The associations between sedentary behaviour and mental health among adolescents: a systematic review

Background With technological developments and modernised sedentary lifestyles has come an increase in diseases associated with inactivity such as obesity and other non-communicable diseases. Emerging evidence suggests that time spent sedentary may also interact with mental health. This systematic review examined the associations between sedentary behaviour and mental health problems among adolescents. Methods This systematic review followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses, and applied a quality assessment tool for quantitative studies to identity best available evidence. Following stringent search strategy of the databases; Cumulative Index to Nursing and Allied Health Literature, Global Health, Health Source: Nursing and Academic Edition, MEDLINE, PsychARTICLES and PsycINFO, we identified 32 articles eligible for review. Results All studies reported leisure screen time among adolescents, and two thirds of identified studies examined depressive symptomatology. Other mental health measures were; anxiety symptoms, self-esteem, suicide ideation, loneliness, stress, and psychological distress. Strong consistent evidence was found for the relationship between both depressive symptomatology and psychological distress, and time spent using screens for leisure. Moderate evidence supported the relationship between low self-esteem and screen use. Poorer mental health status was found among adolescents using screen time more than 2–3 h per day, and gender differences exist. Essential information was missing for quality of evidence including heterogeneity in mental health and screen time-based measures, and self-report data collection methods. Conclusions The findings are of particular significance given the global public health concern of lifestyle-attributed diseases and the possibility for novel approaches to mental health. Future research should examine the psychological impact of reducing time spent using screens for leisure among adolescents, whilst accounting for possible confounding factors such as physical activity and dietary behaviours. It is critical that the reciprocal relationship between lifestyle behaviours and mental health is represented in both the psychiatric and public health forum. Electronic supplementary material The online version of this article (doi:10.1186/s12966-016-0432-4) contains supplementary material, which is available to authorized users.


Background
Lifestyle behaviours represent a broad public health concern with technological developments and modernised sedentary lifestyles has come an increase in diseases associated with inactivity such as obesity and other noncommunicable diseases (NCDs) [1,2]. In addition to physical health determinants, emerging evidence suggests that time spent sedentary may interact with emotional and mental health outcomes [3]. Sedentary behaviours are any waking behaviours with energy expenditure less than or equal to 1.5 metabolic equivalents while in a sitting or reclining posture [4]. Although not indicative of total daily time spent sedentary, screenbased activities for leisure are considered highly prevalent forms of sedentary behaviour [5]. Adolescence is a period of significant risk for the onset of mental disorders [6]. It is also a time during which independent lifestyle behaviours are developed which can significantly impact on immediate and long-term health [7,8].
There are several international observational studies examining the relationship between sedentary behaviour and mental health during adolescence. However, there is yet to be a cohesive review of the associations between screen time-based sedentary behaviours and mental health outcomes among adolescents specifically. In contrast there are several systematic reviews of the impact of screen time or sedentary behaviours on a broad range of physical health and educational indicators including; body composition, fitness, academic achievement, sleep problems, and musculoskeletal pain [9,10].
Two recent systematic reviews synthesised evidence on sedentary behaviour and health (including mental health) among young people [11,12]. One review [11] focused on objectively measured sedentary behaviour only and while this offered methodological rigour, the literature on associations with mental health is in its infancy and requires broad inclusion criteria such as studies reporting self-reported sedentary behaviour. One other review [12] was the first to comprehensively summarise associations between sedentary behaviour and mental health indicators among young people, however this study had some important limitations. The search strategy included literature on both children and adolescents, however adolescents have previously been identified as a 'specialist' group, with unique health risks that require separate targeted review methods [13]. In addition, this review failed to identify some key empirical research (e.g., [14][15][16][17][18]), and did not include experimental studies. Our review aims to extend on this previous work by identifying all relevant literature to date and by specifically examining adolescent groups.
A review published in 2011 examined physical activity and mental health among children and adolescents, and included a brief summary of observational research for sedentary behaviour and mental health [19]. Findings indicated a small, but consistent relationship between screen-time and mental health in young people, with authors predicting that this represents a growing body of research.
More recently, four systematic reviews of the literature have examined mental health and sedentary behaviours across age groups including adolescents, however outcomes of interest were limited to depression [20,21], anxiety [3], or self-harm and suicide [22].
We feel there is sufficient range of studies that merit synthesis. The objective of this systematic review is to synthesize all available evidence on the associations between different types of sedentary behaviour and mental health among adolescent populations. This overview of the research landscape could offer clearer understanding of the interrelated links between lifestyle behaviours and mental health among the adolescent age group, who are known to experience specific health vulnerabilities.
The research questions answered in this review were: 1. What is the current state of the evidence and magnitude of associations between different types of sedentary behaviour and mental health among adolescents? 2. What are the limitations of the current evidence base and what recommendations can be made for future research?

