Relationship between young peoples' sedentary behaviour and biomedical health indicators: a systematic review of prospective studies

Authors

  • M. J. M. Chinapaw,

    Corresponding author
    1. Department of Public and Occupational Health, EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • K. I. Proper,

    1. Department of Public and Occupational Health, EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • J. Brug,

    1. Department of Epidemiology & Biostatistics, EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • W. van Mechelen,

    1. Department of Public and Occupational Health, EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • A. S. Singh

    1. Department of Public and Occupational Health, EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author

MJM Chinapaw, Department of Public and Occupational Health, EMGO Institute for Health and Care Research, VU University Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands. E-mail: m.chinapaw@vumc.nl

Summary

The aim of this systematic review was to describe the prospective relationship between childhood sedentary behaviour and health indicators. We identified prospective studies from searches in PubMed, EMBASE, PsycInfo and Cochrane, from January 1989 through April 2010. Two reviewers independently screened the titles and abstracts for eligibility, rated the methodological quality of the studies, and extracted data.

We identified 31 papers, examining 27 different cohorts. The quality score of the studies ranged from 38 to 88%. Nine studies were scored as high quality. According to the best evidence synthesis we found insufficient evidence for a longitudinal positive relationship between ‘sedentary time’– mainly TV viewing – and body mass index (BMI) and more specific indicators of fat mass. One high quality and two low quality studies found a significant inverse relationship between sedentary time – mainly TV viewing – and aerobic fitness, leading to moderate evidence for this inverse relationship. There was insufficient evidence for a longitudinal relationship between sedentary time and blood pressure, blood lipids or bone mass.

Our systematic review suggests that there is moderate evidence for a longitudinal inverse relationship between screen time and aerobic fitness during childhood. Thus there is evidence to limit screen time in young people in order to prevent low levels of fitness. The possible detrimental health effects of prolonged or excessive sitting on other health indicators needs further study.

Introduction

Sedentary behaviour is defined as a distinct class of behaviours (e.g. sitting, watching TV, computer use) characterized by little physical movement and low energy expenditure (≤1.5 METs) (1,2). Children in developed countries spend a lot of time in sedentary behaviours and today's children are perceived to be more sedentary than previous generations (3). The increased time spent in sedentary behaviour has been attributed to developments in availability, accessibility and attractiveness of electronic media. Children spend more time with electronic media – TV, videogames and Internet – than any other activity except for sleeping (4). In the lives of today's 8- to 18-year olds, television viewing dominates media consumption, taking up about 4.5 h a day in young people's lives. TV viewing is followed by listening to music and other audio (2 h and 31 min), computer use (1.29 h), playing video games (1.13 h), reading (38 min) and watching movies in a movie theatre (25 min) (5). Thus, it is of great Public Health importance to study the possible detrimental effects of prolonged sitting on health.

Relative to the large amount of evidence regarding the acute and chronic effects of physical activity in adults, little is known about the adverse health outcomes caused by prolonged sitting and other ubiquitous sedentary behaviours, especially in young people (6). Sedentary behaviour and physical activity are two distinct classes of behaviour (7) and are likely to have independent effects on weight and health indicators (8). Individuals may meet the recommendations for moderate-to-vigorous physical activity and may still be at increased risk of ill health effect due to long and prolonged engagement in sedentary behaviours the rest of the day (2).

The increasing amount of scientific publications on the topic of sedentary behaviour, and in particular screen time behaviour, indicates that sedentary time is positively associated with overweight. Based on a meta-analysis, Marshall et al. (9) concluded that a small but statistically significant positive relationship exists between TV viewing time and body fatness in youth, but that the clinical relevance is uncertain. However, the included studies were mainly (83%) cross-sectional; only nine longitudinal samples were included. In their narrative review paper of 15 prospective studies on the relationship between childhood sedentary behaviour and the development of overweight and obesity, Must and Tybor (10) concluded that the prospective relation between sedentary behaviour and the development of obesity appeared to be consistent before adolescence only. Rey-Lopez et al. (11) conducted a review on sedentary behaviour and body composition in children and adolescents including cross-sectional, longitudinal and intervention studies published up to April 2007. In line with Must and Tybor, these authors also concluded that sufficient evidence exists for a positive relationship between time spent watching TV and body composition, especially for younger children. However, they did not include the quality of the included studies in their evidence synthesis.

Since 2005, a number of further prospective studies on the association between sedentary time and overweight/obesity as well as some other health indicators have been published. This warrants a true systematic review on the evidence regarding the prospective relation between sedentary time and health indicators in youth, in order to draw conclusions for interventions and further research. In the present paper, we therefore applied a systematic approach to appraise the quality, and summarize and integrate the results of the available prospective studies examining the relationship between sedentary behaviours and various health outcomes in youth.

