Screen Time Associated with Health Behaviors and Outcomes in Adolescents

Vincent Busch, MSc; Lieke Ananda Manders, MSc; Johannes Rob Josephus de Leeuw, PhD

Objectives: To study the associations problems and being overweight. Conclu- of screen time (Internet / video games / sions: Screen time was of significant im- television) with health-related behaviors portance to adolescent health. Behavioral and outcomes in adolescents. Methods: interrelatedness caused significant con- Regression analyses were performed to founding in the studied relations when assess the associations of screen time behaviors were analyzed separately com- with several health-related behaviors and pared to a multi-behavioral approach, outcomes in 2425 Dutch adolescents. Re- which speaks for more multi-behavioral sults: Screen time was associated with analyses in future studies. bullying, being bullied, less physical ac- Key words: screen time, health behav- tivity, skipping school, alcohol use and ior, overweight, psychosocial problems, unhealthy eating habits. Compulsive and adolescent excessive screen times were associated Am J Health Behav. 2013;37(6):819-830 respectively with several psychosocial DOI: http://dx.doi.org/10.5993/AJHB.37.6.11

ith the advance of technology, time spent conduct disorder.10-12 In particular, self-efficacy is on television, Internet and video games is an aspect receiving increasing attention among in- Wincreasing among today’s youth.1-3 Televi- terventions in the field of adolescent health promo- sion viewing, Internet use and video game play- tion, because it is believed to be a mediating vari- ing are collectively called ‘Screen Time’. As early able in the causal path of unhealthy behavior and as 1983, it was claimed video game playing could psychosocial problems in adolescents.13,14 become an addiction like any other behavioral ad- Furthermore, some literature suggests that un- diction and the same was argued for excessive In- healthy behaviors are associated with and influ- ternet use several years later.4,5 Although a formal encing each other instead of existing independent- medical diagnosis for video game addiction or In- ly. This raises interest for the probable associa- ternet addiction is (still) lacking in current medical tions of screen time behaviors in relation to each practice, discussions are on-going to add them to other and their possible associations with other the future Diagnostic and Statistical Manual of Mental unhealthy behaviors and health outcomes, which Disorders (DSM).2,5 In general, unhealthy screen has relevance for future health promotion inter- time behavior is characterized by 2 aspects: (1) ventions.15-19 whether or not one spends an excessive amount of This study investigates how a range of known un- time on it; and (2) whether or not the behavior is healthy behaviors and health outcomes are associ- considered “compulsive.” ated with several, relatively ‘new’ unhealthy screen The evidence increases that excessive and/or time behaviors, in a sample of Dutch high school compulsive screen time behavior holds the poten- students. In this study these unhealthy behaviors tial to be harmful to one’s health;3,6,7 recent studies consist of marijuana use, alcohol use, smoking, un- support for this belief, eg, in relation to obesity8,9 safe sex, skipping school, bullying, poor nutritional and several psychosocial and psychiatric prob- behavior and less physical exercise, in accordance lems, such as depression, lower self-efficacy and with the Health Behavior in School-aged Children study (HBSC).20 The health outcomes consist of students’ psychosocial problems, being overweight, Vincent Busch, Doctoral Student, Julius Center for Health Sci- and General Self-Efficacy (GSE).21 To demonstrate ences and Primary Care, University Medical Center Utrecht, the confounding effects of the interrelatedness of Utrecht, the Netherlands. Lieke A. Manders, University Me- the screen time behaviors in their relations with dical Center Utrecht, Utrecht, the Netherlands. Rob J.J. de other unhealthy behaviors and health outcomes, Leeuw, Senior Researcher, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, these associations are presented with and without the Netherlands. correcting for (possible) confounding by the remain- Correspondence Vincent Busch; [email protected] ing screen time behavior variables.

Am J Health Behav.™ 2013;37(6):819-830 819 Screen Time Associated with Health Behaviors and Outcomes in Adolescents

Thus, we quantify the associations of different the Internet or playing video games as well as the screen time behaviors with a range of unhealthy compulsiveness of these behaviors was measured. behaviors as well as the noted health outcomes, Spending more than 2 hours/day on screen time while preventing the introduction of bias due to behaviors was defined as “excessive” use, in accor- the screen time behaviors’ interrelatedness and dance with current standards in the literature.6,23 demonstrating the importance of multi-behavioral The compulsiveness of someone’s screen time be- analyses in adolescent health behavior research. havior was measured by the Compulsive Internet Use Scale (CIUS) for compulsive Internet use24 and METHODS by the Videogame Addiction Test (VAT) for compul- Samples sive video game playing.25 Both the VAT and CIUS Data were collected from 5 Dutch high schools consist of 14 questions with a 5-point Likert scale, as part of the Utrecht Healthy School (UHS) study used to evaluate compulsive behavior, respectively (N = 2425),22 the UHS pilot study school and its 4 for compulsive video game playing (CVP) and com- sister schools from the UHS itself. These schools pulsive Internet use (CIU). A score higher than 3.0 were part of a convenience sample of schools that points indicates compulsive behavior. These bi- were recruited for the UHS. All 5 schools partici- nary measures for compulsive and excessive use pated out of intrinsic motivation and were not were used in all presented analyses, because these provided with funds or other incentives to partici- are considered indicators of “problem behavior.” pate. All were all assisted with the questionnaire Definitions of the unhealthy behaviors and procedures by the research team. All are situated student demographics. Questionnaire items in suburban areas of middle-to-large cities in the and operationalization were largely similar to the Netherlands. Therefore, all 5 schools should be Dutch HBSC questionnaire and covered several categorized as in-between rural and urban. health outcomes and a range of different health behaviors and socio-demographics20 “Recent be- Survey Procedures havior” was defined as behavior in the month prior The UHS questionnaire was completed indepen- to completing the questionnaire. Questions on re- dently by participants in classroom settings at the cent behavior were asked with regard to alcohol start of the school year in September. Survey pro- and marijuana use, smoking, bullying and unsafe cedures allowed students to participate voluntarily sex. These measures dichotomous, ie, (0 = did not and anonymously. Prior to the survey, students recently exert a behavior, 1 = did exert a behavior were informed of the questionnaire’s purpose and recently). These questions on unhealthy behav- content by means of a newsletter. These points iors were posed as in the HBSC, which means that were repeated at the time of the survey by a mes- binge drinking was regarded as more than 5 alco- sage presented on the questionnaire and by the holic beverages on a single occasion, and skipping classroom teachers. The only students not com- school as >3 hours of disallowed absence from pleting the survey were ones not present at the time school in the recent month.20 of the survey (due to conflicting course schedules, Also measured, but not a standard part of the according to their teachers) or ones absent on the international HBSC survey, were questions re- day of the survey; the surveys were unannounced, garding bullying and being bullied, based on the so this should have not been a source of bias. Tak- Olweus Bully Score and the Olweus Bully Victim ing into account these 2 reasons for being absent, Score.26 These scores distinguish bullies and bul- the response rate was over 95%. Data cleaning was lied children from non-bullies and non-bullied performed in such a way that when 50% or more of children with a cut-off of “2 to 3 times a month” the data were missing the participant was deleted (0 = not bullied/bullying, 1 = bullied/bullying). from the study; also, when answers were contra- In previous research Solberg and Olweus demon- dictory and/or unreliable on account of at least 3 strated these measures’ validity and reliability for main questionnaire topics (eg, nutrition, alcohol adolescents. They stated that these scores allow for use, physical activity) the entire questionnaire was prevalence estimates of bullying and being bullied deleted from analyses. Data cleaning resulted in to be obtained conveniently, that they have a rea- usable questionnaires of over 95% of the partici- sonably well-defined meaning, and that they are pants. No forms of data imputation were applied. easily and unambiguously understood by users and researchers. Thereafter, Kyriakides, Kaloyirou Measures and Lindsay27 assessed its validity and reliability, Screen time. Watching television, using the concluding it to be a psychometrically sound mea- computer/Internet and playing video games will sure for bullying prevalence among adolescents. be referred to as “screen time behaviors.” Inter- Furthermore, healthy physical exercise pat- net use was defined as use of Internet for non- terns were defined as the following dichotomous school-related purposes. Video game playing was measure: at least one hour of moderately intensive defined as (online) gaming on a game console such physical activity every day, where at least twice a as the X-Box or PlayStation. Game use did not in- week the activity is aimed at improving or main- clude games with monetary awards or gambling. taining physical fitness22 (0 = sufficing, 1 = not Both the time spent on watching television, using sufficing). This measure is commonly used in the

