Methodology
Evidence That Sets No Letter
An EBL letter rests on randomized human trials of an intervention. Many other studies are worth reading: large population studies, genetic studies, studies with no comparison group, and work in animals. The records cite them where they help a reader, and none of them moves a letter.
This page lists every study a graded record cites and deliberately does not count, grouped by the kind of study it is. Each entry names the record that cites it, what the study found, its design and why it sets no letter. Animal and laboratory work cited under a record's Evidence wanted block is labeled by model on that record's page.
Genetic studies have a page of their own explaining the method: Mendelian randomization.
Population studies, not trials
Cohort, case-control and cross-sectional studies, and pooled analyses of them. They show who has which outcome, not what a change would do, because the people compared differ in more ways than the one being studied.
Lineweaver et al. (American Journal of Psychiatry, 2014; PMID PubMed ID: the permanent number the US National Library of Medicine gives a published paper. Each one links to that paper's record on PubMed. 24170170). 144 cognitively normal older adults who knew or did not know their genotype.
Those who knew they carried e4 rated their memory more harshly and performed worse on an objective verbal memory test than carriers who did not know.
Design: Nested case-control (not randomized).
Why it sets no letter: EBL grades only randomized human trials. People who chose to learn their genotype may differ from those who did not.
Fortea et al. (Nature Medicine, 2024; PMID 38710950). 3,297 individuals in the pathology study and 10,039 in the clinical study.
Almost all e4 homozygotes showed Alzheimer pathology; by age 65 nearly all had abnormal spinal fluid amyloid and 75% had positive amyloid scans; symptom onset at 65.1 years.
Design: Observational Research that watches what people already do rather than assigning a treatment. It can show that two things travel together, not that one causes the other. clinical, biomarker A measurable sign in the body, such as a blood test value, used to track a condition or the effect of a treatment. and autopsy cohorts.
Why it sets no letter: EBL grades only randomized human trials. Describes the biology of two e4 copies, not the effect of learning the result.
Langbaum et al. (Lancet Healthy Longevity, 2025; PMID 41319675). 9,496 cognitively unimpaired, psychologically pre-screened adults aged 60 to 75 who learned their APOE genotype while being screened for the API Generation Study 1.
Disease-specific distress 2 to 7 days after disclosure rose more in e4 homozygotes than in heterozygotes and noncarriers, and anxiety rose in homozygotes while it fell in the other groups; depressive symptoms did not change, and the increases did not reach predefined levels of clinical concern.
Design: Cohort study An observational study that follows a group of people over time and compares what happens to them. within trial screening.
Why it sets no letter: EBL grades only randomized human trials. No nondisclosure comparison group; participants were screened for readiness before testing.
Grill et al. (JAMA Network Open, 2026; PMID 41910973). 3,414 adults screened for the AHEAD preclinical trial.
Intrusive thoughts and avoidance scores after disclosure were higher for e4 homozygotes (mean 12.7) than heterozygotes (9.1) and noncarriers (6.2), with group means below clinically significant thresholds.
Design: Cohort study within trial screening.
Why it sets no letter: EBL grades only randomized human trials. No nondisclosure comparison group.
Cited on: Avoiding ultra-processed food
Schnabel et al. (JAMA Internal Medicine, 2019; PMID 30742202). Ultra-processed food consumption and mortality in middle-aged adults.
Higher ultra-processed food consumption associated with higher all-cause mortality.
Design: Prospective observational cohort.
Why it sets no letter: EBL grades only randomized human trials. People who eat more ultra-processed food differ in income, education, smoking and activity in ways no adjustment removes. The randomized evidence above is stronger and does not need this to make the point.
Cited on: Daytime napping (length and timing)
Naska et al. (Arch Intern Med, 2007; PMID 17296887). Greek EPIC cohort, 23,681 adults.
Taking a siesta associated with lower coronary mortality (ratio 0.66), most evident in working men.
Design: Prospective observational cohort.
Why it sets no letter: EBL grades only randomized human trials. People who nap long or often differ from those who do not in age, health, sleep disorders and work patterns, and illness itself can lengthen naps, which no statistical adjustment fully removes. No randomized trial has assigned people to nap or not nap and measured this outcome.
Cited on: Daytime napping (length and timing)
Yamada et al. (Sleep, 2015; PMID 26158892). Dose-response meta-analysis A study that pools the numbers from several trials into one combined estimate. of 11 prospective cohorts, 151,588 participants.
Naps of 60 minutes or more a day associated with higher cardiovascular disease (RR 1.82) and all-cause mortality (RR 1.27); naps under 60 minutes not associated; J-shaped curve for cardiovascular disease lowest near 30 minutes.
Design: Meta-analysis of prospective observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. People who nap long or often differ from those who do not in age, health, sleep disorders and work patterns, and illness itself can lengthen naps, which no statistical adjustment fully removes. No randomized trial has assigned people to nap or not nap and measured this outcome.
