Don't get fooled.

GBD 2018 + Mendelian randomization rule out a J-curve. Any dose raises cancer + all-cause mortality.

Therapeutic doses need >1 L wine/day; real intake is rounding-error noise.

Five UPF mechanisms (reward/hormones/microbiome/eating speed/food matrix) yield a real calorie gap (Hall 2019).

TRE matches plain calorie restriction; no metabolic uplift. Harmful for pregnancy or eating-disorder history.

Monacolin K is chemically identical to lovastatin — 'natural' brings dose variance + citrinin contamination.

HbA1c drop is metformin-tier (~0.7-1%) — nowhere near GLP-1 RA pharmacology.

STEP-1 indication is BMI≥30 (or ≥27+comorbidity); ~2/3 weight returns within a year of stopping (STEP-4).

Sumithran 2011 NEJM: one year after losing 10 kg, ghrelin stays elevated and PYY/leptin stay suppressed — defensive hunger does not normalize.

Lichtman 1992 NEJM: self-reported 'diet-resistant' adults underestimate intake by ~47% and overstate exercise by ~51%. Individual RMR varies only ±200 kcal — far smaller than the reporting bias.

Meal replacements are portion-control tools, nothing more. Stop using them and weight returns; Heymsfield 2003 meta shows no advantage over isocaloric diet.

Hall 2015/2021 metabolic-ward RCTs: at equal calories low-fat lost more body fat than low-carb; ad libitum, the low-fat plant arm ate 689 fewer kcal/day. The carb-insulin model fails in controlled testing.

Timing has no independent effect; total energy decides fat accrual. Sofer 2011 actually showed shifting carbs to dinner improved satiety hormones and weight loss in a small RCT.

One day of overfeeding cannot reverse months of leptin and RMR adaptation; it usually just erases the week's hard-won deficit.

Global BMI Mortality Collaboration Lancet 2016 (n=10.6M): BMI shows a robust J-shaped curve with all-cause mortality. Imprecise at the individual level, strongly informative at the population level.

Heymsfield 1999 JAMA: leptin is already elevated in common obesity (resistance), so exogenous leptin is ineffective. Only the rare congenital leptin-deficiency patients respond.

STEP 1 extension (Wilding 2022): ~2/3 of weight is regained within a year of stopping — effectively lifelong therapy. ~70% have GI side effects and lean-mass loss is elevated.

Trepanowski 2017 JAMA IM: alternate-day fasting matched standard CR at 12 months (−6.0% vs −5.3%, NS) but had worse adherence and higher binge-cycle risk.

An hour of running burns ~500 kcal; a single doughnut is ~400 kcal. Diet drives weight loss (Donnelly 2013 ACSM position); exercise's main value is lean-mass retention, cardio fitness, and weight maintenance.

Routine salt tablets are unsupported and can be counterproductive for most exercisers; sodium is covered by food + electrolyte drinks, and only very salty, high-intensity sweaters need individualized sodium (NATA 2017).

Gritting through is exactly what breeds heatstroke; the right approach is acclimatize first, watch WBGT, drink to thirst, pre-cool, and heed the body's signals (Racinais 2015).

The liver and kidneys already do this. 'Detox' programs have no human clinical evidence; short-term weight loss is mostly water and bowel contents.

GAIT 2006 NEJM primary endpoint was negative; moderate-severe subgroup didn't clear pre-specified analysis.

Oral peptides are degraded to amino acids in the gut; this is the 'eat pork knuckle to feed your skin' fallacy.

Fiatarone 1994 NEJM: 87-year-olds gained +174% strength in 8 weeks of progressive PRT. Replicated for 30+ years.

Testosterone delta is only ~14-15% in RCT meta — still inside normal physiological variation.

Real mechanism is dampening cortisol overshoot — a subset of what sleep + exercise + nutrition already do.

Acute dose shows no robust cognitive benefit in healthy adults; long-term human data is thin.

T uplift is on the same order as Ashwagandha (~10%) — within physiological variation, far below marketing claims.

Human RCTs only show NAD+ blood levels rise — zero data on clinical endpoints (lifespan / healthspan).

IV stops at the same surrogate (blood NAD+); zero clinical endpoint RCTs, plus infection + allergy risk.

0.3 mg is the physiological equivalent; 5-10 mg overloads receptors → next-day grogginess + rhythm disruption.

Main effect is attenuating caffeine jitter; no robust evidence as a standalone insomnia treatment.

NGF upregulation is in cell + rodent models only; human cognition data is small short-term RCTs.

CARET 1996 was halted early — β-carotene supplementation raised lung cancer in smokers. Antioxidant ≠ safe.