820 Busch et al

Netherlands and is known as the Norm Healthy behaviors are dependent variables. Odds ratios Physical Exercise. Finally, for healthy nutrition (ORs) are presented with confidence intervals (CIs) another commonly used composite measure was at the 95% level. In the analyses, collinearity tests used, Norm Healthy Nutrition (0 = sufficing, 1 = were performed for the different independent vari- not sufficing), defined as a composite score of hav- ables and all were non-significant. All ORs were ing breakfast at least 5 times per week and eating controlled for confounding by sex, age, school, year fruits and vegetables at least 5 times per week.23 of school, educational level, ethnicity and socioeco- Health outcomes. Three indicators for adoles- nomic status (SES). The ORs were checked for in- cents’ physical and psychosocial well-being were teraction by sex. When significant sex interactions used as health outcome measures. Firstly, psycho- were present the analyses were further performed social problems were measured by the Strengths on a sex-stratified sample. and Difficulties Questionnaire (SDQ).28 This scale To answer the second research question, both is comprised of 5 subscales (emotional problems, the so-called “single-screen time analyses” and a conduct problems, hyperactivity, peer problems, “multi-screen time analysis” were performed. In and pro-social behavior). The total SDQ score the single-screen time analyses the associations is the sum of the scores on the first 4 subscales were analyzed with only one screen time behavior (maximum score of 40). A problematic total SDQ as the predicting variable together with the demo- score was defined as a score higher than 15, indi- graphic factors. In the multi-screen time analysis cating more psychosocial problems; this dichoto- the group of confounders in the regression analy- mous measure was used in the analyses to refer to sis was expanded to include the remaining screen students either having a ‘normal’ (= 0) or ‘(poten- time behaviors. The differences between the effect tially) problematic’ SDQ score (= 1). sizes (and their significance) of the associations A dichotomous measure for healthy weight was of the screen time behaviors and the other health used by means of the Body Mass Index (BMI), behaviors and outcomes in the single-screen time corrected for age and sex, with appropriate cut- behavior analyses versus those in the multi-screen off scores for adolescents, based on previous re- time behavior analysis can be interpreted as the search.20 This means that a different cut-off score effect of the confounding due to the screen time for a healthy BMI or being over- or underweight behaviors’ clustering. All statistical analyses were was used based on age and sex. performed with SPSS version 17.0. Due to the importance of self-esteem, social anxiety, and assertiveness in adolescent development RESULTS and psychosocial functioning, a composite mea- Overall, 2425 students completed the question- sure of these concepts was integrated into the UHS naire, a response rate of over 95%. Socio-demo- survey. In the literature this concept was referred graphic characteristics and screen time behaviors to as “General self-efficacy”(GSE).21 GSE has a are listed in Table 1. Approximately 45% or partici- broad definition without clear consensus. For this pants were boys, 55% were girls and their average study, Schwarzer’s definition of GSE is applied to age was 14 years (range 11-18 years). Their educa- refer to the concept of how one describes beliefs in tional level and ethnic composition is representa- their capabilities to practice control over challeng- tive for a Dutch sample of adolescents. Students’ ing demands and functioning across different psy- SES was reported to be somewhat higher than that chological domains. The functional domains inves- of their peers in the Netherlands.20 tigated in the current study are self-esteem, social In the following section, several different models anxiety and assertiveness, assessed by a survey are presented by topic. First, the single-behavior consisting of a combination of the Rosenberg’s Self regression analyses will be presented, comprised Esteem Scale29 and Schwarzer’s Generalized Self- of individual screen time behavior and standard Efficacy Scale.30 GSE is quantified here by a score socio-demographic confounders per analysis. Sec- based on a series of 11 questions on a 4-point Lik- ond, these different single-behavior analyses are ert Scale to indicate one’s beliefs about capabilities combined into a multi-screen time analysis. to practice control over challenging demands and over their functioning across the aforementioned Screen Time Behaviors’ Associations with domains. The cut-off score for a low/reduced GSE Other Unhealthy Behaviors was defined as one higher than 2.50, based on pre- Table 2 presents the ORs of the single-screen vious literature (0 = normal score, 1 = problematic time behavior analyses. These analyses concern score).29,30 the influence of screen time behaviors (independent variable) on ‘classic’ unhealthy behaviors (de- Statistical Analyses pendent variables). Furthermore, in Table 3 the To answer the first research question on how results of the multi-screen time behavior analyses the unhealthy behaviors and health outcomes are are presented. In contrast to the results presented associated with the relatively ‘new’ screen time in Table 2, the Table 3 results are corrected for behaviors, logistic regression analyses were con- confounding with regard to these additional screen ducted. The different screen time behaviors serve time behavior variables. Thus, the associations as independent variables whereas the other health between a certain unhealthy behavior (eg, binge

