OUP user menu

Factors associated with media use among adolescents: a multilevel approach

Xavier Garcia-Continente , Anna Pérez-Giménez , Albert Espelt , Manel Nebot Adell
DOI: http://dx.doi.org/10.1093/eurpub/ckt013 5-10 First published online: 8 February 2013

Abstract

Background: During the last few years, several studies have reported a high screen time use among adolescents that can be related to negative health effects. The aims of this study were to describe screen time use among secondary school students and to identify individual- and school-level factors associated with media use. Methods: A cross-sectional study based on a self-reported questionnaire was performed among a representative sample of 2675 secondary school students (13–19 years old). Adolescents reported the amount of time spent viewing television, playing videogames and using the computer as well as other health-related behaviours and attitudes. Multilevel analysis was carried out and prevalence ratios were calculated to determine the association between media use and related factors. Results: Around 50% of the students reported watching television for ≥2 h/day during weekdays. Boys reported playing videogames for ≥2 h/weekday much more often than girls (14.6 and 1.5%, respectively). 68.2% of boys and 61.7% of girls reported using the computer for ≥2 h/weekday. In the multilevel analysis, the main factors associated with screen-related sedentary behaviours were attending schools from a low socio-economic status neighbourhood, eating unhealthy food and not reading books frequently. Conclusion: The prevalence of adolescents reporting an excessive use of media devices is high, especially among students attending schools from deprived areas. Interventions to reduce screen time among adolescents may be necessary to reduce the risk of some metabolic and cardiovascular diseases such as being overweight and obesity in late adolescence or early adulthood.

Introduction

There is a growing concern over the long-term impact of sedentary lifestyles on overweight and obesity, particularly among youth.1,2 Of all sedentary behaviours, television viewing, videogame playing and computer use are currently highly common among adolescents.3 Some studies suggested that spending time in these activities reduces the time available for physical activity.4,5 However, more recent studies and meta-analyses have observed no evidence or weak evidence of this ‘replacement hypothesis’.6 Otherwise, it has been suggested that screen-related sedentary behaviours may take time away from psychological activities such as studying or reading books which could negatively affect the school performance7 as media use has been related to more distractibility for academic activities, whereas reading books was negatively related to distractibility.8 Screen-related sedentary behaviours have also been associated with an increase of the energy intake,9 which may give rise to a high-energy balance and increase the risk of overweight and obesity.9,10 There is also growing evidence that screen-related sedentary behaviours, specifically television viewing, are associated with metabolic and cardiovascular diseases such as diabetes mellitus, hypercholesterolemia and hypertension, which are increasing among youth.3,11,12

The American Academy of Pediatrics (AAP) recommends that children over 2 years and adolescents should not be exposed to media devices for more than 2 h/day of quality programmes.13 More recently, other countries such as Canada, Australia and Spain have launched similar recommendations on the use of media devices among this population.14 Despite these recommendations, a recent European study showed that 35.2 and 32.3% of boys and girls 15–18 years old viewed television for ≥2 h/weekday, whereas 21.6% of boys and 9% of girls played computer games for at least 2 h/weekday. 7.6% of boys and 0.5% of girls played with videogames above the recommended limits.15 In another study carried out in Catalonia (Spain), 34.5% of boys and 29.7% of girls 10–14 years old reported spending ≥2 h/day viewing television, playing video or computer games or surfing on Internet.16

Although a few studies from southern European countries have identified individual factors associated with screen-related sedentary behaviours, there are no previous epidemiological studies that have analysed school environmental factors. Adolescents spend most part of a weekday at school and their behaviour may be influenced by schoolmates, professors or even school characteristics. As it has been seen in studies that consider neighbourhood’s socio-economic status (SES) of residence,17 neighbourhood’s SES of school could be associated with healthy behaviours, including sedentary behaviours. Otherwise, adolescents may also be influenced by the type of school funding. It is known that subsidized and private schools offer more extracurricular activities than state schools and these kind of structured activities are positively associated with healthy behaviours during the adolescence.18

