Preventive Medicine 54 (2012) 201–204

Contents lists available at SciVerse ScienceDirect

Preventive Medicine

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

Habitual active transport, TV viewing and weight gain: A four year follow-up study

Ding Ding a,b,⁎, Takemi Sugiyama c,d, Neville Owen c,d

a University of California San Diego, La Jolla, California, USAb San Diego State University, San Diego, California, USAc Baker IDI Heart and Diabetes Institute, Melbourne, Victoria, Australiad University of Queensland, Brisbane, Queensland, Australia

⁎ Corresponding author at: 3900 5th Ave., Suite 31fax: +1 619 260 1510.

E-mail address: dding@projects.sdsu.edu (D. Ding).

0091-7435/$ – see front matter © 2012 Elsevier Inc. Alldoi:10.1016/j.ypmed.2012.01.021

a b s t r a c t

a r t i c l e i n f o

Available online 8 February 2012

Keywords:Sedentary behaviorTelevision viewingPhysical activityCommutingObesityWeight gain

Objectives. To examine the associations of TV viewing time and domain-specific physical activitywithweightchange; to determine whether domain-specific physical activity moderates the potential association of TVviewing time with weight change.

Methods.We used four-year longitudinal data (baseline: 2003–2004, follow-up: 2007–2008) on 969 adultsfrom selected neighborhoods in Adelaide, Australia (Age: 48.6±10.6 years, 61% females). Mixed models exam-ined four-year weight change as the dependent variable, with TV viewing time, habitual transport and pastweek domain-specific physical activity at baseline as independent variables.

Results. On average, participants gained 1.6 kg over four years. TV viewing time at baseline was positively asso-

ciated with weight gain at follow-up. Each additional hour of TV viewing was associated with 0.24–0.27 kg of extraweight gain. This relationship was not moderated by recent recall of transport, leisure-time, and occupationalphysical activity, but was moderated by habitual transport: an additional hour of TV viewing time at baseline wassignificantly associated with an extra weight gain of 0.65 kg at follow-up among those who were inactive in every-day transport; TV timewas not significantly associatedwithweight change among thosewhowere regularly activein transport.

Conclusion. Habitual active transport may protect adults against risk of weight gain associated with prolongedTV viewing time.

© 2012 Elsevier Inc. All rights reserved.

Introduction

Rising rates of obesity are a global threat to public health. In Austra-lia, approximately 60% of adults are overweight or obese (AustralianInstitute of Health andWelfare, 2010; Thorburn, 2005). Obesity preven-tion requires environmental, social and policy initiatives based on anunderstanding of the determinants of overweight and obesity in popu-lations (James, 2008).

Sedentary behavior (too much sitting) is a class of activities charac-terized by low energy expenditure and is associated with an array ofhealth outcomes, including excess adiposity (Tremblay et al., 2010).Sitting is ubiquitous in contemporary societies and can occupy a largeproportion of adults’ waking hours (Owen et al., 2010). Based onemerging evidence on the health impact of prolonged sitting, publichealth organizations now acknowledge the importance of reducingsitting time in addition to increasing physical activity, as a strategy toaddress the obesity epidemic (National Preventative Health Taskforce,2009).

0, San Diego, CA 92103, USA.

rights reserved.

There has been growing evidence for the association of sedentarytime with excess adiposity and this association often persists whenphysical activity is adjusted for (Dunstan et al., 2005; Tremblay et al.,2010). Two cross-sectional studies have examined whether specificdomains of activity may moderate this relationship: Dunton et al.(2009) found that playing sedentary video gameswas positively associ-atedwith BMI only among thosewho reported less than 60minutes perday of leisure-time physical activity. Sugiyama et al. (2010) found thatprolonged TV viewing time was associated with higher body massindex (BMI) among those who were inactive or occasionally active inhabitual transport behavior, but not among those who were regularlyactive. These studies suggest potential mitigating effects of domain-specific physical activity on obesity risk associated with prolongedsitting. However, evidence fromprospective studies is needed to under-stand how physical activity in particular domains and sedentary behav-iors may contribute to long-term weight maintenance or weight gain(Summerbell et al., 2009).

