What changed. Why. And the study that triggered it.

The record listed antioxidant activity, eye health, skin photoprotection, cardiovascular support and exercise recovery under one B. A PubMed search and a reading of every cited paper found the B holds only for skin moisture and elasticity (Zhou et al., Nutrients, 2021; PMID 34578794). Each other use is now graded on its own evidence and is C. The descriptions of Tominaga et al. (PMID 22428137) and Park et al. (PMID 20205737) were also corrected: Tominaga's moisture result was a trend, and Park found some inflammatory markers rose at 8 mg.

The record said Rabade et al. (Inflammopharmacology, 2024; PMID 38581640) and Gregori et al. (JAMA, 2018; PMID 30575881) found small benefits from chondroitin, often with glucosamine. Gregori found no long-term pain benefit for chondroitin, and Rabade found the combination with glucosamine improved neither pain nor function. A PubMed search for every randomized trial and pooled analysis found the B rests on one preparation: pharmaceutical-grade chondroitin sulfate at 800 mg a day (Reginster et al., Ann Rheum Dis, 2017; PMID 28533290; Honvo et al., Adv Ther, 2019; PMID 30879253). Chondroitin as generally sold, and the combination with glucosamine, are graded C.

The B leaned on an 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. mortality meta-analysis (Cappuccio et al., Sleep, 2010; PMID 20469800), and EBL never grades on observational data. Without it, the graded evidence is sleep extension trials 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.. A PubMed search found them mixed: Zhu et al. (Eur J Cardiovasc Nurs, 2022; PMID 34279625) found no overall change in blood pressure, glucose or insulin resistance, and Beals et al. (Diabetes Care, 2026; PMID 41564347) found no change in clamp-measured insulin sensitivity. Tasali et al. (JAMA Intern Med, 2022; PMID 35129580) found lower energy intake in one trial. Mixed or unreplicated results are the written C definition. The descriptions of Gonzales 2022 and Hartescu 2022 were also corrected.

The B rested on one meta-analysis described as consistent. A PubMed search for every randomized trial and pooled analysis of whole-body vibration on bone density found the two largest independent trials null: Slatkovska et al. (Ann Intern Med, 2011; PMID 22084333), 202 postmenopausal women for 12 months with blinded assessors, and Kiel et al. (J Bone Miner Res, 2015; PMID 25581217), 174 adults with a mean age of 82 for 24 months against a placebo platform. Five meta-analyses disagree site by site. Mixed results with failed large trials meet the written C definition. EBL is paid a commission on a vibration plate listed under this therapy; this change was made through the logged review that payment requires, and it lowers the letter.

The zinc record rested on one citation, Singh and Das (Cochrane Database of Systematic Reviews A review that searches for every study on a question using a method set in advance, then assesses them together., 2013; PMID 23775705), which Cochrane withdrew in 2015. It is replaced by Science et al. (CMAJ, 2012; PMID 22566526), Hemilä et al. (Br J Clin Pharmacol, 2016; PMID 27378206), Hemilä and Chalker (BMC Fam Pract, 2015; PMID 25888289) and Hemilä et al. (BMJ Open, 2020; PMID 31980506). They agree that lozenges supplying about 75 to 92 mg of zinc a day, started early, shorten colds in adults by roughly two and a half days, that a 13 mg lozenge did not, and that children showed no significant effect. The B now names that use, form and dose; children are graded C; deficiency correction is no longer part of the graded use.

The record rested on one pooled analysis and said fenugreek lowered fasting glucose and HbA1c. That analysis, Shabil et al. (Medicina (Kaunas), 2023; PMID 36837450), found HbA1c lower but no change in fasting glucose. A PubMed search for every pooled analysis of fenugreek on blood glucose found three more that agree on both markers: Kim et al. (Int J Mol Sci, 2023; PMID 37762302), 10 randomized trials in type 2 diabetes or prediabetes; Fakhr et al. (Pharmacol Res, 2023; PMID 36470549), 29 randomized trials; and Neelakantan et al. (Nutr J, 2014; PMID 24438170), which also found the effect confined to medium or high doses in people with diabetes and most trials of low quality. Replicated pooled evidence on glycemic surrogates is the written B definition, so the letter is unchanged; the record now states what each analysis found and the search that found them.

The record said both cited meta-analyses found myo-inositol improved insulin resistance in PCOS. Greff et al. (Reproductive Biology and Endocrinology, 2023; PMID 36703143) did, against placebo, across 26 trials. Jethaliya et al. (Reprod Sci, 2022; PMID 35477841) found no significant difference in fasting insulin, fasting glucose or HOMA, but its comparators were folic acid, metformin and oral contraceptives, not placebo. The record now says so. The B rests on the placebo-controlled pooled evidence and is unchanged.

The statin record listed vitamin D3 as a mild depletion while its own evidence said the opposite: across three randomized trials, statins raised serum vitamin D by 2.71 ng/mL (Mazidi et al., Eur J Clin Invest, 2017; PMID 27859044), and only observational designs showed a fall. The check report therefore told statin users their vitamin D could fall, which the trials do not support. The row is removed and the statin record now states why vitamin D is not listed, with the same citation.

Three depletion rows carried the label critical over evidence with no human trial behind it: statins and geranylgeraniol (Osaki et al., 2026; PMID 42323079, a mouse knockout), corticosteroids and magnesium (Rude, Magnes Res, 1999; PMID 10612083, a rat study), and insulin and magnesium (Paolisso et al., Am J Clin Nutr, 1992; PMID 1595589, a supplementation trial, not a depletion finding). The strongest warning on a page may not rest on its weakest evidence, so each now reads moderate, with its evidence label unchanged. A build check now stops any future row that pairs critical with mechanistic evidence.

The October upgrade to A counted Kayumov et al. (Psychosom Med, 2001; PMID 11211063) as the replication of Sletten et al. (PLoS Med, 2018; PMID 29912983). The Kayumov trial enrolled 22 people, and the rule this log publishes requires more than 30 participants before a trial can move a grade. Without it the use rests on one qualifying trial of 116 people, supported by smaller trials that agree (Dahlitz et al., 1991; PMID 1674014; Mundey et al., 2005; PMID 16295212) and a two-trial pooled estimate (Buscemi et al., 2005; PMID 16423108). That is the written B definition. The use returns to B and the headline follows. Applied the published grade rules under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

A new graded protocol covers structured exercise programs after treatment for early-stage cancer, graded on disease-free and overall survival. CHALLENGE (Courneya et al., N Engl J Med, 2025; PMID 40450658) randomized 889 people after colon cancer chemotherapy and found 5-year disease-free survival of 80.3 against 73.9 percent. A 2026 meta-analysis of 21 randomized trials (Tzang et al., Br J Sports Med; PMID 42744658) found the same direction across cancer types. Graded B rather than A because the pooled trials are mostly in breast cancer and in colon cancer the result rests on one trial. CHALLENGE was also added to the landmark trials library.

Antibody lowering replicates across meta-analyses (Huwiler et al., Thyroid, 2024; PMID 38243784) and the CATALYST trial (Larsen et al., Eur Thyroid J, 2024; PMID 38215286), but thyroid peroxidase antibodies are not an established stand-in for a clinical outcome, and quality of life, CATALYST's primary outcome, improved equally on placebo. The editor, who had signed the B in October, returned it to C the next day on the strict reading of the B definition (Arnold Sandlow, D.C.).

The use rests on one randomized trial of 63 people (Hadanny et al., Aging, 2020; PMID 32589613), and a search found no independent replication. The C definition covers a single positive trial nobody has independently repeated. Applied the written grade definition in October 2026.

The heart failure use rests on one trial, WAON-CHF (Tei et al., Circ J, 2016; PMID 27001189), whose primary endpoint was null. The vascular use rests on a small marker trial (Masuda et al., 2004; PMID 15090706) against a null crossover (Hussain et al., 2022; PMID 34954348); a coronary disease trial was also null on vascular measures (Debray et al., 2023; PMID 37650138). A single unrepeated trial is C, and an oxidative-stress marker is not an established stand-in. Applied the written grade definition: the B definition admits a biomarker A measurable sign in the body, such as a blood test value, used to track a condition or the effect of a treatment. only when it is an established stand-in for a clinical outcome, such as HbA1c, LDL, blood pressure or bone density, the strict reading the editor confirmed in October 2026.

Homocysteine lowering by betaine replicates (Olthof et al., J Nutr, 2003; PMID 14652361; Ashtary-Larky et al., 2022; PMID 33764214), but homocysteine is not an established stand-in for a clinical outcome. Applied the written grade definition: the B definition admits a biomarker only when it is an established stand-in for a clinical outcome, such as HbA1c, LDL, blood pressure or bone density, the strict reading the editor confirmed in October 2026.

Positive trials (Knapen et al., Thromb Haemost, 2015; PMID 25694037; Vossen et al., JAMA Cardiol, 2026; PMID 42268593) stand beside null ones (Witham et al., J Am Soc Nephrol, 2020; PMID 32817311; Fulton et al., J Nutr Health Aging, 2016; PMID 26892582; Bartstra et al., 2021; PMID 33068157), and arterial stiffness and coronary calcium are not established stand-ins for a clinical outcome. Applied the written grade definition: the B definition admits a biomarker only when it is an established stand-in for a clinical outcome, such as HbA1c, LDL, blood pressure or bone density, the strict reading the editor confirmed in October 2026.

Three randomized trials agree on the markers (Moludi et al., Nutr Metab (Lond), 2022; PMID 35870947; Maleki et al., Diabetol Metab Syndr, 2020; PMID 32015761; Li et al., iScience, 2025; PMID 40546935), but adhesion molecules, hs-CRP High-sensitivity C-reactive protein, a blood marker of low-grade inflammation., TNF-alpha and vascular function are not established stand-ins for a clinical outcome. Applied the written grade definition: the B definition admits a biomarker only when it is an established stand-in for a clinical outcome, such as HbA1c, LDL, blood pressure or bone density, the strict reading the editor confirmed in October 2026.

Tamoxifen was listed under the aromatase inhibitor record, so a reader who entered it was shown that record's bone-loss row. The trial the row cites (Eastell et al., J Clin Oncol, 2008; PMID 18309940, the ATAC bone substudy) found continued bone loss on anastrozole and not on tamoxifen. Tamoxifen is now its own record with no depletion recorded, and the DIM interaction on record for women taking tamoxifen still appears when both are entered in the check.

The B row was labeled for institutionalized or deficient older adults. The positive trial (Chapuy et al., N Engl J Med, 1992; PMID 1331788) and the institutional benefit in the Cochrane review (Avenell et al., 2014; PMID 24729336) are in people living in institutions, while the two largest community trials were null (Jackson et al., N Engl J Med, 2006; PMID 16481635; Grant et al., Lancet, 2005; PMID 15885294), and vitamin D3 alone in deficient or older adults is graded D on its own row. The label now names the population the trials support. The letter is unchanged.

Earlier in October the use was raised to B on Rossman et al. (Hypertension, 2018; PMID 29661838) and Kirkman et al. (Am J Physiol Renal Physiol, 2023; PMID 37560769). On review the editor returned it to C. The B definition admits a biomarker only when it is an established stand-in for a clinical outcome; flow-mediated dilation predicts events in cohorts (Ras et al., Int J Cardiol, 2013; PMID 23041097) but no trial has shown that raising it prevents one. The second trial is an 18-patient pilot whose groups started apart. Editor decision, October 2026 (Arnold Sandlow, D.C.).

The protocol carried one A for both the training and the weekly dose in its name. The editor split it in October 2026. Measured VO2max after low-to-moderate intensity training keeps A: Gormley et al. (Med Sci Sports Exerc, 2008; PMID 18580415), the DREW trial (Church et al., JAMA, 2007; PMID 17507344) and a 50-trial pooling (Nuuttila et al., 2026; PMID 41543030). The 150-minute weekly dose is C on its own row: only DREW's 136-minute arm comes near it, no other group has repeated it, and DREW found a steady dose response with no threshold (Arnold Sandlow, D.C.).

Zheng et al. (Lancet Respir Med, 2014; PMID 24621680), PANTHEON, reduced exacerbations at 600 mg twice daily in moderate to severe COPD. Tse et al. (Chest, 2013; PMID 23348146), HIACE, found the same as a secondary endpoint An outcome a trial also measures but did not choose as its main test. in 120 patients, with a manufacturer's employee among the authors. Zhou et al. (Nat Commun, 2024; PMID 39349461), from the PANTHEON group and without commercial funding, found no reduction at the same dose in mild to moderate disease, a population that overlaps the moderate band. The editor kept the use at C in October 2026: the replication is not strong enough to carry the letter higher. No letter changed (Arnold Sandlow, D.C.).

Rossman et al. (Hypertension, 2018; PMID 29661838) and Kirkman et al. (Am J Physiol Renal Physiol, 2023; PMID 37560769), independent groups at 20 mg/day, both found flow-mediated dilation improved against placebo. The use was held at C because the populations differed and the marker is not a clinical outcome. The editor decided that a replicated randomized effect on a measured physiological marker carries B, and that the B definition does not require the replicating trials to share a population. The letter is capped at B because no trial measured a cardiovascular event. Editor decision, October 2026 (Arnold Sandlow, D.C.).

Taverniti et al. (Microorganisms, 2026; PMID 42354894) randomized 140 healthy adults to Lacticaseibacillus rhamnosus CRL 1505 at 1 billion CFU a day or placebo for 12 weeks. Fewer people on the probiotic had three or more respiratory infections by week 16, while the one-or-more and two-or-more thresholds of the same primary endpoint were not significant. A single positive trial nobody has independently repeated is C under the written definition, and the authors include the strain's owner and licensee. The use is graded for this strain only; the headline stays C. Applied the written grade definition at the editor's request of October 2026 (Arnold Sandlow, D.C.).

The headline sat on the harm finding (Gliemann et al., J Physiol, 2013; PMID 23878368) while the record's best use, metabolic markers in obese men (Timmers et al., Cell Metab, 2011; PMID 22055504, against null replications), is C. The headline now shows C labeled with that use, and the harm finding stays D on Route 2 on its own row, listed directly beneath. No evidence changed. Headline policy adopted by the editor in October 2026 (Arnold Sandlow, D.C.): the headline letter is the record's best graded use, labeled with that use, and every other use, including any harm finding, is listed beneath it.

The headline showed C on longevity while the record's best use, seizure frequency in children with drug-resistant epilepsy (Neal et al., Lancet Neurol, 2008; PMID 18456557; Lambrechts et al., Acta Neurol Scand, 2017; PMID 27027847; Cochrane, PMID 32588435), is B. The headline now shows B labeled with that use, and longevity and cardiometabolic markers keep C on their own row (Patikorn et al., BMC Med, 2023; PMID 37231411). No evidence changed. Headline policy adopted by the editor in October 2026 (Arnold Sandlow, D.C.): the headline letter is the record's best graded use, labeled with that use, and every other use, including any harm finding, is listed beneath it.

The sarcopenia use rested on a pooled analysis (Gu et al., Maturitas, 2025; PMID 39999663) and was held below A by judgment. The placebo-controlled trials were read individually: Yang et al. (J Nutr Health Aging, 2023; PMID 37248756) found handgrip 4.6 kg higher across the whole sample; Meza-Valderrama et al. (Arch Gerontol Geriatr, 2024; PMID 38171034) found it higher in women only. Handgrip qualifies for A, but a benefit not replicated across independent groups is B under the written definition, so B stands, now for that stated reason. Nasimi et al. (J Am Med Dir Assoc, 2021; PMID 33067129) tested a fortified food, and Wu et al. (Crit Care, 2023; PMID 38102705) found no effect of HMB alone in intensive care. Applied the written grade definition at the editor's request of October 2026 (Arnold Sandlow, D.C.).

The hormone-response use was graded C on Corpas et al. (J Clin Endocrinol Metab, 1992; PMID 1379256), which randomized only the order of two active doses in 10 men, and Vittone et al. (Metabolism, 1997; PMID 9005976), which was uncontrolled. The site's written definition of a 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. is people randomly assigned to treatment or placebo; neither study had a placebo arm, so sermorelin has not been tested in a randomized trial on any outcome. The use moves to D on 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. and the headline follows. Applied the written grade definition at the editor's request of October 2026 to settle the calls the definitions decide (Arnold Sandlow, D.C.).

The perioperative use was held at B because the Montreal Cognitive Assessment is administered rather than recorded by an instrument. The written A definition also accepts a validated outcome scale scored by a clinician blinded to treatment, and two double-blind Neither the participants nor the people assessing them know who received the real treatment until the trial ends. trials from independent groups found less postoperative cognitive dysfunction with active stimulation: Zhou et al. (Aging Clin Exp Res, 2022; PMID 35809206) and Ma et al. (J Geriatr Oncol, 2026; PMID 42456197), with Li et al. (Neuromodulation, 2026; PMID 42104979) agreeing. The use moves to A, with the limits (one country, about 100 patients per trial, one retrospective registration) stated beside the letter. Applied the written grade definition at the editor's request of October 2026 to settle the calls the definitions decide (Arnold Sandlow, D.C.).

The delayed sleep phase use was held at B by judgment while its own note said the endpoint qualifies for A. Under the written A definition, sleep onset measured by polysomnography or actigraphy is function recorded by an instrument, and two independent groups found it earlier on melatonin than on placebo in double-blind trials: Kayumov et al. (Psychosom Med, 2001; PMID 11211063) and Sletten et al. (PLoS Med, 2018; PMID 29912983). The use moves to A with a replication block naming both trials and their limits, and the headline follows the best use. Applied the written grade definition at the editor's request of October 2026 to settle the calls the definitions decide (Arnold Sandlow, D.C.).

The record said no randomized trial had assigned people to drink less and measured mortality. Re-running its stored search found Kristenson et al. (Alcohol Clin Exp Res, 1983; PMID 6135365): 585 middle-aged Malmo men with heavy drinking randomized to counseling to cut down or a letter, with mortality about half as high over 6 years in the intervention group (60-month report, Kristenson et al., Alcohol, 1985; PMID 2862882). A single unreplicated trial on a hard endpoint An outcome that matters directly to a person and is counted without judgment, such as death, heart attack, stroke or fracture. is C, now its own row. The headline stays B on blood pressure, and the record's endpoint field, which still named mortality, now names blood pressure. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The ref called Zugni et al. (Cancer Imaging, 2020; PMID 32393345) a systematic review; it is a review and recommendations without formal risk-of-bias appraisal, now labeled so, with its per-subject figures added (cancer suspected in 1.8 percent, confirmed in 1.1 percent). Added: Kwee et al. (J Magn Reson Imaging, 2019; PMID 30932247), Martins da Fonseca et al. (Eur Radiol, 2026; PMID 40884613) and Ahsan et al. (J Gen Intern Med, 2026; PMID 42823587), all pooling yield studies with no outcome data, and the 2026 Cochrane protocol (Chen et al., Cochrane Database Syst Rev, 2026; PMID 42639816). grade_basis now names Route 1 in words. The absence check was re-run with the same result. D Route 1 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

grade_basis said no prior randomized trial had tested this scheduling question; Meyer et al. (Eur J Cardiovasc Prev Rehabil, 2006; PMID 16874148) randomized adults to two consecutive weekend sessions and Tripp et al. (Scand J Med Sci Sports, 2025; PMID 39921357) to two against four sessions at matched volume. Neither tested one session, so the claim now reads that no other randomized trial has tested a single weekly session in adults, with both trials described. The endpoint now names fat mass as the primary outcome and VO2max as secondary, and endpoint_type reads surrogate. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

grade_basis said the prostate finding carries the letter and set ATBC against SELECT as two trials of the same question. ATBC (N Engl J Med, 1994; PMID 8127329) tested 50 mg/day in smokers; it is now labeled a different dose and population. The D now rests on replicated no-benefit: SELECT's primary report (Lippman et al., JAMA, 2009; PMID 19066370), the Physicians' Health Study II (Gaziano et al., JAMA, 2009; PMID 19066368), the Women's Health Study (Lee et al., JAMA, 2005; PMID 15998891), HOPE-TOO (Lonn et al., JAMA, 2005; PMID 15769967) and the USPSTF review (O'Connor et al., JAMA, 2022; PMID 35727272), all added. The SELECT harm stays as the badge, now with its context: nonsignificant at the primary report, significant on follow-up that ran past the end of supplementation, not reproduced by PHS II. HOPE-TOO's heart failure finding and the hemorrhagic stroke signal (Loh et al., Stroke Vasc Neurol, 2021; PMID 33109618) added. D Route 2, harm found, unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The dose field said the graded trials used 2,000 to 5,000 IU/day; the positive fracture trial used 800 IU/day with calcium, VITAL 2,000 IU/day, D-Health 60,000 IU monthly and FIND 1,600 or 3,200 IU/day, now stated. grade_basis and ref still read Headline Grade C after the October calcium row set the headline at B; both now say B and why. The replete-adult fracture row said it did not contradict a Grade B for deficiency, which moved to D in October; corrected, with D-Health (Waterhouse et al., 2023; PMID 37011645) and FIND (Rikkonen et al., Bone, 2026; PMID 42704968) added, including FIND's higher fracture count at 1,600 IU/day. The cardiovascular row adds D-Health (Thompson et al., BMJ, 2023; PMID 37380191), borderline and consistent with no effect; the cancer row adds Keum et al. (Br J Cancer, 2022; PMID 35676320), null on incidence with a daily-dosing subgroup signal on cancer mortality. The trial populations now read not selected for vitamin D status rather than replete. All letters unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

evidence_type read rct on a record whose only study 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. and uncontrolled and whose best use says no randomized trial has been run; it now reads as a single uncontrolled study. The ref described 12 hypertensive adults; Houston et al. (Cureus, 2023; PMID 37849591) enrolled adults with elevated pressure or mild hypertension, and the diastolic fall was confined to those with raised diastolic pressure. The ref now also says the paper's efficacy conclusion is beyond what the design can show. The absence check was re-run with the same result. D Route 1 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

best_use led with mitophagy and mitochondrial health, which are markers, and said safety is well established on 4 weeks to 4 months of exposure in a few hundred people. It now names muscle function, where both larger trials missed their pre-registered primary endpoints, and states the exposure. Added: Dao et al. (Front Nutr, 2026; PMID 42559151), meta-analysis with the 6-minute walk inconclusive at low certainty; Jamialahmadi et al. (Rev Recent Clin Trials, 2024; PMID 38415449), null in heart failure; Monsalve Acevedo et al. (Front Nutr, 2025; PMID 41245402), a small positive pilot in young soccer players. Andreux et al. (Nat Metab, 2019; PMID 32694802) is relabeled as a phase 1 trial with safety as its primary outcome. The maker's funding of five trials was already stated trial by trial. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record graded immune measures, but its only citation, the Cochrane review (Pilkington et al., 2022; PMID 36445793), studied survival and adverse effects in colorectal cancer, not immune markers, and the ref did not say what it found. The endpoint is now overall survival after curative resection with PSK added to chemotherapy. Added: Sakamoto et al. (Cancer Immunol Immunother, 2006; PMID 16133112) and Oba et al. (Cancer Immunol Immunother, 2007; PMID 17106715), positive pooled survival; Miyake et al. (Surg Today, 2018; PMID 28634730), Okuno et al. (Cancer Chemother Pharmacol, 2018; PMID 29094178) and Ogawa et al. (Anticancer Res, 2024; PMID 38307591), null; Zhong et al. (Front Pharmacol, 2019; PMID 31333449), immune markers pooled with Ganoderma products. The dose field now states PSK at about 3 g/day and that retail products are not shown equivalent. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

