Claim Ledger — MED01: Historical Evidence Landscape, medical cannabis & cannabinoid research 1960s → 2026

Round: MED01 · Part: P1 (1960–1999) · Ledger version: v1.0 — SIGNED OFF by Danny 2026-09-25 (P1 portion; MED01 ledger continues with P2) · text = v0.2 unchanged · trail: v0.1 Claude → v0.2 ChatGPT calibration (13 items) → v1.0. HOLD-marked components (C025, C028, C029, C035, C036 primary-document checks) remain open items carried into the interim synthesis and cross-audit; any change to them is a logged correction, not a silent edit. · Evidence search cutoff: 2026-09-20 (MED01-wide) · claims describe the evidence searched through this date References: PROTOCOL-MEDICAL §M3 (evidence base: source type × study design), §M5 (outcome type), §M7 (drift D1–D14), §M9 (dispositions) · base PROTOCOL.md §3 (grading) · critique.md · verification log Sources: S-241–S-307 appended to the shared SOURCE_LIBRARY.md at v1.0 sign-off (2026-09-25); source-library-additions.md is the round-local copy. Next free ID after P1: S-308. P2 (C041–C085) v1.0 SIGNED OFF by Danny 2026-10-10; S-308–S-445 appended to SOURCE_LIBRARY the same day; next free ID: S-446. (S-446–S-448 added 2026-10-13 by corrections #8 and #11; next free S-449.)

Claim IDs are MED01-Cnnn, permanent, never renumbered; P2 and P3 continue the series. Retired or failed claims stay with their disposition. Medical evidence is not forced into true/false.

Not medical advice. Every claim below is a claim about a literature and its conditions, or about what a named study or body reported in the populations it studied. Nothing here is a statement about what cannabis does for a patient today.

Origin key: Both = proposed independently by ChatGPT and Gemini (weight discounted per critique §1: neither run was clean) · ChatGPT · Gemini · verifier-found = surfaced by the completeness probe or verifier re-run (§M12b; mandatory cross-audit target). Access key: FT = full text read · AB = PubMed abstract · BIB = bibliographic record only · DOC = official document opened · — = not yet opened (capped).

A. Shape of the literature

MED01-C001 — Two frozen PubMed/MEDLINE queries returned 8,663 and 10,630 indexed cannabis/cannabinoid records for 1960–1999; despite search-string dependence, both show a 1970s surge, a 1980s trough, and a 1990s peak exceeding the 1970s.

Claim type: literature-shape · Outcome type: n/a
Origin: verifier-found (Gemini claimed 11,241; ChatGPT obtained no count)
Sources: S-241 (NCBI PubMed/MEDLINE, ESearch, run 2026-09-23). Query A (Gemini’s string, MeSH + tiab): 10,630 = 414 / 3,679 / 2,516 / 4,021 by decade. Query B (ChatGPT’s string, tiab incl. cannabinoid*, tetrahydrocannabinol): 8,663 = 252 / 2,492 / 2,153 / 3,766.
Best access: DOC (verifier ran both queries) · Evidence base: S-241 · official database count · design n/a
Base grading: established · direct · high for the shape; the absolute number is string-dependent
Verification: Gemini’s 11,241 and all four decade figures not reproducible (accepted by Gemini in correction round as web-interface term mapping); D-none. Reproduced by verifier a second time 2026-09-26: 10,630 / 8,663 exactly. §M10a both-model audit could not be completed by ChatGPT (PubMed blocked in its environment) — escalated to an origin-independent human rerun (Danny) per §M10a.
Qualifications: pre-1975 MEDLINE records lack abstracts (title/MeSH only), the MeSH heading “Cannabinoids” was introduced in 1977 (NLM record, verifier-confirmed 2026-09-26; the v1.0 text said 1971 — ChatGPT cross-audit correction), non-English and Eastern-bloc literature under-indexed; counts are of indexed records, not of research effort.
Final disposition: Accepted with qualification · Status: active

MED01-C002 — In a 2022 Scopus bibliometric analysis (title search `cannabi* OR hashish OR marijuana OR marihuana`), the United States was the most productive country overall (n = 12,420 of 29,802 records, 1829–2021), contributed 33 records (10.96%) in 1960–69 and 979 (41.68%) in 1970–79, and U.S. federal bodies (NIH, HHS, NIDA) were prominent among funding sponsors.

Claim type: literature-shape · Origin: Both
Sources: S-242 Ng JY, Chang N. J Cannabis Res 2022;4:25. DOI 10.1186/s42238-022-00133-0; PMID 35610633; PMCID PMC9131698
Best access: FT (Europe PMC XML) · Evidence base: S-242 · primary · observational (bibliometric database study) · retrospective
Base grading: established · direct · high
Verification: ✅ identity; ✅ the 33/10.96%, 979/41.68%, 12,420 and the TITLE() string confirmed in full text. Gemini’s global decade totals (~301 / 2,349 / 1,555 / 3,100) were not found in the text and are not ledgered. ChatGPT’s co-author “Chang YK” corrected to Nathan Chang.
Qualifications: Scopus title-only search; US-share percentages are of records with a US affiliation, not of all research; the paper codes funders and countries, not harm-vs-therapeutic orientation (see C016).
Final disposition: Accepted · Status: active

MED01-C003 — The verified record supports a useful four-phase periodization for this evidence map — constituent chemistry (1960s), experimental human pharmacology and small therapeutic trials (1970s), pharmaceutical cannabinoid development and antiemetic trial programmes (1980s), and receptor/endocannabinoid molecular biology with renewed official review (1990s) — anchored in the publication-count shape (C001) and the dated discovery and trial record below; the boundaries are editorial, not an observed variable.

Claim type: literature-shape / synthesis-conclusion · Origin: Both (phase boundaries drawn slightly differently; the four-phase structure was proposed by each)
Sources: S-241, S-242, and the primary sources of C004–C011, C012–C020
Best access: mixed · Evidence base: inference from verified primary sources; no single source states it
Base grading: probable · direct · medium
Verification: phases are an editorial reading, not a finding of any one source; both researchers were asked for phases with evidence (§3 F1) and both anchored them in dated primary papers.
Final disposition: Accepted with qualification (interpretive; boundaries administrative) · Status: active

B. Discoveries that redirected the field (all non-human evidence — mechanism/plausibility only, §M3). Design tag for pure chemistry (isolation, structure, synthesis) = `mechanistic (chemical/structural; no biological outcome)`, not `in-vitro`.

MED01-C004 — The structure of cannabidiol was published by Mechoulam and Shvo in 1963.

Claim type: discovery · Origin: Both · Sources: S-243 Mechoulam R, Shvo Y. Tetrahedron 1963;19:2073–2078. DOI 10.1016/0040-4020(63)85022-X; PMID 5879214
Best access: BIB · Evidence base: S-243 · primary · mechanistic (chemical/structural) · period
Base grading: established · direct · high · Verification: ✅ identity (PubMed + Crossref); no abstract; title states the result
Final disposition: Accepted · Status: active

MED01-C005 — Δ9-THC was isolated, its structure assigned, and a partial synthesis reported by Gaoni and Mechoulam in 1964; the first total synthesis (racemic) was published by Mechoulam and Gaoni in 1965; stereospecific and further total syntheses followed in 1967 (Mechoulam, Braun & Gaoni; Fahrenholtz, Lurie & Kierstead) and 1969 (Petrzilka et al.).

Claim type: discovery · Origin: Both for 1964; ChatGPT for 1965; Gemini for 1967/1969 (Gemini’s “1967 = total synthesis” corrected)
Sources: S-244 Gaoni & Mechoulam, JACS 1964;86:1646–1647, DOI 10.1021/ja01062a046 · S-245 Mechoulam & Gaoni, JACS 1965;87:3273–3275, DOI 10.1021/ja01092a065, PMID 14324315 · S-246 Mechoulam, Braun & Gaoni, JACS 1967;89:4552–4554, DOI 10.1021/ja00993a072, PMID 6046550 · S-247 Fahrenholtz, Lurie & Kierstead, JACS 1967;89:5934–5941, DOI 10.1021/ja00999a034 · S-248 Petrzilka, Haefliger & Sikemeier, Helv Chim Acta 1969;52:1102–1134, DOI 10.1002/hlca.19690520427
Best access: BIB (all paywalled; titles state the results) · Evidence base: all primary · mechanistic (chemical/structural) · period
Base grading: established · direct · high
Verification: all five DOIs resolve to the stated papers (Crossref titles: 1965 “A Total Synthesis of dl-Δ1-THC”; 1967 Mechoulam “Stereospecific synthesis…”; 1967 Fahrenholtz “Total synthesis of (±)-Δ9-THC…”). Verifier ruling on chronology in critique/correction log.
Qualifications: “first” is bounded to the indexed record; Δ1 (monoterpenoid) = Δ9 (dibenzopyran) numbering.
Final disposition: Accepted · Status: active

MED01-C006 — Other phytocannabinoids were isolated or structurally characterised in the 1960s–70s (cannabichromene, Claussen, von Spulak & Korte 1966; a propyl homologue of THC reported by Gill, Paton & Pertwee 1970), and a 1980 review by Turner, ElSohly & Boeren catalogued the constituents then known. Subclaim (Inconclusive, held to full text): that the 1980 review counts 61 cannabinoids.

Claim type: discovery · Origin: Gemini (CBC, Gill 1970); ChatGPT (61-cannabinoid review, via correction round)
Sources: S-249 Claussen, von Spulak & Korte, Tetrahedron 1966;22:1477–1479, DOI 10.1016/S0040-4020(01)99445-1 · S-250 Gill, Paton & Pertwee, Nature 1970;228:134–136, DOI 10.1038/228134a0, PMID 5466704 · S-251 Turner, ElSohly & Boeren, J Nat Prod 1980;43:169–234, DOI 10.1021/np50008a001, PMID 6991645 · S-252 NIDA Research Monograph 31 (1980), PMID 6775230
Best access: BIB · Evidence base: S-249/S-250 primary · mechanistic (chemical/structural) / animal · period; S-251 narrative-review · period
Base grading: probable · direct · medium
Verification: ✅ identities (after correction: Gemini’s original pages for Gill were wrong, and its correction-round PMID 5460037 was wrong-object; ChatGPT’s original PMID 35817852 was wrong-object, replaced). Not verified: that Gill 1970 is the first THCV isolation (Gemini) — no abstract; that the “61” figure appears in S-251 — no abstract. CBG/CBN priority claims not pursued.
Final disposition: Accepted with qualification (identities established); “61” subclaim Inconclusive (no abstract; not read); Gill-1970-as-first-THCV Inconclusive · Status: active

MED01-C007 — A saturable, stereoselective, high-affinity cannabinoid binding site (Kd ≈ 133 pM for [³H]CP-55,940) coupled to G-protein was characterised in rat brain membranes by Devane, Dysarz, Johnson, Melvin and Howlett in 1988.

Claim type: discovery · Origin: Both · Sources: S-253 Mol Pharmacol 1988;34:605–613, PMID 2848184
Best access: AB · Evidence base: primary · in-vitro / mechanistic · period · Base grading: established · direct · high
Verification: ✅ identity; ✅ Kd 133 pM, saturable, single class of sites confirmed in abstract. Human effect: not tested (D1 guard).
Final disposition: Accepted · Status: active

MED01-C008 — A cannabinoid receptor cDNA was cloned and functionally expressed by Matsuda et al. (NIMH) in 1990, and receptor distribution was mapped autoradiographically across mammalian brains including human tissue by Herkenham et al. (NIMH) the same year.

Claim type: discovery · Origin: Both · Sources: S-254 Matsuda et al., Nature 1990;346:561–564, DOI 10.1038/346561a0, PMID 2165569 · S-255 Herkenham et al., PNAS 1990;87:1932–1936, DOI 10.1073/pnas.87.5.1932, PMID 2308954
Best access: AB · Evidence base: both primary · mechanistic · period · Base grading: established · direct · high
Verification: ✅ both. Human tissue in Herkenham = anatomical mapping, not a treatment study (D1).
Final disposition: Accepted · Status: active

MED01-C009 — An endogenous brain constituent binding the cannabinoid receptor, arachidonylethanolamide (“anandamide”), was isolated from porcine brain and reported by Devane, Hanuš, Mechoulam and colleagues in 1992.

Origin: Both · Sources: S-256 Science 1992;258:1946–1949, DOI 10.1126/science.1470919, PMID 1470919 · Evidence base: primary · mechanistic (biochemical; porcine tissue, mouse bioassay) · period
Verification: ✅ · Base grading: established · direct · high · Final disposition: Accepted · Status: active

MED01-C010 — A second cannabinoid receptor, expressed predominantly in peripheral immune tissue and not detected in brain, was cloned by Munro, Thomas and Abu-Shaar (MRC Cambridge) in 1993.

Origin: Both · Sources: S-257 Nature 1993;365:61–65, DOI 10.1038/365061a0, PMID 7689702 · Evidence base: primary · mechanistic (cloning from human HL-60 cells) · period
Verification: ✅ · Base grading: established · direct · high · Final disposition: Accepted · Status: active

MED01-C011 — 2-Arachidonoylglycerol was identified as a second endogenous cannabinoid-receptor ligand in 1995 by two independent groups (Mechoulam et al., canine gut; Sugiura et al., rat brain).

