What the evidence does not support
Most sites in this field publish only what they can recommend. This page is the other half of the same work: what was tested and did not work, and what nobody has tested at all. A null result is a finding. An absence of evidence is a different finding, and the two are separated below.
Tested in humans, and it did not work
130 entries where a trial was run against a named endpoint The outcome a trial measures to decide whether the treatment worked, such as heart attacks, blood pressure or a symptom score. and it did not show benefit. In 111 the result was null. In 19 the treated arm did worse than placebo An inactive substitute made to look like the real treatment, so participants cannot tell which one they received. on that endpoint, and those are marked HARM FOUND. Each one names the endpoint, who was studied, and the paper.
Endpoint tested: Atrial fibrillation recurrence in the 28 days after ablation, on treatment. Population: 93 adults randomized, 82 analyzed, undergoing atrial fibrillation ablation; 750 mg three times daily against placebo Low to moderate. One single-center pilot trial stopped early for futility; the harm was pre-specified and survived post hoc adjustment, and did not persist after the drug was stopped. PMID 40709466
Endpoint tested: Atrial fibrillation recurrence in the 28 days after ablation, on treatment. Population: 93 adults randomized, 82 analyzed, 750 mg three times daily against placebo Low to moderate. One pilot trial, stopped early for futility. PMID 40709466
Endpoint tested: Proportion reaching a 20 percent or greater reduction in fatigue visual analog score after 8 weeks. Population: 166 adults with inflammatory bowel disease in remission and fatigue, 13 Belgian hospitals Moderate for a null: multicenter, randomized, 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.; 35.6 percent versus 37.6 percent, P = .830; secondary fatigue scales also null. PMID 35940251
Endpoint tested: Whole-body insulin sensitivity (primary). Population: 142 adults with metabolic syndrome, pasteurized MucT for 4 months Moderate. The scale-up of the positive pilot, run by the maker; the primary endpoint The one outcome a trial names in advance as its main test. Results on other outcomes carry less weight. did not differ from placebo and benefit appeared only in exploratory subgroups. PMID 42343233
Endpoint tested: Weight and HbA1c A blood test that reflects average blood sugar over the previous two to three months. between groups. Population: 58 adults with overweight or obese type 2 diabetes, a different strain (AKK-WST01), 12 weeks Low to moderate. Both arms improved with no significant between-group difference; benefit only in a low-baseline-Akkermansia subgroup. PMID 39879980
Endpoint tested: Disability-free survival (composite of death, dementia or persistent physical disability), the primary endpoint; major hemorrhage (secondary). Population: 19,114 healthy adults aged 70 and over, 100 mg enteric-coated aspirin daily, median 4.7 years (ASPREE) High. Large randomized placebo-controlled primary-prevention trial, stopped for futility. 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. 1.01 for the primary endpoint; major hemorrhage increased. PMID 30221596
Why this grade, and what it does not cover
Grade D on Route 2 Human randomized trials have been run and did not show benefit on the stated outcome. One of the two routes to the lowest letter on EBL's scale.: tested, harm found. ASPREE randomized 19,114 healthy adults aged 70 and over to 100 mg aspirin daily or placebo for a median of 4.7 years. Disability-free survival, the primary endpoint, was not improved, cardiovascular events were not meaningfully reduced, and major hemorrhage increased. In adjacent populations, ASCEND (diabetes) found fewer vascular events but more bleeding, and ARRIVE (moderate risk) was null with doubled gastrointestinal bleeding; neither enrolled healthy older adults. During ASPREE, all-cause death was higher on aspirin, mostly from cancer, which the investigators called unexpected; a decade of follow-up found no lasting death excess, no long-term benefit, and a persisting excess of major bleeding. The finding is not that aspirin does nothing, it is that in people who have not had an event, the bleeding it causes outweighs the events it prevents. WHAT THIS GRADE DOES NOT COVER: this letter grades starting low-dose aspirin for PRIMARY prevention in healthy older adults. It does not grade aspirin taken after a heart attack, a stent, a stroke or a transient ischemic attack. That is secondary prevention, a different question with different trials and a different answer, and stopping it abruptly raises the risk of a further event. Nothing on this page is a basis for changing a prescribed medication, and that decision sits with the prescribing physician.
Endpoint tested: Cardiovascular disease (fatal coronary disease, myocardial infarction, stroke, heart-failure hospitalization) and major hemorrhage. Population: The same 19,114 ASPREE participants High. Cardiovascular events hazard ratio 0.95 (95% CI 0.83 to 1.08); major hemorrhage hazard ratio 1.38 (1.18 to 1.62). PMID 30221597
Why this grade, and what it does not cover
Grade D on Route 2: tested, harm found. ASPREE randomized 19,114 healthy adults aged 70 and over to 100 mg aspirin daily or placebo for a median of 4.7 years. Disability-free survival, the primary endpoint, was not improved, cardiovascular events were not meaningfully reduced, and major hemorrhage increased. In adjacent populations, ASCEND (diabetes) found fewer vascular events but more bleeding, and ARRIVE (moderate risk) was null with doubled gastrointestinal bleeding; neither enrolled healthy older adults. During ASPREE, all-cause death was higher on aspirin, mostly from cancer, which the investigators called unexpected; a decade of follow-up found no lasting death excess, no long-term benefit, and a persisting excess of major bleeding. The finding is not that aspirin does nothing, it is that in people who have not had an event, the bleeding it causes outweighs the events it prevents. WHAT THIS GRADE DOES NOT COVER: this letter grades starting low-dose aspirin for PRIMARY prevention in healthy older adults. It does not grade aspirin taken after a heart attack, a stent, a stroke or a transient ischemic attack. That is secondary prevention, a different question with different trials and a different answer, and stopping it abruptly raises the risk of a further event. Nothing on this page is a basis for changing a prescribed medication, and that decision sits with the prescribing physician.
Endpoint tested: All-cause mortality (secondary endpoint An outcome a trial also measures but did not choose as its main test.). Population: The same 19,114 ASPREE participants Moderate. Hazard ratio 1.14 (95% CI 1.01 to 1.29), driven by cancer deaths; the authors call it unexpected; the extended follow-up (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. 41043446) found no lasting excess. PMID 30221595
Why this grade, and what it does not cover
Grade D on Route 2: tested, harm found. ASPREE randomized 19,114 healthy adults aged 70 and over to 100 mg aspirin daily or placebo for a median of 4.7 years. Disability-free survival, the primary endpoint, was not improved, cardiovascular events were not meaningfully reduced, and major hemorrhage increased. In adjacent populations, ASCEND (diabetes) found fewer vascular events but more bleeding, and ARRIVE (moderate risk) was null with doubled gastrointestinal bleeding; neither enrolled healthy older adults. During ASPREE, all-cause death was higher on aspirin, mostly from cancer, which the investigators called unexpected; a decade of follow-up found no lasting death excess, no long-term benefit, and a persisting excess of major bleeding. The finding is not that aspirin does nothing, it is that in people who have not had an event, the bleeding it causes outweighs the events it prevents. WHAT THIS GRADE DOES NOT COVER: this letter grades starting low-dose aspirin for PRIMARY prevention in healthy older adults. It does not grade aspirin taken after a heart attack, a stent, a stroke or a transient ischemic attack. That is secondary prevention, a different question with different trials and a different answer, and stopping it abruptly raises the risk of a further event. Nothing on this page is a basis for changing a prescribed medication, and that decision sits with the prescribing physician.
