Association between electronic media use and sleep habits: an eight-day follow-up study

Abstract

Previous research suggests that single use of television, computer or game console may be associated with sleep problems. In practice, youngsters have a multiple rather than a single use of electronic media (EM) every day. This study examined the association between electronic stimulation throughout the day, the final evening activity every day of the week, EM availability in the bedroom and sleep problems in 332 middle-school children (girls = 53%; mean age = 12.9). Cell phones (odds ratio [OR] = 2.5 [1.20–5.38]) and MP3 players (OR = 2.5 [1.13–5.70]) were found to predict sleep problem. An evening screen time of one hour or more is associated with a higher risk of sleep problems (OR = 3.10 [1.24–7.78]), particularly going to bed late (OR = 3.4 [1.59–7.271]) and difficulty waking up (OR = 2.15 [1.01–4.6]). Possession and pattern of EM use by adolescents are associated with impaired sleep quality and late bedtimes, indicating the need to control adolescents' screen time.

Keywords::

• adolescents
• sleep problems
• electronic media
• sleep
• screen time

Introduction

Sleep disorders and sleepiness have been associated with numerous negative health outcomes among adolescents, including increased risk of mood disorders (Carskadon, Acebo, & Jenni, 2004; Millman, 2005) and poor school performance (James, Kristjansson, & Sigfusdottir, 2011; Shin, Kim, Lee, Ahn, & Joo, 2003; Wolfson & Carskadon, 2003). Adolescents with disturbed sleep report more psychological distress (Dahl & Lewin, 2002; Vignau et al., 1997). Studies of the need for sleep during puberty have demonstrated that adolescents needed about nine hours and 15 minutes of sleep per night and that this need did not change between the ages of 10 and 17 (Carskadon et al., 1980). However, secondary school children very seldom sleep this long, and many of them experience restricted sleep (Dorofaeff & Denny, 2006; Roberts, Roberts, & Xing, 2011), which has an impact on their daytime functioning (Dewald, Meijer, Oort, Kerkhof, & Bögels, 2010; Pizza et al., 2010). Moreover, inadequate sleep has been found to be strongly linked to obesity in both children and adults (Patel & Hu, 2008; Van Cauter & Knutson, 2008) and associated with difficulty controlling emotions, and with mood disorders or behaviour problems such as aggression and violence (Clinkinbeard, Simi, Evans, & Anderson, 2011; Harvey, 2011; Kamphuis, Meerlo, Koolhaas, & Lancel, 2012).

Given the growing body of evidence for an association between disturbed sleep and impaired adolescent functioning, more attention needs to be directed toward identifying the underlying causes. Adolescents spend one-third of their life sleeping, which indicates the importance of sleep for development and health. In western industrialised countries today, in addition to non-electronic activity (non-EA) such as reading, the use of electronic media (EM) (television [TV], computers, game consoles, cell phones, MP3 players, etc.) takes up a large part of children's waking time and evening activities. For example, the proportion of TV users in the evening among adolescents aged 12–18 years is over 80% (Calamaro, Mason, & Ratcliffe, 2009).

The relationship between EM and sleep patterns has been a subject of debate for some time. In a longitudinal study of 759 adolescents with mean ages of 14, 16 and 22 years, Johnson and colleagues suggested that extensive TV viewing during adolescence may be associated with an increased risk of sleep problems in late adolescence or early adulthood (Johnson, Cohen, & Kasen, 2004). Owing to its longitudinal design, this study presents the strongest evidence to date that TV viewing is causally related to sleep problems. Dworak, Schierl, Bruns, and Struder (2007) showed that excessive TV and computer exposure affect children's sleep patterns. In particular, playing computer games resulted in prolonged latency of sleep onset, more sleep time in stage two, and less slow-wave sleep as a percentage of total sleep time in subsequent sleep. Using game consoles before bedtime has been linked to increased sleep latency, later bedtime and shorter sleep duration (Oka, Suzuki, & Inoue, 2008; Weaver, Gradisar, Dohnt, Lovato, & Douglas, 2010). The negative effects of EM suggested by these studies are thus a matter of considerable concern.