Origin: Both · Sources: S-258 Biochem Pharmacol 1995;50:83–90, DOI 10.1016/0006-2952(95)00109-D, PMID 7605349 · S-259 BBRC 1995;215:89–97, DOI 10.1006/bbrc.1995.2437, PMID 7575630 · Evidence base: both primary · mechanistic · period
Verification: ✅ both · Base grading: established · direct · high · Final disposition: Accepted · Status: active

C. What was studied in humans (clinical findings of the period; not present-day verdicts)

Claim type: clinical-finding-of-the-period · Outcome type: symptom relief · Origin: Both
Sources: S-260 Sallan, Zinberg & Frei, NEJM 1975;293:795–797, DOI 10.1056/NEJM197510162931603, PMID 1099449
Best access: AB · Evidence base: primary · rct (crossover, each patient own control) · period
Population: adult cancer patients on chemotherapy of central emetogenic action · Intervention: oral Δ9-THC (dose not in abstract) · Comparator: placebo · Outcome: antiemetic response per course; between-condition
Base grading: established · direct · high for what this trial reported
Verification: ✅ identity; ✅ all figures in abstract. Design-appropriate contrast (within-participant, between conditions) — §M4 rule satisfied. RoB: not appraised (abstract only); blinding integrity questionable — abstract notes no patient vomited while “high”.
Qualifications: N = 20; single centre; dose unreported at abstract level; 1975 antiemetic comparators are not modern ones.
Final disposition: Accepted with qualification · Status: active

Claim type: clinical-finding-of-the-period · Outcome type: symptom relief · Origin: Both (route framing: ChatGPT correct; Gemini corrected in round 1)
Sources: S-261 Chang et al., Ann Intern Med 1979;91:819–824, DOI 10.7326/0003-4819-91-6-819, PMID 293141
Best access: AB · Evidence base: primary · rct (crossover) · period
Verification: ✅ identity (Gemini’s correction-round PMID 389332 was wrong-object); ✅ figures in abstract; route-contrast wording rejected (D5). Gemini’s correction concedes the smoked arm was open-label.
Final disposition: Accepted with qualification · Status: active

Claim type: clinical-finding-of-the-period · Outcome type: symptom relief · Origin: Both
Sources: S-262 Herman et al., NEJM 1979;300:1295–1297, DOI 10.1056/NEJM197906073002302, PMID 375088
Best access: AB · Evidence base: primary · rct (crossover, active comparator) · period · sponsor: Eli Lilly (per ChatGPT; to confirm in FT)
Verification: ✅; both researchers under-reported the doubled side-effect rate — verifier addition from abstract (D14 guard).
Final disposition: Accepted with qualification · Status: active

MED01-C015 — A 2015 Cochrane review identified 23 randomized trials of cannabinoids (nabilone 12, dronabinol 11) for chemotherapy-induced nausea and vomiting conducted 1975–1991, 19 of them crossover designs, most at risk of bias from allocation concealment or attrition; two 2001 BMJ systematic reviews — Tramèr et al. (quantitative, antiemesis) and Campbell et al. (qualitative, pain) — had earlier synthesised the pre-2000 trials.

Claim type: literature-shape · Origin: ChatGPT (Cochrane); verifier-found (Tramèr, Campbell — neither researcher cited them)
Sources: S-263 Smith et al., Cochrane CD009464.pub2 (2015), PMID 26561338 (a newer version exists — cite the version used) · S-264 Tramèr et al., BMJ 2001, PMID 11440936 · S-265 Campbell et al., BMJ 2001, PMID 11440935
Best access: AB (Cochrane summary page read) / BIB (BMJ) · Evidence base: all sr-ma · retrospective · rest on period rct
Base grading: established · direct · high (for the review’s corpus)
Verification: ✅ identities. Used only to map the size and shape of the period literature (rule 6a); Cochrane’s own count is its search, not ours.
Final disposition: Accepted · Status: active · Cross-audit: mandatory (verifier-found component)

Claim type: clinical-finding-of-the-period · Outcome type: symptom relief · Origin: Both (Gemini’s “roughly equivalent to 60–120 mg codeine” withdrawn; its calibrated “20 mg ≈ 60 mg codeine” is not in the abstract)
Sources: S-266 Noyes et al., J Clin Pharmacol 1975;15:139–143, DOI 10.1002/j.1552-4604.1975.tb02348.x, PMID 1091664 · S-267 Noyes et al., Clin Pharmacol Ther 1975;18:84–89, DOI 10.1002/cpt197518184, PMID 50159
Best access: AB · Evidence base: primary · rct (single-dose crossover) · period
Verification: ✅ identities (Gemini’s correction-round PMID 1097148 was wrong-object → Hollister); ✅ wording matches abstracts. Any codeine-equivalence figure is capped pending FT.
Final disposition: Accepted with qualification · Status: active

MED01-C017 — Acute reductions in intraocular pressure after cannabis or THC were reported in humans from 1971 (Hepler & Frank, JAMA letter, no effect size in the record); a 1977 study of intravenous Δ9-THC in ten normotensive subjects reported an average 37% (up to 51%) IOP decrease with 22–65% heart-rate increase; a 1998 review states smoked marijuana lowers IOP in 60–65% of users while judging chronic use for glaucoma untenable because of systemic effects.

Claim type: clinical-finding-of-the-period · Outcome type: surrogate physiological endpoint (not glaucoma outcome — D8) · Origin: Both for Hepler 1971; verifier-found for Cooler 1977 and Green 1998
Sources: S-268 Hepler & Frank, JAMA 1971;217:1392, PMID 5109652 · S-269 Cooler & Gregg, South Med J 1977, PMID 329423 · S-270 Green K, Arch Ophthalmol 1998, PMID 9823341
Best access: BIB (Hepler) / AB (Cooler, Green) · Evidence base: S-268 primary · interventional-nr · period; S-269 primary · interventional-nr (IV THC, N=10) · period; S-270 narrative-review · period
Verification: Gemini’s “25–30% reduction, 11 subjects” for Hepler 1971 REJECTED — not in the cited letter; both replacement identifiers offered in the correction round (AJO 1972 DOI; PMID 4647805) resolve to unrelated papers; no Hepler & Frank 1972 AJO paper found in PubMed. Cooler’s figures are for intravenous THC, not smoked cannabis (D5 guard).
Final disposition: Accepted with qualification · Status: active · Cross-audit: mandatory (verifier overturned a researcher)

Claim type: clinical-finding-of-the-period · Outcome type: surrogate physiological endpoint · Origin: ChatGPT (Williams 1976); Gemini (Tashkin 1973, identifiers corrected)
Sources: S-271 Williams, Hartley & Graham, Thorax 1976;31:720–723, DOI 10.1136/thx.31.6.720, PMID 797044, PMC470501 · S-272 Tashkin, Shapiro & Frank, NEJM 1973;289:336–341, DOI 10.1056/NEJM197308162890702, PMID 4718513
Best access: AB (Williams 1976) / BIB (Tashkin 1973 — PubMed carries no abstract; title-level identity only) · Evidence base: both primary · rct / interventional-nr · period
Verification: Williams ✅ abstract; Tashkin 1973 identity ✅ only (Gemini’s original DOI/pages wrong; correction DOI right, correction PMID 4719469 wrong-object). The 1973 bronchodilation result is therefore not carried in the synthesis until the text is seen. Acute single-dose only (D7 guard).
Final disposition: Accepted with qualification · Status: active

MED01-C019 — In a 1980 Brazilian double-blind study, 15 patients with secondary generalized epilepsy with temporal focus, uncontrolled on existing antiepileptics, were randomized to add-on oral CBD 200–300 mg/day or placebo for up to 4½ months; the investigators reported that 4 of 8 CBD patients remained almost free of convulsive crises, 3 showed partial improvement and 1 did not respond, while 7 of 7 placebo patients were unchanged and 1 improved; 3 mg/kg/day for 30 days was tolerated in 8 healthy volunteers.

Claim type: clinical-finding-of-the-period · Outcome type: symptom relief · Origin: Both
Sources: S-273 Cunha et al., Pharmacology 1980;21:175–185, DOI 10.1159/000137430, PMID 7413719
Best access: AB · Evidence base: primary · rct (parallel, add-on) · period
Verification: ✅ — the 4/8 – 3/8 – 1 and 7/1 breakdown is in the PubMed abstract (verifier re-read 2026-09-26 after the ChatGPT cross-audit; the 2026-09-23 read was truncated — verifier error, correction #2). Later Cochrane (Gloss 2012/2014, verifier-confirmed PMIDs 22696383 / 24595491) judged the historical trials too small for reliable conclusions — reported separately (rule 6a).
Final disposition: Accepted with qualification · Status: active

Origin: ChatGPT · Sources: S-274 Petro & Ellenberger, J Clin Pharmacol 1981;21:413S–416S, DOI 10.1002/j.1552-4604.1981.tb02621.x, PMID 6271839 · Evidence base: primary · rct (crossover) · period · Best access: AB
Verification: ✅ all in abstract. Context (verifier-found, S-286 §5.20): by 1998 the House of Lords counted six published MS trials totalling 41 subjects worldwide · Final disposition: Accepted with qualification (N = 9) · Status: active · Cross-audit: mandatory (verifier-found datum)

Claim type: clinical-finding-of-the-period · Outcome type: symptom relief (appetite); weight = not shown · Origin: Both (Gemini’s Tier-1 “effective … for halting progressive weight loss” corrected — D8/D11)
Sources: S-275 Beal et al., J Pain Symptom Manage 1995;10:89–97, DOI 10.1016/0885-3924(94)00117-4, PMID 7730690
Best access: AB · Evidence base: primary · rct (parallel) · period
Verification: ✅ (Gemini’s original DOI 404’d; corrected). Attrition 139 → 88 evaluable noted.
Final disposition: Accepted with qualification · Status: active

D. Preparations and pharmaceutical development

MED01-C022 — The period’s human studies used smoked plant material, oral Δ9-THC (including later pharmaceutical dronabinol formulations), oral purified CBD, the synthetic analogue nabilone, and investigational synthetics (levonantradol, DMHP, dexanabinol/HU-211), which are pharmacologically and legally distinct and not interchangeable (D10). (US standardized research cigarettes: see C025, held.)

Claim type: literature-shape · Origin: Both (inventory largely ChatGPT’s; Gemini added potency and placebo-preparation detail — unsourced, not ledgered)
Sources: S-276 Lemberger & Rowe, Clin Pharmacol Ther 1975;18:720–726, PMID 1204278 (nabilone human pharmacology) · S-277 Cronin et al., J Clin Pharmacol 1981;21:43S–50S, PMID 7028793 (IM levonantradol) · S-278 Lemberger et al., Clin Pharmacol Ther 1974;15:380–386, PMID 4821441 (DMHP) · S-279 Brewster et al., Int J Clin Pharmacol Ther 1997;35:361–365, PMID 9314087 (dexanabinol IV PK) · plus C012–C021 sources
Best access: AB/BIB · Evidence base: primary · interventional-nr / rct · period
Verification: ✅ identities. Gemini’s “1.0–3.5% THC research cigarettes” and Soxhlet placebo detail: no source — not ledgered.
Final disposition: Accepted · Status: active

MED01-C023 — Dronabinol (Marinol) was approved by the US FDA on 31 May 1985 for chemotherapy-associated nausea and vomiting refractory to conventional antiemetics, with an AIDS-anorexia indication added in 1992; the approved formulation was moved from US Schedule I to II in 1986 and to Schedule III by DEA final rule of 2 July 1999 (64 FR 35928).

Claim type: regulatory · Jurisdiction/date: United States; as of 1999-07-02 · Origin: Both
Sources: S-280 DEA final rule, 64 FR 35928 (2 July 1999) · S-281 FDA congressional testimony 2016 (retrospective official summary)
Best access: DOC (1999 FR page opened by verifier; approval chain confirmed) · Evidence base: official · period (S-280) / retrospective (S-281)
Verification: ✅. Gemini’s exact day for the 1992 supplement (22 Dec) not independently confirmed — month/year only ledgered.
Final disposition: Accepted · Status: active

MED01-C024 — Nabilone (Cesamet) had human clinical pharmacology published from 1975 and antiemetic trials from 1979; it was licensed in the United Kingdom in 1982 for prescription-only, hospital-only use against chemotherapy nausea unresponsive to other treatment (and is not a UK controlled drug), and approved by the US FDA in 1985 for the same indication; a Canadian approval in the early 1980s is reported by Gemini but the year is unverified (1982 in its pass, 1981 in correction).

Claim type: regulatory · Jurisdictions: US (1985); Canada, UK (early 1980s) · Origin: Both (US); Gemini (Canada/UK)
Sources: S-276, S-262, S-281; S-286 House of Lords 1998 ch. 5, Box 4 (opened 2026-09-25: “licensed in 1982 … not a controlled drug”)
Verification: UK ✅ (HoL Box 4, DOC — verifier-found in verified form; cross-audit target); US ✅ S-281 opened by verifier 2026-09-26 (FDA testimony: “Cesamet Capsules for the treatment of nausea and vomiting associated with chemotherapy in 1985”); Canada ⏳ unverified
Final disposition: Accepted with qualification (UK and US verified); Canada component Inconclusive, kept out of the claim sentence · Status: active · Cross-audit: mandatory · Correction log: US HOLD lifted 2026-09-26 (DOC).

F. Conditions of research

MED01-C025 — US federal agencies maintained a standardized research-cannabis supply programme at the University of Mississippi from 1968 (NIMH, later NIDA), with synthetic Δ8/Δ9-THC produced under federal contract, and a 1976 NIDA review listed standardized supply and analytical methods as enabling developments while acknowledging potentially therapeutic pharmacology.