Methods

Literature search

We performed a search in PubMed, EMBASE, PsycINFO and the Cochrane Library from 1 January 1989 up to April 2010. The search strategy focused on terms related to ‘sedentary behaviour’ (TV viewing, inactivity, etc.) in AND-combination with terms representing relevant health indicators (Body Mass Index (BMI), skinfolds, fitness, endurance, health, etc.) and study type (‘Longitudinal Studies’[Mesh] OR ‘Prospective Studies’[Mesh] OR ‘Intervention Studies’[Mesh] OR ‘Randomised Controlled Trial’[Publication Type] OR ‘Prospective Studies’[Mesh] OR ‘relation’ OR ‘association’). The full search strategy can be obtained upon request. For the purpose of the current review, we only included studies that involved children and adolescents. The review of studies among adults is published elsewhere (12).

Inclusion criteria

Studies were included if they were prospective studies (observational cohorts and intervention studies), examining the longitudinal relationship between sedentary behaviour, e.g. TV viewing, videogames, screen time, assessed during youth (mean age ≤18 years of age) and biomedical health indicators, e.g. BMI, fitness, blood pressure, lipid profile, assessed during youth or adulthood. We only included full-text articles published in English in peer-reviewed journals.

Selection process

First, two reviewers (KP and AS) independently checked all titles and abstracts of articles identified through the search process to identify potentially relevant articles. Next, a third reviewer (MC) checked inconsistencies and screened the full papers to determine whether they met the inclusion criteria. In total, 31 studies fulfilled all inclusion criteria describing 27 primary studies.

Data extraction

Two researchers (MC and KP) independently performed the data extraction using a structured form developed for this review (available upon request). The following data were extracted: (i) study sample; (ii) follow-up duration; (iii) measurement (objective or self-report) and type of sedentary behaviour; (iv) measurement (objective or self-report) and type of health indicator; (v) statistical analysis and (vi) main results. Disagreement between the reviewers with regard to the extracted data was discussed until consensus was reached. This information is summarised in Table 1.

Table 1.  Description of the study samples, follow-up duration, type and measurement of sedentary behaviour and heath outcome, and main results of the reviewed studies sorted by health outcome and quality score (QS)
Author (year of publication); QSStudy populationFollow-upType of sedentary behaviourHealth outcomeResults
  1. BMI, body mass index; DEXA, dual-emission X-ray absorptiometry; MVPA, moderate to vigorous physical activity; NR, not reported; NS, not significant; OR, odds ratio; PA, physical activity; QS, quality score; SE, standard error; SEM, standard error of the mean; SES, socioeconomic status.

