Sleep Duration Among US Adolescents, 1991–2023

Abstract

Objectives:

We examined trends in adolescent sleep across recent decades in the US.

Methods:

Data were drawn from a nationally representative study, Monitoring the Future (MTF), measurement years 1991–2023, which represented cohorts of adolescents born from approximately 1972–2011 (n = 401,160). Outcomes were two self-reported survey items, one addressing sleep duration and another on subjective sleep sufficiency. Age-period-cohort models were estimated and sociodemographic differences in trends were examined.

Results:

Adolescent sleep duration declined with increasing age during every period. Adolescents at every age in the last 10 years were more likely to report inadequate sleep duration compared to teens at those same ages in earlier decades. The period 2021–2023 had the lowest prevalences of getting seven or more hours of sleep at every age (ranging from 37.2% at age 12 or 13 to 22.3% at 18 or 19). Disparities in sleep duration between non-Hispanic Black and Hispanic/Latino adolescents and their white peers, and between teens whose parents were more versus less educated, emerged and/or grew steadily over time. For instance, Black and white adolescents in 1991–1995 were equally likely to report seven or more hours of sleep per night (OR – 0.99 [0.92, 1.07]), by 2023 Black teens were less likely to report this (OR – 0.79 [0.67, 0.93]).

Conclusions:

For decades, adolescents’ sleep had been eroding. It is concerning that youth from marginalized sociodemographic groups appear to be at an even greater risk for profoundly short sleep, given that sleep is a resource that confers advantages for health and development.

Introduction

Insufficient sleep has been an ongoing population-wide challenge for adolescents in the US.^1,2 Further, there are distinct sleep inequities where adolescents that have been racially and ethnically minoritized, including Asian, Black, Hispanic/Latino, and Multiracial adolescents, are less likely than white adolescents to get adequate (8–10 hours per night³) sleep.^4–10 Insufficient sleep during one’s youth is an important problem, as it can have long-term health consequences¹¹ and may set the stage for suboptimal adult sleep patterns.¹² But even during adolescence poor sleep appears to correlate with some suboptimal markers of cardiometabolic health concurrently^13–17 and later into young adulthood.¹⁸ It can also adversely impact adolescents’ school performance and psychosocial functioning.^19,20

During adolescence, many obstacles to sleep surface, including social pressure to delay bedtime, as well as intensifying school, peer, and family demands.^21,22 Viewed with a 24-hour activity cycle perspective,²³ when adolescents begin to devote additional time to one or more waking activities, this can displace time spent sleeping. Additionally, many teens must conform to rigid schedules (and in particular the very early school start times of many US secondary schools²⁴) that are misaligned with adolescent circadian biology, which presents a challenge to attaining enough sleep.

Teen sleep deficits appear to be worsening. Prior work, which examined US adolescents’ sleep duration in US-representative Monitoring the Future (MTF) data from 1991 to 2012, revealed a general trend of sleep duration decreasing over time.²⁵ Only a small fraction of adolescents reported sleeping seven hours or more per night, and this proportion declined with adolescents’ increasing age. Teens who attended secondary school in more recent years had shorter sleep duration than earlier cohorts of youth.²⁵ With regard to sociodemographic sleep disparities, adolescents who were girls, had parents with less education, or were Black or Hispanic/Latino were less likely to report sleeping for seven or more hours per night.²⁵ Documenting racial and ethnic, income, and sex sleep disparities among adolescents^26–28 and understanding their current state, trends over time,^2,25 and how they might project into the future, is consequential because these issues appear to be contributing factors to disparities in healthy adolescent development.^29,30

In the last decade, the society and social contexts in which teens are immersed and must navigate daily has rapidly transformed. Over this period, mobile devices became truly ubiquitous in teens’ lives, many genres of social media and newer digital technologies proliferated, and novel, broader community-wide stressors emerged (such as the social unrest following police killings, as well as the COVID-19 pandemic), all of which may impact young people’s sleep.^31–34 In particular, the COVID-19 pandemic dramatically altered the lives and schedules of adolescents in numerous ways. School, work, the economy, family systems, and individuals’ health were disrupted, both in the early months of the crisis and longer term. A review of the research on adolescent sleep suggested that in 2020, after the pandemic began, teens’ sleep duration increased, while sleep quality decreased with more reports of sleep disturbances and insomnia when compared to pre-pandemic.^35,36

The goal of this analysis was to provide an update on age, secular (or period), and cohort trends in adolescent sleep duration and subjective sleep sufficiency, with special attention to sociodemographic disparities, incorporating updated data from Monitoring the Future since 2012²⁵ through 2023, a decade marked by profound shifts in adolescents’ contexts. In our presentation we will also show data since 1991 in order to give full context for the longer-term trends.