Am J Health Behav.™ 2013;37(6):819-830 DOI: http://dx.doi.org/10.5993/AJHB.37.6.11 821 Screen Time Associated with Health Behaviors and Outcomes in Adolescents

Table 1 Student Characteristics

Boys Girls Total N (%) N (%) N (%)

Age in Years 11-12 212 (19.7) 347 (25.8) 559 (23,0) 13-14 497 (46.1) 593 (44.2) 1091 (44,9) 15-16 328 (30.4) 363 (27) 691 (28,4) 17-18 40 (3.7) 40 (3.0) 88 (3,6) Mean in years 13.7 13.9 13.8 Socio-economic Status (FASa score) Low (0-2) – Medium (3-5) 174 (16.3) 260 (19.5) 435 (18.1) High (6-9) 896 (83.7) 1076 (80.5) 1972 (81.9) Watching TV (>14h/week) 271 (25.9) 324 (24.5) 595 (25.1) Internet Use (>14/week) 288 (27.3) 340 (25.5) 628 (26.3) Video Game Playing (>14/week) 48 (4.6) 8 (0.6) 56 (2.3) CIUb 50 (4.7) 45 (3.4) 95 (4.0) CVPc 43 (4.8) 5 (0.7) 48 (2.9) Being Bullied 86 (8.4) 84 (6.4) 170 (7.0) Bullying 57 (5.5) 32 (2.4) 89 (3.7) Alcohol User 300 (27.8) 308 (22.9) 608 (25.1) Binge Drinker 215 (20.9) 209 (15.8) 424 (17.5) Marijuana User 105 (9.7) 67 (5.0) 172 (7.1) Smoker 133 (12.7) 124 (9.3) 257 (10.6) Sufficing to Dutch Norm Healthy Physical Activityd 829 (78.1) 865 (65.0) 1694 (69.8) Sufficing to Dutch Norm Healthy Nutrition e 366 (34.3) 551 (41.3) 917 (37.8)

Note. a FAS = Family Affluence Scale b CIU = Compulsive Internet Use Scale Score >3.0 (range 0-4) c CVP = Videogame Addiction Test Score >3.0 (range 0-4) d Dutch Norm Healthy Physical Activity: “at least one hour of moderately intensive physical activity every day, where at least twice a week the activity is aimed at improving or maintaining physical fitness.” e Dutch Norm Healthy Nutrition: at least having breakfast, eating fruits and vegetables 5 times per week.

drinking) with a certain screen time behavior (eg, screen time behaviors into one multi-screen time excessive television watching), while correcting for analysis, thereby accounting for any confounding the remaining screen time behaviors are shown in due to their interrelatedness, excessive television Table 3. watching was only weakly associated with skipping Excessive television watching. In the single- school among girls (OR 4.14, 95% CI 0.96-17.82, screen time analyses, excessive television watch- .05 < p > .10); its previous associations with drug ing was associated with recent marijuana use (OR use, bullying, nutrition and physical exercise were 1.64, 95% CI 1.11-2.41), bullying (OR 1.51, 95% no longer statistically significant (Table 5). CI 1.09-2.09), poorer nutritional behaviors (OR Using the Internet/PC excessively. Excessive 1.36, 95% CI 1.10-1.69) and being less physically Internet use was associated in the single-screen active (OR 1.32, 95% CI 1.06-1.64). Watching tele- time analyses with recent alcohol use (OR 1.68, vision excessively also was associated with skip- 95% CI 1.32-2.13), binge drinking (OR 1.46, 95% ping school, but only for girls (OR 3.79, 95% CI CI 1.13-1.89), regular smoking (OR 1.53, 95% CI 1.35-10.68); this association was not statistically 1.14-2.05), skipping school (OR 1.73, 95% CI 1.01- significant for boys (Table 4). When combining the 2.95), bullying (OR 2.33, 95% CI 1.74-3.13), poor-

822 Busch et al

Table 2 Odds Ratios of the Association of Screen Time Behaviors with Other Unhealthy Behaviors Excessive Excessive Excessive video watching TV Internet use game playing CIU a CVP b (>14h/week) (>14h/week) (>14h/week) Recent marihuana use 1.64* 1.40+ N.S. 1.95 + Boys: 3.12** (1.11-2.41) (0.98-2.00) (0.96-3.95) (1.32-7.36) Girls: N.S. Recent alcohol use N.S.c 1.68* N.S. 1.69 + 2.02* (1.32-2.13) (0.99-2.86) (1.00-4.08) Binge drinking d N.S. 1.46* N.S. 1.79* 2.44* (1.13-1.89) (1.02-3.12) (1.20-4.98) Regular smoking N.S. 1.53* N.S. N.S. 1.96 + (1.14-2.05) (0.91-4.26) Unsafe sex N.S. N.S. N.S. N.S. N.S. Skipping school Girls: 3.79* 1.73* Boys: 3.30* 5.65** 4.91** (1.35-10.68) (1.01-2.95) (1.06-10.27) (2.64-12.08) (2.01-12.00) Boys: N.S. Girls: N.S. Bullying 1.51* 2.33** N.S. 2.95** 4.27** (1.09-2.09) (1.74-3.13) (1.72-5.04) (2.13-8.57) Being bullied N.S. N.S. N.S. 2.48** N.S. (1.36-4.53) Nutrition norm e 1.36** Boys: 1.36* N.S. 5.35** 6.64** (1.10-1.69) (1.00-1.86) (2.54-11.27) (2.03-21.72) Girls: 2.09** (1.57-2.78) Exercise normf 1.32* 1.61** N.S. 1.51 + N.S. (1.06-1.64) (1.31-1.99) (0.96-2.39)

