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.


Appendix Table E: Summary of findings on the relationship between mobile device screen time/use and sleep outcomes

Reference

Study characteristics

Sample characteristics

Exposure/Intervention Description

Outcome description

Findings

CI=confidence interval

OR=odds ratio

SD=standard deviation

SE=standard error

Mobile device screen time

Patte 2018

Canada

Longitudinal cohort study

Follow-up: 4 years

N=26,205

Age: grade 9-12

Baseline mean age: not reported

Sex: 55% female

Ethnicity: 71% Caucasian, 2.5% Black

Screen time:

Survey which asked participants the average time per day that they spent: "watching/streaming

TV shows or movies," "playing video/computer games," "talking on the phone," "surfing the internet," "texting, messaging, emailing," and "doing homework."

Sleep duration:

Assessed by asking how much time in hours (0–9) and minutes (0, 15, 30, 45) participants usually spend sleeping per day. Responses were classified as either "meets recommendations" (≥8 h) or "insufficient sleep" (< 8 h)

Logistic regression (adjusted for (gender, grade, race/ethnicity)

Talking on the telephone: OR= 1.01 (95%CI 0.98 to 1.03)

Surfing the internet: OR = 1.01 (95% CI 1.00 to 1.02)

Texting, messaging, or emailing: OR= 1.00 (95%CI 0.99 to 1.01

Mobile phone use

Bartel 2018

Australia

Single arm pre-post intervention design

Follow-up: 2 weeks

N=98 (63 included in analysis)

Age: 14-18 years

Baseline mean age: 16.3 years

Sex: 83% female

Ethnicity:

Other: Android phone users only

Pre-bed mobile phone use on school nights: Adolescents were given individualized phone stop times, 1 hour before bed for one school week. At the end of the baseline week, average bedtimes were used to calculate the time which each adolescent needed to stop their mobile phone use, for the school week only (Sunday–Thursday night). This was 1 hour prior to their average baseline weekday bedtime. Instructions were sent to individual email addresses.

Participants installed free screen On/Off Logger Lite' application which records when phone screen is turned on. App was available for Android users only.

An online sleep diary used to collect sleep outcomes for two consecutive weeks; only weekday data were used.

Bedtime:

Baseline: 22:17 pm (SE 0:07)

Follow-up: 22:13 pm (SE 0:08)

Non-significant pre-post difference, F=0.46, p=0.50

Cohen's d = 0.06

Light out

time:

Defined as turning the light off with the intention of sleeping,

after going to bed; obtained from the sleep diary; unit=clock time

Baseline: 22:57 pm (SE 0:07)

Follow-up: 22:40 pm (SE 0:08)

F=9.00, p=0.01

Cohen's d = 0.30

Sleep onset latency:

Sleep diary, minutes

Baseline: 21.0 min (SE2.2)

Follow-up: 19.9 min (SE 1.9)

F=0.34, p=0.57

Cohen's d = 0.06

Total sleep time:

Online sleep diary programme

used an algorithm to calculate total sleep time; unit=hours:min

Baseline: 7 hrs :36 mins (SE 0:07)

Follow up: 7:57 (SE 0:08)

F=7.98, p=0.01

Cohen's d = 0.34

Sleep efficacy:

3-item survey:

  • 1. If their sleep during the intervention week was either 'better than usual', 'the same', or 'worse than usual';
  • 2. If they thought their sleep improved due to the intervention ('improved a bit', 'stayed the same', 'got worse');
  • 3. If they thought the intervention was 'highly effective', 'somewhat effective', 'neither effective nor ineffective',or 'ineffective'

Completed surveys by n=29.