Endpoint tested: Lung cancer incidence; total mortality. Population: 29,133 male smokers aged 50 to 69, Finland, beta-carotene 20 mg/day alone, 5 to 8 years (ATBC) High. Large randomized, double-blind Neither the participants nor the people assessing them know who received the real treatment until the trial ends., placebo-controlled primary-prevention trial. Lung cancer incidence 18 percent higher (95% CI 3 to 36) and total mortality 8 percent higher on beta-carotene. PMID 8127329
Endpoint tested: Lung cancer incidence; death from any cause. Population: 18,314 smokers, former smokers and asbestos-exposed workers, beta-carotene 30 mg/day plus retinyl palmitate 25,000 IU/day, a combination, mean 4 years (CARET) High. Randomized placebo-controlled trial stopped 21 months early: lung cancer 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. 1.28 (95% CI 1.04 to 1.57), death 1.17 (1.03 to 1.33). The combination, not beta-carotene alone, was tested. PMID 8602180
Endpoint tested: Malignant neoplasms, cardiovascular disease, mortality. Population: 22,071 male physicians, 11 percent current and 39 percent former smokers, beta-carotene 50 mg on alternate days, 12 years (Physicians' Health Study) High. Relative risk of any malignancy 0.98 (95% CI 0.91 to 1.06); no difference in lung cancer, cancer death, cardiovascular disease or overall mortality. PMID 8602179
Endpoint tested: Cancer incidence, cardiovascular disease, total mortality. Population: 39,876 women aged 45 and over, beta-carotene 50 mg on alternate days, median 2.1 years of treatment and 4.1 years of follow-up (Women's Health Study) High for a null over the period studied: cancer relative risk 1.03 (95% CI 0.89 to 1.18), no difference in cardiovascular disease or mortality; the beta-carotene arm was stopped early after the ATBC and CARET results. PMID 10601381
Endpoint tested: Plasma testosterone and lean body mass. Population: 19 male bodybuilders aged 20 to 27, 2.5 mg/day for 7 weeks Low. Small controlled trial (PubMed types it as a clinical trial, not a randomized one); null. PMID 8508192
Endpoint tested: Bone mineral density. Population: 28 female college students, 17 athletes and 11 sedentary, one year Low. Small controlled trial (PubMed types it as a clinical trial, not a randomized one); no boron effect. PMID 7889886
Endpoint tested: Mineral balance, plasma sex steroids and urinary pyridinium crosslink markers of bone turnover. Population: Healthy postmenopausal women in a metabolic unit, low-boron diet for 3 weeks then boron 3 mg/day for 3 weeks Low. Small crossover-style metabolic study; no effect on any measure; the low-boron diet appeared to induce calcium hyperabsorption, which may have obscured an effect. PMID 8329361
Endpoint tested: WOMAC total and subscales. Population: 47 adults with moderate to severe knee osteoarthritis, bromelain 800 mg/day for 12 weeks Low to moderate. Small single placebo-controlled trial; null. PMID 17121765
Endpoint tested: Fibrinogen and CRP. Population: Adults with diabetes, bromelain 1,050 mg/day for 12 weeks Low. Small exploratory placebo-controlled trial; null. PMID 27412590
Endpoint tested: Hand grip strength, appendicular lean mass, body fat, visceral fat, tibial bone density and stress-strain index. Population: 49 adults with prediabetes or type 2 diabetes and no musculoskeletal condition, carnosine 2 g/day for 14 weeks Low to moderate. Small double-blind placebo-controlled trial; null on every musculoskeletal measure. PMID 39770949
Endpoint tested: Hand grip strength and physical performance measures overall. Population: 299 adults, carnosine 2 g/day for 12 weeks Moderate for the null on grip strength; endurance findings appeared only in age and sex subgroups. PMID 42308284
Endpoint tested: Inflammatory markers immediately and 1 hour after immersion. Population: Healthy adults in randomized trials of cold-water immersion Moderate for the acute rise; no trial measured a longer-term anti-inflammatory effect. PMID 39879231
Endpoint tested: Mood. Population: Healthy adults in randomized trials of cold showers, baths or plunges at 15C or below Low. Few small trials; the review reports no mood effect. PMID 39879231
Endpoint tested: Mood, depressive symptoms, perceived stress and wellbeing. Population: Adults exposed to cold showers, cold-water immersion or winter swimming, 13 studies, 4,105 participants Low. Most studies at high, serious or critical risk of bias; estimates attenuated in controlled-only analyses. PMID 42819812
Endpoint tested: Strength and muscle mass gains after 12 weeks of resistance training; satellite cell and anabolic signaling responses to a single bout. Population: 21 men randomized to cold-water immersion or active recovery after each session Moderate. Smaller gains in strength and muscle mass and blunted anabolic signaling with immersion; one trial, replicated The same finding reproduced by a separate research group. A result seen once can be chance or bias; a replicated one is harder to explain away. in direction by two pooled analyses. PMID 26174323
Endpoint tested: Strength adaptations to resistance training. Population: Meta-analysis A study that pools the numbers from several trials into one combined estimate. of 8 studies Moderate. Harmful effect on strength adaptations (SMD -0.60); no effect on endurance adaptations. PMID 33146851
Endpoint tested: Strength gains with training. Population: Meta-analysis of 10 studies Moderate. Strength gains attenuated (effect size How large a difference the treatment made, as distinct from whether the difference could be detected at all. -0.23). PMID 35068365
Endpoint tested: Strength and hypertrophy adaptations with regular post-exercise immersion. Population: Umbrella review of 15 systematic reviews A review that searches for every study on a question using a method set in advance, then assesses them together. and meta-analyses, 161 unique studies Low to very low by the review's own rating; consistent in direction with the trial and two meta-analyses above. PMID 42816732
Endpoint tested: Bone mineral density at femoral neck, total hip and lumbar spine. Population: 237 postmenopausal women on creatine plus resistance training and walking Two-year 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., the largest run on this question PMID 37144634
Endpoint tested: Areal bone mineral density at lumbar spine, femoral neck and total femur. Population: 200 postmenopausal women with osteopenia Two-year randomized placebo-controlled trial; bone markers, microarchitecture, falls and fractures also unchanged PMID 31257405
Endpoint tested: Number and surface area of open finger ulcers, infection, inflammation and patient-rated pain. Population: 84 patients with systemic sclerosis Randomized, double-blind, multicenter. No difference among saline, 2% and 70% DMSO on any measure, and over a quarter of the 70% group were withdrawn for skin toxicity. PMID 3884019
Endpoint tested: Development of pressure ulcers. Population: Nursing home patients prone to pressure ulcers Randomized, double-blind crossover. The three interventions did not differ; the authors report that DMSO did not fulfil the expectations raised by earlier work. PMID 17553503
Endpoint tested: ADHD symptoms (blinded ratings) at treatment end and 13 months; 25-month follow-up (PMID 36521694) also without a specific effect. Population: 144 children with ADHD, double-blind sham-controlled Moderate to high. Largest sham-controlled trial; null at every time point. PMID 32853703
Endpoint tested: ADHD symptoms, probably blinded ratings. Population: Meta-analysis of 38 randomized trials Moderate. SMD 0.04 on blinded ratings. PMID 39661381
Endpoint tested: ADHD Rating Scale IV severity and Clinical Global Impressions-Improvement. Population: 41 children aged 8 to 15 with ADHD, double-blind, placebo neurofeedback control, 30 sessions Moderate. Independent trial; null, with blinding confirmed by guess rates at chance. PMID 24021501
Endpoint tested: Systolic and diastolic blood pressure. Population: 10 acute and chronic studies, 187 participants Low. Small studies, randomized and nonrandomized pooled together. PMID 38145410
Endpoint tested: Endurance performance (time trial and time to exhaustion). Population: 8 randomized trials, 80 participants Moderate. Consistent null across trial types; small trials, almost all men. PMID 35042186
Endpoint tested: Peak oxygen uptake and constant-load exercise endurance. Population: 20 adults with heart failure with preserved ejection fraction, randomized placebo-controlled crossover Low. One small acute trial; robust ketosis was confirmed. PMID 40243975
Endpoint tested: Migraine days in the last 4 weeks of treatment. Population: 41 adults with episodic migraine, randomized double-blind crossover Low to moderate. One single-center trial; wide confidence interval The range the true effect probably falls in. A wide interval means the estimate is uncertain.. PMID 34541914
Endpoint tested: Knee osteoarthritis pain and function. Population: Adults with knee osteoarthritis, intra-articular extracellular vesicles Low. Small triple-blind placebo-controlled trial; null. PMID 39465400
Endpoint tested: Incidence of colorectal adenoma and advanced adenoma. Population: 1,021 adults with prior colorectal adenoma, 1 mg/day, colonoscopic surveillance over 3 and 6 years Moderate to high. Randomized placebo-controlled trial. No reduction. Authors flagged a possible increase in neoplasia requiring further study. PMID 17551129
Endpoint tested: Recurrent colorectal adenoma. Population: 672 adults with prior adenoma from two cohorts, 1 mg/day for 3 to 6.5 years Moderate. Placebo-controlled; no overall reduction (RR 0.82, 95% CI 0.59 to 1.13). PMID 19864409
Endpoint tested: Any adenoma and advanced adenoma, pooled. Population: 2,632 adults with prior adenoma across three randomized trials at 0.5 or 1 mg/day High. Combined analysis; RR 0.98 for all adenomas, 1.06 for advanced lesions. PMID 21170989
Endpoint tested: Cardiovascular death, myocardial infarction and stroke (composite). Population: 5,522 adults aged 55 and over with vascular disease or diabetes (HOPE-2) High. Large randomized, placebo-controlled trial. Null on the primary composite; stroke lower as a component. PMID 16531613
Endpoint tested: Recurrent myocardial infarction, stroke and sudden coronary death (composite). Population: 3,749 patients randomized within seven days of a heart attack (NORVIT) High. Randomized, placebo-controlled factorial trial. Null, with a borderline harm signal in the three-vitamin arm. PMID 16531614
Endpoint tested: Total mortality (also cardiovascular and cancer mortality). Population: 251,891 asymptomatic adults across 17 randomized trials High. Cochrane systematic review. RR 1.00 for total mortality. Review concludes health checks are unlikely to be beneficial. PMID 30699470
Why this grade, and what it does not cover
Grade D on Route 2: tested, no benefit demonstrated, at a level of certainty almost nothing else on this site reaches. The 2019 Cochrane review pooled 17 randomized trials and 251,891 participants comparing general health checks against no health checks in adults unselected for disease. All-cause mortality was RR 1.00 (95% CI 0.97 to 1.03, 11 trials, 233,298 participants, 21,535 deaths, high-certainty, I-squared 0%). Cancer mortality at RR 1.01 and ischemic heart disease at RR 0.98 were likewise flat. An interval that tight around 1.00 is not an underpowered null, it is a well-measured absence of effect. Inter99 (PMID 24912589), the largest single modern trial, and a 2026 systematic review through 2025 (PMID 42414985) agree. WHAT THIS GRADE DOES NOT COVER: the pooled trials tested health checks of their era (questionnaires, examination, a few laboratory tests, often counseling), not the 100-plus marker direct-to-consumer panels sold today, and no randomized trial has tested a consumer advanced-biomarker panel against any clinical endpoint. Individual markers measured inside such panels, including ApoB Apolipoprotein B, a protein carried once on each cholesterol particle that can enter the artery wall, so it counts those particles directly., Lp(a) Lipoprotein(a), a cholesterol-carrying particle whose level is set largely by genetics. and HbA1c, carry their own evidence and their own separate grades. It also does not grade targeted screening in someone with a specific indication, which is a different question with different trials behind it.
Endpoint tested: Ischemic heart disease, stroke and total mortality at 10 years. Population: 59,616 adults aged 30 to 60 in Copenhagen suburbs, screening plus repeated lifestyle counseling against no invitation High. Largest single randomized trial of a modern health check; hazard ratios 1.03, 0.98 and 1.00. PMID 24912589
Why this grade, and what it does not cover
Grade D on Route 2: tested, no benefit demonstrated, at a level of certainty almost nothing else on this site reaches. The 2019 Cochrane review pooled 17 randomized trials and 251,891 participants comparing general health checks against no health checks in adults unselected for disease. All-cause mortality was RR 1.00 (95% CI 0.97 to 1.03, 11 trials, 233,298 participants, 21,535 deaths, high-certainty, I-squared 0%). Cancer mortality at RR 1.01 and ischemic heart disease at RR 0.98 were likewise flat. An interval that tight around 1.00 is not an underpowered null, it is a well-measured absence of effect. Inter99 (PMID 24912589), the largest single modern trial, and a 2026 systematic review through 2025 (PMID 42414985) agree. WHAT THIS GRADE DOES NOT COVER: the pooled trials tested health checks of their era (questionnaires, examination, a few laboratory tests, often counseling), not the 100-plus marker direct-to-consumer panels sold today, and no randomized trial has tested a consumer advanced-biomarker panel against any clinical endpoint. Individual markers measured inside such panels, including ApoB, Lp(a) and HbA1c, carry their own evidence and their own separate grades. It also does not grade targeted screening in someone with a specific indication, which is a different question with different trials behind it.