Claim type: regulatory · Jurisdiction/date: United States; 1968–1976 · Origin: ChatGPT (Congressional Record 1969, 1970, 1974; NIDA 1976); Gemini asserted the same structure citing the 1961 Single Convention art. 28 and IOM 1999 ch. 6 (not opened)
Sources: S-282 Congressional Record 1969 (GPO-CRECB-1969-pt27) · S-283 Congressional Record 1970 · S-284 Congressional Record 1974 (GPO-CRECB-1974-pt25) · S-285 NIDA research review 1976 (GOVPUB-HE20-PURL-gpo76421)
Best access: — (govinfo PDFs cited by URL; not yet opened by verifier) · Evidence base: official · period
Base grading: probable · direct · medium (pending page check)
Verification: ⏳ page-level check outstanding; URLs are well-formed govinfo objects.
Final disposition: Inconclusive — HOLD until page-level support is checked in S-282–S-285 (§M9: an unopened source cannot carry an Accepted claim); promote on verification · Status: active

MED01-C026 — Controlled human research with cannabis and cannabinoids was legally possible in the United States after Schedule I placement (1970) and in the United Kingdom under Schedule 1 with Home Office licence (as of 1998), as demonstrated by the completed controlled studies C012–C021; the strong form of the claim that scheduling made research impossible is not supported, while regulated access and single-source supply imposed documented constraints.

Claim type: regulatory / synthesis-conclusion · Origin: Both (ChatGPT “unsupported in strong form”; Gemini “partly supported” — reconciled: same substance)
Sources: C012–C021 sources; S-286 House of Lords 1998 ch. 5 (research licensing) — page to be opened; S-282–S-285
Verification: the existence proof (completed US trials post-1970) is verified; the magnitude of the burden is not quantified by either researcher with sources.
Final disposition: Accepted with qualification · Status: active · Public claim P2: unsupported as stated; partly supported as “created regulatory barriers”

MED01-C027 — The documented searches identified no coded publication or grant dataset capable of quantifying the balance of harms-focused versus therapeutic research attention in 1960–1999; federally funded health-effects research (S-282, figures held pending page check) and sustained therapeutic trial programmes (C012–C021) both existed, and no ratio is supportable.

Claim type: absence / literature-shape · Origin: ChatGPT (position); Gemini’s “>75%” and its replacement (“majority per Ng & Chang”) rejected — the Ng & Chang text codes funders and countries, not orientation
Sources: S-282; S-242 · Verification: verifier read Ng & Chang full text (§3G empirical-basis rule applied)
Final disposition: Accepted with qualification (an absence-of-dataset claim about the documented searches; the balance itself remains indeterminate) · Status: active · Public claim P3: “only harms” unsupported; balance unquantified

MED01-C028 — A US federal single-patient compassionate-access (IND) programme for smoked cannabis existed from the late 1970s and was closed to new applicants in 1992; the commonly stated figures (established 1978 after US v. Randall; 27 patients approved; 13–15 continuing) rest on a participant memoir and secondary accounts, and no primary government record establishing the counts has been located.

Claim type: regulatory · Jurisdiction: United States · Origin: Gemini (specifics); ChatGPT explicitly declined to assert dates/counts
Sources: Gemini cites PHS news release 3 Mar 1992 (not located), IOM 1999 p. 214 (HOLD — page check pending), Randall & O’Leary 1998 memoir (ISBN 1-56025-166-2, participant account)
Verification: ⏳ verifier blocked from IOM 1999 ch. 5; ChatGPT (2026-09-26 review) reports the chapter states marijuana cigarettes were available through “a special FDA-sponsored Compassionate IND Program” closed to new participants in March 1992 — supports existence and the 1992 closure, not the 1978 start or the patient counts. Note for cross-audit: the count figures match Danny’s archive claims A18/A20/A33 verbatim; agreement between a researcher and the archive is not verification.
Final disposition: Inconclusive — existence and 1992 closure probable; dates and counts unverified · Status: active · Cross-audit: mandatory

MED01-C029 — In the US rescheduling proceeding on the 1972 NORML petition, a DEA administrative law judge (Young) recommended transfer of marijuana to Schedule II in 1988 and the DEA Administrator denied the petition in a 1989 final order (reported at 54 FR 53767); a further denial and remand followed in 1992 (57 FR 10499); marijuana remained in Schedule I through 1999.

Claim type: regulatory · Jurisdiction: United States; 1972–1992 · Origin: Gemini (1988/1989 specifics); ChatGPT (1992 FR citation; declined to quote the 1988 ruling)
Sources: S-287 54 FR 53767 (29 Dec 1989) — HOLD: 1989 predates govinfo’s FR coverage; page not yet retrieved · S-288 57 FR 10499 (26 Mar 1992) · S-289 FR 2016 summary (ChatGPT’s retrospective pointer)
Verification: the sequence is consistent across both passes and the 2016 FR history; Gemini’s quoted denial sentence is unverified (no page opened).
Final disposition: Accepted with qualification (sequence); quoted text unverified · Status: active

G. Harms research

MED01-C030 — In a 15-year follow-up of 45,570 Swedish conscripts (1987), cannabis use on more than 50 occasions was associated with a relative risk of schizophrenia of 6.0 (95% CI 4.0–8.9) versus non-use; the authors described cannabis as “an independent risk factor,” a causal reading the observational design cannot by itself establish.

Claim type: harms-research · Origin: Both · Sources: S-290 Andréasson et al., Lancet 1987;2:1483–1486, DOI 10.1016/S0140-6736(87)92620-1, PMID 2892048
Evidence base: primary · observational (cohort) · period · Best access: AB · Verification: ✅; D2 attribution applied
Final disposition: Accepted with qualification · Status: active

MED01-C031 — Respiratory research of the period distinguished endpoints: 47–59 days of heavy smoking in 28 young men produced modest but significant falls in FEV1, mid-expiratory flow, airway conductance and diffusing capacity (1976); an 8-year cohort of 394 young adults found tobacco but not marijuana associated with accelerated FEV1 decline (1997); bronchoscopy in 40 subjects found airway inflammation and mucosal pathology in marijuana, tobacco and combined smokers versus non-smokers (1998).

Claim type: harms-research · Origin: ChatGPT (all three); Gemini (Tashkin lineage generally; its “1999 AJRCCM 160:1658” withdrawn) · Sources: S-291 Tashkin et al., NEJM 1976;294:125–129, PMID 1244507 · S-292 Tashkin et al., AJRCCM 1997;155:141–148, PMID 9001303 · S-293 Roth et al., AJRCCM 1998;157:928–937, PMID 9517614 · (S-294 Tashkin 1987 ARRD 135:209, PMID 3492159 — Gemini, identity ✅, not read)
Evidence base: primary · interventional-nr / observational · period · Best access: AB · Verification: ✅ all figures in abstracts
Final disposition: Accepted · Status: active

MED01-C032 — A human drug-interaction study (1980) found that CBD 600 mg/day for 5–12 days lowered hexobarbital oral clearance by 36% in ten subjects (systemic clearance also 36% lower in the four given IV hexobarbital), establishing that cannabinoid–drug metabolic interactions were being investigated in humans by 1980.

Claim type: harms-research · Origin: ChatGPT · Sources: S-295 Benowitz et al., Clin Pharmacol Ther 1980;28:115–120, DOI 10.1038/clpt.1980.139, PMID 7389248
Evidence base: primary · interventional-nr · period · Best access: AB · Verification: ✅ · D6/D13 guard: 600 mg/day is a research exposure, not a statement about any product.
Final disposition: Accepted · Status: active

MED01-C033 — Early widely publicised harm claims — chromosome breakage in users (Stenchever 1974) and inhibited cell-mediated immunity in smokers (Nahas 1974) — were published in the period; Gemini states they were later refuted, a characterisation not yet verified against period rebuttal literature.

Claim type: harms-research · Origin: Gemini · Sources: S-296 Stenchever et al., Am J Obstet Gynecol 1974;118:106–113, PMID 4808863 · S-297 Nahas et al., Science 1974;183:419–420, PMID 4271816
Evidence base: primary · observational / in-vitro · period · Best access: BIB · Verification: ✅ identities; “systematically refuted” unsourced (rule 6a).
Final disposition: Accepted with qualification (existence of the claims); refutation Inconclusive · Status: active

H. What official syntheses concluded (each attributed and dated — never this programme’s finding)

MED01-C034 — The 1997 WHO report Cannabis: A Health Perspective and Research Agenda (WHO/MSA/PSA/97.4) reviewed health effects across organ systems and included a therapeutic-use section and research recommendations.

Origin: Both (WHO 1997); Gemini (excision episode) · Sources: S-298 WHO/MSA/PSA/97.4 (institutional archive) · Hall et al. 1999 chapter — unverified grey literature
Verification: WHO document ✅ (TDR page). Q8 note (not a claim): Gemini reports that a comparative-risk chapter drafted for the project was excised and later published as a CAMH book chapter (Hall, Room & Bondy 1999) — commonly stated; no primary record seen; stays out of the ledger and synthesis until one is.
Final disposition: Accepted · Status: active

MED01-C035 — The UK House of Lords Science and Technology Committee (Ninth Report, 1998, HL 151) judged the evidence insufficient to establish conclusively whether cannabis itself had medical value, called for urgent trials in MS and chronic pain, and recommended moving cannabis from Schedule 1 to Schedule 2 to permit named-patient prescribing; the UK Government declined the rescheduling recommendation in December 1998 pending clinical-trial evidence.

Origin: Both (report); ChatGPT (Government response, Hansard 3 Dec 1998) · Sources: S-299 HL Paper 151 (1998) · S-300 Lords Hansard 3 Dec 1998
Best access: DOC — verifier opened ch. 4 and ch. 5 via browser 2026-09-25 (sandbox 403). Ch. 5 §5.11: antiemesis “the only medical indication for which adequate data from controlled clinical trials exist, mostly from studies in the 1970s with pure THC and … nabilone … as effective as prochlorperazine”; §5.20: six published MS trials, 41 subjects worldwide (verifier-found datum → C020 context); Box 5: UK moved dronabinol from Schedule 1 to 2 in 1995 (verifier-found) · Evidence base: official · period
Final disposition: Accepted — ch. 4, 5 and 8 opened by verifier (§8.1 “not enough rigorous scientific evidence to prove conclusively that cannabis itself has, or indeed has not, medical value”; §8.3 trials for MS and chronic pain “as a matter of urgency”; §8.6 transfer cannabis from Schedule 1 to Schedule 2 for named-patient prescribing); Lords Hansard 3 Dec 1998 cols 704–705 opened (Lord Hoyle for the Government: “unable to agree” with the Schedule 1→2 recommendation; would move “only if scientific evidence to support the change were available”; the Advisory Council on the Misuse of Drugs had advised against rescheduling on 19 Nov 1998). HOLD lifted 2026-09-26. Box 5 dronabinol-1995 datum is verifier-found (cross-audit target if promoted) · Status: active · Public claim P6: documented instance of a government declining a commissioned body’s recommendation; “disregarded” as a universal is too broad

MED01-C036 — The 1999 US Institute of Medicine report Marijuana and Medicine: Assessing the Science Base (commissioned by ONDCP, 1997) concluded that the accumulated data indicated potential therapeutic value of cannabinoid drugs, particularly for pain, nausea/vomiting and appetite stimulation; characterised smoked marijuana as a crude delivery system delivering harmful substances; and recommended short-term controlled trials and development of non-smoked rapid-onset delivery.

Origin: Both · Sources: S-301 IOM 1999, DOI 10.17226/6376 (NCBI Bookshelf NBK230722)
Best access: DOC (summary chapter read via Bookshelf per ChatGPT; verifier page check pending) · Evidence base: official · period · rests on rct/observational/animal
Final disposition: Accepted with qualification — HOLD for verifier support check (identity secure; NCBI Bookshelf reCAPTCHA/500 and the NAP image reader blocked the verifier on 2026-09-23, -25 and -26; both raw passes read it, and ChatGPT’s 2026-09-26 editorial review reports the Bookshelf Executive Summary states the C036 conclusions verbatim — three independent reads, still convergence rather than verifier check) · Status: active · Public claim P4: partly supported only in the narrow, product-specific sense (approved dronabinol/nabilone indications); unsupported as a broad claim

MED01-C037 — Earlier official bodies also reported in the window — the US National Commission on Marihuana and Drug Abuse (Shafer, 1972), the Canadian Le Dain Commission (1972), the IOM/NAS Marijuana and Health (1982), and the NIH Workshop on the Medical Utility of Marijuana (1997) — with Gemini summarising their conclusions and government responses; none of these summaries has yet been checked against the primary texts.

Origin: Gemini only · Sources: primary documents not yet opened (S-IDs reserved: S-302 Shafer 1972; S-303 Le Dain 1972; S-304 IOM 1982; S-305 NIH 1997)
Verification: ⏳ identities plausible (well-known documents); Gemini’s specific findings/quotes unverified; note Gemini’s “Nixon rejected within hours” is a commonly stated claim needing a dated primary source (Q8 discipline).
Final disposition: Inconclusive pending primary-text check · Status: active · Cross-audit: ChatGPT (Gemini-origin)

I. Disease-modification claims and absences

MED01-C038 — The earliest disease-modification evidence located for this window is non-human: oral Δ9-THC, Δ8-THC and CBN retarded Lewis lung adenocarcinoma growth in mice while CBD did not and THC did not extend survival in L1210 leukaemia (Munson et al. 1975); CBD and THC reduced glutamate neurotoxicity in rat cortical neuron cultures by a receptor-independent antioxidant mechanism (Hampson et al. 1998). No human study of cannabinoid cancer disease-modification or neurodegenerative disease-modification was identified in the documented searches of either researcher or the verifier.