In practice, youngsters have multiple use rather than a single use of EM every day, and little is known about the concurrent effects on sleep patterns of visual stimulation throughout the day, evening electronic activities all week, and having EM in the bedroom. The aim of the present study was to investigate whether EM use by adolescents during their everyday life could influence their sleep quality over the period of a week. We also investigated whether any particular type of EM had a greater effect on adolescents' sleep quality. Based on previous studies, we hypothesised that EM-related evening activities would be more closely linked to sleep disorders and to late bedtime than other activities. We report the results of an eight-day follow-up study that tested several hypotheses regarding the links between EM variables and sleep patterns.

Method

Participants

Participants were adolescents attending five middle schools in the city of Tours (France) and its suburbs (one class per grade: sixth, seventh, eighth and ninth). To reproduce the distribution of middle schools between Tours and its suburbs (58% vs. 42%) reported by the local council (in charge of middle schools), three middle schools were randomly selected from the list of schools in Tours and two from the list of schools in the suburbs.

A preliminary questionnaire was completed by 454 students who agreed to provide some sleep-related data for eight days. Of this group, 122 were excluded from the final sample because they failed to return their sleep diary or because of too many missing data. No differences were observed between these 122 adolescents and those in the study with regard to gender (χ² = 3.47, p = not significant [ns]), grade (χ² = 0.67, p = ns), proportion with working fathers (χ² = 0.87, p = ns) and proportion with working mothers (χ² = 3.21, p = ns). Consequently, the excluded students should not adversely affect the representativeness of the final sample.

The final sample thus consisted of 332 adolescents (response rate = 73%). The mean age of this group was 12.9 years (standard deviation = 1.25), with nearly equal proportions of girls (53%) and boys (47%). Among these adolescents, 84.8% had a working mother and 92.2% had a working father. According to the French data from the last ‘Health Behavior in School-aged Children’ study (Godeau, Navarro, & Arnaud, 2012), in France 85.2% of middle-school children have a working mother and 97.2% have a working father, which is comparable with our sample.

We have no data regarding the ethnic origin of the participants, since the French agency for the defence of individual liberties (Commission Nationale des Informations et des Liberte´s, CNIL) considers this to be a sensitive subject and forbids the gathering of such data for people under the age of 18.

Measures

Three documents were used in this study: a preliminary questionnaire, a sleep diary, and the Karolinska Sleepiness Scale.

The preliminary questionnaire was specifically created to evaluate whether the participants had EM in their bedroom (TV, computer, game console, radio, MP3 player, cell phone or WiFi router), and their utilisation throughout the day (in hours). Internet access at home was also investigated, as well as parental occupation.

The sleep diary was used to record bedtimes, waking times and subjective signs of sleep quality over a week (from Sunday evening to Friday = school week; from Friday evening to Sunday = weekend). We examined signs of poor sleep quality by means  of four variables: tiredness (on a scale of zero to three; 0–1 = no tiredness; 2–3 = adolescent considered to be tired), time taken to fall asleep (on a four-point scale: less than 15 minutes, between 16 and 30 minutes, between 31 and 60 minutes, more than 1 hour – based on previous research by Billiard & Dauvilliers [2011], we considered more than 30 minutes to reflect difficulty falling asleep), nocturnal awakenings (zero, one, more than one: one indicated multiple nocturnal awakenings), and difficulty waking up (on a scale of zero to three; 0–1 = no difficulty; 2–3 = adolescent considered to have difficulty waking up) (Table 1). We added an additional variable to the traditional sleep diary; namely, evening activities before going to bed. These activities are known to affect bedtime and sleep (Cain & Gradisar, 2010). We took the final activity and classified it as follows: non-EA (e.g. reading, drawing, writing); and EA (e.g. watching TV, playing games on a console, using the computer, listening to music, telephoning) (Table 2).

Table 1 Sleep-related variables classification scheme.

