MED01-P2 — Plain-English Claim Ledger (v1.0 — SIGNED OFF by Danny 2026-10-12; P2 LOCK 2026-10-14; corrections #5–#16 applied 2026-10-13/14 after the cross-audit)

What this is: the same 45 claims as the technical ledger (Part 2, v1.0, signed off 2026-10-10, with correction #4), written for a general reader. The language is simpler; the uncertainty is not. Every row keeps the same claim number, so you can look up the technical version and follow it back to the source. Period covered: 2000–2017, as the evidence stood then, found through our search cutoff of 20 September 2026. Anything that happened after 2017 is labelled later. Not medical advice. These are statements about studies and the conditions they were done under. None of them says what cannabis does for any person today.

How to read “How sure”: it describes how confident we are that the row correctly reports its source. It does not rate how strong the evidence is for a treatment: a Solid row can describe a small, uncontrolled or negative study. For the strength of the evidence, read the row itself and the official reviews (C078–C081).

Solid: the source says exactly this and we read it.
Solid, with limits: true as stated, but the study was small, or we could only read the summary (abstract), or one part is still being checked.
Open: we could not settle it from the sources we found.

Words used here:

A randomized trial assigns people to treatment or a dummy (placebo) by chance.
A trial’s main outcome is the result it was designed to test, decided before it started. Other results are side results and count for less.
An association means two things go together in the data. It does not show that one causes the other.

How much was published

#	In plain English	Who this does not tell us about	How sure
C041	Our standard PubMed search finds about 4,300 cannabis papers for 2000–04, 7,500 for 2005–09, 10,300 for 2010–14 and 8,600 for 2015–17, about 30,700 in all. Output rose throughout, fastest in the last three years.	Research PubMed doesn’t index well, such as non-English journals and conference talks. These counts describe the database, not how much work was done.	Solid, with limits
C042	A later study counted medical-cannabis papers growing about nine-fold from 2000 to 2017, against about 2.5-fold for all PubMed papers. The most-studied conditions were HIV, chronic pain, MS, nausea and epilepsy.	The exact multiples depend on that study’s own search words.	Solid, with limits
C043	Reading the record, the period has three turning points. The first big trial programmes and first approval came in 2003–07. A weight-loss drug rose and was withdrawn in 2006–09. The CBD-for-childhood-epilepsy story, and a major US national review, came in 2013–17.	Exact dates. This is our reading, not a measured fact.	Solid, with limits

Lab and animal science

#	In plain English	Who this does not tell us about	How sure
C044	Lab and animal studies showed the body’s own cannabis-like chemicals carry signals backwards across nerve connections. Mice lacking an enzyme that breaks one of them down had more of it. That made the enzyme a drug target.	People. This was rat brain slices and mice.	Solid
C045	A 2015 lab study reported that CBD changes how the main cannabinoid receptor responds.	People. This was cells in a dish, and it was one of several proposed explanations.	Solid, with limits
C046	A 2004 idea, “clinical endocannabinoid deficiency”, proposed that migraine, fibromyalgia and irritable bowel syndrome come from too little of the body’s own cannabinoids. By 2017, reviews had collected some indirect supporting evidence, but our searches did not identify an independent test confirming it.	Whether the idea is true. We only record that it was still unproven in 2017.	Solid, with limits
C047	When healthy users compared whole-plant cannabis with matched pure THC, the “high” felt about the same. The one patient trial comparing them, in cancer weight loss, found no difference between plant extract, THC and placebo. Lab studies in volunteers found CBD can blunt some THC effects, such as paranoia. The claim that plant compounds work better together (“entourage effect”) was put forward as an idea.	Patients’ symptoms. Most of these studies measured how people felt in a lab. No study tested whether whole-plant medicine treats illness better than a single compound.	Solid, with limits