Route 2 rested on Jamialahmadi et al. (Biomed Pharmacother, 2022; PMID 35045351), 15 heart attack survivors given intravenous trehalose, and grade_basis never discussed Kaplon et al. (Aging, 2016; PMID 27208415), the only oral trial in healthy middle-aged and older adults, whose full-sample primary analysis was not significant. Kaplon is now the basis and Jamialahmadi supporting context, with its population corrected. The dose field's 2 g figure matched no trial and is replaced by the doses studied. The ALS platform trial (HEALEY, Lancet Neurol, 2025; PMID 40409314) now records serious adverse events in 16 against 7 percent. Added: Yap et al. (Parkinsonism Relat Disord, 2024; PMID 38843619), oral trehalose in spinocerebellar ataxia, null at interim; Yoshizane et al. (Nutr J, 2020; PMID 32646428), run by a trehalose manufacturer. endpoint_type corrected from hard to surrogate. D Route 2 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record graded a whole modality on one cognition meta-analysis with no population, device or dose, and named an autonomic endpoint with no autonomic citation. It now carries four uses. Postoperative cognitive dysfunction in older surgical patients: three double-blind sham-controlled trials from three centers agree (Zhou et al., 2022; PMID 35809206; Li et al., 2026; PMID 42104979; Ma et al., 2026; PMID 42456197), with less postoperative delirium in a fourth (Shi et al., 2026; PMID 41835922); a validated cognitive test replicated across independent groups meets the written B definition, and the headline follows the best use, C to B. Cognitive test performance in mixed samples stays C on the heterogeneous pools (Liu et al., 2026; PMID 42339645; Ridgewell et al., 2021; PMID 34043386). Mild cognitive impairment is C on one trial (Wang et al., 2022; PMID 36150665). Heart rate variability in healthy adults is D on Route 2, tested and no benefit shown in pooled sham-controlled trials (Souza et al., 2026; PMID 41506123; Soltani et al., 2023; PMID 37119426). Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

grade_basis scoped the A to left prefrontal high-frequency or theta-burst stimulation while two supporting trials used H1-coil deep TMS at 18 Hz (Levkovitz et al., 2015; PMID 25655160; Kaster et al., 2018; PMID 29946106), one in late-life depression; the ref, grade_basis and replication block now say so, and disclose the coil maker's sponsorship of the Levkovitz trial and in-kind equipment for the Kaster trial beside the Neuronetics funding already stated for O'Reardon. The record said a pooled analysis found no effect on suicidal ideation; Mehta et al. (2022; PMID 35044731) found a non-significant reduction in randomized trials, now stated. Added: Eren et al. (2026; PMID 41997374), 30 sham-controlled comparisons with larger effects in single-blind trials; Zhang et al. (2026; PMID 42383744), 141 trials with only 3 at low risk of bias; Orbo et al. (JAMA Netw Open, 2026; PMID 42418203), an independent sham-controlled theta-burst trial positive on blinded MADRS at day 10 and not sustained. The O'Reardon baseline-adjustment caveat is confirmed by that trial's abstract and stands. A unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record's only citation (Li et al., 2026; PMID 42435147) is a network meta-analysis restricted to chronic musculoskeletal pain and does not test tDCS for depression, yet the endpoint was depression scales with no supporting trial. The record now carries the trial base: the individual patient data meta-analysis (Razza et al., 2026; PMID 41498282), SELECT-TDCS (Brunoni et al., JAMA Psychiatry, 2013; PMID 23389323) and ELECT-TDCS (Brunoni et al., N Engl J Med, 2017; PMID 28657871) as positives, DepressionDC (Burkhardt et al., Lancet, 2023; PMID 37414064), the international trial (Loo et al., Brain Stimul, 2018; PMID 29111077) and unsupervised home use (Borrione et al., JAMA Psychiatry, 2024; PMID 38170541) as large nulls, Woodham et al. (2025; PMID 39433921) as the maker-funded home-based positive, and Ren et al. (2025; PMID 40531534) finding tDCS alone not significant. Two use rows authored, clinic-administered and home-based, both C. endpoint_type corrected to clinician-rated; the not-replicated flag removed, the problem being inconsistency. Safety signals from the trials added to the page. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record rested on Leitao et al. (2021; PMID 33541567) described as affecting testosterone-related measures; that trial's primary outcome was erectile function and its largest gains were in the extract-plus-training arm, so it does not isolate the extract. The use is now named as total testosterone in men with low testosterone, following Leisegang et al. (2022; PMID 36013514), whose pooled rise is significant in that group and not in healthy men. Added: Chinnappan et al. (2021; PMID 34262417), the largest placebo-controlled trial, funded and co-authored by the extract's maker, stated beside the letter; Chan et al. (2021; PMID 33559971); Morgado et al. (2024; PMID 37697053); Kotirum et al. (2015; PMID 26365449). A dose field now states the standardized water extracts at 100 to 600 mg/day, and the track field no longer names product brands. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The dose field read 500 to 1,000 mg/day; the trials that lowered homocysteine used 1.5 to 6 g/day, and Alfthan et al. (2004; PMID 15522136) found 1 g/day did not lower it significantly, which the field now states. Lu et al. (2023; PMID 36717385) was presented as a second betaine trial; it tested folic acid, B6 and B12 plus 1 g betaine, so it is relabeled as a combination that cannot grade betaine, with the study products made by a supplement manufacturer that employs one author. Added: Steenge et al. (2003; PMID 12730412) as independent replication; Alfthan 2004; the McRae 2013 (PMID 23997720), Ashtary-Larky 2022 (PMID 33764214) and Zawieja 2021 (PMID 31809615) pools, the latter two documenting higher total and LDL cholesterol at 4 g/day or more; Schwab et al. (2011; PMID 20978525), homocysteine unchanged at 4 g/day for 6 months; Olthof et al. (2006; PMID 16871332), no improvement in flow-mediated dilation despite lower homocysteine. The form_note claim that forms are equally effective, which no trial tested, is removed, and the track field no longer says betaine may lower LDL. B unchanged, now scoped to the studied dose. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record described its only citation (Lin et al., 2026; PMID 42332459) as showing weak, sham-equivalent evidence; that network meta-analysis finds significant pain reduction with ultrasound and ranks it first for pain, mostly through indirect and active-comparator comparisons. The sham-equivalent wording is supported instead by Hussain et al. (2018; PMID 29767650) and Szajkowski et al. (2024; PMID 37721730), both null on pain against sham, now cited. Ansari et al. (2024; PMID 35315745) added for the mixed modality picture, and the Cochrane review of chronic low back pain (Ebadi et al., 2020; PMID 32623724) added so the page says the letter does not extend to pain elsewhere. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

grade_basis counted Borsky et al. (2025; PMID 40592961) as a second positive randomized trial of plasma exchange; it tested plasma removal without replacement, randomized only between 8 and 4 sessions with no untreated control, and reported GrimAge, the Hannum clock and the Dunedin pace of aging increasing. The ref now states that finding and that it does not replicate plasma exchange. Fuentealba et al. (2025; PMID 40424097) is now described as single-blinded with allocation by entry date in the ref's first sentence, and the record discloses that one author co-founded a company that sells plasma exchange, with no specific funding declared. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The acute exercise row said a trial having been run is why the use is Route 2, under a Grade C; the sentence now says the trial found a benefit, which is why the use is C. The aging absence check claimed zero human randomized trials; the re-run query returns 13, and the note now names Van Stijn et al. (2015; PMID 26035756), a week-long perioperative course around hip fracture surgery that reported one-year mortality, null, and Abud et al. (2022; PMID 35700594) and Chupel et al. (2021; PMID 33586039), marker and fitness trials in older women, and explains why none tests the aging claim. Marcangeli et al. (2025; PMID 41061678) added as the observational counterpoint to the taurine-decline premise. Waldron et al. (2018; PMID 29546641) and Deng et al. (2025; PMID 40852891) added as context for the exercise row. The dose field now states 1 g single dose and 2.4 to 3 g/day as studied. All letters unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

best_use led with NRF2 activation and claimed neuroprotective, cancer-preventive and cardiometabolic effects that no cited randomized trial measured; the record's own NRF2 trial (Wise et al., 2016; PMID 27832073) was null and a hemodialysis crossover (Ribeiro et al., 2024; PMID 37619675) found no change in NRF2 mRNA, with one small positive in non-dialysis kidney disease (PMID 38772511). The field now states glucose control in type 2 diabetes, the only clinical question the trials address. Added: Dwibedi et al. (2025; PMID 39929977), the Lund group's prediabetes trial that missed its prespecified primary endpoint with patent and stock interests disclosed; Christiansen et al. (2010; PMID 20805984), no effect on endothelial function or blood pressure in hypertension. Axelsson 2017 (PMID 28615356) is now described as a subgroup result, and PMIDs 22537070, 21559038 and 22325157 as one 81-patient Tehran trial reported three times. The dose field now names what was studied rather than a 10 to 40 mg label. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The dose field read 1 to 2 mg/day from wheat germ extract as if a trial had supported it, and form_note said the supporting study used that preparation; the cited support was a mouse study (Eisenberg et al., Nat Med, 2016; PMID 27841876) and a dietary cohort (Kiechl et al., Am J Clin Nutr, 2018; PMID 29955838), and the one 12-month randomized trial, SmartAge (Schwarz et al., 2022; PMID 35616942), used 0.9 mg/day and was null. Both fields now say so. Added: Senekowitsch et al. (2023; PMID 37111071), 15 mg/day did not raise plasma spermidine; Keohane et al. (2024; PMID 39405978), 40 mg/day safe with minimal polyamine change, authors employed or funded by the product's maker; Pekar et al. (2021; PMID 33211152), an intake comparison without a described placebo arm; Alsaleh et al. (2026; PMID 42169618), immune markers after vaccination, part-funded by the wheat germ product's maker. evidence_type now records the randomized trials. D unchanged on both uses. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The name and description implied a benefit at a 2,300 mg threshold; the pooled trials (He et al., BMJ, 2013; PMID 23558162) tested an average 75 mmol/day reduction regardless of end level, with a larger fall in hypertensive than normotensive people, and DASH-Sodium (Sacks et al., N Engl J Med, 2001; PMID 11136953) shows the fall continuing below 2,300 mg. Added: the 133-trial dose-response meta-analysis (Huang et al., BMJ, 2020; PMID 32094151) and the Cochrane review of cardiovascular events (Adler et al., 2014; PMID 25519688), which is imprecise and explains why the letter stops at B. Potassium-enriched salt substitutes are named as a different intervention. B unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The title read Grade C for the hormone response while the headline letter was D, and the hormone row called two studies from the same National Institute on Aging group independent. Under the site's convention the headline follows the best use, so it is C; every health outcome stays D on Route 1 on its own row. The hormone row now says Corpas 1992 (PMID 1379256) randomized dose order only, with no placebo, and Vittone 1997 (PMID 9005976) was uncontrolled. The cognition row and ref called Friedman 2013 (PMID 23689947) a second positive trial; it is a spectroscopy substudy of Baker 2012 (PMID 22869065). Veldhuis 2004 (PMID 15579798), GHRH(1-44), is added as context beside the other different-molecule trials. DEXA lean and fat mass are retyped as a surrogate. The editor may prefer the headline held at D for health outcomes, as the record formerly had it. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline ref argued its letter from cancer and cardiovascular trials while the title named thyroid antibodies, cited a selenium plus coenzyme Q10 trial (Alehagen et al., Int J Cardiol, 2013; PMID 22626835) as selenium evidence, and called the Nutritional Prevention of Cancer trial suggestive when its skin cancer primary was null (Clark et al., JAMA, 1996; PMID 8971064). The ref now follows the graded use. The replete-adult row rested on one secondary analysis; SELECT (Lippman et al., JAMA, 2009; PMID 19066370), the form-matched 35,533-man trial, the Cochrane review (Vinceti et al., 2018; PMID 29376219) and Sel/Cel (Thompson et al., 2016; PMID 27530657) are its evidence now, with the badge restated as no benefit and the diabetes excess stated as a signal. The thyroid row adds the null clinical outcomes: quality of life unchanged in CATALYST (Larsen et al., 2024; PMID 38215286) and TSH, ultrasound and quality of life unchanged in untreated patients (Winther et al., 2017; PMID 27683225). The form note says the antibody effect is pooled across three selenium forms. Letters unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

Text corrections after the letters were set in October. The dose line attributed 1,600 to 3,200 mg/day to Sarris 2020 (PMID 31712971), which used 800 mg/day in a 49-person pilot whose 3.76-point difference favoring SAM-e did not reach significance; both are now stated. Peng et al. (2024; PMID 38199136) found no difference from imipramine or escitalopram as well as from placebo. Added: a third null add-on trial (Alkema et al., Epigenomics, 2026; PMID 42027147), the 2025 network meta-analysis ranking the add-on effective by indirect comparison (Cheng et al., PMID 40314175) with the note that direct trials govern, and the Soeken osteoarthritis pool (2002; PMID 12019049). The endpoint types are corrected: clinician-rated scales in depression, self-reported pain in osteoarthritis. Safety now records the mania or hypomania reports and the serotonin-syndrome-like case. Letters unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record cited four small trials of one proprietary extract, each with a maker-linked author, and omitted the only fully independent registered trial (Punja et al., PLoS One, 2014; PMID 25268730), in which fatigue was significantly worse on Rhodiola than on placebo. Added with it: the systematic reviews rating every trial at high or unclear risk of bias (Ishaque et al., 2012; PMID 22643043) and noting no independent replication (Hung et al., 2011; PMID 21036578). The best-use line's anti-aging effects via AMPK, measured in no human trial, and the 200 to 600 mg/day range, which the trials do not support, are removed. Mixed evidence stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record rested on Timmers et al. (Cell Metab, 2011; PMID 22055504) alone and described sirtuin activation, NAD+ Nicotinamide adenine dinucleotide, a molecule cells use in energy metabolism. Its level falls with age in some tissues. amplification, a personality's protocol and ongoing longevity trials, none with a trial behind them. The independent attempts to reproduce the Timmers result were null on clamp-measured insulin sensitivity (Poulsen et al., Diabetes, 2013; PMID 23193181; Yoshino et al., Cell Metab, 2012; PMID 23102619; de Ligt et al., Mol Metab, 2018; PMID 29706321), and pooled trials find no effect on cardiovascular risk factors (Sahebkar et al., 2015; PMID 25885871): mixed, C, the headline unchanged. Gliemann et al. (J Physiol, 2013; PMID 23878368), where 250 mg/day blunted the fitness, blood pressure and lipid gains from exercise training in older men, is a D Route 2 row with the harm badge. The Timmers trial's co-author employed by the formulation's maker is stated beside the letter. The dose line now states the trial doses. The headline moves C to D under the convention that a harm finding, where one exists, leads the record; the editor may prefer the headline on the best use. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The A rested on one pooled analysis in adults 75 and over; the Cochrane review of progressive resistance training (Liu and Latham, 2009; PMID 19588334), 73 trials on measured strength, now stands beside it and the population is no longer narrowed to 75 and over. The glycemic use said it rested on a single pooled analysis; Wan et al. (2024; PMID 38623887), 46 trials, is added. The blood pressure use now cites the earlier, smaller Cornelissen pool (2013; PMID 23525435). Wording that implied every measure improved is corrected: handgrip strength and fiber size did not change significantly in Grgic 2020 (PMID 32740889). The Greek replication trial (Kalapotharakos et al., Aging Clin Exp Res, 2010; PMID 20440099) states no funder, so its entry reads unclear. Letters unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record cited one Cochrane review (Jin et al., 2016; PMID 27045603) without saying the trials used different preparations, that no trial measured survival, or that the review rated quality unsatisfying; the use row and form note now say so, and the tumor response figure with its interval crossing 1 is not leaned on. The cardiometabolic use a reader meets in marketing was untouched: the Cochrane review of cardiovascular risk factors (Klupp et al., 2015; PMID 25686270) and the largest placebo-controlled trial (Klupp et al., Sci Rep, 2016; PMID 27511742) are null, D Route 2 on their own row. A case report of acute liver injury (Guedikian et al., 2023; PMID 37885515) is added to the cautions as a signal. The evidence type is corrected to pooled randomized trials. The headline stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The D Route 2 rested on two pooled analyses without naming a trial. Added: Kersten et al. (J Gerontol A, 2013; PMID 22982689), the one trial that clearly lowered the burden, null on cognition at 8 weeks, and van der Meer et al. (BMJ Open, 2018; PMID 30030308), which did not lower the burden and found no difference in falls, admissions or mortality. The endpoint type is corrected to the measured cognition and function outcomes the trials used; cardiovascular events, the only hard outcome the observational study links to burden, are marked as never tested. The sentence calling the observational harm signal real is replaced: association carries no grade in either direction. The track field no longer tells the reader what to do. D Route 2 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The endpoint and ref read recurrent coronary events and mortality as if equal; CCSPS (Lu et al., Am J Cardiol, 2008; PMID 18549841) pre-specified major coronary events and reported mortality as secondary, which the record now says. The meta-analysis of post-infarction patients (Sungthong et al., Sci Rep, 2020; PMID 32066811) is cited with the note that it counts CCSPS subgroup papers as separate studies, so it is not replication. Safety now states the statin-class risks of monacolin K, the pooled randomized safety data (Fogacci et al., 2019; PMID 30844537) and the umbrella review's uric acid and liver enzyme signals (Ma et al., 2024; PMID 38413255). The trial's funder is stated as not given in the indexed record. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record said rapamycin had been given to humans twice on aging questions and that no trial had tested daily dosing. Kraig et al. (Exp Gerontol, 2018; PMID 29408453) randomized 25 adults aged 70 to 95 to 1 mg daily and found no change in cognition, grip strength or walking, with falls in hemoglobin and red cell indices; it is now the third tested-null payload and its safety signal is in the cautions. The Kell 2026 study (PMID 41524558) is restated as a 4-against-5 marker study reporting lower immune-cell p21, not a white-cell-count study. The best-use line no longer says one trial found attenuation of exercise benefit, which the basis itself discounts. PEARL's support by the telehealth company that prescribes rapamycin, with authors employed there, is stated. D Route 2 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record's C rested on two meta-analyses it described as showing modest benefit; both are null on their pooled outcomes (Sahebkar, 2017; PMID 25897620, lipids; Ostadmohammadi et al., 2019; PMID 30848564, glucose), so those uses are D Route 2 rows. The only positive pooled quercetin-alone evidence is blood pressure: systolic about 2 to 3 mmHg lower at 500 mg/day or more in two meta-analyses (Serban et al., 2016; PMID 27405810; Huang et al., 2020; PMID 31940027) with the 2023 umbrella review finding very high heterogeneity and diastolic pressure unchanged (Arabi et al., PMID 37654199), mixed, C, and the headline now attaches to it. The best-use line led with senolytic A compound intended to clear senescent cells: older cells that have stopped dividing but remain in tissue. use in combination with dasatinib or fisetin and the dose line described a quercetin plus fisetin schedule with no trial behind it; quercetin alone as a senolytic is untested on any outcome, D Route 1 with an absence record, and the one quercetin-alone marker study (Mury et al., 2025; PMID 40375481) is cited as such. The phytosome form note's Mayo Clinic claim and 20-fold figure, unsupported, are removed. The headline letter stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record summarized Chang et al. (2026; PMID 42075549) as modest self-reported benefit when its headline result is no VAS pain or total WOMAC improvement at one month, no objective function change and high risk of bias; it is now described as such. The evidence note claimed significant pain effects while Chen et al. (2019; PMID 31583420) found none. Added: the positive syntheses (Wu et al., 2018; PMID 30552258; Yang et al., 2020; PMID 32251502; Vigano et al., 2021; PMID 32508140) so the mixed picture is visible, and the 2026 double-blind sham trial null on WOMAC and function (Lau et al., PMID 41588476). The not-replicated flag, wrong for a nine-trial pool, is replaced: the problem is inconsistency. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The dose line quoted an 80 percent against under 1 percent bioavailability The share of a dose that reaches the bloodstream in a form the body can use. comparison with resveratrol and the form note a blood-brain barrier advantage; neither has a human study on the record, and both are restated as unsupported. Added as context to the D Route 2 set in October: the safety report from the same trial (Riche et al., J Toxicol, 2013; PMID 23431291), a 12-week safety pilot with no efficacy outcome (Otsuka et al., 2025; PMID 40024751) and a megestrol combination trial in endometrial cancer with a null primary (Senguttuvan et al., 2025; PMID 40227411). Letters unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The B rested on a Plantago-genus pool (Zhu et al., Nutr Res, 2024; PMID 38688104) and gave no effect size How large a difference the treatment made, as distinct from whether the difference could be detected at all.. The psyllium-specific meta-analysis at the protocol's dose (Jovanovski et al., Am J Clin Nutr, 2018; PMID 30239559), the 2025 update (Gholami et al., PMID 41366295) and a representative 26-week trial (Anderson et al., 2000; PMID 10837282) are added, with the pooled LDL reduction of about 0.33 mmol/L stated. Zhu 2024 is relabeled as a genus-level pool whose 7 percent risk figure is a projection. Tolerability and the spacing of psyllium from other oral medicines are added. B unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record said the evidence was one mouse study acting on BCL-2 pathways. Later mouse work from other groups exists (Liu et al., PNAS, 2024; PMID 38657052), and the founding paper (Xu et al., Nature Metabolism, 2021; PMID 34873338) describes reactive oxygen species and mitochondrial dysfunction, not BCL-2; its supplement-company funding and the authors' patent and senolytic company interests are now stated beside the Expression of Concern (PMID 41731120). D Route 1 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record was named plyometric and power training and listed box jumps and medicine ball throws, but its one citation (Balachandran et al., JAMA Netw Open, 2022; PMID 35544136) pooled fast-velocity resistance training against traditional lifting, mostly on machines, and found a small low-certainty advantage with self-reported function null. Added: Morrison et al. (2023; PMID 37328359), which calls the effects similar, and the Lopez et al. network meta-analysis (2023; PMID 36378500), which finds traditional training better on two tests; jump training in older adults (Moran et al., 2018; PMID 30341594) is cited as a power outcome, not function. The uncited opening claim about power declining faster than strength is removed, and the track field now names the tests the trials used. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record said piperlongumine kills senescent cells by pushing reactive oxygen species past a threshold, shown in cells and animals. The one cited paper (Wang et al., Aging, 2016; PMID 27913811) reports apoptosis in cultured fibroblasts that did not require reactive oxygen species, and its abstract reports no animal work. Its authors' patent application and senolytic company ties are now stated. The advice-shaped sentence in the form note is replaced with a description. D Route 1 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description graded a consumer longevity application on an osteoarthritis meta-analysis and called fracture non-union evidence solid. No randomized trial of PEMF on a longevity, aging or hard outcome exists, and the Cochrane review of non-union (Griffin et al., 2011; PMID 21491410) found the pooled effect not significant, with a 2024 update (Picelli et al., PMID 39387850) not supporting PEMF for acute fractures. Added beside Wu 2018 (PMID 30552258): the null pain results of Chen et al. (2019; PMID 31583420) and Chang et al. (2026; PMID 42075549). The not-replicated flag, wrong for a 12-trial meta-analysis, is replaced; the evidence is mixed, not unreplicated. The name now states the graded use. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record already named the two null trials in its ref but graded nothing on them. They now carry their own rows: collagen markers in hypertension with raised galectin-3 (Lau et al., JACC Basic Transl Sci, 2021; PMID 33532663) and knee osteoarthritis symptoms (Andrews et al., Contemp Clin Trials Commun, 2020; PMID 31872160), each tested once without benefit, D on Route 2. The prostate use stays D Route 1 on its own row: the 78 percent without PSA progression (Keizman et al., 2021; PMID 34959847) is an uncontrolled figure, now said so, and the 18-month extension (Keizman et al., 2023; PMID 37630724) reports only the 39 men who responded at six months. The developer's review (Eliaz et al., Nutrients, 2019; PMID 31683865) and the combination cell work (Jiang et al., Integr Cancer Ther, 2013; PMID 22532035) are labeled as background and mechanism. The observational flag is removed from a record whose studies are single-arm interventional. Headline D unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The ref said the Cochrane review (Lee et al., 2021; PMID 33871063) showed possible benefit; it found a 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. but trivial gain on erectile function scales, below the minimal clinically important difference, with only self-reported ability to have intercourse improved at low certainty, and the record now says so. The record named no form, dose or duration; the use row states Korean red ginseng or ginseng berry extract at about 600 to 2,700 mg/day for 4 to 12 weeks, with Hong et al. (2002; PMID 12394711) as the representative trial. Added: the 2026 herbal-supplement meta-analysis (Ho et al., PMID 41696741) and the Cochrane adverse-event finding. The vague endpoint and the track field's fatigue, immune and cognitive applications, none graded, are replaced. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description led with REDUCE-IT's 25 percent reduction (Bhatt et al., NEJM, 2019; PMID 30415628), which tested purified EPA, a different agent, in a trial funded by its maker against a mineral-oil placebo; it now leads with the null trials at the protocol's dose, STRENGTH (Nicholls et al., JAMA, 2020; PMID 33190147) and OMEMI (Kalstad et al., Circulation, 2021; PMID 33191772). Added: the dose-related atrial fibrillation increase in the outcome trials (Gencer et al., 2021; PMID 34612056; Abuknesha et al., 2026; PMID 42517224), the most relevant harm at 2 to 4 g/day, and OMEMI's own borderline signal; the Cochrane review's high-certainty nulls on mortality and cardiovascular events (PMID 32114706) beside the slight coronary death reduction the record quoted; the population-specific positive PISCES trial in hemodialysis (Lok et al., NEJM, 2026; PMID 41201837); and the fact that Hu 2019 (PMID 31567003) pooled trials mostly at 1 g/day or less. The Omega-3 Index The share of EPA and DHA among the fatty acids in red blood cell membranes, reported as a percentage. target and the form line are restated as untested. The not-replicated flag, inconsistent with the record, is removed, and the name loses its dash. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The grade basis called Martens 2026 (PMID 42478598) the first randomized trial of NR alone against clinical outcomes; Dollerup 2018 (PMID 29992272), Orr et al. (2024; PMID 37994989), NICE (McDermott et al., Nat Commun, 2024; PMID 38871717) and Wu et al. (2025; PMID 41357333) came first and are now cited. The one positive, six-minute walk in peripheral artery disease at a one-sided 0.10 level, is a C row, unreplicated; blood NAD+ is a C row under the blood-level rule applied to NMN today; clamp insulin sensitivity in obese men (one null) and cognition in mild cognitive impairment (two independent nulls) are D Route 2 rows. Dolopikou 2020 (PMID 30725213) is relabeled: VO2max unchanged. The NR plus pterostilbene trial (PMID 29184669) is labeled as funded by the combination's seller with its employees as authors, and the NR supplier's funding or product supply across most trials is stated. Ratajczak 2016 (PMID 27725675) is labeled cell and mouse work. The headline stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