A recent systematic review of longitudinal studies concluded thatprolonged sitting, after controlling for the effects of moderate-to-vigorous physical activity, increases risks for diabetes, some cancers,and all-cause and cardiovascular disease-related mortality (Thorpet al., 2011). There is consistent evidence for the association of seden-tary behavior with weight gain from childhood to adulthood, but

202 D. Ding et al. / Preventive Medicine 54 (2012) 201–204

insufficient evidence in relation toweight gain during adulthood (Thorpet al., 2011).Wijndaele et al. examined interaction between baseline TVviewing time and overall physical activity levels in relation to change inBMI among colorectal cancer survivors and found a significant associa-tion of BMI increase with prolonged TV viewing time among those whowere physically inactive (Wijndaele et al., 2009). However, no study hasyet examined whether the association of sedentary behavior withweight gain varies by level of domain-specific physical activity.

We examined whether baseline TV viewing time and three specificdomains of physical activity (transportation, leisure-time, and occupa-tion) were independently associated with weight change over fouryears; and whether specific domains of physical activity moderatedthe association of TV viewing time with weight change.

Methods

Data were from PLACE (Physical Activity in Localities and CommunityEnvironments), a longitudinal study examining associations of neighborhoodenvironmental attributes with physical activity. Details of study design andprocedures have been documented elsewhere (Owen et al., 2007). Briefly,households were randomly selected from neighborhoods stratified based oncensus collector district (CCD) level walkability and socioeconomic status inthe city of Adelaide, Australia. In 2003–2004, 2650 adults returned completedbaseline questionnaires (11.5% of the residential addresses initially identified);and in 2007–2008, 1098 (41.4% of the baseline participants) completed thefollow-up survey. After excluding those with missing values and extreme BMI(b15 or >50) values, the final study sample included 969 with complete data.Baseline characteristics of the total baseline sample (n=2650) and the sampleretained for follow-up analyses (n=969) are presented in Table 1. There wasno significant difference between the two samples except for the follow-upsample being older at baseline (pb0.01).

Respondents reported their height and weight at both baseline and follow-up assessments, and TV viewing and physical activity were measured at base-line. TV viewing timewas assessed using previously validated questions regard-ing the number of days and the average amount of TVviewing per day in the lastseven days (Salmon et al., 2003). Daily TV viewing time (hours/day) was calcu-lated bymultiplying the average amount by the number of days and then divid-ed by seven. This measure was found to have excellent test-retest reliability(intraclass correlation coefficient=0.82) and reasonable validity (Spearman'srank-order correlations with a 3-day activity log=0.3, pb .01). Habitual trans-port behavior was measured at baseline using an instrument assessing every-day commuting activities, including going to work, shopping, taking a child toschool/day care and running other errands that lasted at least 10 minutes(Miilunpalo et al., 2000). The original question had eight response categoriesand was re-coded as “inactive,” “occasionally active,” and “regularly active,”consistent with our previous study (Sugiyama et al., 2010). This instrumentwas found to have good reliability and content validity (Miilunpalo et al.,2000). Transport, leisure-time, and occupational physical activities in the past

Table 1Baseline characteristics of the total sample and the follow-up sample (Adelaide, Australia;

Tota

Age (year), mean (SD) 44.5Gender, % women 63.7%Educational attainment, % with tertiary education 45.5%Household income, % median household income and above a 49.3%Employment status, % working 65.0%TV viewing (min/day), mean (SD) 112.9

Habitual active transportInactive 30.4%Occasionally active 35.9%Regularly active 33.7%Transport physical activity (min/day), median (interquartile range) 14.3Leisure-time physical activity (min/day), median (interquartile range) 17.1Occupational physical activity (min/day), median (interquartile range) 0.0 (Body weight (kg), mean (SD) 75.3Body mass index, mean (SD) 26.0

a Median household income=$41600/year **pb0.01

seven days were measured using the long version of the International PhysicalActivity Questionnaire at baseline (Craig et al., 2003).

To account for potential design effects of clustered sampling, mixedmodels were used with CCD as a random effect variable. The dependent var-iable was weight change from baseline to follow-up (positive values indicat-ed weight increase). The independent variables, all measured at baseline,were TV viewing time and physical activity (habitual transport, seven-day re-call transport, leisure-time, and occupational physical activities). The covari-ates adjusted for were baseline weight and demographic variables, includingage, gender, educational attainment, household income, and employmentstatus. In each model, TV viewing time, a physical activity variable, andtheir interaction termwere entered. When a TV time×physical activity inter-action term was significant, stratified analyses were conducted, modeling theassociation of TV time with weight change separately for each physical activ-ity stratum. All independent variables were grand-mean centered to improveinterpretability. Unstandardized regression coefficients for both main effectsand interaction terms were reported (Table 2), which can be interpreted asan adjusted average weight change (in kg) associated with a one-unit in-crease in an independent variable. Data were analyzed using SPSS 17.0.Alpha was set to 0.05 for both main effects and interactions.