 1. Delmas et al. (2007) (22); 75%n = 379, 51% boys, 12 years old adolescents, BMI 18.8 ± 0.2, 23% overweight, France3 yearsTV viewing, reading, no-sport club, drawing, choral (self-report)BMI, waist circumference (objective), fat% (Bioimpedance)In boys but not girls, high baseline TV/video viewing was positively associated with BMI, waist circumference and body fat adjusted for sexual maturity
 2. Treuth et al. (2009) (41); 75%n = 984 girls mean age 11.9 ± 0.4 years (2003), USA2 yearsSedentary time (accelerometer count <50 counts per 30 s)BMI and triceps skinfold (objective)Adjusted association between sedentary time and BMI/%body fat deviation scores: 0.001 (−0.09, 0.10)/−0.26 (−0.57, 0.05) adjusted for race, grade and school
OR for incident overweight/%body fat according to change in sedentary time: 0.99 (0.99, 1.00)/0.99 (0.99, 1.00) adjusted for race and school
 3. Epstein et al. (2008) (24); 75%n = 70, 53% boys aged 4–7 with BMI >75th percentile mean age 6 ± 1.2, USA2 yearsTV and computer use (objective by monitoring device)BMI z-score (objective)TV and computer use mediated the effect of the intervention on BMI z-score values over time. Changes in energy intake but not changes in PA were differentially related to changes in SB
 4. Hancox et al. (2004) (28); 75%n = 991 born in 1972 or 1973 in Dunedin, 52% boys, age 5, New Zealand21 yearsTV viewing (proxy report at age 5–11 and self-report at age 13, 15 and 21)BMI, blood pressure, VO2max, Cholesterol (objective)Average TV viewing between 5–15 years was associated with unfavourable health at 26 years:
 Higher BMI b = 0.48, P = 0.01
 Lower VO2max −0.12, SE = 0.04, P = 0.0009
 Raised cholesterol b = 0.09, SE = 0.04, P = 0.04
Adjusted for sex, SES, BMI age 5, parental BMI, and reported PA at age 15 for VO2max
 5. Landhuis et al. (2008) (35); 75%n = 991 born in 1972 or 1973 in Dunedin, 52% boys, age 5, New Zealand27 years (age 5 −32)TV viewing (5–11 years parent report, 13, 15 and 32 years self-report)BMI (objective)Averaged TV viewing age 5–32 and BMI/obesity age 32:
 BMI: b = 0.76 (0.35; 1.18)
 Obesity: OR = 1.13 (1.02; 1.74)
Adjusted for gender, SES, early BMI, parent BMI, PA, smoking, parental control, sleep time
 6. Robinson et al. (1993) (40); 75%n = 279 girls, mean age 12.4 ± 0.7, USA2 yearsTV viewing (self-report)BMI, triceps skinfold thickness (objective),Multivariate regression of after school TV viewing and: BMI: b = 0.05, P = 0.82/triceps skinfold: b = −0.19, P = 0.67 adjusted for age, race, parent education, parent fatness and baseline PA
 7. Laurson et al. (2008) (36); 63%n = 268, 54% boys, age 10 years, USA18 monthsTV viewing, videogames and computer use (self-report)BMI (objective)Longitudinal correlations between screen time and BMI were low and non-significant: b = 0.029 (±0.011) and −0.087 (±0.014) for boys and girls, respectively controlling for age, baseline BMI, change in height, ethnicity, and state of residence
 8. Boone et al. (2007) (19); 63%n = 9155, 53% boys, aged 13–20, mean age 15.9 ± 0.12, USA6 yearsTV and video viewing (self-report)BMI (objective)Boys/girls:
Predictors of 5 year incident obesity: baseline screen time b = 0.007 (0.0002; 0.01)/0.015 (0.006; 0.02), screen time change b = 0.006 (0.0009; 0.01)/0.01 (0.005; 0.02) adjusted for household income, parent education, race, smoking
MVPA
OR of 5 year incident obesity 4 vs. 40 h screen time = 0.78 (0.61; 0.99)/0.58 (0.43; 0.80) fully adjusted
 9. Chen et al. (2007) (20); 63%n = 307, 48% boys, aged 7–8, mean age 7.5 ± 0.5, China1 yearTV/video viewing and playing video/computer games (self-report)BMI (objective)More TV viewing and PC time on BMI at follow-up: b = 0.156, P = 0.059 adjusted for baseline BMI, maternal BMI, aerobic capacity (PA and dietary intake were NS and removed from model)
10. Davison et al. (2006) (34); 63%n = 169 white girls age 7.3 from well-educated parents, USA4 years (age 7, 9 and 11)TV viewing (mother report)BMI (objective)TV viewing between age 7–11 and overweight at 11year adjusted for income, mothers education, girls BMI at age 5 and pubertal development at age 11:
 1–2 times > 2 h TV per day OR = 2.7 (0.9–8.2)
 3 times > 2 h TV per day OR = 13.2 (2.9–59.5)
TV viewing between age 7–11 and became overweight between 7 and 11 years adjusted for income, mothers education, pubertal development at age 11:
 1–2 times > 2 h TV per day OR = 2.1 (0.7–6.6)
 3 times > 2 h TV per day OR = 4.7 (1.2–17.9)
