You are on page 1of 14

Banks et al.

BMC Public Health 2011, 11:762
http://www.biomedcentral.com/1471-2458/11/762

RESEARCH ARTICLE Open Access

Relationship of obesity to physical activity,
domestic activities, and sedentary behaviours:
cross-sectional findings from a national cohort of
over 70,000 Thai adults
Emily Banks1*, Lynette Lim1, Sam-Ang Seubsman2, Chris Bain3 and Adrian Sleigh1

Abstract
Background: Patterns of physical activity (PA), domestic activity and sedentary behaviours are changing rapidly in
Asia. Little is known about their relationship with obesity in this context. This study investigates in detail the
relationship between obesity, physical activity, domestic activity and sedentary behaviours in a Thai population.
Methods: 74,981 adult students aged 20-50 from all regions of Thailand attending the Sukhothai Thammathirat
Open University in 2005-2006 completed a self-administered questionnaire, including providing appropriate self-
reported data on height, weight and PA. We conducted cross-sectional analyses of the relationship between
obesity, defined according to Asian criteria (Body Mass Index (BMI) ≥25), and measures of physical activity and
sedentary behaviours (exercise-related PA; leisure-related computer use and television watching ("screen-time”);
housework and gardening; and sitting-time) adjusted for age, sex, income and education and compared according
to a range of personal characteristics.
Results: Overall, 15.6% of participants were obese, with a substantially greater prevalence in men (22.4%) than women
(9.9%). Inverse associations between being obese and total weekly sessions of exercise-related PA were observed in
men, with a significantly weaker association seen in women (p(interaction) < 0.0001). Increasing obesity with increasing
screen-time was seen in all population groups examined; there was an overall 18% (15-21%) increase in obesity with
every two hours of additional daily screen-time. There were 33% (26-39%) and 33% (21-43%) reductions in the adjusted
risk of being obese in men and women, respectively, reporting housework/gardening daily versus seldom or never.
Exercise-related PA, screen-time and housework/gardening each had independent associations with obesity.
Conclusions: Domestic activities and sedentary behaviours are important in relation to obesity in Thailand,
independent of exercise-related physical activity. In this setting, programs to prevent and treat obesity through
increasing general physical activity need to consider overall energy expenditure and address a wide range of low-
intensity high-volume activities in order to be effective.
Keywords: Obesity, Thailand, physical activity, inactivity, domestic activity, sedentary behaviours

Background 19% from 1997 to 2004 alone [2]. There have been
The prevalence of obesity is rising rapidly in most Asian accompanying increases in morbidity related to condi-
countries, with increases of 46% in Japan and over 400% tions such as diabetes and cardiovascular disease in
in China observed from the 1980s to early 2000s [1]. In Asian countries [3,4].
Thailand, the prevalence of obesity increased by around It is well established in Western populations that
increasing purposeful or leisure-time physical activity
* Correspondence: emily.banks@anu.edu.au (PA) is associated with reduced rates of obesity [5,6].
1
National Centre for Epidemiology and Population Health, Australian Recent evidence, also from Western countries, suggests
National University, Canberra, Australia that sedentary activities, such as watching television or
Full list of author information is available at the end of the article

© 2011 Banks et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.

at home. work-related factors. computer games? Sitting for any purpose (e. e. sensory resting.11]. The questionnaire requested participant’s response to the question “How many hours information on: socio-demographic factors. dental PA and overall energy expenditure. cise (not exhausting but breathe harder than normal) fers between Asia and Western countries [15] and for more than 20 minutes. e. In brief. The questions on housework/garden- been standardised and validated were used. for: “Strenuous exercise (heart beats rapidly) for more more. per day do you usually spend: Watching TV or playing past and present residence and domestic environment. are associated with increasing obesity. drink-driving. ing.com/1471-2458/11/762 using a computer.g. Total daily leisure-related screen- cross-sectional self-reported data from the Thai Cohort time and sitting time were classified according to the Study questionnaire [17]. e. the transition in risk factor profiles. This paper examines in detail the relationships bowling” and. calculated as “2 × strenuous + moderate to join the study by completing the questionnaire. health outcomes included in these measures. and family structure lowing does your home have now?”. exercise”. because many Asian countries are experiencing rapid cycling at a regular pace". “Walking for at least 10 minutes e. by the square of their height in metres. enrolled STOU students across The responses to this question were used to derive a Thailand who had completed a least one semester weighted measure of overall metabolically-adjusted exer- were mailed a 20-page health questionnaire and asked cise-related PA. in influencing Information on exercise-related PA was obtained obesity has been highlighted [10. use of health services. including microwave oven.g. reading. It incorporates the three major Cohort Study is designed to provide evidence regarding intensities of activity (strenuous. water heater and naires).Banks et al. general include screen-time. these in a reply-paid envelope. physical activity. availability of domestic appliances was classified accord- tobacco and alcohol consumption. social asked to exclude screen-time from this measure. the relationship between domestic activities and than 20 minutes. similar to the ses- emphasis on the interaction between these factors. other Asian populations [19]. Tai-Chi. fied into 5 groups according to the response to the question “How often do you do household cleaning or Data gardening work?” with options ranging from seldom or All of the variables used in this study were derived from never. + mild + walking” exercise sessions. effort) for more than 20 minutes. how many times on average do you do the obesity are not well understood and information rele. screen-time. post-secondary school certifi- and ≥25. Sitting time could also impairment. “Mild exercise (minimal health and lifestyle transitions [16]. “Moderate exer- physical activity related to patterns of daily activity dif. medical history. mental health. yoga. This question is a sessions- and sedentary behaviours in Thailand. respectively.g. sions component of the International Physical Activity Questionnaire and the Active Australia Survey [20]. weight. Education attainment was classified as: secondary neating overweight and obesity were set at BMIs ≥23 school graduation or less. following kinds of exercise?”. domestic activities work. sedentary behaviours.g. use of seat belts and ing to the response to the question “Which of the fol- motorcycle helmets. income. in accordance with International cate or diploma. with particular based measure of physical activity. and returning vious calculations of this quotient [20]. in keeping with pre- viding signed consent for follow-up. A total of 87 134 men The frequency of reported housework and gardening and women aged 15-87 years (median 29 years) joined was used as a measure of incidental PA and was classi- the cohort. to November 2005. aerobics or obesity is unclear [12-14]. soccer. The capital. diet. heavy lifting. from April study to cover common types of activity in Thailand. Personal monthly . divided Study. at between obesity. running. sitting time and availability of domestic Self-reported weight and height were used to calculate appliances were devised specially for the Thai Cohort participants’ BMI. trakraw". digging. carrying light loads. Further. moderate and walking). The interplay through a question asking: “During a typical week (7- between these factors and their combined effects on day period).biomedcentral. and tertiary graduate. pro. questionnaire items that had washing machine. 11:762 Page 2 of 14 http://www. with responses requested vant to Asian populations is particularly scarce. The role of inci.g. as their weight in kilograms. working thinking)?”. Obesity Taskforce recommendations [18] and studies in independent of purposeful PA [7-9]. social networks. Cut-points deli. exercise-related PA. Where possible. to most days. with options and health (See Additional files 1 and 2 for question. as participants were not specifically health. ethnicity. refrigerator. as well as an additional and other factors accompanying development and is “mild” category that was created specifically for this described in detail elsewhere [17]. BMC Public Health 2011. height. This is important because fast bicycling. Methods these session measures have been shown to provide a Study population reliable index of sufficiency of physical activity in non- The Sukhothai Thammathirat Open University (STOU) Thai populations [21].

