Journal Logo


Isotemporal Substitution Analysis for Sedentary Behavior and Body Mass Index


Author Information
Medicine & Science in Sports & Exercise: November 2016 - Volume 48 - Issue 11 - p 2135-2141
doi: 10.1249/MSS.0000000000001002
  • Free


Compelling evidence has shown an inverse relationship between moderate-to-vigorous physical activity (MVPA) and adiposity markers in youth (12,19). It is recommended that children and adolescents engage in at least 60 min of MVPA daily to achieve substantial health benefits, including maintaining a healthy body weight (31,34). However, MVPA typically accounts for less than 5% of a 24-h day. Over 95% of the 24-h day is made up of other components on the movement continuum (sedentary behavior, sleep, and light-intensity physical activity (LPA)), which also have important impacts on health from a young age. In a 24-h day, a decrease in a specific activity means there is an equal time substitution of another activity. Therefore, the influences that specific activity intensities have on health may also depend on the type of activity that it displaces (24).

Isotemporal substitution analysis can model the effects of displacing one specific type of activity with another in an equal amount of time (15). In adults, this statistical technique has been applied to investigate substitution effects of physical activity, sedentary behavior, and sleep on cardiometabolic risk. For children and adolescents, only two studies have applied isotemporal substitution analysis to examine the effects of time reallocation on a selected outcome of interest. It has been shown that substituting 1 h·d−1 of sedentary time with physical activity was favorably associated with percentage body fat (1,20).

Despite recent applications of isotemporal substitution analysis in physical activity research, current understanding of time allocation is limited to the effects of reallocating sedentary time (either objectively assessed or self-reported TV time) with MVPA and/or LPA. Furthermore, no studies conducted in youth have considered sleep. Other than screen time, specific sedentary behaviors have seldom been investigated in relation to obesity. A study conducted with Chinese children found that both studying time and screen time were positively associated with body mass index (BMI) among boys (37); however, the mechanism for this relationship remains unclear. Some studies have suggested that dietary intake is responsible for the observed association between screen time and adiposity (8,13); others have indicated it is partly mediated by the displacement of MVPA (4). Whether sedentary behavior, other than screen time, yields different effects on obesity risk is little known. Such issues may be particularly important with respect to Chinese children because doing academic-related activities occupy large amounts of their discretionary time (16). Therefore, the purpose of this study was to examine the prospective associations of substituting different types of sedentary behavior, LPA, MVPA, and sleep with BMI by using isotemporal substitution analysis. It was hypothesized that statistically displacing screen time with an equal amount of physical activity or other types of sedentary behaviors would lead to a reduced obesity risk.


Study design and participants

The participants were recruited from the Understanding Children’s Activity and Nutrition cohort study in school children in Hong Kong. Recruitment was conducted during June to August 2009 from primary schools in different districts with varying socioeconomic status. Details of recruitment procedure have been reported elsewhere (36). Briefly, after approval was sought from school principals, written consent forms were sent to the parents of children from two randomly selected classes across grades 1–3 from each school. A total of 1666 children from 24 schools consented to participate in the baseline assessment, whereas 1492 (90%) and 1265 (76%) were retained for 1- and 2-yr follow-ups, respectively. Only a subsample of the parents agreed to their child wearing an ActiGraph accelerometer (Pensacola, FL) (details shown in the Physical activity assessment section).

The data collection occurred across the two school semesters (September to November and January to April). At each time point, measurements were taken at a similar time of year to minimize seasonal variations. The children’s anthropometric data were collected during school visits by trained staff. The initialized ActiGraph accelerometers and questionnaires for the parents were distributed to the children during school visits. The parents were instructed to complete the questionnaire at home and to return them to the contact teacher within 1 to 2 wk. Ethical approval was obtained from the Survey and Behavioral Research Ethics Committee of the Chinese University of Hong Kong. Written informed consent was sought from the parents.

Anthropometric measurements

Body weight, in the minimum clothing possible, and standing height without shoes were measured by trained assessors. BMI was calculated by dividing weight (kg) by height squared (m2).

