Table 1 Summary of the key characteristics of included systematic reviews

 Carson et al. (2016)[33]

5–17

From Feb 2010

162 [125 CS; 32 LG; 5 CC

TV, computer use, screen time, total SB

Various

No

Yes

Higher durations or frequencies of screen time and TV viewing were significantly associated with unfavourable measures of body composition across all study designs. But study quality rated very low to low.

No associations between accelerometer assessed sedentary time, breaks and bouts and body composition.

Update of Tremblay et al. (2011) [61]

 Cliff et al. (2016)[34]

2–18

To Nov. 2015

50 [37 CS; 9 LG; 4 CS + LG] (2–4 years = 3; 5–12 years = 37; 13–18 years = 10)

Objectively measured total SB, pattern of SB (i.e. breaks, bouts)

BF%, WC, BMI

Yes (19)

Yes

Overall: ‘no association’ for total volume of SB and adiposity.

Longitudinal studies: level of evidence classification was ‘no association’.

Cross-sectional meta-analysis: weak but significant positive association with high levels of heterogeneity observed.

Most studies measured total SB.

For meta-analysis of adiposity, selection of coefficient followed hierarchy: 1) BF% 2) WC; 3) BMI.

 Costigan et al. (2013)[35]

12–18

To Dec. 2011

19 [13 CS; 6 LG]

Screen-based SB: TV, video, computer, electronic gaming

BMI, body fatness, OW/OB

No

Yes

Strong evidence for a positive relationship between screen-based SB and weight status (especially for low risk-of-bias studies).

Leisure-time domain only.

Girls only.

 Fletcher et al. (2015)[36]

12–19

To Mar. 2014

21 (17 CS; 4 LG)

TV viewing, total screen time, computer use, and video game playing or video viewing (mostly self-report)

BMI (81% assessed objectively) Also: FMI and fat-free mass.

No

Yes

Moderate to strong evidence of the relationships between self-reported television viewing, total screen time and overall sedentary behavior with adiposity, independent of dietary intake.

Only included studies which adjusted for dietary intake.

 Froberg & Raustorp (2014)[37]

6–19

Jan 2000 to Oct 2013

35 [28 CS; 7 LG]

Objectively assessed volume (total time) and patterns (bouts and breaks)

Various

No

Yes

Limited evidence for an association between objectively assessed volume of sedentary time with markers of obesity, when controlling for MVPA.

Evidence on associations between sedentary bouts and breaks with weight status was inconclusive.

 Gorely et al. (2004)[38]

2–18

Not specified

24 [18 CS; 6 LG] (0-6y = 4; 7-18y = 20)

TV/video viewing time

Weight, body fatness

No

No

Body weight was positively associated with TV viewing time (4 samples)

‘Body fatness’ was unrelated to TV viewing time (association: 40% +; 61% no).

Excluded video/computer gaming.

 Leech et al. (2014)[39]

5–18

To Nov 2012

6 [4 CS; 2 LG]

Sedentary behaviors (e.g. TV viewing, video watching, using the computer or internet and playing console games)

BMI

No

No

Findings to support an association between obesogenic cluster patterns (diet, PA, SB) and overweight and obesity were inconclusive with longitudinal research.

Diet, PA and sedentary behavior cluster together in complex ways that are not well understood.

Despite the age group specified in the inclusion criteria, it is also noted that: “With the exception of one study [18] that included children aged 5–12 years, the ages of children and adolescents in these studies ranged from 9–21 years”.

 Marshall et al. (2004)[9]

3–18

1985 - ?

30 Independent samples: TV: 52 [43 CS; 8 LG; 1 IN]. Computer games: 6.

TV viewing, video/computer game use

“Body fatness”: BMI, skinfold

Yes (30)

No

Small but statistically significant relationship between TV viewing and body fatness.

Small and statistically non-significant relationship between video/computer game use and body fatness, but with small number of samples.

30 studies yielded 52 independent samples for the meta-analysis.

Sample proportions by age: < 7 y (8%); 7–12 y (46%); 13–18 y (23%); combination (23%).

 Mistry & Puthussery (2015)[25]

≤18

Jan 1990 to June 2013

5 (all CS) (All 5-18y)

TV and computer games

OW/OB

No

Yes

4/5 studies show positive correlation between TV/computer game time and weight status.