with BMIs between 11 and 50. conclu- ging from never or less than once a month. housework. height or weight (n healthy weight (BMI 18.7-15.7% (3.5-9.1%) of men reported examined how much of the association of certain seden. respectively). housework. while women were more the correlation between the individual measures of phy.7 [sd 10.8-12. mathirat Open University Research and Development gorised as <2 and ≥2 serves per day.8-55. less than a tertiary education. The relationships between a range of personal character. They were asked if Institute (protocol 0522/10) and the Australian National they have ever smoked.001-20.0- 24.0%.8%) reporting ≥18 ses- sitting time were estimated using unconditional logis. using a significance (known as “junkfood”) on a five-point Likert scale ran.6] for women. The mean number of sessions of exercise- and women and adjusted for age (as a continuous related PA per week was 11.8-22. Due to the large sample size.4%).5% (12. all participants. 22. gories for alcohol consumption.4% table to differences in total physical activity level by (33.4% (48. A higher exploration of the effect of additional adjustment for level of exercise-related PA was associated with having factors such as marital status. 14. times and ≥ 3 times per week for fried food and seldom (never or less than once per month) and regularly (≥ Ethical approval once per month) for “junkfood”. We evaluated the and eating more fruit and vegetables. sedentary behaviours and to consumption of fried foods not tertiary educated. 5.000. Fruit and vegetable Ethics approval was obtained from Sukhothai Tham- intakes were noted as serves eaten per day and cate.6%) were underweight (BMI < 18.1%) or obese (22. housework/gardening and 21.9%. Of 74 981 participants with appropriate data. We also 3. computed. istics and exercise-related PA.biomedcentral. Patterns of exercise-related PA varied between men ined.9%) of women and 8.3-22.1% (11. consumption of fried food (data not shown) [22].9-21.5% and 9.000.6-6. 13.1] overall. BMC Public Health 2011.8% (8. between the sexes.9 [sd variable). study cohort.com/1471-2458/11/762 income was in Thai Baht in four categories (≤7. smoking. The pattern ratio test.5-22.6%. as well as (9.0). 10.7%.2%) were overweight but not obese (BMI ≥ 23. income and educational attainment.1- = 1030) or physical activity or inactivity (n = 6863). respectively. Overall.0-22. 41 351 Individuals were excluded from the analyses if they were (55. this was categorised as consumption <3 size. while 3. 10 733 (14. sions of exercise-related PA per week compared to tic regression. leisure-related screen-time and sitting time was exam. fic PA or sedentary behaviour with obesity was attribu. 20.9%) of women and 34.000. 11:762 Page 3 of 14 http://www. men to report strenuous or moderate PA (Table 2).5] for men and 9. obesity prevalence was higher in older par- The proportion of the study population classified as ticipants and urban dwellers and in those with higher obese according to exercise-related PA. comparing the model with and without the of PA making up the total weekly sessions also differed interaction terms. but was not significance of interaction terms using a likelihood strongly related to other factors (Table 1). Variables were categorised (5.5- their two-way and three-way interactions. that they did these seldom or never.9%. screen-time and 0-3 sessions and 26. with 12.001-10.1%).9%) of men reporting obesity according to PA.Banks et al. alcohol con.8%) of men reported doing household cleaning modelling simultaneously the three PA variables and or gardening on most days of the week. ORs were presented separately for men for women. being of lower income sumption and urban/rural residence.7%) than men sical activity and inactivity.9).2.2% (21. Informed written consent was obtained from not current smokers. to once or sions were based on both significance and the effect more a day. 11 241 (15. Respondents mutual adjustment.9) and 11 616 (15. 14.4%.3% (25. with women much less likely than We examined how much of any association of a speci. using 7.2-15.2-12. likely to be underweight (21. 14.9-49.5). 95%CI 54. with similar questions and cate.981 participants. leaving 74. and soft drinks and Western-style junkfood. level of p < 0.05. Housework/garden- tary behaviours could be attributed to the effect of other ing was more common among those who were married.6 [sd 12.7%) were obese (BMI ≥ Statistical methods 25. of lower income and with greater .9%) than women leisure-related screen-time were examined. Analysis was restricted to the 95% of individuals aged Results between 20 and 50 years. with 13. crude and adjusted odds ratios were 22. recorded the frequency of eating deep fried food and All analyses were carried out in STATA version 9. Compared to other members of the into the groups listed in the various tables.8-26. Prevalence odds ratios (OR) and 95% CIs for and women.4%. soft drinks and Western-style fast foods such as pizza All statistical tests were two-sided.6%) and 12. 49. >20.3%.2%. when they started and when University Human Research Ethics Committee (protocol they quit and were categorised as current smokers or 2004344).8-34.000). Men were far more likely to be overweight (21.5%) were classified as being of missing data on age or sex (n = 2). 15.

8 [12.8-22.9] 2.7 [19143] 47.6[15535] 60.1] 3.024 to -0.9%) of women and 36. 17.8 [1.7 [2.2 [11693] 26.9] in women and 6. hours/day) 6.6 [1670] 10.5] 2. 0.9 [1753] 10.2%) of women CI: -0. sessions/week) 3.8 hours [sd 3.5 [41608] sleeping (mean [SD].3 [7.0 [1.0 [3185] 9. 18 or more sessions had a OR of obesity of 0.047) and -0.0 [2325] 21.6%.3 [16136] 44.9 [3323] 20.3 [12555] 30.2] 3.4] physical activity (mean [SD].3] 21.7] 6.8 [33591] age (mean [SD].4 [10029] 30. Women tended to have greater levels of sitting time Obesity and exercise-related physical activity. 95% between men and women. 6.16.1 [10168] 25.2] 30.3 [13.9 [8917] 23.6 [10188] 28.2 [7.6] 21.6 [1388] 9.6 [12044] 35.4 [3.4 [3.7 [13645] 31.9 [5065] 15.8 [12.9 [2. The number of more.9 [2.6] fruit intake (mean [SD]. kg/m2) 21.5 [17656] 49.0 [20958] 57.6] 1.4 [7091] regular alcohol drinker. The correlations between sitting more common among cohort members who were time and number of weekly sessions of exercise-related younger.016.4 [9451] 31.9 [3768] 20.2] 7.9 [16178] 48.4 [1314] 9.1 [3.9 [3. respectively.9] 30.7 [21373] 55.1 [7.6] 4.9]. Higher levels of screen-time were (r = 0.1 [19471] 48.0 [25689] limited physical functiona (%[n]) 10.5 [31775] urban resident (% [n]) 55.0 [17472] 37.2 [7.3] 22.0 [2.029 to -0.9 [1. the OR for being obese decreased steadily and sitting time. housework/gardening and daily screen-time total physical activity (per housework/gardening (per leisure screen-time (per Total week) week) day) <7 sessions ≥7 sessions <2 times ≥2 times <3 hours ≥3 hours TOTAL (n) 31995 42986 33373 41608 35936 39045 74981 male (% [n]) 33.4 [11466] consume soft drinks/junkfood (% [n]) 29.3 [8855] 24. of females are current smokers and regular drinkers fruit and vegetable intake than other cohort members The number of hours of daily screen-time was poorly (Table 1). but significantly inversely correlated with the number of Leisure related screen-time did not vary markedly weekly sessions of exercise-related PA (r = -0.2 hours sions of exercise-related PA.7 [6489] 14.0 [17983] 40.0 [1.6 [12.2] 6.8] 3.8] in men.2] 6.3] married (%[n]) 44.0 [2.6 [3.1] housework ≥2 times/week (% [n]) 48. while 3. Average daily sitting time was 6.8] 6.2 [7.9 [2.9] 1. serves/day) 2.2-37.061 to and who ate fried food daily and soft drinks or Western -0.8% (2.8] 6.041 (-0.2% (21.7 [3.0] 12.9] in women and 2. males onlyb (% [n]) 12.1 [2.9 [1.8 [1.5 [14303] 43. housework.4 [2.2] leisure screen-time (mean [SD].2 [1.2 [3535] 21.5] 2.2 [3912] 9.1 [11. Average daily leisure related screen-time was 2.8 [38614] tertiary educated (% [n]) 27.9] 11.9 [13437] 30.5] 29.022.5] 2. -0. serves/day) 1.6 hours [sd significantly with increasing weighted total weekly ses- 3.2 [3423] eat fried food daily (% [n]) 15.16).7 [6472] 15.7 [3556] 22.8 correlated with doing household cleaning or gardening hours [sd 1. hours/day) 3.8 [21255] 51. style junkfood once a month or more often (Table 1).9%) of men reported 8 hours or ting per day (r = 0.6%) of strongly and positively related to number of hours sit- women and 2.8 [3.4] 11. BMC Public Health 2011.3] 21.2 [13117] 46.1 [6.5 [8778] 26.9] 6.15 to 0.9 [12430] 29.3 [20235] 54.8] overall. it was 3. than men.6] 21.9 [2.3 [10668] 53.9 [1. unmarried.9] 6.6 [2.8% and gardening (36.7] 1.0 [4994] 16.5] 17.014) but was more hours of daily screen-time.054 (-0.5] 2.6 [17359] 54.9] 2.1 [1.com/1471-2458/11/762 Table 1 Characteristics of study population according to total physical activity. 0.6-2.7 [26073] 56.9 [2.009) and doing household cleaning or and 22.3] 2.7 [1.7%) of men reported less than two gardening (r = -0.8 [2.3 [18946] income ≥10.6 [3.2] 30.3] 6.034) respectively.16).2 [12252] 38.9 [2.5 [17455] 41.4-18.0 [6401] 15.2 [21881] vegetable intake (mean [SD].1 [9326] 29.4] Participants with missing values are not included in the percentages a Measured as top decile from 3 items assessing physical limitations in the past 4 weeks (eg how much bodily pain did you have in the past 4 weeks?) b Only 1% and 0.0 [1.6] 1.8] in men.8% (17.2 [7.9] 11.8 [11539] 33. with 46. urban residents and of lower income PA and cleaning/gardening were -0.0 [3. years) 30.9 [3.biomedcentral.0 [4977] 15.3] 2.63- .Banks et al.9 weekly sessions of exercise-related PA was positively hours [sd 1.0 [1.4% (3.69 (0.7 [2035] 8.5 [7097] current smoker.3%) of men reporting 8 or more hours of daily In men.4] 30.0 [3.7] 2. such that those reporting [sd 1.3] 10.6% (46.2-3.9] 7.14 to 0.1 [13792] 43.8 [3.3 [22928] 52.2 [18658] 50.3] 31.049 to -0.0 [2. males onlyb (% [n]) 22.6 [3432] 9.7 [3.4 [11.9 [1. hours/day) 6.8] body-mass index (mean [SD].0-46.7 [14150] 40.1 [4766] 21.5 [2.0 hours [sd 1.000 Baht per month (% [n]) 36.2 [17413] 38.9] sitting (mean [SD].6] 2.7 [4024] 9.5] 21.3 [10091] 21.3] 7.6 [3.4 [3665] 9.1] 6.3 [3073] 9. 11:762 Page 4 of 14 http://www.3 [2109] 11.15.5 [0.