Note. All results were adjusted for sex, age, school, year of school, education level, ethnicity and socioeconomic status. a CIU = Compulsive Internet Use Scale Score> 3.0, (range 0-4) b CVP = Videogame Addiction Test Score> 3.0, (range 0-4) c N.S. = Not Significant, p > .10 d Among drinkers e Dutch Norm Healthy Physical Activity: “at least one hour of moderately intensive physical activity every day, where at least twice a week the activity is aimed at improving or maintaining physical fitness.” f Dutch Norm Healthy Nutrition: at least having breakfast, eating fruits and vegetables 5 times per week. +.10 > p> .05; *: p < .05; **: p < .01

er nutritional behaviors (in boys OR 1.36, 95% CI analyses, despite slightly smaller effect sizes/re- 1.00-1.86 and in girls OR 2.09, 95% CI 1.57-2.78) gression slopes. and less physical exercise (OR 1.61, 95% CI 1.31- Playing video games excessively. Skipping 1.99) (Table 2). In the multi-screen time analysis school was the only behavior that was significantly (Table 3) a variety of classic unhealthy behaviors associated with excessive video game playing (OR were still associated with excessive Internet use: 3.30, 95% CI 1.06-10.27) (Table 2). This associa- recent alcohol use (OR 1.51, 95% CI 1.09-2.08), tion was only significant among boys. This - asso bullying (OR 2.12, 95% CI 1.41-3.18) and less ciation remained virtually the same in the multi- physical exercise (OR 1.77, 95% CI 1.34-2.34). screen time analysis (OR 3.69, 95% CI 1.09-12.54) For girls only, a significant association of exces- (Table 3). sive internet use and poorer nutritional behaviors Compulsive Internet use. In the single-screen (OR 1.87, 95% CI 1.22-2.86) emerged. These find- time analyses CIU was associated with binge drink- ings were similar to those of the single-screen time ing (OR 1.79, 95% CI 1.02-3.12), skipping school

Am J Health Behav.™ 2013;37(6):819-830 DOI: http://dx.doi.org/10.5993/AJHB.37.6.11 823 Screen Time Associated with Health Behaviors and Outcomes in Adolescents

Table 3 Multivariate Regression Analysis: Odds Ratios of the Association of Screen Time Behaviors with Other Unhealthy Behaviors Excessive Excessive Excessive video watching TV Internet use game playing CIU a CVP b (>14h/week) (>14h/week) (>14h/week) Recent marijuana use N.S.c N.S. N.S. N.S. Boys: 2.73 + (0.94-7.93) Girls: N.S. Recent alcohol use N.S. 1.51* N.S. N.S. N.S. (1.09-2.08) Binge drinking d N.S. 1.42 + N.S. N.S. N.S. (0.99-2.03) Regular smoking N.S. 1.45 + N.S. N.S. N.S. (0.96-2.20) Unsafe sex N.S. N.S. N.S. N.S. N.S. Skipping school Boys: N.S. N.S. Boys: 3.69* 4.16** N.S. (1.09-12.54) (1.45-11.96) Girls: 4.14+ Girls: N.S. (0.96-17.82) Bullying N.S. 2.12 ** N.S. N.S. 2.90* (1.41-3.18) (1.22-6.89) Being bullied N.S. N.S. N.S. 3.22** N.S. (1.43-7.22)

Nutrition norm e N.S. Boys: N.S. N.S. 5.79** 3.51* Girls: 1.87** (2.01-16.70) (1.02-12.07) (1.22-2.86) Exercise norm f N.S. 1.77** N.S. N.S. N.S. (1.34-2.34)

Note. All results were adjusted for sex, age, school, year of school, education level, ethnicity, socioeconomic status and the remaining screen time variables. a CIUS = Compulsive Internet Use Scale Score> 3.0, (range 0-4) b CVP = Videogame Addiction Test Score> 3.0, (range 0-4) c N.S. = Not Significant, p > .10 d Among drinkers e Dutch Norm Healthy Physical Activity: “at least one hour of moderately intensive physical activity every day, where at least twice a week the activity is aimed at improving or maintaining physical fitness.” f Dutch Norm Healthy Nutrition: at least having breakfast, eating fruits and vegetables 5 times per week + .10 > p > .05 *: p < .05; **: p < .01.

(OR 5.65, 95% CI 2.64-12.08), bullying (OR 2.95, Compulsive videogame playing. Compulsive 95% CI 1.72-5.04), being bullied (OR 2.48, 95% CI gamers were, in comparison to students that did 1.36-4.53) and poorer nutritional behaviors (OR not play videogames compulsively, more likely to 5.35, 95% CI 2.54-11.27) (Table 2). Many of these report recent alcohol use (OR 2.02, 95% CI 1.00- associations were still present in the multi-screen 4.08), binge drinking (OR 2.44, 95% CI 1.20-4.98), time analysis. Here, CIU was significantly associ- skipping school (OR 4.91, 95% CI 2.01-12.00), ated with skipping school (OR 4.16, 95% CI 1.45- bullying (OR 4.27, 95% CI 2.13-8.57) and hav- 11.96), being bullied (OR 3.22, 95% CI 1.43-7.22) ing poorer nutritional behaviors (OR 6.64, 95% CI and having poorer nutritional behaviors (OR 5.79, 2.03-21.72). For boys also an association with re- 95% CI 2.01-16.70), although the effect sizes/recent marihuana use was significant (OR 3.12, 95% gression slopes somewhat different (Table 3). CI 1.32-7.36) (see Table 2). In the multi-screen

824 Busch et al

Table 4 Odds Ratios of the Association of Screen Time with Psychosocial Problems, Being Overweight and GSE

Excessive Excessive Excessive video watching TV Internet use game playing CIU a CVP b (>14h/week) (>14h/week) (>14h/week)

Emotional problems N.S. c Boys: N.S. N.S. 3.94 ** 6.90 ** Girls: 1.65 ** (2.40-6.48) (3.45-13.80) (1.15-2.37) Conduct problems Boys: 1.38+ Boys: 1.81 ** 2.33 ** 3.47 ** 3.47 ** (0.97-1.97) (1.30-2.53) (1.28-4.25) (2.15-5.60) (1.84-6.55) Girls: 1.95** Girls: 1.55 * (1.27-3.00) (1.01-2.38) Hyperactivity N.S. 1.97 ** N.S. 2.64 ** 2.90 ** (1.59-2.43) (1.68-4.14) (1.55-5.41) Peer problems 1.31* N.S. 1.71 + Boys: 3.83 ** 4.09 ** (1.02-1.68) (0.93-3.13) (2.00-7.32) (2.15-7.78) Girls: 2.30 * (1.13-4.70) Pro-social problems N.S. 1.48 ** 1.73 + Boys: 2.29 ** 1.75 + (1.15-1.90) (0.94-3.17) (1.23-4.29) (0.91-3.37) Girls: 2.48 * (1.15-5.36) Problematic SDQ d -score 1.30 * 1.77 ** N.S. 5.06 ** 4.76 ** (1.00-1.68) (1.39-2.26) (3.23-7.93) (2.54-8.93) Overweight 1.77 ** N.S. N.S. N.S. N.S. (1.31-2.40) Low GSE e N.S. Boys: 1.85 + N.S. Boys: 3.35 * 2.38 + (0.91-3.77) (1.18-9.56) (0.87-6.49) Girls: 1.86 ** Girls: 4.81 ** (1.26-2.73) (2.45-9.47)

