Scrolling Your Sleep Away: The Effects of Bedtime Device Use on Sleep Among Young Adults with Poor Sleep

Abstract

Background

Many young adults report sleep problems, including insufficient sleep and poor sleep quality. Young adults are heavily reliant on electronic devices, even using them during bedtime with adverse effects on sleep. Given the importance of adequate sleep, the present study examined the daily association between using electronic devices during bedtime and sleep in a diverse sample of young adults with poor sleep.

Method

We analyzed data from a pilot randomized controlled trial in which young adults with poor sleep [n = 46; 84% female; mean age 19.3 (SD = 2.9); 30% Asian, 19% Black/African American, 9% multiracial; 34% Hispanic/Latino] wore an electronic sleep tracking device (Fitbit Charge 3) and completed daily sleep diaries including questions about sleep and bedtime routine for 4 weeks following a behavioral sleep intervention. The effect of bedtime device use on sleep latency—time needed to fall asleep—and sleep duration was estimated by generalized linear mixed models (GLMM), adjusting for weeknights/weekend nights.

Results

Bedtime device use on a given night was significantly associated with shorter self-reported sleep duration (b = − 19.80, p = .011), but not with sleep latency. Concordance between the self-reported and Fitbit-measured sleep variables was low, and bedtime device use was not associated and Fitbit-measured sleep variables.

Conclusion

Using electronic devices before bed negatively affected self-reported sleep duration the following night. This finding highlights the importance of minimizing nightly device use among young adults with poor sleep and suggests that the inconsistency between self-reported sleep and device use warrants further investigation.

Introduction

Insufficient sleep is a common problem among young adults in the USA, with 60% suffering from poor sleep quality and 40% reporting insufficient sleep at least two nights a week [1, 2]. The consequences of poor sleep have been well documented and range from difficulties with cognitive performance to issues with physical health [3]. Young adulthood is a distinct developmental period between adolescence and adulthood, usually defined as occurring between the ages of 18 and 24 years, characterized by new societal challenges and expectations, and the development of healthy behaviors. It is also a period when mood disturbances commonly emerge [4]. Thus, inadequate sleep among young adults is a major health concern.

Prior research indicates that exposure to electronic devices (phones, tablets, computers) is associated with reduced sleep quality and duration [5]. However, research examining the relationship between bedtime device use and sleep has been centered on children and adolescents. In these populations, using devices at bedtime has been associated with negative sleep outcomes, including less sleep, poorer sleep quality, and insomnia symptoms [5, 6]. As a result, the widespread recommendation to establish a screen-free bedtime routine has been aimed mostly at children and adolescents, who are encouraged to sleep in bedrooms free of devices or to turn them off 30 min before bed [7, 8].

Young adults rely heavily on electronic devices and habitually use devices before going to sleep [9]. They average 6–7 h of daily smartphone usage [10], the highest screen time average globally [11]. Using digital devices at bedtime may lead to shorter sleep time [12], increases in the time it takes to transition from full wakefulness to sleep, or sleep latency, which plays a significant role in predicting sleep quality and duration [13]. Using a mobile phone (without a blue light filter) for at least 30 min after lights have been turned off has been correlated with greater self-reported sleep latency, poorer sleep quality, and daytime sleepiness [14]. Young adults who accessed emotionally charged content on their phones before going to bed were likely to report longer sleep latency and sleep disturbance [15]. Increased digital media use within 2 h before bedtime was associated with lower sleep duration and a later bedtime [16]; however, device use in bed was not associated with any sleep outcomes [17]. Despite these findings, the impact of bedtime device usage on young adults remains understudied as most studies have relied solely on self-report measures or focused only on those with no sleep difficulties.

Given these gaps in the literature on the impact of device use on sleep outcomes among young adults, we aimed to examine how bedtime device use affects two dimensions of sleep—total sleep duration and the time it takes to fall asleep—essential outcomes to meet recommended sleep guidelines in young adulthood. Further, we aimed to examine these relationships specifically among young adults with reported poor sleep (compared with most prior studies focusing on “healthy” samples), using both self-reported and objective data to overcome limitations in prior studies. Young adulthood is a critical developmental period in which health behaviors are being established, and thus, identifying the impact of device use on sleep outcomes in a population with high prevalence of device use is important to create future health behavior change interventions.

