We used data from the 2007 National Survey of Children’s Health (NSCH), which was conducted between April 2007 and July 2008 by the Centers for Disease Control and Prevention (CDC), National Center for Health Statistics (NCHS), with funding from the Maternal and Child Health Bureau of the Health Resources and Services Administration.^22,23 The survey includes data on a variety of indicators of child health and well-being.^22,23

A random-digit dial sample of households with landline telephones and children under 18 years of age was selected from each state and Washington, DC. The sample included approximately 1,800 households per state, with a total sample of 91,642 children from birth through 17 years of age.^22,23 The survey was conducted in English, Spanish, and 4 Asian languages.^22,23 For households with multiple eligible children, 1 child was randomly selected to be the participant of the survey. The parent or guardian who knew most about the child’s health status and health care was selected as the respondent; therefore, all data are based on parental/guardian reports.

The NSCH data are weighted to be representative of noninstitutionalized children ages 0 to 17 years nationally and within each state.²³ The interview completion rate was 66%.²³

In accordance with Department of Health and Human Services (HHS) regulations (45 CFR 46), the NCHS Ethics Review Board and the National Opinion Research Center Institutional Review Board approved all study procedures and modifications.²³

We conducted 2 analyses. In Analysis I, we examined the association of sociodemographic and behavioral characteristics with excess screen time among 6 to 11 and 12 to 17-year-olds. We limited analyses to this age range because the screen time questions were asked only for children 6 to 17 years of age. In Analysis II, we examined the association of sociodemographic and behavioral characteristics associated with obesity among 12 to 17-year-olds only. We limited Analysis II to this age group because of the lack of accurate height and weight data for 6 to 11-year-olds.

Prevalence estimates were calculated for 3 variables: screen time >2 hours per day, for both age groups; obesity status for the older age group; and for each of the independent variables. We used t-tests for differences in proportions for each independent variable. This allowed us to determine if there were statistically significant differences between each category of the independent variables; differences were considered statistically significant if p < 0.01. Furthermore, we used logistic regression to calculate unadjusted and adjusted odds ratios (OR) for screen time >2 hours per day and obesity status.

On the basis of previous findings,²⁸ we tested for an interaction effect for sex × race/ethnicity among 12 to 17-year-olds and present results with this interaction term. We did not test for this interaction effect among 6 to 11-year-olds because sex was not significantly associated with screen time for this age group. In addition, we examined if there was an interaction between having a TV in the bedroom and exceeding screen time recommendations. For analyses, we used SAS-Callable SUDAAN statistical analysis software that accounted for the complex sample design.²⁹

The NSCH 2007 sample included 64,076 children who were 6 to 17-years-old. We excluded those whose race/ethnicity was non-Hispanic multiracial (N = 2,776) or non-Hispanic other (N = 2,613) or missing (N = 1,091), or the response for screen time was either missing (N = 318) or the respondent did not know or refused (N = 3,067). After these exclusions, there was a total of 54,211 youth 6 to 17 years of age, with 23,416 aged 6 to 11 years and 30,795 aged 12 to 17 years. Because BMI was included for 12 to 17-year-olds only, we also excluded those who had an extreme value for height (N = 411), weight (N = 504), or both (N = 44), or were missing data to calculate BMI (N = 831). Our final sample of 12 to 17-year-olds was 29,005. With the exception of household income described above, missing data for each variable ranged from 0.02% for the TV-in-the-bedroom variable to 4.8% for the screen time variable.

For this age group, 20.8% engaged in screen time more than 2 hours per day (Table 1). We found statistically significant differences in prevalence of excess screen time for all variables examined except sex. By race/ethnicity, non-Hispanic blacks and Hispanics had the highest prevalence (37.8% and 24.4%, respectively). For FPL, the highest prevalence was among those at <100% FPL (30.9%); for physical activity participation, those who did not participate in exercise in the week preceding the survey had the highest prevalence (35%); for adequacy of sleep, those never getting enough sleep in the preceding week had the highest prevalence (37.9%); and those with a bedroom TV had nearly double the prevalence than those who did not (27.6% and 14.7%, respectively).

Results of multivariable logistic regression show several subgroups of sociodemographic and behavioral variables were significantly associated with engaging in excessive (>2 hours per day) screen time (Table 1). Non-Hispanic black children (OR = 2.5) had significantly higher odds than Non-Hispanic white children, and children living at an FPL of <100% or 100% to 199% had higher odds than children at ≥400% FPL (OR = 1.7 and 1.6, respectively). In addition, not engaging in any physical activity the preceding week, or engaging 1 to 3 days (OR = 1.8 and 1.4, respectively) was associated with excessive screen time and children who did not get enough sleep any night in the preceding week were more likely to engage in excessive screen time, compared with children obtaining enough sleep every night in the same week (OR = 2.2). Finally, children with a bedroom TV were more likely to engage in excessive screen time per day, compared with those who had none (OR = 1.7).

For this age group, 26.1% of 12 to 17 year olds engaged in excessive screen time (Table 2). For prevalence of screen time, t-tests found statistically significant differences within subgroups for all variables examined. The prevalence of engaging in excessive screen time varied by sex × race/ethnicity subgroup. Non-Hispanic black males and females had the highest prevalence (45.5% and 40.2%, respectively). For FPL status, the highest prevalence was found for those at <100% FPL (37.6%); for physical activity participation, those who did not participate in the week preceding the survey had the highest prevalence (38.5%); for adequacy of sleep, those never getting enough sleep in the week preceding survey had the highest prevalence (34%); and those with a TV in their bedroom had approximately 50% higher prevalence than those who did not (30.5% and 20.0%, respectively).

