Screen Time and Sleep among School-Aged Children and Adolescents: A Systematic Literature Review

Inclusion and exclusion criteria

We include all studies that investigate the association between any type of screen time and any of the following sleep outcomes: sleep timing, sleep duration, sleep quality, sleep onset latency (SOL), subjective assessment of daytime tiredness or daytime sleepiness, or other reported outcomes including subjective assessment of insomnia symptoms. In order to be consistent with the prior literature review(5), we narrowed the articles down to include any study that contained participants that were primarily between 5 and 17 years old. We did not limit the population to any geographical area or gender. When data from the same cohort were presented in more than one study, we included only the most relevant article. We divided studies into type of media included (e.g., television, computer, mobile phone, video gaming device), allowing each study to be categorized in more than one media type, if appropriate.

This project updates a 2010 literature review(5) that used similar inclusion criteria. In the four years since publication of that paper, we found 31 additional articles meeting the inclusion criteria. All articles that fit the above criteria, including those in the 2010 review, are included in this review (See Table 1).

Television Watching and Sleep

Tables 2a summarizes the studies involving television screens and television watching at bedtime; 32 out of 42 studies (76%) found an adverse association between television watching and adverse sleep outcomes, whereas 10 found no association.

Among the 32 articles that investigated the association between television watching and delayed bedtime or TST, 25 studies (78%) found that television watching was associated with significantly delayed bedtime or shortened TST. For studies that reported a significant negative association between television watching and sleep duration, higher levels of television screen watching was associated with either a larger effect size for shortened total sleep time or higher odds of short sleep duration.

Some of the studies had results with statistically significant associations between television and sleep only when stratified by certain variables. For example, television watching was associated with shorter sleep duration on weekdays compared to weekends(53, 63, 78). Other studies found that older adolescents were more likely to have associations between television and later bedtimes compared to younger children (19, 76) Finally, male study participants were, in some studies, more likely than females to show an association between television watching and shorter sleep duration(56, 66).

Of the studies that we coded as not reporting a significant association between television watching and TST, there are nuances worth discussing. Two studies(24, 25) observed that television watching alone was not sufficient to significantly impact sleep (however, use of multiple screens did have a significant negative association with TST). One study (26) found that sleep duration did not mediate the association between the presence of a television in a bedroom and body weight. Another study (37) did not find an association between television watching and TST across individuals, but reported a positive daily-level association between these two measures within individual adolescents. The study author observed that adolescents watched more television on days where they studied less and were less stressed. These results were derived from a 14-day daily diary which is the longest daily diary prospective study that we catalogued (other studies that utilized a daily diary often used a 2-day daily diary or asked about sleep on a weekday and weekend (e.g.,(72)).

Some studies found that television presence in the bedroom was associated with less sleep, even when reported use of television was not associated less sleep(35, 67). The only study that relied upon objectively measured TST (using actigraphy) did not report a significant association between television watching and TST(62). The lack of association in this study raises important questions about the quality of self-reported and parental-reported sleep data.

Studies that estimated the magnitude of lost sleep time observed bedtime delays within the range of 5 to 10 minutes per hour of television watching (19, 27, 28). These magnitudes can add up when one considers the total number of hours youth spend in front of a television(12, 13). Two separate studies, by the same lead author but on different samples, simply measured whether television watching occurred at bedtime (as a yes/no variable), and both found about a 20 minute delay in sleep time(20, 21).

The evidence for an association between television use and sleep onset latency (SOL) is equivocal: 4 of the 7 studies that examined SOL found a significant association between television use and later sleep onset, with the other 3 studies reported no finding (33, 39, 58). Several studies found an association between television watching and subjective assessment of daytime tiredness(34, 54, 66), while one study found no association(51). Twelve studies included other measures of sleep outcomes and 75% of the studies found significant associations.

Computer Use and Sleep

Table 2b includes only the studies involving computer use. Twenty-nine out of 31 studies (94%) found an association between computer use and at least one of the sleep outcomes, while 2 found no association.