Methods

This is a secondary analysis of data from a pilot randomized controlled trial of a brief, single-session sleep hygiene intervention [18] which compared whether the intervention was stronger for individual receiving the intervention one-on-one vs. with a supportive companion [18]. As no effects were found, the intervention groups were combined.

Recruitment

Fifty-one undergraduate students were recruited through the subject pool and flyers posted on the campus. Students were eligible if they met the following inclusion criteria: substantial sleep difficulties as indicated by a score > 54 on the PROMIS Sleep Disturbance short form [19]; not taking prescription or over-the-counter sleeping aids; being able to complete assessments in English; and being willing to bring a companion (family member, friend, partner) to the study. Participants received one course credit or a $5 Amazon gift card for completing the baseline survey, and one course credit or a $10 Amazon gift card for each study component (intervention, follow-up session, returning sleep diaries that were at least 80% complete).

Intervention

All participants received the single-session Sleep Treatment and Education Program for Students (STEPS) [20], a 30-min scripted PowerPoint presentation about the negative impact of sleep disturbances on physical health and academic performance delivered in person. STEPS includes evidence-based suggestions to improve sleeping habits: establish regular wake time, avoid alcohol, caffeine, and nicotine several hours before bed, limit naps, turn down lights and minimize noise as part of their bedtime ritual to avoid physiological arousal, and use the bed only for sleep and sex (e.g., avoid other activities such as studying or arguing on the phone). Take-home handouts of the suggestions were provided per the original intervention.

Procedure

Data were collected between September 2019 and December 2021. The study was paused from March 2020 through August 2021 because of the COVID-19 pandemic. No differences were found between those who participated at the two data collection periods. Pre-intervention, participants completed an online baseline survey of sociodemographic characteristics and sleep habits. After the intervention was delivered, they were given paper sleep diaries to complete twice daily for 4 weeks, upon waking and before going to sleep and a wearable sleep tracking device (Fitbit Charge 3) [21] to wear continuously, day and night, except during showers, for the 4-week period. Participants were asked to charge the device every 5 days, download the Fitbit phone application, and sync the device daily. After 4 weeks, participants returned to the lab to return the diaries and Fitbits and complete follow-up questionnaires online. Research assistants accessed participants’ Fitbit accounts twice weekly to monitor charging and sync status, sending email reminders if needed to prevent data loss.

Measures

Sleep and device use measures were obtained from the 4-week sleep diaries and actigraphy data. Sociodemographic characteristics were obtained at baseline. Gender, race, ethnicity, and age were assessed with items from the National Health Interview Survey [22].

Sleep

Self-reported sleep included questions adapted from the Pittsburgh Sleep Diary [23] and completed every morning. Sleep latency — the time taken to fall asleep — was assessed by the following item: “Last night it took me ___ minutes to fall asleep.” Sleep duration was measured through one item: “Last night, I slept for ___ hours.”

Objective sleep measures were collected with the Fitbit Charge 3, a wearable device that has been validated [21] and shown to be in good agreement with polysomnography as well as self-reported sleep metrics [24, 25]. Sleep latency was measured as time awake while in bed; Sleep duration was the total number of minutes of sleep recorded.

Use of Electronic Devices at Bedtime

Frequency of electronic device use at bedtime was measured by an item from the nightly sleep diary: “In the hour before going to sleep, my bedtime routine included ___.” These free text responses were coded as a dichotomous variable (yes/no, yes = 1) indicating whether or not each participant explicitly stated using an electronic device (“using my phone,” “watching TV or a screen,” “working on laptop”) or described an activity that implied the use an electronic device (“watching a show,” “gaming,” “Tik Tok”) at bedtime.

Data Analyses

SPSS Version 27 was used to analyze the data. The effect of bedtime device use on sleep latency and sleep duration was estimated by generalized linear mixed models (GLMM), adjusting for weeknights/weekend nights. Days of missing data were excluded from analyses. The covariance matrix structure was specified as first-order autoregressive or “AR(1).” The final model tested for the fixed and random effects of device use on each of the four outcomes (self-reported and objective sleep latency; self-reported and objective sleep duration), adjusting for whether it was a weeknight or a weekend night.