Cited on: Daytime napping (length and timing)
Häusler et al. (Heart, 2019; PMID 31501230). Swiss CoLaus cohort, 3,462 adults.
Napping once or twice a week associated with fewer cardiovascular events (HR 0.52); daily napping not associated after adjustment.
Design: Prospective observational cohort.
Why it sets no letter: EBL grades only randomized human trials. People who nap long or often differ from those who do not in age, health, sleep disorders and work patterns, and illness itself can lengthen naps, which no statistical adjustment fully removes. No randomized trial has assigned people to nap or not nap and measured this outcome.
Cited on: Daytime napping (length and timing)
Pan et al. (Sleep Med, 2020; PMID 32858276). Dose-response meta-analysis of 20 cohorts, 313,651 participants.
Napping associated with higher all-cause mortality (HR 1.19); J-shaped curve for cardiovascular disease lowest near 25 minutes, risk rising with longer naps.
Design: Meta-analysis of observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. People who nap long or often differ from those who do not in age, health, sleep disorders and work patterns, and illness itself can lengthen naps, which no statistical adjustment fully removes. No randomized trial has assigned people to nap or not nap and measured this outcome.
Cited on: Daytime napping (length and timing)
Li et al. (Alzheimers Dement, 2023; PMID 35297533). Rush Memory and Aging Project, 1,401 older adults measured by actigraphy for up to 14 years.
Longer and more frequent naps associated with higher risk of Alzheimer's dementia, and worse cognition followed by more napping a year later.
Design: Prospective observational cohort.
Why it sets no letter: EBL grades only randomized human trials. People who nap long or often differ from those who do not in age, health, sleep disorders and work patterns, and illness itself can lengthen naps, which no statistical adjustment fully removes. No randomized trial has assigned people to nap or not nap and measured this outcome.
Cited on: Daytime napping (length and timing)
Vizmanos et al. (Obesity, 2023; PMID 37140401). ONTIME study, 3,275 Spanish adults.
Siestas over 30 minutes associated with higher BMI, waist, glucose, blood pressure and metabolic syndrome; short siestas associated with lower odds of elevated systolic pressure.
Design: Cross-sectional observational study.
Why it sets no letter: EBL grades only randomized human trials. People who nap long or often differ from those who do not in age, health, sleep disorders and work patterns, and illness itself can lengthen naps, which no statistical adjustment fully removes. No randomized trial has assigned people to nap or not nap and measured this outcome.
Cited on: Daytime napping (length and timing)
Yang et al. (Sleep Med Rev, 2024; PMID 39153335). Meta-analysis of 44 cohort studies, 1,864,274 participants.
Habitual napping associated with higher all-cause mortality, cardiovascular disease, metabolic disease and cancer, and with lower cognitive impairment and sarcopenia Age-related loss of muscle mass and strength.; naps of 30 minutes or longer carried the higher risks, naps under 30 minutes no significant risk.
Design: Meta-analysis of observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. People who nap long or often differ from those who do not in age, health, sleep disorders and work patterns, and illness itself can lengthen naps, which no statistical adjustment fully removes. No randomized trial has assigned people to nap or not nap and measured this outcome.
Cited on: Infrared Sauna
Laukkanen et al., JAMA Internal Medicine 2015 (PMID 25705824). 20-year follow-up of 2,315 Finnish men.
Sauna bathing 4 to 7 times per week was associated with a 63 percent lower risk of sudden cardiac death than once a week (hazard ratio How often an event happened in the treated group compared with the control group over the same period. A ratio of 0.80 means 20 percent fewer events. 0.37), with similar associations for coronary, cardiovascular and all-cause mortality. Traditional Finnish sauna, not infrared.
Design: Prospective observational cohort.
Why it sets no letter: EBL grades only published human randomized trials. This is a cohort study, so it cannot carry the grade no matter how large or how long the follow-up. It is shown because it is the best-known evidence on sauna and longevity and readers will encounter it elsewhere; it is excluded because men who sauna 4-7 times a week in Finland differ from men who do not in ways no adjustment fully removes. The grade rests on the randomized trials of surrogate and functional endpoints The outcome a trial measures to decide whether the treatment worked, such as heart attacks, blood pressure or a symptom score., not on this.
Cited on: Ketogenic diet (very-low-carbohydrate)
Qu et al. (Scientific Reports, 2024; PMID 39353986). Dietary ketogenic score and all-cause mortality.
A higher dietary ketogenic ratio was associated with lower all-cause mortality.
Design: Cross-sectional / prospective observational cohort (NHANES-based).
Why it sets no letter: EBL grades only randomized human trials. People who eat more ketogenically differ systematically from those who do not, and no adjustment removes that. The randomized evidence above is what sets the grade; this mortality association is context only and must not raise the letter.
Bohmer et al. (Sleep Health, 2021; PMID 34420891). Systematic review A review that searches for every study on a question using a method set in advance, then assesses them together. of 25 studies in healthy adults.