Methods

Monitoring the Future (MTF) is an ongoing study of US adolescents’ behaviors surveying cross-sectional samples in grades eight, ten, and twelve annually. Participants complete annual in-school surveys each Spring.³⁷ Any variability in the specific data collection windows for each school were non-systematic across time and other factors potentially pertinent to our outcomes of interest. Schools are invited using a multi-stage random sampling design; schools that declined participation were replaced by others with a similar urbanicity, location, and size.

For these analyses, data were drawn from measurement years 1991 to 2023, representing cohorts born from approximately 1972 to 2011. During this time, school participation ranged from 50% (2023) to 99% (e.g., 1993), with an outlier in 2020 (26%) when data collection was halted in March due to the COVID-19 pandemic.³⁸ Among schools that participated, student response rates varied by grade and year, ranging from 72% (grade 12, 2023) to 91% (e.g., grade 8, 1996).²² A total of 401,160 respondents who completed relevant survey questions on at least one of two study outcome measures were included in these analyses. Those without any sleep outcome data were excluded. Additional information on MTF sampling, participation, and characteristics is available elsewhere.³⁸ All analyses were weighted, accounting for the complex survey design of MTF, with provided weight, cluster, and stratum variables based on selection probabilities of students within schools within regions based on geography and urbanicity. MTF data collection protocols were approved by a University of Michigan IRB (HUM00217920), while data analysis was approved by a Columbia University IRB (IRB-AAAK6107). Informed consent (either passive or active [i.e., written] consent, per school policy) was obtained, either from parents (for students younger than 18 years) or from students 18 years or older at time of assessment.

Measures: Sleep Duration and Sufficiency

Sleep was measured with two items, one addressing duration and one addressing subjective sleep sufficiency. Sleep duration was examined with the following: “How often do you … get at least 7 hours of sleep?” This measure has been used by other authors to examine sleep duration among adolescents.^34,39,40 Responses ranged from “never” to “every day” on a six-point scale. Adolescents reporting ≥7 hours of sleep “every day” or “nearly every day” were considered to regularly sleep ≥7 hours per day, versus those reporting doing so only sometimes or less. Sleep sufficiency was examined with the following: “How often do you … get less sleep than you should?” Responses ranged from “never” to “every day” on a six-point scale. Adolescents reporting “never” or “seldom” getting less sleep than they should were considered to be subjectively indicating they felt they were “getting enough sleep,” versus those who reported more frequently getting less sleep than they should.

Measures: Demographic Characteristics

Demographic characteristics included sex (male: 48%) and urbanicity (urban: 77%) based on residence in metropolitan statistical areas. Respondents were queried about the highest level of education of each parent, which was combined into a single parental education dichotomy of at least one parent with a college degree (56%) versus no parent with a college degree. Race and ethnicity was assessed with a single item through 2004, with response options of “White/Caucasian”, “Black/African American”, “Chicano/Mexican American”, “Cuban American”, “Puerto Rican American”, “Other Latin American”, “Asian American”, “American Indian”, and “Other.” From 2005 onward, race and ethnicity was assessed with several items, allowing respondents to select multiple. Items included identifying as “Black/African American”, “Chicano/Mexican American”, “Cuban American”, “Puerto Rican”, “Other Hispanic/Latino”, “Asian American”, “White (Caucasian)”, “American Indian or Alaskan Native”, and “Native Hawaiian or Other Pacific Islander.” Respondents were recategorized into mutually exclusive groups, including non-Hispanic Black: 12%, non-Hispanic white: 62%, Hispanic/Latino: 15%, and non-Hispanic Asian/Pacific Islander: 4%. Those reporting other races and ethnicities were coded as ‘Other’ (7%) due to small numbers (percentages listed are among those with non-missing data). Demographic data missingness ranged from 0% (urbanicity, assigned based on geography) to 11.0% (race and ethnicity), 11.2% (sex) and 22.7% (parental education – includes missing and “don’t know” response). As demographic factors were only used as subpopulations for supplemental, demographic-stratified analyses and for demographic-based comparisons of outcomes without mutual adjustment, we aimed to preserve the fidelity of demographic group membership and elected not to impute this missingness, instead reflecting comparisons and outcomes among those known to belong to various groups.