Endpoint tested: Diet quality and health-related behavior change at 6 months, genotype-based advice against personalized advice without genotype. Population: Adults in seven European countries; 1,269 completed Moderate. One large randomized trial; endpoint is self-reported diet. PMID 27524815
Endpoint tested: Dietary and physical activity behavior, genotype-based advice against general advice. Population: 11 studies pooled; adults in the general population or at risk of cardiovascular disease, type 2 diabetes or obesity Moderate. Pooled analysis; endpoints are behavior measures, mostly self-reported. PMID 36779907
Endpoint tested: Incident dementia and Alzheimer's disease. Population: 3,069 community volunteers aged 75 and over with normal cognition or mild cognitive impairment (GEM) High. Large randomized, double-blind, placebo-controlled trial, median 6.1 years. Null. PMID 19017911
Endpoint tested: Progression to probable Alzheimer's disease. Population: 2,854 adults aged 70 and over with spontaneous memory complaints (GuidAge) High. Large randomized, double-blind, placebo-controlled trial, 5 years. Null. PMID 22959217
Endpoint tested: Total glutathione and the reduced-to-oxidized glutathione ratio (GSH-F:GSSG), the primary endpoint. Population: 114 healthy older adults (mean age 65), randomized to GlyNAC 2.4, 4.8 or 7.2 g/day for 2 weeks Moderate. Randomized controlled trial, n=114, run by a group other than the originating investigators and sponsored by a company that holds patents on and markets glycine plus NAC products, so a maker-funded null. GlyNAC did not increase GSH-F:GSSG (p=0.739) or total glutathione (p=0.278). Safe and well tolerated. A post-hoc subset with high oxidative stress and low baseline glutathione did show an increase (p=0.016), which is hypothesis-generating, not the primary result. The trial's dose was 2.4 to 7.2 g/day total for 2 weeks, against the originating group's roughly 14 g/day for 16 weeks. PMID 35821844
Endpoint tested: Muscle strength and hypertrophy. Population: 44 resistance-trained men, HMB free acid or calcium HMB 3 g/day with training for 12 weeks Moderate. One placebo-controlled trial of both forms; no effect on either outcome. PMID 30859862
Endpoint tested: Invasive breast cancer, stroke, pulmonary embolism, dementia in women 65 and over (harms); coronary heart disease not significant at final adjudication; all-cause mortality unchanged. Population: 16,608 healthy postmenopausal women aged 50-79 with an intact uterus, combined estrogen plus progestin High. Large randomized placebo-controlled primary-prevention trial, STOPPED EARLY because overall risks exceeded benefits; harms confirmed in the 13-year follow-up (PMID 24084921); no mortality effect over 18 years (PMID 28898378). PMID 12117397
Why this grade, and what it does not cover
Grade D on Route 2: tested, harm found. The Women's Health Initiative randomized 16,608 healthy postmenopausal women aged 50 to 79 with an intact uterus to conjugated equine estrogen plus medroxyprogesterone acetate or placebo, to find out whether it prevented chronic disease. It was stopped early because overall risks exceeded benefits. The firm harms are invasive breast cancer (hazard ratio 1.24 in the final intervention-phase analysis, PMID 24084921), stroke, pulmonary embolism, dementia in women 65 and over, and gallbladder disease, alongside fewer hip fractures and less diabetes. The coronary heart disease excess in the 2002 interim report (hazard ratio 1.29) was only nominally significant; at final adjudication it was 1.18 (95% CI 0.95 to 1.45), and the 2026 Cochrane review rates the coronary effect as probably little to no difference (RR 1.17, 0.95 to 1.44). All-cause mortality was not changed during the trial or over 18 years of follow-up (hazard ratio 1.02, PMID 28898378); the D records harm on specific outcomes, not excess deaths. WHAT THIS GRADE DOES NOT COVER, AND THIS IS THE MOST OVER-READ RESULT IN MODERN MEDICINE: this letter grades one formulation, one route, one dose pair, given for chronic-disease prevention, in women averaging more than a decade past menopause. It does not grade hormone therapy for menopausal symptoms, a different question with different evidence that this site does not grade. It does not grade estrogen-alone therapy after hysterectomy, which produced a materially different risk profile within the same program. It does not grade transdermal routes, other progestogens, or initiation close to menopause. This letter is a statement about prevention trials, not about any individual prescription.
Endpoint tested: Gross motor function and secondary outcomes. Population: Children with cerebral palsy. Slightly pressurized air at 1.3 ATA was the SHAM A fake version of a procedure or device used as the comparison, so participants cannot tell which one they received./control arm; both arms improved equally Moderate. Randomized multicenter trial. 1.3 ATA performed no better than the condition it was designed to control for. PMID 11558483
Endpoint tested: Postprandial blood glucose handling. Population: Individuals with type 2 diabetes mellitus, single sauna session Low. Single cross-over randomized controlled trial. Null. PMID 39209309
Endpoint tested: Apnea-hypopnea index. Population: Adults with obstructive sleep apnea, 10 randomized trials, 166 trained and 157 controls Moderate. Mean difference 1.00 (95% credible interval -2.57 to 4.56); Chen et al. 2023 and Silva de Sousa et al. 2024 agree that the index did not improve. PMID 41721779
Endpoint tested: Aerobic performance, perceived exertion, VO2 kinetics, lactate. Population: Meta-analysis of 10 randomized trials of citrulline in exercise Moderate. Pooled null on every aerobic outcome. PMID 36079738
Endpoint tested: Time to exhaustion and time to completion. Population: Meta-analysis of 9 randomized trials in young healthy adults Moderate. Pooled null on both endurance outcomes. PMID 37155582
Endpoint tested: Prevalence of infectious disease over 12 weeks (primary). Population: 310 healthy adults, placebo, 200 or 600 mg/day Moderate. The maker's own larger trial; no difference in prevalence or immune markers. PMID 32151562
Endpoint tested: Symptomatic and asymptomatic SARS-CoV-2 infection over 90 days. Population: 209 health workers in two Peruvian hospitals, 600 mg/day Low to moderate. Independent, stopped early for vaccine availability; 10.6 against 8.6 percent infected. PMID 36474100
Endpoint tested: Respiratory tract infection incidence in adults, pooled. Population: Meta-analysis, adult trials Moderate. OR 1.00 (0.76 to 1.32) in adults; a reduction appeared only in infants and children. PMID 35481594
Endpoint tested: Gulf War illness symptom severity. Population: 21 male veterans, placebo-controlled pseudo-randomized crossover, lower and higher luteolin doses Low. Small screening study of nine botanicals; luteolin no different from placebo at either dose (p = 0.72 and 0.49) while resveratrol was. PMID 33802381
Endpoint tested: Proportion of delirium-free assessments over 14 days (CAM-ICU). Population: 847 critically ill adults in 12 Australian intensive care units, 4 mg enteral melatonin nightly High. The largest trial, multicenter, double-blind; 79.2 against 80.0 percent delirium-free assessments, no difference in any secondary outcome. PMID 35220473
Endpoint tested: Delirium incidence by day 14 (CAM-ICU). Population: 355 mechanically ventilated adults in 20 French centers, 0.3 or 3 mg nightly Moderate to high. Adaptive phase 2b/3, double-blind; risk ratio 0.99 against placebo, secondary outcomes also similar. PMID 40608082
Endpoint tested: Delirium incidence, pooled. Population: Six low-to-moderate risk-of-bias randomized trials, 2,209 critically ill patients (13 trials, 2,830 patients in the secondary analysis) High for no effect at the pooled level: RR 0.89 (0.73 to 1.09) and RR 0.86 (0.70 to 1.04); no effect on mortality. PMID 41051561
Endpoint tested: Muscle hypertrophy in response to progressive resistance training (primary outcome). Population: 94 healthy adults aged 65 and over without diabetes, metformin 1,700 mg/day or placebo, 14 weeks Moderate. Randomized, double-blind, placebo-controlled, multicenter. The hypothesis was that metformin would AUGMENT the muscle response. It did the opposite: the authors report metformin blunted the hypertrophic response to training. PMID 31557380
Endpoint tested: Mitochondrial adaptation to aerobic exercise training. Population: Older adults without diabetes, metformin or placebo during supervised aerobic training Moderate. Randomized controlled trial. Metformin inhibited the mitochondrial adaptations exercise normally produces, consistent with the earlier finding that it antagonizes exercise-induced gains in insulin sensitivity and cardiorespiratory fitness. PMID 30548390
Endpoint tested: Endurance exercise performance, pooled. Population: 8 placebo-controlled trials, 188 subjects, acute and chronic MitoQ Moderate. Pooled SMD -0.50 (95% CI -1.39 to 0.40); oxidative damage markers fell while performance did not improve. PMID 38981985
Endpoint tested: VO2peak, 20 km time trial and mitochondrial content after 3 weeks of interval training. Population: 23 untrained men aged about 44, 20 mg/day, parallel-group Low to moderate. No difference in the primary adaptations; peak power during the VO2peak test rose more on MitoQ, a secondary finding. PMID 35623315
Endpoint tested: Measured physical function (strength, power, motor function). Population: 18 high-functioning adults aged 60 to 79, randomized placebo-controlled crossover, 20 mg/day for 6 weeks Low to moderate. Small crossover; the authors report no convincing effect, with exploratory subgroup signals in participants aged 70 and over. PMID 41733122