Contains pseudo-B12 that cannot activate human B12-dependent enzymes — may even displace real B12 at receptors.

Hemilä 2013 Cochrane (n>11,000): ≥200 mg/d doesn't reduce incidence; only shortens duration ~8%.

PHS-II: zero on cardiovascular, total cancer down just 8%. The gap was never a pill.

Curhan 1993 NEJM: low-calcium diets actually raise oxalate stone risk by leaving more free oxalate in the gut.

1000 mg is total fish oil weight; EPA+DHA is usually 300-500 mg. Read the actives, not the headline.

USPSTF 2022 withdrew the 60+ recommendation — bleeding risk outweighs CVD benefit (net harm).

Poor absorption plus rapid phase II conjugation means almost no free compound reaches hepatocytes; Cochrane pooling found no effect on mortality or complications, and SyNCH at 3-5x customary dose was no better than placebo.

Transaminases are enzymes that leak from ruptured hepatocytes — a leakage reading, not a function reading; in end-stage disease with nothing left to leak it reads normal, and it fluctuates on its own, so any before-after pair of draws may show a fall.

No supplement has been shown to offset ethanol's liver injury, and in alcoholic liver disease milk thistle showed no significant effect on mortality or complications; treating it as permission only postpones the one thing that works — drinking less.

PPAR-γ signal only in vitro; human RCT effect << metformin, and stacking with antidiabetics raises risk.

Measured broth has ~26-80 mg calcium/L — a bowl is <5% of daily need vs ~300 mg in a cup of milk (7-30x more); what renders out is mostly fat.

Jujube iron is non-heme, low and poorly absorbed; the 'nourish blood' signal is only cell/animal, with no human evidence it corrects iron-deficiency anemia.

Bird's nest is mostly glycoprotein, e-jiao is basically collagen/gelatin; claimed benefits come mostly from lab and animal work, with little quality human evidence.

No scientific basis and harmful: with produce avoided, vitamin A and C intake fell below half the recommended amount; the postpartum body needs vitamin C, folate, and fiber.

Sweat heavy-metal concentration is 100× lower than urine; total 'toxin' excretion <1% via sweat.

Oral HA is broken down in the gut; topical / injected delivery is the only route with skin endpoint data.

IR + androgens is a two-pronged mechanism; sugar-cutting only addresses one. 5-10% weight loss + myo-inositol matters.

MHT moves bone density + vasomotor + GSM together; phytoestrogens have scattered effects and don't substitute.

H₂ solubility in water is tiny (~1.6 mg/L); a 500 mL bottle delivers ≤0.8 mg — orders of magnitude below animal dosing.

Suez 2018 Cell: generic probiotics delayed post-antibiotic recovery 3-6 months. It's a strain × indication × dose match.

WHO 2023 advises against NSS for weight; IARC lists aspartame 2B; Suez shows sweeteners worsen glucose tolerance via microbiome.

Stable anti-inflammatory signal comes from whole dietary patterns (Mediterranean / DASH), not 'avoid nightshades' lists.

ORAC in-vitro ranking doesn't translate to body endpoints; clinical (CVD / lifespan / skin aging) data is thin.

Originated from a 1960s Japanese pedometer ad; Lee 2019 + Paluch 2022 show benefit plateaus at ~7000-8000 steps.

Lactate clears in 1-2h; DOMS is eccentric-loading microdamage + inflammatory cascade peaking 24-72h post.

Vispute 2011: six weeks of ab training didn't shift abdominal fat. In Kostek 2007 the localised difference showed up only on skinfold callipers — MRI saw generalised loss.

Higher fat % at low intensity ≠ more fat burned in absolute terms; total expenditure + EPOC favor high intensity.

Knees travelling past toes is normal squat biomechanics; restricting it shifts load to hips + low back.

Hypertrophy is driven by progressive overload + total volume; exercise variety ranks far below intensity/frequency.

Lactate is a fuel, not waste; the 'burn' comes from H+ accumulation co-produced — correlation, not causation.

Subjective recovery ≠ objective recovery. Van Dongen 2003: when participants self-rate as 'adapted', reaction times still sit 30-40 % below baseline.

Receptor ceiling is 400 mg. What coffee pushes through is the feeling of sleepiness, not the underlying impairment — working memory + error rate stay degraded.

NREM3 rebounds that night, REM not until night 2-3; metabolic + immune markers need 3-7 days. Subjective feeling lies.

Doesn't clear adenosine, doesn't produce NREM3, doesn't clear β-amyloid. Better than scrolling — not a substitute for sleep.