Statistical Analysis

For comparability we followed the analytic plan used in the earlier published analysis.²⁵ Survey design was incorporated for all analyses, producing weighted estimates of means, prevalences, odds ratios, and inputs for subsequent age-period-cohort (APC) models.

We graphed data by time period and age, then estimated APC effect models for each of our two sleep outcomes. Modelling was conducted using the “apc.fit” function from the “Epi” package in R. Models were fit overall for each of the two sleep outcomes and then subsequently stratified by sex, race and ethnicity, parental education, and urbanicity. Models iteratively estimated parameters for age, period, and cohort effects using the methodological approach described in prior work.²⁵ Models were given an “APC” parameterization. These models used natural splines and a binomial distribution; apc.fit models with a binomial distribution use a logit link function. In an APC parameterization, cohort effects are constrained to be 0 on average with 0 slope.⁴¹ For these analyses, 1995 was chosen as period referent, as it was in the prior iteration of these analyses, while 1983 was chosen as cohort referent.

Data were then subset into five-year periods, except for the most recent period group, 2021–2023. We examined the associations between demographics and sleep outcomes, testing interactions with year to examine potential heterogeneity. Specifically, within each period group, weighted logistic regressions used either sex, race and ethnicity, urbanicity, parental education, or age to predict one of our two sleep outcomes. Models did not mutually adjust predictors. Each model included one predictor and outcome and aimed to describe group differences. We then expanded these models by adding an interaction term for each demographic factor and year to test temporal heterogeneity. Yearly trends in outcomes were then visualized to describe patterns and differences in sleep duration and sufficiency over time.

Results

Trends in Sleep by Age and Time Period From 1991 to 2023

Figure 1 shows that the percent of respondents regularly getting seven or more hours of sleep a night as well as the prevalence of “getting enough sleep”, decreased with age. Further, adolescents in more recent years typically had lower prevalences of both outcomes than adolescents in earlier years, with the lowest prevalences at every age exhibited either from 2016 to 2020 or from 2021 to 2023. The 2021–2023 prevalences of getting seven or more hours of sleep were lowest for every age (ranging from 37.2% at age 12 or 13 to 22.3% at 18 or 19). With regard to “getting enough sleep,” 2021–2023 prevalences were lowest for those ages 12 to 16 (ranging from 29.1% among those 12 or 13 to 21.3% among those who were 16). For those 17 and older, prevalences were lowest between 2016 and 2020 (from 17.4% among 17-year-olds to 17.6% among those age 18 or 19).

Figure 1.

Prevalence of Sleep Outcomes by Age and Cohort, 1991–2023

Age, Period, and Cohort (APC) Effects in Adolescent Sleep Patterns From 1991 to 2023

Model fit statistics are available in Supplemental Table 1. For getting seven or more hours of sleep, an APC model was superior to a model with age and either period or cohort (p<.001 for both). For “getting enough sleep,” an APC model was superior to a model with just age and period (p<.001), and was comparable to an age-cohort model (p=0.06). APC model results for the outcome of getting seven or more hours of sleep a night are available in Figure 2; for “getting enough sleep,” results are shown in Figure 3.

Figure 2.

Age-Period-Cohort Model of Sleeping 7+ Hours per Night, with Trends in Prevalence by Age (Panel A) and Trends on the Odds Ratio Scale across Cohort (Panel B, Left) and Period (Panel B, Right)

Figure 3.

Age-Period-Cohort Model of “Getting Enough Sleep”, with Trends in Prevalence by Age (Panel A) and Trends on the Odds Ratio Scale across Cohort (Panel B, Left) and Period (Panel B, Right)

Interpreting Figures 2 and 3 begins with the line on the left side of each plot over the “age” segment of the x-axis, which maps outcome prevalence (y-axis on left) from ages 12 to 19. This captures the age effect of the APC model. Note, the study design is yearly cross-sectional data collection and so these estimates of prevalence are not from the same students over time. They are the typical outcome prevalence among students who were a given age at assessment, regardless of assessment year. For both outcomes, prevalence declined across age.

