Teen Sleep Timing and Consistency, Not Hours, Tracked Heart Health

It is close to midnight on a school night, and the strip of light under a fifteen-year-old's door has not gone out. The parent in the kitchen has read the week's coverage, which said, in one form or another, that teenagers who sleep well grow into young adults with healthier hearts. That sentence is easy to agree with and almost impossible to act on. It leaves open the only question a household actually faces at that hour, which is which part of the sleep is the part that counts: the total, the hour it starts, or the fact that it starts at a different hour every night.

The study underneath the coverage answers that question more precisely than the coverage did, and its most useful finding is a null one. Published in JACC: Advances in late 2025 and picked up by a fresh round of coverage this month, it separated adolescent sleep into several distinct dimensions and found that some of them predicted later cardiovascular health while the one most people track did not.

Gina Marie Mathew, Lauren Hale, Orfeu Buxton and colleagues drew on the Future of Families and Child Wellbeing Study, a birth cohort of 4,898 children born between 1998 and 2000 in twenty large American cities, deliberately oversampled for nonmarital births. A randomly selected subsample was invited into a sleep sub-study at age 15, and 1,049 adolescents assented to wear a wrist accelerometer for seven consecutive days. Years later, a subset attended an in-person cardiovascular examination as young adults. The analysis reported here is the 307 participants who have both: objective sleep at 15 and a measured cardiovascular profile at a mean age of 22.1.

That sample was 57% female, 52% Black, 23% White and 26% Hispanic or Latino, and contributed an average of 6.8 nights of usable monitoring each. Four dimensions of sleep came out of the actigraphy: total sleep time, sleep timing (the clock hours of falling asleep and waking), sleep maintenance efficiency (the share of time in bed actually spent asleep), and the night-to-night variability of each, calculated as an individual standard deviation across that week.

The outcome was a modified version of the American Heart Association's Life's Essential 8, a 0-100 composite of diet, physical activity, nicotine exposure, body-mass index, blood lipids, blood glucose and blood pressure, with the last four measured rather than self-reported. The sleep component of the standard metric was deliberately dropped, so the outcome could not be partly made of the exposure. The sample averaged 69.68 on that score, with a standard deviation of 13.45. Models adjusted for age-22 sociodemographics and household income, and for the age-15 characteristics that were themselves associated with the later score: body-mass index percentile, breakfast frequency, fruit and vegetable intake, sweetened drinks and vigorous activity.

First, what these adolescents' sleep looked like. Average total sleep time was 6.98 hours, and only 14.3% averaged the eight hours that adolescent guidance calls for. The average night began at 00:28 and ended at 08:13. Sleep maintenance efficiency averaged 90.81%. Bedtimes moved around a great deal: the typical participant's night-to-night standard deviation in sleep onset was 1.37 hours.

In the adjusted models, four of those measures predicted the age-22 score. An hour later in average sleep onset went with a 1.11-point lower score (95% CI: -2.06 to -0.15), and an hour later in wake time with 1.08 points lower (95% CI: -1.96 to -0.21). Each percentage point of higher sleep maintenance efficiency went with 0.66 points higher (95% CI: 0.18 to 1.15). An hour more night-to-night variability in total sleep time went with 1.97 points lower (95% CI: -3.87 to -0.07), and the same swing in bedtime with 1.93 points lower (95% CI: -3.72 to -0.14).

Those are modest effects. On a score whose spread in this sample was 13.45 points, the standardized coefficients ran between 0.11 and 0.14 standard deviations, and the confidence intervals for the two variability measures reach to within a rounding error of zero. What the study documents is a gradient across a population, not a switch in an individual.

Average total sleep time did not predict the age-22 score in any form the authors tested: not as a continuous measure (b = -0.27; 95% CI: -1.77 to 1.22; P = 0.72), not with a curved term allowing very long sleep to behave differently, and not as the clean threshold of eight or more hours a night (b = -0.56; 95% CI: -4.53 to 3.41; P = 0.78). Variability in wake time was also unassociated (P = 0.54), as was variability in efficiency, which landed just outside significance (P = 0.06).

That null is the finding with the most consequence, because duration is the dimension the guidance counts. Life's Essential 8 added age-appropriate sleep duration in 2022 and included no measure of timing, regularity or quality. In this cohort, the components it left out are the ones that carried the association, and the one it included did not.

The authors read that as a possible developmental difference: the well-documented link between sleep duration and later cardiovascular health comes largely from middle-aged and older adults, and may simply emerge after adolescence. There is a plainer arithmetic caution alongside it, visible in the study's own descriptive table. A sample averaging 6.98 hours, with fewer than one in seven reaching eight, contains very little of the long-sleep end of the range, and a comparison cannot find a difference across ground it does not cover.

Cardiovascular health at 22 is not a heart attack; nobody in this sample was near an age where events occur. It is a risk profile, and its value here is that it was measured rather than remembered, and that adolescent sleep was measured by a device rather than recalled on a questionnaire. Both halves of the association are unusually hard-edged for a study of habits.

It also names something a household can observe without instruments. Bedtime consistency across a week is visible in a way that sleep efficiency is not, and in this cohort it belonged to the group of measures that tracked the later score. The authors go on to raise later school start times as a structural lever, on the reasoning that a fixed early wake time paired with a late bedtime is what produces both short weeknights and the weekend swing. That is their inference about policy, and it travels further than their data, which measured seven days of one year in one cohort.

The honest headline here is the negative result. Hours slept is the number families and clinicians count, and in this cohort it was the one dimension that predicted nothing, while when sleep started and how much that hour moved from night to night did. That is worth taking seriously precisely because it is inconvenient: it points at the part of adolescent sleep least under a teenager's control and least captured by the standard metric, and it suggests the guidance that counts only duration is measuring the easiest thing rather than the most informative one. The effect sizes are small and the confidence intervals are wide enough that a replication could shrink them toward nothing, and a week of monitoring in one year of one cohort cannot promise any individual household an outcome. What it can do is redirect attention from a target that showed no signal to one that did, and that is a more useful thing for a study to have produced than another confirmation that sleep is good.