A survey of lifestyle risk factors [Factores de Riesgo en ESColares (FRESC)]19 among secondary school students (13–19 years old) is performed periodically by the Agència de Salut Pública de Barcelona (ASPB) (Public Health Agency, Barcelona) since 1987. This survey allows us to describe, monitor and identify risk- and health-related behaviours among adolescents. In the 2008 edition, the FRESC survey collected information about screen-related sedentary behaviours for the first time. The variables included in this study were selected based on a theoretical model created according to associations found in previous studies cited above. The aims of the study were to describe screen-related sedentary behaviours (viewing television, playing videogames and using the computer) and to identify the relevant individual and school ecological factors associated with media use among secondary school students.

Methods

Design and subjects

During the 2007–08 academic year, a cross-sectional study was carried out among a sample of eighth- (14 years old), tenth- (16 years old) and twelfth- (18 years old) grade students from state, private and subsidized secondary schools in Barcelona. Data came from the FRESC survey (2008 edition) and its design took into consideration questions from previous editions and other validated national and international surveys. The questionnaire was anonymous and collected information about substance use, personal relationships, leisure time and eating patterns. Furthermore, students’ height and weight were objectively measured.19

This sample size was fixed to obtain a maximum error of 3% (α = 0.05) to estimate a prevalence of health-related behaviours of 50%. The sample unit was the classroom to take into account the school size. A random sample of classrooms was selected considering six stratum according to the type of school (state or private/subsidized) and SES of the school’s neighbourhood [low, medium or high SES based on the Family Economic Capacity Index (FECI)]20. When possible, schools declining to participate were replaced by other ones from the same stratum. Twenty-one classrooms were replaced by others, whereas 5 classrooms could not be replaced. From 1404 eligible classrooms (34.6% were from a state school), 141 of them were sampled. Finally, 136 classrooms (response rate was 96.5% and 32.4% were from state school) from 80 secondary schools were included in the study.19 The overall response rate of the students who were eligible was 87.2%. The sample was representative of the classroom distribution in Barcelona according to the FECI of the neighbourhood where the school was located and the type of school which the classrooms belonged to.

The questionnaire was administered between February and May 2008, during regular school hours in the presence of the classroom teacher and by trained personnel from the ASPB. The data set was created using Teleform® v.10.2 software.21

Study variables

Dependent variable

Regarding screen-related sedentary behaviours, adolescents reported the amount of time viewing television, playing videogames and using the computer for weekdays during leisure time. The questionnaire gathered the following questions: ‘How many hours per weekday do you usually spend watching television?’; ‘How many hours per weekday do you usually spend playing videogames?’ and ‘How many hours per weekday do you usually spend playing with the computer, chatting on messenger and surfing the Internet?’. All the questions owned an open-ended format but the three sedentary behaviours were collapsed into two categories (<2 h or ≥2 h/day). According to these three variables, the dependent variable ‘media use’ was defined as viewing television, playing videogames or using the computer (excluding for academic purposes) ≥2 h/day on weekdays (Yes/No).

Independent variables

The questionnaire included the following socio-demographic variables: sex (boy or girl), age (considered as a continuous variable), type of school (state, private or subsidized), place of birth, family structure (living with two parents or other situations), the student’s SES and the school’s SES. To obtain the student’s SES, the Family Affluence Scale (FAS) was used.22 This scale is based on a few material conditions of the student’s households (car, computer, holidays and bedroom). The school’s SES was established by using the FECI, an ecologic score based on socio-economic indicators of the neighbourhood (job, electricity consumption, home rental prices or car power).20 Both SES measures were categorized as high, medium or low. Place of birth was categorized according to students’ and parents’ birthplace, yielding three groups: all born in Spain; student born in Spain and mother or father born abroad; and student born abroad (independently of the parents’ birthplace).