All included studies were cross-sectional; limited evidence for a link between light exposure and rest-activity rhythm, conflicting evidence for sleep, circadian phase and mood.
Design: Systematic review of cross-sectional studies.
Why it sets no letter: EBL grades only randomized human trials. The authors note the lack of intervention studies.
Burns et al. (Journal of Affective Disorders, 2021; PMID 34488088). UK Biobank, about 400,000 adults.
Each additional hour outdoors in daylight associated with lower odds of depression, easier waking, less tiredness, fewer insomnia symptoms and earlier chronotype.
Design: Cross-sectional and longitudinal cohort analysis of self-reported time outdoors.
Why it sets no letter: EBL grades only randomized human trials. Time outdoors was self-reported, and mood and sleep themselves change how much time people spend outside.
Windred et al. (Proceedings of the National Academy of Sciences of the United States of America, 2024; PMID 39405349). UK Biobank, 88,905 adults who wore light sensors for 1 week, followed 8 years.
Brighter day light associated with lower all-cause mortality (adjusted hazard ratios as low as 0.66 to 0.83 in the top decile) and brighter night light with higher mortality.
Design: Prospective cohort.
Why it sets no letter: EBL grades only randomized human trials. People who spend their days in bright light differ in health, mobility, work and outdoor activity, and illness keeps people indoors.
Boyle et al. (Psychosomatic Medicine, 2009; PMID 19414613). Rush Memory and Aging Project and Minority Aging Research Study, 1,238 older adults without dementia.
Higher purpose in life associated with lower mortality over up to 5 years (hazard ratio 0.60, 95% CI 0.42 to 0.87).
Design: Prospective cohort.
Why it sets no letter: EBL grades only randomized human trials. A cohort cannot show that raising purpose would lower the risk of death.
Boyle et al. (Archives of General Psychiatry, 2010; PMID 20194831). Rush Memory and Aging Project, 951 older adults without dementia.
Greater purpose in life associated with lower risk of Alzheimer disease (hazard ratio 0.48, 95% CI 0.33 to 0.69) and of mild cognitive impairment.
Design: Prospective cohort.
Why it sets no letter: EBL grades only randomized human trials. Early brain changes that precede dementia by years can lower purpose, so the direction of cause cannot be settled in a cohort.
Okun et al. (Psychology and Aging, 2013; PMID 23421326). Meta-analysis of organizational volunteering by adults 55 and older.
Volunteering associated with a 24% lower mortality risk after adjustment (95% CI 16% to 31%).
Design: Meta-analysis of observational studies.
Why it sets no letter: EBL grades only randomized human trials. Healthier people are more able to volunteer, so the association cannot show that volunteering lengthens life.
Cohen et al. (Psychosomatic Medicine, 2016; PMID 26630073). Meta-analysis of 10 prospective studies, 136,265 participants.
Higher purpose in life associated with lower all-cause mortality (adjusted pooled relative risk The risk in one group expressed as a share of the risk in another. It can make a small absolute difference look large. 0.83, 95% CI 0.75 to 0.91) and fewer cardiovascular events (0.83, 95% CI 0.75 to 0.92).
Design: Meta-analysis of observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. People with a strong sense of purpose differ from others in health, income, mood and activity, and illness itself can erode purpose, which no statistical adjustment fully removes.
Sutin et al. (Journal of Alzheimer's Disease, 2021; PMID 34275900). Meta-analysis of six samples, 53,499 participants, 5,862 incident dementia cases.
Greater sense of purpose associated with lower dementia risk (HR 0.77, 95% CI 0.73 to 0.81).
Design: Meta-analysis of observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. No randomized trial has tested whether a purpose program lowers dementia.
Farvid et al. (Circulation, 2014; PMID 25161045). Meta-analysis of 13 cohort studies, 310,602 people, 12,479 coronary events.
Highest against lowest linoleic acid intake associated with 15 percent fewer coronary events (RR 0.85) and 21 percent fewer coronary deaths (RR 0.79).
Design: Meta-analysis of prospective observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. People who eat more vegetable oil differ in many other habits, which no statistical adjustment fully removes; the randomized trials of the same swap did not show a consistent fall in events or deaths.
Zhou et al. (Journal of the American College of Cardiology, 2020; PMID 32616158). Analysis of 18,096 participants aged 70 or older in the ASPREE aspirin trial, comparing the 5,629 who took statins at baseline with those who did not.
Statin use was not associated with disability-free survival, death or dementia, but was associated with lower risks of physical disability and cardiovascular events.
Design: Observational comparison within a randomized trial of a different drug; nobody was randomized to a statin.
Why it sets no letter: EBL grades only randomized human trials. People who take statins differ from those who do not in health, care and risk, which weighting cannot fully remove. STAREE later tested the question directly.
Cited on: Sulforaphane (Broccoli Sprout Extract)
Baladia et al. (Nutrients, 2024; PMID 38892516). Meta-analysis of 23 case-control and 12 cohort studies of broccoli consumption.