Sleep-related variable considered	Response choices	Cut-off point used to show sleep problem
• In the sleep diary
	• 0
	• 1
Tiredness	• 2
	• 3	Daytime tiredness^(3) = response 2 or 3 reported three times a week or more
Time to fall asleep	• ≤ 15 minutes
	• 15 < t ≤ 30 minutes
	• 30 < t ≤ 60 minutes• < 60 minutes	Difficulty falling asleep^(3) = more than 30 minutes reported three times a week or more
	• 0
Nocturnal awakenings	• 1
	• < 1	Multiple nocturnal awakenings^(3) = more than 1 reported three times a week or more
	• 0
Difficulty waking up	• 1
	• 2
	• 3	Difficulty waking up^(3) =  response 2 or 3 reported three times a week or more
Bedtimes	• ≤ 10:00 pm
	• >10:00 pm →	Mean bedtime during the study period after 10:00 pm (late bedtime)
• Through the Karolinska Sleepiness Scale
Sleepiness at 8:00 am, 10:00 am, 2:00 pm and 4:00 pm	Reported on a nine-point scale during a single day:
	• 0 → 4
	• 5 → 9 →	Daytime sleepiness

Table 2 Electronic media-related variables classification scheme.

EM-related variable	Response choices or response to report		Criteria used to create new EM-related variables
• In the preliminary questionnaire
	• TV• Computer• Game console	• Radio• MP3 player• Cell phone	Less than four EM devices vs. four devices or more
EM available in the bedroom
	• before 8 pm (response in hours)		ST ≥ 3 hours throughout the day vs. ST < 3 hours throughout the day
Use of a screen-based EM device throughout the day	• after 8 pm (response in hours)		Evening ST ≥ 1 hour (after 8:00 pm) vs. evening ST < 1 hour (after 8:00 pm)
• In the diary
Final activity carried out before going to bed	Adolescents had to enter this activity in the diary for each day of the study		Non-EA (e.g. reading, drawing, writing) vs. EA (e.g. watching TV, playing game console) ⇓
			• Participants with electronic activity three times a week or more were classified in the EA group
			• Participants with non-electronic activity three times a week or more were classified in the non-EA group

Next, we assessed daytime sleepiness at school using the Karolinska Sleepiness Scale (Akerstedt & Gillberg, 1990). We illustrated it and verbally designated nine alertness states (compared with only five in the original version) to make it easier for the youngest participants to understand. Only scores ranging from one to four indicate a verbally alert state, and for this reason pupils with a score above four were classified as sleepy (Table 1).

Procedure

While all the schools agreed to cooperate in this study, the participation of the children was voluntary. Parents were informed and were sent a passive consent form (whereby they indicated if they did not want their children to take part) via their son/daughter. In France, passive parental consent meets the guidelines of the national agency for the defence of individual liberties (CNIL).

The study took place in 2009, at the beginning of the second term. The preliminary questionnaire was completed anonymously in the classroom. A research assistant gave instructions on how to respond, reading the items aloud. Then the diary was handed out, some instructions were given, and each participant completed an example. They were all asked to complete the diary for the same consecutive eight days, from Tuesday to Tuesday. On the second Tuesday, they were asked to bring the diary back and estimate their sleepiness at school (Karolinska Sleepiness Scale) at 8:00 am, 10:00 am, 2:00 pm and 4:00 pm. The three documents were completed anonymously.

The protocol was submitted to and approved by the school health services of the local education authority in the département of Indre-et-Loire (France) and by the schools' governing board.

Analysis

In the analysis, several response options were dichotomised. Sleep difficulty could be categorised according to different criteria such as frequency (Blais, Morin, Boisclair, Grenier, & Guay, 2001). As suggested by the International Classification of Sleep Disorders-2 (American Academy of Sleep Medicine, 2005) and the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2000), we used the cut-off point of ‘three times or more a week’ to indicate a sleep problem. Thus, children with difficulty getting up, feeling tired, having difficulty falling asleep, experiencing multiple nocturnal awakenings three times a week or more were respectively classified in the following groups: ‘difficulty waking-up⁽³⁾’, ‘daytime tiredness⁽³⁾’, ‘difficulty falling asleep⁽³⁾’, and ‘multiple nocturnal awakenings⁽³⁾’ (⁽³⁾ refers to three times a week or more). When at least one of these poor sleep quality aspects (occurring three times or more a week) or daytime sleepiness, or having a mean bedtime during the study period after 10:00 pm was reported, we considered that the participant had a sleep problem (Table 1).