77-0.77-0.9) 1.7) 0.1) 1.00 1.87-1.5) 0.01) 0.10) 4 sessions/week 1826 375 (20.00 1.1) 1.78-0.42) 1.58-0.77-0.07 (0.02-1.8) 1.00 1.16) 5+ sessions/week 17767 3744 (21.18) 1.05-1.75) compared to those with 0-3 sessions (p(trend) < moderate and strenuous PA in women and the relation- 0.17) 1.0001).4) 0. There was no nificantly between men and women (p(interaction) < apparent relationship between being obese and 0.18) 1.96) 1081 131 (12.93) 0.81 (0.00-1.68-0.72) 0.89-1.4) 1.99-1.5) 0.66-0.85) 0.91-1.85) 0.30) 3538 391 (11.76-0.8) 1.9) 0.04) 5677 550 (9.77 (0.05) 0.91 (0.76 (0.01-1.90) 0.06 (0.03 (0.82-1.002 *adjusted for age.85 (0.93) 0.75-0.71 (0.77-0.48) 5+ sessions/week 7136 1281 (18.9) 0.98 (0.17) 1.83 (0.77-0.00 1.67-0.86 (0.94-1.29) 815 65 (8.96 (0.91) 4 sessions/week 939 237 (25.98) 0.25) 4194 392 (9.0001 weighted total sessions exercise-related physical activity 0-3 sessions/week 4201 1180 (28.83-1.81-1.5) 0.81-1.12 (1.03 (0.0001 0.5) 0.81-0.6) 0.72-0.15 (0.13) 1.99) 0.00 23972 2341 (9.28) 1.00 1 sessions/week 3482 815 (23.92) 3333 316 (9.94) 0.99) 7-11 sessions/week 8640 1997 (23.3) 0.06 <0.20) 1.73 (0.0) 0.10 (0.89) 0.0001 0.82 (0.0001 0.84 (0.81-1.75-0.9) 1.94) 7639 744 (9.83) 6783 612 (9.7) 1.90-1.70 (0.4) 1.04 (0.85 (0.73-0.88 (0.78 (0.03) 4 sessions/week 2390 517 (21.00 1.85 (0.97 (0.87-1.05) 0.00) 0.90 (0.00 9199 1022 (11.74 (0.85-1.81 (0.27) 1. income and education .00) 0.24) 1.81 (0.04) 0.86 (0.09) 3 sessions/week 4066 859 (21.98-1.00 1 sessions/week 2696 698 (25.96) 4647 429 (9.11 (1.4) 1.00 23459 2321 (9.24) 3 sessions/week 4409 938 (21.94) 5+ sessions/week 4313 871 (20.11 (1.92) 0.1) 1.94-1.96 (0.004 <0.92 (0. 11:762 Page 5 of 14 http://www.4) 0.20 (0.92) 1081 124 (11.1) 0.90) 3721 353 (9.27) 1.1) 0.00-1.06 (0.84 (0.84 (0.08) 2486 210 (8.90 (0.76-0.00 27176 2838 (10.0) 1.90-1.0001) and this relationship was observed particularly ship between strenuous activity and obesity differed sig- for moderate and strenuous PA (Table 2).00 1.97) 0.75-0.77-0.70-0.94-1.17) 3 sessions/week 2636 658 (25.03) p(trend) <0.8) 0.81 (0.87 (0.biomedcentral.2) 0.95) 0.04) 5774 543 (9.79 (0.77) 2571 271 (10.02) 0.17 (1.82-1.00 1 sessions/week 4297 1003 (23.72 (0.90) 12775 1190 (9.95 (0.04 (0.09) 0.65-0.3) 0.95) 0.83-1.3) 0.94-1.92 (0.96-1.17 (1.90 (0.09) 0.82-0.6) 0.13) p(trend) <0.85-1. BMC Public Health 2011.90) 0.03 (0.1) 0.75-0.18) 2 sessions/week 4677 1022 (21.85-1.6) 1.01) 0.93-1.2) 0.85 (0.75) 0.5) 0.07) 0.73-0.87-1.82-0.19) 3484 375 (10.00 1.96-1.6) 1.1) 1.94-1.17 (0.90 (0.25) 1.14 (1. in women an inverse relationship Table 2 Relationship between being obese and measures of exercise-related physical activity (PA) MALES FEMALES Total Obese age adjusted adjusted Total Obese age adjusted adjusted n n (%) OR (95% CI) OR (95% CI)* n n (%) OR (95% CI) OR (95% CI)* strenuous exercise-related physical activity >20 mins 0 sessions/week 10924 2772 (25.13 (1.94-1.91) 0.85-1.0) 0.96 (0.1) 0.88) 18+ sessions/week 8829 1672 (18.88-1.10 (0.90 (0.71-0.27) 2039 211 (10.5) 0.81 (0.00 1 sessions/week 4518 1073 (23.99 (0.86-1.90-1.97-1.00-1.78-0.92) 12-17 sessions/week 7722 1645 (21.5) 0.07) 0.03 (0.89 (0.96 (0.2) 0.12) 4598 419 (9.71-0.94 (0.19 mild exercise-related physical activity >20 mins 0 sessions/week 20590 4671 (22.05-1.81-1.81-1.00 1.14) 1.80-0.93 (0.00) 3219 298 (9.90 (0.45) 1.10 (0.97-1.98 (0.60-0.88-1.97 (0.00-1.00 1.3) 1.87 (0.5) 1.87 (0.92) p(trend) 0.4) 1.96-1.10 (0.02 (0.00 1.98) 4 sessions/week 1559 378 (24.3) 0.77-0.30) 1.14) 0.00 (0.96 (0.4) 1.64-0.4) 0.66 (0.6) 1.98 moderate exercise-related physical activity >20 mins 0 sessions/week 12171 2991 (24.85 (0.91 (0.84) p(trend) 0.01 (0.89-1.81-0.74 (0.76-0.69 (0.88-1.87-1.79 (0.79-1.00 6528 726 (11.15) 2 sessions/week 4214 945 (22.78-0.86-1.0) 0.99 (0.1) 1.5) 0.40) 5+ sessions/week 6554 1276 (19.84 (0.89-1.66-0.56-0.97) 0.Banks et al. However.18) 2 sessions/week 2345 560 (23.95 (0.80-0.03 (0.0001 walking for ≥10 minutes 0 sessions/week 6588 1488 (22.com/1471-2458/11/762 0.25) p(trend) <0.00 4-6 sessions/week 4199 1032 (24.65-0.83 (0.11 (0.95-1.81-1.05) 3 sessions/week 1879 423 (22.7) 1.63-0.94-1.3) 0.92 (0.90 (0.2) 1.3) 0.83-0.05) 21607 1995 (9.87) 3193 350 (11.97 (0.94) 3096 260 (8.0) 1.04) 2 sessions/week 2388 509 (21.81) 0.9) 1.09 (0.99) 4271 451 (10.90) 0.75) 4994 525 (10.75 (0.22 (1.