Text corrections after the letters were set earlier today. The Route 1 longevity row cited a blood-pressure meta-analysis (Zhang et al., 2026; PMID 41901064), which now sits in the ref described as what it is: diastolic pressure about 2 mmHg lower, systolic unchanged. The insulin-sensitivity row omitted the one other direct measurement, null at 2,000 mg/day (Pencina et al., 2023; PMID 36740954), and three meta-analyses finding no effect on glucose, insulin, HbA1c or HOMA-IR (PMIDs 39531138, 39116016, 42514320), now stated. The dose line said sublingual may improve absorption; the only randomized comparison (Wakabayashi et al., 2026; PMID 42304075) found no NAD+ difference between routes. The liposomal claim had no trial and is removed. The industry authorship of Yi 2023 (PMID 36482258) is stated. Letters unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record described HPS2-THRIVE (Landray et al., NEJM, 2014; PMID 25014686) as extended-release niacin and badged it harm found; the trial gave niacin combined with laropiprant in one pill, so its excess serious adverse events (Haynes et al., 2019; PMID 31447131) cannot be assigned to niacin alone. The large trial of extended-release niacin alone on a statin, AIM-HIGH (Boden et al., NEJM, 2011; PMID 22085343), was missing and is now the record's clean test, null. The badge reads TESTED, NO BENEFIT with the combination trial's harm signal stated in the certainty line, the title and the basis. Added: the Cochrane review of 23 trials with no effect on death, infarction or stroke (Schandelmaier et al., 2017; PMID 28616955), both trials' maker co-funding, the run-in that excluded niacin-intolerant people, and the Coronary Drug Project (Canner et al., J Am Coll Cardiol, 1986; PMID 3782631) as the reason immediate-release niacin is excluded rather than graded. D Route 2 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The replication block listed the Cleveland trial (Klika et al., 2022; PMID 32610358) as unclear funding when the report states funding by Cymedica Orthopaedics, the device maker, now stated; and it omitted the cleaner independent replication on dynamometer strength, Yoshida et al. (2017; PMID 28610968), now listed. The as-studied dose and duration fit the Colorado trial only and now state the three trials' differing regimens. Avramidis 2011 (PMID 21410130) is relabeled as vastus medialis stimulation with mobility and self-report outcomes, and DeJong 2020 (PMID 32360105) as patterned stimulation, a different form. Added: Levine 2013 (PMID 23464951), home stimulation noninferior to supervised therapy; Cheuy 2023 (PMID 35856287), same Colorado group; Monaghan 2010 Cochrane (PMID 20091621) as context. The letter stays A: instrument-measured strength replicated by independent groups is what the written A asks for; the basis now states the case for B so the editor can weigh it. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The best-use line and ref said intact oral NAD+ had no controlled human trial while the form note described one. The one randomized trial (Kornilov et al., GeroScience, 2026; PMID 42530810) is a 5-day pharmacokinetic study of a proprietary formulation whose authors are linked to its developer; it raised whole-blood intracellular NAD and left plasma NAD unchanged, and measured no outcome. The record now says no randomized trial has tested oral NAD+ on any health outcome, cites the 2026 systematic review that found none for oral or intravenous NAD+ (Gallagher et al., Aging Res Rev, 2026; PMID 41655607), corrects the preclinical-only flags, and notes that NADH trials are a different form in combination. D Route 1 unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline C rested on one meta-analysis in schizophrenia while the best-use line listed glutathione, antioxidant, liver, mucolytic and insulin-sensitivity claims none of the cited trials tested, and the ref said cognition was not an outcome when Yolland et al. (2020; PMID 31826654) measured it. Schizophrenia is now a row: Berk 2008 (PMID 18436195) and Yolland positive after 24 weeks, Zhang et al. (2024; PMID 37712506) null across 8 trials, Neill et al. (2022; PMID 35857752) and Conus et al. (2018; PMID 29462456) null on symptoms, mixed, C. COPD exacerbations, NAC's best-studied use and absent from the record, are a row: PANTHEON (Zheng et al., Lancet Respir Med, 2014; PMID 24621680) and HIACE (Tse et al., Chest, 2013; PMID 23348146) positive at 600 mg twice daily, BRONCUS (Decramer et al., Lancet, 2005; PMID 15866309) and Zhou et al. (Nat Commun, 2024; PMID 39349461) null, mixed, C. The condition-page PCOS row (Rizk et al., 2005; PMID 15705376) is given its own row at C as a single cited trial. The endpoint type is corrected to clinician-rated, the dose line to the trial doses. The headline stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description said one trial reported reversal of an epigenetic clock A test that estimates biological age from chemical marks on DNA (DNA methylation).. The trial (Fitzgerald et al., Aging, 2021; PMID 33844651) randomized 43 men aged 50 to 72 to a program that included two supplements from the funder; the within-group change was not significant (p=0.066) and only the between-group difference was (p=0.018), so reversal overstated it. The record now names the clock (the 2013 Horvath clock, in saliva), the population, the supplements, the two published corrections (PMID 35907207, PMID 38488762) and the authors' uncontrolled case series in six women (PMID 36947707), which is not a replication. The track field, which described a FINGER-style cognitive battery the trial never measured, now follows the trial's own endpoint. In the pattern applied to other protocols the name states what was studied. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The grade basis said stage IV alone improved; the rate ratio was 0.86 with a confidence interval The range the true effect probably falls in. A wide interval means the estimate is uncertain. reaching 1.00 (Sasieni et al., NEJM, 2026; PMID 42776074), a trend, not a demonstrated improvement, and the text now says so. Late-stage incidence is a stage-shift surrogate for cancer mortality, so the endpoint type is corrected from hard to surrogate, with the UK National Screening Committee position statement cited (Batson et al., BMJ, 2026; PMID 42185047). The Mann 2026 spillover analysis (PMID 42217468) now states its result, a transient rise in diagnostic delay the authors judged immaterial, and is labeled observational. The AHRQ systematic review (Kahwati et al., Ann Intern Med, 2025; PMID 40953446), which found no completed controlled study showing screening benefit, is added. D Route 2 unchanged; funding disclosure retained. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record cited two positive sleep trials and said tart cherry raises endogenous melatonin; Howatson et al. (2012; PMID 22038497) attribute the rise to melatonin in the juice and Chung et al. (2022; PMID 36011907) found no change. Added: the two larger null trials (Tucker et al., 2024; PMID 39683518, 500 mg powder; Hillman et al., 2026; PMID 35790450, juice or capsules for 30 days) and the two small older positives (Losso et al., 2018; PMID 28901958; Pigeon et al., 2010; PMID 20438325). The dose line listed an extract dose no positive sleep trial used; it now states what was studied. Exercise recovery, claimed without a citation or a letter, now has its own row: pooled analyses positive on measured strength (Hill et al., 2021; PMID 33440334; Daab et al., 2026; PMID 41945263) against independent nulls (Lamb et al., 2019; PMID 31337122; Abbott et al., 2020; PMID 31188696; Morehen et al., 2021; PMID 32672095), mixed, C. The headline stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record's only citation (Zanini et al., 2021; PMID 34601077) was a 40-person pilot that found no lipid or metabolic difference, while the evidence note claimed multiple positive trials and the flags claimed a meta-analysis nobody cited. The lipid claim now rests on the three meta-analyses that exist and disagree (Ye et al., 2026; PMID 41888952; Jamialahmadi et al., 2024; PMID 39839806; Todorovic et al., 2023; PMID 37259294), with the key 24-week trial's part-funding by the tablet maker and its lead author's industry consultancy stated beside the letter (LeBaron et al., 2020; PMID 32273740). Zanini 2021 is relabeled as a pilot null on the graded endpoint. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

One C covered COPD symptoms, fatty liver and sleep as if one finding, and the record omitted the only hard-endpoint trial. Each indication now has its own row at C as a single unreplicated small trial (Zheng et al., 2021; PMID 33985501; Tao et al., 2022; PMID 35734974; Gao et al., 2026; PMID 40826930), with the sleep trial's funding by a device company, which took part in analysis and writing, stated beside the letter. HYBRID II (Tamura et al., EClinicalMedicine, 2023; PMID 36969346) missed its primary neurological outcome and is D on Route 2 for that use, with its favorable secondary outcomes and early stop stated. Ohsawa 2007 (PMID 17486089) is relabeled as cell culture and rat work. The endpoint type is corrected: the COPD and sleep scores are self-reported or actigraphic, only liver fat is a surrogate. Shi 2023 (PMID 37970554) and Ono 2017 (PMID 28669654) are added as context. The headline stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The best-use line described a mechanism and named no endpoint; the C now attaches to brachial flow-mediated dilation in older adults (Rossman et al., 2018; PMID 29661838; Kirkman et al., 2023; PMID 37560769), with the manufacturer's contract and capsule supply stated beside the letter. Added as tested nulls at the record's own 20 mg dose: physical function in healthy older adults (Murray et al., J Physiol, 2026; PMID 41733122) and endurance performance (Gonzalo-Skok et al., 2024; PMID 38981985; Shill et al., 2016; PMID 27501153; Broome et al., 2022; PMID 35623315 and PMID 35201920), each D on Route 2. Added as context: no slowing of Parkinson's disease over a year (Snow et al., 2010; PMID 20568096) and a single 80 mg dose that lowered VO2max (Hughes et al., 2023; PMID 38061764). Broome 2022 is relabeled as an exercise-training trial, and Gane 2010 (PMID 20492507) as a different dose and population. The headline stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description still said sleep outcomes were not graded here and cited only Goyal et al. (2014; PMID 24395196), while the grade basis had been rebuilt on the sleep meta-analyses (Rusch et al., 2019; PMID 30575050; Chen et al., 2020; PMID 32590218) and the randomized trials (Black et al., 2015; PMID 25686304; Ong et al., 2014; PMID 25142566; Perini et al., 2023; PMID 34193328). The description now matches: B on self-reported sleep quality, no difference from specific active treatments. The name said 20 minutes a day self-guided, a dose and format no cited trial tested; in the pattern applied to other protocols it now names the weekly group course with home practice the trials used. B unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record cited two meta-analyses and no trial. Added: the 2025 Cochrane review (Wang et al., PMID 40552569), which rates the ALT benefit very low certainty and finds silymarin complexes no better than placebo; the NIH-funded phase II trial null on histology at 420 and 700 mg three times daily (Navarro et al., 2019; PMID 31536511); and the Malaysian biopsy trial whose primary endpoint was not met (Wah Kheong et al., 2017; PMID 28419855). Li et al. (2024; PMID 38579127) is relabeled: it pools complexes and combination products with plain silymarin, its histology claim conflicts with the biopsy trials, and its enzyme effect sizes are implausible as labeled. The best-use line now says which form and dose range were tested and that complexes are different agents. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record's endpoint, memory and cognitive measures, rested on Telch et al. (2014; PMID 25018057), whose primary question was retention of fear extinction; memory was secondary. The closest test of the stated endpoint, Rodriguez et al. (Radiology, 2016; PMID 27351678), one oral dose in 26 healthy adults with 7 percent better retrieval, was not cited; it is from the same laboratory, so the two do not replicate each other. The one trial measuring cognition over months, 195 mg/day in bipolar disorder (Alda et al., 2017; PMID 27284082), was null and is added. The name says low-dose, but the trials used about 260 mg once; the daily few-milligram use consumer products supply has no randomized trial and now carries its own D Route 1 row with an absence record. The headline stays C on the single-dose row. The serotonin syndrome and G6PD cautions were already on the record. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record's title covered allogeneic intravenous MSCs in general while its grade rests on one developer's bone marrow-derived product, so the title now says so. The grade basis said nothing had replicated Ruiz 2026 (PMID 41747733); it now names the open-label umbilical cord trial (Nguyen et al., EBioMedicine, 2026; PMID 42019090) and the Zhu 2024 secondary outcomes (PMID 38679727) and explains why a different product in an unblinded design does not count, with the systematic review that could not pool the three trials (Poutouri et al., 2026; PMID 42737497). Tompkins 2017 (PMID 28977399) is relabeled: safety primary, walk gain in one dose arm only. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The protocol named 0.5 to 1 mg at 90 minutes before sleep and said a low dose was as effective as a high one. No trial tested that dose range at that lead time; the low-dose trials used 0.3 mg at 30 minutes (Zhdanova et al., 2001; PMID 11600532) and 0.5 mg at one hour (Sletten et al., 2018; PMID 29912983), and the dose-response evidence runs the other way (Cruz-Sanabria et al., 2024; PMID 38888087; Herxheimer and Petrie, 2002; PMID 12076414). The name now states the studied dose and timing, in the pattern already applied to other protocols. The endpoint type is corrected to measured function, since the trials used polysomnography and actigraphy. Added: the adult-only insomnia null (Choi et al., 2022; PMID 36179487) and the glucose-tolerance signal at 5 mg (Qian et al., 2026; PMID 42346809). C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

One B row covered sleep onset in primary sleep disorders on two pooled analyses. Read against the literature it was three claims. Ordinary adult insomnia: the pooled gain is a few minutes (Buscemi et al., 2005; PMID 16423108; Ferracioli-Oda et al., 2013; PMID 23691095) and the adult-only analysis is null (Choi et al., Sleep Med Rev, 2022; PMID 36179487), mixed evidence, C. Delayed sleep phase disorder: sleep onset earlier by polysomnography or actigraphy in four independent trials (Kayumov et al., 2001; PMID 11211063; Dahlitz et al., 1991; PMID 1674014; Mundey et al., 2005; PMID 16295212; Sletten et al., 2018; PMID 29912983) and a pooled 38.8 minutes, B, with A left to the editor because the only large trial paired melatonin with behavioral scheduling. Jet lag: nine of ten trials positive on self-rated symptoms in the Cochrane review (Herxheimer and Petrie, 2002; PMID 12076414) against the largest trial null (Spitzer et al., 1999; PMID 10484950), B at the self-report ceiling. The headline stays B on the two B uses. Brzezinski 2005 (PMID 15649737) is relabeled as 17 small studies mostly in healthy people. The antioxidant claim, with no randomized clinical evidence, is removed from the best-use line. Added to safety: 5 mg impaired glucose tolerance in MTNR1B risk-allele carriers (Qian et al., 2026; PMID 42346809) and the Cochrane case reports in epilepsy and on warfarin. The timed-release claim is now cited (Maruani et al., 2023; PMID 37434463) and limited to what the review found. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description said a 32-trial meta-analysis; Goyal et al. (JAMA Intern Med, 2014; PMID 24395196) pooled 47 trials of mindfulness meditation programs generally, not MBSR alone, found only low-strength evidence for stress, did not assess rumination or emotional regulation, and found no superiority over active treatments, none of which the record said. The uncited claim that apps have smaller effects than instructor-led MBSR is removed. Added: Khoury et al. (J Psychosom Res, 2015; PMID 25818837), 29 MBSR studies in healthy adults; Galante et al. (PLoS Med, 2021; PMID 33428616), 136 trials, with its null against specific active controls; and the TAME trial (Hoge et al., JAMA Psychiatry, 2023; PMID 36350591), MBSR noninferior to escitalopram on a blinded clinician scale. Replicated pooled self-report evidence is B, unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description presented about 2 mmHg as a general effect and carried two uncited claims (deficiency is common; red-cell magnesium is the more sensitive measure). Argeros et al. (Hypertension, 2025; PMID 41000008), 38 trials, is added: no significant effect in people with normal blood pressure, falls of 7.7 and 6.0 mmHg in treated hypertensives and people with low magnesium, no dose-response. The glycinate form remains untested on blood pressure, so D Route 1 stands; the reading says the class evidence would be B in hypertensive or magnesium-depleted adults and not in normotensive ones. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record claimed multiple randomized trials improved sleep onset while citing one pooled analysis of three small trials of other salts (Mah et al., 2021; PMID 33865376), asserted a GABA and melatonin mechanism as settled, and carried a not-replicated flag beside pooled evidence. The one randomized trial of the glycinate form (Schuster et al., Nat Sci Sleep, 2025; PMID 40918053; 155 adults, 250 mg elemental, borderline positive on insomnia severity) is now the basis, with its glycine co-dose stated; the null citrate trial (Nielsen et al., 2010; PMID 21199787), the positive older-adult trial (Abbasi et al., 2012; PMID 23853635) and the 2026 systematic review concluding the evidence does not support magnesium for insomnia (Lopresti et al., 2026; PMID 42661485) are added. The written letter attaches to the dose and form studied, so the name moves from 300 to 400 mg to the 250 mg that was tested. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description said magnesium improved insulin resistance in humans on the strength of Morais et al. (Nutrition, 2017; PMID 28526383), a narrative review of 12 trials of other salts whose own conclusion limits benefit to magnesium-deficient, insulin-resistant people. Added: Amiri et al. (BMC Nutr, 2026; PMID 42426860), 15 trials in diabetes and prediabetes, no significant change in insulin or HOMA-IR; Simental-Mendia et al. (Pharmacol Res, 2016; PMID 27329332) and Mohammadi et al. (Nutrients, 2026; PMID 42588058), modest HOMA-IR reductions. The glycinate form remains untested on this endpoint, so D Route 1 stands; the reading now says the class evidence would be C, not B. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record's D Route 1 for the olfactory use stands: every olfactory trial tested palmitoylethanolamide plus luteolin, and a combination does not grade one ingredient. Three corrections: smell tests are measured performance, not surrogates, so the endpoint type is functional; the combination's own evidence is mixed, with the independent Dresden trial finding no difference in clinically meaningful improvement (Gellrich et al., ORL, 2024; PMID 38861957) and the four-arm trial finding the combination without training no better than training with placebo (Di Stadio et al., 2023; PMID 37380908); and the one luteolin-alone placebo-controlled trial (Hodgin et al., 2021; PMID 33802381), null for Gulf War illness symptoms, is a tested null, D on Route 2 on its own row, not an absence. The search note that said all 14 records were combination products is corrected. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record held C on two pilots presented as trials, one small positive crossover, one patent-holders' Crohn's trial and a null glioma trial listed as if supportive. Three larger fibromyalgia trials were null on pain (Bested et al., Pain Rep, 2023; PMID 38226027; Due Bruun et al., Lancet Rheumatol, 2024; PMID 38258677; Rodriguez-Freire et al., Eur J Pain, 2026; PMID 42385209), the independent Crohn's trial was stopped for futility (van de Pol et al., Dig Dis Sci, 2026; PMID 42701992), the Cochrane review found clinical remission not significant (Parker et al., 2018; PMID 29607497), and the 2025 pooled analysis (Hegde et al., PMID 40540205) found more adverse events than placebo. Smith et al. 2007 (PMID 17222320) and Younger and Mackey 2009 (PMID 19453963) are relabeled as pilots, Peters et al. 2022 (PMID 35001215) as null, and the narrative review (Li et al., Int Immunopharmacol, 2018; PMID 29885638) is removed from the grade support. The Smith and Zagon patent (PMID 21380937) is stated beside the Crohn's letter. Both indications are C under the rule's own words, a replication that failed; the headline stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record typed its endpoint as self-report and called measured cognitive tests surrogates; Mori et al. (Phytotherapy Research, 2009; PMID 18844328) used a clinician-administered scale and Saitsu et al. (Biomedical Research, 2019; PMID 31413233) measured tests, which are functional outcomes under the written rule, so the type is corrected. Saitsu is restated as partial: one of three tests improved. The best-use line listed nerve growth factor, neuroplasticity, mood and nerve support, none of which any cited trial measured; it now names what was measured. The dose line led with 500 to 1,000 mg of extract, which no positive trial used; it now states the 3 g/day of powder the trial gave. Added: three null or mostly null trials in healthy young adults (PMIDs 36582308, 38004235, 40276537), the mycelium trial as a different form (Li et al., 2020; PMID 32581767) with its withdrawals for stomach discomfort, nausea and rash now in the caution, the 2025 systematic review (Menon et al., Front Nutr, 2025; PMID 40959699), and the lead author's commercial affiliation beside the letter. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The protocol, already renamed to the studied 3 g/day, cited Mori et al. (Phytotherapy Research, 2009; PMID 18844328) alone and said nothing about the lead author's affiliation with a commercial mushroom producer, now stated beside the letter. Added: Saitsu et al. (Biomedical Research, 2019; PMID 31413233), one of three cognitive tests improved in healthy older adults at a dose the abstract does not state; Li et al. (Front Aging Neurosci, 2020; PMID 32581767), the only chronic trial near 1,000 mg/day, which used erinacine-enriched mycelium in mild Alzheimer's disease; and three null or mostly null trials in healthy young adults (Grozier et al., 2022; PMID 36582308; Docherty et al., 2023; PMID 38004235; Surendran et al., 2025; PMID 40276537). C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record rested on one maker-run trial and called it the only adult trial. The same maker's larger trial (Oda et al., J Microbiol Immunol Infect, 2021; PMID 32151562; 310 adults) shortened illness by about a day and found no difference in how many people got sick; an independent 90-day trial in health workers (Navarro et al., Biometals, 2023; PMID 36474100) found no reduction in SARS-CoV-2 infection; and the pooled adult odds ratio for respiratory infection is 1.00 (Berthon et al., Adv Nutr, 2022; PMID 35481594). Graded by use: symptom scores C on the maker's trials, with the funding stated; infection prevention in adults D on Route 2. The 2023 trial's full analysis set and dropout imbalance are now stated. Headline unchanged at C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record named acute stress as its endpoint while the cited trial (McAllister et al., Stress, 2024; PMID 38975711) found no effect on any stress marker and a single positive Stroop count. The endpoint is now working memory and reaction measures under acute stress, which is what the positive trials measured: Mahoney et al. (Physiol Behav, 2007; PMID 17585971) in cold and Deijen et al. (Brain Res Bull, 1999; PMID 10230711) in combat training. Two heat-stress trials were null (Coull et al., 2016; PMID 26285023; Donnan et al., 2025; PMID 39820592), and van de Rest et al. (Nutrients, 2017; PMID 29168741) found working memory declining with dose in adults aged 60 to 75, now in the caution. No dose was stated; the trial doses are, and they are far above a capsule. C unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The use held B on COLONPREV alone (Castells et al., Lancet, 2025; PMID 40158525), with its second citation the same trial's baseline report (PMID 22356323), and the review of October owed a replication search. The search found none: CONFIRM (Dominitz et al., Am J Gastroenterol, 2017; PMID 29016565) has published only its design, and SCREESCO (Westerberg et al., Nat Med, 2026; PMID 41721101) compares each test with usual care and reports incidence, not mortality. A single trial nobody has repeated is C. The gFOBT A and the FIT-specific D Route 1 are unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The review of October asked that the trials pooled by Shu et al. (Front Nutr, 2025; PMID 41280379) be identified before the letter moved. The full text names them: the Cordyceps sinensis pools for VO2peak and ventilatory threshold rest on Parcell et al. (Int J Sport Nutr Exerc Metab, 2004; PMID 15118196), null; a 2021 Thai crossover in 12 runners, rated high risk of bias by the reviewers and published in a journal PubMed does not index, positive on VO2max; and a 2022 Brazilian marathoner trial, also unindexed, with no VO2max change. The pooled VO2peak gain is carried by the one high-risk trial. Two named null trials on VO2max, one positive small trial, and a threshold gain in older adults (Chen et al., 2010; PMID 20804368) is thin, mixed randomized evidence: C, unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The October split left the editor to confirm: glycemic markers in prediabetes or type 2 diabetes at C on small mixed trials (Houjeghani et al., 2018; PMID 29420997; Hariharan et al., 2024; PMID 38172006; Peng et al., 2020; PMID 31987255), musculoskeletal measures at D Route 2 on null trials (Saadati et al., 2024; PMID 39770949; O'Toole et al., 2026; PMID 42308284). Small positive trials with high heterogeneity and no effect on fasting glucose or insulin resistance in the pooled data is thin randomized evidence, which is C by definition; tested without benefit is D Route 2 by definition. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The decision sheet asked whether drug-resistant epilepsy in children should be A. Two independent randomized trials (Neal et al., Lancet Neurol, 2008; PMID 18456557; Lambrechts et al., Acta Neurol Scand, 2017; PMID 27027847) and a Cochrane pooling (Martin-McGill et al., 2020; PMID 32588435) agree on fewer seizures, but seizure counts come from unblinded diaries kept by families, and the written A admits three endpoint kinds only: a hard clinical event, an instrument reading, or a scale scored by a blinded clinician. A diary in an unblinded trial is none of these, so the use is B at the self-report ceiling. The headline stays C on the longevity question: a use in children with drug-resistant epilepsy does not lift the headline of a record read by adults asking about diet and aging, and the record says so. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The protocol was named for 600 mg or more at 55 percent bacosides. The pooled chronic trials (Kongkeaw et al., J Ethnopharmacol, 2014; PMID 24252493; 9 trials) used 300 to 450 mg/day, the only chronic 600 mg trial (Peth-Nui et al., 2012; PMID 23320031) used a 5 percent saponin extract, and the 600 mg advantage in Tiemtad et al. (Phytomedicine, 2026; PMID 41678913) is an indirect network comparison. The written B and C attach to the dose and form the trials used, so the name now reads Bacopa monnieri standardized extract 300 to 450 mg/day. The letter stays C: pooled gains on two measures against a null 2025 trial of 300 mg (Lopresti et al., 2025; PMID 41091332) is mixed evidence. The three options on the decision sheet collapse to this one under the rule. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record described in prose that immersion after lifting blunted strength and muscle gains (Roberts et al., J Physiol, 2015; PMID 26174323) and that two pooled analyses agree (Malta et al., Sports Med, 2021; PMID 33146851; Grgic, Eur J Sport Sci, 2023; PMID 35068365), but carried no row for it. Under the written rule a treated arm that did worse than control on the stated endpoint is D on Route 2 with the harm badge, so the finding now has its own row where a reader meets it. The headline C rests on the Cain 2025 pooling (PMID 39879231) and the Buijze 2016 trial (PMID 27631616), pooled randomized evidence, so the not-replicated flag contradicted the record and is removed; the letter is unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline C covered a borderline blood-pressure result and a performance claim the pooled trials do not support. Graded by use: blood pressure rests on two pooled analyses that disagree (Barkhidarian et al., 2019; PMID 30788274; Mirenayat et al., 2018; PMID 30284051), mixed, C; aerobic and endurance performance of free L-citrulline rests on two pooled nulls (Viribay et al., 2022; PMID 36079738; Harnden et al., 2023; PMID 37155582), D on Route 2, with the citrulline malate strength findings (Trexler et al., 2019; PMID 30895562; Wang et al., 2026; PMID 42356270) stated as a different form. Headline unchanged at C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline moved B to C in October because joint pain and skin were bundled and the evidence across them was mixed. Graded by use, as the decision sheet proposed: activity-related joint pain in athletes and active adults rests on three randomized placebo-controlled trials that agree (Clark et al., 2008; PMID 18416885; Zdzieblik et al., 2017; PMID 28177710; Kviatkovsky et al., 2023; PMID 37551682), replicated randomized evidence on self-reported pain, B at the self-report ceiling; skin rests on pooled analyses whose benefit disappears in independent and high-quality trials (Myung and Park, 2025; PMID 40324552; Pu et al., 2023; PMID 37432180), mixed, C. The headline follows the best use to B. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline C bundled a replicated instrument-measured skin result with a mixed self-reported knee result. Graded by use: skin hydration, elasticity and wrinkle depth rest on four randomized trials (PMIDs 25834304, 34203487, 38009035, 41422283) and a 7-trial meta-analysis (Amin et al., 2025; PMID 40911749), replicated randomized evidence on surrogates, B, with every trial manufacturer-run and the funding stated beside the letter under the September rule; knee symptoms rest on a trend (Tashiro et al., 2012; PMID 23226979), a null primary (Kalman et al., 2008; PMID 18208600) and a self-rating (Sugiyama et al., 2024; PMID 38234616), mixed, C. The headline follows the best use to B. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline C bundled three populations whose evidence points three ways. The written rule grades by use: in diagnosed sarcopenia, Gu et al. (Maturitas, 2025; PMID 39999663) pool five randomized trials with higher muscle mass index and handgrip strength, replicated randomized evidence on an instrument-measured outcome, B; in adults over 50 without sarcopenia the poolings disagree (Li et al., 2025; PMID 40248035; Wang et al., 2026; PMID 41934514) and the largest factorial trial was null (Osuka et al., 2021; PMID 34081113), C; in resistance-trained young men the one trial found no effect (Tritto et al., 2019; PMID 30859862), D Route 2. The headline follows the best use to B. The sarcopenia row is held at B rather than A by a stated judgment about the width of the pooled confidence intervals, which the editor may overrule. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The endothelial and inflammatory marker use in type 2 diabetes rose C to B in October on three randomized trials (PMIDs 35870947, 32015761, 40546935) plus a 2026 pooled analysis; replicated randomized evidence on surrogates is B. The review noted that two trials are Iranian academic trials whose Tabriz cohorts may overlap; independence of groups is the A requirement, not the B one, and the record states the overlap. The glycemic use is C on one null and two positive trials (PMIDs 35870947, 32472292, 32861603), mixed by definition. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The use rose C to B in October on four randomized trials that agree on higher blood NAD+ (PMIDs 36482258, 36740954, 38789831, 36797393), every one run or funded by an NMN maker, which the replication funding rule of September counts as replication and states beside the letter. Replicated randomized evidence on a surrogate is B, and the surrogate is what keeps it from A; a blood level is not a health outcome, which the record states. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline moved D to C in October to follow the monotherapy use, and the grade reading still opened 'Grade D on Route 2', so the page argued for a letter it did not carry. The reading now states the C: pooled analyses of randomized trials disagree (Limveeraprajak et al., 2024, PMID 38423354, positive; Galizia et al., Cochrane, 2016, PMID 27727432, and Peng et al., 2024, PMID 38199136, null), which is mixed randomized evidence and C by definition; the add-on and osteoarthritis uses stay D Route 2 on null trials. The 'editor to confirm C or hold D' review is closed: Route 2 requires that the trials showed no benefit, and one moderate-certainty pooling says they did. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The headline was held at D Route 1 because the record was framed on mortality, which no trial has tested, while its uses carried B (blood pressure above two drinks a day; Xin et al., Hypertension, 2001; PMID 11711507; Roerecke et al., Lancet Public Health, 2017; PMID 29253389) and C (AF recurrence; Voskoboinik et al., N Engl J Med, 2020; PMID 31893513). The site's headline letter is the best graded use on every other record, and a D on the index beside a B inside the record read as 'tested or untested' for a thing with replicated randomized evidence. The headline now follows the best use; the absence of mortality trials is stated in the reading and carries no letter. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record held C on a positive pilot (Depommier et al., Nat Med, 2019; PMID 31263284) whose 142-person scale-up (Suenaert et al., Gut Microbes, 2026; PMID 42343233) missed its primary endpoint, with a second trial null between groups (Zhang et al., Cell Metab, 2025; PMID 39879980). Under the written rule a use tested without benefit is D Route 2, whatever the pilot said. The weight-regain trial (Mount et al., Nat Med, 2026; PMID 42120725) is a different endpoint in a different population, one positive randomized trial run by the maker: C on its own row with the funding stated beside the letter. The headline follows the best use and stays C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

