There is a growing body of evidence on the detrimental associations of sedentary behavior (prolonged sitting time) with cardiometabolic risk biomarkers and health outcomes (28). Much of this research has focused on television (TV) viewing time: a recent review of measures of nonoccupational sedentary behavior found that all relevant articles (n = 60) had assessed TV viewing time by self-report, and in 39 articles, it was the sole sedentary behavior measured (9). TV viewing is a highly prevalent leisure time behavior (4,12,36) and has been shown to have strong and consistent detrimental associations with cardiometabolic risk biomarkers and health outcomes, including premature mortality (13-15,19-22). It is also a specific behavior that may be recalled relatively accurately (9,26), thus has an advantage over using more imprecise overall measures (11). However, TV viewing time is one of several sedentary behaviors in which adults engage and thus may or may not be representative of overall sedentary time. In this context, Pate et al. (29) highlight inconsistencies in studies on sedentary behavior that report findings solely on TV viewing time, yet discuss these findings in terms of overall sedentary behavior. Evidence is needed to clarify whether TV viewing time can be representative of overall sedentary time.
The prevalence of high TV viewing time has been seen to differ by gender, age, race/ethnicity, and work status, with higher levels of TV viewing time observed in men, older adults, those of African American race/ethnicity, and among those not in paid employment (3,10,35). Interestingly, those population subgroups who report watching the most TV (3,10,35) are not always those who are identified objectively as being the most sedentary overall (25). Thus, the extent to which measures of TV viewing time could be indicative of overall sedentary time may differ between population subgroups. Such variations may provide insights into the differential associations of TV viewing time with health outcomes that have previously been observed across gender and ethnicity groups (2,15,16,18,19,22,35). To date, no studies have examined this issue, although TV viewing time has been shown to be a marker for self-reported total leisure time sedentary behavior in women (36).
The concurrent assessment of self-reported TV viewing time and accelerometer-measured physical activity in the 2003-2006 population-representative US National Health and Nutrition Examination Surveys (NHANES) provides a unique opportunity to address the relationship of TV viewing time with accelerometer-derived sedentary time in a large, diverse, population-based sample. We examined this relationship in the overall adult population, as well as stratified by gender, age, race/ethnicity (non-Hispanic white, non-Hispanic black, and Mexican American), and work status. Relationships were also examined in terms of relative agreement to explore the extent to which TV viewing time may or may not be representative of total sedentary time across the day.
Study Population and Design
The study sample was drawn from the 2003-2004 and 2005-2006 cycles of NHANES: a continuous, cross-sectional, observational study conducted by the Centers for Disease Control and Prevention (7,25,37). NHANES used a stratified, complex multistage probability design to obtain a nationally representative sample of noninstitutionalized civilian US citizens aged ≥6 yr (7,25,37). Certain populations were oversampled, including Mexican and non-Hispanic black Americans. The National Center for Health Studies Ethics Review Board approved the survey protocols, and informed consent was obtained from all participants. Data collection involved an interviewer-administered questionnaire conducted in participants' homes by trained interviewers, and health measurements were carried out in specially designed and equipped mobile examination centers. In the 2003-2004 and 2005-2006 cycles, ambulatory participants were asked to wear an accelerometer (ActiGraph model 7164; ActiGraph, LLC, Ft. Walton Beach, FL) on an elasticized belt over the right hip during waking hours for the seven consecutive days after their examination in one of the mobile examination centers. Details of the accelerometer protocol have previously been reported (5,25,37).
For the purposes of the following analyses results were included from participants who were adults (age ≥20 yr; n = 10,020), wore the accelerometer for ≥10 h·d−1 on at least 4 d (and including at least one weekend day; n = 5742), and completed the TV viewing item of the questionnaire (n = 5738). A smaller sample of participants aged 20-65 yr (nominal working age) was taken from the 2003-2004 survey to examine possible variations by work status (n = 2069). At the time of the current analyses, data on work status were not available for the 2005-2006 survey.