In line with previous studies dealing with similar data (Oka et al., 2008), participants who reported an EA or non-EA in their diary three times a week or more were classified as EA or non-EA, respectively (Table 2).

Screen time (ST) was estimated through the preliminary questionnaire by asking how many hours were spent a day (before and after 8:00 pm) using a computer, watching TV and playing video games. The time of 8:00 pm was chosen because it is two hours before the expected bedtime in France allowing about nine hours of sleep (the start school in France generally entails children waking up at about 7:00 am). Hours per day of these ST activities were added up and summed to create an overall ST score = daytime ST (before 8:00 pm) + evening ST (after 8:00 pm) (Table 2).

The findings are presented in three sections: EM possession and use by middle-school children per grade, sleep habits and sleep quality signs per grade, and analysis of the differences in sleep habits/quality for students in different groups on several EM parameters: multiple electronic items in the bedroom (four or more EM) versus few electronic items in the bedroom (less than four EM), EA versus non-EA, ST of three or more hours per day versus ST less than three hours per day, and ST of one or more hour in the evening (after 8:00 pm) versus ST less than one hour in the evening (after 8:00 pm) (Table 2).

First, the univariate effects were examined using the chi-square test on STATA^®. Then we examined the association between EM parameters and sleep quality/habit variables using logistic regression analysis models. Odds ratios (OR) and 95% confidence intervals (CIs) were used to examine the associations. Statistical significance was established at p < 0.05.

Results

Electronic media possession and use

Sixty-eight per cent of sixth-graders had a music player or radio in their bedroom (Table 3). This proportion increased with age: 76% in seventh grade, 83% in eighth grade, and 84% in ninth grade (p < 0.05). A similar pattern was observed for cell phones (p < 0.001): 36% in sixth grade to 69% in ninth grade. Cell phones were kept in the bedroom at night by 92% of the children who owned one. There was a tendency for MP3 player availability in the bedroom to increase with age: 73% in sixth grade to 85% in ninth grade (p = ns). By contrast, having a game console in the bedroom decreased with age (p < 0.05): 73% in sixth grade to 46% in ninth grade. More than one-third of participants had a TV or computer in their bedroom. Less than 15% of children had a WiFi router in their bedroom. Seventy-one per cent had four or more EM devices in their bedroom.

Table 3 Electronic media available in adolescents' bedroom per grade (n = 332).

	Students' grade
EM in bedroom	Sixth	Seventh	Eighth	Ninth	p value (χ^2 test)
Music player	68 (54)	76 (64)	83 (73)	84 (67)	< 0.05
Cell phone	36 (29)	48 (40)	68 (60)	69 (55)	< 0.001
TV	34 (27)	49 (41)	43 (38)	33 (26)	ns
Game console	71 (57)	70 (59)	57 (50)	46 (37)	< 0.01
Computer	36 (29)	42 (35)	42 (37)	40 (32)	ns
MP3 player	73 (58)	80 (67)	81 (71)	85 (68)	ns
WiFi router	5 (4)	14 (12)	18 (12)	15 (12)	ns

Note: Data presented as % (n). Sixth grade, ∼11–12 years old; seventh grade, ∼12–13 years old; eighth grade, ∼13–14 years old; ninth grade, ∼14–15 years old.

Forty-nine per cent of our sample had more than three hours of ST per day. No relationship was found between ST and age. After 8:00 pm, 47% of children had more than one hour of ST. There were significantly fewer children in the sixth grade reporting one hour of evening ST (p < 0.05).

The final evening activities of 69% of the participants were electronic three times or more per week. There were no age differences.