Depressive symptoms rest on randomized evidence that is thin: a Cochrane review that could pool two small trials and called the result inconclusive (Shaw et al., 2002; PMID 11869656), a weak pooled analysis (Javelle et al., Nutr Rev, 2020; PMID 31504850) and two small trials (Jangid et al., 2013; PMID 23380314; Meloni et al., 2020; PMID 32067288). The written C is exactly this: human randomized evidence that exists but is thin. D would require that the depression trials were null, which they were not; the null trial (Truyens et al., Gastroenterology, 2022; PMID 35940251) was in fatigue and carries that use's D Route 2. Review closed. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

Three pooled analyses of randomized trials agree that selenium lowers thyroid peroxidase antibodies: Wichman et al. (Thyroid, 2016; PMID 27702392), Huwiler et al. (Thyroid, 2024; PMID 38243784; 35 studies, moderate certainty) and Zhang et al. (Medicine, 2025; PMID 40898469). Replicated randomized evidence on a surrogate is B; the antibody titer, with no shown link to symptoms or thyroid function, is what keeps it from A, and the record states that no pooled analysis shows a clinical change. The headline, previously C on mixed cancer and cardiovascular trials, follows the best use to B; those trials keep their own rows and letters. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The one randomized trial (Riche et al., Evid Based Complement Alternat Med, 2014; PMID 25057276) raised LDL 17.1 mg/dL and lowered blood pressure 7.8/7.3 mmHg at the higher dose in the same 80 adults. The written rule grades by use: one positive randomized trial is C, and a tested harm is D Route 2, so each result gets its row. The headline keeps the harm finding, since a record whose only trial raised LDL in people enrolled for high cholesterol should lead with that. Boron's Route 2 (two null marker trials) is confirmed on the same rule without change. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The description claimed improved sleep quality and sleep onset in multiple randomized trials and named delivery apps; the only citation, Goyal et al. (JAMA Intern Med, 2014; PMID 24395196), pooled anxiety, depression and pain. A claim with no trial behind it does not stay on a graded record, so the description now states what the pooled analysis found and says sleep is not graded. The not-replicated flag contradicted the pooled evidence and is removed. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The B rests on Langade et al. (Cureus, 2019; PMID 31728244) and a five-trial meta-analysis of randomized sleep trials (Cheah et al., PLoS One, 2021; PMID 34559859). Pooled randomized evidence is replication under the written rule; what the pooling does not show is a second trial of the same extract at the same dose, which the record states and which is not what the flag means. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

Placebo-controlled replication is on the record: Zhang et al. (J Clin Endocrinol Metab, 2008; PMID 18397984), PREMOTE (PMID 33024120), Ming 2021 (PMID 34365978) and Zhao 2023 (PMID 37598753), so the flag applied only to the metformin comparison, which the reading already confines to the 36-person pilot (Yin et al., Metabolism, 2008; PMID 18442638). Flag removed. The written B attaches to the dose the trials used, and those trials gave 1.0 to 1.2 g/day where the name said 1.5 g/day, so the name now reads Berberine 1.0 to 1.2 g/day (with meals). Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record carried a not-replicated flag while its B rested on a pooled analysis of randomized trials (Goessl et al., Psychol Med, 2017; PMID 28478782). The flag describes a single trial nobody has repeated; a meta-analysis of independent randomized trials is replication by definition, and what caps this letter at B is the endpoint, not the absence of a second trial. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record carried a not-replicated flag while its B rested on a 2026 meta-analysis of randomized trials on bone mineral density (Chen et al., BMC Musculoskelet Disord, 2026; PMID 41559705). The flag describes a single trial nobody has repeated; a meta-analysis of independent randomized trials is replication by definition, and what caps this letter at B is the endpoint, not the absence of a second trial. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record carried a not-replicated flag while its B rested on a meta-analysis of randomized trials in neck pain (Chow et al., Lancet, 2009; PMID 19913903). The flag describes a single trial nobody has repeated; a meta-analysis of independent randomized trials is replication by definition, and what caps this letter at B is the endpoint, not the absence of a second trial. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record carried a not-replicated flag while its B rested on the Goyal 2014 pooled analysis (JAMA Intern Med; PMID 24395196). The flag describes a single trial nobody has repeated; a meta-analysis of independent randomized trials is replication by definition, and what caps this letter at B is the endpoint, not the absence of a second trial. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record carried a not-replicated flag while its B rested on a pooled analysis of randomized trials (Goyal et al., JAMA Intern Med, 2014; PMID 24395196). The flag describes a single trial nobody has repeated; a meta-analysis of independent randomized trials is replication by definition, and what caps this letter at B is the endpoint, not the absence of a second trial. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record carried a not-replicated flag while its B rested on a 20-trial meta-analysis (Trauer et al., Ann Intern Med, 2015; PMID 26054060). The flag describes a single trial nobody has repeated; a meta-analysis of independent randomized trials is replication by definition, and what caps this letter at B is the endpoint, not the absence of a second trial. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record carried a not-replicated flag while its B rested on a 124-study meta-analysis of LDL lowering (Ras et al., Br J Nutr, 2014; PMID 24780090). The flag describes a single trial nobody has repeated; a meta-analysis of independent randomized trials is replication by definition, and what caps this letter at B is the endpoint, not the absence of a second trial. Flag removed, letter unchanged. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record held B on Hadanny et al. (Aging, 2020; PMID 32589613), one randomized trial, with Hachmo et al. (Aging, 2020; PMID 33206062), an uncontrolled study from the same center, beside it, and had sat in the replication baseline since September as a known contradiction of Rule 03. A search in October found no independent randomized trial. A single positive trial nobody has repeated is C. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The replication search of October found four further randomized trials in which inspiratory muscle training lowered blood pressure against sham or control: Vranish et al. (Sleep, 2016; PMID 27091540) and Ramos-Barrera et al. (J Appl Physiol, 2020; PMID 32730174) in adults with sleep apnea, Ferreira et al. (Int J Cardiol, 2013; PMID 21985749) in hypertensive patients, and Ublosakka-Jones et al. (Exp Gerontol, 2018; PMID 29604403) in older adults with treated systolic hypertension. They repeat the direction and size of Craighead et al. (J Am Heart Assoc, 2021; PMID 34184544) in different populations. The written B asks for a trial supported by further trials, which this is; independence of groups is the A requirement, and blood pressure is a surrogate that caps the letter at B regardless. The flag comes off, and the record names the populations in which the finding has been repeated and the one in which it has not. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

Same evidence as the Zone 2 protocol: Gormley et al. (Med Sci Sports Exerc, 2008; PMID 18580415) replicated by the DREW trial (Church et al., JAMA, 2007; PMID 17507344) and pooled in Nuuttila et al. (Scand J Med Sci Sports, 2026; PMID 41543030). Replicated instrument-measured VO2max is A. The record keeps its statement that higher intensities produce larger gains. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The record rested on Gormley et al. (Med Sci Sports Exerc, 2008; PMID 18580415) alone. The replication search found Church et al. (JAMA, 2007; PMID 17507344), the DREW trial, 464 sedentary women trained at 50 percent of peak VO2 for 6 months with a dose arm of 136 minutes a week, and Nuuttila et al. (Scand J Med Sci Sports, 2026; PMID 41543030), 50 pooled low-intensity trials. Replicated instrument-measured VO2max is A under the rule applied to VO2max training in October, and a B beside an A for the same evidence would have graded one fact two ways. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The replication search of October found what the record already cited: Morisco et al. (Clin Investig, 1993; PMID 8241697), an independent Naples multicenter trial of 641 NYHA class III to IV patients, in which hospitalization for worsening heart failure fell from 118 to 73 over one year, with the Cochrane review (Al Saadi et al., 2021; PMID 35608922) pooling it with Q-SYMBIO (Mortensen et al., 2014; PMID 25282031) for that outcome. The written B requires a trial supported by further trials rather than standing alone, which this meets; it does not require independent replication of the pivotal endpoint, which is the A requirement. The flag said the finding was never repeated, which is false for hospitalization and true for mortality, so the flag comes off and the use is named for the events the two trials share, with the mortality result stated as single-trial in the reading. Applied the written grade definition under the editor's standing instruction of October 2026 (Arnold Sandlow, D.C.).

The use was built at C in October and the editor corrected it the same night: the Cochrane pooling (Avenell et al., 2014; PMID 24729336) is high-quality evidence of a hip-fracture reduction (RR 0.84) across nine randomized trials, and Chapuy et al. (N Engl J Med, 1992; PMID 1331788) is a 3,270-patient trial on the same endpoint, so the combination has replicated randomized evidence on a hard outcome, which is B. The null community trials (Jackson et al., N Engl J Med, 2006; PMID 16481635; Grant et al., Lancet, 2005; PMID 15885294) keep the scope to institutionalized or deficient older adults and keep it from A. The record's headline follows its best use to B. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The protocol held B on one trial, Q-SYMBIO (Mortensen et al., JACC Heart Fail, 2014; PMID 25282031; 420 patients, ubiquinone 100 mg three times daily), with the not-replicated flag set, and was named for ubiquinol, a form no cited trial gave. The pooled analysis (Xu et al., BMC Cardiovasc Disord, 2024; PMID 39462324) rests on the same trial for mortality. Under the single-trial rule that is C, and the name now states the form and dose studied, CoQ10 (ubiquinone) 300 mg/day. The CoQ10 supplement record's heart-failure use is held at B pending a replication search. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

Weight loss rested on one phase 2 trial, Jastreboff et al. (N Engl J Med, 2023; PMID 37366315; 338 adults), with the not-replicated flag accurately set, and the first phase 3, Bajaj et al. (Lancet, 2026; PMID 42250575), reports glycemic and anthropometric endpoints in a different population. The written rule says a single positive trial nobody has repeated is C, and the compound is not approved anywhere. Moved to C; it returns to B when a second peer-reviewed obesity trial reports. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The C rested on Cenacchi et al. (Aging (Milano), 1993; PMID 8323999), a trial of bovine-cortex phosphatidylserine, a preparation no longer sold, while the soy-derived form on the shelf had two null trials and a protocol at D Route 2. The search for this decision found a second positive bovine trial, Crook et al. (Neurology, 1991; PMID 2027477; 149 adults), so the bovine evidence is replicated; it stays C because the written B attaches to a form that can be obtained. Decided: the record follows the evidence for each form. The headline C is stated for the bovine-cortex preparation and says it is not sold; the soy-derived form is its own use at D Route 2 on Jorissen et al. (Nutr Neurosci, 2001; PMID 11842880) and Kato-Kataoka et al. (J Clin Biochem Nutr, 2010; PMID 21103034). The review opened October is closed. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The ward row disclosed four intensive-care poolings that split two null and two positive, and the question put to the editor was whether to add an intensive-care use. The trials point in opposite directions by unit, so one combined row would misstate both. General intensive care: the two largest trials are null, Pro-MEDIC (Wibrow et al., Intensive Care Med, 2022; PMID 35220473; 847 patients) and DEMEL (Mekontso Dessap et al., Intensive Care Med, 2025; PMID 40608082; 355 ventilated patients), as is the lowest-bias pooling (Lakbar et al., Intensive Care Med, 2025; PMID 41051561); D on Route 2. Cardiac surgery: small trials and a pooling are positive (Javaherforoosh Zadeh et al., Iran J Med Sci, 2021; PMID 33753956; Han et al., Front Cardiovasc Med, 2022; PMID 35665270) against the largest single trial, Healthy Heart-Mind (Ford et al., J Am Geriatr Soc, 2020; PMID 31595489; 210 patients), which found no difference; C. The ward row's review is closed. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The protocol was named R-ALA 300 to 600 mg fasted. Every cited trial gave racemic alpha-lipoic acid with no fasting instruction: Ansar et al. (Saudi Med J, 2011; PMID 21666939), 300 mg a day; Ziegler et al. (Diabetes Care, 2006; PMID 17065669, SYDNEY 2), 600 to 1,800 mg a day; Ziegler et al. (Diabetes Care, 2011; PMID 21775755, NATHAN 1), 600 mg a day for four years; and Han et al. (Eur J Endocrinol, 2012; PMID 22837391) pooled intravenous trials. The name now reads Alpha-Lipoic Acid (racemic) 300 to 600 mg/day, and the tracking note no longer asserts an R-form equivalence no trial measured. Letter unchanged at C; the review opened October is closed. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The protocol was named for 200 mg taken 30 to 60 minutes before sleep. Hidese et al. (Nutrients, 2019; PMID 31623400) gave 200 mg a day for four weeks with no stated time, and the Bulman et al. meta-analysis (Sleep Med Rev, 2025; PMID 40056718) pools trials of varied timing, mostly inside combination products. No trial tested the bedtime timing in the name, so the name now states the dose studied, L-Theanine 200 mg/day. Letter unchanged at C. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The fracture row graded vitamin D3 alone (D Route 2) while citing Chapuy et al. (N Engl J Med, 1992; PMID 1331788), a trial of vitamin D3 with calcium. The combination is a different intervention with its own evidence: positive in 3,270 institutionalized women (Chapuy 1992) and in the Cochrane pooling (Avenell et al., 2014; PMID 24729336; hip fracture RR 0.84, benefit concentrated in institutions), null in the Women's Health Initiative (Jackson et al., N Engl J Med, 2006; PMID 16481635) and RECORD (Grant et al., Lancet, 2005; PMID 15885294) outside institutions. Mixed by population: C, scoped to institutionalized or deficient older adults. Added because a reader asking about D3 with calcium was being answered with the D3-alone result. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The protocol is named for the common retail pairing, which no trial has tested; the nearest combination trial, AVADEC (Diederichsen et al., Circulation, 2022; PMID 35465686), used MK-7 720 mcg with D3 1,000 IU and was null on its primary endpoint. Decided: keep the name readers search for and open the record with the statement that these doses are untested, rather than rename to doses no reader buys. The citation_incomplete review is closed on that basis; the letter stays C. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The question put to the editor was whether the C should rest on independent trials only or on all of them with the investigator affiliation stated. Decided: as built. Schneider et al. (Circ Cardiovasc Qual Outcomes, 2012; PMID 23149426) and Norris et al. (Front Med (Lausanne), 2025; PMID 40177279) both count, the Maharishi affiliation of the investigators is stated in the record, and the Cochrane extraction (Rees et al., 2024; PMID 38358047) showing no clear difference stands beside the positive trial. Letter unchanged. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The record was held at B while its own ref said the written rule allowed A. The rule names measured VO2max as a qualifying endpoint, and the gain from structured aerobic training is replicated by independent groups: Gormley et al. (Med Sci Sports Exerc, 2008; PMID 18580415), Church et al. (JAMA, 2007; PMID 17507344, the DREW trial) and Helgerud et al. (Med Sci Sports Exerc, 2007; PMID 17414804), with three pooled analyses in agreement (An et al., 2024, PMID 38878596; Nuuttila et al., 2026, PMID 41543030; Mattioni Maturana et al., 2026, PMID 42730846). The earlier hold rested on the view that a fitness measure is not a health outcome; the rule makes no such distinction and was written so that an instrument-recorded function could reach A without a mortality trial. The grade reading states that the trials tested structured aerobic training at either intensity, with intervals giving the larger gain for the same work, so the A is not read as a claim about Zone 2 alone. The mortality association (Kodama et al., JAMA, 2009; PMID 19454641) stays observational context. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The ref said Gonzales 2017 (PMID 28940638) showed exercise-performance effects; it measured femoral blood flow during calf exercise in 25 older adults, with a benefit only in men that lost significance after adjustment, and is relabeled as a vascular surrogate study. Mirenayat 2018 (PMID 30284051), null on brachial and aortic blood pressure, is added beside Barkhidarian 2019 (PMID 30788274), whose diastolic effect appeared only at 6 g/day or more and whose systolic interval was borderline. Viribay 2022 (PMID 36079738) and Harnden 2023 (PMID 37155582), null on aerobic and endurance performance, are added; Trexler 2019 (PMID 30895562) and Wang 2026 (PMID 42356270) show the small strength effects rest largely on citrulline malate, now stated as a different form. Letter unchanged; a split into blood pressure and performance rows is on the decision sheet.

The record carried one C across every use while its own umbrella review, Patikorn 2023 (PMID 37231411), rated seizure reduction as high-quality evidence. Drug-resistant epilepsy in children and adolescents is now its own use: Neal 2008 (PMID 18456557) and Lambrechts 2017 (PMID 27027847), two independent positive randomized trials, and Martin-McGill 2020 (PMID 32588435), the Cochrane pool; seizure diaries are unblinded self-report, so B. Schoeler 2023 (PMID 37977712), KIWE, is added so the use excludes infants and the comparison with a further drug. The LDL harm now cites Wang 2024 (PMID 39097343) and Chang 2026 (PMID 41621647), and Żurawski 2026 (PMID 42588148) states the matched-energy weight difference. The description names the epilepsy population. Headline letter unchanged; whether the epilepsy use should be A is on the decision sheet.

The ref said 'Grade C on consistent RCT 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. evidence limited to cardiometabolic surrogates', which under the site's rule describes a B. The letter is C because against daily calorie restriction the fasting pattern adds nothing, and the record now says so with the trials: Trepanowski 2017 (PMID 28459931), one year, weight loss equal to daily restriction, LDL 11.5 mg/dL higher and 38 percent dropout in the fasting arm; Schübel 2018 (PMID 30475957), HELENA, 50 weeks, equivalent not superior; Semnani-Azad 2025 (PMID 40533200), 99 trials, an edge only in short trials; Garegnani 2026 (PMID 41692034), Cochrane, little to no difference from regular dietary advice. Cioffi 2018 (PMID 30583725) is stated as pooling 8 to 24 week trials only. Kord Varkaneh 2022 (PMID 35958257) is relabeled as a small open trial in fatty liver disease against usual diet, with its null lipid and glycemic results. Letter unchanged.

The ref said Tashiro 2012 (PMID 23226979) improved knee osteoarthritis symptoms; both groups improved, hyaluronic acid showed a trend overall and significance only in those aged 70 or under, and the record now says so. Kalman 2008 (PMID 18208600), null on its primary pain endpoint, is added. Kawada 2014 (PMID 25014997) is a review and is relabeled as background. The best-use line claimed joint lubrication and cartilage health and gut-barrier support; no trial measured cartilage or lubrication and no human trial tested the gut, so both are removed. Added on skin: Hsu 2021 (PMID 34203487), Dolečková 2025 (PMID 41422283), the largest trial, and Amin 2025 (PMID 40911749), the 7-trial meta-analysis; on knees, Sugiyama 2024 (PMID 38234616), self-rated improvement without objective change. The dose line now states each trial's dose and that one often-cited joint trial used a multi-ingredient preparation. Every trial's manufacturer tie is in the funding block. Letter unchanged; splitting skin from joints is on the decision sheet.

The grade basis listed excess coronary heart disease as a finding; the 2002 interim 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. was only nominally significant and the final adjudicated result, Manson 2013 (PMID 24084921), was 1.18 (95% CI 0.95 to 1.45), which the 2026 Cochrane review, Bofill Rodriguez (PMID 42695385), rates as probably little to no difference. The firm harms, invasive breast cancer, stroke, venous thromboembolism, dementia in women 65 and over and gallbladder disease, are now what the record names. Manson 2017 (PMID 28898378) is added so the record says all-cause mortality was not changed during the trial or over 18 years, which the earlier text could be read as implying. Gartlehner 2022 (PMID 36318128), the USPSTF review, is added. Rossouw 2002 (PMID 12117397) is relabeled as the interim report. Whether the manufacturer supplied study drug is not asserted, because the full text was not read. Letter unchanged.

The only citation, Prado 2022 (PMID 35301826), is a vote-counting review of patients with active cancer in which 73 percent of studies gave HMB combined with amino acids, with authors employed by or paid by a maker of HMB-containing oral nutritional supplements; the ref attributed to it findings in older and trained adults it never studied. It is relabeled. The record now rests on the population-specific evidence: Li 2025 (PMID 40248035), small gains in adults over 50; Wang 2026 (PMID 41934514), no added benefit with resistance training; Osuka 2021 (PMID 34081113), null primary in older women with low muscle mass; Gu 2025 (PMID 39999663), small gains in diagnosed sarcopenia; Tritto 2019 (PMID 30859862), null in trained men. Rathmacher 2020 (PMID 32857128) is context as a manufacturer-authored combination trial. Dose and form are stated. Letter unchanged; a split by population is on the decision sheet.

The record said CRP was null; that was true of Lorzadeh 2019 (PMID 30991044), but Huang 2024 (PMID 39279783), Ouyang 2026 (PMID 41883255) and Zhu 2026 (PMID 42434414) each pool a modest CRP reduction, so the inflammatory evidence is restated. The VCAM-1 effect came from Salden 2016 (PMID 27797708), where it was a trend overall and significant only in an unadjusted exploratory subgroup; the record now says so, and that the pool mixes hesperidin 2S and purified hesperidin at different doses in different populations. Gilcher 2026 (PMID 42605853), null on flow-mediated dilation at a dietary dose, is added. Letter unchanged.