Gender, age, race/ethnicity, education, and work status were self-reported in the interviewer-administered questionnaire. Age categories were established based on age in whole year at the time of interview as follows: 20-39, 40-59, and ≥60 yr. Racial/ethnic groups were categorized into non-Hispanic white, non-Hispanic black, Mexican American, and other race (not included in the previous categories) according to NHANES analytic guidelines (6). Work status categories (nonworking, part-time, and full-time work) were derived from participants' self-reported work status in the previous week and the number of hours worked in that week (or hours usually worked for workers who reported <35 h in the previous week). The minimum requirement for full-time work was 35 h worked in the previous week or usually worked; <35 h in the previous week was classified as part-time in those that reported working. Education was dichotomized as <12 yr of schooling or ≥12 yr, or equivalent, including post-high school training. Waist circumference was measured by trained staff during the physical examination, with measurements taken at the upper border of the right iliac crest (7).
TV viewing time.
Amount of time spent watching TV or videos was self-reported in the household interview questionnaire as total time on a typical day during the past 30 d, with the following response options: <1, 1, 2, 3, 4, ≥5 h, or do not watch TV or videos. The category of "do not watch TV or videos" was combined with "<1 h" because of low numbers reporting these categories.
Accelerometer-derived overall sedentary time.
Accelerometer-derived sedentary time was calculated as time spent in <100 counts per minute (cpm). This cut point has previously been shown to approximate sitting time (25); however, as accelerometers are not able to discriminate between sitting and standing very still, time recorded as sedentary will not strictly represent sitting time. An automated program (SAS 9.1) (25) was adapted and used to derive wear time and summarize sedentary time data. Non-wear time was defined as bouts of ≥60 min of consecutive zero counts with interruptions of up to two counts of ≤50 cpm (41). Sedentary time (h·d−1) was corrected for monitor wear time using the residuals method (39). Non-wear time includes time spent asleep, for showering or bathing and for water-based activities.
Analyses were conducted using Stata Statistical Software Release 11.0 (Stata Corporation; College Station, TX). To account for the complex survey design used in NHANES (6), the STATA survey commands with sampling weight, stratification, and clustering (primary sampling unit) variables were used for most analyses. Four-year examination weights (2003-2006), reweighted to correct for nonresponse to the accelerometry component of NHANES were used. Analyses that included occupational data, which were only available for the earlier NHANES cycle, used 2-yr (2003-2004) examination weights with reweighting to correct for nonresponse to the accelerometry component. Weights were not used when testing interactions (1) or in calculating the Spearman rank-order correlations. Strata were collapsed when required because of low numbers in some subgroups. Significance was set at <0.05 for main effects and <0.1 for interactions. Characteristics of the sample (weighted) were described as percent (n) or mean (SD).
Simple linear regression analyses (with linearized variance estimation and weighting) were used to examine the bivariate association between TV viewing time and sedentary time, with data reported as population-weighted mean accelerometer-derived sedentary time across categories (<1, 1, 2, 3, 4, and ≥5 h) of self-reported TV viewing time. To examine whether these associations differed across population subgroups of interest (gender, age, race/ethnicity, and work status), stratified analyses were performed and interactions were tested. Interactions were examined unadjusted, then adjusted for age (in completed years), gender, waist circumference, race/ethnicity, and educational level to ensure that any difference in the degree of relationship was not due to imbalance of these other characteristics. The "Other Race" category was excluded from the racial/ethnic comparisons due to the diverse ethnic backgrounds within the category (n = 379). In adjusted models, participants with missing data for covariates were dropped (missing data for covariates: total sample = 166, non-Hispanic white = 92, non-Hispanic black = 44, Mexican Americans = 26, full-time workers = 29, nonworking = 17).