45% improved sleep a bit

45% sleep

stayed the same

7% sleep became

worse

7% reported the intervention to be

highly effective

38% be somewhat effective

48% reported it to be neither effective nor ineffective

7% reported it to be ineffective

Schwiezer 2017

Switzerland

Longitudinal cohort study

Follow-up: 2 years

N=591

Age: range not reported

Baseline mean age: 14.3 years

Sex: 50% females

Nationality: 83.5% Swiss

Other: SES 5% below average, 38% above average

Smartphone ownership:

Assessed using an online questionnaire, YES/NO response; answers categories into:

  • - Owners (ownership at baseline and follow-up; n=383)
  • - New owners (ownership at follow-up only, n=153)
  • - Non-owners (those not owning a smartphone at any time-point; n=55)

Sleep duration:

Participants indicated how many hours on average they slept during school days and during weekends/vacation. Minutes are given on a decimal scale.

Sleep problems:

Assessed by a single question: "Over the last six months have you ever had sleep problems?". There were five possible answers dichotomized as 'at least once a week' (at least once a week, most days) and 'others' (never, less than monthly, about once a month). Yes = sleep problems

One-way ANOVA

School days:

Owners 7.28h (SD 0.09) vs Non-owners 8.00h (SD 0.20) p=0.002

Owners 7.28h (SD 0.09) vs new-owners 7.54h (SD 0.09) p=0.104

New-owners 7.54h (SD 0.09) vs Non-owners 8.00h (SD 0.20) p=0.075

Weekend/vacation:

Owner vs New-owner: p=0.10

Owner vs Non-Owner: p=0.94

New-owner vs non-owner: p=0.91

Bivariate analysis comparing Owners vs New-owners vs Non-owners:

Baseline sleep problems [yes]: p<0.001

Owners: 35.2%

New-Owners: 19.8%

Non-Owners: 15.4%

Follow-up sleep problems [yes]: p=0.49

Owners: 33.7%

New-owners: 33.6%

Non-Owners: 23.4%

Vernon 2018

Australia

Longitudinal cohort study

Follow-up: 1 year and 2 years

N=1101

Age: 13-16 years

Baseline mean age: 13.5 years

Sex: 57% female

Ethnicity: 56.9%Caucasian, 7.1% Asian, 2% Aboriginal or Torres Strait Islander, 21.9% other

Other: 44% from lower SES

Night-Time Mobile Phone Use: Students were asked if they had a mobile phone and if they answered yes they were then asked, "At what time of the night do you usually send or receive messages and/or phone calls?" 6 response options: never text or phone after lights out; immediately after lights out; 10–11 p.m.; 11 p.m.–12 a.m.; 12–1 a.m.; 1–2 a.m.; 2–6 a.m.; at any time of the night. Coded on 6-point scale (0-5) as 0 = no mobile phone, 1 = never text or phone after lights out, 2 = immediately after lights out, 3 = before midnight, 4 = after midnight, and 5 = at any time of the night

Sleep quality:

Assessed using a scale which consisted of the mean of eight items drawn from the School Sleep Habits Survey. The sleep scale tapped perceptions about sleep quality and behavior during the previous 2 weeks, and included: "How often have you needed more than one reminder to get up in the morning."

Responses for all sleep items were 1 = never, 2 = once, 3 = twice, 4 = several times, and 5 = every day/night. Higher scores = lower sleep quality

Zero-order correlation

1 year follow-up: r=0.17, p<0.05

2 year follow-up: r=0.16, p<0.05

Social media use

Garett 2018

USA

Longitudinal cohort study

Follow-up: 10 weeks

N=197

Age: 17-20 years

Baseline mean age:

18.1 years

Sex: 60% female

Ethnicity: 29% Hispanic, 27% Asian, 22% White non-Hispanic, 12% black, 10% other

Other: Students had to be active Twitter users, tweeting at least three times a week

Twitter use: All tweets and retweets were downloaded and categorized into five emotion categories: fear, anger, love, joy, or neutral using machine learning model (a Naïve Bayes classifier). The classifier used a bag-of words approach. Monograms that appeared in at least three tweets, bigrams that appeared in at least six tweets, trigrams that appeared in at least three tweets were included. Time of the day and weekday were reported.