11. Elgar et al. (2005) (23); 63%n = 355, 45% boys aged 11–14, mean age 12.3 years ± 6.3, UK4 yearsTV/video/computer games (self-report)BMI (objective)Screen time in year 7 and BMI in year 11: B = 0.04 SE = 0.01, P < 0.01 adjusted for sex, age, # parents, family size, SES, h per week sports, meal skipping and # snacks per day
B = 0.00 SE = 0.01, P > 0.05 adjusted for sex, age, # parents, family size, SES, h per week sports, meal skipping and # snacks per day and BMI at age 7
12. Gorden-Larson et al. (2002) (26); 50%n = 12 759, 49% boys, aged 11–19, USA1 yearTV/video viewing, computer video game use (self-report)BMI (NR)Baseline TV/Video viewing ≥35 h week−1 boys/girls: OR = 1.49 (1.14–1.97)/1.43 (1.07; 1.90)
One year change in TV/video viewing ≥7 h week−1 boys/girls OR = 1.08 (0.89; 1.30)/1.15 (0.96; 1.39)
Video ≥4 h week−1 boys/girls OR = 1.11 (0.95; 1.30)/1.08 (0.85; 1.37) adjusting for age, maternal education, family income, presence of mother in household, urban residence, cigarette smoking and region
13. Gortmaker et al. (1999) (27); 50%n = 1295 52% boys students from grade 6,7 mean age 11.7 ± 0.7 years, USA2 schoolyearsTV and video viewing, playing video games (self-report)BMI, tricep skinfold (objective)TV viewing mediated the intervention effect on obesity prevalence: Each hour reduction in TV viewing was associated with a reduction in obesity prevalence (OR = 0.85; 95% CI, 0.75–0.97; P = 0.02) adjusting for baseline TV. Among girls who were obese at baseline, controlled for baseline TV viewing and change in TV viewing, each hour of reduction in TV viewing was independently associated with increased remission of obesity (OR = 1.92; 95% CI, 1.37–2.70; P = 0.002).
14. Hancox and Poulton (2006) (29); 63%n = 991 born in 1972 or 1973 in Dunedin, 52% boys, age 5, New Zealand12 yearsTV viewing (proxy report at age 5–11 and self-report at age 13–15)BMI (objective)Significant association between TV viewing and BMI raging from 0.09 at age 5 to 0.28 at age 15 adjusted for sex, SES, baseline BMI, parental BMI
15. Jago et al. (2005) (32); 63%n = 138 in 1986 3–4 years old children 47–49% boys, Anglo/Afro/Hispanic, USA3 yearsStationary non-moving non trunk movement (observation)
TV viewing (observation)
BMI (objective)BMI across all 3 years adjusted for baseline BMI, gender, ethnicity, and PA (diet was NS and removed from model):
 TV/H*year 1 b = 0.04, P = 0.02
 TV/H*year 2 b = 0.04, P = 0.009
16. Henderson (2007) (30); 63%n = 2379 white and black girls, age 9 or 10, USA10 yearsTV viewing (self-report)BMI (objective)For white girls only baseline TV viewing (at age 9 or 10) was associated with a steeper trajectory of BMI over following 4 years (ages 11–14) (b = 0.03 SEM = 0.01 P = 0.005) adjusted for baseline BMI, sexual maturation, parent education, parent income and baseline PA.
TV viewing at wave 5 was not associated with BMI trajectory over wave 6–10 adjusted for baseline BMI, sexual maturation, parent education, parent income and baseline PA adjusted for baseline BMI, pubertal stage, parent education, parent income and baseline PA.
17. Maffeis et al. (1998) (37); 63%n = 112, 52% boys, mean age 8.6 ± 1.0, Italy4 yearsTV viewing (parent report)BMI (objective)No significant independent relationship between TV viewing and BMI (age, gender, energy intake, TV time and VPA were rejected in the model)
18. Must et al. (2007) (38); 63%n = 156 girls age 8–12 years mean age 10.0 ± 0.9, USAUntil 4 years post menarche, mean = 7.5 yearsTime spent TV viewing, video and computer games (self-report)BMI (objective), % body fat (bio-impedance),BMI z-score slope: r = 0.03, P = 0.29, n = 139 adjusted for age, age at menarche, parental overweight, soda intake and % of calories from fat
% Body fat change: r = 0.24, P = 0.26, n = 139 adjusted for age, age at menarche, parental overweight, and % of calories from fat
19. Proctor et al. (2003) (39); 63%n = 106, 50% male, mean age at baseline 4.0 years, USA7 years (until age 11.1)TV and video viewing (parent report)BMI, triceps skinfold and sum of skinfolds (objective)Children ≥3 h day−1 TV watching had statistically significant higher BMI, triceps, and sum of five skinfolds than children watching <1.75 h day−1 controlled for parental body fat, physical activity, and dietary variables.
20. Ventura et al. (2006) (42); 50%n = 154 girls, mean age 5, USA8 yearsTV and computer usage (self-report)BMI (objective), adiposity (DXA), blood lipids and glucose (fingerprick age 13)No significant relationship between sedentary time and metabolic risk score