Spearman rank-order correlations were used to examine the relative agreement between the participants' rankings of TV viewing time and accelerometer-derived sedentary time overall and separately for each subgroup of interest. Magnitude of relative agreement was described using the scale reported by Landis and Koch (23) as follows: poor = below 0, slight = 0.00-0.20, fair = 0.21-0.40, moderate = 0.41-0.60, substantial = 0.61-0.80, almost perfect = 0.81-1.00. This scale was developed for the κ statistic; however, the range of values and concept of relative agreement examined in this article using Spearman rank-order correlation is similar. Assessment of agreement between quintiles of TV viewing time (only approximate due to the categorical nature of the data) and sedentary time using the weighted κ statistic revealed similar results to the Spearman correlation and are therefore not reported in this article.
Attributes of the 5738 participants included in the study are presented in Table 1. More than half the sample reported their race/ethnicity status as non-Hispanic white. Those reporting the highest category of TV viewing time (≥5 h·d−1) were more likely to be in the older age than the younger age category (P < 0.001) of non-Hispanic black than non-Hispanic white race/ethnicity (P < 0.001) and working full-time than not working (among the working-aged subgroup, P < 0.001). Mexican Americans were less likely to report this high level of TV viewing time than non-Hispanic whites (P = 0.02).
Linear Regression Findings: Associations of TV Viewing Time with Total Sedentary Time
Linear regression models revealed a positive association between self-reported TV viewing time and total accelerometer-derived sedentary time (B = 0.22, 95% confidence interval (CI) = 0.19-0.26, P < 0.001). However, the R2 (an indication of the proportion of variance in sedentary time that is explained by the model) was low (R2 = 0.039).
Associations by gender.
TV viewing time was associated with total sedentary time in both women and men (Fig. 1). There was no significant interaction between gender and TV viewing time (F(df = 5, 26) = 1.52, P = 0.22 without adjustment, F(df = 5, 26) = 2.03, P = 0.11 with adjustment), and the linear trend was similar for women and men (Fig. 1; see also Table, Supplementary Digital Content 1, for population-weighted percentages of participants in self-reported TV viewing time categories; https://links.lww.com/MSS/A61).
Associations by age
There was an association between TV viewing time and sedentary time in all age groups (Fig. 1). The linear trend was least evident in the younger age (20-39 yr) adults, among whom sedentary time was only significantly different in those reporting the highest category (≥5 h·d−1) versus the lowest category (<1 h·d−1) of TV viewing time (Fig. 1). The age interaction was significant, with the association between TV viewing time and sedentary time stronger at midage (40-59 yr) and older age (>60 yr) than at a younger age (20-39 yr) (F(df = 10, 21) = 3.52, P = 0.01 without adjustment, F(df = 10,21) = 2.65, P = 0.03 with adjustment).
Associations by race/ethnicity.
Stratified linear regression analyses revealed a positive association between TV viewing time and total sedentary time within all racial/ethnic groups (Fig. 1). Interaction terms showed no significant differences in this association by race/ethnicity (F(df = 10, 21) = 1.27, P = 0.31 without adjustment, F(df = 10, 21) = 0.75, P = 0.67 with adjustment). However, for those of Non-Hispanic black and Mexican American race/ethnicity, the association between TV viewing time categories and mean sedentary time was only significant for those reporting ≥5 h of TV viewing per day compared with <1 h (Fig. 1). In contrast, the relationship between the two variables was more linear for non-Hispanic whites.
Associations by work status.
In working-aged adults (20-65 yr, NHANES 2003-2004), the association between TV viewing time and total sedentary time was present in the nonworking subgroup and marginally significant for those in part-time work but not for those in full-time work (Fig. 1). For full- and part-time workers, an association between TV viewing time categories and mean sedentary time was only significantly different in those reporting the highest category (≥5 h·d−1) versus the lowest category (<1 h·d−1) of TV viewing time, possibly more indicative of a threshold shift at this level than an overall relationship. Despite the apparent differences, the association did not differ significantly by work status (F(df = 5, 11) = 2.19, P = 0.13 without adjustment, F(df = 5, 11) = 1.14, P = 0.40 with adjustment).