Claim type: absence + discovery · Outcome type: disease modification (non-human only — D1, D3) · Origin: Both
Sources: S-306 Munson et al., JNCI 1975;55:597–602, DOI 10.1093/jnci/55.3.597, PMID 1159836 · S-307 Hampson et al., PNAS 1998;95:8268–8273, DOI 10.1073/pnas.95.14.8268, PMID 9653176
Evidence base: primary · animal (S-306) / in-vitro (S-307) · period · Best access: AB · Verification: ✅ both abstracts match exactly
Final disposition: Accepted with qualification (“none identified” ≠ “none exists”) · Status: active · Public claim P7: the US patent application filed 21 April 1999 (US 09/674,028; granted 2003 → P2) rested on S-307’s in-vitro evidence; a filing asserts utility, not clinical proof — unsupported as phrased

MED01-C039 — Areas for which no human study was identified in the documented searches (not a claim that none exists): human cannabinoid cancer disease-modification; human neurodegenerative disease-modification; large or long-duration randomized trials of smoked/whole cannabis for chronic disease; paediatric CBD epilepsy programmes; interventional or observational human studies in inflammatory bowel disease; PTSD; structured cannabis-as-opioid-substitution trials; human trials of anandamide or 2-AG themselves.

Claim type: absence · Origin: Both (union of both lists; Gemini’s CAND-24 narrowed after EMBASE admission)
Verification: verifier completeness probe found no counter-example; the searches are documented in both passes’ logs and critique §2.
Final disposition: Accepted with qualification · Status: active

MED01-C040 — Public claim P1 (“almost no research before legalization”) is unsupported: the large pre-1996 record in C001–C038 — a chemistry, pharmacology, harms and clinical literature that existed decades before, and continued through, the first modern medical-cannabis legalization measures (California 1996 falls inside P1) — is sufficient to reject it; what is supportable is that controlled clinical-efficacy research was much smaller than basic and pharmacological output. Public claim P5 (pharmaceutical-cannabinoid findings = cannabis findings) is unsupported (C022, D10).

Claim type: synthesis-conclusion · Origin: Both (identical verdicts) · Sources: as cited · Final disposition: Accepted · Status: active

End-of-era statement (brief §3 I-4) — draft, for the interim synthesis

Reasonably established in humans by end-1999 (controlled trials, small): pharmaceutical cannabinoid antiemesis in chemotherapy (C012–C015); dronabinol appetite effect in AIDS anorexia — appetite, not weight (C021); acute IOP reduction and acute bronchodilation as physiological effects (C017, C018); acute heart-rate increase with IV THC (C017). Acute cognitive/psychomotor effects are not yet ledgered for P1 — no claim carries them; add before the synthesis uses that language. Established in molecular/animal systems, not as human treatment evidence: cannabinoid receptors and endogenous ligands (C007–C011). Suggested but not established: cancer-pain analgesia (C016), spasticity (C020), CBD in epilepsy (C019) — tiny trials with compromised blinding. Asserted in period sources but for which this round did not identify traceable supporting evidence: human cancer disease-modification or neuroprotection (C038); a broad “medical cannabis” proven across diseases (C036, C040); suppression/destruction of specific official reports (C034 — commonly stated).

Held for primary-document checks before v0.2

C024 (Canada year only) · C025 (Congressional Record PDFs) · C028 (IOM 1999 p. 214; PHS release) · C029 (54 FR 53767 text) · C035 (conclusions chapter + Hansard 3 Dec 1998 wording) · C037 (four commission texts) · C006 (Turner 1980 “61”; Gill 1970 THCV) · C016 (codeine equivalence) · C019 (Cunha per-patient breakdown).

Part 2 — Modern Clinical Development, 2000–2017 · ledger v1.0 — SIGNED OFF by Danny 2026-10-10 · P2 LOCK 2026-10-14

Status: v1.0 SIGNED OFF by Danny 2026-10-10. Corrections #5–#13 applied 2026-10-13 after the §M10a cross-audit (approved by Danny; corrections.md, dispositions). Changed: C041, C048, C053, C055, C062, C064, C065, C067, C071, C074, C077, C078 and the held list. C078 is Accepted with qualification (only the Whiting patient-rated NRS figure is still held); C077 returned to Accepted after the Table 3 check (corrections #14–#16, 2026-10-14). Trail: v0.1 Claude (2026-10-09) → ChatGPT calibration (7 findings; raw, dispositions) → v0.2 (changed C048, C052, C056, C067, C076, C080, C084, end-of-era statement) → ChatGPT review of v0.2: sign-off YES with two documentation corrections, C067 denominator and held-list heading (raw) → v1.0. The held items stay open into the interim P2 synthesis and §M10a cross-audit; any change to an active claim is a logged correction, not a silent edit. Sign-off covers the historical evidence-map ledger. It does not mean every full-text assessment or study-level risk-of-bias evaluation is complete. Built from the passes of record (ChatGPT Stage 1 + 2; Gemini Stage 1 rerun + Stage 2), both correction rounds, and the verifier’s checks in critique-P2.md §2–§6c and the verification log. Evidence search cutoff: 2026-09-20 (MED01-wide); official-record checks dated 2026-10-09 describe the records as opened that day. Sources: provisional S-308–S-445 in source-library-additions-P2.md; P1 sources S-242, S-263, S-264, S-265 reused. Keys: origin and access keys as in Part 1. Origin weighting (critique-P2 §1): ChatGPT ran with memory on; Gemini’s identifiers were unreliable (27 of 62 PMIDs wrong-object or nonexistent in run 1; four canonical counts fabricated, admitted in correction round). For Gemini-origin items, topic convergence counts, but every identifier and figure below was re-resolved by the verifier; no Gemini figure is ledgered unless the verifier found it in the source. Period rule (brief rule 6a): every claim states what was reported in 2000–2017. Later events (2018+) appear only as labelled retrospective notes.

A. Shape of the literature

MED01-C041 — The canonical PubMed/MEDLINE query (brief §4a) returns 4,335 (2000–04), 7,451 (2005–09), 10,316 (2010–14) and 8,599 (2015–17) indexed records, 30,701 in total; output rose through the window, and the three-year 2015–17 period had the highest annual rate (≈2,866 per year vs ≈2,063 in 2010–14).

Claim type: literature-shape · Outcome type: n/a
Origin: verifier-found (ChatGPT could not run the query and did not estimate, as required; Gemini’s four figures were not reproducible and were presented as live index figures and then admitted to be “estimated from background training approximations”; withdrawn)
Sources: S-308 (verifier run 2026-10-06, same term and date type as P1 Query A, so the P1→P2 series is continuous)
Best access: DOC · Evidence base: official database count · design n/a
Base grading: established · direct · high for the shape
Qualifications: counts are of indexed records, not of research effort; non-English and non-indexed literature and conference abstracts are under-represented; MeSH indexing lag affects the most recent years.
Final disposition: Accepted with qualification · Status: active

MED01-C042 — Bibliometric analyses report that, between 2000 and 2017, PubMed publications on cannabis grew about 4.5-fold and on medical cannabis almost 9-fold, against 2.5-fold growth of PubMed overall; psychiatry was the largest field, and the indications with most medical-cannabis papers were HIV (261), chronic pain (179), MS (118), nausea/vomiting (102) and epilepsy (88).

Claim type: literature-shape · Origin: ChatGPT (Treister-Goltzman); Both cited Ng & Chang 2022 and Gemini cited Matielo 2018 (as “Sweileh et al.”)
Sources: S-309 Treister-Goltzman et al. 2019 · S-242 Ng & Chang 2022 · S-310 Matielo et al. 2018
Best access: AB (S-309); FT for S-242 in P1, but no 2000–2017 subset extracted; BIB (S-310) · Evidence base: observational (bibliometric) · retrospective (published 2019/2022)
Base grading: probable · direct · medium
Verification: S-309 figures confirmed in abstract. Gemini’s “>40%” US share and its author attributions are not ledgered (wrong authors on both papers, corrected in correction round).
Qualifications: the fold-changes depend on the authors’ search terms (PubMed + Web of Science), not the canonical query in C041; the sources are retrospective summaries of the period, used for shape only.
Final disposition: Accepted with qualification · Status: active

MED01-C043 — The period’s own record shows three turning points: (1) 2003–2007 — the first large-scale randomized programmes of cannabis-based medicines (CAMS; nabiximols in MS and pain) and the first marketing authorization of a cannabis-derived product (Canada 2005); (2) 2006–2009 — the rise and withdrawal of the CB1 antagonist rimonabant; (3) 2013–2017 — the paediatric CBD epilepsy pathway, ending in a controlled trial (2017), alongside the first national evidence assessment of the era (NASEM, January 2017).

Claim type: literature-shape / synthesis-conclusion · Origin: Both (each proposed phases; boundaries differ)
Sources: C041 counts and the dated primary sources in C048, C055, C062, C064, C079
Best access: mixed · Evidence base: inference from verified dated sources; no single source states it
Base grading: probable · direct · medium
Final disposition: Accepted with qualification (interpretive; boundaries editorial) · Status: active

B. Science that redirected the field (non-human unless stated — mechanism/plausibility only, §M3)

MED01-C044 — Endocannabinoids were shown to act as retrograde messengers at synapses (Wilson & Nicoll 2001, rat hippocampal slices; reviewed by Alger 2002), and mice lacking fatty-acid amide hydrolase (FAAH) showed elevated brain anandamide and supersensitivity to it (Cravatt et al. 2001) — findings that made enzyme inhibition (C064) a drug target.

Claim type: discovery · Origin: Gemini (all three; PMIDs for Alger and Cravatt and Cravatt’s DOI were wrong-object or nonexistent — corrected by verifier)
Sources: S-311, S-312, S-313
Best access: AB (S-311, S-313); BIB (S-312) · Evidence base: primary · animal / in-vitro · period
Base grading: established · direct · high (as non-human findings)
Verification: ✅ identities after correction; Gemini’s Wilson & Nicoll title (“physiological synapses”) was not the published title (“hippocampal synapses”).
Qualifications: no human treatment effect implied (D1).
Final disposition: Accepted · Status: active

MED01-C045 — In 2015, cannabidiol was reported to act as a negative allosteric modulator of the CB1 receptor in cell systems (Laprairie et al.), one of several mechanisms proposed in the period for CBD’s distinct pharmacology.

Claim type: discovery · Origin: Gemini · Sources: S-314
Best access: AB · Evidence base: primary · in-vitro · period · Base grading: probable · direct · medium
Qualifications: in-vitro only; no human mechanism established (D1). Gemini’s further mechanism claims (5-HT1A, GPR55 mediation of clinical effects; PC8 answer) were not checked against sources and are not ledgered.
Final disposition: Accepted with qualification · Status: active

MED01-C046 — “Clinical endocannabinoid deficiency” was proposed in 2004 (Russo) as a hypothesis linking migraine, fibromyalgia and irritable bowel syndrome to deficient endocannabinoid signalling; follow-up reviews in 2014 and 2016 (the latter by the hypothesis’s author) collected supportive but indirect data; no independent confirmatory test of the hypothesis was identified in the documented searches.

Claim type: hypothesis-status · Origin: Both
Sources: S-315 Russo 2004 · S-316 Smith & Wagner 2014 · S-317 Russo 2016
Best access: AB · Evidence base: narrative reviews · period
Base grading: probable · direct · medium (for the hypothesis’s status)
Qualifications: the claim is about the hypothesis’s evidentiary status in 2017, not whether it is true.
Final disposition: Accepted with qualification · Status: active

MED01-C047 — Period human studies comparing cannabis with isolated THC, or varying minor constituents, found the subjective effects of smoked or oral whole-plant cannabis similar to those of matched THC (Wachtel 2002; Ilan 2005), and the one identified therapeutic head-to-head comparison found no difference between cannabis extract, THC alone and placebo in cancer cachexia (Strasser 2006). Experimental studies in healthy volunteers found that CBD can oppose some THC effects (Bhattacharyya 2010, fMRI; Englund 2013, THC-induced paranoia). Constituent “synergy” was argued in a 2011 review as a hypothesis (Russo).