Broccoli consumption associated with lower cancer risk in case-control studies (OR 0.64) and cohort studies (RR 0.89).
Design: Meta-analysis of observational studies.
Why it sets no letter: EBL grades only randomized human trials. People who eat more broccoli differ in diet, income and health habits, and the food is not the supplement; no randomized trial has tested sprout products on cancer incidence.
Dennis et al. (Annals of Epidemiology, 2008; PMID 18652979). Meta-analysis of 51 study populations.
Risk of melanoma rose with the number of sunburns in childhood, adolescence, adulthood and over a lifetime.
Design: Meta-analysis of observational studies.
Why it sets no letter: EBL grades only randomized human trials. Sunburn history is recalled, and people who burn differ in skin type and sun habits.
Boniol et al. (BMJ, 2012; PMID 22833605). Meta-analysis of 27 observational studies of sunbed use.
Ever use of sunbeds associated with melanoma (summary relative risk 1.20, 95% CI 1.08 to 1.34); first use before age 35, relative risk 1.87.
Design: Meta-analysis of case-control and cohort studies.
Why it sets no letter: EBL grades only randomized human trials. Sunbed users differ from non-users in outdoor sun exposure and skin type, and no trial can assign people to sunbeds.
Wehner et al. (BMJ, 2012; PMID 23033409). Meta-analysis of 12 studies, 9,328 cases of nonmelanoma skin cancer.
Ever use of indoor tanning associated with squamous cell carcinoma (relative risk 1.67, 95% CI 1.29 to 2.17) and basal cell carcinoma (1.29, 1.08 to 1.53).
Design: Meta-analysis of observational studies.
Why it sets no letter: EBL grades only randomized human trials. Associations in observational data cannot carry a letter.
Silva et al. (European Journal of Dermatology, 2018; PMID 29620003). Meta-analysis of 29 studies, mostly case-control, 313,717 participants.
No significant association between reported sunscreen use and skin cancer (odds ratio 1.08, 95% CI 0.91 to 1.28); early studies showed a positive association that faded over time.
Design: Meta-analysis of observational studies with one controlled trial.
Why it sets no letter: EBL grades only randomized human trials. People who use sunscreen are often those with fair skin and heavy sun exposure, which confounds the comparison.
Weber et al. (JAMA Dermatology, 2025; PMID 40991277). Meta-analysis of 17 studies, 321,473 participants.
Painful, blistering or severe sunburns associated with squamous cell carcinoma (odds ratio 1.51 for medium and 1.69 for high lifetime frequency).
Design: Meta-analysis of observational studies.
Why it sets no letter: EBL grades only randomized human trials.
Humphrey et al. (Journal of General Internal Medicine, 2008; PMID 18807098). Meta-analysis of 7 prospective cohorts prepared for the US Preventive Services Task Force.
Periodontal disease associated with higher coronary heart disease incidence, summary relative risks 1.24 (95% CI 1.01 to 1.51) to 1.34 (95% CI 1.10 to 1.63).
Design: Meta-analysis of observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. People with gum disease smoke more, have more diabetes and less access to care; the trials that treated gum disease have not shown fewer heart attacks or strokes.
Romandini et al. (Journal of Dental Research, 2021; PMID 32866427). Meta-analysis of 57 studies, 48 cohorts, 5.71 million participants.
Periodontitis associated with higher all-cause mortality (risk ratio 1.46, 95% CI 1.15 to 1.85) and losing all teeth with higher all-cause mortality (1.66, 95% CI 1.46 to 1.88).
Design: Meta-analysis of observational cohorts.
Why it sets no letter: EBL grades only randomized human trials. No randomized trial has tested whether treating gum disease or brushing more lowers death rates.
Asher et al. (Journal of the American Geriatrics Society, 2022; PMID 36073186). Meta-analysis of 47 longitudinal studies.
Poor periodontal health associated with cognitive decline (OR 1.23) and dementia (HR 1.21); the authors judged the evidence low quality and the associations at least partly due to reverse causality.
Design: Meta-analysis of observational longitudinal studies.
Why it sets no letter: EBL grades only randomized human trials. Early dementia makes tooth care harder, so poor gums can follow memory loss rather than cause it.
Cited on: Traditional Finnish Sauna (dry sauna)
Laukkanen et al., JAMA Internal Medicine 2015 (PMID 25705824). 2,315 men in eastern Finland followed for a median of 20.7 years.
Compared with one sauna a week, 4 to 7 a week was associated with a lower risk of sudden cardiac death (hazard ratio 0.37), fatal coronary disease, fatal cardiovascular disease and death from any cause.
Design: Prospective observational cohort.
Why it sets no letter: EBL grades on randomized trials. Men who use a sauna 4 to 7 times a week differ from men who use it once, in ways no adjustment fully removes, so a cohort cannot carry the letter however long it runs. It is shown because it is the most quoted sauna finding and readers will meet it elsewhere.
