eTable. Association (ß Coefficient and 95% CI) Between Television Viewing Duration and Sleep Duration in Cross-sectional Analysis
Marinelli M, Sunyer J, Alvarez-Pedrerol M, Iñiguez C, Torrent M, Vioque J, Turner MC, Julvez J. Hours of Television Viewing and Sleep Duration in ChildrenA Multicenter Birth Cohort Study. JAMA Pediatr. 2014;168(5):458-464. doi:10.1001/jamapediatrics.2013.3861
This study used longitudinal data to examine potential associations between hours of television viewing and sleep duration in children.
To examine the association between hours of television viewing and sleep duration in preschool and school-aged children.
Design, Setting, and Participants
Longitudinal, multicenter study among birth cohorts in Menorca, Sabadell, and Valencia from the Spanish Infancia y Medio Ambiente (environment and childhood) project. The study sample included 1713 children (468 from Menorca, 560 from Sabadell, and 685 from Valencia).
Parent-reported child television viewing duration measured in hours per day at 2 and 4 years of age in Sabadell and Valencia and at 6 and 9 years of age in Menorca.
Main Outcomes and Measures
Parent-reported child sleep duration measured in hours per day at 2 and 4 years of age in Sabadell and Valencia and at 6 and 9 years of age in Menorca.
In cross-sectional analysis, children with longer periods of television viewing reported at baseline (≥1.5 hours per day) had shorter sleep duration. Longitudinally, children with reported increases in television viewing duration over time (from <1.5 to ≥1.5 hours per day) had a reduction in sleep duration at follow-up visits. Results were similar when examining television viewing duration as a continuous variable, with each 1 hour per day of increased viewing decreasing sleep duration at follow-up visits (β = −0.11; 95% CI, –0.18 to −0.05). Associations were similar when television viewing duration was assessed during weekends and after adjusting for potential intermediate factors (child executive function and attention-deficit/hyperactivity disorder symptoms) and confounders (child physical activity level, parental mental health status, maternal IQ, and maternal marital status).
Conclusions and Relevance
Children spending longer periods watching television had shorter sleep duration. Changes in television viewing duration were inversely associated with changes in sleep duration in longitudinal analysis. Parents should consider avoiding long periods of daily television exposure among preschool and school-aged children.
Sleep is an important survival need, and inadequate sleep duration has been associated with increased risk of psychiatric disorders, overweight, and obesity and with poor school performance among children and adolescents.1- 14 Previous findings suggested that television viewing is an important determinant of irregular sleep habits and duration.15- 17 An increased time spent in front of the screen may displace sleep time, and the content of television programs could also increase sleep problems, including nightmares and reduced sleep duration.16 In 2009, the American Academy of Pediatrics18 recommended that parents should avoid any media exposure among children younger than 2 years and stated that it should be limited to 1 to 2 hours per day among older children. Violent scenes in animated features may also lead to impulsive aggressive behaviors among children younger than 8 years because they may be unable to distinguish between real life and fantasy.18
A longitudinal study19 reported that adolescents who watched television for 3 or more hours per day experienced a greater number of sleep problems in early adulthood compared with adolescents who watched television for less than 3 hours per day. Regarding changes in television viewing over time, increases in television viewing duration have also been observed to relate to subsequent increases in sleep problems.19 To our knowledge, no longitudinal study has yet assessed the potential association between the daily number of hours spent watching television and sleep duration among preschool and school-aged children. Although some previous investigations assessed this relationship cross-sectionally, results are limited because of the potential for reverse causality. Furthermore, most previous studies16,17,20 did not control for potentially important confounders such as parental mental health status and maternal IQ or intermediate factors such as child executive function and attention-deficit/hyperactivity disorder (ADHD) symptoms.
The aim of the present study was to examine the potential association between hours of television viewing and sleep duration in preschool and school-aged children using prospective data from 3 birth cohorts (2 of them assessed at 2 and 4 years of age and the third at 6 and 9 years of age). The potential role of various intermediate factors (child executive function and ADHD symptoms) was also studied.