Methods
This study followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [23].

Inclusion/exclusion criteria
Inclusion criteria were designed to identify all previous studies examining the sedentary behaviour and mental disorders and symptomatology among adolescent populations. Adolescence was defined as 10 to 19 years of age [24]. Studies that included data on young people within adult/child populations were analysed and reported where appropriate. Studies of adolescents experiencing general good health were selected due to possible confounding effects of experiencing a chronic physical health condition. Importantly, the negative impact of symptomatology associated with such disorders, despite not necessarily amounting to clinical diagnosis, is known [25]. Therefore the outcome of interest in this review was extended to include both mental disorders and associated symptomatology. Sedentary behaviour was defined as activities that require little energy expenditure including sitting or lying down (not sleeping) while watching television or playing electronic games, reading, studying, writing or working at a desk or computer (not for schoolwork). This included the use of electronic media for entertainment/ leisure such as television, electronic gaming, and computer use, and is henceforth referred to as 'leisure screen time'. Studies that examined Internet or gaming addiction or other diagnosed sedentary behaviour disorders were not eligible for review. This was due to the complexity of such conditions in the context of mental health, and the focus of this review on habitual sedentary behaviours in typical adolescent populations.
Studies were included in the review if they were; (1) peer-reviewed primary research, (2) reported data on adolescents, (3) included mental health measure/s (diagnostic or symptomatic), (4) included sedentary behaviour or screen time measure/s including TV, gaming, computer/ internet use, (5) reported cross-sectional or longitudinal associations, or an intervention study on typically developing adolescents, and, (6) were published from the start of the selected database through to January 2016. Studies were excluded if they were; (1) treatment and management studies, (2) studies examining child or adult populations, (3) studies focused on Internet or gaming addiction, or, (4) studies of specific populations (such as those with a chronic physical condition).

Search strategy
All articles were sourced from databases accessed through EBSCOhost. Databases searched were; Cumulative Index to Nursing and Allied Health Literature (CINAHL), Global Health, Health Source: Nursing and Academic Edition, MEDLINE, PsychARTICLES and Psy-cINFO. Cochrane Database was searched to identify all possible articles eligible for review. Search terms were selected based on the eligibility criteria and outcomes of interest as previously described. The full search strategy and search terms are reported in Fig. 1.

Data extraction and synthesis
Two members of the research team [EH, KM] screened the titles and abstracts of all articles identified via the search. Full-text articles deemed eligible for review were sourced and read in full to determine final eligibility. A standardised tool for data extraction was created including study characteristics (author, year, country), participants, study design (longitudinal and intervention studies only), mental health and screen time measures used, and main findings. Findings were categorised into cross-sectional, longitudinal and intervention study designs, and data synthesized by sedentary behaviour measure and mental health outcome examined.
A total of 535 articles were identified through the search strategy. A further 29 were identified through Cochrane database searching, and 10 through reference list checks of previous systematic reviews. After removing duplicates, all titles and abstracts were screened (n = 551) and 72 studies were identified as potentially eligible for the review. Full-text of these 72 articles were sourced and read in full to determine final eligibility. Based on the full-text review, a further 40 articles were excluded; 14 focused specifically on internet content and associated behaviour (such as problem gambling) as opposed to sedentary behaviour specifically, eight examined children or adult populations, seven did not include an appropriate mental health measure, six focused on atypical or specific adolescent groups such as those experiencing a chronic health condition, three did not include an appropriate sedentary behaviour or screen time measure and two compared sedentary behavioural and mental health differences across different groups as opposed to the relationship between the two variables specifically. A total of 32 articles met the eligibility criteria and were subsequently included in this review. A flow chart of study selection process is reported in Fig. 2.