Endpoint tested: Cerebral Performance Category 1 or 2 at 90 days. Population: 73 comatose adults after out-of-hospital cardiac arrest, multicenter double-blind placebo-controlled, stopped early Low to moderate. Primary outcome not significant (RR 0.72, 95% CI 0.46 to 1.13); the trial reached a quarter of its planned size, and secondary outcomes favored hydrogen, so a benefit is neither shown nor excluded. PMID 36969346
Endpoint tested: Change in the Movement Disorder Society Unified Parkinson's Disease Rating Scale total score at 16 weeks. Population: 20 people with Parkinson's disease, double-blind, placebo air, 15 analyzed Low. A pilot of 20 with 5 excluded for low adherence; no difference from placebo and no adverse events. Too small to exclude a modest effect. PMID 34319514
Endpoint tested: Spasm relief rate and effective rate. Population: 53 infants with epileptic spasms, double-blind, against medical air, all on standard treatment Low. One small trial; neither primary endpoint nor any secondary endpoint differed (spasm relief p = 0.705). Raised heart muscle enzymes, digestive upset and abnormal liver tests were each more frequent in the hydrogen group. PMID 41908285
Endpoint tested: Clinical worsening by day 14. Population: 675 primary care outpatients with mild to moderate COVID-19, phase 3, triple-blind, placebo-controlled, multicenter Moderate. A phase 3 trial of 675 people found clinical worsening in 46.1 percent on hydrogen water against 43.5 percent on placebo (hazard ratio 1.09, 90% CI 0.90 to 1.31); adverse events were similar in both arms. PMID 39124575
Endpoint tested: Incidence of stage III or IV cancer among 12 prespecified cancer types (primary endpoint). Population: NHS-Galleri: 142,250 adults aged 50 to 77 in England, randomized 1:1 to the test plus usual care or usual care alone, up to three annual screening rounds High for the primary endpoint. Incidence rate ratio 1.03 (95% CI 0.92 to 1.14, P = 0.63): no reduction. Stage IV alone, a key secondary endpoint, had a rate ratio of 0.86 (95% CI 0.74 to 1.00), a nonsignificant trend toward fewer stage IV cancers that cannot override the primary result. Trial funded by GRAIL, which sells the test. PMID 42776074
Endpoint tested: Blood lipids (total cholesterol, LDL-C Low-density lipoprotein cholesterol, the cholesterol fraction most closely tied to plaque in the arteries., triglycerides). Population: 47 adults with hyperlipidemia, randomized to nattokinase alone, nattokinase plus red yeast rice, or placebo Clear. The trial's own title states that the COMBINATION with red yeast rice, but NOT nattokinase alone, had potent effects on blood lipids. Nattokinase monotherapy did not lower lipids. Recorded here so the null result is visible rather than sitting unexplained in the citation list. The Grade B rests on blood pressure (Kim 2008, PMID 18971533), not on lipids. PMID 19786378
Endpoint tested: Composite of coronary death, myocardial infarction, ischemic stroke, acute coronary syndrome hospitalization or revascularization. Population: 3,414 adults with established cardiovascular disease on simvastatin, extended-release niacin 1,500 to 2,000 mg/day alone, mean 3 years High. Stopped early for lack of efficacy; hazard ratio 1.02 (95% CI 0.87 to 1.21) despite HDL raised and triglycerides lowered. Funded by NHLBI and Abbott, the product's maker. PMID 22085343
Endpoint tested: Major vascular events (nonfatal myocardial infarction, coronary death, stroke or arterial revascularization). Population: 25,673 adults with atherosclerotic vascular disease on statin-based therapy, 2 g extended-release niacin with 40 mg laropiprant, median 3.9 years High. Events 13.2 against 13.7 percent, not significant. Serious adverse events increased (new-onset diabetes HR 1.32, bleeding, infection, gastrointestinal), a harm signal that belongs to the niacin-laropiprant combination and cannot be assigned to niacin alone. PMID 25014686
Endpoint tested: Montreal Cognitive Assessment and neurocognitive battery. Population: 20 older adults with mild cognitive impairment, randomized placebo-controlled pilot, 1 g/day for 10 weeks Low to moderate. Blood NAD+ Nicotinamide adenine dinucleotide, a molecule cells use in energy metabolism. Its level falls with age in some tissues. 2.6-fold higher; cognitive and psychometric measures stable in both groups. Small pilot. PMID 37994989
Endpoint tested: Cognitive function (primary), total cerebral blood flow and blood pressure. Population: 42 completers with amnestic mild cognitive impairment, phase-II randomized double-blind placebo-controlled, 12 weeks Moderate. Blood NAD+ doubled; no improvement on the primary or secondary outcomes. Two independent null trials. PMID 42478598
Endpoint tested: MDS-UPDRS parts I to III total score (primary) and Non-Motor Symptoms Scale score. Population: 410 randomized, 393 analyzed, phase 3 double-blind placebo-controlled trial Moderate. One large phase 3 trial. The total rating score worsened by 2.72 points more on nicotinamide riboside than on placebo (95% CI 0.47 to 4.98; P = .02) and non-motor symptom burden by 4.22 points more (P = .009). Both differences are below the 5.2-point threshold for clinically important worsening on the primary scale, and serious adverse events were fewer on nicotinamide riboside (8.3 against 14.1 percent). PMID 42848371
Endpoint tested: Insulin sensitivity by hyperinsulinemic euglycemic clamp. Population: 40 obese insulin-resistant men, investigator-initiated double-blind placebo-controlled trial, 2,000 mg/day for 12 weeks Moderate. The gold-standard clamp measure and whole-body glucose metabolism did not improve; energy expenditure, lipolysis and body composition also unchanged. One trial. PMID 29992272
Endpoint tested: Five cognitive domains (objective tests). Population: 390 cognitively healthy adults aged 65 to 90, DHA 1,720 mg/day for 18 months Moderate. Large, long, placebo-controlled; null with a small adverse signal on psychomotor speed. PMID 29722833
Why this grade, and what it does not cover
Grade D, Route 2: tested in older adults and no benefit found. The trials at or near this protocol's 1 to 2 g/day dose were null (EPOCH at 1,720 mg/day for 18 months; MAPT at 800 mg/day for 3 years), and the DHA-specific meta-analysis was null on every domain. The only positive trial (MIDAS) used 900 mg/day, below the protocol's dose, was positive on one test of several, and was led by an employee of the maker of the DHA it tested. Moved from B in October 2026 by editor adjudication: the B rested on MIDAS alone, and the Cochrane review previously cited (PMID 32114706) studied cardiovascular disease, not cognition, and has been removed from this record. WHAT THIS GRADE DOES NOT COVER: DHA in infants and children, which is a different population; mild cognitive impairment, where small trials are mixed and the pooled analyses combine DHA with EPA; and EPA or mixed fish oil for cardiovascular outcomes, which is graded on the omega-3 record.
Endpoint tested: Composite cognitive score scored by blinded neuropsychologists. Population: 1,680 adults aged 70 and over with memory complaints, DHA 800 mg/day plus EPA 225 mg/day for 3 years Moderate to high. Large, 3-year, placebo-controlled; null on the primary endpoint. PMID 28359749
Why this grade, and what it does not cover
Grade D, Route 2: tested in older adults and no benefit found. The trials at or near this protocol's 1 to 2 g/day dose were null (EPOCH at 1,720 mg/day for 18 months; MAPT at 800 mg/day for 3 years), and the DHA-specific meta-analysis was null on every domain. The only positive trial (MIDAS) used 900 mg/day, below the protocol's dose, was positive on one test of several, and was led by an employee of the maker of the DHA it tested. Moved from B in October 2026 by editor adjudication: the B rested on MIDAS alone, and the Cochrane review previously cited (PMID 32114706) studied cardiovascular disease, not cognition, and has been removed from this record. WHAT THIS GRADE DOES NOT COVER: DHA in infants and children, which is a different population; mild cognitive impairment, where small trials are mixed and the pooled analyses combine DHA with EPA; and EPA or mixed fish oil for cardiovascular outcomes, which is graded on the omega-3 record.
Endpoint tested: Memory, attention, working memory, executive function. Population: Meta-analysis of 10 DHA studies, 2,327 participants with age-related cognitive decline Moderate. DHA-specific pooled analysis; null on every domain. PMID 32060571
Why this grade, and what it does not cover
Grade D, Route 2: tested in older adults and no benefit found. The trials at or near this protocol's 1 to 2 g/day dose were null (EPOCH at 1,720 mg/day for 18 months; MAPT at 800 mg/day for 3 years), and the DHA-specific meta-analysis was null on every domain. The only positive trial (MIDAS) used 900 mg/day, below the protocol's dose, was positive on one test of several, and was led by an employee of the maker of the DHA it tested. Moved from B in October 2026 by editor adjudication: the B rested on MIDAS alone, and the Cochrane review previously cited (PMID 32114706) studied cardiovascular disease, not cognition, and has been removed from this record. WHAT THIS GRADE DOES NOT COVER: DHA in infants and children, which is a different population; mild cognitive impairment, where small trials are mixed and the pooled analyses combine DHA with EPA; and EPA or mixed fish oil for cardiovascular outcomes, which is graded on the omega-3 record.