time analysis CVP was significantly associated Watching television excessively. In the single- with bullying (OR 2.90, 95% CI 1.22-6.89) and screen time analyses excessively watching televi- having poorer nutritional behaviors (OR 3.51, 95% sion was the only screen time behavior that was CI 1.02-12.07) (Table 3). associated with being overweight (OR 1.77, 95% CI 1.31-2.40). Furthermore, excessively watching Screen Time Behaviors’ Associations with television was also associated with peer problems Health Outcomes (OR 1.31, 95% CI 1.02-1.68), a problematic total In Table 4 the findings of the single-screen time SDQ score (OR 1.30, 95% CI 1.00-1.68), and for behaviors’ associations with the health outcomes girls also with conduct problems (OR 1.95, 95% CI psychosocial problems, being overweight and GSE 1.27-3.00) (Table 4). In the comprehensive multi- are shown and in Table 5 the multi-screen time screen time analysis being overweight was still analyses of the associations of different screen significantly associated with excessively -watch time behaviors with those health outcomes. ing television (OR 1.71, 95% CI 1.17-2.51) and a

Am J Health Behav.™ 2013;37(6):819-830 DOI: http://dx.doi.org/10.5993/AJHB.37.6.11 825 Screen Time Associated with Health Behaviors and Outcomes in Adolescents

Table 5 Multivariate Regression Analysis: Odds Ratios of the Association of Screen Time with Psychosocial Problems, Being Overweight and GSE

Excessive Excessive Excessive video CIU a CVP b watching TV Internet use game playing (>14h/week) (>14h/week) (>14h/week)

Emotional problems N.S. c N.S. N.S. 2.70 ** 3.56 ** (1.32-5.53) (1.53-8.27)

Conduct problems Boys: N.S. N.S. 2.46 * 3.29 ** N.S. Girls: 1.87+ (1.29-4.69) (1.71-6.33) (0.99-3.54) Hyperactivity N.S. 1.88 ** 0.40 ** 2.10 * N.S. (1.41-2.52) (0.18-0.90) (1.15-3.84) Peer problems N.S. N.S. N.S. N.S. 2.90 ** N.S. (1.39-6.04) Pro-social problems N.S. N.S. N.S. 4.72 ** N.S. (2.57-8.67) Problematic SDQ d-score N.S. N.S. N.S. N.S. N.S. Overweight 1.71** N.S. N.S. N.S. N.S. (1.17-2.51) Low GSE e N.S. Boys: N.S. N.S. Boys: N.S. N.S. Girls: 1.94* Girls: 4.39** (1.05-3.58) (1.70-11.34)

Note. All results were adjusted for sex, age, school, year of school, education level, ethnicity, socioeconomic status and the remaining screen time variables. a CIU = Compulsive Internet Use Scale Score> 3.0, (range 0-4) b CVP = Videogame Addiction Test Score> 3.0, (range 0-4) c N.S. = Not Significant, p > .10 d SDQ = Strengths and Difficulties Questionnaire e GSE = General Self-Efficacy + .10 > p > .05 *: p < .05; **: p < .01

statistical trend with regard to its association to atic SDQ score were not (Table 5). conduct problems was found, but the remaining Playing video games excessively. Excessive associations from the single-screen time analyses video game playing was only significantly associat- were insignificant in this analysis (Table 5). ed with conduct problems (OR 2.33, 95% CI 1.28- Using the Internet/PC excessively. Excessive 4.25) in the single-screen time analyses (Table 4). Internet use was significantly associated with This association to conduct problems remained emotional problems (OR 1.65, 95% CI 1.15-2.37, virtually similar (OR 2.46, 95% CI 1.29-4.69), but only significant in girls), conduct problems (in boys the weak association to peer problems and pro-so- OR 1.81, 95% CI 1.30-2.53 and in girls OR 1.55, cial behavior were now not statistically significant. 95% CI 1.01-2.38), hyperactivity (OR 1.97, 95% However, in this more comprehensive analysis, its CI 1.59-2.43), pro-social behavioral problems (OR association to less hyperactivity (OR 0.40, 95% CI 1.48, 95% CI 1.15-1.90), a problematic total SDQ- 0.18-0.90) now emerged as being significant (Table score (OR 1.77, 95% CI 1.39-2.26) and a low GSE 5). in girls (OR 1.86, 95% CI 1.26-2.73) (Table 4). In Compulsive Internet use. In the single-screen the multi-screen time analysis the association to time analyses CIU was significantly associated hyperactivity (OR 1.88, 95% CI 1.41-2.52) and (in with emotional problems (OR 3.94, 95% CI 2.40- girls) to low GSE (1.94, 95% CI 1.05-3.58) were still 6.48), conduct problems (OR 3.47, 95% CI 2.15- the only associations that were significant. How- 5.60), hyperactivity (OR 2.64, 95% CI 1.68-4.14), ever, the associations to emotional problems, con- peer problems (in boys OR 3.83, 95% CI 2.00-7.32 duct problems, pro-social behavior and a problem- and in girls OR 2.30, 95% CI 1.13-4.70), pro-social

