The grade table

Every graded claim in the register. The letter is always present, never colour alone.

InterventionClaimGradeIn one line
2-deoxyglucoseTreating cancerDIt reached people at a tolerable amount and produced no efficacy signal.
AlbendazoleTreating cancerDMinimal human cancer data.
AlcoholLimiting or avoiding itBAn established human carcinogen for seven cancers, and cutting down lowers the risk.
Antibiotic timingAvoiding unnecessary courses around immunotherapyBConsistent observational evidence, and it already changes practice.
Apricot kernels, laetrileTreating cancerXTested in 178 people, no benefit, and it releases cyanide.
AspirinColorectal cancer in Lynch syndromeBA double-blind randomised trial in an inherited condition, with a real reduction.
AstragalusAdded to chemotherapy in lung cancerCMany randomised trials from one region, poor quality by the authors' own account.
BerriesTreating cancerDThirteen completed trials, and not one measured what a cancer did.
Beta-blockersTreating cancerCNull overall, with a replicated signal in one breast cancer subtype.
Beta-carotenePreventing lung cancerEA trial stopped early: more lung cancer and more deaths in the supplemented group.
Black salveTreating skin cancerXIt burns. In the dish, a normal cell died at a lower concentration than any cancer line.
Broccoli sproutsTreating cancerDFourteen completed trials in people, and every one of them measured a marker.
Cachexia drug treatmentTreating cachexiaCNo single agent established. The problem itself is real and well understood.
CannabinoidsChemotherapy-induced nausea and vomitingBPooled randomised evidence, against comparators that modern practice has moved past.
CannabisAnxiety and sleepCWidely reported by people using it. Little cancer-specific trial evidence.
CannabisAppetite and weightCIt increases appetite. It has not been shown to preserve muscle.
CannabisCancer painCMixed randomised results, and several trials missed their primary endpoints.
CannabisTreating the tumourDThe laboratory concentrations are not reachable by any route a person can use.
CoffeeNo action neededCRemoved from the possible-carcinogen list in 2016.
Consumer microbiome testsGuiding treatmentDSame sample, two companies, opposite verdicts. No approved clinical test exists.
Culinary mushroomsTreating cancerDMechanism, and some small studies. Nothing measuring what a cancer did.
Cutting sugarStarving an existing tumourDBlood sugar is held steady by the liver whatever you eat.
DairyEither directionCLower bowel cancer risk. The prostate signal was downgraded in 2018.
DichloroacetateTreating cancerDTarget engagement shown in human tumour tissue. No controlled efficacy trial.
Dietary fibreImmunotherapy responseCObservational, in melanoma, with a randomised trial registered.
Eradicating H. pyloriPreventing gastric cancer in infected peopleAA cheap generic treatment that prevents a cancer, with randomised evidence behind it.
EssiacTreating cancerDNo human benefit data, and animal and cell work suggesting tumour promotion.
ExerciseAnxiety and depressive symptomsBRandomised evidence, with specific prescriptions attached to it.
ExerciseDisease-free survival, resected colon cancerAA phase 3 randomised trial met its primary endpoint over nearly eight years.
ExerciseFatigue during and after treatmentAA large and consistent randomised literature, written into guidelines.
ExercisePhysical function and fitnessBGood human evidence, held at one remove: the trials behind it are not on this page.
ExerciseQuality of lifeARandomised evidence, across a range of tumour types.
ExerciseSurvival in cancers other than colonBConsistent observational evidence. The trials are running.
Faecal transplantImmunotherapy-refractory melanomaCEarly-phase responses, and it is being tested properly.
Fasting and FMDAlongside chemotherapyCThe declared endpoints were missed. Two secondary findings survived.
FenbendazoleTreating cancerDA dog wormer. Nobody has measured what it does in a person.
Fermented foodsTreating cancerDSix completed trials, none measuring what a cancer did.
FolatePreventing colorectal adenomasEThe folic acid arm had more advanced lesions, and more people with several of them.
GarlicPreventing gastric cancerCA randomised trial: fewer deaths, no fewer cancers, in a very high-risk population.