Claim type: mechanism / comparative · Origin: ChatGPT (Wachtel, Ilan, Bhattacharyya, Russo 2011); Gemini (Englund; Gemini’s PMID was wrong-object); verifier-found (Strasser)
Sources: S-319, S-320, S-357, S-321, S-322, S-318
Best access: AB · Evidence base: primary · RCT (healthy users / volunteers; crossover or within-subject) + one patient RCT · period
Base grading: probable · direct · medium
Qualifications: the volunteer studies measured subjective and psychotomimetic effects, not therapeutic outcomes; doses and CBD ratios were limited. No study of therapeutic superiority of whole-plant preparations over matched single compounds was identified.
Final disposition: Accepted with qualification · Status: active

C. What was studied in humans (findings of the period, by condition; not present-day verdicts)

Claim type: clinical-finding (period) · Outcome type: clinician-rated and patient-reported spasticity
Origin: Both (CAMS, Novotna); Gemini (Wade 2004, Collin 2007/2010, MUSEC); verifier-found (Corey-Bloom, Vaney 2004, CAMS 12-month, Wade 2010 MA, Kavia 2010, Pooyania 2010)
Sources: S-323, S-324, S-325, S-326, S-327, S-328, S-329, S-330, S-331, S-332, S-384, S-385
Best access: AB · Evidence base: primary · RCTs (parallel; one enriched-withdrawal; crossovers) + one meta-analysis · period
Base grading: established (that this was trialled and the results as stated) · direct · high
Verification: ChatGPT’s CAMS figures exact (correction round 1). Gemini’s CAMS THC interval “−0.36 to 2.24, p=0.14” and “MRC-funded” not supported (abstract; funder per PubMed = MS Society). The “up to 78%” correct-guess figure and capsule composition are full-text only — Inconclusive, held. Novotna’s adjusted difference “−0.84” (Gemini) is not in the abstract; a secondary NCBI Bookshelf table gives −0.84 (−1.29 to −0.40) (ChatGPT cross-audit; convergence, held). Wade 2004 primary-outcome label corrected 2026-10-13 (cross-audit; correction #5).
Qualifications: recurring split between patient-reported benefit and clinician-measured (Ashworth) null results; the enriched design selects early responders, so its effect applies to responders; most trials ≤15 weeks. Also reported: Vaney 2004 (crossover, “might lower spasm frequency”); CAMS 12-month follow-up; Wade 2010 meta-analysis (nabiximols “reduces spasticity”); Kavia 2010 bladder trial (primary not significant); Pooyania 2010 (nabilone in spinal-cord injury, 11 completers, Ashworth reduced, P=.003). Whiting 2015’s spasticity estimates as corrected by the journal show the same split, with patient-rated NRS significant and clinician-rated Ashworth not (C078; convergence, verifier read pending).
Final disposition: Accepted with qualification · Status: active

Claim type: clinical-finding (period) · Origin: Gemini (Rog; PMID corrected); verifier-found (Svendsen, Berman, Schimrigk via registry NCT00959218, Turcotte)
Sources: S-334, S-335, S-336, S-337, S-338
Best access: AB · Evidence base: primary · RCTs (crossover and parallel) · period
Base grading: established · direct · high (as reports)
Qualifications: Schimrigk completed in 2009 (registry) and was published in 2017 with an abstract conclusion emphasising long-term safety; neither researcher identified it. Turcotte 2015 (nabilone added to gabapentin) reported benefit in a small trial.
Final disposition: Accepted with qualification · Status: active

MED01-C050 — Short-term trials of smoked or vaporised cannabis in neuropathic pain reported analgesia: Abrams 2007 (HIV sensory neuropathy, parallel-group: median pain reduction 34% vs 17%, p=0.03; ≥30% responders 52% vs 24%); Ellis 2009 (HIV, crossover); Wilsey 2013 (vaporised, 39 patients, crossover); Ware 2010 (23 recruited, 21 completed; 25 mg of 9.4% THC three times daily for the first five days of each 14-day cycle; difference 0.7 points, 95% CI 0.02–1.4, vs 0% THC); Wallace 2015 (diabetic neuropathy, 16 patients). An individual-patient-data meta-analysis of five trials estimated short-term relief for about 1 in 5–6 patients (Andreae 2015).

Claim type: clinical-finding (period) · Origin: Both (Abrams 2007, Ware 2010); Gemini (Ellis, Wilsey, Wallace); verifier-found (Andreae)
Sources: S-339, S-340, S-341, S-342, S-343, S-344
Best access: AB; registry DOC for Abrams (NCT00046722) and Ware (ISRCTN68314063) · Evidence base: primary · RCTs (mostly crossover, days to weeks) + IPD meta-analysis · period
Base grading: established · direct · high (as reports)
Verification: ChatGPT exact on Abrams and Ware. Gemini’s Abrams “crossover … p=0.016” and Ware “−0.7, CI −1.37 to −0.03, p=0.023 … three times daily for 14-day periods” not supported. Verifier-found: Ware 2010 was registered on 2006-06-21, after its recorded end date (retrospective registration).
Qualifications: small samples, short exposure, research cannabis of fixed potency, experienced users over-represented, psychoactive unblinding risk; effects are short-term.
Final disposition: Accepted with qualification · Status: active

MED01-C051 — In cancer pain and peripheral neuropathic pain, nabiximols results were mixed: Johnson 2010 (177 patients) reported THC:CBD superior to placebo on its primary pain outcome while THC alone was not; Portenoy 2012’s primary 30%-responder analysis was not significant (P=.59), with secondary continuous-responder analyses favouring the low and medium doses; Nurmikko 2007 (125 patients, allodynic peripheral neuropathic pain) reported benefit; Selvarajah 2010 (30 patients, diabetic neuropathy) found no difference from placebo.

Claim type: clinical-finding (period) · Origin: Both (Portenoy — ChatGPT survey line only); Gemini (Johnson, Nurmikko); verifier-found (Selvarajah)
Sources: S-345, S-346, S-347, S-348
Best access: AB · Evidence base: primary · RCTs (parallel) · period
Base grading: established · direct · high (as reports)
Verification: Gemini’s CAND-23 reported only Portenoy’s secondary result as “statistically significant pain reduction” — D11 primary-outcome omission; confirmed in correction round 1.
Final disposition: Accepted · Status: active

Claim type: clinical-finding (period) · Origin: Both (Skrabek); ChatGPT (Ware 2010 nabilone; Blake 2006, rheumatoid arthritis pilot); verifier-found (Frank, Walitt; Narang via registry)
Sources: S-349, S-350, S-351, S-352, S-353, S-354
Best access: AB · Evidence base: primary · RCTs (crossover, parallel; active comparators) + Cochrane review · period
Base grading: established · direct · high (as reports)
Qualifications: Blake 2006 (rheumatoid arthritis, nabiximols, 58 patients over 5 weeks) reported mostly mild or moderate adverse effects and no adverse-effect withdrawals or serious adverse effects on active treatment. Primary-outcome status and adverse events are given here only where the abstract states them.
Final disposition: Accepted · Status: active

MED01-C053 — For chemotherapy-induced nausea and vomiting, the period’s evidence consisted mainly of pre-2000 trials synthesised in 2001 (Tramèr: 30 randomised comparisons, 1,366 patients) and 2015 (Smith, Cochrane: cannabis-based medications “may be useful” for refractory CINV, limited by trial methods); the one identified new comparison with a modern 5-HT3 antiemetic reported similar total response rates for dronabinol and ondansetron in delayed CINV (54% vs 58%; combination 47%), in a small trial not designed to show equivalence (Meiri 2007: 64 randomized, 61 analysed).

Claim type: clinical-finding / synthesis-conclusion · Origin: Both (Tramèr); verifier-found (Meiri, Smith 2015 as a period source)
Sources: S-264, S-263, S-355
Best access: AB · Evidence base: systematic reviews + one RCT · period
Base grading: established · direct · high
Verification: Tramèr’s abstract reports “30 randomised comparisons”; its full text says “We analysed data from 30 randomised controlled trials”. The verifier’s earlier remark that Gemini’s “30 randomized controlled trials” was an error is withdrawn (cross-audit; correction #10).
Qualifications: most underlying trials predate modern antiemetic regimens (5-HT3 and NK1 antagonists), which limits applicability; see C079 for NASEM’s rating.
Final disposition: Accepted · Status: active

MED01-C054 — For appetite and cachexia, controlled trials did not show advantage over standard care or placebo in cancer: megestrol acetate gave better appetite palliation than dronabinol (Jatoi 2002, 469 patients: 75% vs 49%, P=.0001; combination no better), and cannabis extract, THC and placebo did not differ (Strasser 2006, 243 randomized). In HIV, a within-subject laboratory study found that both dronabinol and smoked cannabis increased food intake in cannabis-experienced patients (Haney 2007), a 21-day trial found no adverse effect of smoked or oral cannabinoids on viral load, CD4/CD8 counts or protease-inhibitor levels (Abrams 2003), and a 2013 Cochrane review judged efficacy evidence lacking (Lutge). A small anorexia-nervosa trial reported slight weight gain (Andries 2014).

Claim type: clinical-finding (period) · Origin: verifier-found (all; Gemini’s era table alluded to Jatoi without citation); Andries and Haney via registry
Sources: S-356, S-357, S-358, S-359, S-360, S-361
Best access: AB · Evidence base: primary · RCTs + Cochrane review · period
Base grading: established · direct · high (as reports)
Qualifications: neither researcher identified the two largest cancer-cachexia trials (keystone omission, critique §6a).
Final disposition: Accepted · Status: active

MED01-C055 — Paediatric treatment-resistant epilepsy became a cannabinoid target between 2013 and 2017 along this documented path: a 2014 Cochrane review found four small CBD trials (48 patients) and no reliable conclusions (Gloss); a 2013 parent survey of CBD-enriched cannabis use (Porter & Jacobson) and a 2016 Israeli retrospective series of 74 children (Tzadok) were uncontrolled reports; an open-label expanded-access programme (Devinsky 2016: patients aged 1–30 with childhood-onset epilepsy; 214 enrolled, 162 in the safety and 137 in the efficacy analysis) reported a 36.5% median reduction in motor seizures with adverse events in 79% and serious adverse events in 30%; and the first randomized trial (Devinsky 2017, Dravet syndrome, 120 patients, GW-funded, NCT02091375) reported median monthly convulsive seizures falling from 12.4 to 5.9 with CBD vs 14.9 to 14.1 with placebo (adjusted difference −22.8 percentage points, 95% CI −41.1 to −5.4; P=0.01), with more adverse events. Open-label studies documented interactions with clobazam (Geffrey 2015: N-desmethylclobazam +500% in 13 children) and other antiepileptics (Gaston 2017).

Claim type: clinical-finding / development pathway · Origin: Both (Devinsky 2016/2017, Porter & Jacobson, Geffrey); ChatGPT (Gloss, Gaston); Gemini (Tzadok — PMID wrong-object, verifier located 26800377)
Sources: S-362, S-363, S-366, S-364, S-365, S-367, S-368
Best access: AB; registry DOC for NCT02091375 · Evidence base: Cochrane review; observational (survey, retrospective series); open-label interventional; one RCT · period
Base grading: established · direct · high (as reports)
Verification: Gemini’s “driven by CYP2C19 inhibition” (Geffrey) is not in the abstract — Inconclusive, held. FDA orphan-designation dates were not verified (OOPD query failed) — not ledgered.
Qualifications: uncontrolled change is labelled uncontrolled; the expanded-access and survey populations were self-selected. Retrospective: the Lennox-Gastaut trials (2018) and FDA approval of Epidiolex (NDA 210365, 2018-06-25) fall in P3.
Final disposition: Accepted · Status: active

Claim type: clinical-finding (period) · Origin: Gemini (Bergamaschi, Jetly, van den Elsen — Jetly and van den Elsen PMIDs wrong-object, corrected); verifier-found (Leweke, Müller-Vahl, Curtis, Krishnan)
Sources: S-369, S-370, S-371, S-372, S-373, S-374, S-375, S-376
Best access: AB · Evidence base: primary · RCTs (small) + Cochrane reviews · period
Base grading: established · direct · medium (as reports; small samples)
Qualifications: primary-outcome status is stated only where the abstract names it (van den Elsen). For Leweke, Müller-Vahl, Bergamaschi and Jetly, the outcome given is the authors’ abstract-level finding, and its primary status was not separately verified.
Final disposition: Accepted with qualification (single small trials; no replication within the window) · Status: active

MED01-C057 — In Crohn’s disease, a 21-patient trial of smoked cannabis did not meet its primary outcome (remission 5 of 11 vs 1 of 10, P=.43) while reporting clinical response in 10 of 11 vs 4 of 10 (Naftali 2013), and a later low-dose CBD trial found no benefit (Naftali 2017).

Claim type: clinical-finding (period) · Origin: Both (Naftali 2013); verifier-found (Naftali 2017)
Sources: S-377, S-378
Best access: AB; registry DOC (NCT01040910) · Evidence base: primary · RCTs (small) · period
Base grading: established · direct · medium
Verification: Gemini’s “without endoscopic mucosal healing” comes from a 2021 study — anachronism (rule 6a); not ledgered. Verifier-found: the current registry record lists “reduction of CDAI by 70 points” (response) as primary, while the paper’s primary was remission — registered-vs-published mismatch; registry history not yet read (held).
Final disposition: Accepted with qualification · Status: active

MED01-C058 — In a six-patient crossover study in ocular hypertension or early glaucoma, a single 5 mg sublingual dose of THC lowered intraocular pressure at 2 hours (23.5 vs 27.3 mm Hg, P=0.026) with return to baseline by 4 hours, while CBD 20 mg did not lower it and 40 mg produced a transient rise (Tomida 2006).

Claim type: clinical-finding (period) · Origin: Both (Stage 2) · Sources: S-379
Best access: AB · Evidence base: primary · RCT (crossover, single dose) · period · Base grading: established · direct · medium
Qualifications: acute effect only; the period produced no trial of sustained IOP control. See C079 (NASEM 4-14).
Final disposition: Accepted · Status: active

MED01-C059 — The only human study of cannabinoid anticancer treatment identified in the window was a phase I pilot in nine patients with recurrent glioblastoma given intratumoral THC (Guzmán 2006), whose primary endpoint was safety; it reported reduced tumour-cell proliferation markers in two patients and drew no efficacy conclusion. Other human reports were single case reports (e.g. Singh & Bali 2013); a 2014 systematic review of glioma studies drew mainly on experimental work (Rocha).

Claim type: clinical-finding / absence · Origin: Both (Guzmán, Stage 2); Gemini (Singh & Bali; Rocha — Gemini’s Rocha identifiers were wrong-object in the pass and its correction-round substitute was a different 2008 paper; verifier located 24142199)
Sources: S-380, S-382, S-381
Best access: AB · Evidence base: primary · phase I uncontrolled (n=9); case report; systematic review · period
Base grading: established · direct · high (for what exists)
Final disposition: Accepted · Status: active

MED01-C060 — The one large human disease-modification trial identified in the window, CUPID (Zajicek 2013: 498 patients with progressive MS randomized to oral THC or placebo for three years), found no effect on progression (HR 0.92, 95% CI 0.68–1.23; p=0.57).