Cited on: Traditional Finnish Sauna (dry sauna)
Laukkanen et al., Age and Ageing 2017 (PMID 27932366). The same 2,315 Finnish men, median follow-up 20.7 years.
Compared with one sauna a week, 4 to 7 a week was associated with a lower risk of dementia (hazard ratio 0.34) and Alzheimer's disease (0.35).
Design: Prospective observational cohort.
Why it sets no letter: Observational. No trial has tested sauna against dementia.
Cited on: Traditional Finnish Sauna (dry sauna)
Zaccardi et al., American Journal of Hypertension 2017 (PMID 28633297). 1,621 Finnish men without hypertension, median follow-up 24.7 years.
4 to 7 sessions a week was associated with a lower risk of developing hypertension than one a week (hazard ratio 0.53 after full adjustment).
Design: Prospective observational cohort.
Why it sets no letter: Observational. The blood pressure trials above are the randomized evidence on the same question.
Cited on: Traditional Finnish Sauna (dry sauna)
Kunutsor et al., Neurology 2018 (PMID 29720543). 1,628 Finnish men and women aged 53 to 74, median follow-up 14.9 years.
Compared with one sauna a week, 4 to 7 a week was associated with a lower risk of stroke (hazard ratio 0.38 after adjustment for activity and socioeconomic status).
Design: Prospective observational cohort.
Why it sets no letter: Observational. No trial has tested sauna against stroke.
Owsley et al. (JAMA, 2002; PMID 12186601). Prospective cohort of 277 drivers with cataract, Alabama.
Drivers who chose surgery had half the crash rate of those who did not (rate ratio 0.47, 95% CI 0.23 to 0.94).
Design: Prospective observational cohort.
Why it sets no letter: EBL grades only randomized human trials. Patients chose whether to have surgery; no randomized trial has measured crashes.
Tseng et al. (Ophthalmology, 2016; PMID 26854033). 5% Medicare sample, 1,501,420 patients with cataract.
Cataract surgery associated with lower all-cause mortality (hazard ratio 0.73, 95% CI 0.72 to 0.74).
Design: Retrospective observational cohort.
Why it sets no letter: EBL grades only randomized human trials. Patients well enough for elective surgery are healthier on average, so lower death rates after surgery are expected without any causal effect. A separate cohort by the same group in the Women's Health Initiative was retracted and replaced (PMID 30352119) and is not relied on.
Lee et al. (JAMA Internal Medicine, 2022; PMID 34870676). Adult Changes in Thought cohort, 3,038 adults aged 65 and older with cataract or glaucoma, Kaiser Permanente Washington.
Cataract extraction associated with lower dementia risk (hazard ratio 0.71, 95% CI 0.62 to 0.83); glaucoma surgery, which does not restore vision, showed no association (1.08).
Design: Prospective observational cohort.
Why it sets no letter: EBL grades only randomized human trials. People who choose and are offered surgery differ in health from those who are not, which adjustment cannot fully remove.
Yeo et al. (Ophthalmology, 2024; PMID 38336283). Meta-analysis of 24 articles, 558,276 participants.
Cataract surgery associated with 25% lower risk of long-term cognitive decline (HR 0.75, 95% CI 0.72 to 0.78); GRADE certainty very low to low; the authors call for randomized trials.
Design: Meta-analysis, mainly of observational studies.
Why it sets no letter: EBL grades only randomized human trials. The pooled hazard ratio comes from cohorts; the only randomized trial on cognition (PMID 16961944) found no benefit.
Cited on: Vitamin D3
Garland et al. (American Journal of Public Health, 2014). Meta-analysis of all-cause mortality by serum 25-hydroxyvitamin D.
Lower serum 25-OH-D is associated with higher all-cause mortality.
Design: Meta-analysis of prospective observational studies of a serum LEVEL, not of supplementation.
Why it sets no letter: EBL grades only published human randomized trials. This pools cohorts of a blood level, and low vitamin D tracks illness, obesity and time spent indoors. It is shown because it is the single most-cited paper in support of vitamin D supplementation and it does not test supplementation at all. VITAL did, and found no benefit in replete adults.
Cited on: Whole-food plant-based diet
Tong et al. (BMJ, 2019; PMID 31484644). EPIC-Oxford, 18 years of follow-up.
Vegetarians had lower rates of ischemic heart disease than meat eaters and 20 percent higher rates of total stroke, mostly hemorrhagic stroke.
Design: Prospective observational cohort.
Why it sets no letter: EBL grades on randomized human trials. It is listed because it is the clearest population signal of a possible harm alongside the benefit, and a reader weighing the observational evidence is entitled to see both directions.
Genetic studies (Mendelian randomization), not trials
Studies that use inherited gene variants as a natural experiment. They describe whether an exposure sits on the path to an outcome. They never set a letter.
Cited on: Daytime napping (length and timing)
Dashti et al. (Nat Commun, 2021; PMID 33568662). Genome-wide study of napping in UK Biobank and 23andMe.