826 Busch et al problems (in boys OR 2.29, 95% CI 1.23-4.29 and behaviors than their peers that did not watch tele- in girls OR 2.48, 95% CI 1.15-5.36), a problem- vision excessively. Our findings differed from those atic total SDQ-score (OR 5.06, 95% CI 3.23-7.93) in other studies that did report on significant as- and a low GSE (in boys OR 3.35, 95% CI 1.18-9.56 sociations with bullying,31 poorer eating habits32 and in girls OR 4.81, 95% CI 2.45-9.47) (Table 4). and less physical exercise.33 However, those stud- In the multi-screen time analyses its associations ies used less comprehensive multi-screen time with emotional problems (OR 2.70, 95% CI 1.32- behavioral correction for confounding. In our sin- 5.53), conduct problems (OR 3.29, 95% CI 1.71- gle-screen time behavior analyses that are more 6.33), hyperactivity (OR 2.10, 95% CI 1.15-3.84), comparable to those performed in these studies, and a problematic total SDQ score (OR 4.72, 95% similar significant associations were in fact also CI 2.57-8.67) remained significant. Additionally, in found between watching television excessively and girls, CIU also was related to a low GSE (OR 4.39, peer bullying, poorer nutritional habits, and less 95% CI 1.70-11.34) (Table 5). physical exercise. Therefore, the current findings Compulsive video game playing. In the single- demonstrated the possible overestimation of as- screen time analyses compulsive gaming was asso- sociations due to multi-screen time behavior con- ciated with emotional problems (OR 6.90, 95% CI founding. 3.45-13.80), conduct problems (OR 3.47, 95% CI Secondly, excessive Internet users were more 1.84-6.55), hyperactivity (OR 2.90, 95% CI 1.55- likely to be active alcohol users that non-excessive 5.41), peer problems (OR 4.09, 95% CI 2.15-7.78) Internet users (Table 3), a finding comparable to and a problematic total SDQ score (OR 4.76, 95% several previous studies34,35 In contrast to the study CI 2.54-8.93) (Table 4). The associations that re- by Kim,35 we found no associations between exces- mained significant in the multi-screen time analy- sive Internet use and smoking or marijuana use, sis were related to emotional problems (OR 3.56, when correcting for confounding by other screen 95% CI 1.53-8.27) and peer problems (OR 2.90, time behaviors. However, in the single-screen time 95% CI 1.39-6.04) (Table 5). behavior analyses, excessive Internet use was associated with smoking. This again illustrates the DISCUSSION importance of multi-behavioral analyses. Shi and Associations of screen time behaviors with a Mao36 reported a significant association between broad range of other unhealthy behaviors and sev- excessive Internet use and poorer nutritional be- eral health outcomes (being overweight, GSE, and haviors, similar to our findings, although the asso- psychosocial problems) were analyzed. In these ciation in our study was only significant for girls. It analyses we compared so called single-screen time was found that girls that used Internet excessively behavior analyses with multi-screen time behav- were more likely to report unhealthy eating pat- ior analyses. In the single-screen time behavior terns than non-excessive Internet users. Because analyses the association of a certain screen time no past studies reported this sex difference, this behavior with a ‘classic’ unhealthy behavior or finding could not be interpreted using existing lit- outcome was quantified, while only correcting for erature. Furthermore, the findings that excessive relevant socio-demographic factors. Thereafter, Internet users were less physically active than we compared these analyses with the so-called those who were not excessive Internet users was multi-screen time behavior analyses in which we confirmed by earlier research.33 presented the associations between a certain un- Thirdly, excessive video game players were more healthy behavior (eg, recent binge drinking) with likely to report skipping school (only significant a certain screen time behavior (eg, excessive tele- for boys) than students not reporting to play vid- vision watching), while correcting for both simi- eo games excessively. This effect was impossible lar socio-demographic factors and the remaining to determine for girls, because too few reported to screen time behaviors. Therefore, in contrast to the excessive use of video games. This finding is com- presented single-screen time behavior analyses, parable to other research that also found that girls these multi-screen time behavior analyses’ results to engage in less video game playing. Furthermore, are “unconfounded” with regard to the other screen excessive video game playing was not associated time behaviors. From here on when references are with any other unhealthy behaviors, as other re- made to associations of screen time behaviors with search (Desai et al37 and Koezuka et al33) has dem- other health behaviors or outcomes, reference is onstrated. to the multi-screen time behavior analyses’ results In addition, students reporting to be compul- unless explicitly stated otherwise. sive screen time users were more likely to report skipping school more often, being bullied by their Association of Screen Time Behaviors with peers, and having poorer nutritional habits. How- Other Unhealthy Behaviors ever, in contrast to the study of Ko et al,38 it was not The first study aim was to identify the associa- significantly related to alcohol use. However, once tions of the different screen time behaviors with more, in the single-screen time analyses students other unhealthy behaviors. Students that reported that were compulsive Internet users were more watching television excessively were not signifi- likely to be recent alcohol users than students that cantly more likely to report more other unhealthy were not compulsive Internet users. However, no