Relative Agreement between TV and Total Sedentary Time
Table 2 presents correlations between participants' self-reported TV viewing time and total accelerometer-derived sedentary time (Spearman ρ). The measures were correlated more than would be expected by chance (P < 0.001) for the overall population, although the strength of the correlation was only fair.
Except for full-time workers and young adults (20-39 yr), the relative agreement between the TV viewing and sedentary time was either fair or at the upper end of slight classification for all population subgroups. As expected from the linear regressions, relative agreement was lower among the younger age category (20-39 yr) than other ages, lower (and not significant) for full-time and part-time workers than nonworking adults of working age, and slightly higher among non-Hispanic whites than the other two racial/ethnic groups. Across the board, relative agreement was much below the "substantial" or "almost perfect" relationships that would be expected if TV viewing time was considered to be representative of total sedentary time.
Our findings show a relationship of TV viewing time with total accelerometer-derived sedentary time in the overall study population, evidenced by a significant association and rank-order correlation between the two measures. However, the strength of the agreement was only fair, and adults that reported <1 h·d−1 of TV viewing still spent nearly 8 h·d−1 in sedentary time. Thus, self-report TV viewing time may not be truly representative of accelerometer-derived overall sedentary time. To capture a more complete picture of daily sedentary time, measurement of sitting time across other domains (work and travel), as well as other leisure time behaviors (e.g., reading, computer use) may also be important.
The correlation we observed between our measures was similar or slightly lower than findings from criterion validity studies of total sitting time questionnaires with accelerometer data as the criterion (11,30), indicating that error associated with self-report may be at issue here. However, despite concerns regarding comparing a self-report to an objective instrument, the effect of TV viewing time on sedentary time cannot be ignored, given that it is a highly prevalent leisure time sedentary behavior (4,36). A recent time use study showed TV viewing was the most common leisure time activity; more than 80% of US adults reported watching TV on the surveyed day and for those who reported watching TV, viewing time averaged close to 3.5 h (4). We found an association between TV viewing time and sedentary time for all groups, except those in full-time work. Therefore, those reporting higher levels of TV viewing time had higher sedentary time on average than those who reported lower levels of TV viewing time. TV viewing time may, to some extent, reflect a broader pattern of sedentary behavior in the population as a whole, which is important to consider in light of the findings regarding its detrimental associations with risk biomarkers and health outcomes (13-15,19-22,40).
No gender or race/ethnicity differences were observed in the relationship of TV viewing time with accelerometer-derived sedentary time, suggesting that TV viewing time is a useful measure across these groups. However, there was a nonlinear relationship between the two measures for Mexican American and non-Hispanic blacks with a stronger association at high (>5 h·d−1) levels of TV viewing time. Where possible, continuous measures of TV viewing time should be used to enable selection of cut points suitable for ethnically diverse populations.
TV viewing time was more strongly related to total sedentary time for those in the mid- (40-59 yr) and older-age groups (>60 yr) compared with the younger age group (20-39 yr), indicating that TV viewing may better reflect total sedentary time in older than younger adults. Other media-based behaviors, such as social networking on computers and mobile phones, may be contributing to sedentary time in younger adults. A recent study from Canada showed that the proportion of screen time reported as computer use (as distinct from TV viewing time) was significantly higher in younger compared with older adults (34), therefore measuring these other screen-based behaviors may be important in research involving younger adults. The age difference may also in part be due to the employment status of the participants, particularly in the oldest age group, many of whom could have retired. In the working-age population, the relationship of TV viewing time with sedentary time was present in the nonworking subgroup but not those in full-time work. Previous research has shown those in full-time work are less likely to watch high levels of TV (10). Full-time workers may have less leisure time available for TV viewing, compared with those who do not work, with other sedentary behaviors comprising more of their total sedentary time. Measuring sedentary time in the working population requires development of measures of sedentary behavior in the workplace. The development of such self-report measures has been recommended in a recent review of articles reporting the association between workplace sitting and detrimental health outcomes (38).