Results

At baseline, the study sample had an average weight of 75.2 kg(SD=15.8) and an average BMI of 26.1 (SD=4.73), 35%were classifiedas overweight (BMI≥25) and an additional 18% as obese (BMI≥30).From baseline to follow-up, the average weight gain was 1.6 kg(SD=5.8). The association of baseline TV viewing time with weightchange at follow-up was significant or approaching significance inmost models (Table 2): every additional hour in TV viewing time atbaseline was associated with an extra weight gain of 0.24 to 0.27 kgover the four-year period. Habitual transport behavior, past-seven-daytransport, leisure-time, and occupational physical activities were notsignificantly associated with weight change. There was, however, asignificant interaction for TV viewing time with habitual transport.The interaction terms for TV viewing time and the three past-seven-day physical activity variables (transport, leisure-time, occupational)were non-significant (Table 2).

Based on the results of these interaction analyses, we examinedspecific associations of TV viewing time with weight gain for adultswith each category of habitual transport (Fig. 1). For those who wereinactive, an hour's increase in TV viewing per day was associated withan additional weight gain of 0.65 kg at follow-up (β=0.65, 95% CI:0.21, 1.09). For thosewhowere occasionally active, therewas a positivetrend but it did not reach statistical significance (β=0.30, 95%CI: −0.12, 0.72). For those who were regularly active in habitual

2003–2004).

l sample at baseline (n=2650) Follow-up sample at baseline (n=969)

(12.3)** 48.6 (10.6)61.0%46.2%52.5%64.1%

(97.5) 112.5(87.8)

30.5%38.7%30.4%

(2.9, 38.6) 12.6 (2.9, 34.3)(0.0, 40.0) 17.1 (0.0, 38.6)0.0, 65.7) 0.0 (0.0, 51.4)(18.5) 75.2 (15.8)(5.1) 26.1 (4.7)

Table 2Weight changea (kg) according to baseline TV viewing time and physical activity from baseline (2003–2004) to follow-up (2007–2008) (n=969; Adelaide, Australia).

Habitual transport Transport PAb Leisure-time PA b Occupational PA b

Intercept 0.74 (−0.49, 1.98) 0.49 (−0.61, 1.58) 0.61 (−0.49, 1.70) 0.72 (−0.39, 1.84)TV time (hour/day) 0.26⁎ (0.01, 0.54) 0.26† (−0.02, 0.54) 0.24† (−0.04, 0.51) 0.27⁎ (0.01,0.54)Habitual transport

Occasionally activec −0.22 (−1.15,0.71)Regularly activec 0.05 (−0.95, 1.05)TV time×occasionally active −0.24 (−0.85, 0.36)TV time×regularly active −0.81⁎ (−1.47, −0.15)

Seven-day recall PATransport PA (hour/day) −0.26 (−0.79, 0.28)TV time×Transport PA 0.25 (−0.14, 0.64)Leisure-time PA (hour/day) −0.44 (−1.02, 0.14)TV time×Leisure-time PA −0.31 (−0.77, 0.16)Occupational PA (hour/day) 0.07 (−0.10, 0.24)TV time×Occupational PA 0.02 (−0.09, 0.14)

Presented numbers are unstandardized regression coefficients (95% confidence interval).All models adjusted for age, gender, educational attainment, household income, employment status, and baseline weight. Individuals clustering within district was adjusted for byincluding census collector district as a random effect variable in mixed models.

a Positive values mean weight gain.b PA=physical activity.c Reference: inactive.† pb0.10.⁎ pb0.05.

203D. Ding et al. / Preventive Medicine 54 (2012) 201–204

transport, no significant association was found between TV time andweight change (β=− 0.23, 95% CI: −0.82, 0.35).

Discussion

This is the first longitudinal study to examine interacting effects ofTV viewing and specific domains of physical activity on weight change.We found an association of TV viewing time at baseline with weightgain at follow-up. In contrast, physical activity in the transport, leisure,and occupational domains was not significantly associated with 4-yearweight change. With regard to interactions, habitual active transportmoderated the association of TV viewing timewithweight gain. Consis-tentwith our previous cross-sectional study (Sugiyamaet al., 2010), ourfindings suggest that habitual active transport may protect adultsagainst the risk of obesity associated with prolonged TV viewing time.

