Skip to main content

Table 1 The relationship between sedentary behaviour and adiposity

From: Systematic review of the relationships between sedentary behaviour and health indicators in the early years (0–4 years)

No. of participants (No. of studies) Design Quality assessment Absolute effect Quality
Risk of bias Inconsistency Indirectness Imprecision Other
The range of mean ages at time of exposure measurement was ~0.75 to 4.95 years; the oldest mean age at follow-up was 15.5 years. Data were collected by randomized trial, case-control, cross-sectionally, and up to 12 years of follow-up. Adiposity measures were: BMI (absolute, z-score, SD score, percentile); fat mass index, lean mass index, trunk fat mass index; % body fat (measured using DXA); skinfold ratio (triceps skinfold thickness to subscapular skinfold thickness); sum of skinfolds; waist-to-height ratio; waist-to-hip ratio; weight-for-height (z-score); weight-for-age (z-score); waist circumference (absolute, z-score for age); weight status (CDC, IOTF, or WHO cut-points; Flemish reference data; French reference standards; Rolland Cachera reference curves; United Kingdom reference standards in 1999); total fat mass (SD score); lean mass (SD score).
412 (1) Randomized triala Serious risk of biasb No serious inconsistency No serious indirectness No serious imprecision None Screen time c was significantly lower in the intervention vs control group at 2, 6, and 9 months post-interventiond. BMI z-scores were not different between the intervention and control groups at baseline or 9-month follow-up, but BMI z-scores increased in both groups [34]. Moderatee
32,699 (13) Longitudinalf Serious risk of biasg No serious inconsistency No serious indirectness No serious imprecision None Screen-based sedentary behaviours:
Computer (duration):
1/1 studies reported null associations [85]
Computer games (frequency):
1/1 studies reported null associations [82]
Screen time (duration):
2/3 studies reported unfavourable associations [33, 84]
1/3 studies reported null associations [87]
TV time (duration):
6/10 studies reported unfavourable associations [33, 54, 81, 83, 88, 90]
1/10 studies reported null associations [86]
3/10 studies reported mixed unfavourable and null associations [82, 85, 89]
Watching DVDs (duration):
1/1 studies reported unfavourable associations [83]
Other sedentary behaviours:
Time in baby seats (duration):
1/1 studies reported mixed unfavourable, null, and favourable associations [45]
Time in the car (duration):
2/2 studies reported null associations [81, 82]
Very lowh
1242 (2) Case-controli Serious risk of biasj No serious inconsistency No serious indirectness No serious imprecision None TV time [35, 36] and total sedentary time [36] were not different between children with overweight/obese (case group) or normal weight (control group) status, but watching TV for ≥ 1 h/day was unfavourably associated with having overweight status (OR = 1.71, 95% CI: 1.07, 2.75, p = 0.02) [35]. Very lowk
94,191 (47) Cross-sectionall Serious risk of biasm No serious inconsistency No serious indirectness No serious imprecision None Objectively measured sedentary time:
Sedentary time 30-min bouts (accelerometer derived):
1/1 studies reported null associations [40]
Total sedentary time (accelerometer-derived):
10/11 studies reported null associations [3740, 47, 53, 60, 75, 78, 80]
1/11 studies reported mixed unfavourable and null associations [77]
Screen-based sedentary behaviours:
Computer (duration):
3/4 studies reported null associations [63, 67, 79]
1/4 studies reported mixed unfavourable and null associations [71]
Screen time (duration):
6/18 studies reported unfavourable associations [32, 33, 46, 50, 59, 73]
10/18 studies reported null associations [44, 52, 57, 58, 62, 64, 65, 71, 72, 79]
2/18 studies reported mixed unfavourable and null associations [41, 61]
TV time (duration):
5/23 studies reported unfavourable associations [33, 55, 66, 67, 71]
11/23 studies reported null associations [31, 42, 43, 49, 50, 56, 60, 63, 69, 75, 76]
5/23 studies reported mixed unfavourable and null associations [48, 51, 54, 68, 79]
1/23 studies reported mixed null and favourable associations [74]
1/23 studies reported mixed unfavourable, null, and favourable associations [70]
Using the internet (duration):
1/1 studies reported null associations [69]
Video games (duration):
1/1 studies reported unfavourable associations [69]
Watching DVDs/videos (duration):
1/1 studies reported null associations [69]
Other sedentary behaviours:
Sedentary quiet play (duration):
1/1 studies reported mixed unfavourable and null associations [79]
Time in baby seats (duration):
1/1 studies reported null associations [45]
Using books (duration):
1/1 studies reported null associations [69]
Very lown
  1. BMI Body Mass Index, CDC Centers for Disease Control and Prevention, DXA dual-energy X-ray absorptiometry, IOTF International Obesity Task Force, SD standard deviation, WHO World Health Organization
  2. aIncludes 1 randomized controlled trial [34]
  3. bSerious risk of bias. Unclear if allocation was adequately concealed prior to group assignment; group allocation was adequately concealed from control, but not intervention group during the study; unclear if height and weight were directly measured or proxy-reported; baseline data were not reported, making it impossible to determine if baseline imbalances existed between groups [34]
  4. cScreen time was significantly lower in the intervention vs control group at 2 mo, 6 mo, and 9 mo follow-up post-intervention (mean ± SD: 2 mo: 39.48 ± 16.36 vs 86.64 ± 21.63 min/day; 6 mo: 24.72 ± 4.45 vs 84.95 ± 14.77 min/day; 9 mo: 21.15 ± 6.12 vs 93.96 ± 18.84 min/day; all p < 0.001)
  5. dIntervention: 3 printed materials and interactive CDs and one counselling call intended to decrease screen time; 8-week duration. Control: Usual care; unaware of counselling interventions
  6. eThe quality of evidence from the randomized trial was downgraded from “high” to “moderate” because of a serious risk of bias that diminished the level of confidence in the observed effects