Endpoint tested: Serious vascular events. Population: 15,480 adults with diabetes and no cardiovascular disease, randomized to 1 g/day marine n-3 or placebo Clear and directly relevant. ASCEND tested the dose and formulation most people buy, on modern background therapy, and found nothing. It is the reason the post-myocardial-infarction use is graded B rather than A. PMID 30146932
Endpoint tested: Major adverse cardiovascular events. Population: 13,078 statin-treated adults at high cardiovascular risk with raised triglycerides, randomized against corn oil Relevant and it is the strongest published challenge to this record's Grade A, which is why it is named here rather than left off. It is not a replication failure of REDUCE-IT: STRENGTH tested omega-3 carboxylic acid combining EPA and DHA rather than purified EPA ethyl ester, against corn oil rather than mineral oil, and reached substantially lower plasma EPA. The endpoint is the same and the agent is not. PMID 33190147
Endpoint tested: Major cardiovascular events (composite). Population: 25,871 generally healthy US adults Large, well-powered randomized trial PMID 30415637
Endpoint tested: Markers of collagen metabolism. Population: 68 adults with hypertension and raised galectin-3, randomized double-blind placebo-controlled, NIH-funded with product donated by the maker Low to moderate. Single proof-of-concept trial; collagen markers, echocardiographic measures and vascular function unchanged. PMID 33532663
Endpoint tested: WOMAC knee, SF-36 and RAPID3 scores. Population: 50 patients with knee osteoarthritis, randomized 1 to 1 to pectin or placebo, 12 weeks Low. Small pilot; both groups improved and did not differ; serum galectin-3 unchanged. PMID 31872160
Endpoint tested: Delayed recall and recognition of a learned word list (primary); learning, reaction time, planning and attention (secondary). Population: 120 adults over 57 with age-associated memory impairment, soy phosphatidylserine 300 or 600 mg/day for 12 weeks Moderate. Three-arm placebo-controlled trial, two doses, no difference on any outcome at either dose. PMID 11842880
Endpoint tested: Neuropsychological memory test scores over 6 months. Population: 78 adults aged 50 to 69 with memory complaints, soy phosphatidylserine 100 or 300 mg/day for 6 months Low to moderate. All groups including placebo improved; the reported gain is within a low-baseline subgroup, not a between-group difference. PMID 21103034
Endpoint tested: Delayed recall and recognition of a learned word list (primary); learning, choice reaction time, planning and attention (secondary). Population: 120 adults over 57 with age-associated memory impairment, soy phosphatidylserine 300 or 600 mg/day for 12 weeks Moderate. Three-arm placebo-controlled trial with two doses; no difference on any outcome at either dose. Single trial of this size. PMID 11842880
Why this grade, and what it does not cover
Grade D on Route 2: tested, no benefit. Soy-derived phosphatidylserine was given to older adults with memory complaints in two randomized placebo-controlled trials, Jorissen et al. (Nutr Neurosci, 2001; PMID 11842880; 120 adults, 300 or 600 mg/day, 12 weeks) and Kato-Kataoka et al. (J Clin Biochem Nutr, 2010; PMID 21103034; 78 adults, 100 or 300 mg/day, 6 months), and in neither did the treated groups separate from placebo. WHAT THIS GRADE DOES NOT COVER: bovine-cortex phosphatidylserine, the product in the positive 1993 trial (Cenacchi et al., PMID 8323999), which is a different preparation no longer sold and is graded on the supplement record; phosphatidylserine combined with omega-3 fatty acids or with phosphatidic acid, which are combination products graded, where they are graded at all, as combinations; doses above 600 mg/day, which no trial of the soy form has tested; and any population other than older adults with memory complaints, including children with attention disorders and adults under stress, where small trials exist and are not part of this grade.
Endpoint tested: Neuropsychological memory test scores over 6 months. Population: 78 adults aged 50 to 69 with memory complaints, soy phosphatidylserine 100 or 300 mg/day for 6 months Low to moderate. All groups including placebo improved; the reported gain is a within-group change in a low-baseline subgroup, not a between-group difference. PMID 21103034
Why this grade, and what it does not cover
Grade D on Route 2: tested, no benefit. Soy-derived phosphatidylserine was given to older adults with memory complaints in two randomized placebo-controlled trials, Jorissen et al. (Nutr Neurosci, 2001; PMID 11842880; 120 adults, 300 or 600 mg/day, 12 weeks) and Kato-Kataoka et al. (J Clin Biochem Nutr, 2010; PMID 21103034; 78 adults, 100 or 300 mg/day, 6 months), and in neither did the treated groups separate from placebo. WHAT THIS GRADE DOES NOT COVER: bovine-cortex phosphatidylserine, the product in the positive 1993 trial (Cenacchi et al., PMID 8323999), which is a different preparation no longer sold and is graded on the supplement record; phosphatidylserine combined with omega-3 fatty acids or with phosphatidic acid, which are combination products graded, where they are graded at all, as combinations; doses above 600 mg/day, which no trial of the soy form has tested; and any population other than older adults with memory complaints, including children with attention disorders and adults under stress, where small trials exist and are not part of this grade.
Endpoint tested: Recovery of jump performance and maximal voluntary contraction. Population: Adults following exercise-induced muscle damage, 30-minute IPC application Low. Single randomized trial. No improvement in objective recovery kinetics; a possible reduction in perceived soreness is self-reported and unreplicated. PMID 40555415
Endpoint tested: Lipids: LDL cholesterol rose 17.1 mg/dL on pterostilbene alone; blood pressure fell as another endpoint. Population: 80 adults with high cholesterol, 6 to 8 weeks, four arms Low to moderate. Single small placebo-controlled trial; LDL moved in the harm direction in people enrolled for raised cholesterol; primary-endpoint status not stated in the abstract. PMID 25057276
Endpoint tested: Lipids: LDL cholesterol rose 17.1 mg/dL on pterostilbene alone; blood pressure fell as another endpoint. Population: 80 adults with high cholesterol, 6 to 8 weeks, four arms Low to moderate. Single small placebo-controlled trial; LDL moved in the harm direction in people enrolled for raised cholesterol; primary-endpoint status not stated in the abstract. PMID 25057276
Endpoint tested: LDL, HDL and triglycerides. Population: 5 randomized trials, 442 subjects Moderate. No significant change in LDL, HDL or triglycerides; borderline, clinically irrelevant rise in total cholesterol. Confirmed null in a 17-trial pool and the 2023 umbrella review. PMID 25897620
Endpoint tested: Fasting glucose, HOMA-IR and HbA1c. Population: 9 randomized trials in metabolic syndrome and related disorders Moderate. No overall change; glucose lower only in subgroups at 500 mg/day or more for 8 weeks or longer. Glucose and HOMA-IR also null in a 17-trial pool and the 2023 umbrella review. PMID 30848564
Endpoint tested: Visceral adiposity (primary outcome). Population: Healthy normative-aging adults, low-dose intermittent rapamycin, 48 weeks (PEARL) Moderate. Randomized, placebo-controlled, 48 weeks. The primary outcome did not change (p=0.942). Rapamycin was relatively safe. Lean tissue mass and self-reported pain improved in WOMEN at 10 mg, a subgroup finding the authors explicitly frame as requiring future work to establish efficacy. PMID 40188830
Endpoint tested: Chair-stand performance added to a home exercise program (primary outcome). Population: 40 sedentary adults aged 65-85, sirolimus 6 mg once weekly, 13 weeks (RAPA-EX-01) Moderate. Randomized, double-blind, placebo-controlled. The primary intention-to-treat Analyzing everyone in the group they were assigned to, even if they stopped the treatment, so the comparison stays randomized. analysis favored PLACEBO (-2.13 repetitions, p=0.089) and sensitivity analyses favored placebo and reached statistical significance. The authors conclude sirolimus "did not enhance, and in sensitivity analyses, it may have modestly attenuated" the gains from exercise, and that it "increased the burden of minor adverse events and may have contributed to one serious infection." PMID 41985884
Endpoint tested: Cognition (three instruments), handgrip strength and 40-foot walk time. Population: 25 generally healthy adults aged 70 to 95, randomized placebo-controlled pilot, 1 mg rapamycin daily, 8 weeks or more Low. Small pilot powered for feasibility and safety; no change in any cognitive or physical measure; hemoglobin, hematocrit and red cell indices fell on rapamycin. PMID 29408453
Endpoint tested: Mortality, falls, cardiovascular disease, cognition. Population: Cochrane review of deprescribing Reducing or stopping a medication under a prescriber's supervision. trials in older adults Very low. Few small trials; no between-group differences. PMID 38063254
Why this grade, and what it does not cover
Grade D, Route 2. Randomized deprescribing trials have been run, and the Cochrane review (PMID 38063254) and a meta-analysis of 7 trials (PMID 37740900) found no clear benefit on cognition, falls, quality of life, cardiovascular disease or mortality. The one trial that clearly lowered the burden (Kersten et al., 2013; PMID 22982689) found no cognitive improvement at 8 weeks; most other trials, including van der Meer 2018 (PMID 30030308), did not manage to lower the burden, so they tested whether an intervention changes prescribing more than whether a lower burden improves health. The 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. association between anticholinergic burden and harm carries no grade in either direction; people taking more of these medicines are sicker, so it cannot be read as cause. Moved from Route 1 No human randomized trial has been run, so the record rests on animal or mechanistic evidence only. One of the two routes to the lowest letter on EBL's scale. in October 2026; the letter is unchanged. Incident cardiovascular events, the outcome the observational study links to burden, have never been tested in a trial, so for that outcome alone the position is untested rather than tested without benefit. WHAT THIS GRADE DOES NOT COVER: whether a particular anticholinergic medicine is still needed for a particular person, which is a prescribing judgment for their clinician; deprescribing of other drug classes; and avoiding anticholinergic drugs around surgery, which is a different question.
Endpoint tested: Falls, cognition, burden score, quality of life. Population: Meta-analysis of 7 randomized trials Low. No significant effects. PMID 37740900
Why this grade, and what it does not cover
Grade D, Route 2. Randomized deprescribing trials have been run, and the Cochrane review (PMID 38063254) and a meta-analysis of 7 trials (PMID 37740900) found no clear benefit on cognition, falls, quality of life, cardiovascular disease or mortality. The one trial that clearly lowered the burden (Kersten et al., 2013; PMID 22982689) found no cognitive improvement at 8 weeks; most other trials, including van der Meer 2018 (PMID 30030308), did not manage to lower the burden, so they tested whether an intervention changes prescribing more than whether a lower burden improves health. The observational association between anticholinergic burden and harm carries no grade in either direction; people taking more of these medicines are sicker, so it cannot be read as cause. Moved from Route 1 in October 2026; the letter is unchanged. Incident cardiovascular events, the outcome the observational study links to burden, have never been tested in a trial, so for that outcome alone the position is untested rather than tested without benefit. WHAT THIS GRADE DOES NOT COVER: whether a particular anticholinergic medicine is still needed for a particular person, which is a prescribing judgment for their clinician; deprescribing of other drug classes; and avoiding anticholinergic drugs around surgery, which is a different question.