More frequent napping genetically linked to higher blood pressure and waist circumference.
Design: Genome-wide association study with Mendelian randomization A method that uses gene variants people are born with as a natural experiment, to test whether a factor causes an outcome..
Why it sets no letter: EBL grades only randomized human trials. Mendelian randomization uses gene variants linked to napping as a stand-in for the habit; it describes an exposure, not an intervention anyone was assigned to, and its assumptions cannot be tested from the data alone.
Cited on: Daytime napping (length and timing)
Yang et al. (Hypertension, 2022; PMID 35876003). UK Biobank, 358,451 participants, with Mendelian randomization.
Usual napping associated with hypertension (HR 1.12) and stroke (HR 1.24); the genetic analysis supported a causal role for nap frequency in hypertension.
Design: Prospective observational cohort plus Mendelian randomization.
Why it sets no letter: EBL grades only randomized human trials. Mendelian randomization uses gene variants linked to napping as a stand-in for the habit; it describes an exposure, not an intervention anyone was assigned to, and its assumptions cannot be tested from the data alone.
Cited on: Daytime napping (length and timing)
Paz et al. (Sleep Health, 2023; PMID 37344293). Mendelian randomization in UK Biobank.
Habitual napping genetically linked to larger total brain volume, with no difference in hippocampal volume, reaction time or visual memory.
Design: Mendelian randomization.
Why it sets no letter: EBL grades only randomized human trials. Mendelian randomization uses gene variants linked to napping as a stand-in for the habit; it describes an exposure, not an intervention anyone was assigned to, and its assumptions cannot be tested from the data alone.
Genetic association studies, not trials
Studies that measure how a gene variant relates to risk. They describe the variant, not anything a person can do about it.
Bennet et al. (JAMA, 2007; PMID 17878422). Meta-analysis of 121 studies of coronary outcomes, 37,850 cases and 82,727 controls.
Against e3/e3, coronary disease odds ratio 1.06 (95% CI 0.99 to 1.13) in e4 carriers and 0.80 (95% CI 0.70 to 0.90) in e2 carriers; LDL Low-density lipoprotein cholesterol, the cholesterol fraction most closely tied to plaque in the arteries. cholesterol rose roughly linearly from e2/e2 to e4/e4.
Design: Genetic association meta-analysis.
Why it sets no letter: EBL grades only randomized human trials. The coronary association for e4 is small and its interval includes no effect.
Genin et al. (Molecular Psychiatry, 2011; PMID 21556001). 7,351 cases and 10,132 controls of European ancestry, with US incidence data.
Lifetime risk of Alzheimer disease by age 85: 11% in men and 14% in women overall; 23% (men) to 30% (women) for e3/e4; 51% (men) to 60% (women) for e4/e4.
Design: Genetic association study with modeled lifetime risk.
Why it sets no letter: EBL grades only randomized human trials. Lifetime risk estimates are modeled from observational data and differ by ancestry and data source.
Belloy et al. (JAMA Neurology, 2023; PMID 37930705). 68,756 individuals of East Asian, Hispanic, Black and White ancestry.
Odds ratio for Alzheimer disease with e3/e4: 4.54 in East Asian, 3.46 in White, 2.18 in Black and 1.90 in Hispanic individuals.
Design: Genetic association study.
Why it sets no letter: EBL grades only randomized human trials. Shows that the risk attached to the same genotype differs by ancestry.
Farrer et al. (JAMA, 1997; PMID 9343467). Meta-analysis by 40 research teams, 5,930 patients with Alzheimer disease and 8,607 controls.
In White clinic or autopsy samples, odds of Alzheimer disease against e3/e3: e3/e4 OR 3.2 (95% CI 2.8 to 3.8), e4/e4 OR 14.9 (95% CI 10.8 to 20.6), e2/e3 OR 0.6. The e4 association was weaker in African American and Hispanic people and stronger in Japanese people (e4/e4 OR 33.1), and diminished after age 70.
Design: Case-control genetic association meta-analysis.
Why it sets no letter: EBL grades only randomized human trials. These figures describe risk; they do not show that knowing the result changes it. Odds ratios from case-control samples overstate the absolute risk The actual chance of an event in a group. A fall from 2 percent to 1 percent is a 1-point absolute reduction, even though it is a 50 percent relative one. an individual carrier faces.
Human studies with no comparison group
Studies in which everyone received the treatment. Without a comparison group, change over time cannot be separated from what would have happened anyway.
Cited on: Infrared Sauna
Imamura et al., J Am Coll Cardiol 2001 (PMID 11583886).
Two weeks of daily far-infrared sauna raised flow-mediated dilation from 4.0 to 5.8 percent in 25 men with coronary risk factors.
Design: Uncontrolled before-and-after study.
Why it sets no letter: Shown on the page as background; not a randomized trial of this modality on a graded endpoint, so it does not enter the grade.