80) for housework/gardening on most days.23 (1.83).6) 1.83-1.38-1.76 (0.64-0.87) for 1-2 times per 1.57-0.27-1. 0.85-1.Banks et al.18) 1.92) 0.85 (1.59) p(trend) <0.0001 screen-time 0-1 hours/day 7452 1568 (21.9) 1.31) p(trend) 0.98) 4709 464 (9.00 1550 198 (12.58 (1.39-1.30).00 1.57 (1.80 (1.82-1.0001 sitting-time 0-1 hours/day 2115 510 (24.71-0.50) 10112 1089 (10.33) 6-7 hours/day 4248 928 (21.00 1.59-0.04-1.85) 0.26) 1.5) 0.8) 0. respectively.4) 1.0001 <0.66-0.7) 0. with 33% (26-39%) and 33% (21-43%) lower (0.87 (0.5) 0.35- (0.99-1.22 (1.36-1.72 (0. being obese was still present following additional adjust- the ORs (95% CI) for being obese in men were: 0.94 (0.18) 1398 191 (13. and 1.72-0.96-1.92 (0.00 1.68-0.00) 7797 701 (9.51 (1. relationship between housework/gardening and obesity although a significant trend was observed in women was still present following additional adjustment for (Table 3).84).29) 20694 1935 (9.85) 5076 482 (9. the ORs for the same categories were: 0.00 2-3 hours/day 7651 1661 (21.50 (1.00 1.80 (0.16) 0.76 (0.60-0.6) 0. 1.2) 1.00 1.20) 8+ hours/day 12296 2792 (22.16 (1.biomedcentral.88 (0.65 (0.79-1.13) 1-2 times/week 9491 2092 (22.27).15-1.69-0.67) 6-7 hours/day 1227 300 (24. 1.75 (0.18) 1.75-0.2) 1. 0. for 8 or more hours adjustment for exercise-related PA (see below) and a of daily screen-time versus <2 hours (Table 3.87) 3-4 times/week 4688 1040 (22. 1.71-0.31-1.71 gardening.00 2-3 hours/day 17109 3862 (22.85 0.95) 0.79-1.79) p(trend) <0.72) 0.63-0.85 (0. in those reporting these activities daily versus seldom or never Obesity and leisure related screen-time (Table 3).03) 19177 1933 (10.92 (0.96 (0.63-0.77-0.54) 1. < 0.90).79-0.42 (1. for people Table 3 Relationship between being obese and gardening/housework.18 (1.76 (0.0001 <0.82-1.49 (1.85-1.8) 1. changed by <10%) following additional 116% increases in risk. For For both sexes.35-1.86 (0.92) and 2. 11:762 Page 6 of 14 http://www.32). with 85% and materially (i. 0.95 (0.83-2.56-2. screen-time and exercise-related PA.81-1.65) 1.5) 0.97 (0.14 (0.84) 1847 202 (10. education. screen-time and walking (Table 2).84-1.60-0.15 (1.283) 1.87) 1.68-0.9) 1. income and education numbers do not always sum to total due to missing values .05-1.61 (1.86) 1.76-0.00 1.60) 2. the risk of being obese was consis.8) 1. Overall.76-0.74 (0. respectively.04) 0.8) 0.21 (1.02 (0.94) for people who did housework or gardening the ORs (95% CI) for obesity in men were: 1. 1.74) 20354 1966 (9.6) 1.98-1.64-0. p(trend) similar relationship was observed within separate cate.00 7339 673 (9.98-1.61-0.64-0.0) 0.73 2. and 0.78-2.7) 1.86) 0.15 (1.63 (1.82-1.38 (1.04-1.04) 6850 714 (10.86) 0. women.5) 1.99) 0.com/1471-2458/11/762 with being obese was observed mainly for mild PA and adjusting for age.95 tently lower with increasing frequency of housework/ (0.1) 1. sitting-time was not significantly gories of exercise-related PA (Figure 1).53- week.13) 0. The lower risk of obesity with increasing Increasing leisure-related screen-time was associated housework or gardening was independent of the level of with significantly and substantially increasing risk of exercise-related PA.7) 0.90) for 3-4 times per week.11) 4-5 hours/day 7102 1598 (22.83 (0.32) 1. 1.38 (1.36-1.81-2. and 0.85 (1.67 (0.06) 0. respectively.67).14 (0.2) 0.92 (0.62 (1. adjusted ORs in men and women.001 *adjusted for age.14- 1-3 times per month. 0. BMC Public Health 2011.5) 0. The inverse related to the OR of being obese (p(trend) = 0.1) 1.78-1. leisure-related computer or television use ("screen-time”) and sitting time MALES FEMALES Total Obese age adjusted adjusted Total Obese age adjusted adjusted n n (%) OR (95% CI) OR (95% CI)* n n (%) OR (95% CI) OR (95% CI)* gardening or housework seldom or never 2962 806 (27.33) 1.26) 4-5 hours/day 6875 1560 (22.87) Most days 11490 2371 (20.88 (0.77 (0.e.13 (1.57-2.0) 1.13-1. (0.67 (0.00 (0.14 (0.50).00 2600 250 (9.00 1-3 times/month 4960 1217 (24.27-1.78 (0.85 ment for housework/gardening and exercise-related PA.93) 3200 378 (11.0001).79 (0.0) 0.55).7) 2.76 (0.76 (0.84) 11210 1066 (9.4) 1.95 (0.62-0.99) 0. The positive relationship between screen-time and ple who did housework or gardening seldom or never.11 (0.58 (1. in that the OR did not change being obese in both men and women. weighted weekly sessions of exercise-related PA. Compared to peo.77-0.89) 0.36-1. income.80-1.15).7) 0.90).13) and for women were: 1.94) 8+ hours/day 928 236 (25.15 (1.4) 1.36-1.73 (0.32-1.