Results

The analysis sample included 46 college students, as sleep diary and/or Fitbit data were missing for five participants. The mean age was 19.3 years (SD = 2.9) and most participants (84%) identified as female. The sample was racially and ethnically diverse, with 30.2% of participants identifying as Asian, 18.6% as Black or African American, and 9% as multiracial; over a third (34.1%) reported their ethnicity as Hispanic or Latino (see Table 1).

Table 1.

Sociodemographic characteristics of participants

Characteristics	n	M (SD) or %
Age		19.2 (2.9)
Gender
Female	37	84.1%
Male	6	13.6%
Race
Asian	13	30.2%
Black/African American	8	18.6%
White	8	18.6%
Multiracial	4	9.3%
Other	10	23.3%
Ethnicity
Hispanic/Latino	15	34.1%
Not Hispanic/Latino	29	65.9%
Employment
Working full-time (> 35 h/week)	3	6.8%
Working part-time	23	52.3%
Living off campus	43	97.7%
People in household		4.1 (1.5)
Living with family	40	88.9%
Living alone	2	4.4%
Living with friends/roommates	1	2.2%

Note. Three participants did not provide data on gender, race, and household. Two participants did not provide data on race/ethnicity or employment

The average number of daily diaries completed was 24.3 out of a possible 28 days (86.7%); thus, the models below analyzed a total of 1117 days of data. While all participants scored > 54 on the PROMIS Sleep Disturbance short form [17] during eligibility screening, in the diaries, they reported a lower average sleep duration of 412.4 min (SD = 118.4), or nearly 7 h per night across the daily diary period, and an average self-reported sleep latency of 28.7 min (SD = 40.4) per night. The intra-class coefficient (ICC) for sleep duration was 0.189, indicating that 18.9% of the total variance in self-reported sleep duration was due to within-person fluctuations across days. The intra-class coefficient (ICC) for sleep latency was 0.599, indicating that 59.9% of the total variance in self-reported sleep latency was due to within-person fluctuations across days. In contrast, the Fitbit-measured sleep duration averaged 391 min (SD = 116.3) or about six and a half hours per night and Fitbit-measured sleep latency averaged 53.1 min (SD = 29.6) per night. Thus, the Fitbit data indicated about half an hour less sleep and twice as long to fall asleep as the self-reports.

Effect of Electronic Device Use with Nightly Sleep Duration

We observed a significant negative association between using a device and self-reported sleep duration on that same night. In a model including only device use (i.e., without covariates), using a device before bedtime significantly predicted shorter sleep duration that night (b = − 18.02, p = 0.022). In the full model with covariates (Table 2), adjusting for whether the night was followed by a weekday or weekend day (b = 35.90, p < 0.001), device use remained a significant predictor of shorter sleep duration (b = − 19.80, p = 0.011). In other words, when participants used an electronic device before bedtime, they reported, on average, a reduction in sleep duration of 20 min. Including the covariates, participants reported sleeping for an additional 36 min on nights followed by a weekend day, presumably when they had more flexibility for sleeping longer. We also tested for the random effect of device use on sleep duration but found that the strength of the effect did not vary among individuals and model fit was not improved. We computed parallel models for Fitbit-measured sleep duration but no association was found between bedtime device use and Fitbit-measured sleep duration (b = − 0.95, p = 0.90) (see Table 2).

Table 2.