Sedentary behavior and sleep

The parents completed a validated questionnaire to assess their child’s engagement in 13 types of sedentary behavior and sleep during a typical week (17). The questionnaire was originally developed for children age 10 yr or older to self-report their engagement in physical activity and sedentary behavior. Because the participating children in the current study may be too young to self-report, their behavior was reported by their parents. Parents were asked to report the duration (in minutes or hours) of each of the sedentary behaviors on weekdays and weekends, respectively. Duration estimates exceeding 10 h·d−1 for any given sedentary behavior were considered reporting errors and were excluded from the analyses (16). The minutes spent in different types of sedentary behavior per day were then grouped into three categories: leisure-time screen time (watching TV, playing electronic games, and using the Internet for leisure purpose), academic-related activities (e.g., doing homework, attending tutorial class, and using the Internet for academic-related purposes), and other types of sedentary behavior (e.g., reading excluding those for academic-related purposes, sitting and talking, and listening to music). Sleep duration was assessed by a single question: “How much sleep does your child usually get at night?” The parents were asked to report the sleep duration in hours on weekdays (school days) and weekends (non-school days).

Physical activity assessment

Approval for wearing the accelerometer was sought from a subsample of children (n = 1020) at baseline. At each follow-up, consent forms were resent to the parents and approval was obtained from 874 children at a 1-yr follow-up and 706 children at a 2-yr follow-up. Because of the limited ActiGraph accelerometers available at the beginning of the study, the accelerometer data were only collected at baseline for 448 children (44% of those whose parents had agreed). At the 1- and 2-yr follow-ups, the accelerometer data were collected for all the children whose parents had agreed to wear the device. The children who wore the accelerometers were older and had lower parental education levels than those who did not. No differences were found across the other demographic and anthropometric variables.

The accelerometer (ActiGraph) was attached to an elasticized belt, which was worn by children at hip level for eight consecutive days. The accelerometer was removed during sleeping and water-based activities. The data were recorded in 1-min epochs. Non-wear time was defined as at least 60 consecutive minutes of zero recording (32). At each assessment wave, 3 d of accelerometer data (may or may not include a weekend day) with a minimum of 10 h of recording per day were considered to be valid (5). The accelerometer data were downloaded using ActiLife 6 and analyzed using MeterPlus software (Santech Inc., V.4.3, MVPA was defined as ≥4 METs (3) based on age-specific cut-off counts (33). LPA was classified as >100 counts per minute and below the age-specific cut-off points. Average minutes in daily MVPA and LPA were derived for the whole week.

Other covariates

The parents reported their demographic information, including parents’ age, sex, educational attainment, marital status, body weight, and height, and the children’s date of birth and sex. To assess the children’s snacking habit, parents were asked, “How often your child has a habit of snacking while doing the sedentary activity?” Response categories were never, seldom, sometimes, and often. The parents’ educational attainment was classified based on the categories in Hong Kong, and details have been reported elsewhere (16).

Statistical analysis

The data analyses were restricted to children who provided valid accelerometer data for at least one assessment wave and whose parent returned a questionnaire. Three-level linear mixed models were performed to examine associations between MVPA, LPA, different types of sedentary behavior, sleep, and BMI. This method is appropriate for estimating effects over time, and it has been applied in investigating longitudinal changes in MVPA and sedentary time in youth (3,11). Three levels were the three assessment time points (level 1), children (level 2), and school (level 3). Time spent engaging in each behavior was scaled to 30 min·d−1 (divided by 30) to improve the interpretability of the results. Previous studies among children have used behavior of 60 min·d−1 in regression models to interpret the associations between the replacement of 1 h with different activity intensities and health outcomes (1,20); however, in adults, a range between 10 and 60 min·d−1 has been used depending on the outcomes of interest (6,15,23). Given that children may have already engaged in certain amounts of physical activity in a day, reallocating 30 min·d−1 of one activity type to another may be more practical for informing intervention strategies.