Several studies have documented active transport to be protectiveagainst weight gain. A European study found that middle-aged menwho regularly walked or cycled to work had a lower risk of weightgain over a five-year period (Wagner et al., 2001). Similarly, a recentstudy in the USA reported that the use of light-rail transit is associatedwith a reduction in body mass index (MacDonald et al., 2010). A longi-tudinal study in China showed that men who acquired a motor vehiclegained significantly more weight than those who did not do so (Bellet al., 2002). Along with our research, these studies suggest that

Fig. 1. Unstandardized regression coefficients of TV viewing time on weight change(representing adjusted average weight change in kg associated with a one-hour increasein TV viewing time) among thosewhowere regularly active, occasionally active, and inac-tive in their habitual transport behavior (baseline: 2003–2004, follow-up: 2007–2008;Adelaide, Australia).

promoting active transport is likely to be an effective individual- andcommunity-level strategy for obesity prevention.

In our study, themoderation of the relationship between TVviewingtime and weight gain was only observed for habitual active transport,but not for other physical activity variables, including recent recall oftransport physical activity. There are several possible explanations.First, unlike physical activity variables measured for the last sevendays, the habitual transport behavior measure captures everyday trans-port behavior that is more likely to reflect a long-term lifestyle(Miilunpalo et al., 2000). Given that the weight change was assessedover a four-year period, it is possible that a routine behavior pattern ismore predictive of long-termweight outcomes than shorter-termphys-ical activity. Second, active commuters tend to live in walkable neigh-borhoods where transit stops and other destinations are easy to get to(Saelens and Handy, 2008). Such high walkable neighborhoods mayoffer more health promoting opportunities, such as easily accessiblegrocery stores, which may help local residents maintain a healthyweight (Moudon et al., 2006).

Although prospective design is a major strength, these findingsshould be interpreted in the light of limitations. Reporting biases mayhave influenced TV viewing time and physical activity measures. Also,the weight-change variable was derived from self-report of weight attwo-time points, which could also have introduced elements ofmeasurement error. We did not include long-term measures for otherdomains of physical activity, and therefore could not assess effects ofhabitual leisure-time and occupational physical activity to compareand contrast with active transport.

Conclusions

Based on our findings, adultswhowatchedmore TV at baselineweremore likely to gain weight after four years. Those who were inactive intransport were particularly susceptible to weight gain associated withTV viewing. Although the effect sizes were small, considering the habit-ual nature of both TV viewing and transport behavior, their significantimpacts on body weight throughout lifetime at the population levelshould not be underestimated.

In addressing the prevention of weight gain in populations, there isthe need to promote physical activity that can be sustainable and habit-ual; physically active transport meets these criteria. Research is thusneeded to identify factors that facilitate long-term participation in ac-tive transport. In addition, this study found an association of prolonged

204 D. Ding et al. / Preventive Medicine 54 (2012) 201–204

TV viewing with weight gain, which is consistent with findings fromother longitudinal studies (Ball et al., 2002; Hu et al., 2003). Obesityprevention initiatives need to address prolonged sitting during leisuretime (particularly TV viewing) aswell as in automobiles andworkplace.

Conflict of interest statement

There is no conflict of interest to report.

Acknowledgments

Ding, Owen and Sugiyama were supported by National Health andMedical Research Council of Australia Program Grant #569940, by aResearch Infrastructure Grant from Queensland Health and by Fellow-ship #1003960 (Owen).

References

Australian Institute of Health and Welfare, 2010. Australia's health 2010. Australia'shealth no. 12. Cat. no. AUS 122. AIHW, Canberra.

Ball, K., Brown, W., Crawford, D., 2002. Who does not gain weight? Prevalence and predic-tors of weight maintenance in young women. Int. J. Obes. Relat. Metab. Disord. 26,1570–1578.

Bell, A.C., Ge, K., Popkin, B.M., 2002. The road to obesity or the path to prevention: motor-ized transportation and obesity in China. Obes. Res. 10, 277–283.

Craig, C.L., Marshall, A.L., Sjostrom,M., Bauman, A.E., Booth, M.L., Ainsworth, B.E., Pratt, M.,Ekelund, U., Yngve, A., Sallis, J.F., Oja, P., 2003. International physical activityquestionnaire: 12-country reliability and validity. Med. Sci. Sports. Exerc. 35,1381–1395.

Dunstan, D.W., Salmon, J., Owen, N., Armstrong, T., Zimmet, P.Z., Welborn, T.A., Cameron,A.J., Dwyer, T., Jolley, D., Shaw, J.E., 2005. Associations of TV viewing and physical ac-tivity with the metabolic syndrome in Australian adults. Diabetologia 48, 2254–2261.