Bedtime and sleep quality

The youngest adolescents went to bed earlier, both during the school week and at weekends. During the school week, the youngest (sixth grade) went to bed one hour earlier than the older children (ninth grade). All the participants went to bed later on Sundays than the other evenings of the school week. Taking the week as a whole, the children went to bed latest on Friday and Saturday, ranging on average from 10:56 pm for pupils in sixth grade to 00:11 am for pupils in ninth grade. The older adolescents woke up later at the weekend (9:32 am in ninth grade vs. 8:40 am in sixth grade on Saturday, and 10:08 am in ninth grade vs. 9:10 am in sixth grade on Sunday), whereas waking up time through the school week was about 7:00 am for all grades (Figure 1).

Figure 1 Bedtimes and wake-up times during a week per grade. Note: 6th, ∼11–12 years old; 7th, ∼12–13 years old; 8th, ∼13–14 years old; 9th, ∼14–15 years old.

Adolescents in the seventh grade slept less than nine hours and 15 minutes the first school-week night (from Sunday to Monday). In the eighth and ninth grades, children slept less than nine hours and 15 minutes every school-week night.

During the study period, one-third of pupils in the sixth and seventh grades had difficulty waking up⁽³⁾ in the morning. More of the older children had this difficulty: 55% of children in ninth grade and 41% of those in eighth grade (p < 0.01). Daytime tiredness⁽³⁾ was reported by 28% of the whole sample. More than 20% had difficulty falling asleep⁽³⁾. Less than 5% of participants experienced multiple nocturnal awakenings⁽³⁾. About 23% were drowsy during the school day. Sixty-two per cent went to bed after 10:00 pm, mainly the oldest (p < 0.001) (Table 4).

Table 4 Signs of poor sleep quality and bedtime per gender and grade.

	Signs of poor sleep quality
Sleep problems	Difficulty waking up^(3) (n = 288)		Daytime tiredness^(3) (n = 287)		Difficulty falling asleep^(3) (n = 248)		Multiple nocturnal awakenings^(3) (n = 256)		Daytime sleepiness^a (n = 313)		Mean bedtime during the study period after 10 pm (n = 280)
	%	p	%	p	%	p	%	p	%	p	%	p
Gender
Girls	40	ns	31	ns	20	ns	3		27	< 0.05	70	< 0.01
Boys	39		25		23		4		18		55
Grade
Sixth	30	< 0.01	26	ns	22	ns	3		12	ns	33	< 0.001
Seventh	33		27		25		3		28		21
Eighth	41		26		19		3		25		32
Ninth	55		33		18		4		24		87

Note: Total n values vary due to self-report. Chi-square was used to test the relationship between sleep problems: bedtime (mean bedtime ≥ 10:00 pm throughout the study period), sleep quality (difficulty waking up, daytime tiredness, difficulty falling asleep, nocturnal awakenings, daytime sleepiness) and categorical variables (gender, grade). Sixth grade, ∼11–12 years old; seventh grade, ∼12–13 years old; eighth grade, ∼13–14 years old; ninth grade, ∼14–15 years old. ^aKarolinska Sleepiness Scale score above four at 8:00 am, 10:00 am, 2:00 pm or 4:00 pm.

Associations between electronic media possession and use, bedtime, and sleep quality

Sleep problems (difficulty waking-up⁽³⁾, daytime tiredness⁽³⁾, difficulty falling asleep⁽³⁾, multiple nocturnal awakenings⁽³⁾ or mean bedtime during the study period after 10:00 pm) were reported by 261 participants. In the logistic regression model predicting sleep problems, having a cell phone (OR = 2.5; p < 0.01) or an MP3 player (OR = 2.5; p < 0.05) were significant predictors (Table 5). More than one hour of evening ST was associated with a higher risk of sleep problems (OR = 3.1; p < 0.01).

Table 5 Effects of electronic media possession and use on the occurrence of sleep problems in adolescents.