The other two lines in Figures 2 and 3 over the “Cohort and Period Time-Scale” segment of the x-axis, map the period and cohort effects for each outcome on the odds ratio scale (y-axis on right). The line farther to the right depicts the trends in each outcome by year (or period) relative to the referent year (1995). The line to the left in this pair maps trends by cohort, defined for a given respondent as period minus age, on the same odds ratio scale relative to the referent cohort (1983). Period and cohort effects were marked by recent declines, with estimates lowest in the most recent available period and cohort years for each outcome. For “getting enough sleep,” period effects level off in recent years, while for the outcome of sleeping seven or more hours per night, there is a strong downward trend in recent years.

Subgroup Analyses: APC Models by Sex, Race and Ethnicity, Parental Education, and Urbanicity

We examined APC models by sex, race and ethnicity, parental education, and urbanicity as a series of sensitivity analyses. Results are available in Supplemental Figures 1 through 8. Results were consistent across subgroups, reflecting the trends described above. For sleeping seven or more hours per night, there was declining prevalence by age and declines among recent periods and cohorts for all subgroups. For “getting enough sleep,” there were declines by age and cohort and levelling period effects for all subgroups. The general trends of age, period, and cohort effects were consistent across groups. However, for getting seven or more hours of sleep there were stronger period declines for non-Hispanic Black, Hispanic/Latino, and ‘Other’ race and ethnicity respondents versus non-Hispanic white peers.

Demographic Associations with Sleep Across Time

Respondents were stratified into five-year periods (three years for 2021–2023), and within each period group, we assessed demographic characteristics as predictors of each sleep outcome (Table 1). All models examining demographic associations aimed to describe patterns in group differences across time, and should not be interpreted causally.

Table 1.

ORs (95% CIs) for the Association Between Demographic Variables and Probability of Regular ≥7 Hours of Sleep Per Night and Regularly “Getting Enough Sleep”, By Time Period