Dunton, G.F., Berrigan, D., Ballard-Barbash, R., Graubard, B., Atienza, A.A., 2009. Joint asso-ciations of physical activity and sedentary behaviors with body mass index: resultsfrom a time use survey of US adults. Int. J. Obes. (Lond.) 33, 1427–1436.

Hu, F.B., Li, T.Y., Colditz, G.A., Willett, W.C., Manson, J.E., 2003. Television watching andother sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitusin women. JAMA 289, 1785–1791.

James, W.P., 2008. WHO recognition of the global obesity epidemic. Int. J. Obes. (Lond.)32 (Suppl 7), S120–S126.

MacDonald, J.M., Stokes, R.J., Cohen, D.A., Kofner, A., Ridgeway, G.K., 2010. The effect of lightrail transit on body mass index and physical activity. Am. J. Prev. Med. 39, 105–112.

Miilunpalo, S., Nupponen, R., Laitakari, J., Marttila, J., Paronen, O., 2000. Stages of changein two modes of health-enhancing physical activity: methodological aspects andpromotional implications. Health Educ. Res. 15, 435–448.

Moudon, A.V., Lee, C., Cheadle, A.D., Garvin, C., Johnson, D., Schmid, T.L., Weathers, R.D.,Lin, L., 2006. Operational definition of walkable neighborhood: Theoretical andempirical insights. J. Phys. Act. Health 3, S99–S117.

National Preventative Health Taskforce, 2009. Australia: The Healthiest Country by2020 - National Preventative Health Strategy. Commonwealth of Australia,Canberra.

Owen, N., Cerin, E., Leslie, E., duToit, L., Coffee, N., Frank, L.D., Bauman, A.E., Hugo, G.,Saelens, B.E., Sallis, J.F., 2007. Neighborhood walkability and the walking behaviorof Australian adults. Am. J. Prev. Med. 33, 387–395.

Owen, N., Healy, G.N., Matthews, C.E., Dunstan, D.W., 2010. Too much sitting: thepopulation health science of sedentary behavior. Exerc. Sport Sci. Rev. 38,105–113.

Saelens, B.E., Handy, S.L., 2008. Built environment correlates of walking: a review. Med.Sci. Sports Exerc. 40, S550–S566.

Salmon, J., Owen, N., Crawford, D., Bauman, A., Sallis, J.F., 2003. Physical activity andsedentary behavior: a population-based study of barriers, enjoyment, and prefer-ence. Health Psychol. 22, 178–188.

Sugiyama, T., Merom, D., Reeves, M., Leslie, E., Owen, N., 2010. Habitual active transportmoderates the association of TV viewing time with body mass index. J. Phys. Act.Health 7, 11–16.

Summerbell, C.D., Douthwaite, W., Whittaker, V., Ells, L.J., Hillier, F., Smith, S., Kelly, S.,Edmunds, L.D., Macdonald, I., 2009. The association between diet and physicalactivity and subsequent excess weight gain and obesity assessed at 5 years of ageor older: a systematic review of the epidemiological evidence. Int. J. Obes. (Lond.)33 (Suppl 3), S1–S92.

Thorburn, A.W., 2005. Prevalence of obesity in Australia. Obes. Rev. 6, 187–189.Thorp, A.A., Owen, N., Neuhaus, M., Dunstan, D.W., 2011. Sedentary behaviors and subse-

quent health outcomes in adults a systematic review of longitudinal studies,1996–2011. Am. J. Prev. Med. 41, 207–215.

Tremblay, M.S., Colley, R.C., Saunders, T.J., Healy, G.N., Owen, N., 2010. Physiological andhealth implications of a sedentary lifestyle. Appl. Physiol. Nutr. Metab. 35, 725–740.

Wagner, A., Simon, C., Ducimetiere, P., Montaye, M., Bongard, V., Yarnell, J., Bingham, A.,Hedelin, G., Amouyel, P., Ferrieres, J., Evans, A., Arveiler, D., 2001. Leisure-timephysical activity and regular walking or cycling to work are associated with adiposityand 5 y weight gain in middle-aged men: the PRIME Study. Int. J. Obes. Relat. Metab.Disord. 25, 940–948.

Wijndaele, K., Lynch, B.M., Owen, N., Dunstan, D.W., Sharp, S., Aitken, J.F., 2009. Televisionviewing time and weight gain in colorectal cancer survivors: a prospectivepopulation-based study. Cancer Causes Control 20, 1355–1362.