The key strength of this study is the use of data from a large, multiethnic, population-representative survey, with concurrent objective and self-report measures. Accelerometer-derived sedentary time, being an objective measure, has the advantage of being unaffected by recall error or self-report biases. There are some limitations, however, in the use of accelerometers as a criterion measure for true sedentary time. First, as accelerometers do not detect body position, the measure is indirect. Therefore, periods of low movement (<100 cpm) may include some time spent standing still, resulting in overestimation of sedentary time. In a small criterion validity study, the correlation between accelerometer-derived sedentary time using the <100 cpm cut point and a more direct measure of body posture (sitting, reclining and lying) (42) was only r = 0.59 (25). Second, the sedentary time estimate is affected by monitor wear time, which seems to have been less than the intended coverage of "all waking hours" (average wear time = 14.6 h, range = 10.3-24.0 h). Times that are not captured are likely to be early mornings and late evenings, which are possibly more sedentary periods of the day, thereby resulting in underestimation of sedentary time.
Another limitation of the study is the measure of self-report TV viewing time available in NHANES. Validity of the measure has not been reported, and it has low test-retest reliability (ICC = 0.32, 95% CI = 0.14-0.48) (17) relative to other self-report measures of TV viewing time whose reliability (ICC) typically ranges from 0.7 to 0.9 (27,33,43). We cannot be certain as to whether a different level of relative agreement overall or different patterning across population subgroups might have been seen with other TV viewing time measures. These variations in self-report measures will remain a limitation in the assessment of TV viewing time, until objective measures of this behavior suitable for use in epidemiological and health behavior studies have been developed. Furthermore, the relationship between the two measures may have been underestimated as the measurement period for the TV viewing time question differed from the period of accelerometer wear from which sedentary time was derived. Respondents were asked to recall their average hours of daily TV viewing time during the past 30 d, whereas accelerometer data collection covered 7 d, which has been recommended to obtain a representative measure of habitual physical inactivity (24). As our minimum number of valid days for accelerometer wear was only four, we may not have captured a typical pattern of sedentary time, however, close to half of our sample (48.2%) provided seven valid days of monitor wear. The comparison of a categorical measure to a continuous measure also presented difficulties including limiting the analyses that could be performed, and the highest option for TV viewing time (>5 h·d−1) was lower than the mean sedentary time.
A better understanding of commonly used indices of sedentary behaviors is a fundamental element for the interpretation of the findings of epidemiological and health behavior studies and for guiding the development of evidence-based public health interventions (31). Objective measurement of sedentary behavior, such as by accelerometer, is expensive and often not feasible in large-scale studies. Furthermore, objective measures cannot provide domain-specific data. Therefore, self-report measures of sedentary behavior, typically TV viewing time, have been widely used (8,15,20,22,32). Our findings showed that there was an association between self-report TV viewing time and accelerometer-derived sedentary time, indicating that this behavior has an important influence on sedentary time at a population level and therefore may be useful in epidemiological surveys. However, the relative agreement between the two measures was only fair. Therefore, self-reported TV viewing time did not seem to provide a good reflection of total sedentary time captured using accelerometers at the individual level. Importantly, the relationship between TV viewing time and total sedentary time was consistent for subgroups of gender and race/ethnicity but not age and work status. Future research should take into account these differences in the performance of self-report TV viewing time measures and consider measurement of other sedentary behaviors during leisure time and the domains of travel and work.
This study was supported by funding from Queensland Health and the National Health and Medical Research Council of Australia (#301200).
All data used in this study were collected by the National Center for Health Statistics, Centers for Disease Control and Prevention.