Quality of evidence
Methodological quality of the included studies was determined by a modified version of a quality assessment tool for quantitative studies [26] as previously used to evaluate evidence for sedentary behaviour and mental health [3]. The tool comprises eight components including selection bias, study design, confounders, blinding, data collection methods, withdrawals and dropouts, intervention integrity, and analyses. Components relevant only to intervention study designs (blinding, intervention integrity) were not applied to cross-sectional and longitudinal studies. Each component was applied to the included studies and graded as weak, moderate or strong, leading to an overall study rating of weak (if two or more components scored weak), moderate (if less than three components were strong with no more than one weak score), or strong (if three or more components were scored strong).
Best evidence synthesis was conducted to draw conclusions for each mental health domain based on the methodological quality of studies. Based on previous systematic reviews examining sedentary behaviour and health outcomes [3,27,28], strong evidence was indicated by consistent results in 2 or more strong studies. Consistent was defined as a stable relationship (in the same direction) that was reported in at least 75 % of studies examining that particular outcome. Moderate evidence was consistent results in one high quality study and at least one weak quality study or consistent results in two or less quality studies. Insufficient evidence was defined as only one available study or inconsistent results in two or more studies.

Summary of included studies
Twenty-four of the 32 studies reviewed were crosssectional in design (Table 1)  , six were longitudinal (Table 2) [14][15][16][17][18]53] and one was an intervention study design (Table 2) [54]. One study [55] reported both cross-sectional and longitudinal findings. An overview of the type of sedentary behaviour and mental    Self-esteem: assessed using six items adapted from the Rosenberg Self-Esteem scale, higher scores indicated lower self-esteem.
In females, higher screen time was associated with low self-esteem Strong β standardised beta coefficient; B unstandardized beta coefficient; CI confidence interval; F analysis of variance; OR odds ratio; R correlation coefficient; SD standard deviation; SE standard error; y years a Quality of evidence based on assessment tool for quantitative studies [26] including selection bias, study design, confounders, blinding, data collection methods, withdrawals and drop outs, intervention integrity, and analysis. Strong quality of evidence = if three or more components scored strong. Moderate quality of evidence = if less than three components were strong with no more than one weak score. Weak quality of evidence = if two or more components scored weak  Quality of evidence based on assessment tool for quantitative studies [26] including selection bias, study design, confounders, blinding, data collection methods, withdrawals and drop outs, intervention integrity, and analysis. Strong quality of evidence = if three or more components scored strong. Moderate quality of evidence = if less than three components were strong with no more than one weak score. Weak quality of evidence = if two or more components scored weak health outcomes that each study examined is reported in Table 3. Details pertaining to study aim, sample characteristics, mental health and sedentary behaviour measures and findings are described in Tables 1 and  2 and will not be repeated in the sections below. Results are arranged by study design, and within sections specific findings are stratified by mental health measure and the sedentary behaviour reported in individual studies. All studies examined typically developing adolescents aged 10-19 years, except four studies that included young people aged up to 20-24 years [17,34,36,52]. These studies were accepted for review as a large proportion of the samples in these studies met the 10-19 year age requirement. Two thirds (21/32) of studies eligible for review examined depression or depressive symptomatology. Other mental health outcomes were; anxiety, self-esteem, suicide ideation, loneliness, stress, and psychological distress. Results are stratified firstly by mental health measure, then by type of sedentary behaviour measured.