The study was approved by the appropriate ethical committees, and written informed consent was obtained from the parents of all children before enrollment in the study. The study sample comprised 3 population-based birth cohorts in Menorca (n = 482), Sabadell (n = 657), and Valencia (n = 787) from the Spanish Infancia y Medio Ambiente (environment and childhood) project. The Infancia y Medio Ambiente project is a research network in Spain, initiated in 2001. In Menorca, recruitment of pregnant women in the first trimester of pregnancy began in 1997-1998, whereas in Sabadell and Valencia it started in 2004 to 2007. Women fulfilled the following inclusion criteria: aged 16 years or older, intent to deliver in the city, no assisted conception, and no communication problems. Further protocol information has been published elsewhere.21
Herein, we assessed children with repeated measures on time spent watching television and sleep duration. Because the cohorts were coordinated from different study centers with somewhat varying research objectives, both television viewing duration and sleep duration were collected at different follow-up points. In total, 1713 children (468 from Menorca, 560 from Sabadell, and 685 from Valencia) were included in the study.
The child television viewing duration during weekdays was assessed at 2 and 4 years of age in Sabadell and Valencia and at 6 and 9 years of age in Menorca. Television viewing duration during weekends was evaluated at 2 and 4 years of age in Sabadell and Valencia but not in Menorca. We assessed children’s television viewing duration in all 3 cohorts according to parental report using the following open-ended question: “How many minutes per day does your child watch television?” In Valencia, parents responded to the same question about their child at 4 years of age but based on the following 7 response categories: 0 (none or almost none), 1 (<0.5 hour), 2 (0.5-1 hour), 3 (roughly 1 hour), 4 (2 hours), 5 (3 hours), or 6 (≥4 hours). According to relevant pediatric recommendations,22 the television viewing duration variable was standardized by grouping hours of television viewing per day into the 2 main categories of shorter television viewing (<1.5 hours) and longer television viewing (≥1.5 hours).
We identified the following categories of longitudinal changes in television viewing duration over time: increased television viewing duration, decreased television viewing duration, and no change in television viewing duration. Increased television viewers included children who were reported as shorter television viewers (<1.5 hours per day) at baseline but as longer television viewers (≥1.5 hours per day) at follow-up visits, whereas decreased television viewers included children displaying the opposite pattern. The latter included children who did not change their television habits, who constituted a reference group.
Television viewing duration was also examined as a continuous variable (in hours per day) for viewing duration at baseline and at follow-up visits. In the Valencia cohort, the television viewing variable was categorized according to 0, 0.25, 0.75, 1, 2, 3, and 4 hours per day.
Child sleep duration and sleep-related habits were also assessed by parental questionnaire at 2 and 4 years of age in Sabadell and Valencia and at 6 and 9 years of age in Menorca. To assess sleep duration, parents were asked to reply to the following open-ended question: “How long does the child sleep per day, including naps (in hours)?”
Information was also collected on whether the child sleeps alone (yes or no), the age of the mattress (in years), the presence of humidity in the child’s bedroom (yes or no), and the noise level in the child’s home (high, moderate, low, or absent). Information about whether the child sleeps alone was available in all cohorts except for Menorca at the 9-year follow-up visit. Information about the age of the mattress was reported only in the Sabadell and Valencia cohorts. The presence of humidity in the child’s bedroom was evaluated in Valencia and Menorca at the 2-year and 6-year follow-up visits, respectively. Finally, the noise level in the child’s home was qualitatively assessed in Sabadell and Valencia at the 4-year follow-up visit.
Based on previous data,23 we included the following variables as covariates in regression models: child sex, body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) at baseline, BMI change from baseline, and parental educational level (primary school or less, secondary, or university). Information on weight (in kilograms) and BMI was obtained from pediatric records. Information on parental educational level (in years) was obtained from the baseline questionnaire. In addition, we included the following in sensitivity analyses: parental mental health status, maternal IQ, maternal marital status, child physical activity level, child ADHD symptoms, and child executive function.