Selected studies all school-based.

The review focussed on studies from SE Asia only, with the final included studies being conducted in India only.

 Mitchell & Byun (2014)[40]

6–18

Jan 2008 - Sept 2012

63 [50 CS; 12 LG; 1 IN]

Self-reported SB (inc. screen time); objectively assessed SB

Various

No

No

Cross-sectional; screen time: 77% of studies show positive association with BMI; similar support for WC and fat mass.

Cross-sectional; objective sedentary time: Small number of studies favour null association

Longitudinal; screen time: favours association with changes in BMI and skinfolds Longitudinal; objective sedentary time: favours null association

Moderation analysis with 6 CS studies showed for all that screen time only associated with greater BMI if MVPA was low.

 Must & Tybor (2005)[28]

<22

Not specified

15 (all LG)

Any measure of “Inactivity/SB”

Mostly BMI/BMI z-score. Also: skinfold, DEXA, BF% by BIA

No

No

Most studies (especially with younger subjects) showed a positive association of “inactivity/SB” with weight or adiposity outcomes.

Average follow-up ≥2 years. 9/15 studies focused on children <10 year.

 Pate et al. (2013)[29]

5–18

Jan1990 - Jun 2012

4 (all LG)

Objectively measured SB (accelerometry)

Excessive fatness/body composition (adiposity, BMI, BMI-z, FMI, WC)

No

No

Mixed findings regarding association of SB with excessive fatness in children and adolescents.

No association among younger children (5–9 years); positive association indicated between sedentary time and BMI in older children and adolescents (9–15 years).

Prospective cohort studies.

 Prentice-Dunn & Prentice-Dunn (2012)[26]

2–19

2000–2010

9 (all CS) (11-18y;4-11y; 7-9y; Mean: 6.8y- SD:0.4;7-12y; 6y; 3-5y; 1-12y; Median:15y)

Parent-report or self-report of screen time, accelerometer counts, and direct observation

BMI and BF% (by various measures)

No

No

The majority of studies (7/9 studies) assessing sedentary behaviors (i.e. screen time) found a positive correlation with weight status.

The 2 studies showing no association used objective measures of SB.

Inconsistent reporting of SB studies. Number based on Table 3 - summary of SB and WS findings.

 Rey-Lopez et al. (2008) [41]

2–18

1990 - April 2007

78 [46 CS; 28 LG; 4 IN]

TV viewing, video games, computer use

Various

No

No

Cross-sectional: Positive association with OW/OB: TV viewing (k = 70 samples): 65–69% of studies Video games (k = 12): 50–67% Computer use (k = 18): 40–50%

Longitudinal: Positive association with OW/OB: TV viewing (k = 45): 62–67% of studies Video games (k = 6): 67–100% Computer use (k = 4): 100%

Interventions (k = 4): positive effects for weight status change in 1 of 4 studies.

 Saunders et al. (2016)[14]

5–17

To Jan 2015 (plus additional CINAHL search Jun. 2016)

10 (all CS) (6-18y = 9; mixed =1)

Any (accelerometer and screen-time)

BMI, WC, waist-to-height ratio, BF% (by BIA and DEXA), skinfolds.

No

Yes

A combination of high PA/low SB, compared with low PA/high SB, was associated with lower measures of adiposity and/or reduced risk of obesity.

Minimum sample size ≥300.

SB considered in combination with PA and/or sleep.

 Stierlin et al. (2015)[30, 42]

≤18

Jan. 2000 -May 2014

4 [3LG; 1 IN-RCT] [1 ‘toddlers & pre-schoolers’ (mean age 5y); 1 ‘children’ (mean age 6.3 and 10.3 for cohort 1 &2); 1 ‘adolescents’ (mean age 15.7y); 1 ‘children & adolescents’ (mean age ranging from 10.2–14.5y across countries).

Total SB time; subdomains of SB, inc. time spent watching TV, screen time, homework, reading. For studies using accelerometry, SB defined as <100 counts per minute.

Not reported

No

Yes

Screen time: positive association with weight status at follow-up (based on1 study).

Total SB: no association with weight status (based on 3 studies).

Review of determinants of SB; excluded CS studies.