The record held D Route 1 while linking Yau et al. (J Alzheimers Dis, 2024; PMID 39544014), a randomized double-blind placebo-controlled pilot in 19 adults aged 60 or over with mild cognitive impairment, 25 mg three times a week for a year, in which Rey Auditory Verbal Learning Test performance improved and plasma neurofilament light chain stayed stable against placebo. Under the written rule a single positive randomized trial is C, and holding D while citing the trial read as hedging. The C is scoped to that population and endpoint; the pilot size, the unstated funder and the absence of a second trial are the reasons it is not more. Cheah et al. (Antioxid Redox Signal, 2017; PMID 27488221), the human uptake study, and the observational papers by Cheah 2016 (PMID 27444382), Wu 2022 (PMID 36139790) and Wu 2021 (PMID 34673145) remain context. Moved October 2026 by editor adjudication (Arnold Sandlow, D.C.), closing the review opened October.

The hair row held a single D Route 1 while disclosing Amini et al. (Life (Basel), 2025; PMID 40141845), a randomized saline-controlled trial in 20 men whose plant-extract exosome formulation raised hair count against saline. Under the written rule a single positive randomized trial is C, but a C on a row named for exosome therapy would read as a C for the human stem-cell-derived products clinics sell, which no randomized trial has tested. The row is therefore split: the plant-derived formulation tested is C on one trial, and human stem-cell-derived exosomes for hair stay D Route 1 with their own absence check. The review hold on the hair row is closed. Decided October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The C rested on the view that the human evidence was pilot-scale. It is not: Wei et al. (Sci Transl Med, 2017; PMID 28202779) randomized 100 adults to three cycles, Micarelli et al. (Cell Rep Med, 2025; PMID 39970875) randomized 102 in crossover to six, and Mohammadzadeh et al. (Diabetol Metab Syndr, 2025; PMID 40287774) pooled 11 randomized trials, agreeing on lower blood pressure, HbA1c and IGF-1 and on no effect elsewhere. Replicated randomized evidence on validated surrogates is B under the written rule. The letter is scoped: against a Mediterranean diet, Mishra et al. (NPJ Metab Health Dis, 2023; PMID 40604264) found no between-group difference, and against continuous energy restriction Sadeghian et al. (Obes Surg, 2021; PMID 33420673) found no weight difference, so the B is for the comparison with an unrestricted diet only. Every positive trial used the maker's kit with the equity-holding inventor as coauthor; under the replication funding rule of September those trials count and are disclosed beside the letter. Moved October 2026 by editor adjudication (Arnold Sandlow, D.C.).

The ref said Zuñiga 2017 (PMID 28459647) affected glycemic measures; it did not change the metabolic syndrome components, insulin secretion or insulin sensitivity, and was positive only for weight, BMI and VLDL, so it is now labeled null on glycemic outcomes. The positive glycemic data come from Gaytán Martínez 2021 (PMID 32460589), 2-hour glucose and HbA1c lower in 30 adults with impaired glucose tolerance, and Zamani 2023 (PMID 36580574), a six-trial meta-analysis its authors rate low quality; both are added and the best-use line now rests on them. The record now names the populations studied, none of them healthy adults. Shiyovich 2010 (PMID 20856101), toxic hepatitis attributed to Gymnema, is added to the avoid field. Letter unchanged.

The healthy-older-adults use called Lizzo 2022 (PMID 35821844) an independent trial; its paper discloses that the sponsor holds and licenses patents on glycine plus N-acetylcysteine and markets the products, so it is a maker-funded null, which strengthens rather than weakens it, and its 2.4 to 7.2 g/day for 2 weeks is now set against the originating group's roughly 14 g/day for 16 weeks. The other use said 'randomized trials' in the plural; only Kumar 2023 (PMID 35975308) is randomized and placebo-controlled, the 2021 study (PMID 33783984) is an open-label pilot, and the note now says so. The 'populations with elevated demand' label described a post hoc subgroup from the sponsor trial rather than how Kumar 2023 enrolled people, so the use and title now name older adults. The unverifiable 'BMI above 27' entry criterion is removed. The 'sponsored' flag is replaced by a funding_basis block naming who sponsored which trial. Letters unchanged; the D versus C question on the healthy older adults use is on the decision sheet.

The record rested on Richie 2015 (PMID 24791752) alone and called the grade 'one small RCT'; Allen and Bradley 2011 (PMID 21875351), NIH-funded, found no change in glutathione status or oxidative stress markers at 1,000 mg/day, and Søndergård 2021 (PMID 33740389) found no change in oxidative stress markers, so the grade reason is now mixed small trials and the best-use line no longer lists oxidative stress markers as a supported use. Sinha 2018 (PMID 28853742) is cited for form only: a 12-person uncontrolled pilot of liposomal glutathione funded by a company that sells it. Richie 2015's funder is not in its indexed record and an ingredient maker lists the trial as its own, so the record does not call it independent. Letter unchanged.

The record said the primary human trial of the statin-muscle claim is ongoing; no registered trial was located in this pass, and the claim is withdrawn rather than left unsourced. Osaki 2026 (PMID 42323079) is now described as a mouse genetic model, not as establishing the human mechanism. Added as context, scoped away from the statin claim: D'Adamo 2026 (PMID 42807157), 150 healthy adults, GG alone null on plasma CoQ10 and questionnaires, first author on the advisory board of a company that sells GG; Jurgelewicz 2026 (PMID 42222203), a fixed-order placebo-then-GG study null on testosterone overall, one author employed by the sponsor. The 'no human trial' language is now scoped to the statin-muscle claim. Letter unchanged.

The record was named for serial multi-biomarker screening while its own scope note said the Cochrane trials tested health checks of their era; the name now says multi-organ screening of asymptomatic adults. Added: Jørgensen 2014 (PMID 24912589), Inter99, 59,616 people, null on ischemic heart disease, stroke and mortality at 10 years; Wang 2026 (PMID 42414985), a systematic review of 22 randomized trials through 2025 confirming no mortality or cardiovascular benefit; both beside Krogsbøll 2019 (PMID 30699470). The confidence intervals now read 'to' rather than a dash. Letter unchanged.

The ref said Guimarães 2024 (PMID 38321713) affected stress measures; its abstract reports sleep quality, negative affect, depression scores and heart-rate variability, with no stress result, in 30 overweight women who were also starting an exercise program, and the record now says so. Added: Hepsomali 2020 (PMID 33041752), the systematic review of 14 placebo-controlled trials that rates the stress evidence limited and the sleep evidence very limited; Byun 2018 (PMID 29856155), the one trial with polysomnography, reporting within-group change only. Letter unchanged.

The record said Liu 2025 (PMID 39645105) showed lipid and glucose effects; its abstract reports a smaller rise in glycated serum protein and lower inflammatory markers with no lipid result, and the ref now says so. Added: Wright 2019 (PMID 30312135), a maker-funded placebo-controlled trial null on insulin resistance, lipids, HbA1c and glucose at 1 g/day for 90 days; Hernández-Corona 2014 (PMID 24611906), where HOMA insulin resistance rose on fucoidan; Negishi 2013 (PMID 24005608), the influenza-vaccine antibody trial that carries the immune half of the best-use line. The record now states that fucoidan products differ by seaweed source and molecular weight, so trials of different extracts do not replicate each other. Letter unchanged.

The stroke use was labeled 'folate-deficient'; CSPPT (Huo 2015, PMID 25771069) enrolled hypertensive adults without prior stroke or MI, not people selected for deficiency, gave 0.8 mg/day with enalapril in both arms, and Zhang 2024 (PMID 38824900) found efficacy did not vary by baseline folate, so the label now names the population, the dose and the fortification condition. Yang 2024 (PMID 39155933), the primary-prevention meta-analysis that is null overall and positive only without fortification, now sits beside the positive one. The headline's 'not-replicated' flag is removed: Wu 2009 (PMID 19864409) and Logan 2008, ukCAP (PMID 18022173), replicated the null of Cole 2007 (PMID 17551129) and Figueiredo 2011 (PMID 21170989) pooled all three at RR 0.98. Added: Passarelli 2019 (PMID 31401653), more sessile serrated lesions with extended use, as a safety note; Vollset 2013 (PMID 23352552), no overall cancer increase across 49,621 people; Gao 2013 (PMID 23682073), the open trial on initial occurrence, a different question. The maker's support for CSPPT reported in secondary sources is not asserted, because the JAMA full text was not read. Letters unchanged.

The October adjudication had already scoped the 'no human trial' statement to the senolytic use. Remaining from the review: the dose line gave a 100 to 200 mg/day anti-aging maintenance dose that no trial in healthy or aging adults supports; it is removed, and the doses actually used, 100 mg/day in colorectal cancer (Farsad-Naeimi 2018, PMID 29541713) and 200 mg/day in obese men (Alipour 2026, PMID 42218768 and PMID 41683255), are stated as other uses. The form note claimed liposomal fisetin improves absorption with no citation; the one human bioavailability trial, Krishnakumar 2022 (PMID 36304817), tested a hydrogel formulation, and the note now says so. Yousefzadeh 2018 (PMID 30279143) is labeled mouse data only. Letter unchanged.

The description called the human evidence pilot-scale; Wei 2017 (PMID 28202779) randomized 100 adults, Micarelli 2025 (PMID 39970875) 102 in crossover, van den Burg 2024 (PMID 38546821) 100 for 12 months, and Mohammadzadeh 2025 (PMID 40287774) pooled 11 trials; corrected. Wei 2017's randomized comparison showed weight, fat, blood pressure and IGF-1 only, with the glucose, lipid and CRP results post hoc in at-risk participants; the ref now says so. Espinoza (PMID 41372565) is dated 2026 as PubMed does, with 30 people, 8 days and differences not significant at all time points. Added: Mishra 2023 (PMID 40604264), no difference from a Mediterranean diet, and Sadeghian 2021 (PMID 33420673), no weight difference from continuous energy restriction, as the nulls; Brandhorst 2024 (PMID 38378685) as the inventor group's biological-age analysis. The type 2 diabetes trial (PMID 38546821), co-funded by the maker, is a separate C use rather than support for healthy adults. The kit's product name is no longer on the record. Headline letter unchanged; B versus C is on the decision sheet.

The headline said functional, the skin row said self-report and its note said surrogate; instrument readings are surrogate, and all three now say so. 'One randomized split-face trial' is corrected: Choi 2026 (PMID 42403177) repeated Park 2023 (PMID 37377400) with the same product, funded and supplied by its maker, ExoCoBio, from the same group, so it is not independent replication, and Yen 2026 (PMID 42635327) and Park and Park 2026 (PMID 41752855) are independent but unblinded trials of different products. Huang 2026 (PMID 41800727) is described as n=6 in the ref as well as the note. Stack 2026 (PMID 41800726), which pooled mostly uncontrolled studies and contains no knee or systemic trial, no longer sits under the hair, knee or systemic rows. The hair row's search note now says Amini 2025 (PMID 40141845) lacks the publication-type tag the query relied on, and the knee row's Bolandnazar 2024 (PMID 39465400) is described in full. Letters unchanged; the hair hold is on the decision sheet.

The October adjudication had already corrected the 'no human randomized trial' statement for Yau 2024 (PMID 39544014). Remaining from the review: the dose line said 5 to 30 mg/day with no trial behind it; it now states the one efficacy trial's 25 mg three times a week and the uptake study's 5 or 25 mg/day. Cheah 2017 (PMID 27488221), the first human administration of pure ergothioneine, is added with its mostly non-significant biomarker trends, which is why no antioxidant claim is made. The endpoint type moves from surrogate to functional because the one trial's outcome is a measured cognitive test. The form note says whole-mushroom trials do not support purified ergothioneine. The three observational papers, Cheah 2016 (PMID 27444382), Wu 2022 (PMID 36139790) and Wu 2021 (PMID 34673145), are labeled association only. Letter unchanged, still under review.

The best-use line claimed anti-inflammatory and antioxidant benefit and fasting glucose with no cited outcome; the first two are removed and glucose now rests on Zheng 2013 (PMID 23426037), a 1.48 mg/dL pooled drop with no HbA1c change. Nagao 2005 (PMID 15640470) was run by researchers at the company that makes the catechin beverage it tested, measured body fat in 35 men and did not measure LDL cholesterol or test an extract; relabeled as context. Hartley 2013 (PMID 23780706) now says what it found: no trial reported cardiovascular events and the green tea results were unstable on sensitivity analysis. Xu 2020 (PMID 32434539), already cited, is now described. Peng 2014 (PMID 25176280) is now stated to show its blood pressure effect only in low-dose subgroups. Added: Samavat 2016 (PMID 27806972), the largest extract trial at the record's dose, LDL down about 4 percent and triglycerides up; Dostal 2015 (PMID 26051348), its safety report with ALT elevations in 6.7 percent against 0.7 percent, now in the avoid field; Kim 2011 (PMID 22027055) and Momose 2016 (PMID 27324590) as independent confirmations of the LDL effect; Igho-Osagie 2020 (PMID 33188386), null on lipids and blood pressure in short trials. Letter unchanged.

The October adjudication had already moved this record to D Route 2 on the blinded and sham-controlled evidence. Remaining from the review: van Dongen-Boomsma 2013 (PMID 24021501), the independent double-blind placebo-neurofeedback trial in 41 children, was named in the search record but not cited; it is now on the page with its null result and its chance-level guess rates. Mei 2026 (PMID 42358407) is now described as a network meta-analysis of 37 randomized trials of digital interventions whose neurofeedback ranking rests on unblinded ratings, not as a systematic review. The evidence note names the graded population, children and adolescents, and says that nothing on the page implies benefit for adults or consumer headsets. The sham-controlled trial (PMID 32853703; follow-up PMID 36521694) and the two meta-analyses, Cortese 2016 (PMID 27238063) and Westwood 2025 (PMID 39661381), stand. Letter unchanged.

Thomson 2017 (PMID 28560655) now carries what its full text reports: the primary endpoint was originally breast density and was changed to the urinary 2/16 ratio on the monitoring board's advice after recruitment problems, the original primary was null, 98 of 130 randomized women completed, and the active and placebo capsules were supplied by the maker of the studied product while the NCI funded the work. The best-use line called the ratio change a benefit; no trial has linked it to a cancer outcome, and the line now says so. Castañon 2012 (PMID 22075942), the largest DIM trial, null on cervical cytology and HPV, Dalessandri 2004 (PMID 15623462) and Godínez-Martínez 2023 (PMID 36111381), where 75 mg/day did not raise the ratio, are added. The formulation's maker is no longer named on the record. The tamoxifen-metabolite signal stays in the avoid and drug-conflict fields the page renders. Letter unchanged.

The record described DASH-Sodium as a crossover of three sodium levels within the DASH pattern; Sacks 2001 (PMID 11136953) randomized participants in parallel to a control diet or DASH and crossed sodium levels within each, and the control arm is now stated. Appel 1997 (PMID 9099655) enrolled adults with systolic pressure under 160 and diastolic 80 to 95, with a larger effect in the hypertensive subgroup; both now appear. The B's replication requirement is now met on the page by Saneei 2014 (PMID 25149893) and Filippou 2020 (PMID 32330233), meta-analyses by other groups, with Juraschek 2026 (PMID 41206973) on durability in a community setting. The 'no trial reported events' statement now cites Bensaaud 2025 (PMID 40326569), the Cochrane review that found almost no events across five small trials. Letter unchanged.

The record cited Candow 2015 (PMID 25993883) for creatine plus training on cognition; that trial measured lean mass and strength only and is now labeled as such. Avgerinos 2018 (PMID 29704637) is a systematic review of six trials without pooling, not a trial, and is relabeled. Prokopidis 2023 (PMID 35984306) now carries its own declaration, an author on the advisory board of a creatine manufacturer and another who advised a creatine seller, and the letter by Eckert and Pascher 2023 (PMID 36644917) alleging double counting, with the authors' reply (PMID 36644912). Added at the record's own dose: Sandkühler 2023 (PMID 37968687), the largest preregistered trial, null on both primary tests with side effects more frequent on creatine (RR 4.25), now a safety note; Alves 2013 (PMID 24098469), 24 weeks in older women, null. Rawson 2008 (PMID 18579168) is null in young adults. McMorris 2007 (PMID 17828627), the older-adult positive trial, used 20 g/day for one week, so the description no longer implies the older-adult effect was shown at 3 to 5 g. Xu 2024 (PMID 39070254) is added as the later meta-analysis, pooling patients with healthy adults. Letter unchanged.

The record called three meta-analyses independent replications; Murlasits, Kneffel and Thalib 2018 (PMID 28783467), Eddens, van Someren and Howatson 2018 (PMID 28917030) and Gao and Yu 2023 (PMID 36776981) pool largely the same small trials, and the text now says so, with the third author of each of the first two named and Gao 2023's corrigendum (PMID 37215175) attached. Held 2026 (PMID 41762427), an umbrella review of 17 meta-analyses finding no significant sequence effect, and Wilhelm 2014 (PMID 25449853), a null comparison in older men, are added as the counterweight. Letter unchanged: B on consistent small pooled effects, not A.

The dose line said trials dosed collagen alongside vitamin C; only Shaw 2017 (PMID 27852613) did, and that was an 8-man, 3-day study of vitamin C-enriched gelatin measuring a synthesis marker, now labeled mechanistic context. Clark 2008 (PMID 18416885) used 10 g and Zdzieblik 2017 (PMID 28177710) 5 g without vitamin C; corrected. Choi 2019 (PMID 30681787) is a qualitative review, not a meta-analysis, and is relabeled; the meta-analysis flag now rests on Myung and Park 2025 (PMID 40324552), null in independent and high-quality skin trials, and Pu 2023 (PMID 37432180), positive with biases noted and no difference by collagen source, which also removes the unsupported marine-versus-bovine preference. Kviatkovsky 2023 (PMID 37551682), a 9-month trial with an NIH training grant, is added on joint pain. König 2018 (PMID 29337906), a 12-month bone-density trial, is noted and not graded. Letter unchanged; the joint-pain split is on the decision sheet.

The endpoint line listed inflammation and mood; Cain 2025 (PMID 39879231) found inflammation markers rose for an hour after immersion and mood did not change, so the headline endpoint now reads stress at 12 hours, sleep quality and quality of life, and mood and inflammation are each a D Route 2 use. 'Few small RCTs' is corrected: one trial, Buijze 2016 (PMID 27631616), carries 3,018 of the 3,177 pooled participants and is now cited and graded as a C use on self-reported sickness absence. Roberts 2015 (PMID 26174323) showed smaller strength and muscle gains, not only blunted signaling; Grgic 2023 (PMID 35068365) and Malta 2021 (PMID 33146851) replicate it, and the three now sit together as a harm signal for immersion after lifting. 'Signalling' is now spelled signaling. Headline letter unchanged; whether the headline should be split further is for the editor.

The dose line said most trials used 250 to 500 mg; the trials cited used 1,000 mg/day (Cohen 2003, PMID 12865605; the IDEALE study, Cotroneo 2013, PMID 23403474), 1,000 then 2,000 mg/day (Spiers 1996, PMID 8624220) or 500 mg/day, and the Cochrane review, Fioravanti and Yanagi 2005 (PMID 15846601), reports heterogeneous doses and routes; corrected. The Cochrane review's declared conflict, that its principal reviewer was an investigator in some included trials, and its finding of no benefit on attention are now stated. IDEALE, an open study with a non-randomized control group, is linked as context rather than listed as evidence, and the 'observational' flag that described it is gone. Added: Bonvicini 2023 (PMID 36678257), a positive meta-analysis whose authors rate the studies poor, and Spiers 1996 (PMID 8624220), where the memory benefit appeared only in a subgroup picked after analysis. Letter unchanged.

The record was named 'Choline' with a generic use line while its only evidence was Caudill 2018 (PMID 29217669): 26 mothers, 930 versus 480 mg/day, no placebo, infant reaction time. The population, dose, comparator and n are now stated, with the egg and beef industry part-funding the paper discloses. The larger placebo-controlled trial, Cheatham 2012 (PMID 23134891), was null and is cited, with the 7-year follow-up Bahnfleth 2022 (PMID 34962672) and the Gould 2025 (PMID 40077755) review that finds the evidence insufficient. Adult use is stated as ungraded, with Deuster 2002 (PMID 12502178) null. Letter unchanged.

The ref described Saadati 2024 (PMID 39770949) as a trial of carnosine 'on musculoskeletal and metabolic measures' without saying it was null on all of them, and the best-use line named musculoskeletal measures as the use. That use is now D Route 2, with O'Toole 2026 (PMID 42308284) beside it. The glycemic evidence that is the only basis for a C is now cited: Houjeghani 2018 (PMID 29420997), Hariharan 2024 (PMID 38172006), Peng 2020 (PMID 31987255). The nulls from the same parent trial on vascular, lipid, organ and inflammatory outcomes (PMIDs 38004228, 39599686) are stated, the population (prediabetes and type 2 diabetes, never healthy adults) is stated, and the pooled analyses that mix beta-alanine (PMIDs 34333586, 40999397) are labeled. Headline C held pending editor confirmation.

The record called Waziry 2023 (PMID 37118425) a pre-specified analysis; the paper says post hoc, and it also reports that PhenoAge, GrimAge and other clocks did not change, now stated. Telomere length, Hastings 2024 (PMID 38504468), no net change, added. The trial randomized 218, not 220, and achieved about 12 percent restriction against the 25 percent target, now stated. The safety signal was missing: bone density fell at spine and hip (Romashkan 2016, PMID 26992237; Villareal 2016, PMID 26332798). Cardiometabolic improvement is now sourced to Kraus 2019 (PMID 31303390), and Racette 2017 (PMID 29045325) shows absolute fitness and strength fell. The DunedinPACE An epigenetic test that estimates how fast a person is aging now, in years of biological aging per calendar year. inventors' commercial license is disclosed. Letter unchanged.

The best-use line called the Demidenko 2021 report (PMID 34847066) an analysis of 'self-reported users'; it measured DNA methylation age in 42 customers, retrospectively, without controls, on a multi-ingredient product, and was authored in part from the company that sells it. All of that is now stated. The dose line presented a commercial product's twice-daily schedule as if trial-derived; removed, leaving the 1 g dose the ABLE trial (Lim 2025, PMID 40819772) tests. The endpoint type is now surrogate, since the planned human readout is a methylation clock. Letter unchanged.

The record said the animal data 'disagree with itself'; now the animal studies disagree with each other, and the third one is cited: Shytikov 2021 (PMID 33849306), in which C60-in-oil mice outlived oil-fed but not water-fed controls, pointing at the olive oil. Baati 2012 (PMID 22498298) is labeled small and not reproduced. Letter unchanged.

The record said no human randomized trial 'has been run'; the developer group's review, Sikiric 2012 (PMID 22300085), states a phase II trial in inflammatory bowel disease was run, with no result indexed, so the record now says no result has been published. The dose line omitted the route used in the interstitial cystitis pilot, Lee 2024 (PMID 39325560): injection around the bladder lesion at cystoscopy, 10 mg; added. 'One clinic' is restated as what the abstracts support: one lead author, one named clinic and two unnamed private-clinic reports. Letter unchanged.

The ref said the meta-analysis Hajihashemi 2024 (PMID 38361147) found colostrum 'affected' permeability and 'some immune measures'; it found a fall in one ratio with 99 percent heterogeneity, no effect on I-FABP, and reports no immune measures. Buckley 2009 (PMID 22253980), where permeability rose on colostrum during running training, was missing and is cited. Morrison 2014 (PMID 25068884), null in the heat, added. Immune evidence now has its own citation, Główka 2020 (PMID 32276466), mostly null, and the self-reported respiratory-symptom pooled analysis Jones 2016 (PMID 27462401) is shown as context. The record now says the trials mixed athletes and patients and used doses from 0.5 g/day to 1.7 g/kg/day. Letter unchanged.

The best-use line implied the question was untested; the small controlled trials exist and were null, now stated: Ferrando 1993 (PMID 8508192) and its second report Green 1994 (PMID 7889885), Meacham 1994 (PMID 7889886), and Beattie 1993 (PMID 8329361), added. The steroid-hormone claim is traced to Nielsen 1987 (PMID 3678698), a non-randomized study of 12 women, cited as context. Rondanelli 2020 (PMID 32540741) is described as a narrative review whose conclusion rests on non-randomized studies, with two authors at a commercial ingredient company's research unit. Letter unchanged.

The ref said Kim 2020 (PMID 32951763) 'modestly affected glycemic measures'; HbA1c did not change and the fasting glucose fall is within-group only. The pooled analyses that show the inconsistency are now cited: Peter 2019 (PMID 30385422) and Zhang 2024 (PMID 38784554) positive, Laczko-Zold 2023 (PMID 38274207) and the Cochrane review Ooi 2012 (PMID 22895968) null, with the Cochrane finding that no preparation was tested twice. Mes 2025 (PMID 40199408) added for prediabetes. The record now says no single form or dose carries the letter. Letter unchanged.

The record attributed the CARET harm to 'supplemental beta-carotene' without saying the trial gave beta-carotene with retinyl palmitate, cited the post-intervention follow-up, Goodman 2004 (PMID 15572756), as the primary result and said the excess 'persisted' when it continued as a nonsignificant trend; the primary result, Omenn 1996 (PMID 8602180), is now cited and the follow-up described exactly. The record also said no trial had enrolled anyone who stopped smoking long ago; the Physicians' Health Study, Hennekens 1996 (PMID 8602179), 39 percent former smokers, and the Women's Health Study, Lee 1999 (PMID 10601381), were null and are now cited. The mortality harms in ATBC and CARET, and the pooled lung, stomach and cardiovascular findings of Druesne-Pecollo 2010 (PMID 19876916), Zhang 2023 (PMID 36715090) and O'Connor 2022 (PMID 35727272), are stated. The closing inference that harm is a property of the supplement rather than the dose is withdrawn. Letter unchanged.