What was tested in people

#	In plain English	Who this does not tell us about	How sure
C048	MS muscle stiffness was one of the most-tested uses. In the biggest early trial (630 patients), THC and cannabis extract did not improve stiffness as measured by doctors, the trial’s main outcome, but patients said they felt less stiff. Later spray (nabiximols) trials were mixed. A trial that first picked out early responders found a clear benefit, but that applies to responders only. Doctor-measured and patient-reported results did not consistently match in these trials.	People whose MS stiffness did not respond early; long-term use; other conditions. Whether patients could tell they were on the active drug is still being checked for one trial.	Solid, with limits (abstract only)
C049	For nerve pain in MS, small trials of dronabinol (a THC pill) and the spray reported less pain. But the largest dronabinol trial (240 patients) found no difference from placebo on its main outcome. It finished in 2009 and was published in 2017, with a summary that led on safety.	People with other kinds of pain.	Solid, with limits (abstract only)
C050	Short trials of smoked or vaporised cannabis in nerve pain reported less pain over days to weeks. In one, pain fell by about a third on cannabis against about a sixth on placebo. In another, the difference was small (0.7 points on a 0–10 scale). Pooling five trials suggested about one person in five or six got short-term relief.	Long-term use; people new to cannabis (many volunteers had used it before); other strengths of cannabis. The trials were small, and one was registered only after it ended.	Solid, with limits (small; abstract only)
C051	Four spray (nabiximols) trials in cancer pain and other nerve pain gave mixed results. One found the spray beat placebo on its main outcome while THC alone did not. One found benefit for a type of nerve pain where light touch hurts. One, in diabetic nerve pain, found no difference. One missed its main outcome, although some side results looked better at lower doses.	Other products, or cannabis in general.	Solid, with limits (abstract only)
C052	Nabilone (a man-made THC-like drug) relieved nerve pain less well than a weak opioid painkiller, with more side effects. In a 40-patient fibromyalgia trial it improved pain on the main outcome, with more side effects. In another fibromyalgia trial (31 people) it improved sleep more than the antidepressant amitriptyline. It was slightly better for how rested people felt after sleep, but not for how alert they felt on waking; it did not help pain or mood, and it caused more side effects. A 2016 expert review (Cochrane) found no convincing evidence for nabilone in fibromyalgia, and many people didn’t tolerate it.	Smoked or plant cannabis. These were pills of one synthetic compound.	Solid, with limits (small; abstract only)
C053	For nausea from chemotherapy, the period mostly re-analysed trials done before 2000. In the one new comparison with a modern anti-nausea drug (ondansetron), dronabinol and ondansetron had similar response rates for delayed nausea (54% vs 58%), and combining them added nothing. The trial was small and not designed to show they were equal.	Patients on today’s anti-nausea combinations, which most of the older trials did not use.	Solid, with limits (abstract only)
C054	For weight loss and poor appetite in cancer, two large trials found no advantage. A standard appetite drug (megestrol) worked better than dronabinol (469 patients), and cannabis extract, THC and placebo did not differ (243 patients). In HIV, a small lab study found dronabinol and smoked cannabis both increased eating, and a three-week trial found no worsening of HIV viral load, immune-cell counts or HIV-drug levels. A 2013 expert review found the evidence for HIV lacking.	Other conditions; long-term use.	Solid, with limits (abstract only)
C055	CBD for severe childhood epilepsy developed step by step. A 2014 expert review found too little evidence. Parent surveys and clinic records then described children’s experience, without comparison groups. An open programme with no comparison group enrolled 214 children and young adults (aged 1–30). In the 137 analysed for seizures, motor seizures fell by a median 36.5%. Among the 162 analysed for safety, 79% had side effects and 30% had serious ones. Without a comparison group, it could not show how much of the drop was due to CBD. The first randomized trial (Dravet syndrome, 120 patients, paid for by the company) reported monthly seizures falling from about 12 to 6 on CBD against about 15 to 14 on placebo, with more side effects on CBD. CBD also raised blood levels of another epilepsy drug (clobazam).	Adults; other epilepsies; CBD products other than this purified medicine. Later: the US approved this CBD medicine in 2018.	Solid, with limits (one trial; abstract only)
C056	Other conditions had only single small trials. In one small trial in acute schizophrenia, CBD and the standard drug amisulpride produced similar improvement, with fewer side effects on CBD. Whether that was the trial’s main outcome could not be confirmed from the summary we read. One trial does not show that CBD works as well as standard treatment in general. Small THC trials in Tourette syndrome found too little evidence, according to an expert review. In social anxiety, a single CBD dose before a public-speaking test was an experiment, not a treatment trial. Nabilone for PTSD nightmares was a small pilot. In dementia, low-dose THC did not help on the main outcome.	Treatment of any of these conditions. Each is one small study.	Solid, with limits (small; abstract only)
C057	In Crohn’s disease, a 21-patient trial of smoked cannabis missed its main outcome (remission) while reporting that more patients improved on cannabis. A later low-dose CBD trial found no benefit.	Other bowel conditions; whether the trial’s registered main outcome matched its published one, which is still being checked.	Solid, with limits (small; abstract only)
C058	In glaucoma, one 5 mg THC dose under the tongue lowered eye pressure for about two hours in six patients. A 20 mg dose of CBD did not lower it, and a 40 mg dose briefly raised it.	Long-term glaucoma control. This was one dose in six people.	Solid, with limits (small)
C059	The only study found that gave a cannabinoid to people as a cancer treatment was a nine-patient safety study in brain cancer. It was not designed to show whether the treatment worked, and it drew no such conclusion. The other human cancer-treatment reports our searches found were single case reports, not controlled trials.	Whether cannabinoids treat cancer in people. That was not tested.	Solid
C060	The one large trial testing whether a cannabinoid could slow a disease (498 people with progressive MS, three years of THC) found no effect. Neither AI researcher found it; our own check did.	Other diseases.	Solid