respec. icant interactions were observed (likelihood ratio c392 = ment for consumption of fried foods. with increasing screen-time in unmarried compared to sists. even when screen-time and exercise-related PA married individuals (p(heterogeneity) < 0.0-2. hours of daily screen-time and frequency of housework/gardening. with separate lines 18% (15-21%) increase in the risk of being obese with according to the frequency of housework/gardening.49).9 hours. hours Housework/gardening Most days 1-4 times/week <4 times/month ERPA = sessions of exercise-related physcial activity per week Figure 1 Odds ratios (OR) for being obese in relation to weighted weekly sessions of exercise-related physical activity (ERPA).9 hours. 11:762 Page 7 of 14 http://www. When the relationships between hours of daily screen-time versus 0-1. with: 2.0-3. adjusted for age. being obese and exercise-related PA.biomedcentral. 4-7. BMC Public Health 2011. income and education-adjusted OR of obesity Figure 1 shows the cohort divided into four groups per two-hour increase in daily screen-time is shown according to their level of exercise-related PA (ERPA in separately according to a variety of factors. There was an hours of total daily screen-time. indicating that they were each indepen- junkfood and smoking and alcohol consumption did not dently associated with obesity. in Figure 2. It lation sub-groups examined (Figure 2).Banks et al.0001). housework/ housework/gardening were modelled together.9 hours. Additional adjust. There was a sig- also shows that the finding of lower risks of being obese nificantly greater increase in the risk of being obese with increasing frequency of housework/gardening per. materially alter the OR (data not shown). 3. screen-time and tively.com/1471-2458/11/762 0-3 ERPA 4-6 ERPA 7-11 ERPA 12+ ERPA 4 3 Odds Ratio 2 1 0-1 2-3 4-5 6+ 0-1 2-3 4-5 6+ 0-1 2-3 4-5 6+ 0-1 2-3 4-5 6+ Screen-time. p = 0. every two additional hours of daily screen-time overall This figure shows increasing risk of being obese with and a significant elevation in the risk of being obese increasing screen-time within each exercise-related PA with increasing screen-time was seen in all of the popu- group and within each housework/gardening group. according to total exercise-related PA and according to sented within each group according to the number of housework and gardening. The . education.9 hours. soft drink and 38. including the figure). The sex. no signif- gardening and exercise-related PA. The ORs for being obese were then pre. income. and ≥8 were accounted for.54.

in different population sub-groups. No significant variation in the status. 11:762 Page 8 of 14 http://www. education. The size of the squares is proportional to the amount of statistical information (i. junkfood and fried food . alcohol. adjusted for age.com/1471-2458/11/762 In the figure. income and education. seen according to the other factors examined. vegetable. fruit. the squares represent the OR and the horizontal lines the 95% confidence intervals. relationship between being obese and screen-time was relationship between screen-time and being obese was attenuated significantly in older cohort members (p(het.Banks et al. *in men only p(het)= p value for heterogeneity comparing the OR for obesity per 2 hours of additional daily screen time in the different sub- groups Figure 2 Odds ratios (OR) for being obese per 2 hour increase in daily screen-time. smoking (heterogeneity) = 0.e. including: erogeneity) = 0.02). inversely proportional to the variance of the log OR). sex. where appropriate. BMC Public Health 2011.biomedcentral. urban/rural residence history.02) and in those with higher incomes (p sex.

0) 1.86-1.76) 1.0) 1. smoking.88) 1.00 yes 9513 2769 (29.05-1.4) 1.00 1.48) 1.00 1.89 (1.09) 1. including those in China [23] and obese is consistently higher in those with greater time Korea [24].95-1.55 (1.00 1.00 29129 2623 (9.23-1. shown). Inverse associations include: the lack of data relevant to Asia.81) 1.09-1. Western-style junkfood.2) 1. the specific relationship of leisure-related PA spent in leisure-related television watching and compu.94 (1.40) 8916 585 (6.00 1. The risk of being there is an 18% increase in the risk of obesity with every obese increased significantly with the increasing number two hours of additional daily screen-time.92) 3 7038 1899 (27.34-1.1) 1.2) 1.Banks et al.25-1.00 1.41-1. usually leisure-related PA is well established in Western countries [5. ship between exercise-related PA and being obese that ment for housework/gardening frequency.00 1546 88 (5. The magnitude of the Obesity and domestic appliances association with obesity relating to these risk factors is The risk of being obese was significantly higher in men substantial.99-1.53 (1. the possibility between obesity and total weekly sessions of exercise.47-1.37) 2 11865 2676 (22.19) 1.29 (1.23) 1.00 yes 31526 7303 (23.23 (1.00 1.11 (1.26) water heater no 24269 4810 (19.58 (1.66 (1.75) 40019 4023 (10. and fre.17-1.36 (1.00 1.49 (2.57) 1.51-2. These results The findings reported here show an inverse relation- were not altered materially following additional adjust.00 1.2) 1. disability.38-1.00 yes 9600 2830 (29.44) 1. fruit and vegetables (data not The inverse relationship between obesity and exercise.56 (1.22-1.76 (1.83-2.60 (1.27) 9278 1045 (11.32) 12791 1518 (11. Table 4 Relationship between being obese and ownership of household appliances MALES FEMALES Total obese age adjusted adjusted Total obese age adjusted adjusted n n (%) OR (95% CI) OR (95% CI)* n n (%) OR (95% CI) OR (95% CI)* washing machine no 10414 1499 (14. and may be of lesser magnitude.68) 29849 3296 (11.7) 1. that the proportion of total energy expenditure to lei- related PA are observed in men with a significantly sure-related PA is lower in the Asian context [15].7) 2.00 28139 2482 (8.63-2.96-1.13 (1.26-1.5) 1.3) 1. to those reporting these activities seldom or never and holds with a water heater (Table 4).00 1901 118 (6. Individuals reporting daily housework/gar- and women from households with a refrigerator.8) 1.46-1.19) 1.70 (1.25-2.89-2.24 (1. weaker association seen in women.22 (1.35-1.6) 1.09-1. screen-time and housework/gardening each have inde- pendent associations with obesity. quency of housework/gardening.05) 14515 1458 (10.53 (1.51) 1.biomedcentral.8) 1. micro.61) microwave oven no 24182 4749 (19. income and education .17 (1.9) 1.79) 1.43 (1.20 (1.6) 1.02-1.09 (0.66-2. ter games and inversely associated with time spent The reason for this is not known.00) *adjusted for age.90) 2.07 (0.34 (1.57 (1.49-1.com/1471-2458/11/762 intake.00 1. Exercise-related PA. Potential explanations doing housework or gardening.42) 13781 1659 (12. to obesity is less clear.6.04-1.02 (0. BMC Public Health 2011.0) 1.07-1.26-1. is stronger in men than in women and may be some- alcohol consumption and consumption of fried foods. the risk of being Asian populations.00 12071 845 (7.94) 1. of such appliances within a household.31 (1. Although a reduced risk of obesity Discussion with increasing PA has been demonstrated in certain In this cohort of Thai men and women. 11:762 Page 9 of 14 http://www.15) number of appliances 0 1973 222 (11.08 (0.21-1.0) 1.00 1.61) 7665 965 (12.41) 1. dening have a 33% lower risk of being obese compared wave oven or washing machine and in men from house.58-1.00 1 7487 1065 (14.33 (1. what weaker than that observed in Western populations.3) 1.43) 1.0) 2.68 (1.13-2. level of exercise-related PA.6) 1.47) refrigerator no 2256 276 (12.0) 1.2) 1.6) 1.21].96) 4 5419 1717 (31.1) 1.32 (1.00 yes 23368 6080 (26.0) 1.