Psychometric data included maternal IQ by factor G of the test by Cattell and Cattell24 (in Menorca). Parental psychopathological symptoms were assessed by the Revised Symptom Checklist (in Sabadell and Valencia); this is a self-reported questionnaire widely used in healthy and distressed populations that permits the evaluation of depression, anxiety, and a global severity index of distress.23,25 The similarities subscale of the Wechsler Adult Intelligence Scale–Third Edition was used as a proxy measure of maternal verbal IQ (in Sabadell and Valencia).26,27 Child ADHD Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) symptoms and the McCarthy Scales of Children’s Abilities executive function scores were available for all cohorts.23 Maternal marital status was recorded during the first trimester of pregnancy in Sabadell and Valencia (whether the parents were living together or not) and at the 4-year follow-up visit in Menorca (married and unmarried mothers). Child physical activity level was obtained by parental interview at 6 and 9 years of age in Menorca only according to the following question: “In relation to other children of the same age, do you think that your child does high or roughly high, normal, or low amounts of physical activity?”
We first examined associations between sleep duration and sleep habits (whether the child sleeps alone or with others, the age of the mattress, the presence of humidity in the child’s bedroom, and the noise level in the child’s home) to assess the criterion-related validity of sleep duration. Subsequently, analyses were conducted using linear and mixed regression models to examine the association between television viewing duration and sleep duration.
Linear regression models were used to examine the association between time spent watching television and sleep duration at 2 and 4 years of age in Sabadell and Valencia and at 6 and 9 years of age in Menorca. Multiple linear mixed regression models were used to examine the association between television viewing duration change and sleep duration. The linear mixed regression approach permits simultaneous modeling of the fixed effect and the random effect of the variables in the constant to account for between-child variation. Children who did not change their television habits served as a reference group. Because only 2 individual measurements were available, no specific variance structure was considered. The data assessment to construct the models consisted of creating a new variable representing baseline television viewing duration and creating a new variable representing the change in terms of television viewing duration from baseline.
Data from Sabadell and Valencia were combined because those children were about the same age at baseline and follow-up visits and because sleep duration was also similar. Further joint analyses included Menorca because nonheterogeneity was observed between cohorts. In sensitivity analyses, we assessed the role of the following other potential confounders: maternal IQ (all cohorts had available data), parental psychopathological symptoms (in Sabadell and Valencia), maternal marital status (all cohorts), and child physical activity level (in Menorca). Further sensitivity analyses included potential intermediate factors such as child ADHD Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) symptoms and the McCarthy Scales of Children’s Abilities executive function scores (all cohorts).
We also evaluated the association between time spent watching television during weekends and sleep duration in Sabadell and Valencia. The same final models were used for studying this association during weekdays.
Statistical significance was set at α < .05 (2-tailed). Statistical analyses were conducted using available software (R version 2.11.1; http://cran.r-project.org/bin/windows/base/old/2.11.1/).
Table 1 summarizes the distribution of television viewing duration and sleep duration across the different cohorts. Television viewing duration increased slightly with older child age, while sleep duration decreased. Children with more ADHD symptoms, lower executive function scores, and higher BMIs had shorter sleep duration (Table 2). No important differences in sleep duration were observed according to parental educational level or child sex. Results also did not change by cohort or assessment period (P > .05 for heterogeneity).
Results in cross-sectional analysis suggested no association between sleep duration and sleep habits or child physical activity level (data not shown). A strong inverse association between time spent watching television and sleep duration was observed among children from all cohorts and assessment periods (eTable in the Supplement).
Sleep duration β coefficients of children who were in the longer television viewing group at baseline or who increased or decreased television viewing duration at follow-up visits are summarized in Table 3. Longer television viewing at baseline was associated with reduced sleep duration at follow-up visits across all cohorts. Relative to changes in television viewing duration, children with increased television exposure had reduced sleep duration (β = −0.21; 95% CI, −0.37 to −0.05). Children with decreased television exposure showed a recovery of sleep duration (β = 0.21; 95% CI, 0.00 to 0.42), although this was not statistically significant. Findings were similar when examining television viewing duration as a continuous variable (hours per day watching television) (Table 3). Associations did not vary by cohort (P > .05 for heterogeneity). We observed similar results when considering only television viewing duration during weekends (Table 4).