Three types of regression models were performed: single behavior models, partition models, and isotemporal substitution models. All the models were adjusted for the child’s age, sex, and snacking habit and the parent’s age, BMI, educational attainment, and marital status. Single models examined the association of each individual behavior with BMI, without the adjustment of other behavior variables. Partition models estimated the effects of each individual behavior on BMI while holding the duration of each of the other behavior variables constant. All the behavior variables were entered into the partition models simultaneously. The coefficients for the partition models represented the effect of statistically adding 30 min·d−1 of one behavior type. Isotemporal substitution models were performed by holding the total behavior time (total behavior time = MVPA + LPA + screen time + academic-related activities + other sedentary activities + sleep) constant. Accelerometer-assessed sedentary time was excluded to avoid duplication. The coefficients for the isotemporal models demonstrated the estimated effects of displacing 30 min of one behavior (the dropped one in the models) with an equal duration in another while holding the total behavior time constant. All the analyses were conducted using SPSS version 22.


Of the 1020 children from whom parental approval was sought for wearing the accelerometer, 359 boys and 313 girls provided valid accelerometer data for at least one assessment time point and returned a parental questionnaire (Table 1). No differences were found in the general characteristics of children and sedentary behavior reported by parents between children who provided valid accelerometer data and those who did not. On average, children were 7.6 yr old at baseline. They spent approximately 2 h·d−1 watching TV and doing academic-related activities, respectively. Sleep duration was approximately 9 h·d−1. The percentage of a snacking habit (reporting “sometimes” and “often”) was 25% while doing homework and 27% during screen time. Low correlations were observed between the different behavioral variables, although significant positive associations were found between LPA and MVPA, and between screen time and the other two types of sedentary behaviors (Table 2).

Baseline characteristics of the participants.
Correlation matrix for behavior variables.

Table 3 shows single, partition, and isotemporal substitution models for the associations between the behavior variables and children’s BMI. In single models, screen time tended to be positively associated with BMI (B = 0.03; 95% CI, −0.01 to 0.07), whereas other sedentary behaviors (B = −0.07; 95% CI, −0.13 to −0.01), LPA (B = −0.06; 95% CI, −0.12 to −0.00), and MVPA (B = −0.42; 95% CI, −0.60 to −0.25) were negatively associated with BMI. In the partition models, statistically adding an additional 30 min·d−1 of each activity/sedentary behavior type while holding the other variables constant only changed the associations between LPA and BMI (not significant). In the isotemporal substitution models, replacing 30 min·d−1 of MVPA with each of the other behavior variables resulted in an increase in children’s BMI. Replacing 30 min·d−1 of other sedentary behaviors (e.g., sitting and talking) with screen time (B = 0.12; 95% CI, 0.04 to 0.20) or academic-related activities (B = 0.13; 95% CI, 0.04 to 0.21) resulted in higher BMI.

Single, partition, and isotemporal substitution models of associations between 30 min·d−1 change in MVPA, LPA, different types of sedentary behavior, sleep, and BMI.a


This study applied an isotemporal substitution model approach to examine time allocation effects on BMI in Chinese children. Although this novel technique has previously been used in physical activity research in both adults (15,35) and children (1,20), this study extends previous findings by examining substitution effects of different types of sedentary behavior in children. The results suggest that the substitution of 30 min·d−1 of other types of sedentary behavior (e.g., sitting and talking) with an equivalent amount of screen time or academic-related activities was associated with an increased BMI according to statistical modeling. The findings are also in line with previous studies that show that MVPA is favorably associated with children’s BMI because it was found that displacing any of the other behavior variables, including LPA, with MVPA resulted in a lower modeled BMI. As the isotemporal substitution models controlled for the total behavior time, the observed associations between individual behavior types and BMI are expected to be independent of each other.

This study found that the three types of sedentary behavior had different associations with BMI, which seemed to be independent of MVPA. That is, the overall impact of 30 min of screen time on BMI in Chinese children may depend on what kinds of sedentary behavior it displaces. Current evidence linking sedentary behavior with adiposity is equivocal. Studies primarily focusing on TV viewing or screen time have consistently found associations between these activities and higher risks of obesity (10,30), whereas evidence of association between objectively assessed overall sedentary time (not examined in the current study because of the different outcome of interest) and cardiometabolic risk factors is inconsistent (9,12,14). Our finding that screen time was no longer associated with BMI in single and partition models after adjusting for snacking habit is in line with the notion that the association of screen time with obesity could be partially explained by dietary intake (8,13). It has been also supported by preliminary evidence from experimental studies that screen-based behaviors can increase acute energy intake in children and youth (22).