Outcome: Sleep 7 or More Hours		n=77569	n=57086	n=59606	n=67580	n=63665	n=52558	n=21483
	Predictor	1991–1995	1996–2000	2001–2005	2006–2010	2011–2015	2016–2020	2021–2023	Interaction p-val:
Sex	Male	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	Female	0.68 (0.65, 0.70)	0.68 (0.65, 0.71)	0.70 (0.67, 0.73)	0.67 (0.64, 0.69)	0.64 (0.61, 0.67)	0.68 (0.64, 0.71)	0.61 (0.56, 0.67)	0.22
Race and Ethnicity	Non-Hispanic White	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	Non-Hispanic Black	0.99 (0.92, 1.07)	0.97 (0.89, 1.05)	1.06 (0.97, 1.15)	0.83 (0.77, 0.90)	0.87 (0.79, 0.95)	0.97 (0.87, 1.09)	0.79 (0.67, 0.93)	0.003
	Hispanic/Latino	0.90 (0.83, 0.98)	1.00 (0.92, 1.08)	0.95 (0.88, 1.03)	0.86 (0.81, 0.92)	0.87 (0.81, 0.93)	0.88 (0.81, 0.96)	0.82 (0.72, 0.92)	0.001
	Other	1.03 (0.97, 1.11)	1.06 (0.98, 1.16)	1.06 (0.98, 1.15)	0.83 (0.75, 0.91)	0.88 (0.80, 0.97)	0.84 (0.76, 0.93)	0.73 (0.60, 0.88)	<.0001
	Non-Hispanic Asian/Pacific Islander	0.86 (0.76, 0.96)	0.82 (0.72, 0.94)	0.79 (0.70, 0.90)	0.76 (0.66, 0.86)	0.85 (0.76, 0.96)	0.93 (0.82, 1.05)	0.97 (0.78, 1.20)	0.45
Urbanicity	Urban	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	Nonurban	1.13 (1.05, 1.21)	1.12 (1.04, 1.22)	1.12 (1.03, 1.21)	1.13 (1.04, 1.22)	1.15 (1.07, 1.24)	1.09 (1.00, 1.19)	1.09 (0.97, 1.22)	0.88
Parental Education	No Parent a College Graduate	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	At Least One Parent a College Graduate	1.18 (1.13, 1.23)	1.15 (1.09, 1.20)	1.22 (1.16, 1.28)	1.29 (1.23, 1.35)	1.29 (1.22, 1.36)	1.36 (1.27, 1.45)	1.57 (1.40, 1.76)	<.0001
Age	12/13	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	14/15	0.69 (0.66, 0.73)	0.70 (0.65, 0.75)	0.68 (0.63, 0.72)	0.70 (0.66, 0.75)	0.69 (0.65, 0.74)	0.69 (0.64, 0.74)	0.85 (0.74, 0.97)	0.27
	16/17	0.46 (0.43, 0.49)	0.42 (0.39, 0.45)	0.38 (0.36, 0.41)	0.41 (0.38, 0.44)	0.44 (0.41, 0.47)	0.43 (0.39, 0.46)	0.58 (0.51, 0.67)	0.18
	18/19	0.29 (0.27, 0.32)	0.27 (0.24, 0.29)	0.26 (0.24, 0.28)	0.30 (0.28, 0.33)	0.33 (0.30, 0.35)	0.32 (0.29, 0.35)	0.49 (0.41, 0.57)	<.0001
Outcome: “Getting Enough Sleep”		n=77506	n=57045	n=59584	n=67551	n=63626	n=52448	n=21499
	Predictor	1991–1995	1996–2000	2001–2005	2006–2010	2011–2015	2016–2020	2021–2023	Interaction p-val:
Sex	Male	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	Female	0.81 (0.78, 0.84)	0.81 (0.77, 0.85)	0.87 (0.83, 0.91)	0.82 (0.78, 0.86)	0.77 (0.73, 0.81)	0.79 (0.75, 0.84)	0.78 (0.70, 0.86)	0.20
Race and Ethnicity	Non-Hispanic White	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	Non-Hispanic Black	1.58 (1.48, 1.69)	1.55 (1.44, 1.67)	1.50 (1.39, 1.61)	1.39 (1.29, 1.51)	1.19 (1.10, 1.30)	1.20 (1.08, 1.32)	1.02 (0.85, 1.22)	<.0001
	Hispanic/Latino	1.39 (1.29, 1.50)	1.48 (1.37, 1.59)	1.38 (1.27, 1.49)	1.25 (1.16, 1.33)	1.04 (0.97, 1.12)	1.12 (1.03, 1.21)	1.13 (1.01, 1.27)	<.0001
	Other	1.28 (1.19, 1.38)	1.36 (1.26, 1.48)	1.32 (1.21, 1.44)	1.11 (1.00, 1.24)	0.95 (0.86, 1.06)	0.99 (0.89, 1.09)	0.96 (0.80, 1.15)	<.0001
	Non-Hispanic Asian/Pacific Islander	0.89 (0.79, 1.00)	0.90 (0.80, 1.01)	0.81 (0.70, 0.93)	0.87 (0.77, 0.99)	0.91 (0.80, 1.03)	0.92 (0.81, 1.05)	0.89 (0.73, 1.09)	0.82
Urbanicity	Urban	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	Nonurban	1.19 (1.11, 1.27)	1.17 (1.09, 1.26)	1.18 (1.10, 1.27)	1.13 (1.05, 1.21)	1.19 (1.11, 1.28)	1.09 (1.01, 1.18)	1.13 (1.00, 1.28)	0.26
Parental Education	No Parent a College Graduate	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	At Least One Parent a College Graduate	0.82 (0.79, 0.86)	0.79 (0.74, 0.83)	0.81 (0.77, 0.86)	0.90 (0.86, 0.95)	0.98 (0.92, 1.03)	0.90 (0.84, 0.96)	0.97 (0.86, 1.10)	<.0001
Age	12/13	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)	1.00 (REF)
	14/15	0.74 (0.70, 0.78)	0.75 (0.71, 0.81)	0.78 (0.73, 0.84)	0.76 (0.72, 0.82)	0.72 (0.68, 0.78)	0.71 (0.66, 0.77)	0.80 (0.69, 0.93)	0.72
	16/17	0.52 (0.49, 0.55)	0.50 (0.46, 0.54)	0.52 (0.48, 0.55)	0.49 (0.46, 0.53)	0.48 (0.45, 0.52)	0.49 (0.45, 0.53)	0.64 (0.55, 0.74)	0.93
	18/19	0.44 (0.39, 0.48)	0.42 (0.38, 0.46)	0.42 (0.39, 0.46)	0.41 (0.38, 0.45)	0.46 (0.42, 0.50)	0.43 (0.38, 0.47)	0.62 (0.51, 0.74)	0.04

Female respondents consistently exhibited lower prevalence of both sleep outcomes, and this difference did not vary across periods.