Endpoint tested: HbA1c, total and LDL cholesterol, body mass index. Population: Pooled trials in type 2 diabetes, 1.4 to 3 g/day for 12 to 16 weeks Low. Small trials, most at unclear risk of bias; no statistically or clinically significant change. PMID 25686270
Endpoint tested: HbA1c and fasting glucose (primary), lipids, blood pressure and other secondary outcomes. Population: 84 adults with type 2 diabetes and metabolic syndrome, double-blind placebo-controlled, 3 g/day for 16 weeks Moderate. No effect on any primary or secondary outcome; the largest placebo-controlled trial of the mushroom. PMID 27511742
Endpoint tested: Training gains in maximal oxygen uptake, mean arterial pressure and blood lipids. Population: 27 inactive men aged about 65, randomized to resveratrol or placebo alongside 8 weeks of high-intensity training Low to moderate. One trial; the placebo group's gain in maximal oxygen uptake was 45 percent larger, mean arterial pressure fell only on placebo, and resveratrol abolished the lipid improvements. PMID 23878368
Endpoint tested: Pain and function, knee or hip osteoarthritis. Population: 656 patients across 4 randomized trials versus placebo Low. Pain SMD -0.17, 95% CI -0.34 to 0.01; the review rates trial quality as poor. PMID 19821403
Endpoint tested: Depression rating scales (HamD17, MADRS, IDS-SR30) at 6 weeks. Population: 234 adults with major depressive disorder and inadequate response to an antidepressant, MSI-195 (SAM-e) 800 mg added Moderate. The largest adjunct A treatment tested on top of standard prescribed therapy rather than in place of it. trial; null on all three instruments. PMID 30368163
Endpoint tested: MADRS score at 8 weeks. Population: 107 non-remitting outpatients with major depressive disorder, SAM-e 800 mg/day added to an antidepressant Moderate. Double-blind, placebo-controlled; no between-group difference, high placebo response. PMID 30115553
Endpoint tested: Change in depressive symptoms from baseline to end of treatment, SAM-e as monotherapy versus placebo. Population: 934 adults with major depression across 8 randomized trials Very low quality evidence as rated by the review. SMD -0.54, 95% CI -1.54 to 0.46, P = 0.29, 142 participants in the placebo comparison. PMID 27727432
Endpoint tested: Depression symptom change, SAM-e as monotherapy and as an adjunct to antidepressants. Population: 1,522 subjects across 14 randomized trials, doses 200 to 3,200 mg/day Low. No significant difference from placebo in either comparison; the review notes the small number of studies and the wide dose range. PMID 38199136
Endpoint tested: Pain and function in knee or hip osteoarthritis. Population: 656 patients across 4 randomized trials versus placebo Low. Pain SMD -0.17, 95% CI -0.34 to 0.01, about 0.4 cm on a 10 cm scale; the review rates methodological quality and reporting as poor. PMID 19821403
Endpoint tested: Prostate cancer (primary) and other cancers. Population: 35,533 men aged 50 or 55 and over, 200 mcg/day L-selenomethionine, SELECT, median 5.5 years High. Hazard ratio 1.04 for prostate cancer, no other cancer endpoint reduced; nonsignificant rise in type 2 diabetes (RR 1.07). PMID 19066370
Endpoint tested: Any cancer incidence, pooled. Population: Cochrane 2018, 10 randomized trials, 27,232 participants, low-risk-of-bias pool of 19,475 High. RR 1.01 (95% CI 0.93 to 1.10); a slight increase in type 2 diabetes; no clear influence of baseline selenium status. PMID 29376219
Endpoint tested: Incidence of type 2 diabetes (secondary analysis). Population: 1,202 adults in the Nutritional Prevention of Cancer trial, low-selenium areas of the eastern United States, 200 mcg/day, mean 7.7 years Low to moderate. Diabetes more frequent on selenium (hazard ratio 1.55), concentrated in the highest tertile of baseline plasma selenium; a secondary outcome with mostly self-reported diagnoses, consistent with the diabetes signals in SELECT and Sel/Cel. PMID 17620655
Endpoint tested: Mnemonic discrimination performance at 12 months (primary); secondary memory and biomarker A measurable sign in the body, such as a blood test value, used to track a condition or the effect of a treatment. outcomes. Population: 100 adults aged 60 to 90 with subjective cognitive decline, spermidine-rich wheat germ extract for 12 months Moderate. Single phase 2b placebo-controlled trial, adequately sized for its endpoint; exploratory signals on verbal memory and inflammation unconfirmed. PMID 35616942
Endpoint tested: Time- and frequency-domain heart rate variability. Population: Participants in 6 sham-controlled randomized trials of transcutaneous vagus nerve stimulation, pooled Moderate by the review's own rating; protocols heterogeneous, trials small. PMID 41506123
Endpoint tested: Resistance artery endothelial function (forearm blood flow to acetylcholine); conduit artery function, arterial stiffness, oxidative stress and inflammation. Population: 32 healthy adults aged 50 to 77, 100 g/day oral trehalose against maltose for 12 weeks Low. One small trial; the full-sample analysis was not significant and the positive result is confined to a post hoc weight-stable subgroup. PMID 27208415
Endpoint tested: Arterial wall inflammation (FDG-PET target-to-background ratio) and carotid intima-media thickness. Population: 15 patients with prior myocardial infarction and systemic inflammation, 15 g intravenous trehalose weekly for 12 weeks Low. Very small trial of an intravenous form; supporting context, not the basis. PMID 35045351
Endpoint tested: All-cause mortality. Population: 21,315 Australians aged 60 and over, not screened for vitamin D status (D-Health) High. Large randomized, double-blind, placebo-controlled trial with mortality as the primary endpoint. Null. PMID 35026158
Endpoint tested: Invasive cancer of any type (incidence). Population: 25,871 generally healthy US adults not selected for vitamin D status; men 50+, women 55+ High. Large, randomized, placebo-controlled, median 5.3 years follow-up. Null. PMID 30415629
Endpoint tested: Major cardiovascular events (composite of MI, stroke, CV death). Population: 25,871 generally healthy US adults not selected for vitamin D status High. Large, randomized, placebo-controlled. Null. PMID 30415629
Endpoint tested: Incident cardiovascular disease and death. Population: 5,110 community-resident adults aged 50 to 84 in New Zealand, not selected for vitamin D status (ViDA) High. Randomized, double-blind, placebo-controlled. Null, including in participants deficient at baseline. PMID 28384800
Endpoint tested: Hip fracture and any fracture. Population: Cochrane review of vitamin D alone in older adults and postmenopausal women High. Large pooled randomized evidence. PMID 24729336
Endpoint tested: Any fracture. Population: Meta-analysis of 69 randomized trials, vitamin D alone High. RR 1.00 for any fracture. PMID 42161415
Endpoint tested: Incident total, non-vertebral, and hip fractures. Population: Generally healthy midlife and older adults not selected for vitamin D status (VITAL ancillary) High. Randomized, placebo-controlled. Null. PMID 35939577
Endpoint tested: Incident or recurrent depression or clinically relevant depressive symptoms; mood scores. Population: 18,353 US adults aged 50 and over, 16,657 with no depression history and 1,696 with a history but no recent treatment (VITAL-DEP) High. Large randomized, double-blind, placebo-controlled trial with depression as a primary endpoint. Null. PMID 32749491
Endpoint tested: Prostate cancer incidence. Population: Healthy men, 400 IU/day all rac-alpha-tocopheryl acetate, planned 7-12 year follow-up High. Large randomized placebo-controlled trial. Supplementation SIGNIFICANTLY INCREASED prostate cancer risk. PMID 21990298
Endpoint tested: Myocardial infarction, stroke and cardiovascular death (composite). Population: 9,541 adults aged 55 and over with vascular disease or diabetes, 400 IU/day natural vitamin E (HOPE) High. Large randomized, placebo-controlled factorial trial. Null. PMID 10639540
Endpoint tested: Major cardiovascular events (composite). Population: 14,641 US male physicians aged 50 and over, 400 IU every other day (Physicians' Health Study II) High. Large randomized, double-blind, placebo-controlled factorial trial. Null. PMID 18997197
Nobody has tested it
21 entries where the search for a human randomized trial was run, recorded, and returned nothing. The exact query is printed so anyone can run it again. An absence nobody can check is an assertion, not a finding.
Re-run October 2026: 88 records. One randomized trial assigned heavy drinkers to an intervention to drink less and reported mortality: the Malmo trial (PMID 6135365; follow-up PMID 2862882), now graded C on its own row. The October note missed it. The other records are drug or counseling trials for alcohol dependence, alcoholic liver disease treatment, observational analyses nested in trial cohorts, and unrelated uses of the word alcohol. No randomized trial of reduced drinking reports major cardiovascular events. Search run October 2026 on PubMed, returning 1 human randomized trial(s).