GingerTreatment-related nauseaCRandomised trials mostly null for sickness. Nothing at all for the tumour.
Graviola, soursopTreating cancerXIts acetogenins are linked to an irreversible parkinsonian syndrome.
Green tea, EGCGTreating cancerDNinety-one completed trials, and none isolating green tea against a cancer outcome.
Insomnia programmePersistent insomniaANineteen randomised trials, 1,803 cancer survivors, against real control conditions.
IvermectinTreating cancerDTrials are registered and recruiting. None has reported.
Ketogenic dietTreating cancerCRandomised data exists, and the trial that could have shown a benefit missed its endpoint.
Limiting added sugarMaintaining a healthy weightBThe link that holds is between body fatness and cancer, not between sugar and cancer.
MebendazoleTreating cancerCOne small positive randomised trial, and no trial with a mebendazole-free comparator.
Mediterranean patternPreventing cancerBRandomised evidence, on a secondary endpoint, in a trial that was later corrected.
Mediterranean patternTreating existing cancerCEvidence that a diet prevents a cancer does not transfer to treating one.
MelatoninSleepCTwo small randomised trials in breast cancer, both favourable for sleep.
MelatoninTreating cancerDVery large effects, from one unblinded source, never replicated elsewhere.
MetforminTreating breast cancerDA trial randomised 3,649 patients and found nothing.
Methylene blueTreating cancerDNo human trial, and a documented interaction that has been fatal.
Milk thistleProtecting the liver during treatmentCOne small randomised trial, with a borderline result at delayed follow-up.
Mindfulness approachesAnxiety and depressive symptomsBRandomised, modest effects, mostly in breast cancer, durability unclear.
MistletoeQuality of lifeCA randomised quality-of-life signal. No difference in recurrence or mortality.
Multi-compound protocolsTreating cancerDThe combination has no evidence of its own at all.
MultivitaminGeneral useCNo harm signal in the chemotherapy study, and no benefit either.
NiclosamideTreating cancerDEarly-phase human work only, and it is poorly absorbed by design.
Nutrition and resistance trainingMaintaining muscle during treatmentBHuman evidence, stronger for how people function than for how long they live.
Positive attitudeProlonging survivalDNo evidence, and a documented cost to the people asked for one.
Probiotic supplementsImproving immunotherapy responseDThe signals point the wrong way in the setting studied most.
Processed meatLimiting itBAn established cause of bowel cancer. The size of the effect is modest.
Psychological supportProlonging survivalDThe replication trial found 17.9 months against 17.6.
Red meatLimiting itCA probable carcinogen, and probable is doing real work in that sentence.
Reishi and othersTreating cancerDMechanism, and some small studies.
Seed oilsCausing cancerDPeople looked. The association the claim depends on is not there.
SeleniumPreventionCThe effect appears to depend on how much you had to begin with.
Soy foodsSafe to eat after a breast cancer diagnosisBThe feared harm was looked for repeatedly and not found, including alongside tamoxifen.
StatinsTreating cancerDPooled randomised trials give a hazard ratio of 0.94, with an interval crossing one.
Talking therapyDistress and moodBRandomised evidence, large after treatment, and no effect left at twelve months.
Treating major depressionMajor depression in cancerAA structured programme: 62 percent response against 17 percent in usual care.
Turkey tail (PSK)Adjunct in gastrointestinal cancerCApproved in Japan, with systematic review certainty rated very low.
Turmeric, curcuminTreating cancerDThe bioavailability problem is the whole story.
Vitamin B12Taken through chemotherapyEAssociated with worse disease-free and overall survival when taken through chemotherapy.
Vitamin C, intravenousTreating cancerDThe concentration is reachable. Nobody has run the outcome trial.
Vitamin C, oralTreating cancerDTwo randomised trials found nothing, and the oral ceiling explains why.
Vitamin DCorrecting a measured deficiencyCNull for cancer incidence. The mortality finding is contested.
Vitamin EPreventing prostate cancerESELECT found more prostate cancers in the vitamin E group than on placebo.