	Sleep problem
	OR (95% CI)	p value
EM in bedroom (%)
Music player (77)	0.550 (0.234–1.296)	ns
Cell phone (55)	2.538 (1.209–5.326)	< 0.01
TV (40)	1.320 (0.575–3.032)	ns
Game console (61)	0.546 (0.247–1.206)	ns
Computer (40)	0.570 (0.247–1.314)	ns
MP3 player (79)	2.533 (1.138–5.640)	< 0.05
≥ 4 EM in bedroom (71)	0.740 (0.271–2.014)	ns
Evening activities (%)
EA (69)	2.053 (0.831–5.071)	ns
Non-EA (52)	1.988 (0.798–4.953)	ns
Screen Times (ST) (%)
≥ 3 hours/day (49)	1.134 (0.426–3.024)	ns
≥ 1 hour after 8 pm (48)	3.104 (1.315–7.319)	< 0.01

Note: Association between EM parameters and sleep problems were examined using logistic regression analysis models. When an adolescent reported having at least one sign of poor sleep quality three times or more a week (cf. Table 4), we considered he/she had a sleep problem.

Logistic regression models for predicting each of the variables defined as a sleep problem by EM possession and use revealed a consistent association between evening EA and daytime tiredness⁽³⁾ (OR = 3.376 [CI = 1.160–9.821]; p < 0.01). Adolescents reporting ST of one or more hour after 8:00 pm had a higher risk of difficulty waking up⁽³⁾ (OR = 2.151 [CI = 1.005–4.6]; p < 0.05) and having a mean bedtime during the study period of 10:00 pm or later (OR = 3.398 [CI = 1.588–7.271]; p < 0.01). If they had a cell phone, children had a higher risk of daytime tiredness⁽³⁾ (OR = 2.943 [CI = 1.424–6.078]; p < 0.01) and they had a mean bedtime during the study period of 10:00 pm or later (OR = 2.402 [CI = 1.229–4.694]; p < 0.01). Finally, there was a consistent association between difficulty waking up⁽³⁾ and having a game console (OR = 0.457 [CI = 0.237–0.883]; p < 0.05).

Discussion

This eight-day follow-up study confirms that EM possession and use are associated with children's sleep quality and late bedtime. As expected, EA was associated with tiredness. On the other hand, non-EA was not significantly correlated with any signs of poor sleep quality. Three main possible mechanisms could explain this result. First, it has been hypothesised that the stimulating content of TV or video game might keep children awake (Van den Bulck, 2000). Noise or light associated with EM use is more likely to increase the level of arousal and make children go to bed later (Cain & Gradisar, 2010). Adolescents themselves identified these activities as a reason for delayed or inadequate sleep (Owens, Stahl, Patton, Reddy, & Crouch, 2006). Secondly, an effect of visual stimulation on melatonin onset is conceivable (Cain & Gradisar, 2010); this hormone, whose level rises in the evening close to the usual bedtime, is suppressed by light (Lewy, Wehr, Goodwin, Newsome, & Markey, 1980). Some studies have shown that even room normal light levels can have a suppressing effect on human endogenous melatonin production (Boivin & James, 2002; Zeitzer, Dijk, Kronauer, Brown, & Czeisler, 2000). Recent research has shown that Brazilian adolescents living in houses without electric lighting had earlier sleep and melatonin onset than those with electric lighting (Peixoto, da Silva, Carskadon, & Louzada, 2009), supporting the idea of the effect of artificial light on sleep. Furthermore, Crowley, Acebo, and Carskadon (2007) suggested that light sensitivity changes during development, with pubertal youngsters being more sensitive to light than non-pubertal children or adults. It is conceivable that light emanating from visual EM could delay melatonin and thus sleep onset, particularly among adolescents. However, our results do not support this hypothesis as we failed to show a significant association between having a visual device in the bedroom (TV, computer or game console) and sleep problems. Finally, it is possible that adolescents just used EM instead of sleeping (Cain & Gradisar, 2010).