(alcohol[tiab] AND (reduction[tiab] OR abstinence[tiab] OR abstain*[tiab])) AND (mortality[tiab] OR "cardiovascular events"[tiab] OR "myocardial infarction"[tiab] OR stroke[tiab]) AND randomized controlled trial[pt]
No randomized human trial of standalone apigenin identified. The one randomized trial containing apigenin (Henderson 2024, PMID 38302501) tested a multi-ingredient branded product and is cited as such. Other citations are animal or mechanistic. Re-run October 2026: two records, both trials of multi-ingredient products containing apigenin (PMID 38302501 and a 2026 stress and sleep formula, PMID 42829807); neither tests apigenin alone. Search run October 2026 on PubMed, returning human randomized trial(s).
apigenin supplementation AND randomized controlled trial[pt] AND humans[mesh]
No randomized trial of rhamnan sulfate on plaque, endothelial or glycocalyx outcomes identified. One randomized trial exists for constipation (PMID 34226572), a different use, with no effect on blood lipids. The only human publication on vascular outcomes is an uncontrolled case series. Search run July 2026 on PubMed, returning human randomized trial(s).
rhamnan sulfate OR Arterosil AND randomized controlled trial[pt] AND (atherosclerosis OR endothelial OR glycocalyx)
No randomized trial of boron supplementation reports fracture, osteoporosis or osteopenia as an outcome. The two records returned are a magnetic-field study that mentions boron incidentally (PMID 40299942) and an open multi-nutrient trial (PMID 15658548); neither tests boron. Small randomized studies of boron on surrogate markers A lab value or measurement used in place of the outcome that matters, such as LDL cholesterol standing in for heart attacks. Moving a surrogate does not always move the outcome. exist and are listed on the record’s review status. Absence first confirmed July 2026; query made re-runnable October 2026. Search run October 2026 on PubMed, returning human randomized trial(s).
boron[tiab] AND randomized controlled trial[pt] AND (fracture*[tiab] OR osteoporosis[tiab] OR osteopenia[tiab])
No human randomized controlled trial of BPC-157 was identified. 211 PubMed records mention the compound; none carries a randomized-controlled-trial or clinical-trial publication type. A looser free-text variant returns three records - PMID 42123471, PMID 31266512 and PMID 30598581 - the first a review and the other two rat studies matched on the word randomized. None is a human trial, so none is a counter-example. Re-run October 2026: the developer group's own review (PMID 22300085) says a phase II trial in inflammatory bowel disease was run; no result is indexed, so the statement is now 'no result published' rather than 'no trial run'. Search run October 2026 on PubMed, returning human randomized trial(s).
(BPC-157[tiab] OR "BPC 157"[tiab]) AND (randomized controlled trial[pt] OR clinical trial[pt])
No human randomized trial of oral C60 identified. The randomized fullerene trials found are topical (for example PMID 41861269, radiation dermatitis), a different route and use. Citations for the graded claim are animal or mechanistic only. Re-run October 2026: 21 records; the randomized fullerene trials are topical (radiation dermatitis, PMID 41861269; skin, PMID 31141275 and 21137794). Search run July 2026 on PubMed, returning human randomized trial(s).
C60 OR fullerene AND randomized controlled trial[pt] AND humans[mesh]
No randomized trial of Ca-AKG reporting an aging or biological-age outcome identified. A 6 g/day bone-marker trial in osteopenia (PMID 17896582) is a different dose and endpoint, and the ABLE trial (PMID 40819772) has not reported results. Re-run October 2026: 15 records; the trials returned are the ABLE recruitment report (PMID 40819772), alpha-ketoglutaric acid with hydroxymethylfurfural at altitude (PMID 22893323) and ornithine alpha-ketoglutarate for pressure ulcers (PMID 19621198), none an aging outcome. Search run July 2026 on PubMed, returning human randomized trial(s).
alpha-ketoglutarate OR Rejuvant AND randomized controlled trial[pt] AND (aging OR biological age)
Re-run October 2026: no human randomized trial of fisetin as a senolytic A compound intended to clear senescent cells: older cells that have stopped dividing but remain in tissue. or for an aging outcome has reported results; registered senolytic trials have not published. Small trials in other uses exist and are cited on the record (PMID 33802381, PMID 29541713, PMID 42218768, PMID 41683255), each a different claim. Search run October 2026 on PubMed, returning human randomized trial(s).
fisetin AND (randomized controlled trial[pt] OR randomized) AND humans[mesh]
Re-run October 2026, one result: D'Adamo 2026 (PMID 42807157), a randomized trial of GG on plasma CoQ10 in healthy adults, null for GG alone; it does not test the statin-muscle claim. No human randomized trial of GG for statin-associated muscle symptoms was found, and no registered trial of that claim was located; the earlier 'ongoing trial' statement is withdrawn. Search run October 2026 on PubMed, returning human randomized trial(s).
geranylgeraniol AND randomized controlled trial[pt] AND humans[mesh]
No randomized trial of luteolin without palmitoylethanolamide reports an olfactory, cognitive or neuro-inflammatory outcome. The cited trial (PMID 35450527) is a fixed palmitoylethanolamide plus luteolin combination. The broader search for any randomized trial naming luteolin (14 records) found only combination and mixture products, plus one pseudo-randomized crossover of botanicals in Gulf War illness (PMID 33802381), which is a different claim. Search run October 2026 on PubMed, returning human randomized trial(s).
luteolin[Title] AND randomized controlled trial[pt] AND (olfactory[tiab] OR smell[tiab] OR anosmia[tiab] OR cognit*[tiab] OR neuroinflamm*[tiab]) NOT palmitoylethanolamide[tiab]
One record returned: a trial of magnesium glycinate combined with vitamin D (PMID 36640582) that reported glycemic markers; it does not test glycinate alone. No randomized trial of magnesium glycinate alone on insulin resistance was found. Trials of other magnesium salts are context, not evidence for this salt. Search run October 2026 on PubMed, returning human randomized trial(s).
(("magnesium glycinate"[tiab] OR "magnesium bisglycinate"[tiab] OR "magnesium diglycinate"[tiab])) AND (insulin[tiab] OR HOMA[tiab] OR glucose[tiab]) AND (randomized controlled trial[pt] OR randomized[tiab])
One record returned: a trial of magnesium glycinate combined with vitamin D (PMID 35576873) that reported blood pressure; it does not test glycinate alone. No randomized trial of magnesium glycinate alone on blood pressure was found. Trials of other magnesium salts are context, not evidence for this salt. Search run October 2026 on PubMed, returning human randomized trial(s).
(("magnesium glycinate"[tiab] OR "magnesium bisglycinate"[tiab] OR "magnesium diglycinate"[tiab])) AND ("blood pressure"[tiab] OR hypertens*[tiab]) AND (randomized controlled trial[pt] OR randomized[tiab])
Re-run October 2026: 18 records, no randomized trial of oral NAD+ itself on a health outcome. The returned trials study the precursors NMN and NR, nicotinamide, or NADH combined with coenzyme Q10 in chronic fatigue syndrome; none gives NAD+ itself. The one randomized trial of oral NAD+ (Kornilov 2026, PMID 42530810), found through the review of October, measured blood NAD over 5 days in a proprietary formulation and is not an outcome trial. A 2026 systematic review (PMID 41655607) likewise found no outcome trial of NAD+ itself by any route. Search run October 2026 on PubMed, returning human randomized trial(s).
oral nicotinamide adenine dinucleotide (NAD+) supplementation AND humans[mesh] AND (randomized controlled trial[pt] OR clinical trial[pt])
Re-run October 2026: no randomized trial of modified citrus pectin in prostate cancer is indexed. Two placebo-controlled trials in other uses exist and were null (PMID 33532663, hypertension markers; PMID 31872160, knee osteoarthritis), now graded D Route 2 on their own rows; PubMed does not tag them as randomized, which is why this query does not return them. The prostate studies are single-arm Phase II designs. Search run October 2026 on PubMed, returning human randomized trial(s).
("modified citrus pectin"[tiab] OR PectaSol[tiab]) AND randomized controlled trial[pt]
Re-run October 2026: no human randomized or clinical trial of piperlongumine (piplartine) identified. Records tagged as trials are cell culture studies mis-indexed. The one citation is cell work (PMID 27913811). Search run October 2026 on PubMed, returning human randomized trial(s).
piperlongumine AND randomized controlled trial[pt] AND humans[mesh]
Re-run October 2026: no human randomized or clinical trial of procyanidin C1 identified, and no senolytic trial of grape seed extract. The citations are mouse or cell work (PMID 34873338, PMID 38657052). Search run October 2026 on PubMed, returning human randomized trial(s).
procyanidin C1 OR PCC1 AND randomized controlled trial[pt] AND humans[mesh]
Re-run October 2026 with the same result. No randomized trial of sermorelin itself reports a longevity, body-composition or cognitive benefit; the only randomized exposure to the molecule (Corpas 1992, PMID 1379256) randomized dose order in 10 men and measured hormones. A broader search on GHRH(1-29) returns four randomized trials, of which three used a different molecule ([Nle27]GHRH(1-29)NH2, or the secretagogue L-692,429) and the fourth (PMID 1379256) is the growth-hormone and IGF-1 surrogate trial already cited here. The absence is of outcome evidence for sermorelin, not of research on the hormone axis. Search run October 2026 on PubMed, returning human randomized trial(s).
sermorelin[tiab] AND randomized controlled trial[pt]
No randomized trial has tested serrapeptase for a systemic, anti-inflammatory or longevity outcome. Eleven randomized trials of the enzyme do exist and all of them address narrow, short-term surgical and ENT indications; the trial cited on this record (PMID 33653320) is one of them, on swelling after dental surgery. The absence is of evidence for the graded claim, not of research on the molecule. Search run August 2026 on PubMed, returning human randomized trial(s).
(serrapeptase[tiab] OR serratiopeptidase[tiab]) AND randomized controlled trial[pt] AND (longevity[tiab] OR aging[tiab] OR mortality[tiab] OR "cardiovascular"[tiab] OR atherosclerosis[tiab] OR "chronic inflammation"[tiab])
No randomized trial of spermidine reports mortality, lifespan, longevity or healthspan. The three records returned are a pharmacokinetic study of circulating polyamines at 40 mg/day (PMID 39405978), a fasting study in worms, and a topical DFMO trial; none tests the graded claim. The earlier stored query (July 2026) was unscoped and returned the cognition trials, which are now graded as their own use. Search run October 2026 on PubMed, returning human randomized trial(s).
spermidine[tiab] AND randomized controlled trial[pt] AND (mortality[tiab] OR lifespan[tiab] OR longevity[tiab] OR healthspan[tiab])
No randomized trial of the Vascanox HP formula identified on re-run. The sole human publication (PMID 37849591) is open-label Everyone in the trial knows which treatment they are receiving, so expectations can shape the results, especially for symptoms people report themselves. with no control arm. A placebo-controlled crossover of a different nitric oxide lozenge exists and does not test this formula. Search run October 2026 on PubMed, returning human randomized trial(s).
