Study Inclusion Criteria
The review sought to identify all studies that examined the relationship between sedentary behaviour and a specific health outcome in children and youth (aged 5-17 years). All study designs were eligible (e.g. cross sectional, retrospective, prospective, case control, randomized controlled trial (RCT), longitudinal). Longitudinal studies were included if the data presented in the article was consistent with the age limits that were set (i.e. if the study looked at participants at age 10 and then again at age 30, only baseline measurements from age 10 were used).
Studies were included only if there was a specific measure of sedentary behaviour. Eligible exposures of sedentary behaviours included those obtained via direct (e.g., measurements of sitting, or low activity measured by accelerometer) and self-reported (e.g., questionnaires asking about TV watching, video gaming, non-school computer use, and screen time - composite measures of TV, video games, computers) methods. Sedentary behaviour was often measured as a composite measure of all time engaging in sedentary behaviours including screen time outside of school hours. Six health indicators were chosen based on the literature, expert input, and a desire to have relevant measures from a range of holistic health indicators (i.e. not only physical health, but also emotional, mental and intellectual health). The six eligible indicators in this review were:
1. Body composition (overweight/obesity measured by body mass index (BMI), waist circumference, skin folds, bio-impedance analysis (BIA), dual-energy x-ray absorptiometry (DXA or DEXA));
2. Fitness (physical fitness, physical conditioning, musculoskeletal fitness, cardiovascular fitness);
3. Metabolic syndrome (MS) and cardiovascular disease (CVD) risk factors (unfavourable lipid levels, blood pressure, markers for insulin resistance or type 2 diabetes);
4. Self-esteem (self-concept, self-esteem, self efficacy);
5. Behavioural conduct/pro-social behaviour (child behaviour disorders, child development disorder, pro-social behaviour, behavioural conduct, aggression);
6. Academic achievement (school performance, grade-point average).
No Language or date limits were imposed in the search. The following definitions were used to help guide the systematic review [31
- Sedentary: A distinct class of behaviours (e.g. sitting, watching TV, playing video games) characterized by little physical movement and low energy expenditure (≤ 1.5 METs).
- Sedentarism: Engagement in sedentary behaviours characterized by minimal movement, low energy expenditure, and rest.
- Physically active:
Meeting established physical activity guidelines (e.g. see Tremblay et al. 2011 for Canadian Physical Activity Guidelines [16
- Physical inactivity: The absence of physical activity, usually reflected as the proportion of time not engaged in physical activity of a pre-determined intensity and therefore not meeting established physical activity guidelines.
Study Exclusion Criteria
As the volume of literature on sedentary behaviour was anticipated to be very high, to control the feasibility of this project, the following sample size limits were set a priori: population based studies (observational, cross sectional, cohort, and retrospective studies) were required to have a minimum sample size of 300 participants; RCTs, and intervention studies were required to have at least 30 participants. Studies of 'active gaming' (e.g., Nintendo Wii™, Microsoft Kinect™, Sony's Playstation Move™, video arcades, etc.) were excluded. Finally, studies that defined sedentary behaviour as 'failing to meet physical activity guidelines' were excluded from the review.
The following electronic bibliographic databases were searched using a comprehensive search strategy to identify relevant studies: Ovid MEDLINE(R) (1950 to February Week 2 2010), Ovid EMBASE (1980 to 2010 Week 07), and Ovid psycINFO (1806 to February Week 3 2010). The search strategy was created by a single researcher (JM) and run by a second researcher (AL). The search strategies can be found in Additional file 1
. The search was limited to studies looking at 'school-aged' children and youth (mean age of 5-17 years). Articles were extracted as text files from the OVID interface and imported in to Reference Manager Software (Thompson Reuters, San Francisco, CA). Duplicate articles were first removed using Reference Manager Software, and any remaining duplicates were removed manually. All articles were given a unique reference identification number in the database.
Titles and abstracts of potentially relevant articles were screened by two reviewers (AL and one of GG, MT, RC, RL or TS) and full text copies were obtained for all articles meeting initial screening by at least one reviewer. Two independent reviewers examined all full text articles (AL and one of GG, MT, RC, RL or TS) and any discrepancies were resolved by discussion and consensus between the two reviewers. If the reviewers were unable to reach consensus, a third reviewer was asked to look at the article in question. Consensus was obtained for all included articles.
Twelve key content experts were contacted and asked to identify the most influential papers from their personal libraries examining sedentary behaviour and health in the pediatric age group. Government documents from the U.S [6
], the U.K. [17
], and Australia [18
] were used for reference and to help guide the review process.
Standardized data extraction tables were created; data extraction was completed by one reviewer (AL) and checked by another (one of GG, RC, RL, or TS) for accuracy. Information was extracted regarding study characteristics (i.e. year, study design, country, number of participants, age), type of sedentary behaviour, measure of sedentary behaviour (i.e. direct, or indirect), and health outcome. Reviewers were not blinded to the authors or journals when extracting data.
Risk of bias assessment
The Downs and Black checklist was used to asses study quality [35
]. This 27 point checklist assesses the quality of reporting
(e.g. "Are the main findings of the study clearly described"); external validity
(e.g. "Were the subjects asked to participate representative of the entire population from which they were recruited"); internal validity
(e.g. "Were subjects randomized to intervention groups"); and power
(e.g. "Was there sufficient power such that the difference being due to chance is less than 5%"). The maximum score a study can receive is 32, with higher scores indicating better quality. Inter-rater reliability was calculated using Cohen's kappa.
Quality of evidence was determined by the study design and by Downs and Black score. Level of evidence was used to explain the quality of available studies and the confidence of the findings [36
]. RCTs were considered to have the highest level of evidence while anecdotal reports were considered to have the lowest evidence. See Table for more details. When possible, studies were examined for differences among age and gender subgroups.
Criteria for assigning level of evidence to a recommendation
A meta-analysis was performed with the data that were sufficiently homogeneous in terms of statistical, clinical, and methodological characteristics using Review Manager Software 5.0 (The Cochrane Collaboration, Copenhagen Denmark). Pooled estimates for the meta-analysis and their 95% confidence intervals were obtained using the random effects estimator of DerSimonian-Laird [37
]. Studies were weighted by the inverse of their variance. Cochrane's Q was used to test for heterogeneity among studies and the I2
(squared) index [10
] was used to determine the degree of heterogeneity [38
]. Funnel plots were used to assess publication bias (data not shown). Qualitative syntheses were conducted for remaining studies.