Sleep quality:

Assessed using a weekly survey (items not reported). Rating on a 5-point Likert scale (response options not reported)

Regression model (adjusted for sex, ethnicity, academic major, tweets/week)

Weekday:

Evening tweets β =0.189 (SE 0.097), p<0.05

Late night tweets β = - 0.937 (SE 0.352), p<0.01

Weekend:

Evening tweets β = −0.117 (SE 0.08), p= >0.05 (value not reported)

Late night tweets β =-0.413 (SE 0.139), p >0.05 (value not reported)

Weekdays:

Angry tweets β = -0.205 (SE 0.169), p>0.05

Fearful tweets β = -0.302 (SE 0.131), p<0.05

Loving tweets: β = 0.026 (SE 0.138), p>0.05

Joyful tweets: β = 0.105 (SE 0.128), p>0.05

Neutral tweets: β = -0.135 (SE 0.131). p>0.05

Vernon 2017

Australia

Longitudinal cohort study

Follow-up: 1 year and 2 years

N=874

Age: 12-18 years

Baseline mean age: 14.4 years (SD not reported)

Sex: 59% female

Ethnicity: 57.2% were Caucasian, 7.2% Asian, and 1.6% Aboriginal or Torres Strait Islander, 23.3% other

Social media use assessed using the problematic use of social networking scale consisting of 4 items. Items measured the degree to which adolescents invest emotionally in social networking

  • Item 1: "I prefer to spend time on Facebook/ Myspace/ Bebo rather than attend social activities/ events";
  • Item 2: "I use Facebook/Myspace/Bebo as a way of making me feel good";
  • Item 3: "I get into arguments with other people about the amount of time I spend on Facebook/ Myspace/Bebo."
  • Item 4: "If I can't access Facebook/ Myspace/Bebo, I feel moody and irritable".

Sleep quality: Items were adapted from the School Sleep Habits Survey and asked:

During the during the previous

2 weeks, how often have you:

"felt tired or sleepy during the day"; "had an extremely hard time falling asleep"; "had a good night's sleep (reversed)"; "felt satisfied with your sleep" (reversed). Response option were 1 (never), 2 (once), 3 (twice), 4 (several times), and 5 (every day/night). Higher scores = poorer sleep quality

Bivariate correlation:

1 year: r=0.34, p<0.01

2 years: r=0.26, p<0.01

Adverse digital communication

Barber 2017

USA

Longitudinal cohort study

Follow-up: 5-9 weeks

N=241

Age: 18-28 years

Baseline mean age: 19.0 (SD 1.8)

Sex: 58% female

Ethnicity: 64.0% White/European, 15.6% Black/African-

American, 10.5% Latino/Hispanic, 4.6% Asian, 4.0%

Biracial/Multi-racial.

Other: Introductory Psychology course at a 4-year public university

Telepressure: assessed on a 6-item scale asking to rate the extent to which participants agree (1=strongly disagree; 5 = strongly agree) with statements that describe view on social interaction using information-communication technology (e.g. phones, emails).

  • Item 1: It's hard for me to focus on other things when I receive a message for someone.
  • Item 2: I can concentrate better on the tasks once I've responded to my messages. Item 3: I can't stop thinking about a message until I've responded.
  • Item 4: I feel a strong need to respond to others immediately.
  • Item 5: I have an overwhelming feeling to respond right at that moment when I receive a request.
  • Item 6: It's difficult for me to resist responding to a message right away.

Sleep hygiene: Measured using the 13-item Sleep Hygiene Index. E.g. "I go to bed at different times from day to day". Response options ranged from 1 (never) to 5 (always). Higher scores = poorer sleep hygiene.

Bivariate correlation r=0.18, p<0.05

Multiple regression:

b= 0.11 (SE=0.04), p<0.05

Unemployed:

b = 0.01, SE = 0.05, p = 0.918

Employed: b = 0.27, SE = 0.06, p<0.001

No adjustment for confounders

Jose 2018

New Zealand

Longitudinal cohort study

Follow-up: 1 year & 2 years

N=2179 (baseline)

Age: 10-15 years

Baseline mean age: not reported

Sex: 52% female

Ethnicity: 59% New Zealand European, 28%Māori, and 15% other

Cybervictimisation: defined as being a victim of cyber-aggression which is defined as persistent, hurtful acts perpetrated on another individual through electronic text or pictures. Assessed asking two questions:

  • 1. "In the last month, about how often have you received a mean text message from someone?"
  • 2. "In the last month, how often have you been bullied by others online?"