Total daily sedentary behaviour/screen-time use
When sedentary behaviour was examined as average hours spent using screens for leisure per day, four studies identified a significant association between adolescents reporting higher daily average screen time and heightened depressive scores [31,32,36,43]. This finding was shown to be independent of a broad range of potential confounders including; age, gender, relative wealth (socio-economic status, parent's level of education, geographic location), physical activity levels, weight status, and fruit, vegetable and soft drink consumption.
Studies that treated screen time as a dichotomous measure (whether adolescents met or did not meet an identified daily cut-off ) reported higher levels of depressive symptomatology among those exceeding compared to those who met the cut-off (two hours or less per day in all studies) [39,42,50]. One study reported this relationship being independent of self-reported physical activity levels [50].
Objectively measured sedentary behaviour (total time spent sedentary measured by accelerometry data) was not associated with depressive symptoms in one study [55]. Another study found a null relationship in a sedentary behaviour variable of combined time talking on the phone, watching TV and using the internet and depressive symptomatology [37].

Computer/internet use
One further study identified gender specific findings with females who experienced depressive symptomatology were more likely to use the Internet for three or more hours per day compared to those with mild or no depressive symptomatology [56] and the same relationship was not significant amongst adolescent males in this study.
Low internet use (up to two hours per day) was significantly associated to lower depressive symptoms compared to non-users in one study [30], showing reverse trends once reaching two hours per day, and significantly heightened depressive symptoms at three or more hours of Internet use per day. Another study also reported an increased risk of depressive symptoms with increased Internet use [48] and one other study found a significant correlation between time spent on social networking sites and depressive scores [47]. Durkin and Barber (2002) reported that adolescents who engaged in low computer game use experienced lower depressed mood than both high users of computer games and those who did not use computer games at all [35].

TV viewing
One study found that females who experienced depressive symptoms self-reported an average 168 min more TV viewing per week than those without depressive symptoms, and this difference was significant [55]. Arat (2015) found no significant association between TV viewing and depressive symtpomatology among a group of American adolescents [29].

Video gaming
Casiano et al. (2012) examined the relationship between media use in the preceding 3 months and found that depression was less common among frequent video game users compared to those who rarely engaged in gaming, before and after controlling for household income and gender [33].

Total daily sedentary behaviour/screen-time use
After controlling for potential confounders (gender, relative wealth, weight status, and various diet and activity-related behaviours) two studies [32,43] identified significantly greater severity of anxiety symptoms among adolescents who reported higher daily average hours of screen time, compared to those reporting lower daily estimates. The third study found no significant relationship [37].

Computer/internet use
No studies examined the relationship between anxiety symptoms and computer or internet use.

TV viewing
No studies examined the relationship between anxiety symptoms and TV viewing.

Video gaming
No studies examined the relationship between anxiety symptoms and video gaming.

Total daily sedentary behaviour/screen-time use
Total time spent using screens for leisure was significantly inversely associated with self-esteem amongst a group of adolescent females, after controlling for age, socio-economic status, BMI, and objectively measured physical activity [46]. An examination of a dichotomous measure of screen time found that adolescents who exceeded an average of two hours per day experienced lower self-esteem than those who averaged two hours or less, and this finding was replicated when males and females were examined separately [50]. This study also controlled for the impact of self-reported physical activity.
Computer/internet use Durkin and Barber (2002) found that low computer users reported higher levels of self-esteem than nonusers, however neither group differed significantly in self-esteem from the high computer game users [35].
One study which categorised time spent online by type of activity (communication, entertainment and information-related activities) found no significant associations with self-esteem [34].

TV viewing
No studies examined the relationship between selfesteem and TV viewing.

Video gaming
Jackson et al. (2010) reported lower self-esteem amongst frequent video game users compared to less frequent gamers [40].

Total daily sedentary behaviour/screen-time use
No significant relationship was detected between screen time and suicide ideation in two studies [33,36]. [45] found that after accounting for age, gender, smoking and self-reported 2-week sadness, adolescents reporting five hours or more of video game/ internet use per day were more likely to have experienced suicide ideation compared to adolescents reporting no video game or Internet use. An examination of Internet use among a large (n = 136,589) South Korean adolescent sample [30] revealed lower odds of suicide ideation in those using less than two hours per day of Internet use, and those reporting more than two hours experienced an increased likelihood of suicide ideation.