Clark is supported by an Australian Post Graduate Award Scholarship and Queensland Health funding. Healy is supported by a NHMRC (#569861) / National Heart Foundation of Australia (PH 08B 3905) Postdoctoral Fellowship. Dunstan is supported by a Victorian Health Promotion Foundation Public Health Research Fellowship. Winkler, Sugiyama, and Owen are supported by a Queensland Health Core Research Infrastructure grant and by NHMRC Program Grant funding (#301200). Gardiner is supported by a National Heart Foundation of Australia (#PP 06B 2889) Postgraduate Scholarship. The authors have no conflict of interest to declare.
The results of this study do not constitute an endorsement by the American College of Sports Medicine.
1. Aitken M, Francis B, Hinde J, Darnell R. Statistical Modelling in R
. Oxford (NY): Oxford Statistical Science; 2009. p. 113.
2. Bertrais S, Beyeme-Ondoua J-P, Czernichow S, Galan P, Hercberg S, Oppert J-M. Sedentary behaviors, physical activity, and metabolic syndrome in middle-aged French subjects. Obes Res
3. Bowman SA. Television-viewing characteristics of adults: correlations to eating practices and overweight and health status. Prev Chronic Dis
4. Bureau Labor Statistics. American Time Use Survey 2009 Results
. Washington (DC): Department of Labor US Government; 2009. p. 2.
5. CDC/National Center for Health Statistics. Documentation, codebook and frequencies: physical activity monitor NHANES 2005-2006 [Internet]. Hyattsville (MD): US Department of Health and Human Services, Centers for Disease Control and Prevention; [cited 2009 May 30] Available from: http://www.cdc.gov/nchs/data/nhanes/nhanes_05_06/paxraw_d.pdf
6. CDC/National Center for Health Statistics. National Health and Nutrition Examination Survey: analytic guidelines [Internet]. Hyattsville (MD): US Department of Health and Human Services, Centers for Disease Control and Prevention; [cited 2009 June 18]. Available from: http://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/nhanes_analytic_guidelines_dec_2005.pdf
7. CDC/National Center for Health Statistics. NHANES 2003-2004 and 2005-2006 data files: data, docs, codebooks, SAS code [Internet]. Hyattsville (MD): US Department of Health and Human Services, Centers for Disease Control and Prevention; [cited 2009 June 18]. Available from: http://www.cdc.gov/nchs/nhanes.htm
8. Ching PL, Willett WC, Rimm EB, Colditz GA, Gortmaker SL, Stampfer MJ. Activity level and risk of overweight in male health professionals. Am J Public Health
9. Clark BK, Sugiyama T, Healy GN, Salmon J, Dunstan DW, Owen N. Validity and reliability of measures of television viewing time and other non-occupational sedentary behavior of adults: a review. Obes Rev
10. Clark BK, Sugiyama T, Healy GN, et al. Socio-demographic correlates of prolonged television viewing time in Australian men and women: the AusDiab study. J Phys Act Health
11. Craig CL, Marshall AL, Sjostrom M, et al. International physical activity questionnaire: 12 country reliability and validity. Med Sci Sports Exerc
12. Dong L, Block G, Mandel S. Activities contributing to total energy expenditure in the United States: results from NHAPS study. Int J Behav Nutr Phys Act
13. Dunstan DW, Barr ELM, Healy GN, et al. Television viewing time and mortality: the Australian Diabetes, Obesity and Lifestyle Study (AusDiab). Circulation
14. Dunstan DW, Salmon J, Healy GN, et al. Association of television viewing with fasting and 2-h postchallenge plasma glucose levels in adults without diagnosed diabetes. Diabetes Care
15. Dunstan DW, Salmon J, Owen N, et al. Associations of TV viewing and physical activity with the metabolic syndrome in Australian adults. Diabetologia
16. Dunstan DW, Salmon J, Owen N, et al. Physical activity and television viewing in relation to risk of undiagnosed abnormal glucose metabolism in adults. Diabetes Care