Am J Health Behav.™ 2013;37(6):819-830 DOI: http://dx.doi.org/10.5993/AJHB.37.6.11 827 Screen Time Associated with Health Behaviors and Outcomes in Adolescents literature was found to compare these findings be- other screen time behavior variables in the multi- cause measuring CIU is not yet a standard prac- screen time analysis in comparison to the single- tice in many adolescent health studies. This makes screen time analyses. Such a change would be the persistent findings in our study with regard to indicative of a close interrelationship between the such a wide range of health behaviors and out- screen time behaviors and their relation to a par- comes being related to CIU even more provocative. ticular outcome. Especially with regard to CIU, and Compulsive video game users were significantly to a lesser extent, with excessive Internet use, the more likely to be bullies and to report poorer nu- significant associations of the single-screen time tritional habits, findings not previously reported behavior analyses remained in the multi-screen in the literature. Like CIU, CVP may need to be time behavior analyses. However, the confounding included as part of the standard set of adolescent effects that different screen time behaviors pose health behaviors measured, given our new digital on each other’s associations to the outcomes in age. question in this study (other unhealthy behaviors and health outcomes) often were statistically sig- Association of Screen Time Behaviors with nificant. This indicates that the single-screen time Health Outcomes behavior analyses might overestimate the asso- Part of the first study aim was to examine wheth- ciations with other unhealthy behaviors or health er students that presented ‘problematic’ screen outcomes. It also implies that the “associated” time behaviors were more likely to report negative unhealthy behaviors and health outcomes of ado- health outcomes, ie, having psychosocial prob- lescents that watch television excessively and/or lems, being overweight, and having low GSE. play video games excessively or compulsively can As in previous studies9, students that reported be explained to a great extent by excessive or CIU watching television excessively were more likely to of those adolescents. This is an important finding, be overweight. No other associations were found, particularly given previous studies that only inves- which is comparable to previous research.39 These tigate one screen time behavior and do not take results, however, are in contrast to the findings of a broader spectrum of screen time behaviors into McClure et al,40 who reported that students that account. These differences in analyses seem to watched television excessively had lower self-es- explain the majority of the discrepancies between teem. However, direct comparability is limited by the findings of this study and previous literature. the fact that questions were not identical. Although some of the confidence intervals were Students reporting excessive Internet use were skewed (indicating a small cell size and that the more likely to report being hyperactive (part of the results need to be interpreted with caution) this SDQ survey). Similar to findings by Mathers et al,39 main conclusion still seems to hold true. we found no other significant association between excessive Internet use and total SDQ score. The Strengths and Limitations specific association of excessive Internet use with One strength of the study is that the study pop- psychosocial problems was only significant with ulation is its representativeness of Dutch adoles- regard to the hyperactivity sub-part of the SDQ. cents. A second strength is the specific set of mul- For girls an association with lower GSE was found tivariate analyses, and the comprehensive mea- for excessive Internet use, similar to what Nihill et sures of adolescent health. Many previous studies al41 found. only used what we have referred to in this paper Students who reported playing video games ex- as single-screen time behavior analyses instead of cessively were more likely to report conduct prob- more comprehensive multi-behavioral analyses. lems but less likely to report hyperactivity. Mathers Therefore, the current study is more likely to be et al39 found no association between video game properly conservative in its presented associations playing and psychosocial problems. and conclusions. Compulsive Internet use showed the strongest However, our data were based on self-report and association among screen time behaviors with could have represent some reporting bias. How- psychosocial problems, ie, emotional problems, ever, this effect was minimized by only using vali- conduct problems, hyperactivity, a problematic dated questionnaires, mostly based on the large total SDQ score, and, for girls, a low GSE score. (international) HBSC survey. Secondly, the report- These findings were similar to those of previous re- ed associations cannot be interpreted as causal search.6 relations, due to the cross-sectional nature of this Lastly, students who were compulsive video study. Also, it has to be taken into account that the game players were more likely to report psychoso- cut-off scores for excessive and compulsive screen cial problems. These findings were similar to those time use are standard measures in the surveys from other studies.7,42,43 used in the current study, but they may deviate from other investigations. The cut-off score for ex- Single-screen versus Multi-screen Time cessive screen time as problem behavior is sample Behavior Analyses dependent and study findings should not be gen- The second aim of this study was to quantify the eralized to other populations. Furthermore, many change in ORs that appeared after correcting for statistical tests have been performed, thereby rais-

828 Busch et al ing the danger of an inflated Type-I error. However, 283. despite these precautions, major conclusions from 3. Tremblay M, LeBlanc A, Kho M, et al. Systematic review this study offer provocative implications for future of sedentary behaviour and health indicators in school- aged children and youth. Int J Behav Nutr Phys Act. research. 201;8:98. 4. Yen JY, Ko CH, Yen CF, et al. Psychiatric symptoms in Recommendations for Future Research adolescents with Internet addiction: comparison with As shown, different screen time behavior vari- substance use. Psychiatry Clin Neurosci. 2008;62(1):9- ables are often closely associated. A screen time 16. variable that was not included in our study is the 5. Young KS. Internet addiction: the emergence of a new clinical disorder. CyberPsych Behav. 1996;1(3):237-244. use of smart phones or tablets. As Van Rooij et 6. Meerkerk GJ. Pwned by the Internet: explorative research al previously stated “even during the time span of into the causes and consequences of compulsive Internet writing this thesis, various technological innova- use. PhD Thesis, Erasmus MC: University Medical Cen- tions have been introduced.”7 Technology rapidly ter Rotterdam. Rotterdam, the Netherlands: Erasmus increases and smart phones seem to intensify peo- Universiteit Rotterdam. 2007. ple’s Internet use and with this increase, excessive 7. Van Rooij AJ. Online video game addiction. exploring a new phenomenon. PhD Thesis, Erasmus MC: University and compulsive Internet use may evolve further. Medical Center Rotterdam. Rotterdam, the Netherlands: This study showed the potential for (psychosocial) Erasmus Universiteit Rotterdam. 2011. problems associated with excessive or compulsive 8. Rey-Lopez JP, Vicente-Rodriguez G, Biosca M, Moreno screen time behavior, and therefore, the authors LA. Sedentary behaviour and obesity development in recommend that future studies take this devel- children and adolescents. Nutr Metab Cardiovasc Dis. opment into account. A second important aspect 2008;18(3):242-251. 9. Altenburg T, Singh A, van Mechelen W, et al. Direction of that currently is understudied is the content of the the association between body fatness and self-reported screen time behaviors ie, what adolescents watch, screen time in Dutch adolescents. Int J Behav Nutr Phys do, and play behind screens. As Mathers et al39 Act. 2012;9(4). show, this might provide interesting insights that 10. Gentile DA, Choo H, Liau A, et al. Pathological video shed light on these new screen time behaviors. game use among youths: a two-year longitudinal study. Pediatrics. 2011;127(2):e319-e329. 11. Yen JY, Ko CH, Yen CF, et al. The comorbid psychiatric Conclusions symptoms of internet addiction: attention deficit and hy- These results show several significant associa- peractivity disorder (ADHD), depression, social phobia, tions between screen time behaviors and unhealthy and hostility. J Adolesc Health. 2007;41(1):93-98. behaviors as well as with health outcomes related 12. Jang KS, Hwang SY, Choi JY. Internet addiction and to psychosocial problems, being overweight, and psychiatric symptoms among Korean adolescents. J Sch having low GSE. However, when correcting for the Health. 2008;78(3):165-171. 13. Foxcroft DR, Tsertsvadze A. Universal school-based pre- confounding effects by the other screen time behav- vention programs for alcohol misuse in young people. iors in the multi-screen time analysis, declines and Cochrane Database Syst Rev. 2011(5). The Cochrane changes in these associations were noted. In this Collaboration. multi-screen time behavior analysis, excessive and 14. Vreeman RC, Carroll AE. A systematic review of school- compulsive Internet use was uniquely associated based interventions to prevent bullying. Arch Pediatr with unhealthy behaviors and undesirable health Adolesc Med. 2007;161(1):78-88. 15. De Bruijn G, Van Den Putte B. Adolescent soft drink con- outcomes. Thus, these findings suggest that spe- sumption, television viewing and habit strength. Investi- cial attention should be paid in future studies to gating clustering effects in the Theory of Planned Behav- adolescents’ screen time and associated unhealthy iour. Appetite. 2009;53(1):66-75. behaviors in a clustered, comprehensive fashion. 16. van Nieuwenhuijzen M, Junger M, Velderman MK, et al. Clustering of health-compromising behavior and delin- Human Subjects Statement quency in adolescents and adults in the Dutch popula- tion. Prev Med. 2009;48(6):572-578. This study was approved by the Institutional 17. Zabinski MF, Norman GJ, Sallis JF, et al. Patterns of Review Board of the University Medical Center sedentary behavior among adolescents. Health Psychol. Utrecht, the Netherlands. METC-protocol number 2007;26(1):113-120. 11-397 / C. 18. Ko C-H, Yen J-Y, Liu S-C, et al. The associations between aggressive behaviors and Internet addiction Conflict of Interest Statement and online activities in adolescents. J Adolesc Health. 2009;44(6):598-605. All authors declare to have no conflict of interest. 19. Sanchez A, Norman GJ, Sallis JF, et al. Patterns and correlates of physical activity and nutrition behaviors in Acknowledgment adolescents. Am J Prev Med. 2007;32(2);124-130. Mr Busch and Ms Manders contributed equally 20. van Dorsselaer S, de Looze M, Vermeulen-Smit E, et al. to the research as primary authors. Health, well-being and upbringing of Dutch youngsters, HBSC 2009 (in Dutch: Gezondheid, welzijn en opvoeding van jongeren in Nederland). Trimbos Instituut; Universit- REFERENCES eit Utrecht; Sociaal en Cultureel Planbureau; the Neth- 1. Christakis D. Internet addiction: a 21st century epidem- erlands 2010. ic? BMC Med. 2010;8:61. 21. Bandura A. Self-efficacy: toward a unifying theory of be- 2. Weinstein A, Lejoyeux M. Internet addiction or excessive havior change. Psychol Rev. 1977;84:191-215. internet use. Am J Drug Alcohol Abuse. 2010;36(5):277- 22. Busch V, de Leeuw JRJ, Schrijvers AJP. Study proto-