Claim type: clinical-finding · Origin: verifier-found (neither researcher identified it; both made disease-modification absence statements) · Sources: S-383
Best access: AB · Evidence base: primary · RCT (parallel) · period · Base grading: established · direct · high
Final disposition: Accepted · Status: active (mandatory §M10a cross-audit target)

D. Products and regulatory decisions (approval ≠ efficacy evidence, §M8)

MED01-C061 — Period human studies used distinct, non-interchangeable products: oromucosal THC:CBD extract spray (nabiximols/Sativex), oral cannabis extract, oral dronabinol (later also an oral solution), nabilone, purified oral CBD, CBD-enriched cannabis oils of variable composition, and smoked or vaporised research cannabis of fixed THC content; vaporisation was evaluated as a smokeless delivery route (Hazekamp 2006, device study; Abrams 2007, pilot in volunteers).

Claim type: product-description · Origin: Both (product classes); Gemini (Hazekamp 2006, Abrams 2007b) · Sources: S-400, S-401, and the trial sources in C047–C060
Best access: AB · Evidence base: inference from verified trial records · period · Base grading: established · direct · high
Qualifications: findings for one product do not transfer to another (D10).
Final disposition: Accepted · Status: active

MED01-C062 — Sativex (nabiximols; DIN 02266121) received a Health Canada Notice of Compliance with conditions on 2005-04-15 for adjunctive treatment of neuropathic pain in MS and on 2007-08-01 for advanced cancer pain, a UK marketing authorisation on 16 June 2010 for MS spasticity (named-patient import having been licensed in 2006), and a Spanish authorisation on 20 July 2010.

Claim type: regulatory-event · Origin: Both
Sources: S-386 (Health Canada NOC/DPD), S-387 (GOV.UK/MHRA), S-446 (UK SmPC), S-447 (AEMPS ficha técnica)
Best access: DOC · Evidence base: regulatory records · period · Base grading: established · direct · high
Verification: dates and control numbers confirmed 2026-10-09. Gemini’s DINs 02269926 / 02268571 do not exist (corrected). UK day resolved 2026-10-13: “Date of first authorisation 16 June 2010” (current UK SmPC §9). Spain: “Fecha de la primera autorización: 20/julio/2010” (AEMPS ficha técnica §9, opened 2026-10-13). Australia: ARTG date 26 November 2012 per ChatGPT’s cross-audit — convergence, verifier read pending (correction #11).
Qualifications: conditional approvals required confirmatory studies. Retrospective (2019-12-12): Health Canada removed both conditional indications because confirmatory studies did not confirm benefit — neither researcher reported this.
Final disposition: Accepted with qualification · Status: active

MED01-C063 — In the US, an oral dronabinol solution (Syndros, NDA 205525) was approved on 2016-07-01 and placed in Schedule II by DEA rule of 2017-03-23 (82 FR 14815); the 2006 FDA action on Cesamet (NDA 018677) was a labeling supplement, not an approval (original approval 1985-12-26; P1 C024). No plant-derived cannabinoid medicine was approved by the FDA within the window; Epidiolex was approved on 2018-06-25 (P3).

Claim type: regulatory-event · Origin: Both (Syndros); ChatGPT (Schedule II rule; Cesamet not counted as a P2 approval) · Sources: S-388, S-389
Best access: DOC · Evidence base: regulatory records · Base grading: established · direct · high
Verification: Gemini’s “FDA approved Cesamet, May 15, 2006” — not supported; its correction-round re-description (“efficacy supplement approval”) also wrong.
Final disposition: Accepted · Status: active

MED01-C064 — Drugs targeting the endocannabinoid system reached humans with adverse outcomes: the CB1 antagonist rimonabant produced weight loss in large trials (RIO-Europe, Van Gaal 2005; RIO-North America, Pi-Sunyer 2006, 3,045 patients), was authorised in the EU on 2006-06-19 and withdrawn on 2009-01-16 after psychiatric-risk review; the FAAH inhibitor PF-04457845 was stopped for futility in osteoarthritis pain (Huggins 2012, 74 patients); and the FAAH inhibitor BIA 10-2474 caused an acute neurological disorder in a 2016 phase I trial — an acute progressive syndrome in three of the four participants reported, one of whom became brain dead (Kerbrat 2016) — with off-target lipase inhibition later demonstrated (van Esbroeck 2017). The French regulatory inquiry reported one death and four other participants with severe neurological disorders (IGAS 2016, as quoted in the cross-audit; convergence, verifier read pending).

Claim type: drug-development / harm · Origin: Both (rimonabant; BIA 10-2474; van Esbroeck); Gemini (RIO trials, Huggins — PMIDs wrong-object, corrected)
Sources: S-390, S-391, S-392, S-393, S-394, S-395
Best access: DOC (EMA); AB · Evidence base: regulatory record; RCTs; phase I case series; in-vitro · period
Base grading: established · direct · high
Verification: Gemini’s “five healthy volunteers … haemorrhages and necrosis” is not supported. Its “one death and four hospitalizations” is not in the NEJM abstract, which reports a four-person consenting subset, but it is consistent with the regulatory record as quoted by ChatGPT. The verifier’s earlier ruling that it was unsupported is withdrawn as overbroad (cross-audit; correction #11).
Final disposition: Accepted · Status: active

MED01-C065 — Therapeutic-use patents of the period included US 6,630,507 (US Department of Health and Human Services; cannabinoids as antioxidants and neuroprotectants; granted 2003-10-07, expired 2019-04-21) and GW Pharma CBD-for-epilepsy patents (US 9,066,920, with Otsuka, granted 2015-06-30, claim 1 = partial seizures; US 9,474,726, granted 2016-10-25, claim 1 = FIRES); a Hadasit/Yissum patent claimed CBD for Type II diabetes (US 8,071,641, granted 2011-12-06; claim 1, verified 2026-10-13).

Claim type: regulatory-event (intellectual property) · Origin: Both (US 6,630,507); ChatGPT (US 9,474,726); Gemini (US 9,066,920, US 8,071,641)
Sources: S-396, S-397, S-398, S-399
Best access: DOC (Google Patents) · Evidence base: patent records · Base grading: established · direct · high
Verification: ChatGPT’s description of US 9,474,726 as covering Dravet syndrome mischaracterises claim 1 (FIRES). Gemini’s “Dravet/LGS” for US 9,066,920 and its attribution of US 8,071,641 to GW not supported.
Qualifications: a patent records a claim of invention, not evidence that the use works.
Final disposition: Accepted · Status: active

E. Study quality and reporting

MED01-C066 — The controlled evidence of the period was numerous but mostly small and short: a 2015 systematic review found 79 trials (6,462 participants), only four at low risk of bias (Whiting); designs were often crossover or enriched-enrolment; blinding was a recognised concern for psychoactive interventions (CAMS reported “a degree of unmasking” in active arms), but an analysis of three nabiximols trials (666 patients) found no evidence of widespread unblinding (Wright 2012).

Claim type: quality-characterisation · Origin: Both (Whiting); verifier-found (Wright)
Sources: S-402, S-333, and the trials in C048–C060
Best access: AB · Evidence base: systematic review; secondary analysis of RCTs · period
Base grading: established · direct · high
Qualifications: CAMS’s “correct-guess” rate is full-text only (held, C048).
Final disposition: Accepted with qualification · Status: active

Claim type: reporting-characterisation · Origin: verifier-found · Sources: S-403, S-342, S-377, S-337, S-448
Best access: DOC (registries); AB (matched publications) · Evidence base: registry records · period (publication status assessed as of 2026)
Base grading: probable · direct · medium
Verification: Corrected 2026-10-13 (cross-audit B2-02; correction #8):
- The v1.0 text named NCT01898520 as a trial with no publication identified. It was published in 2020 (Fairhurst, Dev Med Child Neurol; PMID 32342496: no significant difference on the primary spasticity NRS; three cases of hallucinations). The title search scheduled in critique §6b had not been run.
- The re-screen dropped the intervention-name filter, re-ran every lookup (0 errors) and applied the explicit therapeutic rule. Twelve of the original “32 patient trials” were dependence-treatment, healthy-volunteer or pharmacokinetic studies.
- Trial-level lists are in critique-P2 §6d.
Qualifications:
- This is a reporting-disclosure control, not an estimate of unpublished trials or of publication bias.
- The 11 are trials with no publication identified, not trials known to be unpublished. Publication after 2017 is a P3 question.
- WHO ICTRP and EU-CTR were not searched, so EU-registered nabiximols trials are not covered.
Final disposition: Accepted with qualification · Status: active (mandatory §M10a cross-audit target)

F. Conditions of research

MED01-C068 — In the US, research cannabis came from a single federally contracted source throughout the window; a challenge to the DEA’s refusal to register an additional manufacturer failed (Craker v. DEA, 1st Cir., 2013-04-15); the DEA announced in August 2016 that it would accept further manufacturer applications (81 FR 53846), with no new registrant within the window; period commentators described Schedule I as an obstacle to research (Nutt 2013; Stith & Vigil 2016), and NASEM (2017) recommended addressing regulatory and supply barriers (Recommendation 4). A state-funded programme, the University of California Center for Medicinal Cannabis Research, ran controlled trials in the period (described by Grant 2012).

Claim type: conditions-of-research · Origin: Both (single supply, NASEM, 81 FR, CMCR); Gemini (Craker, Nutt, Stith & Vigil, Grant — PMIDs/DOIs wrong-object or nonexistent, corrected)
Sources: S-404, S-405, S-406, S-407, S-408, S-411
Best access: DOC (court, Federal Register, NASEM); AB/BIB (commentaries)
Evidence base: legal and regulatory records; national assessment; commentary · period
Base grading: established · direct · high (records); commentary = attributed opinion
Qualifications: “no new registrant within the window” is from the period record as searched; the documented effect of these conditions on specific studies is limited to the cases named — no quantitative estimate of research forgone is supportable.
Final disposition: Accepted with qualification · Status: active

MED01-C069 — National medical-cannabis programmes produced some published data in the window: Dutch pharmacy records showed 5,540 patients receiving prescription cannabis in 2003–2010 (Hazekamp 2013); a Canadian survey of 271 registered patients reported substitution for prescription drugs (Lucas & Walsh 2017, see C083); and a one-year prospective Canadian cohort found more non-serious adverse events among cannabis users than controls but no difference in serious events (COMPASS, Ware 2015).

Claim type: conditions-of-research / observational finding · Origin: ChatGPT (Hazekamp 2013, Lucas & Walsh); Gemini (COMPASS)
Sources: S-409, S-410, S-412
Best access: AB · Evidence base: observational (records cohort; survey; prospective cohort) · period
Base grading: probable · direct · medium
Qualifications: programme legal bases and start dates (Netherlands 2003; Canada 2001 and 2014 regimes; Israel; US states) as reported by the researchers were not checked against primary legal texts — held for the synthesis.
Final disposition: Accepted with qualification · Status: active

G. Harms research

MED01-C070 — Prospective cohort and meta-analytic evidence associated cannabis use with later psychotic outcomes: in 50,087 Swedish conscripts, heavy use (>50 occasions) was associated with schizophrenia (adjusted OR 6.7, 95% CI 2.1–21.7, cannabis-only users; Zammit 2002); further cohorts reported similar associations (Arseneault 2002; Henquet 2005); a 2007 systematic review reported a pooled adjusted OR of 1.41 (1.20–1.65) for any psychotic outcome among ever-users with a dose–response gradient (Moore); a 2016 meta-analysis confirmed dose–response while stating a causal link could not be unequivocally established (Marconi); heavy use was also associated with depressive disorders (Lev-Ran 2014). NASEM (2017) found substantial evidence of a statistical association with schizophrenia and other psychoses, highest among the most frequent users.

Claim type: harm-finding · Origin: Both (Moore); Gemini (Zammit, Arseneault, Henquet — Zammit PMID and Henquet DOI wrong, corrected); verifier-found (Marconi, Lev-Ran)
Sources: S-413, S-414, S-415, S-416, S-417, S-445, S-408
Best access: AB; DOC (NASEM) · Evidence base: observational (cohort) + meta-analyses · period
Base grading: established (association) · direct · high; causation not established
Qualifications: observational designs cannot exclude confounding or reverse causation; Gemini’s Di Forti 2015 (high-potency cannabis) was not opened.
Final disposition: Accepted with qualification · Status: active

MED01-C071 — In the Dunedin birth cohort (1,037 members), persistent cannabis users showed neuropsychological decline from childhood to age 38, concentrated among adolescent-onset users, with incomplete recovery after cessation (Meier 2012); in the full text, adolescent-onset users who used persistently to age 38 (n = 23) lost an average of 8 IQ points, and the most persistent users overall (n = 41) lost 5–6 points; a 2013 re-analysis argued socioeconomic confounding could reproduce the association and called the causal inference premature (Rogeberg). NASEM rated acute impairment of learning, memory and attention as moderately supported.

Claim type: harm-finding · Origin: Both (Meier); Gemini (Rogeberg — PMID and DOI wrong, corrected)
Sources: S-419, S-420, S-408
Best access: AB; DOC · Evidence base: observational (birth cohort); commentary with simulation · period
Base grading: probable · direct · medium
Verification: “an average decline of 8 IQ points” (Gemini) is not in the abstract. The full text (PMC, read 2026-10-13) supports it for the 23 persistent adolescent-onset users only — held item resolved (correction #12). Best access for this component: FT.
Final disposition: Accepted with qualification · Status: active

MED01-C072 — A cannabis withdrawal syndrome was characterised prospectively (Budney 2003: 18 users, 12 ex-users; onset days 1–3, peak days 2–6, most effects lasting 4–14 days) and reviewed as valid and clinically significant (Budney 2004); dependence was treated in randomized trials (Levin 2011: 156 adults, dronabinol improved retention and withdrawal symptoms), and a 2014 Cochrane review found incomplete evidence for any pharmacotherapy (Marshall).