Associations were similar after adjusting for various potential confounders. These included child physical activity level, maternal IQ (factor G), and maternal marital status in Menorca and included parental psychopathological symptoms, maternal IQ (Wechsler Adult Intelligence–Third Edition), and maternal marital status in Sabadell and Valencia. In addition, we analyzed the role of potential intermediate factors (ADHD symptoms and executive function scores) at 4 years of age; however, the results did not meaningfully change (Table 5).
We observed an inverse association between time spent watching television and sleep duration in a multicenter population-based sample of preschool and school-aged children in Spain. In cross-sectional analysis, television viewing duration was inversely associated with sleep duration across the cohorts assessed separately and in the joint analysis of all 3 cohorts combined. Two main findings emerged from longitudinal models. First, increasing television exposure from baseline led to a reduction in sleep duration at follow-up visits. Second, children who had a decrease in television viewing duration showed a moderate recovery, with longer sleep duration at follow-up visits. To our knowledge, this is the first prospective study assessing the association between television viewing duration and sleep duration in children.
Similar findings were observed by examining television viewing duration during weekends. However, some evidence of a moderate attenuation of the association was observed, probably because of a reduction in child stress during weekends.
In agreement with previous data,28 sleep duration declined with older child age. Investigators have assessed child television habits and intensity relative to sleep disturbances; however, the study designs were cross-sectional.16,17,19 Among children 4 to 10 years of age, Owens et al16 studied whether television viewing habits (television in the bedroom, the use of television as a sleep aid, bedtime television viewing, and the mean weekday television viewing duration) affected sleep disturbances (bedtime resistance, sleep-onset delay, sleep duration, sleep anxiety, and night waking). They found an increase in most sleep disturbances with adverse television viewing practices. Sleep duration was also significantly associated with some television viewing practices before bedtime but not with the mean time spent watching television during the day. Van den Bulck20 investigated similar relationships among older children. Having a television set in the child’s room and spending more time watching television were both associated with a later bedtime during weekdays and with shorter sleep duration during weekends. A study by Paavonen et al17 found some relationships between passive television exposure to adult content programs and sleep problem severity among children 5 to 6 years of age. Finally, data from an aforementioned longitudinal study19 among adolescents suggested that individuals who watched more television (≥3 vs <3 hours per day) developed more sleep problems in early adulthood. However, the study did not consider sleep duration.
Potential intermediate factors such as ADHD symptoms and executive function scores at 4 years of age were included in the sensitivity analyses, without affecting the associations observed. Early and more recent epidemiological studies5,9,22,29 highlighted the association between ADHD symptoms and lower executive function with shorter sleep duration and sleepiness in children and adolescents. Moreover, some evidence suggested that excessive television viewing during early childhood was associated with attention problems in children.30 A longitudinal study by Landhuis et al31 reported a positive association between greater time spent watching television during childhood and more attention problems later in adolescence. In 2011, Lillard and Peterson32 identified an association between watching fast-paced television cartoons for 9 minutes and performing poorly on executive function tasks compared with watching other educational programs or drawing during the same time. Two possible explanations have been suggested. First, the fast-paced images of cartoons and other television programs may overstimulate the child and disrupt normal brain development.30 Second, watching television could reduce child motivation to perform other activities such as drawing, physical activity, or school homework that enhance neurodevelopment.33 However, we did not observe that child executive function and ADHD symptoms were in the pathway between television viewing duration and sleep duration herein.
The main strength of this study is the external validity of findings because of the use of data from large population-based birth cohorts, as well as the ability to conduct longitudinal analyses that strengthen the associations observed. Findings were similar when considering television exposure as a continuous variable and after adjusting for several important potential confounders such as parental social class, parental mental health, maternal IQ, maternal marital status, and child physical activity level.