Products and approvals

#	In plain English	Who this does not tell us about	How sure
C061	The studies used different products: a THC+CBD mouth spray, cannabis extract pills, THC pills, nabilone, purified CBD, CBD oils of varying content, and smoked or vaporised research cannabis. Results for one do not carry over to another.	“Cannabis” in general. No single product stands for all of them.	Solid
C062	Canada gave the mouth spray Sativex conditional approval for MS nerve pain (2005) and cancer pain (2007). The UK approved it for MS stiffness on 16 June 2010, and Spain on 20 July 2010. Later: in 2019 Canada removed both conditional uses because follow-up studies did not confirm benefit.	Other products. The exact UK day, and the Australian and Spanish records, are still being checked.	Solid, with limits
C063	In the US, a liquid THC medicine was approved in 2016. No plant-derived cannabis medicine was approved by the FDA before 2018. A 2006 FDA action on nabilone was a label change, not a new approval.	Approval is not proof that a medicine works for other uses.	Solid
C064	Drugs aimed at the body’s cannabis system went badly. A weight-loss drug (rimonabant) was approved in Europe in 2006 and withdrawn in 2009 over psychiatric risks. One enzyme-blocking drug failed in arthritis pain. In a 2016 early safety trial of another enzyme-blocking drug, three of the four participants described in a published report developed a severe, worsening neurological illness; one was declared brain dead. The French government’s inquiry reported one death and four other volunteers with severe neurological problems (we have not yet read that report ourselves).	Cannabis itself. These were different drugs that act on the same body system.	Solid
C065	Patents were granted for cannabinoids as antioxidants (US government, 2003), CBD for some seizure types (2015, 2016) and CBD for Type II diabetes (2011).	Whether any of these uses works. A patent is a claim of invention, not evidence. Both AI researchers misdescribed at least one patent.	Solid

How good the evidence was

#	In plain English	Who this does not tell us about	How sure
C066	There were many controlled trials, but most were small and short. A 2015 review counted 79 trials with 6,462 people, and only four were at low risk of bias. Whether patients could tell they were on the active drug was a known worry. One analysis of three spray trials found no sign that many could.	Every trial. Quality varied.	Solid, with limits
C067	We checked a trial registry (re-checked 13 October 2026). Of the completed cannabinoid trials, 132 had neither posted results nor a linked paper. Searching PubMed by trial number found papers for 44 and nothing for 88. Of those 88, 28 gave a cannabinoid to patients as a treatment; the rest were healthy-volunteer, drug-level or dependence-treatment studies. Searching by title, researcher, sponsor and design matched 17 of the 28 to a paper. One was a 72-child spray trial in cerebral palsy that finished in 2017 but was only published in 2020. For the other 11, no publication was identified in our searches, so their status is unresolved. That does not prove they were never published. Two other registries were not searched. (Our first check used a looser filter and missed the 2020 paper; corrected.)	Whether results were hidden on purpose. This is a disclosure check, not an estimate of missing studies.	Solid, with limits

