)⁎

64.9 (85.1) 196.3 (93.7) 182.3 (84.5)83.3 (94.5) 190.5 (110.5) 200.1 (146.1)=86 n=60 n=62.83 (1.96) 2.55 (2.27) 2.66 (1.90).52 (2.32) 2.01 (2.30) 2.59 (2.32).58 (1.69) 1.69 (1.64) 2.06 (1.41)=82 n=61 n=61.01 (1.69) 2.81 (2.01) 3.04 (1.50).24 (1.93) 2.93 (2.22) 3.21 (1.47).13 (1.40) 2.56 (1.48) 3.40 (1.41)=89 n=60 n=60

.54 (1.30) 1.43 (0.99) 1.65 (1.17)

.26 (0.86) 1.27 (0.91) 1.34 (1.33)

.18 (0.92) 1.36 (1.06) 1.56 (1.64)⁎

=89 n=60 n=60

.99 (0.79) 0.85 (0.68) 0.85 (0.53)

.81 (0.54) 0.70 (0.53) 0.77 (0.58)

.88 (0.65) 0.72 (0.69) 0.72 (0.63)

y mass index (BMI, kg/m2) calculated and converted as recommended for analyses ofon median (based on US data) (Kuczmarski et al., 2000) from the child's raw BMI score.erred to as BMI units.

Table 2Standardized stability coefficients (unstandardized SD) for all outcome variablesstratified by age group and gender of Australian children using generalized estimatedequations (GEE).

Standardized stabilitycoefficients, β(Unstandardized SD)

Younger cohort Older cohort

Boys Girls Boys Girls

Body Mass Index (BMI units) 0.74 (0.98) 0.78 (0.98) 0.87 (1.02) 0.92 (1.05)TV viewing 0.73 (0.88) 0.71 (0.88) 0.72 (0.95) 0.65 (0.71)Fruit 0.74 (1.06) 0.73 (1.05) 0.84 (1.15) 0.89 (1.10)Vegetables 0.81 (1.09) 0.77 (1.06) 0.86 (1.12) 0.52 (1.03)Energy-dense sweet snacks 0.65 (1.23) 0.47 (1.25) 0.41 (0.96) 0.57 (0.85)Energy-dense savory snacks 0.56 (1.09) 0.40 (1.13) 0.67 (1.01) 0.48 (0.92)

GEE measure the association between an indicator at the first period of measurements(T1: 2002/3) and the same indicator at all other periods (T2: 2006 and T3: 2008).All stability coefficients were significant at the pb0.001 based on unstandardizedstability coefficients.Unstandardized SD is calculated using the stata commands: sum ‘variable’ if time=1 |time=2 and sum ‘variable’ if time=2 | time=3. The produced standard deviations arethen divided to produce a ‘pooled’ SD which is the ‘unstandardized SD’.

270 N. Pearson et al. / Preventive Medicine 53 (2011) 268–270

completedhigh school (26.1%), halfwere in part-timepaid employment(48.9%) and 19.6% were in full-time paid employment. Boys and girlswere equally represented (50.8% girls) and 58.6% were in the youngerage group.

Descriptive statistics and sex differences for each of the outcomevariables are displayed in Table 1.

Stability coefficients for BMI units and each behavior are shown inTable 2. Stability coefficients for frequency of consumption ofvegetables were high for younger boys and girls and for older boys,and moderate for older girls. Stability coefficients for frequency ofconsumption of energy-dense sweet and savory foodsweremoderate-high.

Discussion

This study found a high degree of tracking of BMI units, TV viewingand frequency of fruit and vegetable consumption over 5 years inmiddle childhood and adolescence. In particular, from the age of five itappears that children's BMI position changes little relative to peers.This may in part be due to genetics, however ‘at risk’ groups should beidentified and targeted to prevent increases in BMI; especially sincehigh tracking was accompanied by increases in BMI units over time.Similarly, further research is needed to identify ways to ‘de-stabilize’TV viewing in young people, as both high tracking andmean increasesover time were observed. Evidence suggests that TV and dietaryinterventions that include reinforcement and cognitive strategies maybe most fruitful (Kamath et al., 2008).

Stability coefficients varied between the eating behaviors, withhigher coefficients for frequency of fruit and vegetable consumptioncompared to energy-dense foods, which had the weakest and leastconsistent results across age, corroborating previous findings (Li andWang, 2008). It may be that the types or frequency of energy-densefoods consumed vary more in response to environmental stimuli.However, given that snack foods could be ‘social’ foods consumedwith friends or outside the home, the results could be explained byinaccuracy in parental reports related to these foods.

Conflict of interest statementAuthors declare that there are no conflicts of interest.