“Consistent” evidence for weight status being positively associated with screen time comes from one study including two cohorts of young children (Fuller-Tyszkiewicz et al., 2012),

 Tanaka et al. (2014)[31]

<19

1950s to Dec 2013

3 (All LG) Age range at baseline: 7-9y. Follow up period: 2–7 years.

Objectively measured SB

Various

No

Yes

No clear evidence that increased sedentary time is associated with increased adiposity.

 Van Ekris et al. (2016)[32]

≤18

To Jan 2015

50 (All LG)

TV, computer use, screen time, total SB

BMI, WC, BF, skinfolds, weight, weight for height, OW/OB)

Yes (8)

Yes

TV: strong evidence for positive association with overweight/obesity. Other outcome measures: insufficient evidence. Computer use/gaming: no or insufficient evidence.

Screen time: strong evidence for BMI; moderate for OW/OB.

Total SB: no evidence for BMI, WC, body fat. Insufficient evidence for OW/OB.

Update of Chinapaw et al. (2011). 50 studies on ‘anthropometrics’. Additional papers reviewed considered ‘multiple indicators of cardio-metabolic health’. These are not listed here.

 Zhang et al. (2016)[27]

≤18

To June 2014

14 (all CS) (<6y = 3; 6-18y = 9; mixed = 2)

TV viewing time

“weight/height” (see Comments) For meta-analysis: OW/OB risk

Yes (14)

No (see Comments)

Increased TV watching is associated with increased risk of childhood obesity.

A linear dose–response relationship was found for TV watching and childhood obesity, and the risk increased by 13% for each 1 h/day increment in TV watching.

Minimum sample size for inclusion >200.

Unable to determine whether “weight/height” relates to specific assessment formula or just general categories.

Publication bias assessed but not individual study quality.

 Azevedo et al. (2016)[43]

≤17

1980 – March 2015

67 [17 with 0-5 years.; 35 with 5-12 years.; 15 with 12-17 years.] (61 RCT or cluster RCT; 6 non-randomized CTs) [6 (SB only), 10 (SB + PA), 51 (SB + other behaviour(s))]

Activities undertaken whilst sitting or lying down, such as screen-based activities

Objectively measured BMI or BMI-z

Yes (51)

Yes

SB interventions were associated with a very small and clinically irrelevant effect on BMI or BMI-z when applied to the general population or normal weight population.

By contrast, SB interventions to reduce BMI might be clinically effective for overweight/obese children.

Interventions targeted SB alone or combined with other behavioural components. Interventions appeared to be more successful when they were implemented with other behaviours (e.g. diet).

 Bautista-Castano et al. (2004)[44]

≤18

Jan 1993 - Dec 2003

4 (all RCTs) (1 SB alone, 1 SB + PA, 2 SB + PA + diet) (11.7y, 5-7y, 8.9y, 8-10y)

‘Sedentary activities’, such as watching TV

BMI, triceps skin-fold, WC, WHR

No

No

Decreasing ‘sedentary activity’, such as watching TV, positively influenced the effectiveness of interventions designed to prevent childhood obesity

RCT studies with the school as the unit of randomisation, intervention and analysis.

 De Mattia et al. (2007)[45]

Child-ren or adoles-cents (mean age 3.9 & 14.2 y)

1966 – Feb 2005

6 (all RCT’s) [1 (SB only), 2 (SB + PA), 1 (SB + diet), 2 (SB + PA + diet)] (Mean age: 10.4y; 10.0/10.2y; 14.2y; 3.9/4.0y; 8.9y; 9.5y/9.5y)

Recreational screen time

BMI, BMI-z, OW%, body composition (BIA, WHR, triceps skin-fold, DEXA scan)

No

Yes

SB interventions were associated with a modest improvement of weight parameters.

Controlled IN studies.

As the SB messages in these interventions are often combined with other health information (e.g. healthy eating and exercise), it is not possible to estimate the magnitude of the weight influences because of SB messages alone.

 Leung et al. (2012)[46]

6–19

1980 - Apr 2011

6 (1 SB only; 5 SB + Other)

Screen-based SB, “breaks from activity”, low EE activities (e.g. reading)

BMI, waist/hip circumference, BF%, skinfold thickness

No

No

Interventions targeting SB were effective at reducing SB and/or improving measurements related to weight status.