Body mass index (BMI) was derived from objective measurements of height and weight. Gender and age-specific cut-off points for BMI were used to classify students as being of normal weight, overweight or obese.23 There were also a few questions about eating patterns. We asked about having breakfast before leaving home every day and about unhealthy food intake. According to the recommendations of food pyramid, unhealthy food groups studied were sweets, cakes, cookies, soft drinks and crisps and the variable was defined as having consumed some of this food four or more times in the last week.

Finally, other independent variables included in the study were self-perceived academic performance, reading books frequently and regular physical exercise. Regarding self-perceived academic performance, the students were asked about their perceived relative position (low, medium or high) in their class comparing with their classmates. Students also had to report how often they read books and did physical exercise during leisure time. Both questions had five response categories: ‘never’, ‘a few times a year’, ‘once or twice a month’, ‘at least once a week’ and ‘almost every day’. Reading books frequently was defined as reading at least once a week during leisure time and doing regular physical exercise was categorized as taking physical activity almost every day during leisure time.

Statistical analysis

Students without information on sex (n = 8) or out of the target age range (n = 83) (13 - to 15-year-old in the eighth grade, 15 - to 17-year-old in the tenth grade and 17 - to 19-year-old in the twelfth grade) were excluded from the analysis. In addition, as it has been done in similar studies,24 students who reported doubtful screen times (≥10 h/day) (n = 414) were excluded. A sensitivity analysis was performed to assess the previous defined cut-off point (≥10 h/day) for the variables viewing television, playing videogames and using the computer. Similar associations were found using lower and upper cut-off points.

Percentages and 95% confidence intervals (CIs) were calculated for categorical variables. Mean and standard deviation or the median and interquartile range were calculated for continuous variables. In addition, multilevel analysis using logistic regression models were performed to investigate the variables that were considered to be potentially associated with media use.25 Individual variables were treated as the first level and school variables (type of school and the school’s SES) as the second level. Significant variables in the bivariate analysis were further analysed using multilevel multivariate logistic regression analysis. The odds ratio provided were transformed into prevalence ratios (PRs) using the conversion formula and their 95% CIs were calculated using the Miettinen formula, which was already used in previous studies.26 Multilevel logistic regression models and transformation of the odds ratio provided were performed instead of multilevel Poisson regression models due to convergence problems running Poisson models. The school-level variance and intraclass correlation coefficient were also calculated. All the analyses were performed stratifying by sex.

All statistical analyses were conducted using Stata v.10 and multilevel analyses were obtained using HLM v.6.02.

Results

A total of 2675 students were included in the analyses. The main characteristics of the sample are described in Table 1. The mean age of the students was 15.8 years and 53% were girls. Most students (72.7%) attended subsidized or private schools and 84.9% went to schools located in medium or high SES neighborhoods.20 11.7% of the students were born abroad, whereas 7.9% were born in Spain although either their father or mother was born abroad. Almost 70% of the students lived in two-parent families. About a half of the students declared not reading books frequently and 55.4% did not regularly perform physical exercise. Regarding eating patterns, 57.8% ate unhealthy food almost every day and >60% did not have breakfast before leaving home every day.

