Adolescents' screen time, sleep and mental health: literature review

Systematic review summarising the published experimental and longitudinal evidence on adolescent screen time, sleep and mental health.


Research Question 1: To what extent does adolescents' mobile device screen time impact on sleep outcomes?

Through our literature search it became evident that the number of research outputs on the links between adolescent electronic media use and sleep increased substantially over the last decade from 26 publications in 2007 to 455 publications in 2018 (Figure 3). Despite the increasing interest in this topic in the recent years, the majority of research was limited to cross-sectional studies and studies that explored the association between television screen time and sleep outcomes. Therefore, most studies were not eligible for inclusion in this review (see Figure 2). Only 9 studies met the inclusion criteria for RQ1.

Figure 3. Primary studies returned by year for RQ1

Figure 3. Primary studies returned by year for RQ1

The following paragraphs describe the characteristics of the nine studies included in this literature review, their quality and the findings. The characteristics of included studies and the findings are summarised in Appendix Table E.

Description of studies

Study design, geographical location and setting

Seven studies were longitudinal observational studies (15, 17-22) and two offered an intervention to all study participants (that is, a one group before-after study design without control group (16, 23)). Two of the longitudinal observational studies used the same cohort of adolescents (21, 22). The longitudinal studies primarily tracked some form of mobile phone use (e.g. time spent on device, being cybervictimised, social media use) and collected data on sleep outcomes (e.g. duration, quality) (15, 17, 19-22). Bartel et al (16) conducted an intervention restricting adolescents' screen use before bedtime and measured school night sleep habits. Werner-Seidler et al (23) assessed the feasibility and acceptability of a smartphone application (app) designed to improve sleep and mental health outcomes in young people.

Studies were conducted in the USA (15, 17), Australia (16, 21-23), New Zealand (18), Canada (19), and Switzerland (20).

Most studies collected data through self-report questionnaires, surveys, or self-report time logs (15, 16, 18-22). Garett et al (17) used self-report questionnaires and the number of tweets participants made. Werner-Seidler et al (23) gave participants an app and had them report on their experiences using it as well as conducting in-depth semi-structured interviews.

Sample size and participant characteristics

Sample sizes ranged from 50 to 26,205 participants. One study had a very large sample of 26,205 participants (19). Two studies had fewer than 100 participants (16, 23). The remaining six studies had between 500 and 2200 participants (15, 17, 18, 20-22).

Ages ranged from 10-28 years. Two studies did not report the age range and only provided the mean age of 14.4 years (21) and 14.3 years (20). Patte et al (19) reported neither the age range nor mean age of their sample (n=26,205), but stated that participants were secondary students in Canada, grades 9-12 (typically age 14-17 years. The two Vernon et al. studies (21, 22) used the same cohort of recruited students.

All studies had a majority of female participants ranging from 52-83% of the samples (15-23). Two studies did not report ethnicity (16, 23). Schweizer et al (20) reported on Swiss nationality but not race or ethnicity. The participant sample in Garett et al (17) was predominantly Hispanic (30%), Asian (28%), and White Non-Hispanic (22%). The remaining five studies had participant samples that were White/European majority (56-71%) ethnicity (15, 18, 19, 21, 22).

Exposures to mobile device and screen time

Mobile device screen time. One study measured mobile device screen time as their exposure using self-report questionnaires (19). Questions included how much time participants spent texting, messaging, emailing, or talking on the phone. Some questionnaire items measured time spent in hours and others in units of time. Importantly, the study only used overall time spent on the mobile device (i.e. not just a smartphone) and did not examine whether the phone was used and when as detailed in the paragraph below.

Mobile phone use. Three studies measured mobile phone use (16, 20, 22). Two of these focussed on mobile phone use at bedtime or night time (16, 22). Vernon et al (22) asked students at what time of night they messaged or made calls and offered increments of time they could choose (e.g. 10-11pm). Bartel et al (16) asked students to record the time, in hours and minutes, they spent on their phones on school nights. Schweizer et al (20) measured student ownership of smartphones across a two-year span using self-report questionnaire data.

Social media use. Garett et al (17) and Vernon et al (21) measured social media use in secondary students and college students, respectively. Garett et al (17) tracked Twitter use over a semester by downloading all participant tweets and retweets. Using a machine learning model, they categorised the tweets by five emotions: fear, anger, love, joy, or neutral. Vernon et al (21) measured problematic social media use using a 4-item scale. Participants were asked to respond to statements such as, 'If I can't access Facebook/Myspace/Bebo, I feel moody and irritable' using a Likert-type response (i.e. completely disagree to completely agree).

