33.2% of US adults report sleeping under 7 hours a night (CDC, 2023). 72.7% of US high school students sleep under 8 hours on school nights (CDC MMWR, 2018). A night or more of sleep loss cut exercise performance by 7.56% on average (Sports Medicine, 2022). People with insomnia had a 2.27 times higher risk of later depression (BMC Psychiatry, 2016). Insufficient sleep may cost the US economy up to $411 billion a year (RAND Europe, 2016). 22 of 25 melatonin gummies (88%) were inaccurately labelled (JAMA, 2023).

How Many Adults Sleep Too Little? Sleep Loss and Physical Performance Sleep, Mortality and the Economy Teens and Children Insomnia Shift Work and Caffeine Sleep Supplements: Label Accuracy and Safety Methodology and Limitations "Published" is the year of the paper; "Data" is the year the underlying numbers were collected or the review's search window. Where a study is observational or self-reported, we say so.

All figures from CDC MMWR, "Short Sleep Duration Among Middle School and High School Students, United States, 2015". Published 2018; data from the 2015 Youth Risk Behavior Surveys. Self-reported school-night sleep. 57.8% of middle school students (nine states) slept under 9 hours on school nights, ranging from 50.2% (New Mexico) to 64.7% (Kentucky). The combined state data are not nationally representative. 72.7% of US high school students slept under 8 hours on school nights (national sample of 15,624). High school state estimates ranged from 61.8% (South Dakota) to 82.5% (West Virginia). "Short sleep" was defined as under 9 hours for ages 6-12 and under 8 hours for ages 13-18. What this means: Short sleep is the norm for US teens, not the exception: even the best-performing state had more than six in ten high schoolers under the recommended amount. These are 2015 self-reported data, so current figures may differ.

How common insomnia is depends heavily on how it is defined, so estimates range widely. Read the figures below as a range, not a single number. A meta-analysis of 34 cohort studies (172,077 participants) found insomnia was linked to a pooled relative risk of 2.27 (95% CI 1.89-2.71) for later depression, over a mean follow-up of 60.4 months. Heterogeneity was high (I² = 92.6%) and publication bias was detected. Source: BMC Psychiatry, 2016; searches to October 2014. A 2025 meta-analysis of 47 studies put insomnia disorder prevalence at 12.4% (95% CI 9.0-16.8%) when diagnosed by interview, and 16.3% (11.3-23.0%) by DSM-based self-report; questionnaire-based studies ranged from 7.5% to 32.3%. Source: Journal of Sleep Research, 2025; searches to April 2024. An earlier meta-analysis of 13 articles (n = 22,980) pooled prevalence at 22.0% (95% CI 17.0-28.0%), with women having higher odds than men (OR 1.58, 95% CI 1.35-1.85). Source: Frontiers in Psychiatry, 2020; searches to April 2019. Estimates differ from the 2025 review because of different diagnostic criteria. In China, pooled insomnia prevalence was 15.0% (95% CI 12.1-18.5%) across 17 studies and 115,988 participants; it was 20.4% in samples with a mean age under 43.7 versus 11.6% in older samples. Source: PLoS One, 2017. What this means: The most defensible headline is that about one in eight adults meets diagnostic criteria for insomnia disorder (12.4%), while looser definitions capture closer to one in five. Women are consistently more affected. The depression link is large, but the direction can run both ways: insomnia can precede depression, and depression can disrupt sleep. The authors also found signs of publication bias, so the true risk may be somewhat lower than 2.27.

In 817 elite Japanese athletes (Asian Games candidates; 891 surveyed), mean time in bed was 7 hours 29 minutes, and 229 (28.0%) scored above the clinical cutoff for poor sleep quality on the Pittsburgh Sleep Quality Index. About 40% reported being told they snore loudly or twitch their legs in sleep. Source: Sports Medicine - Open, 2018. Self-reported. A narrative review found adolescent athletes get on average about 6.3 hours of sleep against the 8-10 hours recommended. Source: Nutrients, 2023. This is a narrative review, not a pooled estimate. In a six-week study of 25 professional rugby players, a sleep extension programme raised total sleep time by 6.3% and time in bed by 7.3%, and lowered cortisol by 18.7% and mean reaction time by 4.3% versus control (confidence limits are wide). Source: Sports, 2018. In a crossover study of 24 physically active students, a single night of extended sleep added about 55 minutes (531.3 vs 476.5 minutes), improved 5 m shuttle-run distance at 8 AM (102.8 m vs 93.3 m) and cut simple reaction time (252.8 ms vs 296.4 ms). Source: Life, 2025. Participants were active students, not elite athletes. What this means: Roughly one in four elite athletes in this sample reported poor sleep, so elite status does not protect against it. The sleep-extension results are encouraging but come from two very small studies, one without randomisation, so they suggest benefit rather than prove it. A reasonable takeaway is that extra sleep is a low-cost thing to test, particularly before morning sessions.