For sedentary behavior other than screen-based activities, little is known on their associations with adiposity. It is surprising to find that in the current study, replacement of academic-related activities with screen time did not yield any differences in BMI. The only study reporting a positive relationship between studying time and BMI was conducted on 6- to 11-yr-old Chinese boys, but snacking habits were not examined (37). High-energy snacks are commonly accessible to children in Hong Kong (18). In the current study, one fourth of the Chinese children had a habit of snacking while doing academic-related activities, which is a similar percentage to those who reported eating snacks during screen time. The opportunity for using the Internet to do homework is increasing for school-age children nowadays. It is possible that exposure to the screen per se may increase the likelihood of food intake; no matter if it is for homework or entertainment purpose. Nevertheless, it is important to note that although frequency of snacking habit was controlled, whether other dietary variables, such as energy intake of the snacks, were different or not remains unknown. The other unexpected finding is the negative association between other sedentary behaviors (e.g., sitting and talking, and listening to music) and BMI. Consistent with previous work conducted with Chinese children (16,37), other sedentary pursuits were less prevalent compared with screen time and academic-related activities. Furthermore, we found that screen time and academic-related activities were negatively associated with time in MVPA, whereas other sedentary behaviors were not. It is possible that these activities are less likely to be combined with snacking or they may be indicative of a healthier lifestyle in general. Some laboratory studies have shown that teenagers consumed less energy from drinks and/or solid foods while listening to music or sitting in groups, compared with when the TV was on (25,27). However, longer term effects of different types of sedentary behavior on adiposity need to be investigated by well-designed experimental or cohort studies.

No beneficial effects of reallocating time from sedentary behavior to LPA or sleep on BMI for Chinese children were generally consistent with the previous studies using an isotemporal substitution approach for adults (15) and children (1,20). The observation that sleep duration was not associated with BMI contrasts with previous reviews that suggest that sleep duration is inversely related to obesity (7,26). However, Chinese children observed in this study slept on average 9 h a night, which may be considered to be adequate. Given that, in this study, the strongest associations were found between MVPA and BMI in all of the three types of model, it seems that MVPA remains the most potent behavior along the activity spectrum in relation to children’s BMI. This suggests that the maximal benefits on BMI will be achieved if MVPA can be reallocated from screen time or academic-related activities. Practically, however, the actual amount of time in MVPA that can be reallocated from other physical activities or sedentary behavior may be limited. It will be particularly difficult to limit time spent in academic-related activities for Chinese children because of the academic pressure in Chinese society (16). However, promoting healthy snacking habits or limiting energy intake while engaging in sedentary behavior warrants attention in future interventions. Taken together, these findings support current advocacy in promoting health-enhancing physical activity and reducing screen-based behaviors to maintain a healthy body weight (21,29).

The strengths of this study lie in its longitudinal design and its focus on substitution effects of different types of sedentary behavior. Thus, the associations observed between sedentary behavior and BMI reflect a prospective relationship and are independent of MVPA. Future studies may consider using an objective device that detects postural transition, e.g., activPAL, to distinguish between sitting and LPA in children (2,28). A potential bias of using subjective measures of sedentary behavior is acknowledged. Although sedentary time could be calculated based on the ActiGraph data, the outcome of interest in this study was on the different types of sedentary behavior. As a result, the sum of the three types of reported sedentary behavior instead of accelerometer-assessed sedentary time was computed in the isotemporal models. The questionnaire was developed and validated for Chinese children in Hong Kong (17), and high correlations were found between parent-reported total sedentary behavior and objectively measured sedentary time (r ranged from 0.64 to 0.73 at the three assessment waves, data not shown). In order to maximize the analytical sample, having valid accelerometer data on at least a weekend day was not considered as a requirement. However, 87% of the children who were included in the final data analyses did provide valid accelerometer data on more than 3 d, including a weekend day. In addition, mixed-effects models excluding those children who provided valid data on weekdays only did not yield different findings.

