Counts describe this library, not the strength of evidence.
Comprendre ce type de preuve
Sleep health has several dimensions. Regularity cohorts and sleep-extension trials answer different questions; a prediction from observational data is not proof that changing a score prevents disease. Exercise effects also differ between sleep quality and daytime sleepiness.
Across 11 consumer sleep trackers tested against in-lab polysomnography in 75 people in Korea, sleep-stage agreement varied widely: the best macro F1 score was 0.69 and the worst was 0.26. The tested people were in sleep clinics and a hospital, so results may not match everyday home use.
Among 8,505 US adults in NHANES, reporting sleep problems, sleep apnea symptoms or daytime sleepiness was linked to a higher systemic immune-inflammation index (SII), for example β 30.320 for daytime sleepiness. Sleep was self-reported and measured once, so the data cannot show that poor sleep raises inflammation.
Among 761 medical students in Serbia, 21.7% screened as smartphone addicted, and the strongest independent links were more than 4 hours of daily use (OR 2.39) and higher depression scores (OR 2.51). Data are a one-time cross-sectional questionnaire, so the direction of these links cannot be told.
Across 5 US cohorts of 5946 adults without stroke or dementia, higher sleep maintenance efficiency (pooled β 0.08 per 1% increase) and no obstructive sleep apnea were associated with better global cognition over 5 years. These are observational links from a single overnight home sleep study per person, so they do not show that improving sleep protects thinking.
In 62 cognitively normal older adults, stronger NREM slow wave activity was linked to better face-name memory only in those with high β-amyloid burden (B = 2.694, p = 0.019), not in those without it (B = -0.115, p = 0.876). The study is small and cross-sectional, so it cannot show that improving deep sleep protects memory.
In 138 breast cancer patients with chemotherapy-associated insomnia, real acupuncture was not better than sham acupuncture at lowering Insomnia Severity Index score at 6 weeks (mean difference -0.4, 95% CI -1.8 to 1.1; P = 0.609). Several secondary outcomes favored acupuncture, but they are secondary results from a single-region trial, so they need confirmation.
Among 4,496 matched participants in the INTERSTROKE case-control study, people with a first acute stroke were more likely than controls to report short sleep under 5 hours (OR 3.15, 95% CI 2.09-4.76), long sleep over 9 hours (OR 2.67) and breathing cessation (OR 2.87). Sleep was self-reported in the month before stroke, and stroke itself may change sleep habits, so the direction of the link is uncertain.
In a randomized trial of 60 college students with test anxiety, three weeks of the probiotic Lactobacillus plantarum JYLP-326 relieved anxiety, depression and insomnia scores compared with placebo, and shifted gut bacteria and fecal metabolites. The abstract gives no effect sizes, and the trial was short and small, so the mechanism claims remain tentative.
Among 1,409 postpartum women who completed an online survey, higher fatigue, worse sleep quality and lower resilience were each linked to higher depression scores, together explaining R2 = 0.37 of the variation. Because it is a one-time, self-reported online survey, it shows association and cannot say which factor comes first.
In a survey of 5074 Kaiser Permanente Northern California members aged 65-79, about 30% usually slept less than 7 hours, 13% had frequent insomnia and 8% had potential sleep apnea symptoms, yet only 10% had discussed sleep with a healthcare professional. All data were self-reported by members of one health system.
Among 9004 US adults in NHANES 2017-2020, mean sleep was 7.59 hours on workdays and 8.24 hours on free days, and 46.5% (95% CI 42.6% to 50.3%) had 1 hour or more of social jet lag. All sleep data were self-reported, and the survey is a snapshot that cannot show health effects.
In the Whitehall II cohort, sleeping 5 hours or less at age 50 was associated with higher risk of multimorbidity than 7 hours (hazard ratio 1.30; 95% CI 1.12 to 1.50), and similar at ages 60 and 70. Sleep was self-reported and undiagnosed illness could have shortened sleep, so the link may not be causal.
Among 17 859 US adults in NHANES, both short sleep (OR 1.66) and long sleep (OR 2.75) were linked to clinically relevant depression compared with 7-9 hours, and a poor sleep pattern had an OR of 5.98. The survey is cross-sectional and sleep was self-reported, so it cannot show which came first.
In 270 people with insomnia and depression, 8 weeks of electroacupuncture lowered the PSQI sleep score by 3.6 points more than sham acupuncture (95% CI -4.4 to -2.8; P < .001) and 5.1 more than standard care alone. Everyone also received psychiatric standard care, and the trial was in 3 hospitals in Shanghai, so results may not carry over to other settings.
In 32 cognitively healthy older adults, more wake after sleep onset, lower sleep efficiency and less total sleep time on a home Zeo device were each linked to higher PSQI sleep disturbance scores, but no objective measure related to the PSQI total score. With only 32 people and no effect sizes reported in the abstract, the findings are preliminary.
Among 2067 people with hypertension and obstructive sleep apnea followed for a median of 6.83 years, each standard deviation increase in the cardiometabolic index (CMI) was linked to a higher risk of cardiovascular disease (adjusted HR 1.31, 95% CI 1.20-1.43). This is a retrospective cohort from one study in Urumqi, so it shows association in a specific patient group, not that lowering CMI prevents disease.
Across 227 outcome measures from 69 publications, sleep loss lowered exercise performance by a mean of 7.56% (95% CI -11.9 to -3.13, p = 0.001). Results varied enormously between studies (I2 = 98.1%) and 89% of participants were male, so the average is a rough guide, not a prediction for any one person.
Using genetic variants as proxies, genetically determined insomnia was linked to higher odds of ADHD (OR 1.33), major depressive disorder (OR 1.31) and PTSD (OR 1.32), and PTSD and depression were in turn linked to insomnia. Mendelian randomisation can be biased by genetic effects on several traits, and the genetic proxies reflect lifelong liability rather than a few bad nights.