The rules the research ran under

#	In plain English	Who this does not tell us about	How sure
C068	In the US, research cannabis came from one government-contracted source throughout. A court challenge for a second grower failed in 2013. The government said in 2016 it would accept more growers, but none was added before 2018. Commentators and the 2017 national review named these rules as barriers. A California state-funded research centre still ran trials.	How much research was lost. That cannot be measured from these records.	Solid, with limits
C069	Some national medical-cannabis programmes produced data. Dutch pharmacy records show 5,540 patients on prescription cannabis in 2003–2010. In a one-year Canadian observational study, medical-cannabis users reported more non-serious side effects than a comparison group, with no difference in serious ones. That comparison does not show cannabis caused the events.	The exact legal start dates of these programmes, which are still being checked.	Solid, with limits

Harms

#	In plain English	Who this does not tell us about	How sure
C070	Long-term studies found heavier cannabis use went together with later psychosis, and more use went with more risk. In Swedish conscripts, using more than 50 times was linked to about six to seven times the odds of schizophrenia. A 2017 national review called the link substantial.	Whether cannabis causes psychosis. These studies show an association, not cause.	Solid, with limits
C071	In a New Zealand study following about 1,000 people from birth, those who used cannabis heavily and persistently from their teens showed declines in thinking skills by age 38. A re-analysis argued that other life factors could explain it. The often-quoted “8 IQ points” is in the full paper, but only for the 23 people who started in their teens and kept using heavily to 38. The heaviest users overall lost 5–6 points.	Occasional users; adult starters. Cause is not established.	Solid, with limits
C072	In a small study of 18 users and 12 former users, stopping heavy use was followed by withdrawal symptoms: irritability, poor sleep and low appetite. They began within 1–3 days, peaked around days 2–6, and mostly lasted 4–14 days. A later review recognised cannabis withdrawal as a clinically significant syndrome. Cannabis dependence was treated in trials. A 2014 expert review found no medicine with good evidence.	Occasional users.	Solid, with limits (abstract only)
C073	Recent cannabis use went with about double the odds of a car crash across nine studies. In a driving simulator, moderate THC blood levels caused lane-weaving similar to legal-limit alcohol.	A safe driving level. None was established.	Solid, with limits (abstract only)
C074	People who stopped smoking cannabis saw their bronchitis symptoms clear. Light smoking was not linked to worse lung function in one large study. In an observational study of 3,882 heart-attack patients, smoking cannabis was identified as a rare possible short-term trigger; this did not establish cause. The 2017 national review found substantial evidence of worse breathing symptoms with long-term smoking, and moderate evidence of no link to lung cancer.	Non-smoked products; very heavy long-term use.	Solid, with limits (abstract only)
C075	Using cannabis in pregnancy went with more maternal anaemia and lower birth weight in a 2016 review. US obstetricians advised against use in pregnancy and breastfeeding (2015).	Cause. Tobacco use often goes with it and was hard to separate.	Solid, with limits
C076	A syndrome of repeated severe vomiting in heavy users, often with compulsive hot bathing, was described in 2004. The vomiting stopped in 7 of the 9 cases described when cannabis use stopped. The 2017 national review found moderate evidence of more child poisonings where cannabis was legal. It found too little evidence to say whether cannabis overdose causes death.	Adults’ acute risks beyond these.	Solid, with limits
C077	Inside medical studies, a 2008 review of 31 studies found that, within its 23 randomized trials, most reported side effects were non-serious (96.6%). Non-serious side effects were more common on cannabinoids. Serious ones did not differ in those trials, but long-term risks were poorly studied. A 2015 review of a different, larger set of trials reported higher odds of serious side effects and of stopping treatment because of side effects.	Long-term use, which was barely studied.	Solid, with limits (2008 review: abstract only)