Our literature search yielded 24 scientific articles that specifically analyzed computer use and TST. All studies show shortened TST with computer use. In one study, the magnitude is 0.86 hours (or 51 minutes) less TST for teens that report usually or always using the internet for social reasons at bedtime (20) than for teens who do not report using the internet at bedtime.

All of the 9 studies that looked at delayed bedtime found that youth who use more computer screen time go to bed later. Five of the six studies (83%) that looked at sleep problems found that computer use predicted other sleep problems (20, 33, 51, 83, 84), whereas one study (52) did not find an association between computer use and other sleep problems. While no studies reported on the analysis of the association between computer use and general sleep quality specifically, 3 studies(34, 52, 81) (75% of 4) reported a significant association with subjective assessment of daytime tiredness or daytime sleepiness and only 2 out of 5 studies(33, 49) reported a significant association with SOL. We also found that 5 out of the 6 studies that looked at “other” sleep outcomes found a significant association with computer use(20, 43, 54, 70, 84).

Comparing studies on computer use and sleep is difficult given the wide range of definitions and questions asked about computer time -- for example, some asked about whether the computer was used for social networking, leisure, studying, or video gaming, while other studies grouped all computer use together. Nine of the studies specifically studied computer use at bed time(20-22, 25, 28, 34, 52, 57, 64), 4 of the studies asked about both daytime and bedtime use(53, 73, 78, 84) and the remaining did not differentiate between timing of computer use.

Video Gaming and Sleep

Table 2c presents studies involving video game use. Among these, 18 out of 21 studies (86%) found an association between video games use and sleep patterns, while three found no association. All studies described in this section were included if study authors mentioned some form of screen-based game playing--which will henceforth be referred to as “video gaming.”

Among the 16 articles that examined video gaming and TST or delayed bedtime, 13 articles (81%) reported either a significant negative association with TST or a positive association with delayed bedtime. In several studies, significant results were moderated by age and day of the week. In one study, older study participants were more likely to have video game playing associated with later bedtimes(19). In two studies, weekend video gaming was more likely to be associated with shorter sleep duration whereas this did not hold for weekday game playing(19, 64). Some studies estimated the magnitude of the sleep delay associated with bedtime video gaming as up to 28 minutes(20, 21, 28).

Of the three null findings: one study categorized video gaming as time spent at a game center/arcade (18), which is the only unconventional description of video gaming listed among these studies; and two studies examined low video game use as the unexposed group compared to high video game use (46, 47), rather than comparing the high video game use to a no-use reference category.

For the outcomes of SOL and subjective assessment of daytime tiredness or daytime sleepiness, 4 out of 7 studies (57%) reported a significant positive association between video gaming and higher SOL, while the two studies that looked at subjective assessment of daytime tiredness or daytime sleepiness found a significant positive association with video gaming. Only three studies looked specifically at sleep quality: one found a significant negative association between video gaming and sleep quality(47), whereas two found no significant association(46, 50). Of the five studies that looked at other sleep outcomes, all of them found statistically significant associations with video gaming(19, 20, 49, 51, 75)

Mobile Devices and Sleep

Table 2d summarizes the studies involving mobile devices; among which 15 out of 18 studies (83%) found an association between mobile device use and at least one sleep outcome, while 3 studies found no association.

Of the 12 looking at TST or delayed bedtime, 10 (83%) found a statistically significant association between either shortened TST or delayed bedtime. Of the studies that estimated the amount to which sleep duration was shortened, one study of UK adolescents found that cell phone use at bedtime shortened sleep by 21 minutes(21) and another found that among pre-teens from the UK, usually using a mobile device at bedtime was associated with a shortened sleep duration of around 45 minutes(20). Two studies failed to find the association between mobile phone use and sleep duration. The first collected data from Taiwanese adolescents and found that “problematic cell phone use” did not predict short sleep duration at the p<0.05 level, but the odds ratio (OR=1.295, 95% CI 0.99-1.694) shows that the magnitude was reasonably large and was very close to reaching that level of significance(83). The second study (28) found no significant association between use of a phone to chat or text at nighttime with sleep duration, but they did find a significant negative association between having a cellular phone (or other handheld communication device) in the student's bedroom and TST. Two other studies that measured presence of a cell phone but did not measure its use found a significant association between having a cell phone in the bedroom and adverse sleep outcomes(24, 43). These and other findings on the effects of ownership or presence of devices are not included in the Tables because we wanted to focus on use of the devices.