For example.28]. particularly what younger and more urbanised than the Thai general screen-time [7. Although the cohort is some- increasing time spent in sedentary activities. geographic region of residence. The measures used for tobacco smoking and alco- the different household appliances to obesity and the hol consumption were brief and the physical activity apparently greater effect in men compared to women measure used has. only been .13. suggests that this may well not be a causal relationship. although one study in older US but not obese is similar to the 18% found in the third adults found house cleaning. increasing numbers of appliances within the household. even The proportion of the cohort classified as overweight heavy domestic activity. males and 47% of females had low incomes (<7000 Baht sured and categorised screen-time and other sedentary per month or $5. 35% of [8. The limitations of the study should also be borne in However. We found house. the 18% increase in obesity bution. However. the play of parisons [36]. income. independent of representative of the general population. small sample sizes or other factors. It is therefore examining the relationship between BMI and domestic likely to provide useful early insights into the effects and activity in the Asian context demonstrated a significantly mediators of the health-risk transition in middle-income lower BMI in men with increasing time spent in domes. Studies in Western populations lates of obesity [22] and the relationship between gen- have generally not found a significant relationship der. is not designed to be frequency of housework and gardening. provide sufficient heterogeneity of exposure to allow Although the lack of a positive finding in the Western reliable estimates of relative risk based on internal com- context may reflect measurement error. ahead of national education trends.8.Banks et al. but not gardening. viduals. the major compari- been noted in one Chinese study [29]. 11:762 Page 10 of 14 http://www. it represents well the geographic regions of sedentary behaviours and obesity is observed in both Thailand and exhibits substantial heterogeneity in the cross-sectional [7. Increasing viours and health conditions may be less likely to be use of labour saving devices is part of the transition attending an open university or to participate. The direct relationship between population.37].biomedcentral. countries. sons in this paper are between obese and non-obese hold ownership of domestic appliances to be signifi. which is also likely to lead to more conservative with increasing risks of being obese accompanying estimates of association. However. from that in Western countries. for example. to be Thai National Health Examination Survey (2004). Previous relevant studies from the cohort tic activities and a non-significant relationship in include examination of the broader health-related corre- women.25. due to use since community members with more extreme beha- of labour saving devices or differing practices. socioeconomic status and obesity [35]. adults of the relationship between being obese and tele.26] and prospective studies distribution of other factors [17].com/1471-2458/11/762 differing types and intensities of leisure-related PA com. Participants in the behaviours. and differences in measurement it may instead reflect a broader difference in socioeco- error. like shows a decreasing risk of being obese with increasing the vast majority of cohort studies. The study presented vey) but similar among men in the two studies (23%) here is the largest to date investigating the issue and [2]. educa- risk per 2 hours of additional daily screen-time observed tion and course of study [32]. so it is Thai Cohort Study in 2005 were very similar to the difficult to summarise quantitatively the magnitude of STOU student body in that year for sex ratio.15. Furthermore. Strengths of the current study include its large size vision and computer use in Asia. accompanying industrialisation and is associated with it is important to note that ORs comparing groups reduced energy expenditure in domestic tasks [31]. The “healthy cohort effect” and the 44% chance. it is also response rate for this study means that the estimates of possible that domestic PA in Asian countries differs relative risk shown here are likely to be conservative. mediated by education [33. Much of the health-risk here is consistent with the 23% increase observed in US transition underway in middle income countries is nurses [8] and older Australian adults [9]. pared to the West. Studies have varied in the way they have mea. rather than obese and “healthy weight” indi- cantly associated with increasing risk of being obese. within the cohort remain valid and can be generalised Decreasing domestic physical activity over time has more broadly [36. in China [29]. BMC Public Health 2011. but is meant to exercise-related PA and screen-time. the lack of specificity in the relationship of mind.34] and the cohort is. Studies in Western and inclusion of adults from a wide range of social and populations consistently show increases in obesity with economic backgrounds. individuals. between BMI and domestic activity [12.50 US per day). to our knowledge. as well as obesity related outcomes. It should be noted that the Thai Cohort Study. nomic status and lifestyle between households with and We were unable to locate any previous studies in without appliances. while associated with decreased BMI on multivariate analysis obesity is much lower among STOU women (10% com- [14] and another found decreased all-cause mortality pared to 36% in the National Health Examination Sur- with increasing domestic PA [30].30]. by defi- The one previous publication we were able to locate nition.25-27]. age distri- the risk involved.

Furthermore. as was done here. the latter is also termed “incidental” PA. ambulatory activities. However. PA is often conceptualised as com. following adjustment for terised by a shift in work patterns for a substantial pro- pedometer measured physical activity [42]. which have been shown 16% was attributable to domestic activities and yard to provide a valid measure of body size in this popula. The gen. We were not able to locate tively high levels of exercise-related physical activity. particularly television watching. pollution and other hazards. Although dietary factors are impor.016) goes against this argument. including obesity and Asian populations occurs least when BMI is about 22 or metabolic disorders. domestic .biomedcentral. which typically accounts for 70% of all kilojoules chronic malnutrition has been common in many Asian burned [15]. 11:762 Page 11 of 14 http://www. is also tions. these changes are likely to increase population obesity. All of energy expenditure. work of at least moderate intensity and less than 5% tion. Moreover. volume daily intermittent low-intensity postural or The excellent correlation between measured and self. including global obesity epidemic [10]. Although both domestic activities and screen-time fatty food and in other studies following adjustment for remain predictive of obesity within categories of physical total energy intake [8] and foods eaten while watching activity. recent evidence indi- [38]. weight-related disease in ing cardiovascular risk factors. with correlation coefficients for BMI based on self. any studies validating these measures in Asian popula. This evidence reported versus measured height and weight of 0. A further 10% of energy expenditure comes countries. Cut-points delineating overweight and obesity were cates that one of the most potent mechanisms determin- set at BMIs ≥23 and ≥25. a more sedentary lifestyle. ity than in ascertaining domestic and screen-related In lower. including activities [25]. as food availability improves. inadequate urban individual energy expenditure is the basal metabolic infrastructure. erally reliable for ranking participants according to BMI Sedentary behaviours generally involve sitting or lying in epidemiological studies. although expenditure. since it may co-exist with rela- ber of hours involved [41]. the finding of increased obesity screen-time and overall physical activity are all likely to among those watching greater amounts of television be reported with differing degrees of measurement persisted in this dataset after adjustment for intake of error. with -0. tant. However. the lack of strong correlation Thailand. incidental physical activity associated with other health behaviours. sively [44-46]. portion of the population.and middle-income countries. was attributable to leisure-time PA [43]. which account for as much of 90% reported values mean that self-reported values are gen. the metabolic syndrome ity. level energy expenditure from PA was from sedentary social factors are key upstream determinants of the and low-intensity tasks. it is possible that this is the result of greater television [27]. self-reported height and weight. BMI was based on occupational activity above and beyond sitting at work. as does the previous increasing car and computer use and a higher fat diet observation that television viewing remains a significant dominated by more refined foods [16]. For example. so this is unlikely to explain much of its measurement error in ascertaining overall physical activ. 16% was attributed to major influences on obesity.Banks et al. whereby energy intake exceeds energy-demanding jobs in the service sector [16]. mainly resulting from their impact on inci- respondents have a tendency to under-report the num. A substantial of BMI from self-reported measures are less reliable. from high energy expenditure Obesity and overweight are the result of sustained activities such as farming. industrialisation is generally accompanied by between screen-time and overall physical activity (r = increasing urbanisation. of energy expended in physical activity [11]. mining and forestry to less positive energy balance. effects. The main determinant of environmental factors such as heat. In Thailand. is the amount of time spent in high less [39]. BMI based on dental physical activity are responsible for the greatest self-reported measures was underestimated by an aver. Screen-time.77 kg/m2 for men and 0.95 for women [38]. variations in energy expenditure between individuals age of 0.62 kg/m2 for women and populations [10]. and significantly increases with BMI ≥23 [40]. leading to stunting in significant portions of from the thermic effect of food and the remaining 20% the population and rendering them vulnerable to obesity comes from PA [15].91 for is consistent with the suggestion that differences in inci- men and 0. such as eating is difficult to measure [42] and domestic activities. prised of purposeful and non-purposeful physical activ. amount of time spent in sedentary activities is likely to Self-reported leisure-related television and computer contribute to obesity through reduced overall energy use has good test-retest reliability and validity. However. It is also charac- predictor of obesity in women.com/1471-2458/11/762 validated in Western populations. The importance of obesity. fatty foods. BMC Public Health 2011. dental physical activity. down and are characterised by low energy expenditure eral underestimation of BMI means that absolute values (metabolic equivalent intensity <2) [41]. rate. A recent and diabetes in Thailand has been highlighted exten- study from the US found that over half of population. there is mounting evidence that insufficient energy There are a number of specific barriers to increasing expenditure is likely to be a key factor underlying the physical activity in many Asian countries. and in many other countries. Time spent in habitual.