Adverse digital communication. Three studies explored possible negative effects of digital communication using questionnaire data (15, 18, 19). Barber et al (15) asked participants about telepressure, i.e. participant views on negative social interaction using technology like phones. They used statements like the following: 'It's hard for me to focus on other things when I receive a message from someone' and participants responded using a Likert-type response (e.g. 1=strongly disagree; 5 = strongly agree). Jose et al (18) used two questions that were averaged, to assess cybervictimisation, or hurtful acts perpetrated through electronic text or pictures. Patte et al (19) assessed cybervictimisation through a single question about whether the participant was bullied in the last 30 days.

Smartphone application. There was only one study that used an app that sought to improve sleep outcomes (23). The app promoted good bedtime behaviours that promote good sleep quality (i.e. sleep hygiene) practices through six training lessons, a sleep tracking function, reminders to begin wind-down, and sleep tips.

Outcomes

All studies reported on some aspect of sleep. Four studies measured sleep duration (hours and minutes) (16, 19, 20, 23) or sleep adequacy (e.g. has participant had at least 7 or 8 hours in the last week?) (18). Four studies measured sleep quality using a range of questions and scales (17, 21-23), for example, 'In the previous 2 weeks, how often have you felt tired or sleepy during the day'. Barber et al (15) assessed sleep hygiene using the 13-item Sleep Hygiene Index with statements like, 'I go to bed at different times from day to day' and 5-point scale measuring agreement. Werner-Seidler et al (23) had many measures for sleep outcomes besides sleep duration and sleep quality, including a survey for insomnia, night awakenings, being refreshed after sleeping, sleep medication, time in bed, and sleep efficiency.

Quality of studies

Table 1 shows the rating for the quality of studies. All studies had a high risk of bias in at least one quality item. Across the studies, methodological flaws occurred in the assessment of media device screen time or use (i.e. performance bias) and assessment of sleep outcomes (i.e. detection bias) because it was self-reported in most studies. Jose et al (18) appeared to be of highest quality with selection bias and attrition bias rated as low risk of bias and high risk of bias in only one quality item. The study with the lowest quality rating was Werner-Seidler et al (23) with four quality items rated as being of high risk of bias.

Table 1. Quality assessment of included studies with sleep outcomes

Image showing Table 1. Quality assessment of included studies with sleep outcomes

Findings

Table 2 summarises the findings on the association between adolescent mobile screen time and sleep outcomes. The findings described below must be interpreted with caution since most are based on studies of poor quality. This means that findings are likely to be biased, in that they are not representing the true relationship between mobile screen time and sleep.

Table 2. Summary findings for RQ1

Exposures

Number of studies

Number of participants + ages

Outcomes

Summary findings

Mobile device screen time

1

N=26205

Age: 9-12 graders

Sleep duration

-/+

Mobile phone use

3

N=1755

Age: 14-16 years

Sleep duration

Sleep quality

Bedtime/sleep onset

Sleep problem

-

-

-/+

-/+

Social media use

2

N=1071

Age: 14 & 18 years

Sleep quality

-

Adverse digital communication

3

N=28625

Age: 10-19 years

Sleep hygiene

Sleep duration

-

-

Smartphone app

1

N=28-34

Age: 12-16 years

Insomnia

Sleep quality

Sleep onset

Night awakening

Sleep duration

Sleep refreshingness

Time in bed after waking up

Use of sleep mediation

+

+

+

+

+

+

-/+

-/+

-/+ No association; + positive association; - negative association

Mobile device screen time

One study (19) assessed the relationship between time spent talking on the phone, surfing the internet, texting, messaging or emailing and meeting the recommended amount of at least 8 hours sleep/night in 9-12 graders (n=26,205). Contrary to what the authors expected, Patte et al (19) found no association between these types of screen use and sleep duration over the four-year study period (Appendix Table E).

Mobile phone use

Bedtime mobile phone use. Two studies assessed the association between bedtime mobile phone use and sleep outcomes. Bartel et al (16) tested whether instructions to stop using the mobile phone one hour before bedtime had any effect with 63 adolescents aged 16 years on average (Appendix Table E). Adolescents stopped using mobile phones 80 minutes earlier on average after one week of taking part in the intervention. Before the intervention, they stopped using the phone at 22:19 pm. After the intervention they stopped using the phone at 20:59 pm. This was statistically significant and indicated a large effect size (Appendix Table E). Two sleep-related outcomes were also statistically significantly associated with the intervention. Participants turned off the lights 17 minutes earlier and slept 21 minutes longer compared to baseline indicating a small effect size. On the other hand, findings suggested that adolescents did not go to bed significantly earlier (4 minutes earlier) and did not fall asleep any quicker: 20.9 minutes before the intervention versus 19.9 minutes after the intervention. Vernon et al (22) found that night-time mobile phone use (i.e. sending or receiving text messages or phone calls after bedtime) and poor sleep quality underwent positive linear growth over time in 1101 adolescents aged 13.5 years on average. In other words, longer mobile phone use after bedtime was associated with lower sleep quality at 1-year and 2-year follow-up (Appendix Table E).