300-400 mg caffeine + sugar + taurine + guarana → FDA MedWatch has repeatedly logged acute arrhythmia and hypertension events.

Sympathetic spike lasts 5-15 min, then a deeper crash follows. A stopgap when a nap is impossible — not a substitute.

Strength −5-11 %, proprioception dulled, injury rate up — PR day becomes injury day. The flip side works: sleep banking the week before is real (Rupp 2009).

Selective memory. Chronic short sleep is linked to raised T2D, cardiovascular, dementia, and depression risk — epidemiology's long tail is not friendly.

Water has zero calories and cannot become fat. A 1-2 kg daily swing is water: glycogen binds 3-4× its weight (Olsson & Saltin 1970) + sodium retention + hormones. Watch the 2-week trend.

Heritability is a population statistic, not a personal verdict. Genes give a tendency, not an endpoint; the same genome yields very different weights in different environments (Loos & Yeo 2022).

Lacks human evidence. PUFA replacing saturated fat lowers CHD (Mozaffarian 2010); the appetite hypothesis is mouse-only (Alvheim 2012). The real issue is the 'fried + packaged' food form, not the molecule.

Chronotype is heritable (Jones 2019, 351 loci), but being innately late ≠ staying up is free. Greater social jet lag tracks higher BMI in the overweight (Roenneberg 2012). Sleep enough, aligned to your rhythm.

'Superfood' has no scientific definition; its core ORAC in-vitro basis was withdrawn by USDA in 2012, and human-endpoint evidence for any single star food is sparse.

Smith-Spangler 2012 (Stanford, 237 studies): nutrient content is near-identical to conventional. 'Organic' certifies production method, not nutrition or safety.

The liver (Phase I/II) and kidneys clear metabolic waste 24/7; Klein & Kiat 2015 review found no rigorous RCT support — short-term weight loss is just water and bowel contents.

Blood pH is locked at 7.35-7.45 by lungs and kidneys, and gastric acid neutralises the water first; Fenton 2016 (BMJ Open) systematic review rules out an acid-base/cancer link.

Originates from a 1968 uncontrolled letter; Geha 2000 double-blind challenge found only 2.3% responding, non-reproducibly — glutamate is the same molecule as in tomato and breast milk, safe at dietary doses.

Johnson & Fritsche 2012 (15 RCTs): linoleic acid doesn't raise CRP/IL-6/TNF-α; Mozaffarian 2010: PUFA replacing saturated fat cuts CHD ~19%. The real issue is the fried/ultra-processed form, not the molecule.

Gluten-free is a medical fact for celiac disease (~1%), but Lebwohl 2017 (BMJ, 110k) shows no cardiovascular benefit for those without it — and gluten-free packaged foods are often lower-fiber, higher-sugar.

Liquid sugar bypasses satiety, stacking calories instead of replacing; Imamura 2015 (BMJ, 17 cohorts / 380k): one daily SSB serving raises T2D risk 18-26%, with 100% juice independently associated too.

IARC Group 1 measures certainty of evidence, not magnitude; Bouvard 2015 (Lancet Oncol): 50 g/day processed meat carries colorectal RR≈1.18 vs smoking's lung-cancer RR≈15-30.

Oral collagen is broken into amino acids in the gut, not shipped intact to skin or joints (Paul 2019); hydrolyzed peptides show a weak signal, but Myung 2025 found no effect in non-industry-funded studies — and bone broth's dose is a blind box.

Charcoal only adsorbs just-swallowed poison within ~1 h in the ER (Chyka 2005); taken daily it never reaches the blood and instead indiscriminately soaks up drugs (incl. oral contraceptives) and nutrients — its whitening is enamel abrasion.

Coconut oil is ~82% saturated and its lead fatty acid lauric (C12) raises LDL; Neelakantan 2020 (Circulation, 16 trials) shows it raises LDL vs vegetable oils and burns no fat — AHA 2017 advises against it.

No human trials for anti-aging/brain claims, the antioxidant pitch rests on withdrawn in-vitro ORAC data, and the FTC fined açaí weight-loss false advertising ($1.5M settlement). They are just decent ordinary fruits.

The mechanism is real (mevalonate pathway), but a 6-RCT / 302-person meta shows CoQ10 doesn't reliably relieve the myalgia (Banach 2015).

PPI → calcium/iron is plausible but weak and debated; what's solid is long-term-use hypomagnesemia (FDA 2011) and low B12.

From a 1960 plastic-surgery observation, not a habit study; the real median is ~66 days, range 18-254 (Lally 2010).

The ego-depletion model failed a 23-lab preregistered replication (d=0.04, Hagger 2016); design your environment, don't white-knuckle.