35]. but is increasing rapidly. Suttanit that increasing BMI by overfeeding of lean individuals Hounthasarn. Research focusing ber of activity-related measures and the large numbers on habitual activities and sedentary behaviours is rela- allowed quantification of the association of these factors tively new. the prevalence of overweight and obesity in status was accompanied by increasing risk of being Thailand is lower than that seen in many Western obese in men and decreasing risk of being obese in nations. which further exacerbates inactivity. context are needed urgently. higher socioeconomic tries. Thailand. In this case. women in particular are under Thinkamrop and his team from Khon Kaen University for guiding us pressure to be thin and the increased walking among successfully through the complex data processing. In Thai society. Going against a by the Australian National Health and Medical Research Council. and keeping with other data that show that increasing Western populations where high socioeconomic status exercise-related leisure-time PA alone is unlikely to be is associated with reduced obesity in both men and sufficient to prevent population obesity [15]. due to the extra exertion required because of their Australian NHMRC (268055). School of Human Ecology. This study was supported by the International Collaborative Research Grants Scheme order to lose weight or may reduce their exercise-related with joint grants from the Wellcome Trust UK (GR071587MA) and the PA. low intensity PA is likely to be strate high socioeconomic status to be associated with important for maintaining a healthy weight and are in increased obesity in both men and women [47]. Chaiyun Churewong. Avoiding the transi- women [35]. Anthony McMichael. or indeed a National Centre for Epidemiology and Population Health. Seubsman. since the major risk factor for incident obesity is having a Australia: Chris Bain. An English language translation of a questionnaire administered to students of Sukhothai Thammathirat Open University in decreased domestic PA. Gordon high BMI at baseline [49]. We thank the staff at Sukhothai Thammathirat weight or obesity related health issues (e. Cathy Banwell. are causal i. Lynette Lim. Sam-ang issue is not resolved entirely by using longitudinal data. Australia. Daoruang Pandee. Sukhothai Thammathirat Open University. Nintita Sripaiboonkij. joint pro. Vasoontara Yiengprugsawan be complex. Acknowledgements ple with a high BMI may do more exercise-related PA in We thank the 87 134 Thai students who took part in the Study. Reverse causality occurs when an exposure varies because of the Additional material specific condition under investigation. it is also Sukhothai Thammathirat Open University in 2005. In this cohort. is likely that certain elements of the PA-BMI relationship Additional file 2: Thai Cohort Study baseline questionnaire (Thai). Author details 1 tribution of obesity leading to inactivity [48]. We also thank Dr Bandit blems). related to the risk of being obese. reverse causality would mean that obesity might result in reduced Additional file 1: Thai Cohort Study baseline questionnaire exercise-related PA. which demon. with a causal relationship between inactivity and obesity predominating. This Tippawan Prapamontol. Bruce Caldwell. Leisure- women [2. The data presented here suggest that habi- less-developed countries such as China. high volume. Brisbane. Tarie Dellora. particularly to do In common with many middle to low income coun- the heavier work. Australia. Emily Banks is supported women of higher BMI may reflect this. possible that people with high BMI may change their level of PA. We propose that the relation. does not result in increased sedentary behaviour [49]. Lyndall Strazdins. “spiral” relationship. Australian National University. Suwanee Khamman. Intuitively. Pathumvadee Somsamai. Effective interventions to reduce sedentary with obesity within a range of population subgroups. 2Thai Health-Risk Transition Project. Queensland.48] and experimental studies show Study team: Thailand: Jaruwan Chokhanapitak. Adrian Sleigh. There are a priori reasons why it 2005. University of increases in obesity [50]. Janya Puengson. Matthew ship between sedentary behaviour and obesity is likely to Kelly. Yodyiam Sangrattanakul. How. whereby inactivity leads to obesity.biomedcentral. that reduced energy expenditure due to The Thai language original questionnaire administered to students of reduced PA results in increased BMI. BMC Public Health 2011. ships observed or to exclude reverse causality. Sharon Friel. However.Banks et al. . related television and computer use were strongly The study was able to investigate simultaneously a num. increased sedentary behaviour and (English). We are grateful to Suan Peng Ng for preparing Figure 2 and Isabel Latz for large role for reverse causality is the fact that increasing assistance in preparing the tables. leading to further Nonthaburi. Open University (STOU) who assisted with student contact and the STOU students who are participating in the cohort study. Jane Dixon. Duangkae Vilainerun. This pattern is believed to represent an tion to the obesity patterns seen in the West is a key intermediate stage in the health-risk transition between priority. Carmichael. 11:762 Page 12 of 14 http://www. Wanee Wimonwattanaphan. tual. Emily Banks. this might apply more to exercise- related PA than to screen-time or domestic activities.com/1471-2458/11/762 duties are often divided along gender lines and many Conclusions wealthier households have servants.e. Boonchai Somboonsook. innovative interventions applicable to the Asian not possible to directly attribute causality to the relation. 3School of Population Health. Tanya Mark. Suttinan Pangsap. Tord Kjellstrom. There is likely to be some con. David Harley. inactivity has been shown to result in increased obesity in We also acknowledge the contribution of all members of the Thai Cohort longitudinal data [8. time and increase incidental activity are being devel- ever. the analyses presented here are cross-sectional so it is oped.g. peo. Canberra.