Smartphone ownership. Assuming that ownership of a mobile phone implies use of the mobile phone, we included a third study under the exposure mobile phone use. Schwietzer et al (20) assessed the relationship between smartphone ownership and two sleep outcomes (sleep duration and sleep problems) in 591 adolescents aged 14 years on average. Participants were classed as smartphone owners (owned a smartphone at baseline and 2-year follow-up; n=383), New-owners (acquired a smartphone over a duration of 2 years; n=153), and Non-owners (did not acquire a smartphone; n=55).

Adolescents in all three groups decreased their school day sleep time between baseline and follow-up assessment by 32 minutes (Owners), 41 minutes (New-Owners) and 37 minutes (Non-owners). Although the decrease in school day sleep time is larger for New-Owners and Non-owners, mobile phone owners continued sleeping for the shortest amount of time 2-years later (7.28 hours). In comparison, the average school day sleep time was 7.54 hours for New-owners and 8.0 hours for Non-owners. The difference in sleep time at follow-up between Owners and Non-owners was statistically significant. Smartphone owners were also statistically significantly more likely to have sleep problems at baseline than adolescents who did not own a mobile phone at baseline (i.e. Non-owners and New-owners). The proportion of adolescents reporting sleep problems at baseline was 35.2% (Owners), 19.8% (Non-owners), and 15.4% (New-owners). However, there were no statistically significant differences in the proportion of adolescents reporting sleep problems between groups at 2-year follow-up. The proportion of reported sleep problems was 33.7% (Owners), 33.6% (New-Owners), and 23.4% (Non-owners). In summary, this study suggested that the ownership of a mobile phone alone is not linked to reporting of sleep problems but phone ownership of at least 2 years appeared to be linked to shorter sleep duration.

Social media use

Two studies assessed the relationship between using social media and sleep quality. Vernon et al (21) assessed the relationship between use of Facebook, Myspace, or Bebo and sleep quality in 874 adolescents aged 14 years on average. Garett et al (17) assessed the relationship between Twitter use and sleep quality in 1st year undergraduate students (n=197) aged 18 years on average.

Self-reported baseline scores for sleep quality on 5 point scales were 2.76 in Vernon et al (21) and 3.08 in Garett et al (22) indicating neither poor nor good sleep quality. Both studies reported a decline in sleep quality over time. A higher degree of problematic social media use was associated with poorer sleep quality over time ((21), Appendix Table E). Late-night tweeting (2:00 am – 6:00 am) on weekdays was statistically significantly linked to poorer quality of sleep. Late-night tweeting on weekends and evening tweeting (7 pm-2 am) any day of the week were not associated with sleep quality (17). When considering the emotional states of the tweets, categorised as fearful, angry, joyful, loving, or neutral, only the association between fearful tweets and lower sleep quality reached statistical significance (Appendix Table E).

Adverse digital communication

Telepressure. Telepressure, defined as experiencing pressure to socially engage using a mobile phone, was measured by researchers using six questions that were scored based on a 5-point scale (e.g. 1=strongly disagree; 5 = strongly agree). Higher scores indicated higher levels of telepressure. The baseline average score (Appendix Table E) measuring telepressure for all 241 participants was 2.84 and at the 5-9 week follow up assessment the score was 2.78. Assessment of sleep hygiene at the 5-9 week follow up showed an average score of 2.61 on a 5-point scale with higher scores indicating poorer sleep hygiene. There was a weak statistically significant positive correlation between baseline telepressure and poor sleep hygiene at follow-up and multiple regression analysis confirmed that an increase in telepressure was associated with poorer sleep hygiene. A subgroup analysis of employed versus unemployed college students found that for employed students, increased telepressure was statistically significantly associated with poor sleep hygiene, whereas for the unemployed students there was no association between telepressure and sleep hygiene. (Appendix Table E). This may suggest that staying connected to one's social network using digital media may be more detrimental to the sleep of college students with additional employment obligations (15).

Cybervictimisation. Two studies assessed the association between cybervictimisation and sleep adequacy, defined as meeting the recommended 8 hours/night of sleep one year and two years later (18), and over a period of 4 years (19).