17. Evenson KR, McGinn AP. Test-retest reliability of adult surveillance measures for physical activity and inactivity. Am J Prev Med
18. Ford ES, Kohl HW III, Mokdad AH, Ajani UA. Sedentary behavior, physical activity, and the metabolic syndrome among U.S. adults. Obes Res
19. Healy GN, Dunstan DW, Salmon J, Shaw JE, Zimmet PZ, Owen N. Television time and continuous metabolic risk in physically active adults. Med Sci Sports Exerc
20. Hu FB, Leitzmann MF, Stampfer MJ, Colditz GA, Willett WC, Rimm EB. Physical activity and television watching in relation to risk for type 2 diabetes mellitus in men. Arch Intern Med
21. Hu FB, Li TY, Colditz GA, Willett WC, Manson JE. Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women. J Am Med Assoc
22. Jacoby E, Goldstein J, Lopez A, Nunez E, Lopez T. Social class, family, and life-style factors associated with overweight and obesity among adults in Peruvian cities. Prev Med
23. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics
24. Matthews CE, Ainsworth BE, Thompson RW, Bassett DR Jr. Sources of variance in daily physical activity levels as measured by an accelerometer. Med Sci Sports Exerc
25. Matthews CE, Chen KY, Freedson PS, et al. Amount of time spent in sedentary behaviors in the United States, 2003-2004. Am J Epidemiol
26. Matton L, Wijndaele K, Duvigneaud N, et al. Reliability and validity of the Flemish Physical Activity Computerized Questionnaire in adults. Res Q Exerc Sport
27. McCormack G, Giles-Corti B, Milligan R. The test-retest reliability of habitual incidental physical activity. Aust N Z J Public Health
28. Owen N, Healy GN, Matthews CE, Dunstan DW. Too much sitting: the population health science of sedentary behavior. Exerc Sport Sci Rev
29. Pate RR, O'Neill JR, Lobelo F. The evolving definition of "sedentary." Exerc Sport Sci Rev
30. Rosenberg DE, Bull FC, Marshall AL, Sallis JF, Bauman AE. Assessment of sedentary behavior with the International Physical Activity Questionnaire. J Phys Act Health
. 2008;5(1 suppl):S30-44.
31. Sallis JF, Owen N, Fotheringham MJ. Behavioral epidemiology: a systematic framework to classify phases of research on health promotion and disease prevention. Ann Behav Med
32. Salmon J, Bauman A, Crawford D, Timperio A, Owen N. The association between television viewing and overweight among Australian adults participating in varying levels of leisure-time physical activity. Int J Obes Relat Metab Dis
33. Salmon J, Owen N, Crawford D, Bauman A, Sallis JF. Physical activity and sedentary behavior: a population-based study of barriers, enjoyment, and preference. Health Psychol
34. Shields M, Tremblay MS. Screen time among Canadian adults: a profile. Health Rep
35. Sidney S, Sternfeld B, Haskell WL, Jacobs DR Jr, Chesney MA, Hulley SB. Television viewing and cardiovascular risk factors in young adults: the CARDIA study. Ann Epidemiol
36. Sugiyama T, Healy GN, Dunstan DW, Salmon J, Owen N. Is television viewing time a marker of a broader pattern of sedentary behavior? Ann Behav Med
37. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc
38. van Uffelen JGZ, Wong J, Chau JY, et al. Occupational sitting and health risks: a systematic review. Am J Prev Med
39. Willett W, Stamfler M. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol
40. Williams DW, Raynor HA, Ciccolo JT. A review of TV viewing and its association with health outcomes in adults. Am J Lifestyle Med
41. Winkler AE, Gardiner PG, Healy GN, et al. Distinguishing true sedentary from accelerometer non-wearing time: accuracy of two automated wear-time estimations. Med Sci Sports Exerc
42. Zhang H, Werner P, Sun M, Pi-Sunyer FX, Boozer C. Measurement of human daily physical activity. Obes Res
43. Zhang M, Xie X, Lee AH, Binns CW. Sedentary behaviors and epithelial ovarian cancer risk. Cancer Causes Control