The best-use line listed AMPK activation and microbiome modulation, which are mechanisms, and insulin sensitivity, which Zhang 2008 (PMID 18397984) did not show by clamp; rewritten to the surrogates the trials measured. Added PREMOTE (PMID 33024120), the placebo-only pooled analyses Zhao 2023 (PMID 37598753) and Blais 2023 (PMID 36941490) for lipids, and Lei 2026 (PMID 41543854), null on visceral and liver fat. The dihydroberberine absorption claim now cites its source, Moon 2021 (PMID 35010998), with its supplier funding and null glucose result. The CYP interaction evidence, Guo 2012 (PMID 21870106), is on the record. The unsourced 'higher doses in Chinese medicine' line is gone. Letter unchanged.

The description gave Zhang 2008 (PMID 18397984) within-arm changes as differences versus placebo, called the single 36-person Yin 2008 pilot (PMID 18442638) a metformin equivalence, and said Zhang's primary outcome was HbA1c when it was plasma glucose and lipids; all corrected. Xie 2022 (PMID 36467075) is described as pooling berberine with and without other drugs. The placebo-controlled replication is now cited: PREMOTE (PMID 33024120), Ming 2021 (PMID 34365978), Zhao 2023 (PMID 37598753). The 'mandatory interaction check' wording is replaced by the interaction evidence itself, Guo 2012 (PMID 21870106). The record now says the B is drawn from 1.0 to 1.2 g/day trials and that 1.5 g/day matches only the pilot. Letter unchanged.

The best-use line claimed vascular and exercise support that no cited trial measured; exercise is now stated as not graded. Siervo 2013 (PMID 23596162) pooled trials of 2 hours to 15 days and is described as acute evidence. The null trials the record left out are cited: Sundqvist 2020 (PMID 32091599), Bondonno 2015 (PMID 26135348), Fejes 2024 (PMID 38546454), and the record now says the effect is smaller or absent in people already on antihypertensive treatment. Kerley 2018 (PMID 29553033), an independent positive replication on 24-hour pressure, is added. Letter unchanged.

The description claimed broad cognitive improvement at 600 mg plus from pooled trials that used 300 to 450 mg/day and improved two measures, and said the lack of a hard endpoint is why the letter is not A, which is not the site's rule. Rewritten to what Kongkeaw 2014 (PMID 24252493), Pase 2012 (PMID 22747190) and Tiemtad 2026 (PMID 41678913) show. Added Downey 2013 (PMID 23281132): acutely, 320 mg improved performance and 640 mg did not. Morgan 2010 (PMID 20590480) added for its result and its gastrointestinal side effects, also seen in Lopresti 2025 (PMID 41091332). The not-replicated flag stands for the 600 mg plus dose. Letter unchanged; the name is on the editor's list.

The record said the Hall trial was 'confirmed by' Lane 2024 (PMID 38418082), an umbrella review of observational studies that cannot confirm a trial or carry a grade. The randomized replications are now cited: Hamano 2024 (PMID 39267249) and Dicken 2025 (PMID 40760353). Sciarrillo 2024 (PMID 39163976) answers the no-washout objection. The nulls and limits are shown: Rego 2026 (PMID 41255123) and Aramburu 2024 (PMID 38988861). The description now says what the B covers, short-term intake and weight over 1 to 8 weeks. The replication search owed since October is closed. Letter unchanged.

The record cited only the ASPREE mortality report, McNeil 2018 (PMID 30221595), and attributed the disability-free survival and bleeding results to it. Those come from the companion papers by the same group, now cited: McNeil 2018 (PMID 30221596), the primary endpoint report, and McNeil 2018 (PMID 30221597), the cardiovascular and bleeding report. The mortality finding is stated as the paper states it: higher all-cause death, mostly cancer, called unexpected by the authors, and not confirmed over the decade of follow-up in Wolfe 2025 (PMID 40796244) and Shah 2025 (PMID 41043446), which found no long-term benefit, a higher event rate after stopping, and persisting bleeding. ASCEND, Bowman 2018 (PMID 30146931), and ARRIVE, Gaziano 2018 (PMID 30158069), are now cited and described as the different populations they are. Letter unchanged.

The best-use line claimed sleep quality with no sleep trial cited; Cheah 2021 (PMID 34559859) and Langade 2021 (PMID 32818573) now carry it. The ref treated two different extracts as one and overstated Lopresti 2019 (PMID 31517876): HAM-A and cortisol fell, the DASS-21 did not reach significance, testosterone did not differ from placebo; corrected. The strength signal is no longer called single-trial: Ziegenfuss 2018 (PMID 30463324) is added with its maker funding, and Lee 2026 (PMID 42738987) rates the pooled effect fragile. Albalawi 2025 (PMID 40746175), null on perceived stress, and Arumugam 2024 (PMID 39348746), positive, sit side by side. Björnsson 2020 (PMID 31991029), cholestatic liver injury cases, is added. Letter unchanged.

The record called Burke and Jennings 2023 (PMID 38404609) an uncontrolled case series but not that its six patients were also on drugs, other supplements and lifestyle changes, nor that the first author sits on the maker's scientific advisory board; both are now stated. Shimada 2021 (PMID 34226572), the only randomized trial, now carries its authors' ties to a rhamnan sulfate producer, and the form note no longer says MonitumRS is the only studied extract. Letter unchanged.

The record credited Kellenberger 2011 (PMID 21641214) with showing apigenin inhibits CD38; that paper names other flavonoids and not apigenin. Escande 2013 (PMID 23172919) is the apigenin paper and is now cited for it. Henderson 2024 (PMID 38302501), listed as a 'systems approach', is a randomized crossover trial of a branded multi-ingredient product with the maker's founder and an employee among the authors; it is now described that way and the record says 'standalone apigenin' where it says no trial exists. Letter unchanged.

The best-use line implied a healthy-aging result that no trial has produced, and an athletic power claim no cited paper supported; both are rewritten to the diagnosed populations and doses the trials used. The dose line said 300 to 600 mg/day; the Alzheimer's and diabetes trials used 1,200 mg/day. Added: Jeon 2024 (PMID 39300341), Sohn 2025 (PMID 39703111), Marcus 2017 (PMID 29042830) with its finding that TSH fell at 500 mg/day, and Kerksick 2024 (PMID 39683633), null on performance. Sagaro 2023 (PMID 36683513) is described as a meta-analysis mostly of alpha-GPC added to donepezil. Letter unchanged.

The record leaned on an observational global analysis and said only that no randomized trial exists. The randomized evidence it was missing is now cited: Xin 2001 (PMID 11711507) and Roerecke 2017 (PMID 29253389) pooled 15 and 36 trials and found blood pressure falls with reduced drinking, but only in people drinking more than two drinks a day; that is a new use at B. Voskoboinik 2020 (PMID 31893513), abstinence and AF recurrence, is a new use at C. Gardiner 2025 (PMID 39631226) is described as what it is, acute alcohol-before-sleep studies showing REM disruption, not a trial of reducing intake. The GBD 'zero is safest' conclusion is kept as context only. Headline D Route 1 held pending editor confirmation.

The record called Zhang 2025 (PMID 39879980) an RCT confirming metabolic benefit; it reports no significant between-group difference, with benefit only in a low-baseline subgroup. The largest trial, Suenaert 2026 (PMID 42343233, 142 adults), was null on its primary insulin-sensitivity endpoint and was run by the company that makes the product; it is now cited. Perraudeau 2020 (PMID 32675291) tested a five-strain combination and cannot grade Akkermansia alone; its authors are all employees of its maker. The co-founder and patent ties of the Depommier 2019 (PMID 31263284) senior authors are stated. Two trials the search found are added: Mount 2026 (PMID 42120725), company-run, 90 adults, less weight regain after dieting on pasteurized MucT, a different endpoint from the graded use; and Kang 2024 (PMID 39683431), a different strain tested on leg strength. Strains and forms are no longer pooled as one. Headline C held pending editor confirmation.

The description said protein supports lean mass 'particularly in older adults'. The evidence runs the other way: Morton 2018 (PMID 28698222) found the effect shrinks with age, Bhasin 2018 (PMID 29532075) was null at almost this record's dose in older men, and Obasi 2025 (PMID 41385355) found no benefit in healthy older people. Tagawa 2022 (PMID 36057893) shows the strength benefit depends on concurrent resistance training, now stated. Nunes 2022 (PMID 35187864) added with its industry ties, and the dairy-industry support behind Morton 2018 stated. Wu 2016 (PMID 26797090) relabeled as a narrative review. Endpoint type set to surrogate (lean mass), which is what the grade rests on. Letter unchanged.

The dose line said 500 to 1,500 mg/day, below every dose that reported benefit (2 to 3 g/day) and within the range the Cochrane review Rolim 2019 (PMID 31201734) found ineffective; corrected. The best-use line claimed 'cognitive aging' and 'aging-related neuropathy', neither of which any trial tested; it now names mild cognitive impairment, mild Alzheimer's disease and diabetic neuropathy, and drops the mechanism. Thal 2000 (PMID 10994000), the second large null Alzheimer's trial, and Hershman 2018 (PMID 29361042), showing the chemotherapy neuropathy harm persisting at two years, are added. Li 2015 (PMID 25751285) is now described as a pooled analysis of neuropathic pain, not cognition, and Pennisi 2020 (PMID 32408706) as a narrative review. Letter unchanged.

The record presented Truyens 2022 (PMID 35940251) as support for a C on fatigue; that trial was null on its primary endpoint and every secondary one, so fatigue in inactive inflammatory bowel disease is now a D Route 2 use. The depression evidence the record alluded to is now cited: the Cochrane review Shaw 2002 (PMID 11869656), the pooled analysis Javelle 2020 (PMID 31504850), and the small trials Jangid 2013 (PMID 23380314) and Meloni 2020 (PMID 32067288), with their quality warnings. Safety: the unresolved eosinophilia-myalgia syndrome association and 19.6 percent adverse events after a single dose in Jackson 2026 (PMID 42160304). Headline C held, scoped to depressive symptoms, pending editor confirmation.

The record carried a 'not replicated' flag while describing 11 controlled trials; the flag is removed and the checklist corrected (several small trials, null clinical endpoints, one positive imaging endpoint). Wu 2025 (PMID 40855942), a meta-analysis reporting large cognitive gains, is now cited with the reason it is not relied on: it pools transcranial brain stimulation with sensory stimulation. Letter unchanged.

Funding stated beside the letter. The NHS-Galleri trial, Sasieni 2026 (PMID 42776074) with the test-performance paper Neal 2026 (PMID 42773209), was funded by GRAIL, which sells the test, with GRAIL employees among the authors. The evidence already said so; it now also prints beside the letter.

Funding stated in the evidence. Ruiz 2026 (PMID 41747733) was NIH-funded and its senior author is chief scientific officer and an equity holder of Longeveron, maker of the cells; Tompkins 2017 (PMID 28977399) was foundation-funded with Longeveron employees among the authors; the Chinese trial Zhu 2024 (PMID 38679727) was government-funded. Letter unchanged.

Funding stated in the evidence. The largest sham-controlled trial, OVERTURE, Hajós 2024, and the MRI paper from the same trial, Da 2024 (PMID 38073386), are now cited; both were funded by Cognito Therapeutics, maker of the device, and written mostly by its employees. OVERTURE missed its primary clinical outcome. The meta-analysis the letter rests on, Ang 2025 (PMID 41381439), was publicly funded. Letter unchanged.

Funding stated in the evidence. Wei 2017 (PMID 28202779) was publicly funded but used a diet provided by L-Nutra, which licenses it from USC, and two authors hold L-Nutra equity; Espinoza 2026 (PMID 41372565) was sponsored by L-Nutra, with two of its employees among the authors. Letter unchanged.

Funding stated in the evidence. The single pilot trial, Fitzgerald 2021 (PMID 33844651), was supported by an unrestricted grant from the company that makes the two supplements in the program, and its two lead authors use the program in practice, hold a related patent application and earn from its educational products. Letter unchanged.

Funding stated beside the letter. The single adult trial, Oda 2023 (PMID 37764743), states that research funds and test foods were provided by the company that makes the lactoferrin tested, and five of its nine authors are its employees. Letter unchanged.

Funding checked for the stroke-prevention use. The October review reports nonfinancial support to the CSPPT trial, Huo 2015 (PMID 25771069), from Shenzhen AUSA Pharmed, maker of the combination pill. PubMed's record of the trial carries no conflict statement, so the use note states only what the indexed record shows; the full report is to be read before the review's statement is added.

Ties stated beside the letter. All four trials, Shevtsov 2003 (PMID 12725561), Darbinyan 2000 (PMID 11081987), Spasov 2000 (PMID 10839209) and Olsson 2009 (PMID 19016404), tested one proprietary standardized extract, and each has a scientist from the institute that developed and sells it among its authors; none names a funder. Letter unchanged.

Ties stated beside the letter. Every trial behind the Grade C was run with a manufacturer's product and a manufacturer's researchers among the authors: one manufacturer in Tashiro 2012 (PMID 23226979) and Kawada 2015 (PMID 25834304), a second in Gao 2023 (PMID 38009035). Kawada 2014 (PMID 25014997) is a review by the first manufacturer's employees, not a trial, and is now described as one. Letter unchanged.

Funding stated in the evidence. The neuropathy meta-analysis Li 2015 (PMID 25751285) reports that most of the trials it pooled were funded by the manufacturer of the acetyl-L-carnitine tested, and names a potential commercial bias; the trial that found harm, Hershman 2013 (PMID 23733756), was funded by the National Cancer Institute and the Medical Research Council. Letter unchanged.

Funding stated in the evidence. The one randomized trial behind the letter, De Jesus Moreno Moreno 2003 (PMID 12637119), names no funder in its indexed record; the systematic review Sagaro 2023 (PMID 36683513) and the stroke cohort Lee 2021 (PMID 34817582) were publicly funded. The October review found that alpha-GPC trials which do state a funder were paid for by alpha-GPC makers. Letter unchanged.

Funding stated beside the letter. The prostate study Keizman 2021 (PMID 34959847) was sponsored by the maker of PectaSol, and its co-author Isaac Eliaz is the company's developer; he is also the sole author of the review Eliaz 2019 (PMID 31683865). The earlier study, Guess 2003 (PMID 14663471), names no funder. Letter unchanged.

Ties stated beside the letter. The single open-label study, Houston 2023 (PMID 37849591), has an author on the staff of Calroy Health Sciences, which sells the product, and its authors hold a patent application covering the composition. The indexed record names no funder. Letter unchanged.

Funding corrected. Earlier today this record said PubMed recorded no funding for the one trial cited, Riche 2014 (PMID 25057276); that was true of the indexed entry, but the full text states a grant from the company that makes the pterostilbene tested, which also supplied the pterostilbene and placebo and reports having had no role in data collection or analysis. Now stated beside the letter.

Funding stated in the evidence. The knee-pain trial Zdzieblik 2017 (PMID 28177710) has a Collagen Research Institute author, and the skin meta-analysis Myung 2025 (PMID 40324552) found no effect in trials without industry funding, so the positive skin results come from company-funded trials. Letter unchanged.

Funding stated beside the letter. All three trials behind the Grade B were run with the extract maker's money or product: the standardized root extract supplied by its maker in Chandrasekhar 2012 (PMID 23439798) and Wankhede 2015 (PMID 26609282), and Lopresti 2019 (PMID 31517876) funded by the maker of the extract it tested. Letter unchanged.

Funding stated beside each trial. The company that makes the urolithin A product tested funded or ran five of the six randomized trials cited: Andreux 2019 (PMID 32694802), Liu 2022 (PMID 35050355), Singh 2022 (PMID 35584623), Whitfield 2025 (PMID 40839339) and Denk 2025 (PMID 41174221). Its employees, founder and board members are among the authors, and Liu 2022 (PMID 35050355) states the company had a role in design, conduct and analysis. The one independent trial, Zhao 2024 (PMID 39487653), was publicly funded in China. Whitfield 2025 (PMID 40839339), maker-funded and null on running performance in trained runners, is now cited. Letter unchanged.

Both type 2 diabetes uses rested on one trial, Moludi et al. (Nutr Metab (Lond), 2022; PMID 35870947; 120 adults), null on fasting glucose and positive on endothelial and inflammatory markers. The search found independent trials on both. Endothelial and inflammatory markers: Maleki et al. (Diabetol Metab Syndr, 2020; PMID 32015761; 50 adults) and Li et al. (iScience, 2025; PMID 40546935; 165 adults, Chongqing, blood pressure, vasodilation and pulse wave velocity) replicate the direction, so a replicated surrogate reaches B. Glycemic control: Maleki et al. (Amino Acids, 2020; PMID 32472292) and Esmaeili et al. (Can J Diabetes, 2021; PMID 32861603), the same Tabriz group at the same 3 g/day dose without a diet, found fasting glucose and HOMA-IR fall, and pooled analyses (Tzang et al., 2024; PMID 38755142; Nie et al., 2026; PMID 41275513) find small reductions; one null and two positive trials is mixed, C. The headline follows the best use. Moved October 2026 by editor adjudication (Arnold Sandlow, D.C.) of a use-specific literature search, every trial re-fetched from PubMed before the letter rose; the search is recorded on the record.

The record graded SAM-e D on Route 2 for depression as monotherapy or add-on together, on two null pooled analyses (Galizia et al., Cochrane, 2016; PMID 27727432; Peng et al., Gen Hosp Psychiatry, 2024; PMID 38199136). The largest and most recent pooled analysis, Limveeraprajak et al. (Prog Neuropsychopharmacol Biol Psychiatry, 2024; PMID 38423354; 23 trials, 2,183 patients), finds SAM-e alone superior to placebo (SMD -0.58) and SAM-e as an add-on not. Two questions with two answers were under one letter. Monotherapy, with positive and null pooled analyses side by side and the best modern trials null (Mischoulon et al., 2014; PMID 24500245; Sarris et al., 2020; PMID 31712971), is mixed evidence, C. Add-on use, with one early positive trial (Papakostas et al., 2010; PMID 20595412; 73 patients) and the two larger later trials null (Targum et al., 2018; PMID 30368163; Sarris et al., 2018; PMID 30115553), stays D on Route 2. Moved October 2026 by editor adjudication (Arnold Sandlow, D.C.) of a use-specific literature search, every trial re-fetched from PubMed before the letter rose; the search is recorded on the record.

The blood NAD use held a C on one dose-ranging trial, Yi et al. (GeroScience, 2023; PMID 36482258; 80 adults). Three further randomized placebo-controlled trials report the same result: Pencina et al. (J Clin Endocrinol Metab, 2023; PMID 36740954; 30 adults, 28 days), Morifuji et al. (GeroScience, 2024; PMID 38789831; 60 older adults, 12 weeks) and Katayoshi et al. (Sci Rep, 2023; PMID 36797393; 36 adults, 12 weeks). A replicated surrogate reaches B under the written rule. Every one of the four was run or funded by a company that sells NMN, so the funding and investigator line prints beside the letter under the rules of September and October. The pooled metabolic endpoints across NMN trials (Yang et al., Nutrients, 2026; PMID 42514320) show no clear clinical benefit, which the record states. Moved October 2026 by editor adjudication (Arnold Sandlow, D.C.) of a use-specific literature search, every trial re-fetched from PubMed before the letter rose; the search is recorded on the record.

The stored absence query for spermidine was unscoped and had been returning the cognition trials since July while the record claimed zero; gate_g169 was reporting the mismatch as unreproducible. The aging claim keeps D on Route 1 with a query scoped to mortality, lifespan and longevity (zero randomized trials). Memory in older adults with subjective cognitive decline is now its own use at D on Route 2: the 12-month SmartAge trial of Schwarz et al. (JAMA Netw Open, 2022; PMID 35616942; 100 adults) found no difference from placebo on its primary memory outcome, and the 3-month pilot of Wirth et al. (Cortex, 2018; PMID 30388439) reported only a within-group gain in 30 people. The aging claim's own citations are unchanged: Eisenberg et al. (Nat Med, 2016; PMID 27841876), lifespan in mice, and Kiechl et al. (Am J Clin Nutr, 2018; PMID 29955838), observational dietary intake. The headline letter is unchanged.

The B rested on the medical-ward subgroup of Chen et al. (Mol Neurobiol, 2016; PMID 26189834), a meta-analysis whose pooled primary was not significant (P=0.08). The subgroup's positive trial is Al-Aama et al. (Int J Geriatr Psychiatry, 2011; PMID 20845391; 145 patients, 0.5 mg). Two later medical-ward trials, neither cited, are null: Jaiswal et al. (Am J Med, 2018; PMID 29729237; 3 mg, delirium numerically more frequent on melatonin) and RESTORE (Al Alawi et al., BMJ Open, 2026; PMID 41592826; 5 or 8 mg, stopped for futility). One positive and two null trials on the same ward is mixed evidence, which is C. The endpoint type is corrected to clinician_rated, since delirium by the Confusion Assessment Method is a clinician-scored scale. Moved October 2026 by editor adjudication (Arnold Sandlow, D.C.) of a use-specific literature search; the search is recorded on the record.

A use-specific PubMed search in October 2026 found randomized trials on records that said none existed, and found null trials that move three records from Route 1 (untested) to Route 2 (tested, no benefit). No letter changes. Boron: small randomized trials exist and were null (Ferrando et al., Int J Sport Nutr 1993; PMID 8508192; Meacham et al., Environ Health Perspect 1994; PMID 7889886); the route moves from 1 to 2 and the letter stays D. Exosome therapy: a triple-blind placebo-controlled knee trial was null (Bolandnazar et al., BMC Musculoskelet Disord 2024; PMID 39465400), so the knee use moves from Route 1 to Route 2; a 20-man randomized hair pilot of a plant-derived formulation (Amini et al., Life 2025; PMID 40141845) is now named and the hair letter is held at D under review. Reducing anticholinergic burden: randomized deprescribing Reducing or stopping a medication under a prescriber's supervision. trials have been run and found no clinical benefit (Taylor-Rowan et al., Cochrane Database Syst Rev 2023; PMID 38063254; Braithwaite et al., Age Ageing 2023; PMID 37740900); the route moves from 1 to 2, the false line is removed, and 'prudent practice' wording is withdrawn. Alcohol reduction: the record now names the randomized abstinence trial in atrial fibrillation (Voskoboinik et al., N Engl J Med 2020; PMID 31893513) as a different use, keeps Route 1 for mortality and major events, and drops the word 'prudent'. Arterosil: 'no human randomized trial' was false as written; a small randomized trial for constipation exists (Shimada et al., Sci Rep 2021; PMID 34226572), so the line is scoped to vascular outcomes. C60: 'no human trial has been run' was false as written; topical fullerene trials exist (Liu et al., J Clin Oncol 2026; PMID 41861269), so the line is scoped to oral C60. Ca-AKG: 'no human randomized trial' was false as written; a bone-marker trial at 6 g (Filip et al., Int J Vitam Nutr Res 2007; PMID 17896582) exists and the ABLE trial has enrolled (Lim et al., Exp Gerontol 2025; PMID 40819772); the line is scoped to aging outcomes. Fisetin: 'no human randomized trial has reported results' was false as written; small trials in other uses exist (Hodgin et al., Int J Environ Res Public Health 2021; PMID 33802381), so the line is scoped to the senolytic use. NAD+ (direct): one randomized pharmacokinetic study of an oral NAD+ formulation now exists (Kornilov et al., Geroscience 2026; PMID 42530810), so the line is scoped to clinical outcomes. Spermidine: 'no human randomized trial has been run' was false as written; SmartAge (Schwarz et al., JAMA Netw Open 2022; PMID 35616942) was a 12-month randomized memory trial and was null; the line is scoped to lifespan and mortality. Ergothioneine: 'no human randomized trial' was false; a 19-person randomized pilot in mild cognitive impairment exists (Yau et al., J Alzheimers Dis 2024; PMID 39544014). The letter is held at D under review, not raised, until it is replicated. PectaSol: 'no randomized trial for any use' was false; two placebo-controlled trials in other uses were null (Lau et al., JACC Basic Transl Sci 2021; PMID 33532663; Andrews et al., Contemp Clin Trials Commun 2020; PMID 31872160). The prostate use has no randomized trial, so it stays on Route 1; the manufacturer's role is disclosed. Aged garlic extract: renamed to the doses its trials used, 480 mg/day for blood pressure (Ried et al., Eur J Clin Nutr 2013; PMID 23169470) and 2.4 g/day in the plaque trials; the name previously read 1.2 to 2.4 g/day, a range no trial tested. The letter (B) is unchanged.

The magnesium glycinate elemental protocol held a B on a meta-analysis of oral magnesium salts generally (Zhang et al., Hypertension 2016; PMID 27402922), which the record itself said did not test glycinate. EBL grades magnesium by the salt on the label. The only glycinate trial with a blood pressure outcome gave it with vitamin D (Cheung et al., Nutrition 2022; PMID 35576873). Untested for this salt: D on Route 1. This is not a finding that magnesium does not work. Editor adjudication, October 2026.

The magnesium glycinate 300 to 400 mg protocol held a B on a review of trials that used other magnesium salts (Morais et al., Nutrition 2017; PMID 28526383). EBL grades magnesium by the salt on the label. No randomized trial of magnesium glycinate alone has measured insulin resistance, and a 2024 clamp trial of oral magnesium was null (Drenthen et al., Diabetologia 2024; PMID 37922013). Untested for this salt: D on Route 1. This is not a finding that magnesium does not work. Editor adjudication, October 2026.

Pterostilbene held a C on its single trial (Riche et al., Evid Based Complement Alternat Med 2014; PMID 25057276), whose pre-specified primary outcome was lipids. LDL cholesterol rose by 17.1 mg/dL; blood pressure fell only as a secondary outcome. The trial was funded by the company that makes the pterostilbene tested. Tested on its primary question and adverse: D on Route 2. Editor adjudication, October 2026.