Jose et al (18) reported that adolescents aged 10-15 years met the sleep recommendations on 5.23 nights/week on average. After one year the average was 5.07 nights/week and after two years 4.93 nights/week. After taking sex, ethnicity and age into consideration, there was a statistically significant negative association between cybervictimisation and meeting the recommended amount of sleep of at least 8 hours/night. This means, more frequent incidences of cybervictimisation were linked to fewer nights sleeping 8 hours or more. This association lasted up to two years after experiences of cybervictimisation occurred (Appendix Table E). These findings are consistent with Patte et al (19) who assessed the likelihood of meeting the sleep recommendations after experiences of cybervictimisation in a large cohort of 26,205 adolescents in 9-12th grade. After taking sex, ethnicity, and school grade into consideration, adolescents who newly experienced cybervictimisation in the last 30 days of follow-up had a reduced likelihood of sleeping at least 8 hours/week by 18%.

Smartphone application

One study (23) assessed the preliminary effects of a smartphone app on nine different sleep outcomes: insomnia, sleep quality, sleep-onset latency, night-time awakenings, sleep refreshingness, use of sleep medication, total sleep time, time in bed, and habitual sleep efficiency (Appendix Table E). Of the 50 adolescents who initially took part in the intervention, not all provided outcome data after the intervention had finished. Complete data across outcomes ranged from 68% to 58% obtained from 29 to 34 individuals. The findings from the study presented below should be interpreted with caution due to small sample sizes and poor follow-up rates.

Findings indicated improvements in seven out of the nine sleep outcomes (Appendix Table E). From baseline to post-intervention follow-up, there was a small decrease in insomnia severity scores. However, this decrease did not result in a change of the insomnia severity category and participants remained in the 'subthreshold insomnia' category. Sleep quality improved on average with participants indicating fewer sleep difficulties after using the app. Participants using the app decreased their average time it took to fall asleep by 21 minutes (pre-intervention=72min; post-intervention=51 min). Participants woke less frequently during the night reducing the number of night-awakenings to an average of 0.87 times. Total sleep time improved by 33 minutes from 7 hours 40 minutes to 8 hours 13 minutes. This increase meant that participants met the sleep recommendations of at least 8 hours of sleep per night after using the app. Findings of the study also suggested improvements in adolescents' perception of how refreshing their sleep was on a scale ranging from 1 (=exhausted) to 5 (=very refreshed). The average baseline score was 2.37 points and the follow-up score was 2.78 points. Participants' habitual sleep efficiency improved by 5.5% from 80.1% to 85.6%). Time in bed, which is the time between waking in the morning and getting out of bed, changed minimally to an average of one minute less. There was no change in the proportion of days sleep medication that was used before and after the intervention.

RQ 1 Summary

The nine included studies provided findings for five different types of mobile device exposure in adolescents and ten different sleep outcomes. In total, 16 unique exposure-outcome relationships were assessed across the nine included studies. It became evident that there were only one or two studies available for each exposure-outcome relationship. Findings were mixed depending on the type of exposure and sleep outcome. Table 2 provides a summary overview of the body of evidence on the different exposure-outcome relationships.

Mobile phone use (especially after bedtime) and cybervictimisation, but not overall time spent engaging in mobile phone activities, was linked to lower sleep duration (including meeting the sleep guidelines). Preliminary findings from a small, poor quality study showed that using a smartphone app that teaches about the importance of consistent sleep and wake times, and recommended bedtimes, was linked to improved sleep duration. Sleep quality was also positively influenced by using the smartphone app and negatively influenced by mobile phone use in general and social media use in particular. None of the included studies assessed the link between time spent on mobile screens and adverse digital communication and sleep quality. Experiencing pressure to socially engage using a mobile phone (i.e. telepressure) was associated with poor sleep hygiene. No other mobile device exposure was related to sleep hygiene. In terms of sleep onset, stopping phone use one hour before bedtime was not linked to earlier sleep, whereas using the smartphone app was linked to earlier sleep onset. The app intervention was also associated with improved insomnia, night-awakening, and feeling refreshed after a night of sleep. Those three sleep outcomes were not assessed in other studies included in the review. There was no evidence for an association between any of the included mobile device exposures and sleep problems, time in bed after waking up and use of sleep medication.

Our confidence in the validity of the observed associations is limited for the following three reasons: (i) small number of studies for each exposure-outcome association, (ii) all but one study (19) had a small number of participants recruited, and (iii) the quality of included studies was at unclear or high risk of bias. Therefore, the available evidence on the association between adolescents' mobile device screen time/use and sleep is incomplete.

Contact

Email: socialresearch@gov.scot

Back to top