Two studies estimated the association between mobile phone use and SOL with one finding a positive association(49) and one finding no association (39). This second study analyzed both mobile phone use ownership and SOL in a sample of Japanese schoolchildren in 2004(39). Given that mobile technology in 2004 was significantly different than the capabilities of modern smartphone technologies, there may be major differences in the impacts of mobile technologies on sleep between then and now. We speculate that this change in the technology utilization patterns over time may account for the differential in findings between the two studies.

Only one study investigated mobile phone use and sleep quality, and it found a negative association(60). All 5 studies that investigated mobile phone use and subjective assessment of daytime tiredness or daytime sleepiness found a negative association(51, 60, 61, 77, 79). There were 8 studies that looked at mobile devices and other sleep outcomes, of which 6 found significant associations (21, 42, 49, 51, 60, 70), whereas 2 studies found no associations(74, 83).

Unspecified Screen Time and Sleep

Table 2e lists the 11 articles that clustered multiple types of screens into a single measure of screen time, which we term unspecified screen time. 91% of these found an adverse association between unspecified screen time and at least one of the sleep outcomes. All studies described in this section were included if study authors used an aggregate category of screen time and did not stratify by type of screen. Thus, while this category describes use of multiple screens, usage may not necessarily be simultaneous.

Among the 9 articles that examined aggregate screen time and TST, 8 articles (89%) reported a significant negative association with TST. Reported effect sizes range up to 45 minutes if multiple screens are present in a child's bedroom(24). One study(32) reported differential results by age group, where younger children under 13 years of age were more likely to have aggregate screen time associated with shorter sleep duration than older children, a finding in contrast to the stronger effects found among older teens for television viewing(19, 76). The study that reported a null finding(44) cited a limitation in aggregating screen time into one category.

Of the two remaining studies, one study(51) investigated screen time of middle-school aged children and found that evening screen time of one hour or more was associated with a 3.4-fold increased risk (p<0.01) of going to bed at 10 PM or later. One other study(36) reported a significant association between aggregate screen use and later sleep onset.

Causality Difficult to Ascertain

Given the use of observational data to study population behavioral patterns that occur among real-life children and adolescents, all but the 3 smallest of these studies(46, 50, 82) lack an experimental design in which one group is randomly assigned into a screen time exposure group while the other one has no screen time. As a result, while the rest of the studies provide a relatively good portrait of the screen time and sleep behaviors of youth, along with how these behaviors are associated statistically with each other in real populations, they do not provide enough data to confirm whether the association is causally linked.

There are several ways in which the observed associations may not indicate a causal impact of screen time on sleep outcomes. First, reverse causality might occur because youth who need less sleep or have trouble sleeping for other reasons choose to spend their time in front of more screens simply because they have more time to do so (34). Further, there could be other confounding factors that are associated with both increased screen time and decreased sleep duration that suggest a spurious association. Examples of such potential confounding factors include having low physical activity and being overweight. However, there are ways to improve the likelihood of a causal link using observational data. The first is to have the measure of exposure (screen time) precede the measurement of the outcome (sleep), such as using a daily diary over multiple days, as was done in only 9 of the 67 studies(19, 33, 37, 46, 50, 51, 55, 72, 82). This allows for screen time to be measured the day or night before sleep time and address some key issues of reverse causality. Another advisable approach to address causality is to adjust for as many potentially confounding variables as possible, provided the sample size allows for it. While most studies adjusted for basic covariates such as age and gender, several studies did an excellent job of adjusting for a wide range of characteristics including family structure(53), family income(53), sleep environment, physical activity, bedtime routine, and child health, among other variables(20, 21). Another approach would be to test for within-individual differences on nights when they do and do not have exposure to screen time, as was done in one study(37). The gold standard would be to test the effects of an intervention or experiment that reduces screen time through random assignment as was done in the three trials included in our review(46, 50, 82).