21. Gable et al. (2007) (25); 50%n = 8459 participants from kindergartens, 48% boys, mean age 68.4 ± 4.1 months, USA4 yearsTV and video viewing at home (parent-report)BMI (objective)Overweight at 3rd grade is predicted by child TV hours at kindergarten and first grade OR = 1.02 (1.0; 1.04) adjusted for days per week of aerobic exercise at kindergarten spring, opportunities for activity at kindergarten spring, average number of family meals per week at kindergarten and first-grade spring, and parents' reports of neighbourhood safety at kindergarten and first-grade spring.
22. Kaur et al. (2003) (33); 38%n = 2223 12–17 years olds 52% boys, 68% white, 6% overweight, USA3 yearsTV viewing (self-report)BMI (self-report)Baseline TV and BMI% at follow-up
b = 0.47 se = 0.02, P = 0.02 adjusted for ethnicity and baseline BMI
>2 h TV at baseline and overweight at follow-up:
 Overall: OR = 2.2 (1.4; 3.6)
 Normal weight: OR = 1.9 (1.1; 3.5)
 Overweight: OR = 2.8 (1.3–6.3)
23. Viner and Cole (2006) (43); 38% n = 4461, NR% boys, mean age 16 years14 yearsTV viewing, videos, computer games (self-report)zBMI (self-report)zBMI: r = 0.15 (0.03,0.27) P = 0.01 for ≥4 h day−1 screen time adjusted for sex, social class, height at age 16 and 30 and zBMI at age 16
24. Aires et al. (2010) (18); 38%n = 345, 43% boys, aged 11–16, mean age 14.0 ± 1.4, Portugal3 yearsTV and PC time (self-report)BMI (NR)Screen time > 2 h and BMI: b = −0.08 (−0.55; 0.39)
Multivariate adjusted for parent education, physical activity
25. Berkey et al. (2003) (21); 38%n = 11 887, 51% boys aged 10–15, USA1 yearTV, video, VCR, computer games (self-report)BMI (self-report)1 year change in screen time and 1 year change in BMI: b = 0.004 (−0.03; 0.04) and b = 0.05 (0.02; 0.08) for boys and girls, respectively adjusted for age, race, menarche history (girls), tanner stage, height growth, baseline BMI, energy intake in past years, baseline screen time and PA
26. Horn et al. (2001) (31); 88%n = 198, 48% boys, mean age 7.5 ± 1.3 years Mohawk, Canada2 yearsTV viewing and videogame playing (self-report)Subscapular skinfold (objective)Girls: excessive TV viewing and change in skinfold thickness R = 0.14 SE = 0.07, P < 0.05 adjusted for age, baseline skinfold and PA. In boys excessive TV viewing was not entered in the model
27. Janz et al. (2005) (45); 38%n = 378, 47% boys, mean age 5.6 years, 95% white, USA3 yearsTV viewing, video/computer games (parent report)Fat mass (DEXA)Children maintaining low levels of TV viewing were less likely than peers to be in the upper quartile for adiposity at follow-up, and were less likely to gain adiposity during the study period.
28. Skinner et al. (2003) (44); 50%n = 52, 45% boys, mean age 7, USA1 yearTV viewing, videos, computer games, audiotapes and other non-active activities (mother report)Body fat indexes (DEXA)% Body fat: adjusted for calcium, total fat, polyunsaturated fat and gender
kg body fat: r = 0.66, P = 0.02 adjusted for calcium, total fat, and polyunsaturated fat
29. Bounds et al. (2005) (46); 38%n = 52, 48% boys from 6–7 years old followed to 8 years, white, middle-upper SES2 yearsTV, video, computer games, listening audio, reading (mother report)Bone mass (DEXA)Hours per day in sedentary activities at age 6 and 7 years was not significantly correlated with children's total bone mass at age 8 years according stepwise regression analysis including dietary intake of energy and micronutrients, sex, BMI, age and mothers body mass
30. Aires et al. (2010) (47); 50%n = 345, 43% boys, aged 11–19 years, mean age 14.0 ± 1.4, Portugal3 yearsTV and PC time (self-report)Fitness (curl-ups, push-ups and 20m shuttle run)Change in screen time and change in physical fitness: b = −0.002 (−0.006; 0.002) adjusted for age, gender, baseline fitness, change in PA and change in BMI and interaction term with gender
31. Treuth et al. (2004) (48); 50%n = 91 normal-weight girls, 8 years old, Tanner stage 1, USA2 years (age 9 and 10)TV viewing during school year and summer (self-report)Fitness by treadmill testingTV viewing during school year was after adjusting for weight, fat free mass, tanner stage and ethnicity negatively associated with:
 VO2peak r = −0.17, P < 0.05
 Time on treadmill r = −0.20, P < 0.05
 Final stage r = −0.16, P < 0.05
No significant effects for time spent TV viewing in school or summer adjusted for weight, fat free mass, tanner stage and ethnicity