Sleep Facts and Statistics: What the Primary Research Shows

Q: Shift Work and Caffeine

A review of 48 systematic reviews found moderate-grade evidence linking shift work to breast cancer and long work hours to stroke; evidence for other chronic conditions was low-grade or unclear. Source: PLoS One, 2020. A later pooled analysis of 10 prospective studies (4,660 breast cancers; about 1.4 million women) found a combined relative risk of 0.99 (95% CI 0.95-1.03) for any night shift work, and the authors concluded night shift work has little or no effect on breast cancer incidence. Source: Journal of the National Cancer Institute, 2016. These two reviews disagree. In a meta-analysis of 10 studies (128 athletes), evening caffeine reduced sleep efficiency by 4.87% (95% CI -7.45 to -2.29), though the result was not robust in sensitivity analyses; the 32.47-minute drop in total sleep time was not statistically significant (95% CI -69.93 to 4.99). Athletes nonetheless reported clear subjective sleep impairment. Source: Sports, 2025. What this means: The shift-work evidence is genuinely mixed. The 2020 review rated the breast cancer link as moderate-grade, while the large prospective pooling found no effect, so no firm conclusion is justified yet. For caffeine, the objective sleep data were inconclusive but the athletes consistently felt their sleep was worse. With only 128 athletes across the studies, the safest reading is that evening caffeine may cost some people sleep, and perceived sleep quality matters in its own right.

Q: Sleep Apnea and Screens

A meta-analysis of 11 studies put the pooled prevalence of obstructive sleep apnea among healthy adults aged 18-30 at about 16% (95% CI 8-29%, I² = 92%). Diagnostic thresholds and devices varied widely. Source: Journal of Clinical Medicine, 2024. Across 21 cohort studies (548,338 participants), each additional hour of daily screen time was associated with about 3 to 5 minutes less sleep (β = -0.05, 95% CI -0.08 to -0.03) and a higher risk of short sleep (OR 1.25, 95% CI 1.08-1.40). Source: Frontiers in Psychiatry, 2025. In the same review, each hour of screen time was linked to a 13.2-minute later bedtime and 3.05 times the odds of difficulty falling asleep (95% CI 1.51-6.24). A systematic review of 31 papers found screen time associated with poorer sleep in infants, toddlers and preschoolers; the meta-analysis confirmed this in infants and toddlers but not preschoolers, and overall evidence quality was rated very low. Source: Sleep Medicine Reviews, 2020. What this means: The sleep apnea estimate is wide and the studies disagree (I² = 92%), so treat 16% as a rough indication that apnea is not limited to older adults. The screen-time effect per hour is small (minutes), but the delayed bedtime and trouble falling asleep figures suggest the effect is more about when people start sleeping than total time. These are associations from cohort data, so they do not prove screens cause sleep loss.

Written by Athletic Insight Research · Last verified 8 October 2026

Roughly one in three US adults sleeps less than the recommended seven hours, and the research linking short sleep to worse performance, mood and health keeps growing. This page collects the key figures from peer-reviewed papers, CDC reports and one RAND economic model. Each figure links to the paper it came from, and we say when a number is self-reported, modeled, or based on a small sample.

Key Takeaways

33.2% of US adults report sleeping under 7 hours a night (CDC, 2023).
72.7% of US high school students sleep under 8 hours on school nights (CDC MMWR, 2018).
A night or more of sleep loss cut exercise performance by 7.56% on average (Sports Medicine, 2022).
People with insomnia had a 2.27 times higher risk of later depression (BMC Psychiatry, 2016).
Insufficient sleep may cost the US economy up to $411 billion a year (RAND Europe, 2016).
22 of 25 melatonin gummies (88%) were inaccurately labelled (JAMA, 2023).

How Many Adults Sleep Too Little?

The best recent snapshot of US adult sleep is a CDC analysis of the 2020 Behavioral Risk Factor Surveillance System, a large telephone survey. The figures below are the share of adults reporting under 7 hours in 24.