In summary, this study provides preliminary evidence of the different impacts of various types of sedentary behavior on children’s BMI. Reallocating screen time or academic-related activities with other types of sedentary behavior or MVPA was associated with lower BMI. It is possible that the unfavorable effects of sedentary behavior on adiposity may be mediated not only by replacing MVPA but also by increasing other unhealthy behaviors, such as snacking. Whereas screen time reduction continues to be a potent component of health-enhancing interventions, future studies may consider providing instructions on alternative activities and targeting other unhealthy behaviors while engaging in screen-based behavior and academic-related activities in Chinese children.

The authors are grateful to all the children and their parents who participated in this study, along with the school teachers and student helpers who assisted in the Understanding Children’s Activity and Nutrition project.

The Understanding Children’s Activity and Nutrition study was supported by the General Research Fund (GRF) from the Research Grants Council of the Government of the Hong Kong Special Administrative Region, China (GRF 451308). J. Salmon was supported by a Principal Research Fellowship from the Australian National Health and Medical Research Council (APP1026216).

The authors declare that they have no conflict of interest.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Aggio D, Smith L, Hamer M. Effects of reallocating time in different activity intensities on health and fitness: a cross sectional study. Int J Behav Nutr Phys Act. 2015;12:83.
2. Aminian S, Hinckson EA. Examining the validity of the activPAL monitor in measuring posture and ambulatory movement in children. Int J Behav Nutr Phys Act. 2012;9:119.
3. Arundell L, Ridgers ND, Veitch J, Salmon J, Hinkley T, Timperio A. 5-Year changes in afterschool physical activity and sedentary behavior. Am J Prev Med. 2013;44(6):605–11.
4. Berentzen NE, Smit HA, van Rossem L, et al. Screen time, adiposity and cardiometabolic markers: mediation by physical activity, not snacking, among 11-year-old children. Int J Obes (Lond). 2014;38(10):1317–23.
5. Brooke HL, Atkin AJ, Corder K, Ekelund U, van Sluijs EM. Changes in time-segment specific physical activity between ages 10 and 14 years: a longitudinal observational study. J Sci Med Sport. 2016;19(1):29–34.
6. Buman MP, Winkler EA, Kurka JM, et al. Reallocating time to sleep, sedentary behaviors, or active behaviors: associations with cardiovascular disease risk biomarkers, NHANES 2005–2006. Am J Epidemiol. 2014;179(3):323–34.
7. Cappuccio FP, Taggart FM, Kandala NB, et al. Meta-analysis of short sleep duration and obesity in children and adults. Sleep. 2008;31(5):619–26.
8. Chaput JP, Klingenberg L, Astrup A, Sjödin AM. Modern sedentary activities promote overconsumption of food in our current obesogenic environment. Obes Rev. 2011;12(5):e12–20.
9. Colley RC, Garriguet D, Janssen I, et al. The association between accelerometer-measured patterns of sedentary time and health risk in children and youth: results from the Canadian health measures survey. BMC Public Health. 2013;13:200.
10. Coombs NA, Stamatakis E. Associations between objectively assessed and questionnaire-based sedentary behaviour with BMI-defined obesity among general population children and adolescents living in England. BMJ Open. 2015;5(6 e007172).
11. Corder K, Sharp SJ, Atkin AJ, et al. Change in objectively measured physical activity during the transition to adolescence. Br J Sports Med. 2015;49(11):730–6.
12. Ekelund U, Luan J, Sherar LB, Esliger DW, Griew P, Cooper A. Moderate to vigorous physical activity and sedentary time and cardiometabolic risk factors in children and adolescents. JAMA. 2012;307(7):704–12.
13. Falbe J, Rosner B, Willett WC, Sonneville KR, Hu FB, Field AE. Adiposity and different types of screen time. Pediatrics. 2013;132(6):e1497–505.
14. Fröberg A, Raustorp A. Objectively measured sedentary behaviour and cardio-metabolic risk in youth: a review of evidence. Eur J Pediatr. 2014;173(7):845–60.