View this table:
Table 1

Socio-demographic and relevant variables of the sample by sexa

Boys (n = 1258)Girls (n = 1417)Total (n = 2675)
n (%)n (%)n (%)
Age [mean (SD)]15.8 (1.71)15.9 (1.78)15.8 (1.75)
Grade
    8th grade493 (39.2)517 (36.5)1010 (37.8)
    10th grade461 (36.6)476 (33.6)937 (35.0)
    12th grade304 (24.2)424 (29.9)728 (27.2)
Type of school
    State334 (73.5)395 (27.9)729 (27.3)
    Private/subsidized924 (26.5)1022 (72.1)1946 (72.7)
Socio-economic status of the schoolb
    Low204 (16.2)200 (14.1)404 (15.1)
    Medium562 (44.7)596 (42.1)1158 (43.3)
    High492 (39.1)621 (43.8)1113 (41.6)
FASc,d
    Low127 (10.1)148 (10.4)275 (10.3)
    Medium542 (43.1)629 (44.4)1171 (43.8)
    High575 (45.7)634 (44.7)1209 (45.2)
Place of birthd,e
    Spain1008 (80.1)1103 (77.8)2111 (78.9)
    Spain/Outside Spain96 (7.6)114 (8.1)210 (7.9)
    Outside Spain133 (10.6)179 (12.6)312 (11.7)
Family situationd
    Living with mother and father867 (68.9)946 (66.8)1813 (67.8)
    Other387 (30.8)464 (32.8)851 (31.8)
Self-perceived academic performanced
    High412 (32.8)416 (29.4)828 (31.0)
    Medium628 (49.9)807 (57.0)1435 (53.6)
    Low206 (16.4)182 (12.8)388 (14.5)
Reading books frequentlyd
    No703 (55.9)693 (48.9)1396 (52.2)
    Yes540 (42.9)712 (50.3)1252 (46.8)
Regular physical exercised
    No482 (38.3)999 (70.5)1481 (55.4)
    Yes757 (60.2)408 (28.8)1165 (43.6)
Unhealthy food intaked
    No424 (33.7)649 (45.8)1073 (40.1)
    Yes799 (63.5)748 (52.8)1547 (57.8)
Breakfast before leaving homed
    No373 (29.7)577 (40.7)950 (35.5)
    Yes868 (69.0)831 (58.7)1699 (63.5)
Body Mass Indexd
    Normal weight937 (74.5)1124 (79.3)2061 (77.1)
    Overweight242 (19.2)235 (16.6)477 (17.8)
    Obesity72 (5.7)48 (3.4)120 (4.5)
  • a: FRESC Report [Factores de Riesgo en ESColares (Risk factors in schoolchildren)], 2008, Barcelona, Spain.19

  • b: Socio-economic status of the school’s district based on the FECI (1996).

  • c: FAS.

  • d: Missing values <2%.

  • e: Spain: student, mother and father born in Spain. Spain/outside Spain: student born in Spain and mother or father born abroad. Outside Spain: student born abroad.

  • SD indicates standard deviation.

Table 2 shows information about the amount of time spent viewing television, playing videogames and using the computer on weekdays by boys and girls. Overall, ∼50% of the students (boys and girls) spent at least 2 h/weekday watching television. Compared with girls, a higher percentage of boys played videogames (14.7%) and used the computer (68.2%) for ≥2 h/weekday (1.5 and 61.7% among girls, respectively). Moreover, higher percentages of playing videogames were found among boys and girls of younger age. However, differences by age were not found in television viewing or media use (data not shown).

View this table:
Table 2

Prevalence and 95% CI of television viewing, videogame playing and computer use among secondary school students on weekdaysa

BoysGirls
nb% (95% CI)Median (min)IQR (min)nb% (95% CI)Median (min)IQR (min)
Television viewing (min/day)(N = 1223)10560–150(N = 1401)12060–180
    0282.3 (1.5–3.1)271.9 (1.2–2.6)
    1–11959648.7 (45.9–51.5)61944.2 (41.6–46.8)
    ≥12059949.0 (46.2–51.8)75553.9 (51.3–56.5)
Videogame playing (min/day)(N = 1206)300–60(N = 1362)00–0
    051843.0 (40.2–45.7)109880.6 (78.5–82.7)
    1–11951142.3 (39.6–45.2)24417.9 (15.9–20.0)
    ≥12017714.7 (12.7–16.7)201.5 (0.8–2.1)
Computer use (min/day)(N = 1220)16590–270(N = 1394)15060–260
    0715.8 (4.5–7.1)987.0 (5.6–8.4)
    1–11931726.0 (23.5–28.4)43631.3 (28.8–33.7)
    ≥12083268.2 (65.6–70.8)86061.7 (59.1–64.2)
Media usec(N = 1218)(N = 1391)
    ≥120 min/day in any of screen-related sedentary behaviours97880.3 (78.1–82.5)109078.4 (76.2–80.5)
  • a: FRESC Report [Factores de Riesgo en ESColares (Risk factors in schoolchildren)], 2008, Barcelona, Spain19

  • b: Missing values <5%.