  7. fIncludes 13 longitudinal studies [33, 45, 54, 8190] from 9 unique samples. Pagani et al. [90] and Fitzpatrick et al. [89] reported data from the Quebec Longitudinal Study of Child Development; Reilly et al. [81] and Leary et al. [82] reported data from the Avon Longitudinal Study of Parents and Children (ALSPAC); Gooze et al. [84] and Flores and Lin [83] reported data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B); and Fuller-Tyszkiewicz et al. [54] and Wheaton et al. [85] reported data from the Longitudinal Study of Australian Children (LSAC). Results are presented separately and participants are counted only once
  8. gSerious risk of bias. Questionable validity and reliability of the exposure measure [33, 45, 54, 8190]. Data were reported as missing, but amount and reasons were not provided [89]. Height and weight data were incomplete without explanation for 23% of the analyzed sample and 60.7% of the original cohort [81]. Possible selective reporting: differences between included and excluded participants were reported for confounding variables but not exposure variables without explanation [82]. BMI at age 3 yr was analyzed, but was not reported in the purpose or methods [88]. Did not account for potentially important confounding variables or mediating factors: sugar-sweetened beverage consumption and sleep were assessed but not accounted for [33]; diet was not measured or included in the analysis [45]; adjusted for physical activity [89]; of the potential child and family confounders that were assessed, potential confounders were included or omitted from analyses based on the authors’ determination of what was “likely to be linked to our predictor or outcome variables,” without providing a basis for that determination [89]. Data were pooled from the control and experimental groups of a messaging-based obesity prevention intervention study [33]
  9. hThe quality of evidence from the longitudinal studies was downgraded from “low” to “very low” because of a serious risk of bias that diminished the level of confidence in the observed effects
  10. iIncludes 2 case-control studies [35, 36]
  11. jSerious risk of bias. Questionable validity and reliability of the 1-day physical activity recall questionnaire [36]. Potentially inappropriate statistical analysis: investigators dichotomized participants by category of TV viewing of ≥1 h/day or <1 h/day based on exploratory bivariate analyses that showed 1 h to be the duration most related to children’s weight status [35]
  12. kThe quality of evidence from the case-control studies was downgraded from “low” to “very low” because of a serious risk of bias that diminished the level of confidence in the observed effects
  13. lIncludes 47 cross-sectional studies [3133, 3780] from 40 unique samples. Williams et al. [37], Byun et al. [39], and Byun et al. [38] reported data from the Children’s Activity and Movement in Preschool Study (CHAMPS); Sijtsma et al. [45] and Sijtsma et al. [46] reported data from the Groningen Expert Center for Kids with Obesity (GECKO) Drenthe birth cohort; Manios et al. [48], Kourlaba et al. [49], and van Stralen et al. [50] reported data from the Growth, Exercise and Nutrition Epidemiological Study in preSchoolers (GENESIS); Mendoza et al. [71] reported data from the National Health and Nutrition Examination Survey (NHANES) 1999 to 2002, Fulton et al. [72] from NHANES 1999 to 2006, and Twarog et al. [73] from NHANES 2008 to 2012; Taverno Ross et al. [76] and Espana-Romero et al. [77] reported data from the Study of Health and Activity in Preschool Environments (SHAPES); Brown et al. [55] and Fuller-Tyszkiewicz et al. [54] reported data from the Longitudinal Study of Australian Children (LSAC); Dolinsky et al. [53] and Boling Turer et al. [45] reported data from Kids and Adults Now: Defeat Obesity! (KAN-DO). Results are presented separately and participants are counted only once
  14. mSerious risk of bias. Potentially inappropriate sampling technique: participants were a non-representative convenience sample [66]; sampling deviated from protocol and specific deviations were not documented [57]. Potentially inappropriate measurement tools were used: questionable validity and reliability of the exposure measure [3133, 41, 4346, 4951, 5462, 6476] and outcome measure [65]; questionable validity of exposure measure [42, 52, 63, 79]; poor reliability of exposure measure [42]; height and weight were obtained by parent-report [44, 70]; options for 2–3 h and 4–5 h were missing from the Likert-type scale used to assess screen time [74]; applied accelerometry cut-points were not validated for the age group of interest [47]. Potential attrition bias: amount of unexplained missing exposure or outcome data is unknown [42, 50] or ranged from 14% to 67% [39, 40, 42, 43, 59, 60, 69, 71, 73, 74, 76], and reason may be related to the true outcome of interest [40, 43, 66, 71]. Potential selective reporting bias: statistics for non-significant relationships were not reported [48, 64]; authors decided post-hoc not to report analyses with continuous exposure variables [59]; only final model was reported [44]; results for correlations described in the methods section were not reported [62]; composite outcomes were presented without individual components; results for categorical screen time and total screen time described in the methods section were not reported [32]; outcomes from pooled hierarchical linear regression and variance information of included results were not reported [70]. Did not account for potentially important confounding variables or mediating factors: diet [43, 45, 46, 50, 58, 60, 63, 64, 67, 71, 72, 77, 80]; sugar-sweetened beverage consumption; and sleep [33]. Controlled for physical activity [59, 61, 66, 78]. Sleep during the day was considered sedentary time [40]
  15. nThe quality of evidence from the cross-sectional studies was downgraded from “low” to “very low” because of a serious risk of bias that diminished the level of confidence in the observed effects