Claim type: harm-finding · Origin: ChatGPT (Budney 2003, Levin); Gemini (Budney 2004 — PMID wrong, corrected; Volkow 2014 review); verifier-found (Marshall)
Sources: S-421, S-422, S-423, S-424, S-425
Best access: AB · Evidence base: primary · observational and RCT; reviews · period
Base grading: established · direct · high
Final disposition: Accepted · Status: active

MED01-C073 — Acute cannabis use was associated with motor-vehicle collision risk in a 2012 meta-analysis of nine studies (OR 1.92, 95% CI 1.35–2.73; fatal collisions 2.10, 1.31–3.36; Asbridge), and in a driving-simulator study blood THC of 8.2 and 13.1 µg/L during driving increased lane weaving similarly to breath alcohol of 0.05 and 0.08 (Hartman 2015, 18 completers). NASEM found substantial evidence of an association with motor-vehicle crashes.

Claim type: harm-finding · Origin: Gemini (Asbridge — PMID wrong, corrected); ChatGPT (Hartman) · Sources: S-426, S-427, S-408
Best access: AB; DOC · Evidence base: meta-analysis of observational studies; within-subject RCT · period
Base grading: established · direct · high
Final disposition: Accepted · Status: active

MED01-C074 — Respiratory and cardiovascular research distinguished endpoints: quitting smoked cannabis resolved chronic-bronchitis symptoms in a cohort (Tashkin 2012); in the CARDIA cohort (5,115 enrolled; 5,016 analysed) low cumulative exposure was not associated with lower lung function (FEV1 and FVC were slightly higher at low exposure; the association was nonlinear) (Pletcher 2012); a 2006 systematic review on lung cancer found molecular and histopathologic plausibility (Mehra); and a case-crossover study of 3,882 myocardial-infarction patients identified smoking cannabis as a rare trigger (Mittleman 2001). NASEM found substantial evidence of worse respiratory symptoms with long-term smoking and moderate evidence of no association with lung cancer.

Claim type: harm-finding · Origin: ChatGPT (Tashkin 2012, Mittleman); Gemini (Pletcher; Tashkin 2013 review — PMID wrong, corrected); verifier-found (Mehra)
Sources: S-428, S-429, S-430, S-431, S-432, S-408
Best access: AB; FT (Pletcher analytic N, confirmed in correction round); DOC
Evidence base: observational (cohort, case-crossover); systematic review; narrative review · period
Base grading: established · direct · high
Verification: “FEV1/FVC” in v1.0 named the ratio, which the study did not report as its finding; corrected to FEV1 and FVC (cross-audit; correction #6).
Final disposition: Accepted · Status: active

MED01-C075 — A 2016 meta-analysis of 24 studies found prenatal cannabis exposure associated with maternal anaemia (pooled OR 1.36, 1.10–1.69) and low birth weight (pooled OR 1.77, 1.04–3.01) (Gunn); NASEM found substantial evidence associating maternal smoking with lower birth weight; ACOG advised against use in pregnancy and lactation (2015).

Claim type: harm-finding · Origin: ChatGPT (Gunn, ACOG) · Sources: S-433, S-408, S-437
Best access: AB; DOC · Evidence base: meta-analysis of observational studies · period
Base grading: established (association) · direct · high
Qualifications: confounding by tobacco and other exposures limits causal inference.
Final disposition: Accepted with qualification · Status: active

MED01-C076 — Acute toxicity was documented in two forms. Cannabinoid hyperemesis was described as a syndrome in 2004 (Allen: 19 patients identified, 9 presented; vomiting resolved with cessation in 7; compulsive hot bathing in 9 of 10). NASEM (January 2017) found moderate evidence of increased paediatric cannabis overdose injuries in US states where cannabis use was legal, and insufficient evidence on death due to cannabis overdose; this is a population-level assessment, not evidence from therapeutic trials.

Claim type: harm-finding · Origin: Gemini (Allen; PMID wrong-object, corrected); calibration-added (NASEM 9-4a/9-4b, from the verifier-opened S-408 — ChatGPT finding 7) · Sources: S-434, S-408
Best access: AB; DOC · Evidence base: case series; national evidence assessment · period · Base grading: probable · direct · medium
Final disposition: Accepted · Status: active

MED01-C077 — Harms inside therapeutic use were characterised in the period: a 2008 systematic review of 31 studies of medical cannabinoids (23 randomized trials, 8 observational) found that, within the 23 trials, 4,615 of 4,779 reported adverse events (96.6%) were non-serious; non-serious events were more frequent than in controls (RR 1.86, 1.57–2.21), serious events did not differ significantly (RR 1.04, 0.78–1.39), and long-term risks were poorly characterised (Wang 2008); across a different, larger set of trials, Whiting 2015 reported higher odds of any adverse event (OR 3.03, 2.42–3.80), of withdrawal for adverse events (OR 2.94, 2.18–3.96) and of serious adverse events (OR 1.41, 1.04–1.92) (Table 3 of the corrected article, verified 2026-10-14 via Danny’s browser: any AE 29 trials/3,714 participants; serious 34/3,248; withdrawal 23/2,755); CBD programmes recorded adverse events in most patients and drug interactions (C055); NASEM and a 2017 review (Nugent) found limited evidence on long-term harms in patient populations.

Claim type: harm-finding · Origin: verifier-found (Wang, Nugent); Both (CBD trial harms); Gemini (Whiting AE ORs)
Sources: S-435, S-444, S-364, S-365, S-367, S-368, S-412
Best access: AB · Evidence base: systematic reviews; trial safety data · period
Base grading: established · direct · high
Verification: Wang denominator corrected 2026-10-13: the 96.6% refers to events within the 23 randomized trials (abstract; cross-audit; correction #9). Whiting adverse-event ORs: Gemini (P2 pass) and ChatGPT (cross-audit) agreed. Verified 2026-10-14 from Table 3 of the corrected JAMA article (Danny’s screenshot; correction #16).
Qualifications: neither researcher identified Wang 2008, the principal period harms synthesis for therapeutic cannabinoids (keystone omission). Wang and Whiting used different trials and measures (event rate ratio vs participant odds ratio) and are not repeated measurements of one population.
Final disposition: Accepted · Status: active (mandatory §M10a cross-audit target)

H. What the syntheses said (attributed and dated — never this programme’s finding)

MED01-C078 — Whiting et al. (JAMA 2015; 79 trials, 6,462 participants) reported moderate-quality evidence supporting cannabinoids for chronic pain and spasticity, low-quality evidence for chemotherapy-induced nausea and vomiting, weight gain in HIV, sleep disorders and Tourette syndrome, and an increased risk of short-term adverse events; pooled estimates in the PubMed abstract included CINV OR 3.82 (1.55–9.42) and ≥30% pain reduction OR 1.41 (0.99–2.00). Its spasticity estimate as first published (Ashworth WMD −0.36, −0.69 to −0.05; 7 trials) was later corrected by the journal; the corrected article reports Ashworth WMD −0.12 (95% CI −0.24 to 0.01; 5 trials), which is not significant (verified 2026-10-14). A patient-rated NRS estimate of −0.76 (−1.38 to −0.14; 3 trials) is reported from the full text (convergence; verifier read pending). The corrected abstract gives CINV 47% vs 20% (OR 3.82) and pain reduction 37% vs 31% (OR 1.41), and states “an increased risk of short-term AEs with cannabinoids, including serious AEs”.

Claim type: synthesis-conclusion · Origin: Both · Sources: S-402
Best access: AB · Evidence base: systematic review + meta-analysis · period
Base grading: established · direct · high (as the review’s conclusion)
Verification: Whiting 2015 has four published errata (PubMed ErratumIn: 2015-08-04 “Incorrect Figure Label”; 2015-08-25 “Incorrect Values Reported”; 2015-12-01 “Incorrect Nonproprietary Drug Name and Approved Use”; 2016-04-12 “Incorrect Effect Estimate”). The verifier’s v1.0 check used the uncorrected abstract and missed them (cross-audit B3-01, the most serious P2 finding; correction #13). Read 2026-10-14 via Danny’s browser (JAMA blocks the verifier; Danny’s screenshots were viewed):
- The corrected online abstract gives the Ashworth figure above.
- JAMA’s correction notice “Incorrect Effect Estimate” (JAMA 2016;315(14):1522, doi 10.1001/jama.2016.3470) states the article “was also corrected … on July 13, 2015, for a corrected average reduction to the Ashworth spasticity scale (as reported in the Abstract)”.
- The same notice corrects a sleep sentence: nabilone and insomnia, mean difference −3.25 (95% CI −5.26 to −1.24); sleep restfulness 0.48 (0.01 to 0.95). The ledger does not quote that figure.
- The 2015-08-25 notice did not render for Danny either.
- The patient-rated NRS figure and the adverse-event odds ratios (Table 3) remain held (correction #15). Gemini’s −0.76 is the patient-rated NRS estimate, which Gemini attached to the Ashworth scale; the verifier’s earlier ruling stands only for that mislabel.
Final disposition: Accepted with qualification (the review’s conclusions as attributed; pooled estimates held pending the corrected article) · Status: active

Claim type: synthesis-conclusion · Origin: Both · Sources: S-408 (conclusions 4-1 to 4-20)
Best access: DOC (NAP reader, opened 2026-10-09) · Evidence base: national evidence assessment · period (literature through August 2016)
Base grading: established · direct · high (as NASEM’s conclusion)
Qualifications: NASEM’s epilepsy rating predates the 2017 Dravet RCT (C055); its harms conclusions are in C070–C075.
Final disposition: Accepted · Status: active

MED01-C080 — The American Academy of Neurology’s 2014 systematic review (Koppel; 34 studies, 8 Class I) rated oral cannabis extract effective and nabiximols and THC probably effective for patient-centred measures of MS spasticity, and judged it possible that oral extract and THC were effective for both patient-centred and objective measures at one year. The PubMed record opened during verification (2026-10-06) carries a [RETIRED] designation of unestablished date; that later status is separate from the review’s 2014 conclusions.

Claim type: synthesis-conclusion · Origin: Both (ChatGPT characterised it correctly; Gemini described it as a complementary-and-alternative-medicine guideline with “Level A … established as ineffective” Ashworth ratings — scope and wording do not match Koppel)
Sources: S-436 · Best access: AB · Evidence base: systematic review / guideline · period
Base grading: established · direct · high
Final disposition: Accepted · Status: active

MED01-C081 — Cochrane reviews of the period mostly found evidence insufficient or absent: epilepsy (Gloss 2014: no reliable conclusions), dementia (Krishnan 2009: no evidence), Tourette (Curtis 2009: not enough evidence), HIV/AIDS (Lutge 2013: evidence lacking), fibromyalgia nabilone (Walitt 2016: no convincing evidence), cannabis in schizophrenia (McLoughlin 2014: limited and inconclusive), pharmacotherapy for cannabis dependence (Marshall 2014: incomplete); the exception was CINV (Smith 2015: may be useful for refractory cases).

Claim type: synthesis-conclusion · Origin: ChatGPT (Gloss; Walitt named without identifier); verifier-found (the others)
Sources: S-362, S-375, S-372, S-360, S-353, S-418, S-424, S-263
Best access: AB · Evidence base: Cochrane systematic reviews · period
Base grading: established · direct · high
Qualifications: 15 of the 16 period Cochrane reviews retrieved by the probe were cited by neither researcher.
Final disposition: Accepted · Status: active

MED01-C082 — Professional bodies issued statements in the window, including the American College of Obstetricians and Gynecologists (Committee Opinion 637, 2015, on use in pregnancy and lactation) and the American Academy of Pediatrics (2015 policy statement on marijuana policies and youth, updating its 2004 statement); the ACP (2008) and AMA (2009) statements cited by the researchers were not opened by the verifier.

Claim type: synthesis-conclusion · Origin: ChatGPT (ACOG, AAP, ACP); Gemini (AMA, AAP); Both (ACP)
Sources: S-437, S-438
Best access: AB (ACOG, AAP); — (ACP, AMA)
Base grading: established (existence and date) · direct · high
Verification: the researchers’ summaries of each body’s recommendations (research access, rescheduling) were not checked against the statements — held.
Final disposition: Accepted with qualification (existence and date only); recommendation content Inconclusive · Status: active

I. Opioids, absences and public claims

MED01-C083 — In 2014, an ecological analysis found states with medical-cannabis laws had 24.8% lower mean annual opioid-analgesic overdose mortality over 1999–2010 (95% CI −37.5% to −9.5%; P=.003; Bachhuber), an association the authors did not present as causal; patient surveys reported substitution of cannabis for opioids (Boehnke 2016; Lucas & Walsh 2017: 63% substituted for prescription drugs, 30% for opioids; Reiman 2017); and a pharmacokinetic study in 21 patients on sustained-release morphine or oxycodone found vaporised cannabis did not significantly alter plasma opioid levels while pain decreased (Abrams 2011, uncontrolled). Retrospective (2019): extending the analysis through 2017 reversed the association (from −21% to +23%), and the authors judged it likely spurious (Shover).