  • c: Viewing television, playing with videogames or using the computer for ≥2 h/day.

  • CI indicates confidence interval; IQR, interquartile range.

The school-level variances for the null models were 0.189 (P = 0.009) in boys and 0.216 (P = 0.001) in girls. Intraclass correlation coefficients were 5.43 and 6.16, respectively. Table 3 shows the PR and their 95% CI obtained from multilevel analyses. Among boys, using media for ≥2 h/weekday was associated with attending a lower SES school [low SES: adjusted PR (aPR) = 1.18, 95% CI: 1.07–1.31; medium SES: aPR = 1.16, 95% CI: 1.09–1.25], attending a state school (aPR = 1.10, 95% CI: 1.03–1.18), not living in bi-parental families (aPR = 1.07, 95% CI: 1.01–1.14), reading books frequently (aPR = 0.85, 95% CI: 0.80–0.90) and eating often unhealthy food (aPR = 1.11, 95% CI: 1.04–1.19). In girls, those who attended a state school (aPR = 1.06, 95% CI: 1.00–1.14) with lower SES (low SES: aPR = 1.22, 95% CI: 1.10–1.35; medium SES: aPR = 1.13, 95% CI: 1.06–1.22), did not read books frequently (aPR = 0.80, 95% CI: 0.86–0.96) and ate often unhealthy food (aPR = 1.11, 95% CI: 1.04–1.18) were also more probably to use media ≥2 h. Moreover, reporting a mid self-perceived academic performance (low SES: aPR = 1.21, 95% CI: 1.08–1.35; medium SES: aPR = 1.13, 95% CI: 1.05–1.22) and not having breakfast before leaving home (aPR = 0.94, 95% CI: 0.89–0.99) were associated with the same media use pattern among girls.

View this table:
Table 3

Multilevel analysis of the association between using media (viewing television, playing videogames or using the computer) for ≥2 h/weekday and socio-demographic characteristics, behaviours and other variablesa

BoysGirls
n (%)PR (95% CI)aPR (95% CI)n (%)PR (95% CI)aPR (95% CI)
Individual variables
Family structure
    Living with mother and father662 (78.4)1730 (78.2)1
    Other314 (84.9)1.08 (1.02–1.15)1.07 (1.01–1.14)354 (78.7)1.01 (0.95–1.07)
Self-perceived academic performance
    High314 (77.5)1289 (69.8)1
    Medium495 (81.7)1.05 (0.98–1.12)640 (81.1)1.16 (1.08–1.24)1.13 (1.05–1.22)
    Low162 (81.8)1.06 (0.97–1.16)152 (86.4)1.24 (1.12–1.38)1.21 (1.08–1.35)
Reading books frequently
    No586 (86.2)1574 (83.8)1
    Yes383 (72.5)0.84 (0.80–0.90)0.85 (0.80–0.90)508 (73.0)0.88 (0.84–0.93)0.90 (0.86–0.96)
Regular physical exercise
    No378 (81.3)1782 (79.6)1
    Yes592 (80.0)0.98 (0.93–1.04)302 (75.5)0.95 (0.89–1.02)
Unhealthy food intake
    No304 (73.4)1468 (73.2)1
    Yes652 (83.8)1.14 (1.07–1.22)1.11 (1.04–1.19)607 (82.6)1.13 (1.06–1.19)1.11 (1.04–1.18)
Breakfast before leaving home
    No303 (83.0)1472 (83.3)1
    Yes667 (79.0)0.95 (0.89–1.01)613 (75.0)0.91 (0.86–0.96)0.94 (0.89–0.99)
Body mass index
    Normal weight728 (80.4)1861 (78.1)1
    Overweight189 (79.4)0.98 (0.92–1.06)180 (77.6)0.98 (0.91–1.06)
    Obesity57 (81.4)1.00 (0.89–1.14)39 (83.0)1.06 (0.90–1.23)
Contextual variables
Type of school
    Private/subsidized700 (77.9)1771 (76.6)1
    State278 (87.2)1.13 (1.05–1.21)1.10 (1.03–1.18)319 (83.1)1.09 (1.01–1.18)1.06 (1.00–1.14)
Socio-economic status of the schoolb
    High347 (72.3)1440 (71.8)1
    Medium456 (84.6)1.17 (1.10–1.25)1.16 (1.09–1.25)479 (81.6)1.13 (1.06–1.22)1.13 (1.06–1.22)
    Low175 (87.9)1.22 (1.11–1.33)1.18 (1.07–1.31)171 (89.5)1.24 (1.13–1.37)1.22 (1.10–1.35)
  • a: FRESC, Factores de Riesgo en ESColares (Risk factors in schoolchildren)19.