Cited on: Sulforaphane (Broccoli Sprout Extract)
Galan et al. (Dig Dis Sci, 2004; PMID 15387326). Nine patients with Helicobacter pylori ate broccoli sprouts for 7 days.
Eradication was temporally associated in three of nine patients.
Design: Uncontrolled preliminary report.
Why it sets no letter: No control group.
Cited on: Sulforaphane (Broccoli Sprout Extract)
Riedl et al. (Clin Immunol, 2009; PMID 19028145). Placebo-controlled dose escalation of broccoli sprout homogenate.
Phase II enzyme RNA rose in nasal lavage cells, maximal at 200 g of sprouts.
Design: Dose-escalation trial, not indexed or described as randomized.
Why it sets no letter: Not a randomized comparison; a mechanism marker only.
Cited on: Sulforaphane (Broccoli Sprout Extract)
Zeng et al. (Eur Arch Psychiatry Clin Neurosci, 2024; PMID 37728803). 45 patients with schizophrenia given sulforaphane tablets 90 mg/day for 24 weeks.
PANSS negative subscale scores fell from baseline.
Design: Single-arm study without a control group.
Why it sets no letter: No randomized comparison; improvement from baseline cannot be separated from time, placebo An inactive substitute made to look like the real treatment, so participants cannot tell which one they received. response or regression to the mean.
Controlled studies that were not randomized
Studies with a comparison group chosen some other way, such as by hospital or by choice, so the groups may differ before treatment begins.
Turpeinen et al., Finnish Mental Hospital Study (International Journal of Epidemiology, 1979; PMID 393644). Two mental hospitals near Helsinki, 1959 to 1971.
Coronary heart disease during the cholesterol-lowering diet periods was about half that during the normal-diet periods in both hospitals.
Design: Controlled trial allocated by hospital: one hospital received the diet and the other served as control, then the diets were reversed after 6 years.
Why it sets no letter: The abstract describes allocation by hospital, two units in all, not randomization of people or of enough clusters to balance them, and the populations changed over time. EBL gives no letter to it; some meta-analyses pooled it.
Cited on: Sulforaphane (Broccoli Sprout Extract)
Lozanovski et al. (Invest New Drugs, 2020; PMID 31250356), POUDER pilot. 40 patients with advanced pancreatic cancer on palliative chemotherapy, 29 on broccoli sprout capsules and 11 on placebo, up to 1 year.
Lower early death rate on sprouts, not statistically significant Unlikely to be explained by chance alone at the threshold the study set. It says nothing about whether the difference is large enough to matter. (p = 0.291 at day 180); high dropout; digestive problems, nausea and vomiting with the capsules.
Design: Allocation described as unblinded, with unequal groups.
Why it sets no letter: The report does not describe randomized blinded allocation and the result was not significant; the harms it reports are carried in the avoid field.
Cited on: Traditional Finnish Sauna (dry sauna)
Ernst et al., Annals of Medicine 1990 (PMID 2248758). 25 volunteers using a sauna and 25 controls, followed for 6 months.
Fewer colds in the sauna group, roughly half as many in the last 3 months; duration and severity of colds did not differ.
Design: Controlled trial; the abstract does not describe randomization.
Why it sets no letter: Not described as randomized, and 50 people. Shown because it is the source of the claim that sauna prevents colds.
Trial evidence not counted toward this letter
Randomized evidence that does not answer the use graded on the record: a different combination, a different endpoint or a single session. The reason is stated on each entry.
Cited on: Audiovisual / Brainwave Entrainment
de Souza et al., Pain Medicine 2026 (PMID 42116620). 84 women with fibromyalgia randomized to brainwave entrainment via a BrainTap headset or to music therapy, 20 minutes daily, five days a week, eight weeks, with six-month follow-up.
Pain and anxiety fell in the entrainment arm; quality of life, depression and oxidative-stress measures improved in both arms.
Design: Randomized controlled trial People are assigned to the treatment or a comparison by chance, so the groups differ only in what they received. It is the design that can show cause and effect. by description. PubMed indexes it as Journal Article only, with no interventional publication type.
Why it sets no letter: Under Rule 01 a grade may rest only on evidence independently indexed as interventional. An unindexed design is a fact about the indexing rather than about the study, which is why it is recorded here in full instead of being dropped. It is also the only trial anywhere that tested a light-and-sound headset as a unit, and its comparator was music therapy, not a sham A fake version of a procedure or device used as the comparison, so participants cannot tell which one they received..
Cited on: Cold Water Immersion
Roberts et al. Post-exercise cold water immersion attenuates acute anabolic signaling and long-term adaptations in muscle to strength training.
Immersion after each resistance session for 12 weeks produced smaller gains in strength and muscle mass than active recovery; satellite-cell and anabolic-signaling responses to a single bout were blunted.
Design: Randomized human trial, 21 men, plus a 9-man crossover A trial design in which each participant receives both the treatment and the comparison, one after the other, and serves as their own control..