Quality assessment

All included studies were examined using the 13 quality criteria listed in Table 2 adapted from existing quality criteria lists (13–15). The list of 13 items is categorized in four dimensions: (i) study population and participation (three items); (ii) study attrition (four items); (iii) data collection (three items) and (iv) data analysis (three items). Further, the criteria distinguish between informativeness (I, n = 5) and validity/precision (V/P, n = 8). Criteria have a ‘yes’ (= 1), ‘no’ (= 0) or ‘don't know’ (= 0) answer format. Two independent reviewers (MC and KP) performed the quality assessment. We gave a positive score if the publication provided an informative description of the criterion at issue and met the quality criterion. A negative score was given in case of an informative description, but an inadequate performance or lack of description. We gave a question mark in case of an unclear or incomplete description of the item. If the study referred to another publication describing the design or other relevant information about the same study, we retrieved the additional publication to score the criterion of concern. Disagreements were discussed and resolved. For each study, we calculated a total methodological quality score by counting the number of items scored positively on the validity/precision (V/P) criteria divided by the total number of validity/precision criteria (i.e. 8). We considered a study of high quality if the methodological score was at least 0.75, i.e. 75%. A lower score was defined as low quality.

Table 2.  Criteria List for Assessment of the Methodological Quality of Prospective Studies (based on Singh et al. (15), Hayden et al. (14), Hoogendoorn et al. (13))
Criteria (rating of criteria: + = yes, − = no, ? = not or insufficiently described)I, V/P*% of studies scoring +
  • *

    I = criterion on informativeness, V/P = criterion on validity/precision.

  • Adequate = sufficient information to be able to repeat the study.

  • ‘+’ is given only if adequate information is given on all items.

  • §

    ‘+’ is given only if non-selective dropout on key characteristics (age, gender, sedentary behaviour, health outcomes) is reported in the text or tables.

  • ‘+’ is given only if a multivariate regression model was used adjusting for PA or diet in case the dependent variable was BMI, and for PA in case the dependent variable was fitness.

  • BMI, body mass index.

Study population and participation (baseline): The study sample represents the population of interest on key characteristics  
  1. Adequate description of sampling frame, recruitment methods, period of recruitment, and place of recruitment (setting and geographical location)I59
  2. Participation rate at baseline at least 80%, or if the non-response was not selective (show that baseline study sample does not significantly differ from population of eligible subjects)V34
  3. Adequate description of baseline study sample (i.e. individuals entering the study) for key characteristics (number of participants, age, gender, sedentary behaviour, and health outcome)I59
Study attrition: Loss to follow-up is not associated with key characteristics (i.e. the study data adequately represent the sample)  
  4. Provision of the exact number of participants at each follow-up measurementI66
  5. Provision of exact information on follow-up durationI97
  6. Response at short-term follow-up (up to 12 months) was at least 80% of the number of participants at baseline and response at long-term follow-up was at least 70% of the number of participants at baselineV63
  7. Not selective non-response during follow-up measurement(s)§V/P16
Data collection  
  8. Adequate measurement of sedentary behaviour: done by objective measures (i.e. accelerometry, heart rate monitoring, observation) and not by self-report (self-report = −, no/insufficient information = ?)V1
  9. Sedentary behaviour was assessed at a time point prior to the measurement of the health outcomeV100
 10. Adequate measurement of the health outcome: objective measurement of the health outcome done and not by self-report (self-report = −, no/insufficient information = ?)V81
Data analyses  
 11. The statistical model used was appropriateV/P94
 12. The number of cases was at least 10 times the number of the independent variablesV/P91
 13. Presentation of point estimates and measures of variability (confidence interval or standard error)I59

Levels of scientific evidence

After summarizing the included studies, it appeared that the studies were heterogeneous, especially with regard to the type and measurement of sedentary behaviour and the health outcome. Additionally, of those studies that examined the same health outcome, the statistical analyses varied between the studies, including the categorization of the independent variable (sedentary behaviour) resulting in different types of effect sizes (odds ratios and regression coefficients) making statistical pooling impossible. Therefore, to synthesize the methodological quality of the studies and to be able to draw conclusions regarding the relationship between sedentary behaviour and the health outcome, we applied a best evidence synthesis (13,15). This rating system consists of three levels and takes into account the number, the methodological quality (V/P), and the consistency of outcomes of the studies:

  • • Strong evidence: consistent findings in multiple (≥2) high-quality studies.
  • • Moderate evidence: consistent findings in one high-quality study and at least one low-quality study, or consistent findings in multiple low-quality studies.
  • • Insufficient evidence: only one study available or inconsistent findings in multiple (≥2) studies.

Similar to previous reviews that applied this best evidence synthesis (13)(15–17), we considered results to be consistent when at least 75% of the studies showed results in the same direction, which was defined according to significance (P < 0.05). If there were two or more high-quality studies, we disregarded the studies of low methodological quality in the evidence synthesis; those studies were thus not incorporated in the conclusion.

Results

The literature search yielded a total of 7723 hits: 3837 in PubMed, 3034 in Embase, 277 in Cochrane and 575 in Psychinfo. After checking of abstracts and full articles, we included 31 publications examining 27 different study samples (18,19)(20–49). Table 1 summarizes the details of the reviewed studies.

Sample characteristics

Sample sizes ranged from 52 to 12 759. The mean age at baseline varied from 3 years up to around 17 years old, with the majority of studies (n = 21) performed in children with a mean baseline age < 12. The mean age at follow-up varied from 5–6 years up to around 32 years old. The follow-up period varied from 1 to 29 years.

Study quality

Table 3 presents the methodological quality scores. The quality of the trials ranged from 38% to 88%. Eight of the 31 studies were of high quality. Table 2 shows the % of studies scoring positive on each quality item. Only three studies used an objective measure of sedentary behaviour. Only six papers reported that the non-response at follow-up was not selective (item 7). Ten studies had a participation rate at baseline of at least 80% or reported that the non-response was not selective (item 2).