For race and ethnicity, however, there were shifts in existing inequities. For non-Hispanic Black and Hispanic/Latino respondents, as well as those categorized as ‘Other’ race and ethnicity, compared with non-Hispanic white peers, there were no or minimal differences for getting seven or more hours of sleep in 1991 to 1995. However, by 2021 to 2023 these groups were less likely than non-Hispanic white respondents to report getting seven hours of sleep (e.g., 0.79, 95% CI: 0.67, 0.93 for non-Hispanic Black respondents from 2021–2023). The same groups went from being more likely to report “getting enough sleep” in early period groups (e.g., 1.58, 95% CI: 1.48, 1.69 for non-Hispanic Black respondents from 1991 to 1995) to being similar or only minimally more likely to report “getting enough sleep” in recent cohorts (e.g., 1.02, 95% CI: 0.85, 1.22 for non-Hispanic Black respondents from 2021 to 2023), as is evident from the interactions between race and ethnicity and period predicting “getting enough sleep.”

For urbanicity, non-urban respondents had higher odds of each sleep outcome with a consistent difference compared to urban peers across time.

Parental education patterns varied by outcome. Those with at least one parent who graduated college had higher odds of getting seven hours of sleep per night (e.g., OR = 1.18, 95% CI: 1.13, 1.23 from 1991 to 1995), and this difference grew larger over time (e.g., OR = 1.57, 95% CI: 1.40, 1.76 from 2021 to 2023). However, with self-reported “getting enough sleep,” those in early period groups with a college graduate parent had lower odds compared to peers without a college graduate parent (e.g., OR = 0.82, 95% CI: 0.79, 0.86 from 1991 to 1995). This gap narrowed over time.

Age differences were largely consistent, with older adolescents having a lower odds of both sleep outcomes. These patterns are reflected in Supplemental Figures 9 through 14, detailing declining trends in sleep outcomes overall and by demographic subgroup. While group differences were evident along the patterns described above, all groups declined in sleep outcomes from 1991 to 2023.

Discussion

We found that adolescents’ sleep duration appears to have further deteriorated in the last 10 years. This extends a trend that had been documented in MTF from 1991–2012.²⁵ Similar to the prior work, this updated analysis through 2023 shows that older teens were less likely to sleep more than seven hours per night compared to younger teens. However, teens at every age in the last 10 years were more likely to report poor sleep duration compared to teens at those same ages in earlier decades. We also noted an increasing divergence in racial/ethnic sleep equity, wherein 2001–2005 non-Hispanic Black and Hispanic/Latino adolescents had similar odds of reporting seven or more hours of sleep per night as white adolescents; starting in 2006, generally, non-Hispanic Black and Hispanic/Latino adolescents were less likely to report seven hours of sleep. Further, the sleep duration-gap between Black and Hispanic/Latino adolescents and their white peers grew steadily through the 2021–2023 period. Prior to 2016, Asian adolescents were less likely to report getting more than seven hours of sleep per night but subsequently began to report similar odds of meeting this bar as white respondents. While adolescents with at least one parent who graduated from college were more likely to report getting seven or more hours of sleep in all periods compared to teens whose parents were less educated, the gap had grown such that by 2021–2023 teens from families where a parent graduated from college had 57% greater odds of reporting seven or more hours of sleep per night.