Vascanox OR Noxa24 OR (S-allylcysteine AND nitrate) AND randomized controlled trial[pt] AND (blood pressure OR hypertension)
No randomized trial has tested whether whole-body MRI screening of asymptomatic people reduces mortality from cancer or anything else. The query returns two records carrying the randomized-controlled-trial publication type: a psoriatic arthritis imaging protocol and a 2009 comparison of whole-body MRI with bone scintigraphy for metastases; neither is an outcome trial of population screening. Every pooled analysis found (PMID 30932247; PMID 32393345; PMID 40884613; PMID 42823587) pools single-arm yield studies, and the 2026 Cochrane protocol (PMID 42639816) is only now setting out to search for trials. Search run October 2026 on PubMed, returning human randomized trial(s).
("whole body MRI"[tiab] OR "whole-body magnetic resonance"[tiab]) AND screening[tiab] AND randomized controlled trial[pt]
Everything holding a Grade D
39 entities. A D means the evidence for the named endpoint is absent, contradicted, or too weak to support the claim being made for it.
| Grade | Intervention | Endpoint |
|---|---|---|
| D | 2-HOBA (2-Hydroxybenzylamine) Supplement | Atrial fibrillation recurrence after ablation |
| D | Apigenin Supplement | No human randomized trial of standalone apigenin has been run. Route 1: cell, animal and mechanistic work only. Proposed as a CD38 inhibitor that preserves NAD+ by blocking its main degrading enzyme, and taken alongside NMN in some longevity protocols on that reasoning. The reasoning is not the evidence: a PubMed search for randomized human trials of apigenin on its own returned none; the one randomized trial that contains apigenin tested a multi-ingredient product. |
| D | Arterosil HP (Rhamnan Sulfate) Supplement | No randomized trial has tested rhamnan sulfate on plaque or vascular outcomes; one small randomized trial for constipation found no effect on blood lipids. Route 1: one uncontrolled case series plus preclinical data. |
| D | Aspirin (low-dose, primary prevention in healthy older adults) Protocol | Disability-free survival in healthy adults aged 70 and over |
| D | Beta-Carotene (supplemental) Supplement | Tested at scale for cancer prevention in six large trials. Beta-carotene alone raised lung cancer in male smokers (ATBC); beta-carotene with retinyl palmitate raised lung cancer and death in smokers and asbestos workers (CARET, stopped early); in mostly non-smoking physicians and women it did nothing. A harm finding in smokers, a null in everyone else. |
| D | Boron Supplement | Bone and steroid-hormone markers |
| D | BPC-157 (Body Protection Compound 157) Supplement | Tendon, ligament and gut tissue healing. No result of a human randomized controlled trial of the compound has been published; the developer group states a phase II trial in inflammatory bowel disease was run, and no result of it is indexed. |
| D | Bromelain Supplement | Osteoarthritis pain and inflammation markers, bromelain alone |
| D | C60 (Carbon-60 / Buckminsterfullerene) Supplement | No human trial of oral C60 has been run; the only human randomized trials of fullerene are topical. Route 1: animal and mechanistic data only, and the animal studies disagree with each other. C60 in olive oil extended lifespan in a small 2012 rat study that has not been reproduced; a 2021 mouse study found C60-in-oil mice outlived oil-fed controls but not water-fed controls, pointing at the vehicle; another 2021 mouse experiment reported light-dependent toxicity and no lifespan extension. Sold as an antioxidant with no established human use case. |
| D | Calcium Alpha-Ketoglutarate (Ca-AKG) Supplement | No randomized trial of Ca-AKG on an aging outcome has reported. A placebo-controlled trial of 1 g Ca-AKG on a DNA methylation clock (ABLE) has enrolled and not yet reported; an older trial of 6 g AKG calcium in women with osteopenia measured a bone-turnover marker. Route 1 Grade D: animal lifespan data plus one retrospective, uncontrolled report of customers of a multi-ingredient product, measured by a commercial DNA methylation test and authored from the company that sells it. |
| D | EEG Neurofeedback Device or modality | ADHD symptoms in children and adolescents (self-report, blinding-sensitive) |
| D | Fisetin Supplement | No randomized trial of fisetin as a senolytic for aging has reported results. Small randomized trials exist in other uses, including a null placebo-controlled crossover in Gulf War illness. Route 1: animal senolytic data only. |
| D | Folic Acid (supplemental, 1 mg/day) Supplement | Tested for colorectal adenoma recurrence in three placebo-controlled trials and a pooled analysis of 2,632 patients. No reduction. The largest trial raised the possibility of more advanced neoplasia, and its long-term follow-up found more sessile serrated lesions with extended use; no overall cancer increase was found across 13 trials. |
| D | General health checks (multi-organ screening of asymptomatic adults) Protocol | All-cause, cardiovascular and cancer mortality in adults unselected for disease |
| D | Genetic testing to choose diet or supplements (nutrigenomic panels) Device or modality | Diet quality and health measures with advice tailored to genotype, compared with personalized advice that does not use genotype |
| D | Geranylgeraniol (GG) Supplement | No human randomized trial of the statin-muscle claim has reported, and no registered trial of that claim was located in this pass. Route 1 for the statin-muscle use: mechanistic and animal work. GG has now been tested in humans on other endpoints: a 150-person randomized trial found GG alone had no effect on plasma CoQ10 or on health and vitality questionnaires in healthy adults, and a 34-person fixed-order placebo-then-GG study found no change in testosterone overall. Popularity is not evidence: this is taken as a statin adjunct on the strength of a mechanism, not a trial result. |
| D | Hormone replacement therapy (combined estrogen + progestin, primary prevention) Protocol | Chronic disease in healthy postmenopausal women, including breast cancer, stroke and blood clots |
| D | Luteolin Supplement | Olfactory test performance after COVID-19; tested only inside a palmitoylethanolamide combination |
| D | Magnesium Glycinate 300–400 mg Protocol | Insulin resistance |
| D | Magnesium Glycinate 300–400 mg elemental Protocol | Blood pressure |
| D | Metformin (for longevity) Longevity Rx | Muscle and mitochondrial gains from exercise training in healthy older adults |
| D | Multi-cancer early detection blood test (cell-free DNA) Device or modality | Incidence of late-stage (III and IV) cancer at diagnosis, a stage-shift surrogate for cancer mortality |
| D | NAD+ (Direct Oral Supplementation) Supplement | No randomized trial has tested oral NAD+ itself on a clinical, functional or validated-scale outcome, and a 2026 systematic review found none for oral or intravenous NAD+. The one randomized trial of oral NAD+ (Kornilov 2026) tested a proprietary formulation for 5 days with blood NAD as its primary endpoint; its authors are linked to the formulation's developer. Route 1. The trials commonly cited for NAD+ supplementation are trials of its precursors, NMN and NR, which are different molecules. |
| D | Niacin (extended-release, for lipids) Supplement | Tested at scale on top of statins. Extended-release niacin alone (AIM-HIGH) and extended-release niacin with laropiprant (HPS2-THRIVE) did not reduce vascular events; the laropiprant combination increased serious adverse events. |
| D | Omega-3 DHA 1–2 g/day Protocol | Cognitive function in older adults |
| D | PectaSol Modified Citrus Pectin (MCP) Supplement | PSA kinetics in biochemically relapsed prostate cancer, from single-arm studies only |
| D | Phosphatidylserine 300 mg/day (soy-derived) Protocol | Memory in older adults with memory complaints |
| D | Piperlongumine Supplement | No human trial has been run, and no human dose or safety data exist. Route 1: preclinical senolytic work only. In cultured human fibroblasts, piperlongumine killed senescent cells by apoptosis, and the cited paper reports that this did not require the induction of reactive oxygen species. The authors filed a patent on piperlongumine and its analogs as anti-aging agents, and two are tied to a senolytic company. |
| D | Procyanidin C1 (PCC1) Supplement | No human trial has been published. Route 1: mouse and cell work only. The founding mouse study, now under an Expression of Concern from its journal, reported that PCC1 killed senescent cells, possibly by promoting reactive oxygen species and mitochondrial dysfunction; later mouse studies from other groups, for example in the aged retina, report senolytic effects. Nothing in humans. |
| D | Pterostilbene Supplement | LDL cholesterol in adults with raised cholesterol |
| D | Rapamycin (Sirolimus) Longevity Rx | Visceral adiposity and physical function (chair-stand repetitions) in older adults |
| D | Reducing anticholinergic drug burden Protocol | Cognition, falls, quality of life and mortality |
| D | Sermorelin (GHRH 1-29) Supplement | One placebo-controlled trial of GHRH(1-29), in healthy older adults (PMID 16399214), is under review. Growth hormone and IGF-1 rose in two tiny studies from one research group, one randomizing only the order of two doses and the other uncontrolled; a hormone level shows the drug acted, not that health changed. Body composition, frailty and mortality have no randomized trial of sermorelin itself. All uses D, Route 1. |
| D | Serrapeptase (Serratiopeptidase) Supplement | Postoperative swelling (measured) and pain (self-reported) |
| D | Spermidine Supplement | No randomized trial of spermidine on lifespan or mortality has been run. A 12-month randomized trial in older adults with subjective cognitive decline (SmartAge) found no effect on memory. Route 1: animal lifespan data plus observational epidemiology. |
| D | Trehalose Supplement | Proposed autophagy The process cells use to break down and recycle their own worn-out parts. inducer, untested for that claim in people. The one oral trial in healthy middle-aged and older adults (100 g/day for 12 weeks) found no significant change in its primary vascular measure in the full sample, and an intravenous trial in heart attack survivors found no change in arterial inflammation. |
| D | Vascanox HP (Nitric Oxide / Noxa24) Supplement | No randomized trial has been run. Route 1: one open-label, uncontrolled trial of the branded formula. |
| D | Vitamin E (supplemental alpha-tocopherol) Supplement | Tested at scale for cancer prevention with no benefit in any large trial. In SELECT, 400 IU/day of synthetic vitamin E significantly increased prostate cancer in healthy men on extended follow-up; a second large trial in healthy men at 400 IU every other day found no prostate effect either way. |
| D | Whole-body MRI screening (asymptomatic adults) Device or modality | Cancer mortality or all-cause mortality in asymptomatic screened adults |
Every letter and every null result here is copied from the record that owns it. The full graded index →