Table 3.  Quality assessment of prospective studies on sedentary behaviour (SB) and health outcomes in youth sorted by quality score
AuthorRecruitmentParticipation rateDescription baseline sampleNumbers at follow-upFollow-up durationResponse rate at follow-upNot-selective non-responseMeasure SBSB measured before health outcomeMeasure health outcomeAppropriate statistical model# Cases at least 10* # independent variablesPoint estimates and measures of variabilityQuality score (%)
  1. +, yes; −, no; ?, not or insufficiently described; SB, sedentary behaviour.

 1. Horn et al. (31)++++++++++++88
 2. Treuth et al. (41)++++++?+++++75
 3. Delmas et al. (22)+++++++++75
 4. Epstein et al. (24)?+++++++++75
 5. Landhuis et al. (35)+++++++++75
 6. Hancox et al. (28)++++++++++75
 7. Robinson et al. (40)+++++?+++++75
 8. Laurson et al. (36)?++++++++63
 9. Boone et al. (19)+?++++?+++++63
10. Chen (20)++++++++63
11. Elgar et al. (23)+++++++++63
12. Henderson (30)+?+++++++63
13. Jago et al. (32)??+++++++++63
14. Davison et al. (34)?++++?+++++63
15. Maffeis et al. (37)+?+++++++++63
16. Must et al. (38)+?+++?++++63
17. Proctor et al. (39)+?++++?++++63
18. Gordon-Larsen et al. (26)+++++?++50
19. Gortmaker et al. (27)++++++++50
20. Hancox and Poulton (29)?+++++?++50
21. Ventura et al. (42)?+++?+++50
22. Aires et al. (47)?+?+??+++++50
23. Gable et al. (25)?++++++50
24. Skinner et al. (44)++??++++50
25. Treuth et al. (48)+?++++?+++50
26. Kaur et al. (33)++++++++38
27. Viner and Cole (43)++++++++38
28. Aires et al. (47)?+?+??+?+++38
29. Berkey et al. (21)??++++++38
30. Bounds et al. (46)??++?+++38
31. Janz et al. (45)??+?+??+++38

Measurement of sedentary behaviour

All studies using self-report measures examined TV viewing; other frequently assessed sedentary behaviours were playing computer or video games (n = 12) and computer use (n = 5). Except for three studies (24,32,41) all measurements of sedentary behaviour were self-reported and/or parent-reported. One of the studies using an objective measure for sedentary time used observations to measure time spent in stationary non-moving, non-trunk movement and TV viewing (32). One study measured sedentary time by accelerometers using a cut-point of <100 counts per minute (41) and one study by an automated device controlling and monitoring the use of televisions or computer monitors (24).

Association between sedentary behaviour and health outcomes

Twenty-six studies examined the longitudinal relationship between sedentary time and BMI or BMI z-score. All studies looked at TV viewing and 12 included computer use as well. Four of six high-quality studies observed a significant positive relationship between sedentary time and BMI of which one study in boys only (22). Two high-quality studies in girls only found no significant association between sedentary time and BMI (40,41). According to the best evidence synthesis we found insufficient evidence for a positive longitudinal relationship between self- or proxy-reported sedentary time and BMI when considering boys and girls together. However, for girls, it appeared that there was no evidence for such a relationship.

Ten studies examined sedentary time in relation to indicators of fat mass such as waist circumference (22), fat percentage (22,38,44,45,50) or skinfold thickness (27,31,39–41). Five studies found a significant positive relationship while five did not. Two high-quality studies found a significant positive relationship, one in girls but not boys (31) and one in boys but not girls (22). Two other high-quality studies in girls only found no significant relationship meaning that we found insufficient evidence for a longitudinal relationship between self- or proxy-reported sedentary time and indicators of fat mass.

Few studies examined the longitudinal relationship between sedentary time and other health outcomes such as blood pressure (28), blood lipids (28,42), fitness/VO2max (28,47,48) or bone mass (46). Based on these findings, we can conclude that there is moderate evidence for a longitudinal inverse relationship between sedentary time and aerobic fitness (based on one high-quality and two low-quality studies) and insufficient evidence for a longitudinal relationship between sedentary time and blood pressure, blood lipids or bone mass.

Discussion

This article appraised the peer-reviewed literature published between January 1989 and April 2010 that reported on the longitudinal relationship between sedentary behaviour and health indicators in children and adolescents. The large majority of the studies focused on a relationship between TV time or screen time and BMI or indicators of fat mass. We found insufficient evidence for a positive longitudinal relationship between sedentary time – mainly TV viewing – and BMI and indicators of fat mass. Few studies examined the relationship with other health indicators. Moderate evidence was found for a significant inverse relationship between sedentary time and aerobic fitness. There was insufficient evidence for a longitudinal relationship between sedentary time and blood pressure, blood lipids and bone mass because these outcomes were studied in a single study.