15. Hamer M, Stamatakis E, Steptoe A. Effects of substituting sedentary time with physical activity on metabolic risk. Med Sci Sports Exerc. 2014;46(10):1946–50.
16. Huang WY, Wong SH. Time use clusters in children and their associations with sociodemographic factors. J Public Health. 2015; doi:10.1093/pubmed/fdv088.
17. Huang YJ, Wong SH, Salmon J. Reliability and validity of the modified Chinese version of the Children’s Leisure Activities Study Survey (CLASS) questionnaire in assessing physical activity among Hong Kong children. Pediatr Exerc Sci. 2009;21(3):339–53.
18. Hui LL, Nelson E. Dietary characteristics of Hong Kong Young children: implications for nutrition education. Hong Kong J Paediatr. 2006;11:255–62.
19. Jiménez-Pavón D, Kelly J, Reilly JJ. Associations between objectively measured habitual physical activity and adiposity in children and adolescents: systematic review. Int J Pediatr Obes. 2010;5(1):3–18.
20. Loprinzi PD, Cardinal BJ, Lee H, Tudor-Locke C. Markers of adiposity among children and adolescents: implications of the isotemporal substitution paradigm with sedentary behavior and physical activity patterns. J Diabetes Metab Disord. 2015;14:46.
21. Lynch BM, Owen N. Too much sitting and chronic disease risk: steps to move the science forward. Ann Intern Med. 2015;162(2):146–7.
22. Marsh S, Ni Mhurchu C, Maddison R. The non-advertising effects of screen-based sedentary activities on acute eating behaviours in children, adolescents, and young adults. A systematic review. Appetite. 2013;71:259–73.
23. Mekary RA, Lucas M, Pan A, et al. Isotemporal substitution analysis for physical activity, television watching, and risk of depression. Am J Epidemiol. 2013;178(3):474–83.
24. Mekary RA, Willett WC, Hu FB, Ding EL. Isotemporal substitution paradigm for physical activity epidemiology and weight change. Am J Epidemiol. 2009;170(4):519–27.
25. Mekhmoukh A, Chapelot D, Bellisle F. Influence of environmental factors on meal intake in overweight and normal-weight male adolescents. A laboratory study. Appetite. 2012;59(1):90–5.
26. Nielsen LS, Danielsen KV, Sørensen TI. Short sleep duration as a possible cause of obesity: critical analysis of the epidemiological evidence. Obes Rev. 2011;12(2):78–92.
27. Péneau S, Mekhmoukh A, Chapelot D, et al. Influence of environmental factors on food intake and choice of beverage during meals in teenagers: a laboratory study. Br J Nutr. 2009;102(12):1854–9.
28. Ridgers ND, Salmon J, Ridley K, O’Connell E, Arundell L, Timperio A. Agreement between activPAL and ActiGraph for assessing children’s sedentary time. Int J Behav Nutr Phys Act. 2012;9:15.
29. Tremblay MS. Major initiatives related to childhood obesity and physical inactivity in Canada: the year in review. Can J Public Health. 2012;103(3):164–9.
30. Tremblay MS, LeBlanc AG, Kho ME, et al. Systematic review of sedentary behaviour and health indicators in school-aged children and youth. Int J Behav Nutr Phys Act. 2011;8:98.
31. Tremblay MS, Warburton DE, Janssen I, et al. New Canadian physical activity guidelines. Appl Physiol Nutr Metab. 2011;36(1):36–46; 7–58.
32. Troiano RP, Berrigan D, Dodd KW, Mâsse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc. 2008;40(1):181–8.
33. Trost SG, Pate RR, Sallis JF, et al. Age and gender differences in objectively measured physical activity in youth. Med Sci Sports Exerc. 2002;34(2):350–5.
34. U.S. Department of Health and Human Services. 2008 Physical Activity Guidelines for Americans. Washington (DC): Department of Health and Human Services; 2008. pp. 15–8.
35. Wellburn S, Ryan CG, Azevedo LB, et al. Displacing sedentary time: association with cardiovascular disease prevalence. Med Sci Sports Exerc. 2016;48(4):641–7.
36. Wong SH, Huang WY, He G. Longitudinal changes in objectively measured physical activity differ for weekdays and weekends among Chinese children in Hong Kong. BMC Public Health. 2015;15:1310.
37. Zhang J, Seo DC, Kolbe L, Middlestadt S, Zhao W. Associated trends in sedentary behavior and BMI among Chinese school children and adolescents in seven diverse Chinese provinces. Int J Behav Med. 2012;19(3):342–50.


© 2016 American College of Sports Medicine