TV viewing
Arat et al. (2015) [29] focused on specific ethnicityrelated findings, reporting lower odds of suicide ideation among African adolescents who spent longer hours watching television, compared to those who spent less time watching television.

Video gaming
No studies examined the relationship between suicide ideation and video gaming.

Total daily sedentary behaviour/screen-time use
Two studies examining Canadian adolescents reported that exceeding 2 h per day average screen time was significantly associated with higher odds of psychological distress compared to two hours or less per day of average screen time [50,52]. One of these studies [50] demonstrated this relationship to be independent of physical activity levels, and that the association was most pronounced among males. One study found no significant associations between loneliness and total daily average time spent talking on the phone, watching TV and using the Internet [37]. Total amount of time spent using screen time per day was positively associated with perceived stress [36]. Self-rated mental health was negatively associated with total daily screen time in that adolescents reporting more than two hours per day were 30-50 % more likely to rate their mental health sub-optimally (good/fair/ poor) [38]. Parent reported mental health status indicated similar trends, with adolescents who spent more than two hours of screen time each day experiencing poorer mental health status compared to those using less than two hours per day [49]. This study controlled for self-reported physical activity levels.

Computer/internet use
Mathers et a. (2009) revealed that adolescents who reported low or high computer use experienced lower levels of psychological distress compared to those who reported no computer use. One study examining loneliness reported no significant associations with time spent online (categorised by time spent communicating, entertainment purposes, or information-related activities) [34].
Self-reported sadness was less frequent among adolescents reporting one hour or less of daily game/Internet use compared to those reporting no use at all [45]. In addition, adolescents who reported 5 h or more of video game/Internet use per day were more likely to experience sadness compared to those reporting no use [45].

TV viewing
No studies exmained the relationship between TV viewing and other mental health indicators.

Video gaming
Among Australian adolescents, after adjusting for gender, age, socio-economic status and BMI z-score, those who reported a high level of video game use were more likely to report high/very high levels of psychological distress compared to those who did not play games [44].

Depressive symptomatology
Of the reviewed longitudinal studies, five examined the association between depression or depressive symptoms and sedentary behaviour [14,16,17,53,55].
Total sedentary behaviour/screen time use Sund et al. (2011) found that time spent on sedentary activities predicted higher scores of depressive symptoms among male adolescents only [53].

Computer/internet use
After controlling for past symptoms and media use, Romer et al. (2013) found that recent depression was associated with greater Internet use and video game playing in a cohort of American adolescents [17]. Bickham et al. (2015) demonstrated that, examined individually, higher TV use and mobile phone use reported at baseline significantly predicted higher participant depression scores at 1-year follow-up, after controlling for baseline depressive scores [14]. This relationship remained significant following adjustments for gender, ethnicity and parental education.

TV viewing
Similar predictive findings were reported by Primack et al. (2009) in that each additional hour of daily TV use reported at baseline and overall total media exposure, increased the odds of developing depression at 7-year follow-up [16]. A gender interaction was found, suggesting that females are less likely than males to develop depression based on the same total media exposure. Hume et al. (2011) reported no significant longitudinal relationships between depressive symptoms, objectively measured sedentary behaviour and TV viewing, in a community-based sample of Australian adolescents.

Video gaming
No longitudinal studies examined the relationship between video gaming and depressive symptomatology.

Total sedentary behaviour/screen time use
When adolescents were clustered into activity level which included a sedentary group, being categorised in the most sedentary group at baseline predicted lower self-esteem at 1-year follow-up, compared to those categorised as most active [15].

Computer/internet use
No longtudinal studies examined the relationship between computer/internet use and self-esteem.

Video gaming
In one other study [18] self-esteem at baseline was negatively associated with time spent using video games over a 3-year period, and positively associated with computer use over the same period.

Intervention findings
One study [54] examined the impact of an intervention on sedentary behaviours and mental health among adolescents.