33.2% of US adults (age-adjusted) reported sleeping fewer than 7 hours per 24 hours; the crude figure was 32.7%. Source: Pankowska et al., Preventing Chronic Disease (CDC). Published 2023; data 2020 BRFSS, 390,193 respondents. Self-reported.
Short sleep was most common at ages 25-44 (36.4%) and lowest at 65 and over (26.0%). Same source; crude prevalence.
Men 33.3% vs women 32.1% reported short sleep (crude). Same source.
By race and ethnicity (age-adjusted), short sleep ranged from 30.8% (non-Hispanic Asian) to 46.5% (non-Hispanic Native Hawaiian or Pacific Islander); non-Hispanic Black adults reported 43.6% and non-Hispanic White adults 31.8%. Same source.
Prevalence rose as income fell: 29.3% at $75,000 or more vs 38.1% under $15,000 (age-adjusted). Same source.
Prevalence was 27.0% among college graduates vs 36.8% among adults with some college (age-adjusted). Same source.
Prevalence rose with decreasing urbanicity, from 32.0% in large central metro counties to 35.0% in noncore counties (age-adjusted). Same source.
County-level modelled estimates ranged from 23.8% (Boulder County, Colorado) to 48.4% (Greene County, Alabama). Same source; model-based, not direct survey estimates.
About 45% of the Western adult population fail to get the recommended 7-9 hours per night, according to the introduction of Craven et al., Sports Medicine. Published 2022; this is a literature statement, not a new measurement.

What this means: Short sleep is not evenly spread. The working-age group (25-44) sleeps least, and the gaps by income, education and county are large: a 24.6-point spread between the lowest and highest counties. That pattern points to work schedules, money and local conditions as drivers, not just personal habits. Because the survey is self-reported, true sleep time is probably somewhat different, but the group-to-group differences are the useful signal.

Sleep Loss and Physical Performance

All figures from Craven et al., “Effects of Acute Sleep Loss on Physical Performance”, Sports Medicine. Published 2022; studies searched from inception to September 2020, with an update search on 31 December 2021. The review pooled 69 publications and 227 outcome measures; 89% of participants were male and heterogeneity was very high (I² = 98.1%).

Acute sleep loss (6 hours or less in 24 hours) lowered exercise performance by a mean of 7.56% (95% CI -11.9 to -3.13).
Performance fell about 0.4% for every hour awake before the exercise task under deprivation and late-restriction protocols.
Mean change by task type: skill -20.9%, strength-endurance -9.85%, anaerobic power -6.26%, HIIE -6.15%, endurance -5.55%, speed/power endurance -2.90%, strength -2.85%.
Evening tests suffered more than morning tests: -8.31% in the PM vs -5.42% in the AM.
Early restriction (delayed sleep onset) did not produce a statistically significant overall effect (-5.85%, 95% CI -13.4 to 1.66, p = 0.125).
The authors note professional athletes, shift workers and military personnel appear particularly susceptible to sleep loss.

What this means: Not every kind of effort suffers equally. Pure strength barely moved (-2.85%), while skilled, technical tasks dropped the most (-20.9%). In practical terms, a short night is more likely to hurt a sport that depends on precision, timing or decision-making than a single heavy lift. Two cautions apply. These are acute, one-night effects, not the effect of months of mild sleep debt. And the very high heterogeneity means the 7.56% average hides large differences between studies, so it should not be read as a prediction for any one athlete.

Sleep, Mortality and the Economy

All figures from Hafner et al., RAND Europe, “Why sleep matters” (RR-1791). Published 2016 (30 November). Modelled estimates for five OECD countries; the study was prepared for Vitality Health.

Sleeping under 6 hours was linked to a 10% higher mortality risk than sleeping 7-9 hours; 6-7 hours was linked to a 4% higher risk.
Insufficient sleep costs the US economy up to $411 billion a year, and Japan up to $138 billion.
As a share of GDP the estimated loss was 1.56-2.28% for the US, 1.86-2.92% for Japan, 1.36-1.86% for the UK, 1.02-1.56% for Germany and 0.85-1.56% for Canada.
Estimated working days lost per year: about 1.23 million (US), 604 thousand (Japan), 207 thousand (UK), 209 thousand (Germany), 78 thousand (Canada).