Multilevel model results predicting daily self-reported sleep duration and sleep latency adjusted for whether the night was followed by a weekday or weekend day

	Self-report			Fitbit-measured
	b	95% CI	p	b	95% CI	p
Sleep duration
Intercept	415.17	401.55, 428.80	< .001	384.48	371.07, 397.90	< .001
Device use (referent, no = 0)	− 19.80	− 34.98, − 4.61	.011	− 0.95	− 16.23, 14.33	.90
Day of week (referent, weekday = 0)	35.90	20.60, 51.20	< .001	27.93	12.18, 43.68	< .001
Sleep latency
Intercept	28.32	22.26, 34.38	< .001	51.20	47.33, 55.06	< .001
Device use (referent, no = 0)	− 0.28	− 4.53, 3.97	.89	1.88	− 2.26, 6.02	.37
Day of week (referent, weekday = 0)	2.89	− 1.10, 6.89	.16	1.86	− 2.27, 6.00	.38

Note. CI, confidence intervals

Effect of Electronic Device Use on Nightly Sleep Latency

In a model including only device use (i.e., without covariates), using a device before bedtime did not predict longer sleep latency that night, whether self-reported (b = − 0.24, p = 0.911) or Fitbit-measured (b = 1.91, p = 0.365). In the full model (Table 2), adjusting for whether the night was followed by a weekday or weekend day (b = 2.89, p = 0.16), device use did not predict longer self-reported sleep latency (b = − 0.28, p = 0.89): on average, participants reported taking only 3 min longer to fall asleep on nights followed by a weekend day. We also tested for random effect of device use on sleep latency; however, the strength of the effect did not vary among individuals and model fit was not improved. The GLMM (Table 2) for device use and Fitbit-measured sleep latency was not significant (b = 1.19, p = 0.37).

Discussion

Consistent with prior findings on bedtime device use and self-reported sleep outcomes among young adults, this study found that using a device within an hour of going to bed was associated with shortened self-reported sleep duration on the subsequent night, especially on weeknights, and a decrease in self-reported time spent asleep (sleep duration) of almost 20 min per night. We did not find an association between device use and self-reported sleep latency, measured either by self-report or by objective wearable technology.

Bedtime device use was associated with a decrease in self-reported sleep duration on both weeknights and weekend nights; not surprisingly, participants reported sleeping longer on nights followed by a weekend day. However, it is possibly that participants overestimated their sleep time, as we did not see this pattern with Fitbit-measured sleep duration.

Why was sleep latency not affected by device use? Participants may have used blue light filters, or avoided engaging with emotionally charged content, which cause emotional arousal and alertness leading to longer sleep latency and sleep disturbances [15]. There was a greater discrepancy between self-report and wearable technologies than anticipated; Fitbits indicated fewer hours asleep and a longer time to fall asleep. It is possible that participants had poor sleep recall, did not approximate time asleep carefully, or”rounded up” in the diaries. Although this finding could explain the discordance between self-report and objective measures; it underscores the importance of using objective measures in sleep research. It is also of note that while all participants were screened for substantial sleep difficulties using the PROMIS scale [19], they did not appear to have significant issues with sleep latency or duration during the study period.

As part of the intervention, participants were given evidence-based recommendations to improve their sleep. As bedtime device use was high after the intervention, information alone is unlikely to create major changes or fully account for these findings.

Strengths and Limitations

The sample allowed us to examine device use among racial and ethnic minoritized college students, while previous studies have not. The multilevel modeling analyses included over 1000 data points and were sufficiently powered. All participants had poor sleep at study entry, in contrast with past studies; this suggests that an educational intervention was needed.

The study had several limitations including the small sample size characteristic of a pilot study. We did not record the particular activities when using the device (e.g., texting with friends, watching action videos). Similarly, we did not have data on whether device usage occurred in bed specifically, whether the participant sleep in a room alone or with others, or shared a bed; nor did we assess whether participants used blue light glasses or a filter, which may reduce the negative effects of the light emitted from electronic devices [26]. Because paper diaries have no digital timestamp, we are unable to verify whether participants may have filled them in at a later time. It is also possible that depression and anxiety could impact the association between bedtime device use and sleep dimensions, although this sample was low on these symptoms at baseline.

Conclusion

This study provides evidence that using electronic devices before bedtime negatively affects self-reported sleep duration, supporting widespread guidelines to limit device use before bedtime. The findings are congruent with past studies [16]. As sufficient sleep is crucial for the mental and physical health of young adults, researching how sleep may be affected by nighttime habits and creating evidence-based guidelines for device use targeted for young adults with poor sleep may increase sleep quality and quantity.

Data Availability

The study dataset cannot be publicly shared as consent was not sought from participants to allow for the publication of de-identified datasets.