What the official reviews said (their conclusions, not ours)

#	In plain English	Who this does not tell us about	How sure
C078	A major 2015 review (JAMA) rated the evidence moderate for chronic pain and MS stiffness, low for chemotherapy nausea, HIV weight gain, sleep and Tourette syndrome, and found more short-term side effects. The journal later corrected one of its MS stiffness figures; the corrected results show benefit on patient-reported stiffness but not on the doctors’ scale.	Today. This is that review’s 2015 judgement.	Solid, with limits (the doctors’-scale correction is confirmed; the patient-scale figure is still being checked)
C079	In 2017 the US National Academies judged the evidence conclusive or substantial for cannabis in adult chronic pain, and for oral cannabinoids in chemotherapy nausea and vomiting and in patient-reported MS stiffness. It found moderate evidence for short-term sleep, limited evidence for several other uses, limited evidence of no effect in dementia, glaucoma and depression, and too little evidence for cancer, cancer weight loss, irritable bowel syndrome and epilepsy.	Today, or every cannabis product. These were the review’s evidence ratings at the time, and its epilepsy rating came before the 2017 epilepsy trial.	Solid
C080	The American Academy of Neurology (2014) judged cannabis extract pills effective, and the spray and THC probably effective, for MS stiffness as patients report it. Later status, checked in 2026: the guideline’s PubMed record is marked “retired”; the date was not established. That does not change what it concluded in 2014.	Doctor-measured stiffness; current guidance.	Solid
C081	Most expert (Cochrane) reviews of the period found too little or no evidence: epilepsy, dementia, Tourette syndrome, HIV, fibromyalgia, schizophrenia and dependence treatment. The exception was chemotherapy nausea, where cannabinoid medicines “may be useful” when other drugs fail.	Today. These are the reviews’ judgements at the time.	Solid
C082	US obstetric and paediatric medical societies issued statements in 2015. Other societies’ statements cited by the researchers were not opened by us.	What those societies recommended. That is still being checked.	Solid, with limits (existence only)

Opioids, gaps and public claims

#	In plain English	Who this does not tell us about	How sure
C083	A 2014 study found states with medical-cannabis laws had about 25% lower opioid-painkiller overdose deaths in 1999–2010. That is a link, not proof the laws caused it. Patient surveys reported using cannabis instead of opioids. No controlled trial tested this. Later: when the same analysis was extended to 2017, the link reversed, and its authors judged it likely spurious.	Individual patients. State-level data cannot say what any person did.	Solid, with limits
C084	Not found in our searches, which does not mean none exist: a controlled human trial of whether cannabinoids treat cancer (the one nine-patient study tested safety, not effect); any demonstrated disease-slowing benefit (the one large trial testing it was negative); a controlled trial of cannabis replacing opioids; a controlled trial of plant cannabis or THC for PTSD (a small nabilone trial studied PTSD nightmares); controlled trials longer than one year of smoked or vaporised cannabis; and published results for several completed registered trials.	—	Solid, with limits
C085	Eight common public claims, checked against the period evidence. “No rigorous evidence by 2017”: wrong. “Benefits proven for many conditions”: partly true, for only a few. “Cannabinoids cure cancer in people”: not supported. “Approving a cannabis medicine = approving medical cannabis”: wrong. “Medical-cannabis laws cut opioid deaths”: not supported for the period, and later contradicted. “Cannabis causes schizophrenia”: partly true, since there is a substantial link but cause is not shown. “Cannabis isn’t addictive”: wrong. “CBD has no mental effects and no real risks”: wrong as worded. It is not intoxicating like THC, but it had side effects, drug interactions and measurable effects on the brain in trials.	—	Solid

Source file in the research repository: medical/MED01/ledger-plain-language-P2.md. Private links, some names and practitioner-record passages are removed from the public copy, and broken character encoding from file conversion is repaired; nothing else is changed.