What this means: The mortality figures describe an association, not proof that short sleep causes earlier death; people who sleep badly often have other health problems. The economic numbers are model outputs built on assumptions, which is why RAND reports ranges. The consistent finding across all five countries is the direction: lost productivity from poor sleep is measured in percent of GDP, not rounding errors.

Teens and Children

All figures from CDC MMWR, “Short Sleep Duration Among Middle School and High School Students, United States, 2015”. Published 2018; data from the 2015 Youth Risk Behavior Surveys. Self-reported school-night sleep.

57.8% of middle school students (nine states) slept under 9 hours on school nights, ranging from 50.2% (New Mexico) to 64.7% (Kentucky). The combined state data are not nationally representative.
72.7% of US high school students slept under 8 hours on school nights (national sample of 15,624).
High school state estimates ranged from 61.8% (South Dakota) to 82.5% (West Virginia).
“Short sleep” was defined as under 9 hours for ages 6-12 and under 8 hours for ages 13-18.

What this means: Short sleep is the norm for US teens, not the exception: even the best-performing state had more than six in ten high schoolers under the recommended amount. These are 2015 self-reported data, so current figures may differ.

Insomnia

How common insomnia is depends heavily on how it is defined, so estimates range widely. Read the figures below as a range, not a single number.

A meta-analysis of 34 cohort studies (172,077 participants) found insomnia was linked to a pooled relative risk of 2.27 (95% CI 1.89-2.71) for later depression, over a mean follow-up of 60.4 months. Heterogeneity was high (I² = 92.6%) and publication bias was detected. Source: BMC Psychiatry, 2016; searches to October 2014.
A 2025 meta-analysis of 47 studies put insomnia disorder prevalence at 12.4% (95% CI 9.0-16.8%) when diagnosed by interview, and 16.3% (11.3-23.0%) by DSM-based self-report; questionnaire-based studies ranged from 7.5% to 32.3%. Source: Journal of Sleep Research, 2025; searches to April 2024.
An earlier meta-analysis of 13 articles (n = 22,980) pooled prevalence at 22.0% (95% CI 17.0-28.0%), with women having higher odds than men (OR 1.58, 95% CI 1.35-1.85). Source: Frontiers in Psychiatry, 2020; searches to April 2019. Estimates differ from the 2025 review because of different diagnostic criteria.
In China, pooled insomnia prevalence was 15.0% (95% CI 12.1-18.5%) across 17 studies and 115,988 participants; it was 20.4% in samples with a mean age under 43.7 versus 11.6% in older samples. Source: PLoS One, 2017.

What this means: The most defensible headline is that about one in eight adults meets diagnostic criteria for insomnia disorder (12.4%), while looser definitions capture closer to one in five. Women are consistently more affected. The depression link is large, but the direction can run both ways: insomnia can precede depression, and depression can disrupt sleep. The authors also found signs of publication bias, so the true risk may be somewhat lower than 2.27.

Shift Work and Caffeine

A review of 48 systematic reviews found moderate-grade evidence linking shift work to breast cancer and long work hours to stroke; evidence for other chronic conditions was low-grade or unclear. Source: PLoS One, 2020.
A later pooled analysis of 10 prospective studies (4,660 breast cancers; about 1.4 million women) found a combined relative risk of 0.99 (95% CI 0.95-1.03) for any night shift work, and the authors concluded night shift work has little or no effect on breast cancer incidence. Source: Journal of the National Cancer Institute, 2016. These two reviews disagree.
In a meta-analysis of 10 studies (128 athletes), evening caffeine reduced sleep efficiency by 4.87% (95% CI -7.45 to -2.29), though the result was not robust in sensitivity analyses; the 32.47-minute drop in total sleep time was not statistically significant (95% CI -69.93 to 4.99). Athletes nonetheless reported clear subjective sleep impairment. Source: Sports, 2025.

What this means: The shift-work evidence is genuinely mixed. The 2020 review rated the breast cancer link as moderate-grade, while the large prospective pooling found no effect, so no firm conclusion is justified yet. For caffeine, the objective sleep data were inconclusive but the athletes consistently felt their sleep was worse. With only 128 athletes across the studies, the safest reading is that evening caffeine may cost some people sleep, and perceived sleep quality matters in its own right.