Dwyer T. Jorm L. physical activity? Am J Clin Nutr 2008. Bauman A: Screen-time. Obes Rev 2008. Frejka T. International Journal 1999-2000 Australian Diabetes. Healy GN. obesity in the Asia-Pacific region. 56:473-478. 27(11):1905-1916. Clin Nutr 2003. Stampfer MJ: between occupational and domestic physical activity patterns and body Activity level and risk of overweight in male health professionals. Crawford D: Who does not gain weight? Prevalence and Dalton M. Luben R. Zderic TW: Role of low energy expenditure 34. Open University students. 12. type 2 diabetes. Health transition review: the cultural. 368(9548):1681-1688. AS and ageing and disability: findings from 91. Timperio A. Primatesta P: Domestic physical activity in 36. weight in China. Hosoi Y. J determinants of health in the Third World. Int J Epidemiol 2008. 11. Welch A. CB. Owen N: Joint associations 3. Bain C. Frenk J. 5. Seubsman S. 9:119-126. Clements M. Lawlor D. obesity. 11:S732. metabolic syndrome. Ching PL. Bobadilla JL. Welborn T. social and behavioural 10. behavioural determinants of health 1991. sex. Lee JH. Xu WH. Hamilton DG. relative risk estimates from studies of the same population with 14. 36(2):125-135. Nysse LJ. 1(1):21. mortality and 7. Implications for population studies of obesity in developing countries. 23. Sung J: Leisure-Time Physical Activity is Associated with a Reduced Risk for Metabolic Syndrome. Takahashi Y. WHO Technical Report Series Three major dietary patterns are all independently related to the risk of 894. Clements M. Ball K. Thai Cohort Study Team: Cohort profile: The Population Health Metrics 2009. Crawford D. 14(1):34-43. Lee SA. Banwell C. Banwell C.134 Open University students. Jenkins KR. 1. 38. 19(11):784-792. Banks E. Sugiyama T. Colditz GA. mass index. Lanningham-Foster L. Wen W. Yoon KH. Dixon J. Lim LLY. 11(S667-751):681-684. Boonthum A. Eur J Clin Nutr 2008. 27. Seubsman S. Apr 22:1-10. Tatsanavivat P.liberating the life-force. Steel D. All authors were involved in the interpretation of the analyses and 21. 29. Couper D: Is this finding relevant? Generalisability activity alone may not be a sufficient public health approach to prevent and epidemiology. 169:1191-1200. Bain C. Cho JH. Adair LS. Day NE.a focus on China. Xiang YB. AS and SS contributed to the study concept and design. Hillsdon M. Seubsman SA. Zhai F. participating in varying levels of leisure-time physical activity. 2003. Shaw J: Overweight and obesity in Australia: the predictors of weight maintenance in young women.biomedcentral. and classification of obesity in Japan and Asia-Oceania. 5. 15(12):3113-3121. Clark BK. Cameron A. Salmon J. Geneva: World Health Organization. 26(12):1570-1578. Obesity 2003. Wareham NJ: Television viewing and low participation in vigorous 2. Gortmaker SL. Fultz NH: The relationship of older adults’ activities and body contrasting response rates and designs. Osaka T. Clements M. 13. Up Study. Lim L. Schmidt MD. 20:13-20. obesity among 3760 Japanese women aged 18-20 years. Results from the British Women’s Heart and Health 87-year-old Open University students in Thailand. Stamatakis E. The Lancet 2008. Owen N: The association cardiovascular disease: Is domestic physical activity beneficial? Am J between television viewing and overweight among Australian adults Epidemiol 2009. Tiptaradol S. Numba Y. Bauman A: Screen-time. Rimm EB. Aekplakorn W. Am J Prev Med 2007. Bingham S. and reliability of measures of television viewing time and other non- . Obesity and Lifestyle Study (AusDiab). Hogan MC. Seubsman S. Australian Institute of Health and Welfare: The Active Australia Survey: a EB. Obesity 2007. Banks E. and cardiovascular disease. calories lost: the International Journal of Obesity 2000. The authors declare that they have no competing interests. Dunstan DW. Eur J associations with education and urban or rural residence in Thailand. 63:366-372. social. Choi YH. Cho ER. The Lancet 2006. Huxley R. Sleigh AC: health? Levels of physical activity and factors associated with activity in Socioeconomic status. Rogers K. lifestyle changes and the nutrition transition. and body mass index among university students in Thailand: World Development 1999. Khaw KT. Lim LLY. Bauman A. 289(14):1785-1791. Salmon J. Obesity 2008. Annals of References epidemiology 2009. Jakes RW. ageing and disability: findings from 91. Son HY: of multiple leisure-time sedentary behaviours and physical activity with Epidemic obesity and type 2 diabetes in Asia. critically revising the manuscript for important intellectual content. Rogers K. All authors read and approved the final manuscript. Bain C. height. 20(2):217-234. Ponsonby AL. Bauman A. Lim SS: Trends in obesity and cardiovascular disease risk: EPIC-Norfolk population-based study. obesity. Hu FB. energetic impact of domestic labor-saving devices. and obesity in a large national cohort of 15- elderly women. and beverage consumption during viewing time or reduced leisure-time 372(9650):1697-1705. 8. Journal of Epidemiology and Community Health 2009. Jee SH. 16(6):1440-1447. Brown W. 39. Med J Aust 2003.266 participants in the 45 and SS provided study supervision and were involved in the acquisition of data. Sugiyama T. 11:762 Page 13 of 14 http://www. Up Study. Rogers K: Investigation of relationship to multiple CVD risk factors. Levine JA: Labor saved. 32:320-327. 2010. Pub Health Nutr 2011. 28. Venn AJ: Television viewing and 4. J Aging Health 2008. 86(1):25. Zimmet P. 87(5):1148-1155. analysis and reporting Canberra: LL wrote the initial draft of the manuscript. Hamer M. Kanazawa M. 24(5):600-606. Sleigh A. Stamatakis E. Pub Health Nutr 2010. Li H. Owen N. Kim MK. obesity . Banks E. and the health transition. James WPT: Leisure-time physical 37. Lozano R: Elements for a theory of and sitting in obesity. Chen LC.com/1471-2458/11/762 Authors’ contributions 20. Intern Med 2007. Yiengprugsawan V. Study of the Health-Risk Transition based on Sukhothai Thammathirat 9. Zhai Y. Popkin BM: Longitudinal relationships 6. Sripaiboonkit N. LL. Report of a WHO Consultation. Zhao W. 40. Kim J. Zimmet P. Healy GN. Diabetes 2007. Monda KL. Hamilton MT. Manson JE: Television watching and 11(10):1178-1181. LL performed the statistical AIHW. Jorm L. Journal of Epidemiology Study. of clinical nutrition 2002. 2000. Aust NZ J Public Health 1996. Shin A. Sleigh A: BMI and health- related behaviors in a national cohort of 87. Li TY. Kim JW. recreation are independently associated with obesity and markers of Chariyalertsak S. 178:427-432. Dunstan D. Owen N: Validity journal of obesity 2008. 20:54-56. Shu XO: Prevalence of obesity and correlations with lifestyle and dietary factors Received: 16 December 2010 Accepted: 4 October 2011 in Chinese men. J Epidemiol Community Health 2002. Thai Cohort of 87. Sasaki S. Int J Behav Nutr Phys Act 2008. Caldwell JC: Health transition: the cultural. Colditz GA. BMC Public Health 2011. 16. Lawlor D: Physical activity. 62:1318-1325. Allman-Farinelli M. Yoshiike N. Sleigh A: A large national Thai Cohort diabetes mellitus in women. ASEAN Journal of Open Distance Learning . 30. Published: 4 October 2011 24. 32(3):541-549. 31. Levine JA: Nonexercise activity thermogenesis . 22. 26. obesity in Australian adults. other sedentary behaviors in relation to risk of obesity and type 2 32. World Health Organization: Obesity: preventing and managing the global 19. Dwyer T. Inoue S: Criteria 37:266-272. Jolley D. 8:191-196. Asia Pacific Cohort Studies Consortium: The burden of overweight and 25. 17. American journal of public health 1996. 262:273-287. Yang G. 56(11):2655-2667.134 Thai open university Competing interests students. 10:26. 57(9):1089-1096. Wan X. Jorm L. Chongsuvivatwong V. Koplan JP: Emergence of abdominal obesity in young adults: is the association mediated by food chronic non-communicable diseases in China. Dunstan DW. Sleigh A: Validity of self-reported weight. Itabashi M: epidemic. Ko SH. JAMA 2003.Banks et al. Asia Pacific journal journal of clinical nutrition 2002. BMC Res Methodology 2010. Willett WC. Cleland VJ. Murakami K. Kong L. analysis. 33. 7(1):15. Taylor M. EB and guide and manual for implementation. Stern C. Mealing N. Seubsman S. International 41.266 participants in the 45 and Accepted October 2010. Popkin BM: Urbanization. Kelly M. Okubo H. Yang G. Willett WC. Social Science & Medicine 1993. Asia Pacific 18. Weisell RC: Body mass index as an indicator of obesity. Bedford C. Ebrahim S: Is housework good for your 35. Zimmet P. Salmon J. of Obesity 2002. Salmon J. Obes Rev 2006. 15. Oakes S. Zheng W.

Dunstan D: The inverse relationship between number of steps per day and obesity in a population-based sample . Sharp S. Olson LR. Granger RH. Siegler H. Leslie E. Ware R. BMC Public Health 2011. Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www. Jensen MD.com/submit .com/1471-2458/11/762 occupational sedentary behaviour of adults: a review. Hosmer D. Sorensen T: Prospective associations between sedentary lifestyle and BMI in midlife. 14:1462-1471. Pre-publication history The pre-publication history for this paper can be accessed here: http://www. Zimmet PZ. Obesity 2006. 88(3):612-617. Dwyer T. 307(5709):584-586. Krizan AC. Block G. Ekelund U.biomedcentral. Ministry of Public Health. Mortensen L. Lanningham-Foster LM. Levine JA. Office of Policy and Strategy: Tenth national health development plan 2007-2011 (in Thai). 44. Kane PH. 49. Blizzard CL. 8(1). Nonthaburi: Department of Health.: Relationship of obesity to physical activity. 46. 45. 1(1):4.com/1471-2458/11/762/prepub doi:10.biomedcentral. 31(5):797-804. domestic activities. Besson H. Owen N: Family average income and body mass index above the healthy weight range among urban and rural residents in regional Mainland China. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed. Nonthaburi: Ministry of Public Health. Blair SN. BMC Public Health 2011 11:762. 50. Brage S.Banks et al. National Health Security Office: Guidebook for Community Health Security Fund (in Thai) Nonthaburi: National Health Security Office. 42. Science 2005. Barefoot J. Department of Health (Thailand): Strategies for reducing obesity in Thailand (in Thai). Yin XM. Obes Rev 2009. Shaw JE. 10(1):7-16. Hosmer T. 43. and sedentary behaviours: cross-sectional findings from a national cohort of over 70. 2007. Clark MM: Interindividual variation in posture allocation: Possible rote in human obesity. CAS. 2010. 48. Cochrane JA.biomedcentral. Mandel S: Activities contributing to total energy expenditure in the United States: results from the NHAPS Study. Zhang M. 11:762 Page 14 of 14 http://www. McCrady SK.the AusDiab study. Public health nutrition 2005. 47. Venn AJ. Xu F. International Journal of Obesity 2007. Int J Behav Nutr Phys Act 2004.1186/1471-2458-11-762 Cite this article as: Banks et al. Wareham NJ: Time spent being sedentary and weight gain in healthy adults: reverse or bidirectional causality? Am J Clin Nutr 2008. Gronboek M. Dong L. 2010.000 Thai adults.