Multivariate logistic regression revealed that several subgroups of sociodemographic and behavioral variables were significantly associated with excessive screen time. Non-Hispanic black males (OR = 3.4) and non-Hispanic black females (OR = 2.7) had the highest likelihood of exceeding screen time recommendations, compared with non-Hispanic white females (Table 2). Youth living at less than 400% FPL also had higher odds than youth at 400% or greater FPL (<100% FPL, OR = 2.1, 100% to 199%; OR = 2.0; 200% to 399% OR = 1.6). For the behavioral variables, similar patterns were found for 12 to 17-year-olds as among 6 to 11-year-olds. Lower levels of physical activity participation were all significantly associated with higher odds of engaging in excessive screen time: Never, OR = 2.0, 1 to 3 days, OR = 1.4; never sleeping enough during the week preceding survey (OR = 1.5) and having a TV in the bedroom (OR = 1.4).

Among 12 to 17-year-olds, the overall obesity prevalence was 13.5%. As shown in Table 3, within each variable subgroup except for sleep, there were statistically significant differences in the prevalence of obesity. By age, obesity prevalence was highest among 12-year-olds at 18.2%. Males had a higher prevalence than females (16.7% vs. 10.0%), Hispanics (19.3%) had the highest prevalence for race/ethnicity, as well as those living at <100% FPL (20.8%). There were variations in obesity prevalence by frequency of participation in physical activity with those who participated 1 to 3 days (15.1%) or did not participate in the previous week having the highest prevalence (14.7%). Adolescents with a TV in their bedroom and engaging in excessive screen time had nearly triple the prevalence of obesity compared with adolescents without a TV in their bedroom and meeting screen time recommendations (20.9% and 7.6%, respectively).

In multivariate logistic regression, we found several sociodemographic variables to be significantly associated with higher odds of obesity (Table 3). Children aged 12 years compared with those aged 17 years (OR = 1.8); males compared with females (OR = 1.8); non-Hispanic blacks (OR = 1.4) and Hispanics (OR = 1.6) had higher odds of obesity compared with non-Hispanic whites; and those living at a FPL below 200% (<100% OR = 2.4, 100% to 199% OR = 1.9) also had higher odds compared with youth at ≥400% FPL. For the behavioral variables, lower than daily frequencies of physical activity (Never, OR = 1.6; 1 to 3 days, OR = 1.6, 4 to 6 days, OR = 1.4) and never obtaining enough sleep (OR = 1.5) in the week preceding the survey were associated with increased odds of obesity. Last, the interaction term between having a TV in the bedroom × exceeding screen time recommendations showed that in each combination compared with the reference category, the results were significant, and that youth who had a TV in their bedroom and exceeded >2 hours per day had the highest odds of obesity (OR = 2.5).

There are limitations to our findings. First, it is cross-sectional and only allows for examination of associations and not causal relationships. Second, because the respondents were parents or knowledgeable caregivers, it is possible they were unable to provide valid responses to some questions; for example, the amount of time their child spends watching TV, watching videos, or playing video games. However, prior research has shown parent report of TV viewing is acceptable although parents tend to underestimate viewing time, especially for those children with a TV in their bedroom.^37,38 Furthermore, the questionnaire only assessed weekday TV viewing, videos and playing video games, which may lead to underestimation since viewing time may be higher on weekends than weekdays.³⁹ Another limitation is presence of a TV in the bedroom was not measured objectively, although due to its simplicity there is speculation this is answered fairly accurately.²¹ Questions are determined by NSCH and many have not undergone psychometric testing. Third, our obesity estimates are limited by 3 factors. Age was reported in years, and the midpoint of each age group was used for calculating BMI; we excluded 959 (3.0% of analytic sample) respondents who had extreme values for height, weight, or both. Also, the respondent reported the sample child’s weight and height. Despite the high correlations between measured and parent-reported weight, height, and BMI among adolescents, mean weight is typically underestimated from reported data, with larger differences between reported and measured data for females than males.⁴⁰–⁴⁴ Studies have demonstrated that reported data have high specificity (92% to 99%). However, the sensitivity of self-reported data to detect obesity has ranged from 45% to 76%.^40,42,43 This low sensitivity likely accounts for the lower prevalence of obesity that we observed, compared with measured data from NHANES 2007 to 2008 showing 18.1% of adolescents aged 12 to 19 years were obese,⁴⁵ our obesity prevalence estimate from NSCH 2007 was 13.5% among adolescents aged 12 to 17 years. For younger children, we did not examine obesity status due to the 2003 NSCH experience where parents significantly underreported their child’s height, which led to too many children being classified as overweight. The 2007 NSCH took measures to ensure researchers do not use these data by suppressing height and calculated BMI categorizations for children less than 10 years of age.²³ Fourth, the question on sleep is subjective. An alternative to asking about adequate sleep is asking about sleep duration to allow for comparison against the National Sleep Foundation’s recommended amounts of sleep for youth.

In conclusion, given that more than one fifth of youth exceed screen time recommendations, and that both excess screen time and having a bedroom TV are associated with increased odds of obesity, a public health challenge exists to raise awareness among families and schools about the impact of excessive TV time and having a bedroom TV, and to design effective interventions to reduce screen time among youth. On the basis of our findings, interventions targeting males, non-Hispanic blacks, and those of lower FPL are needed, given the greater risks for excess screen time and obesity in these subgroups. Further research should identify modifiable family, home, and school intervention opportunities. In addition, psychometric research is needed to determine the best measures for assessing screen time and presence of a TV in the bedroom for parent-report questionnaires, especially during adolescence because parents may have less control over time use. This will help move the field forward when assessing media use.