Total sedentary behaviour/screen time use
The intervention program was set in an Australian secondary school and ran for 20-weeks. The aim of the intervention was to target health behaviours of adolescent boys at risk of obesity, with the specific focus on avoiding screen-time. Intervention activities pertaining to sedentary behaviour and mental health specifically were parental newsletter focused on recreational screen time and interactive seminars addressing key behavioural messages. One intervention component was a smart phone application and website for the project. Eightmonth follow-up in 2012 found that the intervention had a positive effect on daily screen time, and that reduced screen time was significantly related to improved psychological well-being. Psychological well-being was defined using the Diener et al. (2010) Flourishing Scale [57]. Results were significant after controlling for the school that the participant attended, baseline values, and objectively measured physical activity levels.

Quality of evidence
Overall methodological quality ratings are reported in Tables 1 and 2 with study characteristics. The outcomes of individual components assessed using the quality assessment tool are provided in Additional file 1. Overall, 12 studies demonstrated strong methodological quality, 17 were rated as moderate and 3 studies demonstrated weak methodology. All reviewed studies demonstrated some methodological weaknesses. There were a large number (23/32) failed to report validity and reliability of measures of both mental health and sedentary behaviour, close to two-thirds (21/32) did not report randomised sampling, and many studies were missing key information such as participation and retention rates. Best evidence synthesis was conducted to determine relative strength of each mental health domain and is summarised here.
Based on consistent findings of seven strong [30-32, 42, 48, 50, 55], seven moderate [14,16,17,39,41,43,51] and two weak [35,36] studies, the evidence for the positive association between depressive symptoms and sedentary behaviour among adolescents was considered strong. Of the three studies that examined anxiety, one strong [32] and one moderate [43] quality study presented a positive association between symptoms and sedentary behaviour. Due to the low number of studies examining this relationship (n = 3), and the requirement of consistency to be determined by 75 % of studies, this evidence was considered insufficient. Based on one strong [50] and four moderate [15,18,40,46] studies demonstrating an association between low self-esteem and sedentary behaviour, this evidence was considered moderate. Due to only three out of five findings (60 %) examining suicide ideation and sedentary behaviour reported significant positive associations, the evidence for this relationship was considered insufficient. The evidence for a positive association between psychological distress and sedentary behaviour was indicated by two strong studies [44,50,52] and one moderate study, and the evidence for this relationship was rated strong. The evidence was considered insufficient for the mental health outcomes of loneliness, stress, mental well-being and sadness, due to the low study numbers examining associations with sedentary behaviour.

Principle findings
While a large body of cross-sectional research was identified for review, there is a lack of longitudinal studies examining sedentary behaviour and mental health, and only one intervention was included in the review.
Directionality was not possible to determine due to the large cross-sectional evidence base. Most commonly examined mental health indicators were; depressive symptomatology, anxiety symptoms, self-esteem, suicide ideation, and psychological distress, while a few studies examined loneliness, stress, mental well-being and sadness. Whilst the eligibility criteria allowed for a broad range of sedentary behaviours, the most frequently examined behaviour in this age group is total screen time for leisure/entertainment purpose, and computer/internet use and TV viewing in particular were most frequently examined.
Findings of this review indicate strong evidence for the positive relationship between depressive symptomatology and screen time for leisure among adolescents, based on a mix of high quality cross-sectional and longitudinal studies. There was some evidence to suggest that low levels of screen time for leisure was associated with lower levels of depressed mood, with adverse findings only appearing at more than two to three hours per day of average of screen time [30,35,51].
Although few studies reported psychological distress and sedentary behaviour (n = 3), the best evidence synthesis suggests that these studies demonstrate strong evidence for a relationship between high levels of screen time (total screen time, video and computer use) and high levels of psychological distress.
Moderate evidence was found for the relationship between self-esteem and sedentary behaviour, indicating lower levels of self-esteem amongst those who reported higher levels of screen time (total screen time, TV viewing, and computer use).