Sleep in Athletes

In 817 elite Japanese athletes (Asian Games candidates; 891 surveyed), mean time in bed was 7 hours 29 minutes, and 229 (28.0%) scored above the clinical cutoff for poor sleep quality on the Pittsburgh Sleep Quality Index. About 40% reported being told they snore loudly or twitch their legs in sleep. Source: Sports Medicine – Open, 2018. Self-reported.
A narrative review found adolescent athletes get on average about 6.3 hours of sleep against the 8-10 hours recommended. Source: Nutrients, 2023. This is a narrative review, not a pooled estimate.
In a six-week study of 25 professional rugby players, a sleep extension programme raised total sleep time by 6.3% and time in bed by 7.3%, and lowered cortisol by 18.7% and mean reaction time by 4.3% versus control (confidence limits are wide). Source: Sports, 2018.
In a crossover study of 24 physically active students, a single night of extended sleep added about 55 minutes (531.3 vs 476.5 minutes), improved 5 m shuttle-run distance at 8 AM (102.8 m vs 93.3 m) and cut simple reaction time (252.8 ms vs 296.4 ms). Source: Life, 2025. Participants were active students, not elite athletes.

What this means: Roughly one in four elite athletes in this sample reported poor sleep, so elite status does not protect against it. The sleep-extension results are encouraging but come from two very small studies, one without randomisation, so they suggest benefit rather than prove it. A reasonable takeaway is that extra sleep is a low-cost thing to test, particularly before morning sessions.

Sleep Apnea and Screens

A meta-analysis of 11 studies put the pooled prevalence of obstructive sleep apnea among healthy adults aged 18-30 at about 16% (95% CI 8-29%, I² = 92%). Diagnostic thresholds and devices varied widely. Source: Journal of Clinical Medicine, 2024.
Across 21 cohort studies (548,338 participants), each additional hour of daily screen time was associated with about 3 to 5 minutes less sleep (β = -0.05, 95% CI -0.08 to -0.03) and a higher risk of short sleep (OR 1.25, 95% CI 1.08-1.40). Source: Frontiers in Psychiatry, 2025.
In the same review, each hour of screen time was linked to a 13.2-minute later bedtime and 3.05 times the odds of difficulty falling asleep (95% CI 1.51-6.24).
A systematic review of 31 papers found screen time associated with poorer sleep in infants, toddlers and preschoolers; the meta-analysis confirmed this in infants and toddlers but not preschoolers, and overall evidence quality was rated very low. Source: Sleep Medicine Reviews, 2020.

What this means: The sleep apnea estimate is wide and the studies disagree (I² = 92%), so treat 16% as a rough indication that apnea is not limited to older adults. The screen-time effect per hour is small (minutes), but the delayed bedtime and trouble falling asleep figures suggest the effect is more about when people start sleeping than total time. These are associations from cohort data, so they do not prove screens cause sleep loss.

Sleep Supplements: Label Accuracy and Safety

All figures from Cohen et al., JAMA research letter. Published 2023; products identified in the NIH Dietary Supplement Label Database in September 2022 (30 brands entered in 2021, 25 analysed). Small sample: one sample per brand, gummies only.

22 of 25 melatonin gummy products (88%) were inaccurately labelled; only 3 (12%) were within ±10% of the declared melatonin.
Measured melatonin ranged from 74% to 347% of the labelled amount, and 1.3 mg to 13.1 mg per serving.
One product had no detectable melatonin but contained 31.3 mg of CBD.
US Poison Control calls about pediatric melatonin ingestions rose 530% from 2012 to 2021, with 4,097 hospitalizations, 287 ICU admissions and 2 deaths (the letter cites this from the CDC MMWR report by Lelak et al.).

What this means: The study tested one product per brand and only gummies, so it cannot speak for all melatonin products. Even so, a spread from 74% to 347% of the labelled dose shows why third-party testing matters when choosing any sleep supplement. Keep all melatonin products out of children’s reach, given the poison control trend.

Methodology and Limitations

Every figure comes from a source we opened and read: full text for the CDC, Craven, RAND and JAMA papers, and abstracts for the insomnia, shift-work and caffeine reviews.
Prevalence figures are self-reported and drawn from a telephone survey with a median 47.6% response rate.
Performance effects come from mostly male, mostly lab-based studies and should not be read as a prediction for any individual.
Economic and mortality figures are modelled estimates, not direct counts.
Each figure is traceable to a quote in the verification ledger. We refresh this page as new data is published.

Reuse of any statistic here is welcome with a link back to this page.

Interested in sleep and recovery support? Browse our guides at Athletic Insight.