EEG neurofeedback held a C on a network meta-analysis of digital interventions whose neurofeedback signal rests on unblinded ratings (Mei et al., Front Psychiatry 2026; PMID 42358407). The largest double-blind sham-controlled trial was null at every time point (Neurofeedback Collaborative Group, J Am Acad Child Adolesc Psychiatry 2021; PMID 32853703), and meta-analyses limited to probably blinded ratings found no effect (Cortese et al., J Am Acad Child Adolesc Psychiatry 2016; PMID 27238063; Westwood et al., JAMA Psychiatry 2025; PMID 39661381). Tested, no benefit: D on Route 2. Editor adjudication, October 2026.

The general 2 to 4 g/day omega-3 protocol held a B on pooled analyses mostly at 1 g/day or less and on REDUCE-IT, which tested purified EPA. At this dose of mixed EPA and DHA the large trials were null: STRENGTH (Nicholls et al., JAMA 2020; PMID 33190147) and OMEMI (Kalstad et al., Circulation 2021; PMID 33191772). C. Editor adjudication, October 2026.

The melatonin 0.5 to 1 mg protocol held a B on a meta-analysis that pooled all doses and found larger effects at higher doses (Ferracioli-Oda et al., PLoS One 2013; PMID 23691095), which does not test a low-dose protocol. Direct low-dose trials are small and population-specific (Zhdanova et al., J Clin Endocrinol Metab 2001; PMID 11600532; Sletten et al., PLoS Med 2018; PMID 29912983), and a jet lag trial with 0.5 mg arms was null (Spitzer et al., Am J Psychiatry 1999; PMID 10484950). C. Editor adjudication, October 2026.

Collagen held one B across joint and skin uses. Joint pain rests on small self-reported trials in athletes (Clark et al., Curr Med Res Opin 2008; PMID 18416885; Zdzieblik et al., Appl Physiol Nutr Metab 2017; PMID 28177710). For skin, Myung et al. (Am J Med 2025; PMID 40324552) found no effect in trials without industry funding and in high-quality trials. Mixed across the named uses: C. Editor adjudication, October 2026.

Citicoline held a B on a 2005 Cochrane review of 14 short, heterogeneous trials (Fioravanti and Yanagi, Cochrane Database Syst Rev 2005; PMID 15846601). The only 12-month double-blind trial was null (Cohen et al., Cerebrovasc Dis 2003; PMID 12865605), and the European Food Safety Authority found a memory benefit not established (EFSA J 2024; PMID 38966137). Mixed: C. The Cochrane authors are now named correctly as Fioravanti and Yanagi. Editor adjudication, October 2026.

Beetroot held a B on one positive trial in hypertensive adults (Kapil et al., Hypertension 2015; PMID 25421976) and an early meta-analysis of mostly acute trials. The largest trial (Sundqvist et al., Am J Clin Nutr 2020; PMID 32091599; 231 adults) was null on 24-hour blood pressure, and pooled analyses find the effect only on clinic readings (Ashor et al., J Hypertens 2017; PMID 28319596; Gronroos et al., Nutr Metab Cardiovasc Dis 2024; PMID 39069465). Mixed: C. Editor adjudication, October 2026.

The Bacopa 600 mg+ protocol held a B on meta-analyses whose trials mostly used 300 to 450 mg/day (Kongkeaw et al., J Ethnopharmacol 2014; PMID 24252493; Pase et al., J Altern Complement Med 2012; PMID 22747190). The only chronic trial at 600 mg/day (Peth-Nui et al., Evid Based Complement Alternat Med 2012; PMID 23320031) used a 5 percent saponin extract, and a recent trial was null (Lopresti et al., Clin Drug Investig 2025; PMID 41091332). At the named dose this is C. Editor adjudication, October 2026.

ALCAR held a B on two meta-analyses and one neuropathy trial. The largest Alzheimer's trial (Thal et al., Neurology 1996; PMID 8797468) was null, the Cochrane review of dementia (Hudson and Tabet, Cochrane Database Syst Rev 2003; PMID 12804452) found no benefit on objective cognition, and the Cochrane review of diabetic neuropathy (Rolim et al., Cochrane Database Syst Rev 2019; PMID 31201734) found very low certainty and no benefit at 1,500 mg/day or less. A 409-patient trial (Hershman et al., J Clin Oncol 2013; PMID 23733756) found it worsened chemotherapy neuropathy. Mixed: C. Editor adjudication, October 2026.

Bromelain held a C on a trial of a three-ingredient enzyme product (Henrotin et al., RMD Open 2025; PMID 40803821) sponsored by the product's maker. The bromelain-alone search found two placebo-controlled trials on the graded uses and both were null: Brien et al. (QJM 2006; PMID 17121765) in knee osteoarthritis and Ley et al. (Chin J Integr Med 2016; PMID 27412590) on inflammation markers. Tested alone, no benefit: D on Route 2. Editor adjudication, October 2026.

Vitamin D3 held a B headline on fracture reduction in deficient or older adults. The trial it rested on, Chapuy et al. (N Engl J Med 1992; PMID 1331788), gave vitamin D3 with calcium. For vitamin D alone, the Cochrane review (Avenell et al., Cochrane Database Syst Rev 2014; PMID 24729336) and a 69-trial meta-analysis (Masse et al., BMJ 2026; PMID 42161415) found no fracture benefit: D on Route 2. Falls moved to C: a later meta-analysis was null (Ling et al., Clin Nutr 2021; PMID 34656949) and high doses increased falls (Sanders et al., JAMA 2010; PMID 20460620). Respiratory infection moved to C: the same group's update is no longer significant (Jolliffe et al., Lancet Diabetes Endocrinol 2025; PMID 39993397). The headline follows to C. Editor adjudication, October 2026.

The general red and near-infrared light page held one B across pain, wound healing, skin aging and cognition, on one skin trial (Wunsch et al., Photomed Laser Surg 2014; PMID 24286286) and a mechanism review. Searched use by use in October 2026, those uses rest on small, mixed or self-reported trials, which is C. Oral mucositis prevention keeps its A on its own page. Editor adjudication, October 2026.

The D3 2,000 to 4,000 IU plus MK-7 100 mcg protocol held a B on one MK-7-alone trial at 180 mcg (Knapen et al., Osteoporos Int 2013; PMID 23525894) measuring bone density, a surrogate. No trial has tested the combination at these doses. The largest combination trial, AVADEC (Diederichsen et al., Circulation 2022; PMID 35465686; 720 mcg MK-7 plus 1,000 IU D3), missed its primary endpoint, and vitamin D alone did not prevent fractures in VITAL (LeBoff et al., N Engl J Med 2022; PMID 35939577) or D-Health (Waterhouse et al., Lancet Diabetes Endocrinol 2023; PMID 37011645). C, with the name's doses flagged as untested. Editor adjudication, October 2026.

The TM record held a B on one trial, Schneider et al. (Circ Cardiovasc Qual Outcomes 2012; PMID 23149426), 201 adults with coronary heart disease. No independent trial has repeated it. The Cochrane review (Rees et al., Cochrane Database Syst Rev 2024; PMID 38358047) extracts the same trial as null, and a later trial from the same group (Norris et al., Front Med 2025; PMID 40177279) missed its primary endpoint. A single positive trial nobody has repeated is C. Editor adjudication, October 2026.

The L-theanine 200 mg sleep protocol held a B on Hidese et al. (Nutrients 2019; PMID 31623400), a 30-person crossover trial of stress in which sleep quality was a secondary, self-reported outcome, funded by the company that makes the L-theanine tested. The review it cited as confirmation (Gerolymos et al., Mol Psychiatry 2026; PMID 42410082) does not pool sleep outcomes. The sleep meta-analysis (Bulman et al., Sleep Med Rev 2025; PMID 40056718) is positive on self-report only and pools combination products. One small unreplicated trial is C. Editor adjudication, October 2026.

Time-restricted eating held a B on the claim that its trials isolate a timing effect independent of calories. One of the two trials cited, Wilkinson et al. (Cell Metab 2020; PMID 31813824), is single-arm with no control group. The other, Sutton et al. (Cell Metab 2018; PMID 29754952), is a crossover in 8 men. The larger randomized trials that matched calories found no added effect of the eating window: Liu et al. (N Engl J Med 2022; PMID 35443107; 139 adults, 12 months) and Lin et al. (Ann Intern Med 2023; PMID 37364268; 90 adults, 12 months). One small positive trial without replication is C, and the isocaloric-timing claim is withdrawn. Editor adjudication, October 2026.

The DHA 1 to 2 g/day protocol held a B on one trial, MIDAS (Yurko-Mauro et al., Alzheimers Dement 2010; PMID 20434961), which gave 900 mg/day, improved one memory test of several and was led by an employee of the company that made the DHA tested. The larger and longer trials are null: EPOCH (Danthiir et al., Am J Clin Nutr 2018; PMID 29722833) at 1,720 mg/day for 18 months, MAPT (Andrieu et al., Lancet Neurol 2017; PMID 28359749) over 3 years, and the DHA-specific meta-analysis (Balachandar et al., Eur J Clin Pharmacol 2020; PMID 32060571). A Cochrane review of cardiovascular prevention that the record cited for cognition has been removed. Tested, no benefit: D on Route 2. Editor adjudication, October 2026.

Sardeli et al. (Nutrients, 2018; PMID 29596307) pooled six randomized trials in older adults with obesity who were restricting calories. Adding resistance training three times a week prevented 93.5 percent of the lean mass lost with restriction alone, while fat and total weight loss did not differ significantly. The record now cites it beside CALERIE-2. It addresses the muscle cost of restriction rather than the aging endpoint, so the C stays.

Harmsen et al. (Cell Metab, 2026; PMID 41418772) ran a randomized crossover in 13 adults with type 2 diabetes: 4.5 days of daylight through office windows kept glucose in the normal range for longer than constant artificial light, and shifted metabolism toward fat burning. Two large UK Biobank analyses add observational context: brighter days and darker nights went with lower mortality across 88,905 people wearing light sensors (Windred et al., PNAS, 2024; PMID 39405349), and each daily hour outdoors in daylight went with fewer insomnia symptoms and better mood (Burns et al., J Affect Disord, 2021; PMID 34488088). The trial is small and short and measures glucose, not disease, so daylight carries no grade.

Kane et al. (bioRxiv preprint, 2024; PMID 38979132), in a version revised in 2026, gave NMN to mice from middle age for life. Frailty accumulated more slowly in both sexes, but median lifespan rose 8.5 percent in females only, and the female control mice were short-lived, so the result may reflect correcting poor baseline health. It is a mouse study and not peer reviewed. No human trial has tested NMN against lifespan, healthspan or mortality, so that use stays D.

Hawkinson et al. (Nat Metab, 2026; PMID 42265388) found oral glucosamine worsened cognition in Alzheimer's mouse models, and in health records glucosamine use went with faster disease progression and shorter survival. A preprint by Nakashima et al. (medRxiv, 2026; PMID 42619916) found the same progression estimate for multivitamin and calcium plus vitamin D records and no mortality excess, which points to the record marking the patient. Both are cited on the record as context. The C for knee osteoarthritis is unchanged.

Liao et al. (J Prev Alzheimers Dis, 2026; PMID 42000571) reported faster cognitive decline among older adults in the ADNI cohort who took omega-3 supplements, and Dyall and Plourde (J Prev Alzheimers Dis, 2026; PMID 42612546) replied that confounding and method limits may explain it. Supplement type and dose were not recorded, so this is not a test of the DHA dose the protocol names. Both are now cited on the DHA cognition protocol, whose B was already under review for resting on a single trial. The letter does not move on observational data.

Hua et al. (BMJ Open, 2026; PMID 41571412) followed 5,065 US adults for six years and found slightly faster cognitive decline among vitamin D supplement users, confined to those whose blood level was already normal at baseline. An observational association cannot show the supplement caused it. It is now cited on the record as context beside the existing D grades for already-replete adults, where randomized trials found no benefit. No letter moves.

The 2023 animal work behind the taurine longevity claim argued that taurine falls with age. Fernandez et al. (Science, 2025; PMID 40472098), from the National Institute on Aging, measured it in three human cohorts, in monkeys and in mice, both in the same individuals over time and across age groups, and found it rose or held steady. The aging use was already D on Route 1, because no human trial has tested taurine against an aging outcome, and it stays D. The record now cites the failed replication alongside the original claim.

Mattioni Maturana et al. (Sports Med, 2026; PMID 42730846) pooled 69 randomized trials with 2,387 participants. Interval training raised VO2max more than moderate continuous training, and about 80 percent of its maximal effect arrived at roughly 11 minutes per session including recovery, against about 52 minutes for continuous moderate training. Effects on blood pressure, lipids, glucose and body composition were comparable. This is a second, independent meta-analysis of randomized trials on the endpoint the B rests on, and it is now part of the open review of the record's not-replicated flag. The letter stays B pending that review.

STAREE (Zoungas et al., N Engl J Med, 2026; PMID 42670961) randomized 9,971 adults aged 70 and over without cardiovascular disease, diabetes or dementia to atorvastatin 40 mg or placebo for a median 5.9 years. Major cardiovascular events fell about 30 percent (hazard ratio 0.70, 95% CI 0.61 to 0.82). Disability-free survival, the trial's second primary endpoint, did not change (hazard ratio 0.94, 95% CI 0.84 to 1.05). The library now carries both halves side by side, because together they show a hard endpoint moving while the healthspan endpoint did not. No grade changes.

The protocol was named "Lion's Mane 1,000 mg (fruiting body)" while its only trial, Mori et al. (Phytother Res, 2009; PMID 18844328), gave four 250 mg tablets of fruiting-body powder three times daily, about 3 g/day, for 16 weeks. A dose in a name is read as the dose that was tested, and 1,000 mg is a consumer dose no trial has tested on cognition. The protocol is renamed at the studied dose with an as-studied block; the letter (C) and the record identifier are unchanged. Found by the dose-in-name check that became gate_g183 the same day.

Luteolin held a C on Di Stadio et al. (Curr Neuropharmacol, 2022; PMID 35450527), a randomized trial of a fixed palmitoylethanolamide plus luteolin combination with olfactory training. The trial does not isolate luteolin, and under the published scale a combination product grades the combination, not its parts. A PubMed search in October 2026 for randomized trials naming luteolin returned 14 records, all combination or mixture products; the only luteolin-alone arm is a pseudo-randomized botanical crossover in Gulf War illness (Hodgin et al., Int J Environ Res Public Health, 2021; PMID 33802381), a different claim. With no randomized trial of luteolin alone on any graded outcome the letter is D on Route 1, untested. Moved October 2026 by editor adjudication; the absence check and the search are recorded on the record.

The R-ALA protocol held a B on the same bibliography as the supplement record: Ansar et al. (Saudi Med J, 2011; PMID 21666939), one 8-week glycemic trial of 57 adults; SYDNEY 2 (Ziegler et al., Diabetes Care, 2006; PMID 17065669); NATHAN 1 (Ziegler et al., Diabetes Care, 2011; PMID 21775755), null on its primary; and Han et al. (Eur J Endocrinol, 2012; PMID 22837391), a mostly intravenous pool. None used R-alpha-lipoic acid or fasted dosing. One positive glycemic trial without replication is C. Moved October 2026 by editor adjudication; the name's R-form and fasted claims stay on the review roster.

Alpha-lipoic acid held a B on SYDNEY 2 (Ziegler et al., Diabetes Care, 2006; PMID 17065669), a 5-week oral trial in 181 adults, with replication claimed from Han et al. (Eur J Endocrinol, 2012; PMID 22837391), a meta-analysis of mostly intravenous trials its own authors rate as poor. The 4-year oral trial, NATHAN 1 (Ziegler et al., Diabetes Care, 2011; PMID 21775755; 460 adults, 600 mg/day), missed its primary composite endpoint and recorded more serious adverse events on treatment. The glycemic claim rests on Ansar et al. (Saudi Med J, 2011; PMID 21666939), one 8-week trial of 57 adults. One positive short oral trial, one null long oral trial and an intravenous pool do not meet the B definition of an independently replicated randomized result; the letter is C. Moved October 2026 by editor adjudication; the search is recorded on the record.

The soy-derived phosphatidylserine protocol held a C on the strength of Cenacchi et al. (Aging (Milano), 1993; PMID 8323999), a trial of bovine-cortex phosphatidylserine, a different product. The two randomized trials of the soy form in the population it is sold to, 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), found no between-group difference from placebo. Under the published scale a product tested in its own form and population without benefit is Grade D on Route 2. The letter moved in October 2026 after an editor adjudication of a machine second check and five independent reviews; the search behind the move is recorded on the protocol.

A second check of 143 machine-judged grades, and five independent reviews of that check, were reconciled against PubMed's own records in October 2026. Nine records keep their letter and gain corrected text. Benfotiamine cited only its two null trials; the two positive short trials (Haupt et al., Int J Clin Pharmacol Ther 2005, BEDIP; Stracke et al., Exp Clin Endocrinol Diabetes 2008, BENDIP) and the null 24-month trial (Fraser et al., Diabetes Care 2012) are now on the record. Glucosamine called the GAIT dose the sulfate dose when GAIT used the hydrochloride; the two positive (Reginster et al., Lancet 2001; Pavelka et al., Arch Intern Med 2002) and two null (Hughes and Carr, Rheumatology 2002; Fransen et al., Ann Rheum Dis 2015) sulfate trials are now cited by salt and endpoint. Chromium gains the two further meta-analyses that disagree with the one it cited (Asbaghi et al., Pharmacol Res 2020; Yin and Phung, Nutr J 2015). PQQ claimed two cognition trials and cited two papers that measured none; the three trials that did are now cited (Itoh et al., Adv Exp Med Biol 2016; Shiojima et al., J Am Nutr Assoc 2022; Tamakoshi et al., Food Funct 2023). Phosphatidylserine states that its positive trial used the bovine form and that the soy form has a null trial (Jorissen et al., Nutr Neurosci 2001). Berberine credited the wrong paper with the metformin comparison. Alpha-lipoic acid gains the null 4-year NATHAN 1 trial (Ziegler et al., Diabetes Care 2011) and its adverse event rate. Vitamin E's prostate cancer increase in SELECT (Klein et al., JAMA 2011) is significant, not a trend, and ATBC's hemorrhagic stroke deaths are now stated. Methylene blue's one trial is described as what it was, a single dose after fear-extinction sessions. Eighteen further records carry an internal Grade under review status with the reason, and no letter moves on them until the missing trials are searched.

The DIM record cited Thomson 2017 for a favorable change in an estrogen-metabolism biomarker. The same trial, in 130 women taking tamoxifen for breast cancer, found plasma endoxifen, 4-hydroxytamoxifen and N-desmethyltamoxifen lower on DIM than on placebo (P<0.001), and breast density unchanged. Endoxifen is tamoxifen's main active form. Whether the drop reduces tamoxifen's benefit has not been tested. The record now states the finding in its caution, its drug-conflict text and a new interaction entry. The grade is unchanged; the omission was the defect.

The record's title said Grade D, no human randomized trial, while its own citation was a randomized trial of swelling after dental surgery. Five randomized placebo-controlled trials exist: four found less postoperative swelling (Tachibana 1984, Al-Khateeb 2008, Tamimi 2021, Kee 1989) and one found no effect (Chopra 2009), and a 2013 systematic review rates them small and poorly designed. The record now grades two uses. Swelling after oral or sinus surgery is C: small, mixed, short-term trials. Systemic, cardiovascular and longevity use stays D on Route 1, because no trial has tested it, which is the use the enzyme is sold for. The headline letter stays D.

The record carried a C on two prostate cancer studies that measured PSA doubling time in 13 and 59 men (Guess et al., Prostate Cancer Prostatic Dis 2003; Keizman et al., Nutrients 2021). Neither had a control group. A PubMed search in October 2026 for randomized controlled trials of modified citrus pectin returned none, for any outcome. Under EBL's rule a C needs human randomized evidence, so the letter moves to D on Route 1: untested, which is different from tested and failed. The record's notes also carried a stale sentence arguing for a B on galectin-3 reduction; it is gone. Five independent reviews of the second citation check reached the same reading.

EBL joined Holy Hydrogen's affiliate program and now earns a commission when a reader buys the Lourdes Hydrofix through the link on the hydrogen water page. Holy Hydrogen is the US seller the manufacturer links to, and it was already where that link went, unpaid. The generator carries no grade. The therapy it sits under, Molecular Hydrogen Water, keeps its C, now locked: it changes only through a logged review that names the evidence. The disclosure page now lists three commercial relationships instead of two.

The general page stated that the evidence for knee osteoarthritis pain and wound healing was small and mixed without citing it. It now carries four uses with their own letters and citations: skin aging C, on small trials including Lee et al. (J Photochem Photobiol B, 2007; PMID 17566756); knee osteoarthritis pain C, where meta-analyses disagree, Stausholm et al. (BMJ Open, 2019; PMID 31662383) positive and others not, mostly on lasers outside 630 to 850 nm; diabetic foot ulcer healing C, on three positive meta-analyses of small trials including Li et al. (Diabetes Res Clin Pract, 2018; PMID 30009935); venous leg ulcer healing C, on small trials with mixed results including Lundeberg and Malm (Ann Plast Surg, 1991; PMID 1793238). Headline letter unchanged at C. Editor adjudication, October 2026.

The B rested on Goyal 2014 (PMID 24395196), which graded anxiety, depression and pain and found insufficient evidence on sleep, while the record described sleep. The letter now rests on the sleep evidence: Rusch 2019 (PMID 30575050), 18 trials, moderate strength against attention-matched controls, with supporting meta-analyses and trials. Three statements were removed for lack of support: that the effect adds to cognitive behavioral therapy for insomnia (de Entrambasaguas 2023, PMID 37271575, could not show it), that it works best for middle-of-the-night waking, and that benefit needs 20 or more minutes a day. The named app was removed. Letter unchanged at B. Editor adjudication, October 2026.

The review asked whether the D Route 2 should stand on the LDL rise or blood pressure should be graded as its own use. Under the written definitions both hold, so both are now on the record: LDL cholesterol in adults with raised cholesterol, D Route 2, tested and harm found; blood pressure, C, one positive trial (Riche et al., Evid Based Complement Alternat Med, 2014; PMID 25057276). Headline letter unchanged at D. Closes the review opened October. Applied by the written grade definitions, October 2026.

Incident type 2 diabetes rests on one trial, the SURMOUNT-1 three-year analysis (Jastreboff et al., N Engl J Med, 2025; PMID 39536238), and the cardiovascular trial is a non-inferiority comparison with dulaglutide (Nicholls et al., N Engl J Med, 2025; PMID 41406444). Under Rule 03 a single positive trial without independent replication is C, the rule applied to retatrutide in October. Closes the review opened October. Applied by the written grade definitions, October 2026.

The B rested on a Cochrane review that rates its five small placebo-controlled trials very low quality, with withdrawals tied to the outcome and evidence of publication bias (Stewart et al., Cochrane Database Syst Rev, 2016; PMID 27021597). The written definition of B requires at least one well-designed trial supported by others; small trials of very low quality are C. Walking distance gains are unchanged as findings. Applied by the written grade definitions, October 2026.

Four randomized trials agree that oral NMN raises blood NAD (Yi et al., GeroScience, 2023; PMID 36482258; Pencina et al., J Clin Endocrinol Metab, 2023; PMID 36740954; Morifuji et al., GeroScience, 2024; PMID 38789831; Katayoshi et al., Sci Rep, 2023; PMID 36797393). The editor settled the open question of what counts as a qualifying biomarker in October 2026 and added it to the B definition: a biomarker qualifies only when it is an established stand-in for a clinical outcome, and a blood level of the supplement or of what the body makes from it shows absorption and qualifies for nothing above C. Blood NAD is that kind of level, so the use and the headline move to C, the same reasoning that holds nicotinamide riboside at C. This supersedes the correction of earlier the same day that confirmed the B. Editor decision, Arnold Sandlow, D.C., October 2026.

The use-level A rested on one pooled review and did not cite the largest trial. PLACIDE (Allen et al., Lancet, 2013; PMID 23932219), 2,981 older inpatients, found no significant reduction in C. difficile diarrhea. The 2025 Cochrane update (Esmaeilinezhad et al.; PMID 40931979) still finds fewer cases overall, about one prevented for every 65 people treated, and rates that evidence low certainty. The positive trials tested different products, so the benefit does not attach to probiotics as a class. The record's headline grade was already C and does not change.

Grade A requires the effect a record reports to replicate across independent research groups. CoQ10's A reported fewer deaths and major cardiovascular events, and that result rests on one trial, Q-SYMBIO (Mortensen et al., Journal of the American College of Cardiology: Heart Failure, 2014; PMID 25282031). Cochrane (Al Saadi et al., 2021; PMID 35608922) could pool only that study for mortality. An independent 1993 trial from Naples (Morisco et al.; PMID 8241697) reported fewer heart-failure hospitalizations, which keeps the use above C. The record now names what would raise it back to A: an independent trial on current heart-failure therapy with mortality or major events as its primary endpoint. The change was prompted by an outside review of the site.

SAM-e (S-adenosyl-L-methionine) enters the database at Grade D on Route 2: tested, no benefit demonstrated. Galizia et al. (Cochrane Database of Systematic Reviews, 2016; PMID 27727432) pooled 8 randomized trials in 934 adults and found no strong evidence of a difference from placebo as monotherapy. Peng et al. (General Hospital Psychiatry, 2024; PMID 38199136) pooled 14 trials in 1,522 subjects and found no significant difference from placebo as monotherapy or as an adjunct A treatment tested on top of standard prescribed therapy rather than in place of it.. Rutjes et al. (Cochrane Database of Systematic Reviews, 2009; PMID 19821403) found a pain difference of about 0.4 cm on a 10 cm scale in knee or hip osteoarthritis, with a confidence interval crossing zero. In the three-arm trial of Mischoulon et al. (Journal of Clinical Psychiatry, 2014; PMID 24500245), neither SAM-e nor escitalopram beat placebo.