Potential mechanisms
While the evidence examining lifestyle behaviours and mental health is growing, there is a notable gap exploring and revealing mechanisms underlying such associations [20]. Studies have typically examined associations, and then discussed possible explanations for findings as opposed to being driven by clear theoretical concepts from the outset. One possible explanation may be the complex range of mediators driving lifestyle behaviours, and the difficulties in extrapolating associated impact.
Although evidence of the independent effect of sedentary behaviour on health is emerging, it has been assumed that time spent sedentary is linked to physical activity [58]. Physical inactivity has been shown to interact with mental health [19,59,60], and exercise has been shown to have a positive effect in mental health treatment studies [61][62][63]. It is possible that the beneficial pathophysiological, social and general health effects of being active may be omitted when sedentary, which may have a negative impact on mental health. Adolescents who experience poorer mental health may lack motivation to be physically active and may turn to screen based activities requiring little effort as a coping mechanism, and therefore lose such protective effects of physical activity.
Although the relationship between mental health and sedentary behaviour remained significant after controlling for physical activity in six of the reviewed studies here [31,43,46,49,50,54], only two used objectively measured physical activity [46,54], and this relationship requires further investigation. Previous findings have indicated that the relationship between sedentary behaviour and physical activity among children and adolescents is negative, but small [64]. This suggests these behaviours do not directly displace one another, and support the examination of sedentary behaviour as a distinct behaviour. In addition, an adolescent reporting high levels of physical activity might be experiencing poor mental health alongside high levels of sedentary behaviour. The issue of whether physical activity moderates the assoiciation between sedentary behaviour and mental health is beyond the scope of this systematic review, but forms an important avenue for future research.
As with physical activity, there are known complex relationships between sedentary behaviour, overweight/ obesity and other associated obesogenic risk behaviours, and this may have a significant negative impact upon adolescent mental health [20]. The relationship between unfavourable body composition [9], weight status [10] and increased levels of sedentary behaviour has been demonstrated among adolescents. Importantly, overweight and obese young people are known to experience particular mental health vulnerabilities including; weight-based teasing and bullying, stigmatisation, poor body image, and self-esteem, and this in turn is expected to impact on potential sedentary behaviour and mental health associations [65]. Weight status/body mass index was included as a covariate in nine of the reviewed studies [30,32,38,39,43,44,46,50,53]. It is reccomended that future research includes such weight-based measures as covariates in asessing the relationship between sedentary behaviour and mental health.
There is evidence supporting the role and impact of diet on mental health during adolescence [66]. The relationship between dietary and sedentary behaviours among this age group has also been previously reported [67]. Recently, an updated comprehensive systematic review [68] revealed that among young people, higher levels of sedentary behaviour (TV viewing in particular) is associated with less healthy dietary behaviours such as low fruit and vegetable consumption, and increased consumption of energy dense foods and sugar sweetened beverages. With known impact of nutritional deficiencies on mood [69] and the increased appetite for sweeter and processed foods when experiencing poorer mental health [70], it is possible that diet is implicated in the relationship between leisure screen time and mental health. Diet-related measures were included in four of the reviewed studies [32,36,39,49], and as with weightbased measures, it is recommended that future research include diet as a covairate in assesssing the relationship between sedentary behaviour and mental health.
While outside the scope of this review, the type of activity being undertaken during screen time for leisure may have important implications in the context of adolescent mental health. It has been suggested that increasing time spent sedentary using screens may represent a population increased dependency on social media platforms for connectedness and interaction with peers [71]. Such experiences may hold protective benefits for mental health and an examination beyond time spent sedentary is needed to identify this potential mechanism. Similarly, it has been suggested that the Internet may offer increased access to health information, including mental health support [72,73], and this relationship requires further investigation. The type of sedentary activity may be appropriate to examine in relation to mental health compared to objectively measured duration spent sedentary. In addition, screen use does not nessesarily equate to sedentary behaviour in metabolic terms and this further supports comprehensive examination of specific experiences in the context of mental health.