  • b: Socio-economic status of the school’s district based on the FECI (1996).

  • PR indicates prevalence ratio; aPR, adjusted prevalence ratio; CI, confidence interval.

Discussion

To our knowledge, this is the first study that investigates the association between screen-related sedentary behaviours and school variables (SES and type of school). The results show an association between screen-related sedentary behaviour and these contextual variables as well as individual variables such as not reading books frequently, unhealthy eating habits, low self-perceived academic performance and family structure. This study also shows a high prevalence of students spending excessive daily screen time on weekdays.

Based on AAP recommendations for screen time (<2 h/day), half of our adolescents exceeded the recommendations for television viewing and around two-thirds for the computer use. Our results are similar to the prevalence of television viewing observed in adolescents from UK (49.8%).27 However, recent European15 and Spanish28 data showed lower rates (30%) except the HBSC study,22 which showed higher prevalences among Spanish adolescents (62–69%). Overall, previous literature shows a wide range of prevalences of viewing television above the recommended levels.15,22 These discrepancies in prevalence may be due to methodological differences across the studies in measuring screen time (i.e. using of closed-ended vs. open-ended questions, asking for an usual weekday vs. for weekdays in a specific time period, etc.) and comparisons should be made with caution.

Regarding videogames playing and computer use, our results showed higher prevalences of these activities than those observed in a study performed in similarly aged European adolescents.15 However, these differences must be taken with caution because of methodological differences across the studies. On the other hand, like the previous research, we found that boys reported playing videogames more than girls.15,24,29

Several studies have found an inverse relationship between sedentary behaviours and SES.27,30,31 Most of these studies have used individual indicators such as the parents’ education or household income to assess adolescent or child SES. The HBSC study, which used FAS, also showed that higher levels of television viewing were associated with lower family affluence in most of the countries studied.22 Contrary to these findings, our data did not show this association. These differences may be related to the limitations of adolescents’ SES indicators, as indicators used for the adult population are not considered sufficiently valid to determine adolescents’ SES.32 Nevertheless, we observed a relationship between sedentary behaviours and the school’s SES (ecological indicator). Thus, independently of the SES of the adolescents’ family, students who attended schools located in lower SES neighbourhoods were more likely to overuse media. Otherwise, students who attended a state school reported using media more often than those who attended private or subsidized ones. Private or subsidized schools provide more extracurricular activities than state ones and students from private or subsidized schools located in a high SES neighbourhood are usually from higher income families which allows them to enrol into these activities.33 Thus, we hypothesize that accessibility to extracurricular activities might displace time spent at home and therefore it reduces the amount of screen time. Nevertheless, there is limited research on screen-related sedentary behaviours by type of school. Further research is needed to enlarge our knowledge.

Some studies have described a positive association between television viewing and being overweight among children and adolescents,1,9,34 whereas others have reported no association3,35,36 or weak association.37 A meta-analysis35 showed a slight association between television viewing and being overweight, but the clinical relevance of the relationship was questioned. Television viewing has been hypothesized to be related to unhealthy eating habits such as skipping meals37 and poor nutrition,9 which could be reflected in being overweight or obese in the medium or long term.1,6 Despite this, our data did not show association between media use and being overweight, and we found a positive association with unhealthy eating habits such as unhealthy food intake and not having breakfast before leaving home.

