Evening Smartphone Use Tracks With Later Bedtimes and Shorter Sleep in Adolescents
Passive smartphone sensing is drawing a sharper line between screen time and sleep than self-report ever could. In a JAMA Pediatrics intensive longitudinal cohort study, evening phone use in adolescents was associated with later sleep onset, later wake time, and shorter total sleep time on a night-to-night basis, with the strongest signal emerging after 10 pm.
The study moved beyond the familiar limitations of recall and cross-sectional survey work. Instead of asking youths how much they thought they used their phones or slept, investigators paired passively sensed evening smartphone data with Fitbit-derived sleep measures and examined both within-person and between-person associations. That design matters because it asks a more bedside-relevant question: when the same adolescent uses the phone more than usual on a given night, does sleep measurably shift?
The answer was yes.
Night-to-night phone use mapped to measurable sleep loss
The cohort included 425 adolescents, with a mean age of 14.8 years and just over half female, contributing 7,591 person-nights of data. Participants were Android users in the Adolescent Brain Cognitive Development Study and had at least 14 days of concurrent EARS application data and Fitbit-measured sleep. Data collection ran from late 2020 through early 2024, and the analysis used within-between decomposition in multilevel models to separate an adolescent’s own nightly deviations from differences across adolescents.
On nights when adolescents used their phones more than their own average after 10 pm, they slept a mean of 20.05 minutes less and fell asleep about 64.73 minutes later. Those numbers are clinically meaningful because adolescent sleep debt rarely arrives as a single dramatic event; it accumulates in small losses that compound across school nights, weekend recovery sleep, and circadian drift.
The adjusted within-person model sharpened the association further. Each additional minute of smartphone use after 10 pm above a participant’s own average was associated with sleep onset 0.87 minutes later, sleep offset 0.37 minutes later, and total sleep time 0.43 minutes shorter. These associations were statistically precise, with 95% confidence intervals that stayed tightly bounded away from null for all three outcomes.
The clock mattered more than the device itself
Evening use before midnight was not treated as a single undifferentiated block. Investigators examined smartphone use after each hourly cutoff from 8 pm to midnight, as well as use in the 5 to 30 minutes before sleep onset. The pattern pointed to timing rather than mere exposure. The later the phone remained active, the more sleep was pushed back.
That fits adolescent sleep biology. The teenage circadian system already trends toward delayed sleep phase, with endogenous sleep onset naturally drifting later during puberty. Add cognitively and emotionally activating screen engagement at the end of the evening, and the sleep window narrows further. Bright light is one plausible contributor, but the behavioral mechanism may be broader: notifications, scrolling, social reinforcement, and the difficulty of disengaging from an interactive device can all prolong pre-sleep wakefulness. A passive evening routine is easier to interrupt than an active one.
The study also assessed sleep outcomes beyond duration, including onset, offset, and both sustained and total wake after sleep onset. Those measures matter because sleep quality is not just about how long a teenager stays in bed; fragmentation and delayed initiation can shape daytime alertness, school performance, mood regulation, and next-day attention. Even modest shifts in sleep timing can ripple into chronic schedule misalignment during the school week.
Within-person analysis makes the finding harder to dismiss
Between-person comparisons often mix behavior, temperament, household routine, and baseline sleep differences into one estimate. A youth who routinely stays up later may also differ in family structure, school demands, or evening autonomy. The within-person approach used here reduces some of that noise by asking whether an individual adolescent sleeps differently on heavier-use nights than on lighter-use nights.
That internal comparison strengthens the clinical relevance of the results. It suggests the association is not explained solely by the fact that certain adolescents are habitual short sleepers or heavy phone users. Nightly deviation from personal baseline carried its own signal, and that signal was concentrated after 10 pm.
The practical implication is straightforward for pediatric counseling, though the mechanism is not purely punitive. Evening phone restriction may help because it targets a modifiable pre-sleep behavior at the point when circadian delay is already most likely to intensify. For adolescents, a later bedtime often means less total sleep, not just a shifted schedule. On school nights, that can translate into chronic sleep restriction even if weekend catch-up sleep partially masks the deficit.
Objective sensing strengthens the case, but not every question is closed
Fitbit-derived sleep estimates and passive smartphone sensing are both stronger than self-report, yet neither eliminates all uncertainty. Wearable sleep algorithms infer sleep from movement and physiology rather than direct polysomnography. Passive phone sensing captures use patterns but not every aspect of media exposure, emotional arousal, or whether the device was the last stimulus before lights out. The study also focused on Android users with complete data, which narrows the analytic sample and may limit generalizability.
Still, the design is difficult to ignore because it aligns behavior with physiology across thousands of nights. The data suggest that the nightly price of late smartphone use is paid in delayed sleep onset first, then in reduced total sleep time, with wake timing shifting later as well. For adolescents already living near the margin of adequate sleep, even a 20-minute loss can be the difference between a tolerable school night and cumulative sleep deprivation.
The study leaves a few important clinical questions open. It does not prove that turning off phones earlier will reliably improve sleep in every adolescent, and it does not establish which component of evening use matters most: light, content, social stimulation, or the simple act of staying engaged. It does, however, show that later-night use tracks with objectively worse sleep on the same person’s own nights, and that is the sort of signal clinicians can use when advising families about bedtime routines, device boundaries, and adolescent sleep hygiene.