The insufficient evidence for a positive relationship between screen time and BMI support the earlier meta-analysis of Marshall et al. (9) of mainly cross-sectional studies and the narrative review of Must and Tybor (10) and Rey-Lopez (11). This inconsistency is probably due to the fact that we applied a best evidence synthesis based on high-quality studies only. Of the three high-quality studies reporting a significant positive association between screen time and BMI included in our review, two reported on the same sample and focused on the average TV viewing time over a longer period (5–15 years and 5–32 years, respectively). The third study included children aged 4–7 with a BMI >75th percentile.

All studies reviewed focussed on associations between total sedentary time or a particular kind of sedentary behaviour and health outcomes. Recent studies in adults suggest that sedentary patterns, e.g. prolonged non-interrupted sedentariness, may be important for metabolic health risk. For example, a study by Healy et al. (51) in Australian adults found that independent of total sedentary time and moderate-to-vigorous intensity activity time, increased breaks in sedentary time were beneficially associated with waist circumference, BMI, triglycerides and 2 h plasma glucose. None of the included studies examined the effects of such sedentary patterns, because to date, no prospective studies on this topic have been published in youth.

As with any systematic review, this review is limited by the quality of the included studies. Eight of 31 studies (26%) were rated as high quality. A common limitation of the studies reviewed was the measurement of sedentary behaviour. The large majority of studies solely measured TV viewing time and all but three studies used self- or proxy-reported TV time. Measurement of sedentary time by self-report is prone to misclassification typically biasing the relationship to null. Bryant et al. (52) systematically reviewed studies related to overweight in children that had included a measure of TV exposure. They concluded that validity and reliability of self- or proxy-reported sedentary measures was often not examined. The majority of tools that were evaluated were compared with another self-report measure or an objective measure of physical activity. The use of accelerometry as an objective measure of physical activity and sitting time has opened new possibilities for studying the health effects of sedentary time and sedentary patterns. In our review only one study (41) used accelerometry to assess usual sedentary time. This study found that an increase in objective sedentary time was not associated with changes in BMI or per cent body fat.

Other common methodological limitations were that the required information to assess the participation rate or selective non-response at follow-up was often lacking or unclearly reported. A possible source of bias is publication bias: because studies with significant findings are more likely to be published, this can lead to an overestimation of a possible association. Another observation was that despite the prospective design of the included studies, sedentary time and the health outcome were measured at the same time points while preferably you would have also measurements of sedentary time at intermediate time points.

TV viewing was the most frequent surveyed type of sedentary behaviour. However, children spend time in much more sedentary activities such as sitting in the classroom, doing homework, playing computer games, talking on the phone and reading. Thus, TV may not be a good marker of sedentary behaviour in young people (53). Another potential limitation of TV viewing as an indicator of sedentary behaviour is that especially youth do many things simultaneously such as watching TV while using the mobile phone and doing homework. More research is needed to examine the relationship between different types of sedentary behaviours and health indicators. Another point of attention is the growing popularity of playing active video games. Future studies collecting data on video game playing need to distinguish between traditional and active gaming. Because the latter cannot be defined as sedentary behaviour.

While the evidence base is growing on potential detrimental health effects of prolonged or excessive sitting in adults, there is a lack of evidence in children and adolescents. Therefore, we need mechanistic studies and well-designed interventions to extend the small but growing evidence base. For future studies we recommend to (i) focus on other health indicators besides BMI or fat mass, e.g. metabolic health indicators and mental health; (ii) the use of a variety of measures of sedentary time, including objective measures; and (iii) examining the health effects of the way sedentary time is accumulated.

We must keep in mind that relationships between physical activity and health outcomes in youth are also often not strong because some of the outcomes may not be easily manifested in young age. Hence, not finding strong associations between sedentary behaviour and health indicators is unsurprising. In addition, the majority of studies focussed on the relationship between sedentary behaviour and health indicators both assessed before the age of 18 years. Thus, small relationships during that period could still be reflective of health problems later in life.

Our systematic review suggests that there is moderate evidence for an inverse longitudinal relationship between screen time and aerobic fitness during childhood. This evidence supports the development and implementation of interventions aiming to reduce screen time. There is insufficient evidence for a relationship between childhood sedentary time and BMI. To date, only few studies have examined the relationship between sedentary behaviour and other adverse health outcomes in children, and most studies have focused primarily on TV viewing. To be able to conclude that time spent in sedentary behaviours is causally related to adverse health outcomes, more high-quality prospective studies on different health outcomes are needed.

Conflict of Interest Statement

None declared.

Acknowledgements

We would like to thank Ilse Jansma, MSc, and Teatske Altenburg, PhD, from the VU University Medical Center, for their support in performing the literature searches for this review.

Ancillary