Our results showed that boys who lived with both father and mother were less likely to use media excessively than those who lived in single-parent or reconstructed families. In girls, we did not observe this association. It is possible that parental control on their children’s behaviour, including media use, is greater in bi-parental families than in single-parent or reconstructed families, especially among boys who usually obey less than girls to similar patterns of guidance.38 Thus, the greater parental control needed towards boys than towards girls could explain the gender differences observed. Otherwise, adolescents who overuse media reported reading books less frequently and a lower self-perceived academic performance, especially among girls. These results are consistent with those studies that have negatively related screen time to school performance.4,7,39,40 Most of these studies8 suggested that screen time could displace time spent on cognitive activities such as reading or doing homework. If time available for academic tasks is replaced by using media, school performance could be negatively affected. A study carried out among American youth showed that reading books was related to less distractibility for academic activities, while media use was positively related to distractibility.8 Sharif et al.7 further observed that some television programmes may influence adolescent behaviours, such as aggressive behaviour or substance abuse, indirectly also affecting school performance. A relationship has also been found between screen time and sleeping patterns. Adolescents who viewed television and used Internet more often went to bed later. Thus, they spent less time in bed and reported higher levels of tiredness, affecting school performance.41

Some potential limitations of this study should be considered. First, selection bias may have occurred, since some students were excluded from the analysis mainly because of bizarre responses. Excluded students were mainly boys of younger age and attended state schools in low SES neighbourhoods. Therefore, their exclusion may affect external validity. The high number of screen hours reported by some students could be due to the misinterpretation of the questions. Second, the questionnaire gathered a large number of health indicators as it was not a survey specially designed to study media use. Thus, some variables such as physical exercise were not accurately defined. Otherwise, the use of self-reported information raises some concern about the reliability and accuracy of the data. However, anonymous questionnaires for self-reported behaviours have previously been proven to have good validity and reliability.42 Moreover, to minimize this threat, we used a previously validated questionnaire that has been used for periodic monitoring of schoolchildren’s lifestyles. A final limitation would be the cross-sectional design that does not let determination of causal relationships. However, because of their low cost and high feasibility, this kind of study is considered a robust approach to obtain descriptive information and to detect new emergent problems.

The elevated percentage of adolescents who overuse media is a public health concern because of its negative consequences on adolescents’ health. This study shows social inequalities among schools and the importance of the implementation of preventive programmes focused on reducing screen time, especially in state schools located in low SES neighbourhoods. Nevertheless, further longitudinal studies are needed to confirm our findings and establish the direction of the relationships. More detailed information about family behaviour (such as parental control in using media devices) would also be useful to analyse the contextual processes of leisure time use. In addition, information on the quality of videogames and television programmes, the location of the media devices or patterns such as eating while viewing television, should be taken into account.

Key points

  • Adolescents reported a high prevalence of excessive media use.

  • Attending a state school with a lower SES was associated with using media more than the recommended levels.

  • Some individual factors previously identified as associated to media use such as unhealthy eating habits or not reading books have been confirmed by multilevel analysis when they are analysed together.

  • Preventive programmes focused to reduce screen time should take into account social inequalities at school.

Acknowledgements

The authors would like to thank the teachers and students of the participating schools. We would also like to thank personnel of the Community Health Service and of the Evaluation and Intervention Methods Service of the Agència de Salut Pública de Barcelona who collaborated in the administration of the questionnaires, height and weight measurement and to the creation of the data set. This work was supported by the “Agència de Gestió d'Ajuts Universitaris i de Recerca de la Generalitat de Catalunya” (AGAUR SGR2009-1345). This article represents partial fulfilment of the requirements for the PhD programme of Xavier Garcia Continente at the Pompeu Fabra University (Barcelona, Spain).

Conflicts of interest: None declared.

Footnotes

  • Dr. Manel Nebot died in October 18, 2012. This work would not have been possible without his convincement, perseverance and dedication.

References

View Abstract