This study also had some limitations. First, both television viewing duration and sleep duration were by parental report. However, the consistency of associations across the 3 cohorts may suggest a lack of significant misclassification error. Second, data on the content of television programs were unavailable, which limits the possibility of investigating the longitudinal association between types of television programs watched (eg, educational or fast-paced animated feature films) and sleep duration. Third, we also lacked data on child physical activity level and parental mental health in some of the cohorts. Fourth, we were unable to investigate the potential effect of the timing of television exposure (eg, watching television just before going to bed). Fifth, the fact that both television viewing duration and sleep duration were assessed at only 2 time points did not permit us to consider an adequate correlation structure between observations.
Several potential mechanistic explanations have been suggested. Exposure to a bright screen before bedtime may have a role in salivary melatonin concentration changes and subsequent reduction in sleep duration.34- 36 In addition, children with more television exposure before sleeping may be more prone to being excited and displacing sleep time. Another explanation may be related to the content of television programs. As reported elsewhere,18 all American animated feature films for children produced between 1937 and 1999 contained violent acts. At age 18 years, an individual may have been exposed to more than an estimated 200 000 violent acts. Children younger than 8 years may also be more vulnerable to the content of television programs because they may be unable to discriminate between fantasy and reality, leading to increasing perceived stress.
This multicenter birth cohort study showed that children spending longer periods watching television everyday had subsequent reductions in sleep duration during the short term. Most of the associations were also robust to adjusting for various potential intermediate factors such as child executive function and ADHD symptoms and for confounders such as maternal educational level, maternal IQ, maternal marital status, parental mental health, and child physical activity level. Further prospective studies are required to confirm these findings and to investigate the mechanisms that may underline the possible association.
Accepted for Publication: August 7, 2013.
Corresponding Author: Marcella Marinelli, MSc, PhD, Center for Research in Environmental Epidemiology, Calle Dr Aiguader 88, 08003 Barcelona, Spain (email@example.com).
Published Online: March 10, 2014. doi:10.1001/jamapediatrics.2013.3861.
Author Contributions: Dr Sunyer had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Marinelli, Sunyer, Julvez.
Acquisition of data: Marinelli, Sunyer, Alvarez-Pedrerol, Iñiguez, Torrent, Vioque, Julvez.
Analysis and interpretation of data: Marinelli, Sunyer, Julvez.
Drafting of the manuscript: Marinelli.
Critical revision of the manuscript for important intellectual content: Sunyer, Alvarez-Pedrerol, Iñiguez, Torrent, Vioque, Turner, Julvez.
Statistical analysis: Marinelli.
Obtained funding: Sunyer, Torrent.
Conflict of Interest Disclosures: None reported.
Funding Sources: This study was funded by grants Red INMA G03/176 and CB06/02/004; FIS-FEDER 03/1615, 04/1509, 04/1112, 04/1931, 05/1079, 05/1052, 06/1213, 07/0314, 11/01007, and 09/02647; and FIS 97/0588, 00/0021-2, PI061756, and PS0901958 from the Instituto de Salud Carlos III; by grants FIS-PI041436 and FIS-PI081151 from the Spanish Ministry of Health; by grant 1999SGR 00241 from the Generalitat de Catalunya–Consell Interdepartamental de Recerca i Innovació Tecnològica; by grant NEWGENERIS FP6-2003-Food-3-A-016320 from the European Union Sixth Framework Programme; by the Conselleria de Sanitat Generalitat Valenciana; by the Beca de la IV Convocatoria de Ayudas a la Investigación en Enfermedades Neurodegenerativas de la Caixa; by contract QLK4-CT-2000-00263 from the European Commission; and by Fundación Roger Torné.
Role of the Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Silvia Fochs, Anna Sànchez, Maribel López, Nuria Pey, and Muriel Ferrer assisted in contacting the families and administering the questionnaires. Jonathan Whitehead provided useful comments. We thank all the participants for their generous collaboration. A full roster of the Infancia y Medio Ambiente project investigators is available (http://www.proyectoinma.org/presentacion-inma/listado-investigadores/en_listado-investigadores.html).