A sizeable proportion of our sample (almost one-half) used a screen for one hour or more just before going to bed. This result is very close to those recently presented by the National Sleep Foundation (2011) in which one-half of 13–18-year-olds were also concerned. In our study, evening ST emerged as a predictor of sleep problems, particularly going to bed late (after 10:00 pm). Late bedtime could explain the correlation between evening ST of one or more hour and difficulty waking up⁽³⁾. The importance of limiting ST is underlined by previous findings that extensive screen viewing may be associated with substance abuse, aggressive and risky behaviours, and eating disorders (Carson, Pickett, & Janssen, 2011; Strasburger, Jordan, & Donnerstein, 2010). Based on these findings, the American Academy of Pediatrics has recommended that young people should not have more than one to two hours of media time per day (Bar-on et al., 2001). The present findings support the further recommendation that adolescents should not spend more than one hour in the evening in front of a screen, either TV, computer or game console.

Among EM devices available in the bedroom, only cell phones and MP3 players were associated with sleep problems. More surprisingly, our results revealed no association between TV, computer or game console and sleep problems. This is in line with the findings of Van den Bulck who showed that cell phone use after lights out is very prevalent among adolescents and is related to increased levels of tiredness. He also found that there are no safe doses or times for using the cell phone for text messaging or calling after lights out (Van den Bulck, 2007). We hypothesise that cell phones and MP3 players, which are less subject to parental control than TV, computers or game consoles, could thus be prolonged until late in the night to the detriment of sleep. With the development of smartphones, which are visually attractive and can be used as all-in-one cell phone, computer and music player, more and younger children will become competent and frequent users of this new technology out of parents' control. This is likely to contribute to a continuing increase in sleep disorders, and become a public health issue. Sleep and EM use by adolescents should thus be the subject of future studies.

As with any other research, the current study suffers from some limitations. Owing to the cross-sectional nature of the study, it was not possible to determine causal relationships. We did not control for baseline sleep problems. Moreover, no information was provided with regard to the pubertal status of participants, which previous studies have linked to sleep variables such as delayed bedtime (Sadeh et al., 1995). It is possible that pubertal status would be a confounding factor if more biologically mature subjects reported higher media use and had more sleep problems. Another obvious limitation is that we had no objective data from physiological studies on disturbed sleep. While such data would be useful, self-reports and interview-based methods remain the measures of choice in community surveys. Moreover, it should be noted that there is evidence that subjective measures of children and adolescents' sleep patterns are correlated with objective measures of disturbed sleep (Carskadon, Acebo, Richardson, Tate, & Seifer, 1997).

Based on the findings of our study, we hypothesise that it is the way children use EM that is deleterious to their sleep, particularly during adolescence. Given that sleep disorders have been found to be associated with a number of negative health outcomes, the use of behavioural modification strategies to improve sleep, such as reducing the amount of ST, may have a wide range of health benefits. Future research should investigate more extensively the longitudinal association between EM use and sleep problems throughout life, as well as the sleep-related outcomes of intervention protocols to reduce EM use.

Our study may help clinical and public health interventions. Our findings suggest that by restricting the amount of time that adolescents spend in front of a screen, they may be able to reduce the likelihood of the onset of disturbed sleep signs. Parents, healthcare professionals and other caregivers may be able to help young people avoid developing sleep problems by encouraging them to avoid excessive ST and evening electronic activities. It is important that they do not encourage EM in the bedroom, as adolescents' pattern of EM use is inappropriate and affects their sleep.

Additional information

Notes on contributors

Violaine Kubiszewski

Violaine Kubiszewski is a PhD student in the Psychologie des Ages de la Vie Laboratory at the University François Rabelais of Tours, France.

Roger Fontaine

Roger Fontaine, PhD, is professor in the Department of Psychology, University François Rabelais of Tours, France. He is also the director of the Psychologie des Ages de la Vie Laboratory at the University François Rabelais of Tours, France.

Emmanuel Rusch

Emmanuel Rusch, MD, PhD, is a public health professor at the University François Rabelais of Tours, France. He is also an associate professor in the Psychologie des Ages de la Vie Laboratory.

Eric Hazouard

Eric Hazouard, MD, is the chief of the Sleep Medicine Service in Alliance Clinic, Saint-Cyr-sur-Loire, France.