New device record, graded for one use. Two large sham-controlled trials, Gollwitzer et al. (J Bone Joint Surg Am 2015; PMID 25948515) and Gerdesmeyer et al. (Am J Sports Med 2008; PMID 18832341), found less heel pain at 12 weeks, and two meta-analyses of sham-controlled trials, Aqil et al. (Clin Orthop Relat Res 2013; PMID 23813184) and Sun et al. (Medicine 2017; PMID 28403111), agree. An independent trial, Buchbinder et al. (JAMA 2002; PMID 12234230), found no benefit and is cited on the record. Self-reported pain caps the letter at B. Shoulder, elbow, sexual function and regeneration uses are not covered.

New device record under the Grade A rule as revised in September 2026, which counts a clinician-assigned outcome grade scored blind to treatment. Low-level light applied inside the mouth during head and neck radiotherapy or stem cell transplant cut severe mucositis in sham-controlled trials from independent groups in Brazil, India and the United States (Antunes et al., Radiother Oncol 2013; PMID 24044799. Gautam et al., Radiother Oncol 2012; PMID 22884841. Hu et al., Support Care Cancer 2026; PMID 42360517), and the MASCC/ISOO guideline recommends it for prevention (Zadik et al., Support Care Cancer 2019; PMID 31286228). Not covered: treating sores already present, children, where trials found no benefit (Andriakopoulou et al., In Vivo 2024; PMID 38688628), and any US FDA clearance, since none exists for this use.

New device record, and the first letter set under the Grade A rule as revised in September 2026, which now counts a validated outcome scale scored by a clinician blinded to treatment. On the Hamilton and Montgomery-Asberg scales, sham-controlled trials in treatment-resistant depression replicate across more than 100 randomized trials (Mutz et al., BMJ 2019; PMID 30917990), including an NIH trial with an active sham (George et al., Arch Gen Psychiatry 2010; PMID 20439832). The record keeps what the letter does not cover: no trial shows fewer suicide attempts or hospitalizations, about one extra remission per 8 to 12 people treated, an effect that fades after the course ends, and a Veterans Affairs trial with no difference from sham (Yesavage et al., JAMA Psychiatry 2018; PMID 29955803).

New device record, graded for one narrow claim: early quadriceps strength and mobility after total knee replacement, compared with usual care. Strength measured on a dynamometer and timed mobility tests improved in randomized trials from independent groups in the United States, Japan and Greece (Stevens-Lapsley et al., Phys Ther 2012; PMID 22095207. Klika et al., J Knee Surg 2022; PMID 32610358), and a meta-analysis of nine trials still found a strength benefit at 12 months (Peng et al., Front Med 2021; PMID 34926522). The letter does not say stimulation beats a progressive strengthening program: in the largest trial, 200 patients, it added nothing when combined with one (Petterson et al., Arthritis Rheum 2009; PMID 19177542).

Same gap again. The record carries no letter: a test measures aerobic capacity and does not raise it, and the training that raises it is graded on its own records. It cites the methods literature on what makes a test maximal, and the validation work on the estimates consumer wearables report. Cited: Midgley et al. (Sports Med, 2007; PMID 18027991); Schaun (Sports Med Open, 2017; PMID 29218470); Carrier et al. (Sensors (Basel), 2025; PMID 39797066).

Same gap as DEXA: a published page with no record. The record carries no letter. It describes the instrument only, and says plainly that grip strength as a prognostic marker belongs with the biomarkers and that isometric handgrip training is a separate graded therapy, so a reading from this device says nothing about whether squeezing a handgrip treats anything. Cited: Roberts et al. (Age Ageing, 2011; PMID 21624928); Dos Santos et al. (Hand Ther, 2026; PMID 40626051).

The DEXA page has been published for months with no record behind it, which left its safety and medication text written by hand in the page's own HTML and invisible to every check that reads the data. The record carries no letter, by design: DEXA measures body composition and bone density and treats nothing, and the A to D scale grades treatment efficacy. It cites two measurement-accuracy studies rather than outcome trials. Cited: Schubert et al. (Clin Physiol Funct Imaging, 2019; PMID 30325573); Nickerson et al. (Int J Sport Nutr Exerc Metab, 2017; PMID 28787184).

The Energy Lounger was listed under pulsed electromagnetic field devices. It is a red and near-infrared light lounger, so it now sits under photobiomodulation. It carries no grade of its own. The modality grade rests on targeted low-level laser trials, and one randomized trial of whole-body light (Forsey et al., Lasers Med Sci 2023; PMID 37099210) found no effect on exercise performance.

New measurement entry, ungraded by design. PNOE agreed strongly with a laboratory metabolic cart on a cycle ergometer (Tsekouras et al., Front Sports Act Living 2019; PMID 33344948) and moderately with another portable system during walking (Kwok et al., J Funct Morphol Kinesiol 2025; PMID 40407443). Accuracy varies widely between metabolic systems (Van Hooren et al., Scand J Med Sci Sports 2024; PMID 37697640).

New measurement entry, ungraded by design because an analyzer is an instrument, not an intervention. The InBody 770 agreed well with a four-compartment model and with DXA for tracking, with a known systematic offset against DXA: Brewer et al. (Clin Physiol Funct Imaging 2021; PMID 33752260) and Potter et al. (BMJ Nutr Prev Health 2022; PMID 36619314).

New device record. Randomized trials by independent groups, including two with placebo insoles, and a 2026 meta-analysis of seven trials (Lin et al., Disabil Rehabil 2026; PMID 40844962) found less low back pain and disability with custom-made foot orthotics. Every outcome is self-reported, which caps the letter at B, and the effect is short term: seen at 4 to 6 weeks and gone from the largest trial (Cambron et al., Arch Phys Med Rehabil 2017; PMID 28465224) by 12. The two placebo-insole trials are Castro-Méndez et al. (Prosthet Orthot Int 2013; PMID 23327838) and Castro-Méndez et al. (Int J Environ Res Public Health 2021; PMID 34201981). Custom orthotics are fitted from a practitioner-taken impression; off-the-shelf insoles are not what this letter covers.

New record for the two-ingredient combination as studied, separate from Boswellia serrata alone, which holds its own Grade B. One randomized, placebo-controlled trial of 62 adults with knee osteoarthritis (Vaidya et al., Pharm Res 2025; PMID 39875757) found less pain and stiffness over 90 days. The trial was funded by the company that sells the ingredient and has not been repeated by an independent group, so the record holds at C.

New record. Two randomized trials have given C15:0 to people. In the TANGO trial (Chooi et al., Am J Clin Nutr 2024; PMID 38035997), 300 mg a day in soymilk lowered LDL cholesterol beyond what a weight-loss diet achieved in 88 women with fatty liver disease. In a 30-person trial of 200 mg a day (Robinson et al., J Nutr 2024; PMID 39069269), the pre-specified outcome was the blood level of C15:0, which rose; the liver-enzyme changes it reported compare responders rather than randomized arms. One surrogate result, not yet repeated, and no clinical endpoint: Grade C. The common retail serving of 100 mg is below both studied doses.

The dose field credited "Macpherson et al." alongside the trial it does cite. This record has never cited a Macpherson paper. Its vascular dose comes from Rossman et al. (PMID 29661838), which used 20 mg/day, and its other two citations are Broome (PMID 35623315) and Gane (PMID 20492507). The attribution was removed. Found in the same pass as the glucosamine attribution and for the same reason: a sweep for author names in dose and form fields that do not appear in the record's own citation list. A named author with no identifier beside it is invisible to every gate on this site that checks PMIDs, which is all of them.

The dose field credited its 1,500 mg/day figure to "the Zheng et al. mortality analysis". No Zheng paper is cited by this record and none was ever attached to it. The dose actually comes from Clegg et al. (PMID 16495392), the randomized trial the record does cite. The record's two cohort mortality citations are Li (PMID 32253185) and King (PMID 33219063), and both analyze regular glucosamine use without establishing any dose. The attribution was removed and the dose is now credited to the trial that used it. Nothing about the grade changed, because the grade never rested on the misattributed paper. This is the same class of defect as the two fabricated attributions found in the July audit: a named author standing where a citation should be, passing every check that reads identifiers because it contains none.

The record carried Grade A on the replicated hard-outcome route, and the reading published to readers said so in those words. It was not replicated. The two PMIDs on that use were REDUCE-IT (Bhatt et al., New England Journal of Medicine, 2019; PMID 30415628), positive, and STRENGTH (Nicholls et al., JAMA, 2020; PMID 33190147), null. One positive trial and one null does not meet the Grade A requirement that an effect replicate across independent research groups. JELIS (Yokoyama et al., Lancet, 2007; PMID 17398308) was assessed as the replication candidate and rejected as insufficient at this dose: 1,800 mg/day against the 4 g/day this record grades, open-label against a statin-only control, in a Japanese population not selected for elevated triglycerides. It is now cited on the record as supporting evidence rather than replication. Grade C was considered and rejected. C describes a single unrepeated trial, and two positive randomized trials in roughly 26,000 patients is not that. Grade B is the honest reading: supported by further trials, not standing alone, not replicated at the graded dose.

Resistance training was carried as THREE separate top-level records: blood pressure and lipids at Grade B, glycemic control at Grade B, and progressive overload for strength at Grade A. One intervention, three endpoints, three letters, and the glycemic record's entire evidence base was a single paper the blood-pressure record already cited. With the records split that way, nothing checked the letters against each other: gate_g13 compares a headline against the uses beneath it, and three separate top-level records slipped past that check entirely. They are now one record graded by use, the shape already used for creatine, omega-3 and CoQ10. The surviving record's headline is A, matching the site convention that a headline is no better than its best use, and the A is scoped on the record to muscle size and strength. Blood pressure with lipids, and glycemic control, remain at B as separate uses because every endpoint behind them is a surrogate. No letter changed its meaning in this consolidation: the A, and both Bs, rest on exactly the citations they rested on before.

Two uses on the consumer wearables record carried letters: sleep staging agreement against polysomnography at Grade C (de Zambotti 2019, PMID 28323455) and VO2max estimate agreement against a metabolic cart at Grade D (Caserman 2024, PMID 39083800). Both letters have been withdrawn. A letter on this site means published human randomized evidence, and an agreement study cannot be randomized: it compares one instrument against a reference standard in the same people at the same time. The record's own grade note already said consumer wearables are measurement instruments rather than interventions, so the letters contradicted the record that carried them. The findings are unchanged and remain on the record: the Oura ring agrees reasonably with polysomnography on total sleep time and poorly on staging, and the Apple Watch underestimated VO2max against the reference standard in 19 adults. What changed is that they are now stated as findings rather than graded as efficacy. Separately, 27 citations across the site were amended to name their first author, and one misattribution was corrected: PMID 33071237 was cited as Petr et al. and is Kulkarni et al.

The NMN supplement record moved to Grade C. The dosing protocol built on the same two trials stayed at B, and its own description said so out loud: one strong randomized trial, replication still needed. Yoshino 2021 (PMID 33888596) carries the insulin-sensitivity result and Yi 2023 (PMID 36482258) measured blood NAD in a different population, so neither repeats the other. Both records now read C.

This record entered the Longevity Rx list in August at Grade B on STOP-NIDDM (PMID 12086760) alone, and carried the not-replicated flag from the day it was written. Rule 03 places a single unrepeated randomized trial at Grade C. The letter is corrected. The evidence is unchanged: acarbose delayed conversion from impaired glucose tolerance to type 2 diabetes in one multicenter trial, there is no randomized human trial of acarbose on mortality or healthspan, and the mouse lifespan result that puts it on longevity lists moves no letter here.

In September this record moved from A to B, on the finding that its two citations are the China Coronary Secondary Prevention Study and that study's own elderly hypertensive subgroup, rather than two trials. The B did not follow from that finding. Rule 03 states that a single positive trial without independent replication is a Grade C, however large or well conducted it is, and CCSPS is one trial. Its size, roughly 4,870 adults after myocardial infarction, and its hard endpoints are why the trial is worth citing; neither is a substitute for a second independent group reporting the same thing. The letter is corrected to C. Nothing about the trial's result has changed, and the record still names it in full.

This record grades omega-3 by use. The use reading secondary prevention after myocardial infarction held Grade A on GISSI-Prevenzione Investigators, group-authored (Lancet, 1999; PMID 10465168), a single randomized trial of 11,324 patients reporting reduced all-cause mortality. The endpoint is hard and the trial is large, but the result has not replicated on modern background therapy. ASCEND (New England Journal of Medicine, 2018; PMID 30146932) randomized 15,480 adults to the same 1 g/day dose and found no reduction in serious vascular events. One large randomized trial on a hard clinical endpoint supports B; what is absent is a second trial of the same intervention in the current treatment era. Nothing about the 1999 result has changed. The high-dose icosapent ethyl use remains Grade A and the general-population primary prevention use remains Grade D, both unchanged. ASCEND is now recorded on this record explicitly rather than left out.

This use carried Grade C on observational evidence. PREDIMED (Guasch-Ferre et al., Journal of Nutrition, 2014; PMID 24259558) reports an association between the highest tertile of dietary magnesium intake and lower all-cause mortality, and Qu et al. (PLoS One, 2013; PMID 23520480) is a meta-analysis of prospective cohort studies An observational study that follows a group of people over time and compares what happens to them.. Nobody in either was randomized to magnesium, and both measure dietary intake rather than supplementation. EBL excludes observational data from grades under the same rule applied to the Finnish sauna cohorts and to the resistance-training mortality association, so this use is corrected to Grade D Route 1, no qualifying trial. The headline grade for magnesium is unchanged at C, which rests on the sleep use.

This record carried Grade A on two PMIDs. Both come from the China Coronary Secondary Prevention Study. Lu et al. (American Journal of Cardiology, 2008; PMID 18549841) is the trial itself, roughly 4,870 adults after myocardial infarction, and Li et al. (Journal of Clinical Pharmacology, 2009; PMID 19602720) is that same trial's elderly hypertensive subgroup. Two papers from one trial population are not independent replication, and Grade A requires replication of the same intervention in a separate trial. The letter is corrected to B, which is what one large randomized trial on hard clinical outcomes supports. Nothing about the trial's result has changed. The citation for the subgroup paper is corrected in the same edit: it was attributed to the Journal of Geriatric Cardiology and is in the Journal of Clinical Pharmacology.

This record characterized Balshaw et al. (Amino Acids, 2013; PMID 22855206) as Route 2, tested and no benefit. The trial reported the opposite: taurine significantly improved 3-km running time by 1.7%, p=0.013, in a double-blind randomized crossover design in eight trained middle-distance runners. The use is corrected to Grade C, which is what a single unrepeated trial in eight people supports, and the Route 2 designation is removed. The error was first surfaced by this repository's own FINDINGS_2026-08-x30d1.md, which flagged it and correctly declined to change the letter without adjudication.

STOP-NIDDM (PMID 12086760) randomized adults with impaired glucose tolerance to acarbose or placebo, with conversion to type 2 diabetes as the primary endpoint. One landmark trial supports B and not A under Rule 03. Acarbose appears on longevity lists because it extended median lifespan in the NIA Interventions Testing Program in mice. Animal lifespan data does not move a grade here at any effect size, and there is no randomized human trial of acarbose on mortality, healthspan, or in a healthy population.

EMPA-REG OUTCOME (PMID 26378978), DAPA-HF (PMID 31535829) and DAPA-CKD (PMID 32970396) are three independent randomized trials in three different populations, all on hard clinical outcomes. That is the replicated-hard-outcome route to Grade A. The record states the boundary in the same breath as the letter: every one of those trials enrolled people who already had cardiovascular, cardiac or renal disease. No trial has given an SGLT2 inhibitor to a healthy adult to see whether they live longer.

Metformin is the most-cited longevity drug and had no graded record on EBL. It has one now, and the letter is D. Walton et al. (PMID 31557380, the MASTERS trial) randomized 94 adults aged 65 and over without diabetes to metformin 1,700 mg/day or placebo through 14 weeks of progressive resistance training. The stated hypothesis was that metformin would augment the muscle response; it blunted it. Konopka et al. (PMID 30548390) found metformin inhibited the mitochondrial adaptations to aerobic training in the same population. TAME, the trial designed to test metformin against age-related disease, has not reported, and the observational survival signal is not randomized. So the D is Route 2: tested in humans on the actual molecule, in the target population, result adverse on the endpoint that matters.

PMID 42172592 screened 1,960 publications and found 15 randomized trials of multidimensional healthspan in 4,656 people, 11 of them testing exercise, with the supplement and caloric restriction evidence too sparse to support a conclusion. PMID 42494770 reaches the same position from dietary patterns and cognition. Cited as methodology. Neither sets a letter for any entity.

Kell et al. (Aging Cell, 2026; PMID 41524558) randomized healthy older men to 1 mg/day or placebo for four months and reported no difference in white cell counts at eight weeks. A safety and mechanism result in a small randomized sample. No randomized human trial has tested rapamycin against a clinical aging endpoint, so the grade does not move.

A positive randomized pilot on cognition in mild cognitive impairment (PMID 41310471, n=46, untreated control) and an NHANES cohort of 31,044 adults reporting higher all-cause mortality for a last meal before 19:00 (PMID 42538400) are now both stated in the grade basis. Neither moves the letter: the pilot is too small and tests a different endpoint, the cohort is observational. Grade B unchanged.

A randomized trial of 3 g/day for 8 weeks in 120 adults with type 2 diabetes (PMID 35870947) named fasting blood glucose as a primary outcome and did not improve it, nor HbA1c, nor the lipid profile. Serum insulin and HOMA-IR, secondary outcomes, did improve, as did endothelial and inflammatory markers listed alongside glucose as primary. Glycemic control is filed as D Route 2, tested and not supported on the primary endpoint. Endothelial and inflammatory markers are filed separately at C. The headline aging grade is unchanged at D Route 1.

Two phase 2 randomized trials (PMID 37366315, PMID 37385280) met their pre-specified primary endpoints of body weight and HbA1c, with large dose-dependent effects. Phase 2 trials on anthropometric and biomarker endpoints cap at B however large the effect, and no phase 3 outcome trial had reported at review. The material sold online under this name sits outside any pharmacy supply chain and none of this evidence applies to an unverified vial.

SURMOUNT-1 (PMID 39536238) reports a reduction in incident type 2 diabetes over 176 weeks against placebo, a hard clinical outcome from a single trial. SURPASS-CVOT (PMID 41406444) is a non-inferiority trial against dulaglutide, which shows tirzepatide is not worse than a drug that works rather than that it beats placebo. That combination is B, not A.

Graded A on the cardiovascular composite, not on weight loss. SELECT (PMID 37952131) and SUSTAIN-6 (PMID 27633186) are two independent randomized trials in different populations, both reporting a reduction in cardiovascular death, myocardial infarction or stroke. Weight reduction is the reason most readers arrive at this record and is not what earns the letter.

Added as a device modality separate from 40Hz Gamma Sensory Stimulation, which sits in the same category and shares none of its literature. Three meta-analyses of randomized trials carry the letter, all of binaural beats delivered through headphones: pooled g = 0.45 across cognition, anxiety and pain (Garcia-Argibay 2019, PMID 30073406), g = 0.40 on memory and attention (Basu 2023, PMID 35842538), and SMD -1.38 on perioperative anxiety at 91.6 percent heterogeneity (Xiong 2025, PMID 41176178). Every pooled endpoint is self-report or a cognitive test score, so B is the ceiling before anything else is weighed. Graded C rather than B for three reasons: a systematic review of EEG recordings during stimulation found five studies supporting entrainment and eight contradicting it (Ingendoh 2023, PMID 37205669), so the named mechanism is not established; the comparator in these trials is blank audio or music and cannot be blinded; and no trial has reported a longevity endpoint. The graded form is tones alone. No light-and-sound headset has been tested as a unit in any pooled trial, which is why the one consumer product listed under this modality carries a low delivery fidelity.

The cognitive protocol carried a Grade A resting on Prokopidis 2023 (PMID 35984306). Rereading the meta-analysis against the letter: the pooled memory effect is SMD 0.29 (95 percent CI 0.04 to 0.53) with I-squared of 66 percent, concentrated in adults aged 66 to 76 (SMD 0.88) and absent in adults aged 11 to 31 (SMD 0.03, -0.14 to 0.20). A signal that reverses to nothing across half the age range is mixed trial data. Regraded C. This was found while splitting the creatine supplement record by use, which had carried one letter across muscle strength, cognition and brain energy reserve. Creatine's Grade A is scoped to measured strength, where it rests on instrument-recorded one-repetition maximum in two mixed-sex meta-analyses, and is unchanged. The bone density use is newly graded D by Route 2 on two 2-year randomized trials that found no effect.

Curcumin now carries two graded uses instead of one headline letter. Systemic inflammatory markers stay at C: only one cited trial isolates curcumin, the other tests a curcuminoid-piperine combination, and every endpoint is a surrogate. Knee osteoarthritis pain and function is newly graded C on two head-to-head randomized trials against ibuprofen (PMID 24672232, PMID 19678780) and an 11-trial meta-analysis in which curcuminoids beat placebo (PMID 29622343). It is C rather than B for three reasons the meta-analysis states about itself: study quality was low overall, most trials enrolled under 100 participants, and the pooled arm mixes formulations. The trials used 1,500 to 2,000 mg/day of unenhanced curcuminoid extract, not the enhanced-bioavailability form at a lower dose that is commonly sold.

Probiotics is not one intervention. Prevention of antibiotic-associated Clostridioides difficile diarrhea holds Grade A on pooled randomized evidence. Inflammatory markers in type 2 diabetes is C on surrogate endpoints. The headline letter is now C, the grade of the general-purpose use, so that an A earned on one narrow clinical endpoint is not read as a verdict on the category.

The headline Grade B sat above per-use grades of C, C and D. A headline letter may not exceed the best use it rests on, so the headline is now C. The longevity, healthspan and mortality use remains D: no cited trial measures it.

The Grade B rested on two citations and both tested GlyNAC, glycine combined with N-acetylcysteine, so neither isolated glycine. The record's own reference field said so and the grade did not follow. The only placebo-controlled human study of glycine alone bearing on sleep is Bannai 2012 (PMID 22529837), which gave 3 g before bed to healthy volunteers under experimental sleep restriction and found reduced next-day fatigue and improved psychomotor vigilance. Small, unreplicated, not a population with insomnia, and PubMed assigns it no trial publication type. The widely quoted Yamadera 2007 glycine sleep study appeared in a journal PubMed does not index, so it carries no identifier and is not used. Regraded C.

A new graded protocol covers compressing the week's vigorous exercise into a single session. Siu et al. 2026 randomized 315 inactive adults with overweight and central obesity to once-weekly HIIT, thrice-weekly HIIT, or control, both training arms performing 75 minutes of high-intensity interval work per week for 16 weeks. Against control, the once-weekly arm lost 0.8 kg of fat mass and the thrice-weekly arm 1.0 kg. The comparison usually quoted from this trial, once weekly against three times weekly, is not a finding: that between-arm difference did not reach significance. Grade C on a single trial.

A new graded protocol covers the order of resistance and aerobic work inside a single session. Three independent syntheses agree on the direction: lower-body 1RM higher by 3.96 kg when resistance work came first (Murlasits and Kneffel, 2018), lower-body dynamic strength higher by 6.91 percent (Eddens and van Someren, 2018), and a pooled SMD of 0.19 across 19 randomized trials (Gao and Yu, 2023). All three report no difference in VO2max in either order. Graded B rather than A because the endpoint is performance rather than a hard clinical outcome, and the lower confidence bound sits close to zero.

The Grade B rested on four citations, none of which was a chronic randomized trial of nicotinamide riboside alone against a meaningful endpoint: a pharmacokinetic study, a cell-biology paper, an acute exercise crossover, and a repeat-dose trial of nicotinamide riboside combined with pterostilbene, which cannot isolate a single ingredient. A phase-II randomized placebo-controlled trial in older adults with amnestic mild cognitive impairment (Martens et al., Alzheimers Dement, 2026) has since reported that 12 weeks of nicotinamide riboside doubled blood NAD+ and produced no improvement in cognitive function, total cerebral blood flow, or blood pressure. Raising the marker is established. Benefit from raising it is not.

Every study named on the senolytics page was re-resolved against PubMed. The resveratrol trial in Alzheimer's disease reported no cognitive benefit and greater brain-volume loss rather than benefit (Turner, Neurology 2015), and the resveratrol blood-pressure trial reported LDL rising by 17.1 mg/dL on monotherapy (Riche, 2014). Five further citations named the wrong year, journal, or paper. Every claim on the page now carries a PubMed link.

The dose field named a single daily dose and described the studied range as 250 to 500 mg. No cited trial used 500 mg, and the range was wrong in both directions. The baker's yeast beta-1,3/1,6-glucan trials used 250 mg per day; the insoluble brewers' yeast preparation trials used 900 mg per day. The field now states that there is no single studied dose, names each preparation with its own dose and duration, and records that the two preparations are not interchangeable.

97 supplement pages carried a sentence saying that EBL might be paid when a reader bought through the link below it. That was not true. EBL is paid nothing on any item carrying an evidence grade, and its only product income is device hardware, which carries no grade. The sentence had been accurate when the site was built and nobody revisited the pages when the economics changed. A build gate now fails on any wording that asserts or implies payment on a graded item.