Acknowledgments

The Health Eating and Play study (HEAPs) was funded by theVictorian Health Promotion Foundation (baseline) and the AustralianResearchCouncil (follow-ups,DP0664206). JS is supportedbyaNationalHeart Foundation of Australia/Sanofi aventis Career DevelopmentAward. KC, AT and DC are supported by Victorian Health PromotionFoundation Public Health Research Fellowships. NP and AT conceivedthe manuscript. NP analyzed the data and drafted the manuscript. AT,DC and JS designed the Health Eating and Play study (HEAPs) project,and all authors provided critical feedback on drafts, read and approvedthe final manuscript.

References

Australian Bureau of Statistics, 1998. National nutrition survey 1997. Nutrient Intakesand Physical MeasurementsAustralian Bureau of Statistics, Canberra.

Biddle, S.J., Pearson, N., Ross, G.M., Braithwaite, R., 2010. Tracking of sedentarybehaviours of young people: a systematic review. Prev. Med. 51, 345–351.

Boreham, C., Robson, P.J., Gallagher, A.M., Cran, G.W., Savage, J.M., Murray, L.J., 2004.Tracking of physical activity, fitness, body composition and diet from adolescenceto young adulthood: The Young Hearts Project, Northern Ireland. Int. J. Behav. Nutr.Phys. Act. 1, 14.

Cole, T.J., Faith, M.S., Pietrobelli, A., Heo, M., 2005. What is the best measure of adipositychange in growing children: BMI, BMI%, BMI z-score or BMI centile? Eur. J. Clin.Nutr. 59, 419–425.

Freedman, D.S., Khan, L.K., Serdula, M.K., Dietz, W.H., Srinivasan, S.R., Berenson, G.S.,2005. The relation of childhood BMI to adult adiposity: the Bogalusa Heart Study.Pediatrics 115, 22–27.

Fuemmeler, B.F., Pendzich, M.K., Tercyak, K.P., 2009. Weight, dietary behavior, andphysical activity in childhood and adolescence: implications for adult cancer risk.Obes. Facts 2, 179–186.

Harsha, D.W., Smoak, C.G., Nicklas, T.A., Webber, L.S., Berenson, G.S., 1987. Cardiovas-cular risk factors from birth to 7 years of age: the Bogalusa Heart Study. Tracking ofbody composition variables. Pediatrics 80, 779–783.

Hopkins, P.N., Williams, R.R., 1986. Identification and relative weight of cardiovascularrisk factors. Cardiol. Clin. 4, 3–31.

Kamath, C.C., Vickers, K.S., Ehrlich, A., McGovern, L., Johnson, J., Singhal, V., Paulo, R.,Hettinger, A., Erwin, P.J., Montori, V.M., 2008. Clinical review: behavioral in-terventions to prevent childhood obesity: a systematic review and metaanalyses ofrandomized trials. J. Clin. Endocrinol. Metab. 93, 4606–4615.

Kelder, S.H., Perry, C.L., Klepp, K.I., Lytle, L.L., 1994. Longitudinal tracking of adolescentsmoking, physical activity, and food choice behaviors. Am. J. Public Health 84,1121–1126.

Kemper, H.C., Post, G.B., Twisk, J.W., van Mechelen, W., 1999. Lifestyle and obesity inadolescence and young adulthood: results from the Amsterdam Growth AndHealth Longitudinal Study (AGAHLS). Int. J. Obes. Relat. Metab. Disord. 23 (Suppl.3), S34–S40.

Kuczmarski, R.J., Ogden, C.L., Grummer-Strawn, L.M., Flegal, K.M., Guo, S.S., Wei, R., Mei,Z., Curtin, L.R., Roche, A.F., Johnson, C.L., 2000. CDC growth charts: United States.Adv. Data 1–27.

Li, J., Wang, Y., 2008. Tracking of dietary intake patterns is associated with baselinecharacteristics of urban low-income African–American adolescents. J. Nutr. 138,94–100.

Malina, R.M., 1996. Tracking of physical activity and physical fitness across the lifespan.Res. Q. Exerc. Sport 67, S48–S57.

Salmon, J., Campbell, K.J., Crawford, D.A., 2006. Television viewing habits associatedwith obesity risk factors: a survey of Melbourne schoolchildren. Med. J. Aust. 184,64–67.

Serdula, M.K., Ivery, D., Coates, R.J., Freedman, D.S., Williamson, D.F., Byers, T., 1993. Doobese children become obese adults? A review of the literature. Prev. Med. 22,167–177.

Singh, A.S., Mulder, C., Twisk, J.W., van Mechelen, W., Chinapaw, M.J., 2008. Tracking ofchildhood overweight into adulthood: a systematic review of the literature. Obes.Rev. 9, 474–488.

Twisk, J.W.R., Kemper, H.C.G., vanMechelen, W., Post, G.B., 1997. Tracking of risk factorsfor coronary heart disease over a 14-year period: a comparison between lifestyleand biologic risk factors with data from the Amsterdam growth and health study.Am. J. Epidemiol. 145, 888–898.

World Health Organization, 2008. 2008–2013 Action Plan for the Global Strategy for thePrevention and Control of Noncommunicable Diseases. World Health Organization,Geneva.