The two items were averaged to produce a single score. Responses ranges from 1 ("never"), 2 ("1 to 3 times"), 3 ("4 to 6 times"), 4 ("7 or more times") to 5 ("almost daily/daily").

Sleep adequacy: Measured using a single survey question "In the last week, on how many nights did you get at least 8 h of sleep? Responses were provided on a scale from 0 to 7 days.

Bivariate correlation:

1 year: r= -0.09, p<0.01

2 years: r= -0.04, p= >0.05 (value not reported)

Regression model (adjusted for sex, age, ethnic group):

1 year: β=−0.05, p=0.008

Averaged over 2 years: β=−0.08, p=0.011

Patte 2018

Canada

Longitudinal cohort study

Follow-up: 4 years

N= 26,205

Age: grade 9-12

Baseline mean age: not reported

Sex: 55% female

Ethnicity: 71% Caucasian, 2.5% Black

Cybervictimisation: assessed using a single question "In the last 30 days, in what ways were you bullied by other students?" Response option: cyber-attacks (e.g. being sent mean text messages or having rumours spread about you on the internet) Response options included: "I have not been bullied in the last 30 days:" YES/NO scale

Sleep adequacy:

Assessed by asking how much time in hours (0–9) and minutes (0, 15, 30, 45) participants usually spend sleeping per day. Responses were classified as either "meets recommendations" (≥8 h) or "insufficient sleep" (< 8 h)

Logistic regression (adjusted for (gender, grade, race/ethnicity)

Adjusted OR=0.82 (95%CI 0.74 to 0.91)

Smartphone application

Werner-Seidler 2019

Australia

pilot study (single arm pre-post intervention design)

Follow-up: 6 weeks

N=50 (baseline)

Age: 12 to 16 years

Baseline mean age: 13.71 (SD 1.35)

Sex: 66% female

Other: With mild insomnia; 94% born in Australia

Sleep Ninja App aiming to teach users about the importance of consistent sleep and wake times, and recommended bedtimes. The structure of the Sleep Ninja app includes six training lessons, a sleep tracking function, recommended bedtimes based on sleep guidelines, reminders to start a wind-down routine each night, a series of sleep tips and general information about sleep. Training sessions were delivered through a chat-bot format where the sleep ninja essentially acts as a sleep coach. Training sessions took approximately 5–10 min to complete.

Intervention duration: 6 weeks (locked sessions thereafter)

Insomnia: Insomnia Severity Index, higher scores=more severe insomnia

β=−4.29 (95%CI −5.63 to 2.95)

Sleep Quality: Pittsburgh Sleep Quality Index, higher scores=poorer quality

β=−1.88 (95%CI −2.85 to 0.90)

Sleep onset latency [time taken to fall asleep]

ß= −0.37 (95%CI

−0.70 to –0.03)

Night-time awakenings [number]

ß=−0.46 (95%CI

−0.81 to –0.11)

Sleep refreshingness [scale from 1=exhausted to 5=very

refreshed]

ß=0.43 (95%CI 0.19 to 0.68)

Use of sleep medication [proportion of days]

ß=−0.01 (95% CI −0.02 to 0.01)

Total sleep time (calculated by subtracting sleep-onset latency, wake after sleep onset

and time between waking and getting up in the morning,

from time in bed)

ß=0.53 (95%CI

0.17 to 0.90)

Time in bed (time between waking in the morning and getting out of bed)

ß=−0.01 minute (95%CI −0.42 to 0.41)

Habitual sleep efficiency

(total sleep time/time in bed)

ß=5.25 (95%CI 1.03 to 9.47)

Contact

Email: socialresearch@gov.scot

Back to top