Am J Health Behav.™ 2013;37(6):819-830 DOI: http://dx.doi.org/10.5993/AJHB.37.6.11 829 Screen Time Associated with Health Behaviors and Outcomes in Adolescents

col - an exploratory trial on health promoting schools at 33. Koezuka N, Koo M, Allison KR, et al. The relationship be- Dutch secondary schools. Int J Integr Care. 2012;12(2). tween sedentary activities and physical inactivity among 23. van Dorsselaer S, Zeijl E, van den Eeckhout S, et al. adolescents: results from the Canadian Community Health, well-being and upbringing of Dutch youngsters, Health Survey. J Adolesc Health. 2006;39(4):515-522. HBSC 2005 (in Dutch: Gezondheid en welzijn van jon- 34. Epstein JA. Adolescent computer use and alcohol use: geren in Nederland, HBSC 2005). Trimbos Instituut, So- what are the role of quantity and content of computer ciaal en Cultureel Planbureau; the Netherlands, 2007. use? Addict Behav. 2011;36(5):520-522. 24. Meerkerk GJ, van den Eijden RJ, Vermulst AA, Gar- 35. Kim JY. The nonlinear association between internet us- retsen HF. The Compulsive Internet Use Scale (CIUS): ing time for non-educational purposes and adolescent some psychometric properties. CyberPsych Behav. health. J Prev Med Public Health. 2012;45(1):37-46. 2009;12(1):1-6. 36. Shi L, Mao Y. Excessive recreational computer use and 25. Van Rooij AJ, Schoenmakers TM, Vermulst AA, et al. On- food consumption behaviour among adolescents. Ital J line video game addiction: identification of addicted ado- Pediatr. 2010;36(52);1-4. lescent gamers. Addiction. 2011;106(1):205-212. 37. Desai RA, Krishnan-Sarin S, Cavallo D, Potenza MN. 26. Solberg ME, Olweus D. Prevalence estimation of school Video-gaming among high school students: health cor- bullying with the Olweus Bully/Victim Questionnaire. relates, gender differences, and problematic gaming. Pe- Aggress Behav. 2003;29(3):239-268. diatrics. 2010;126(6):e1414-e1424. 27. Kyriakides L, Kaloyirou C, Lindsay, G. An analysis of 38. Ko C-H, Yen J-Y, Yen C-F, et al. The association between the Revised Olweus Bully/Victim Questionnaire us- internet addiction and problematic alcohol use in adoles- ing the Rasch measurement model. Br J Educ Psychol. cents: the problem behavior model. CyberPsych Behav. 2006;76(4):781-801. 2008;11(5):571-576. 28. Muris P, Meesters C, van den Berg F. The Strengths and 39. Mathers M, Canterford L, Olds T, et al. Electronic media Difficulties Questionnaire (SDQ): further evidence for its use and adolescent health and well-being: cross-section- reliability and validity in a community sample of Dutch al community study. Acad Pediatr. 2009;9(5):307-314. children and adolescents. Eur Child Adolesc Psychiatry. 40. McClure AC, Tanski SE, Kingsbury J, et al. Character- 2003;12(1):1-8. istics associated with low self-esteem among US adoles- 29. Rosenberg M. Society and the Adolescent Self-image. cents. Acad Pediatr. 2010;10(4):238-244. Princeton, NJ: Princeton University Press; 1972. 41. Nihill GFJ, Lubans DR, Plotnikoff RC. Associations be- 30. Teeuw B, Schwarzer R, Jeruzalem M. Dutch adaptation tween sedentary behavior and self-esteem in adolescent of the General Self-Efficacy Scale. Available at: http:// girls from schools in low-income communities. Ment userpage.fu-berlin.de/~health/dutch.htm. Accessed Health Phys Act. 2012;5(1):1-8. June 12, 2012. 42. Ghassemzadeh L, Shahraray M, Moradi A. Prevalence of 31. Kuntsche E, Pickett W, Overpeck M, et al. Television Internet addiction and comparison of Internet addicts viewing and forms of bullying among adolescents from and non-addicts in Iranian high schools. CyberPsych Be- eight countries. J Adolesc Health. 2006;39(6):908-915. hav. 2008;11(6):731-733. 32. Pearson N, Biddle SJH. Sedentary behavior and dietary 43. Lemmens JS, Valkenburg PM, Peter J. Psychosocial intake in children, adolescents, and adults: a systematic causes and consequences of pathological gaming. Com- review. Am J Prev Med. 2011;41(2):178-188. put Human Behav. 2011;27(1):144-152.

830 Copyright of American Journal of Health Behavior is the property of PNG Publications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.