Why it sets no letter: Not evidence for stress, sleep or quality of life, and not excluded from the record: it is a harm signal for a common use. Two meta-analyses, Grgic 2023 (PMID 35068365) and Malta 2021 (PMID 33146851), replicate the blunted strength gains. Recorded here so the reader sees that a cold plunge after lifting appears to reduce the adaptation lifted for.
Cited on: Infrared Sauna
Gayda et al., Journal of Clinical Hypertension 2012 (PMID 22863164).
Exercise followed by sauna lowered 24-hour systolic and mean blood pressure in 16 patients with untreated hypertension.
Design: Small randomized crossover of single sessions, traditional sauna.
Why it sets no letter: Shown on the page as background; not a randomized trial of this modality on a graded endpoint, so it does not enter the grade.
Mitchell et al., Sleep Medicine Reviews 2019 (PMID 31377503).
CBT-I did not reliably improve polysomnography sleep measures; sleep diaries improved.
Design: Meta-analysis of 15 studies.
Why it sets no letter: Shown on the page as background; not a randomized trial of this modality on a graded endpoint, so it does not enter the grade.
Cited on: Sulforaphane (Broccoli Sprout Extract)
van Die et al. (Prostate, 2017; PMID 28181675). 22 men with biochemically recurrent prostate cancer, a combination of turmeric, resveratrol, green tea and broccoli sprouts for 12 weeks.
Non-significant rise in PSA log-slope on the combination.
Design: Randomized pilot of a four-ingredient combination.
Why it sets no letter: The effect of broccoli sprouts cannot be separated from the other three ingredients.
Cited on: Whole-food plant-based diet
Ornish et al. (JAMA, 1998; PMID 9863851). The Lifestyle Heart Trial, five-year results.
In 48 patients with coronary heart disease, an intensive program of a very low-fat vegetarian diet, exercise, stress management, smoking cessation and group support slowed or reversed narrowing of the coronary arteries, and the program group had fewer cardiac events than usual care.
Design: Small randomized trial of a multi-part lifestyle program.
Why it sets no letter: The diet was one of five changes made together, so the trial cannot say how much of the result came from the diet. It is the most cited interventional result for this pattern and it is listed so a reader can find it, but it does not test the graded intervention.
Measurement studies, not outcome trials
Studies of how well a test or device measures something, not of whether using it changes health.
Cited on: Home blood pressure monitoring
Bliziotis et al., Journal of Hypertension 2012 (PMID 22499289).
Home blood pressure predicted target organ damage as well as ambulatory monitoring and better than office readings.
Design: Systematic review and meta-analysis of 23 comparative studies.
Why it sets no letter: Shown on the page as background; not a randomized trial of this modality on a graded endpoint, so it does not enter the grade.
Reviews and consensus statements, not trials
Narrative reviews, position papers and consensus statements. They summarize opinion and evidence; they are not new evidence.
Cited on: Audiovisual / Brainwave Entrainment
Ingendoh et al., PLoS One 2023 (PMID 37205669). Systematic review of 14 studies measuring EEG activity during binaural-beat stimulation.
Five studies supported the brainwave entrainment hypothesis, eight contradicted it, one was mixed.
Design: Systematic review of mechanistic EEG studies. No clinical outcome.
Why it sets no letter: It measures whether the brain entrains, not whether anyone feels or performs better. It cannot carry a letter because it reports no clinical endpoint. It is shown because it is the reason this modality is C rather than B: the mechanism the whole category is named after is not established.
Cited on: DEXA Body Composition Scan
Ross et al., Nature Reviews Endocrinology 2020 (PMID 32020062).
Waist circumference should be measured routinely alongside body-mass index.
Design: Consensus statement.
Why it sets no letter: Shown on the page as background; not a randomized trial of this modality on a graded endpoint, so it does not enter the grade.
Cited on: Home blood pressure monitoring
Parati et al., Journal of Hypertension 2021 (PMID 34269334).
European Society of Hypertension working group recommendations on home blood pressure monitoring.
Design: Position paper.
Why it sets no letter: Shown on the page as background; not a randomized trial of this modality on a graded endpoint, so it does not enter the grade.
Irwin, Nature Reviews Immunology 2019 (PMID 31289370).
Sleep and the immune system regulate each other; disturbed sleep is linked to inflammatory and antiviral responses.
Design: Narrative review.
Why it sets no letter: Shown on the page as background; not a randomized trial of this modality on a graded endpoint, so it does not enter the grade.
Animal and laboratory evidence
Work in animals or cells. It can explain a mechanism. It cannot show an effect in people.
Cited on: Exosome Therapy (extracellular vesicles)
Systematic review of exosome-driven hair regrowth (Mol Biol Rep 2026).
Reports supportive findings for hair regrowth.
Design: Systematic review of PRECLINICAL studies.
Why it sets no letter: EBL grades only published human randomized trials. This review pools animal and laboratory work. It is noted because it is frequently cited as clinical support for exosome hair treatments and is not clinical evidence.