Methodological considerations
In interpreting these findings, it is important to acknowledge that few longitudinal and intervention studies were eligible for review compared to a large body of crosssectional research, and this limits conclusions on causality. Study design aside, all reviewed studies were missing some essential information required for methodological quality.
Heterogeneity existed in mental health and sedentary behaviour measures, with studies often failing to report validity or reliability estimates for the respective measure of sedentary behaviour. An indicator for screen time has often been derived from participant self-reported response to very few or single items, thus decreasing methodological quality of evidence to date. In addition, there is some debate surrounding the negative health impact of types of sedentary behaviours, and although this is currently being considered and tackled [74], it remains unclear across these pooled studies. This impacts on the specificity of the exposure (based on types of behaviours examined) and consequential evidence synthesis.
Both continuous and categorical measures of mental health have been used thus limiting comparisons across outcomes. Some studies utilised the same tool however employed different cut-off levels and therefore differences exist in the interpretation of mental health outcomes. Although validity and reliability measures were frequently reported for mental health symptomatology, no study examined objective clinical diagnosed mental disorders. Important differences in statistical analysis limited a comparison of findings, as did the varying adjustments for potential confounders.
A final methodological consideration was that although the intervention study design included in this review aimed to reduce leisure screen time, one intervention component included a smart phone application and a website dedicated to the intervention. Although the purpose of this component was to monitor physical activity including recording of fitness challenge results and goal setting for physical activity and screen-time, such initiatives may impact on the sedentary behaviour and mental health relationship under examination.

Strengths and weaknesses of this review
This review employed stringent systematic review methodology in line with the PRISMA guidelines [23] to ensure all relevant literature to date was identified and evaluated with the best possible scientific rigour. In addition, the conclusions drawn by this review are strengthened by the use of the quality assessment tool [26] and best evidence synthesis consistent with previous literature examining sedentary behaviour and health outcomes [3,27,28].
A limitation of this review was that in order to identify all relevant evidence to date, no exclusion criteria were set on size of samples examined in cross-sectional research. Such methods are advised to ensure studies examined represent those large enough and with sufficient power to detect significant change. This limitation was accepted as necessary due to sedentary behaviour and mental health research in this age group being in its infancy. In addition, our review was not able to infer on the direction and magnitude of the associations between sedentary behaviour and mental health problems among adolescents, due to a lack of longitudinal research in this area.
Although the rigorous search strategy aimed to identify all relevant literature, mental health outcomes emerged from the search strategy a posteriori. Various symptoms are associated with mental disorders and it is plausible that some literature may have been overlooked. In addition, literature demonstrating the relationship between sedentary behaviour and physiological responses that are also associated with stress and mood responses (such as hypothalamic-pituitary-adrenal axis activity [75]) may have also been overlooked. Due to heterogeneity of the outcome measures a meta-analysis was not possible in this review.

Conclusion
Although limited by varying quality of the methodological design of the included studies, this systematic review suggests there is emerging evidence to link increased time spent using screens for leisure, with poorer mental health experiences among adolescents. Specifically, strong consistent evidence suggests higher levels of screen time for leisure or entertainment is associated with heightened depressive symptomatology and psychological distress among this age group. Methodological considerations include heterogeneity in sedentary behaviour and mental health measures, and few longitudinal and intervention study designs. Future research should examine the psychological impact of reducing time spent using screens for leisure among adolescents, whilst accounting for possible confounding factors such as physical activity and dietary behaviours. The challenge remains to tease out the complex interaction between types of sedentary behaviours, i.e., what is being performed while sedentary (e.g., Video game v reading), their physiological and cognitive impacts, and pathways to mental health. The opportunities for sedentary behaviours have increased rapidly and older studies may have been confined to few sedentary behaviours, e.g., TV viewing, however the ability of adolescents to utilize these greater opportunities at younger ages may contribute more exposure to impacts on mental health. Further analysis of time use data may offer more depth to any secular shifts of type and exposure.
The current findings are of particular significance with increasing burden of poor mental health among adolescents, and growing understanding of the negative health impacts of sedentary lifestyles. It is critical that the reciprocal relationship between mental health and lifestyle behaviour is represented in the public health discussion.

Additional file
Additional file 1: Methodological quality assessment checklist (XLSX 15 kb)