RCTs lasting ≥ 12 weeks; interventions aimed at reducing SB in school-aged children

 Liao et al. (2014)[47]

≤18

To July 2012

25 (study design not specified) [(5 SB only), (10 SB + PA), (10 SB + PA + diet)] Mean age [5 (<6 years), 15 (6–12 years), 5 (>12 years)]

Watching TV/DVD/VCR, playing sedentary video/ computer games and general sitting time

BMI

Yes (25)

Yes

Interventions seeking to decrease sedentary behaviours among children significantly reduced BMI when compared with control groups; mean BMI mean difference (g = −0.073, P = 0.021) at post-intervention.

Multi-component interventions (SB + PA or SB + PA + diet) were not more effective in reducing BMI than SB interventions alone.

Authors stated that although the mean BMI mean difference may not be considered clinically significant for the treatment of obese children, it may achieve public health significance in obesity prevention interventions among non-obese children.

 Luckner et al. (2012)[48]

≤18

To Nov 2008

9 (7 RCTs; 2 controlled non-randomized) [(1 SB only- TV); (8 SB-TV + other, e.g., PA)]. No age breakdown provided in the meta-analysis.

TV viewing and other (not specified)

BMI or BF% (by skin-fold, BIA or DEXA)

Yes (8)

Yes

In children (0–18 years), the highest reductions in mean BMI were achieved through promoting reduced television viewing [-0.27 kg/m2 (95% CI -0.4 to -0.13 kg/m2)]. The meta-analysis suggested that interventions which aimed to reduce TV viewing led to a significant reduction in BMI.

Interventions targeting weight status.

Studies conducted with general (i.e. non-obese) populations of children.

Only 1 of the 9 studies aimed solely at reducing TV viewing; others also incorporated PA and other components.

 Ramsey Buchanan et al (2016)[49]

All ages, inc. adults and ‘Child-ren’ mainly </=13y

1966 - June 2013

46 ‘behavioural interventions’ of’screen time’ and ‘screen time-plus’ studies (with children)

Recreational screen time

BMI, BMI-z, BF%

Yes (some calculations of effect sizes)

Yes

Reductions in BMI from screen-time-only interventions, mainly for ‘high intensity’ interventions. “Strong evidence that screen-time only interventions are effective at reducing recreational sedentary screen time … and improving or maintaining weight status”

IN’s primarily targeted recreational sedentary screen time. While review addressed all ages, most were with children.

 Stice et al. (2006)[50]

≤22

1980 - Oct. 2005

5 (all with “random assignment”) [(2 SB + Ed), (3 SB + PA + Ed)] (mean age 8.9–11.7years)

Sedentary behaviours, such as media (TV, video games) use

Mostly BMI, skinfold thickness

Yes (5)

No

Sedentary behavior reduction (as a moderator) was not associated with significantly larger effects.

This meta-analysis focused solely on effect sizes for weight gain prevention effects.

Inconsistency noted in the number of designated SB studies across various tables.

 Wahi et al. (2011)[51]

≤18

To Apr. 2011

9 [2 (≤6 years); 9 (>6 years)] [6 SB-only; 1 SB + PA; 4 SB + PA + Diet]

Screen time (hrs/week)

BMI

Yes (6)

Yes

Interventions to reduce screen time were not effective and mean changes in BMI (−0.10 (95% CI: −0.28 to 0.09) not significant (P = .32).

RCTs aimed at reducing screen-time in children.

 Wu (2016)[52]

≤18

To August 24, 2015

7 [<6 years = 2; 6–17 years = 5] [4 SB-only; 1 SB + PA; 2 SB + Diet]

Screen time (hours per week).

BMI

Yes (7)

Yes

Based on pooled analysis, including one with adults, interventions targeting screen time reduction had a significant effect on BMI reduction (-0.15 kg/m², P <0.001)

Notably, when looking specifically at 6–17 year olds, the mean BMI change was clearly not significant: -0.02 (95% CI: -0.18-0.15; P = 0.846).

Includes only RCT studies.

According to authors, subgroup analyses stratified by baseline age and presence of co-interventions suggested no significant differences.