The prior report, which was published in 2015 in Pediatrics, was entitled “The Great Sleep Recession…”²⁵ Ten years later, with no respite in sight from a prolonged downturn in adolescent sleep health, it appears we are in the midst of a Great Sleep Depression. As in an economic depression, which causes prolonged and widespread damage to an economy, this Great Sleep Depression is poised to exact a toll on these generations of young people who grow up exhausted, having consequences for them that are greater than just the immediate discomfort of their tiredness. Sleep insufficiency is linked to increased risk for injury, reduced executive functioning and academic performance, poor mental health, and a range of chronic disease and chronic disease risk factors among adolescents.^{39,40,42–44} Society-wide issues can have a more intense negative impact on those with lesser social advantage. Prior research has illustrated adolescent sociodemographic sleep inequities.^7,9 Our analyses demonstrate an alarming trend, also suggested in a recent report using US Youth Risk Behavior Survey data,² that adolescents of certain marginalized demographics are becoming even less likely to report they are sleeping as much as they should and/or are getting seven or more hours of sleep per night.

Though this study could not directly examine how particular factors such as social media use might influence sleep, these trends do coincide with rapid increases in adolescent time spent with smartphones and social media.^45,46 Even with the increase in restrictive school phone policies in the US⁴⁷ or the adoption of social media age restriction rules that we have seen elsewhere,⁴⁸ it seems unlikely that our society will revert to screen-free lives. Looking to the future, there will likely be unanticipated novel and mounting challenges that youth face in getting healthy sleep (both technology-based and otherwise). These include what Schlarb and colleagues have termed “chronosystem factors,” including climate change and greater exposure to heat and humidity as well as shocks, such as natural disasters (i.e., hurricanes, floods), community violence and political instability which cause disruption and stress.⁴⁴ Despite these potential sleep obstacles, it is critical that we not allow barriers to adolescent sleep to outrun sleep promotion efforts.

While the causes of adolescent sleep decline are likely multifactorial, an effective, population-wide intervention to improve these trends could be changing school start times. Schools’ start times are a highly salient structural component of adolescents’ 24-hour cycle that can be, and have been, effectively improved and aligned with adolescents’ biological needs through policy. These efforts have been shown to extend adolescents’ sleep. There is now a robust body of literature supporting later secondary school start times as an effective population-scale intervention for extending and expanding adolescents’ sleep and reducing circadian misalignment and social jet lag.^24,49

A limitation with what we have presented is that the sleep measures in MTF are both subjective and coarse. No anchors for “getting enough sleep” were provided to respondents; no distinction was made between school night and free day sleep. Though this measure may be subject to variability of interpretation among respondents, we believe it is valuable as it provides information on the adolescents’ insights regarding their own sleep, which is an underexamined facet of adolescent health. Additionally, the sleep duration item asked about getting “at least 7 hours of sleep” but any length of sleep shorter than eight hours should considered suboptimal given that the recommendation is 8–10 hours of sleep per night.³ Our findings may be impacted by temporal heterogeneity in sampling and measurement, particularly due to the COVID-19 pandemic in 2020, though sensitivity analyses comparing results with and without this year suggest minimal differences. Our findings may also not reflect older adolescents who have dropped out or those who are chronically absent, both of whom may experience greater disadvantage and worse sleep than their peers. However, these limitations are balanced by the strengths of MTF, namely that it has a very large, US-representative sample that spans adolescence, and includes consistent measures of adolescent sleep over three decades.

Conclusion

For decades, adolescents’ sleep had been declining; here, we demonstrated that, in the last decade, it has regrettably continued to plummet. Even the teen participants of MTF are aware they are being shortchanged on sleep, as evidenced by majority of the sample that indicated they were aware that they regularly slept less than they should. It is concerning that youth from sociodemographic groups that are marginalized appear to be at even greater risk for profoundly short sleep, given that sleep can be seen as a resource that confers advantages for health, learning, and development. Regularly getting less than seven hours of sleep per night is insufficient for adolescents of any age. This is will exact a toll, both immediate and long-term, on this generation of teens.

What’s Known on This Subject:

Insufficient sleep during one’s youth is an important problem that can have both immediate and long-term health consequences and may set the stage for suboptimal adult sleep patterns. In the US, most adolescents do not meet nightly sleep duration recommendations.

What This Study Adds:

Adolescent sleep has been in sharp decline. Further, sociodemographic sleep gaps have widened over time; fewer teens from disadvantaged groups (i.e. Black adolescents, those with less educated parents) report sufficient sleep compared to adolescents with greater socioeconomic advantage.

Abbreviations:

MTF

Monitoring the Future

OR

Odds Ratio

APC

age, period cohort