Claim type: observational finding · Origin: Both (Bachhuber, Shover); ChatGPT (Lucas & Walsh, Reiman, Abrams 2011); Gemini (Boehnke)
Sources: S-439, S-441, S-410, S-442, S-443, S-440
Best access: AB · Evidence base: observational (ecological; surveys); open-label pharmacokinetic · period + one retrospective
Base grading: probable · direct · medium
Qualifications: no controlled trial of cannabis as an opioid substitute was identified in the window. Gemini’s “proving … an artifact” overstates Shover, who called the association likely spurious.
Final disposition: Accepted with qualification · Status: active

Claim type: absence · Origin: Both (absence statements); verifier-found (CUPID, registry) · Sources: as cited
Final disposition: Accepted with qualification (bounded to documented searches) · Status: active

MED01-C085 — Public-claim rulings (probe labels, not §M9; each describes what period evidence supported by end-2017):

PC1 “essentially no rigorous clinical evidence by 2017” — contradicted (Both): 79 trials with 6,462 participants (C078), a NEJM RCT (C055), and NASEM’s conclusive/substantial ratings (C079).
PC2 “benefits established across many conditions” — partly supported (Both): substantial or moderate evidence for a few indications (C079); limited, null or insufficient for most others (C054, C056–C059, C081).
PC3 “cannabinoids shrink or cure cancers in people” — unsupported (Both): no controlled human anticancer trial; one phase I safety study and case reports (C059).
PC4 “approval of a cannabinoid medicine = approval of medical cannabis” — contradicted (Both): approvals named specific products, formulations and indications (C062, C063); approval is not efficacy evidence for other products (C061, D10).
PC5 “medical-cannabis laws reduced opioid overdose deaths” — unsupported (verifier ruling; ChatGPT unsupported, Gemini partly supported for the period): the period evidence was a single ecological association that cannot establish that the laws reduced deaths (C083); retrospectively contradicted by the 2019 reversal.
PC6 “cannabis use causes schizophrenia” — partly supported (Both): substantial, dose-related association; causation not established (C070).
PC7 “cannabis is not addictive” — contradicted (Both): withdrawal syndrome and dependence treatment trials (C072).
PC8 “CBD has no psychoactive effects and no meaningful risks” — contradicted as worded (verifier ruling; ChatGPT partly supported, Gemini contradicted): the “no meaningful risks” half is contradicted by controlled-trial adverse events and documented drug interactions (C055, C077); the narrower statement that CBD is not THC-like intoxicating is supported (C047), but CBD had measurable central effects in trials (C055, C056).
Claim type: synthesis-conclusion · Origin: Both · Final disposition: Accepted · Status: active

End-of-era statement, P2 (brief §3 I-4) — draft, describes 2017 not today

The three group headings below are the descriptive groups prescribed by the frozen brief (§3 I-4). They are not a certainty scale, and the groups are not ranked against each other. Each entry states its design and number of trials, so that a single trial is not read as replicated evidence. Formal certainty ratings belong only to the period syntheses that issued them: Whiting 2015 (C078) and NASEM 2017 (C079), each attributed and dated.

Replicated or controlled human evidence identified:

Several randomized trials:
- MS spasticity — patient-reported relief with nabiximols and oral cannabinoids; clinician-rated results inconsistent; one enriched-enrolment design (C048).
- Neuropathic pain — short-term relief with smoked or vaporised cannabis in small crossover trials, plus an individual-patient-data meta-analysis (C050); nabiximols results mixed, with two trials whose primary outcome was not met (C051).
- CINV — mainly pre-2000 trials, plus one comparison with ondansetron (C053).
- Cancer appetite and cachexia — two large controlled trials (469 and 243 patients), neither showing an advantage for dronabinol or cannabis extract over megestrol or placebo (C054). (Moved from the second group by correction #4, 2026-10-10.)
A single randomized trial: CBD for convulsive seizures in Dravet syndrome (C055).
Harms: association with psychosis (cohorts and meta-analyses, C070); withdrawal and dependence (C072); crash risk and driving impairment (C073); excess non-serious adverse events in trials (C077); CBD–antiepileptic interactions (open-label studies, C055); paediatric overdose injuries (NASEM, C076).

Preliminary human evidence, or non-human evidence only:

Single or small trials, or uncontrolled human reports: appetite in HIV (a laboratory study and a 21-day safety trial) and anorexia nervosa (one small trial) (C054); schizophrenia, Tourette syndrome, social anxiety, PTSD nightmares (C056); Crohn’s disease (C057); glaucoma — a transient IOP effect (C058); fibromyalgia and sleep (C052); anticancer use — one phase I safety study and case reports (C059).
Healthy-volunteer experiments: constituent interaction (C047).
Non-human only: endocannabinoid biology and CBD mechanisms (C044, C045); anticancer effects in experimental systems (C059).

Claims appearing in period sources for which this search did not identify traceable support: human tumour shrinkage or cure (C059, PC3); disease modification (tested once and negative, C060); endocannabinoid deficiency as an established disease mechanism (C046); a causal reduction of opioid deaths by medical-cannabis laws (C083, PC5); risk-free CBD (PC8).

Held for full-text or primary-document checks — unresolved at P2 ledger v1.0; carried forward to the interim P2 synthesis and §M10a cross-audit

(Updated 2026-10-13 after the §M10a cross-audit. “Convergence” = a cross-auditor quoted the primary record; the verifier has not yet reproduced it.)

C048: CAMS correct-guess rate and capsule composition (held); Novotna adjusted difference (convergence, secondary table).
C055: Geffrey CYP2C19 mechanism (held); FDA orphan designation 2013-11-14 for Dravet (convergence).
C057: NCT01040910 registry history (convergence via a Cochrane risk-of-bias note; held).
C062: TGA ARTG 26 Nov 2012 (convergence). UK day and AEMPS date resolved.
C064: IGAS regulatory-cohort text (convergence; the verifier’s URL returned 404).
C067: ICTRP/EU-CTR limitation (errored lookups and title searches resolved by the re-screen).
C069: programme legal bases (Netherlands 2003-09-01 convergence; others held).
C071: resolved.
C078: Whiting patient-rated NRS spasticity estimate (−0.76) — full text only; held. (The Ashworth correction and the Table 3 adverse-event ORs were verified on 2026-10-14.)
C082: ACP/AMA position text (convergence; amendment check pending).
New: the content of published errata for Devinsky 2016 (C055), Bachhuber 2014 (C083), Van Gaal 2005 and Pi-Sunyer 2006 (C064), and Wilson & Nicoll 2001 (C044). Check that no ledgered figure changes.

Mandatory §M10a cross-audit targets (verifier-found): C041, C054, C060, C067, C077, and the verifier-found sources in C048–C057, C070, C081.

Source file in the research repository: medical/MED01/ledger.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.

Claim Ledger — MED01: Historical Evidence Landscape, medical cannabis & cannabinoid research 1960s → 2026

A. Shape of the literature

MED01-C001 — Two frozen PubMed/MEDLINE queries returned 8,663 and 10,630 indexed cannabis/cannabinoid records for 1960–1999; despite search-string dependence, both show a 1970s surge, a 1980s trough, and a 1990s peak exceeding the 1970s.

B. Discoveries that redirected the field (all non-human evidence — mechanism/plausibility only, §M3). Design tag for pure chemistry (isolation, structure, synthesis) = `mechanistic (chemical/structural; no biological outcome)`, not `in-vitro`.

MED01-C004 — The structure of cannabidiol was published by Mechoulam and Shvo in 1963.

MED01-C007 — A saturable, stereoselective, high-affinity cannabinoid binding site (Kd ≈ 133 pM for [³H]CP-55,940) coupled to G-protein was characterised in rat brain membranes by Devane, Dysarz, Johnson, Melvin and Howlett in 1988.

MED01-C008 — A cannabinoid receptor cDNA was cloned and functionally expressed by Matsuda et al. (NIMH) in 1990, and receptor distribution was mapped autoradiographically across mammalian brains including human tissue by Herkenham et al. (NIMH) the same year.

MED01-C009 — An endogenous brain constituent binding the cannabinoid receptor, arachidonylethanolamide (“anandamide”), was isolated from porcine brain and reported by Devane, Hanuš, Mechoulam and colleagues in 1992.

MED01-C010 — A second cannabinoid receptor, expressed predominantly in peripheral immune tissue and not detected in brain, was cloned by Munro, Thomas and Abu-Shaar (MRC Cambridge) in 1993.

MED01-C011 — 2-Arachidonoylglycerol was identified as a second endogenous cannabinoid-receptor ligand in 1995 by two independent groups (Mechoulam et al., canine gut; Sugiura et al., rat brain).

C. What was studied in humans (clinical findings of the period; not present-day verdicts)

MED01-C012 — In a 1975 randomized, double-blind, placebo-controlled study, oral Δ9-THC produced an antiemetic response in 14 of 20 THC courses versus 0 of 22 placebo courses in patients receiving emetogenic chemotherapy (22 entered, 20 evaluable; completers 12/15 vs 0/14, P < 0.001).

MED01-C020 — In a 1981 double-blind study, oral THC 10 mg reduced clinically measured spasticity versus placebo in a principal group of nine patients (P < 0.01); the blinded examiner correctly identified THC trials in seven of nine cases, indicating compromised blinding.

D. Preparations and pharmaceutical development

F. Conditions of research

G. Harms research

H. What official syntheses concluded (each attributed and dated — never this programme’s finding)

MED01-C034 — The 1997 WHO report Cannabis: A Health Perspective and Research Agenda (WHO/MSA/PSA/97.4) reviewed health effects across organ systems and included a therapeutic-use section and research recommendations.

I. Disease-modification claims and absences

End-of-era statement (brief §3 I-4) — draft, for the interim synthesis

Held for primary-document checks before v0.2

Part 2 — Modern Clinical Development, 2000–2017 · ledger v1.0 — SIGNED OFF by Danny 2026-10-10 · P2 LOCK 2026-10-14

A. Shape of the literature

B. Science that redirected the field (non-human unless stated — mechanism/plausibility only, §M3)

MED01-C045 — In 2015, cannabidiol was reported to act as a negative allosteric modulator of the CB1 receptor in cell systems (Laprairie et al.), one of several mechanisms proposed in the period for CBD’s distinct pharmacology.

C. What was studied in humans (findings of the period, by condition; not present-day verdicts)

MED01-C057 — In Crohn’s disease, a 21-patient trial of smoked cannabis did not meet its primary outcome (remission 5 of 11 vs 1 of 10, P=.43) while reporting clinical response in 10 of 11 vs 4 of 10 (Naftali 2013), and a later low-dose CBD trial found no benefit (Naftali 2017).

MED01-C060 — The one large human disease-modification trial identified in the window, CUPID (Zajicek 2013: 498 patients with progressive MS randomized to oral THC or placebo for three years), found no effect on progression (HR 0.92, 95% CI 0.68–1.23; p=0.57).

D. Products and regulatory decisions (approval ≠ efficacy evidence, §M8)

E. Study quality and reporting

F. Conditions of research

G. Harms research

H. What the syntheses said (attributed and dated — never this programme’s finding)

I. Opioids, absences and public claims

MED01-C085 — Public-claim rulings (probe labels, not §M9; each describes what period evidence supported by end-2017):

End-of-era statement, P2 (brief §3 I-4) — draft, describes 2017 not today

Held for full-text or primary-document checks — unresolved at P2 ledger v1.0; carried forward to the interim P2 synthesis and §M10a cross-audit

Claim Ledger — MED01: Historical Evidence Landscape, medical cannabis & cannabinoid research 1960s → 2026

A. Shape of the literature

MED01-C001 — Two frozen PubMed/MEDLINE queries returned 8,663 and 10,630 indexed cannabis/cannabinoid records for 1960–1999; despite search-string dependence, both show a 1970s surge, a 1980s trough, and a 1990s peak exceeding the 1970s.

B. Discoveries that redirected the field (all non-human evidence — mechanism/plausibility only, §M3). Design tag for pure chemistry (isolation, structure, synthesis) = mechanistic (chemical/structural; no biological outcome), not in-vitro.

MED01-C004 — The structure of cannabidiol was published by Mechoulam and Shvo in 1963.

MED01-C007 — A saturable, stereoselective, high-affinity cannabinoid binding site (Kd ≈ 133 pM for [³H]CP-55,940) coupled to G-protein was characterised in rat brain membranes by Devane, Dysarz, Johnson, Melvin and Howlett in 1988.

MED01-C008 — A cannabinoid receptor cDNA was cloned and functionally expressed by Matsuda et al. (NIMH) in 1990, and receptor distribution was mapped autoradiographically across mammalian brains including human tissue by Herkenham et al. (NIMH) the same year.

MED01-C009 — An endogenous brain constituent binding the cannabinoid receptor, arachidonylethanolamide (“anandamide”), was isolated from porcine brain and reported by Devane, Hanuš, Mechoulam and colleagues in 1992.

MED01-C010 — A second cannabinoid receptor, expressed predominantly in peripheral immune tissue and not detected in brain, was cloned by Munro, Thomas and Abu-Shaar (MRC Cambridge) in 1993.

MED01-C011 — 2-Arachidonoylglycerol was identified as a second endogenous cannabinoid-receptor ligand in 1995 by two independent groups (Mechoulam et al., canine gut; Sugiura et al., rat brain).

C. What was studied in humans (clinical findings of the period; not present-day verdicts)

MED01-C012 — In a 1975 randomized, double-blind, placebo-controlled study, oral Δ9-THC produced an antiemetic response in 14 of 20 THC courses versus 0 of 22 placebo courses in patients receiving emetogenic chemotherapy (22 entered, 20 evaluable; completers 12/15 vs 0/14, P < 0.001).

MED01-C020 — In a 1981 double-blind study, oral THC 10 mg reduced clinically measured spasticity versus placebo in a principal group of nine patients (P < 0.01); the blinded examiner correctly identified THC trials in seven of nine cases, indicating compromised blinding.

D. Preparations and pharmaceutical development

F. Conditions of research

G. Harms research

H. What official syntheses concluded (each attributed and dated — never this programme’s finding)

MED01-C034 — The 1997 WHO report Cannabis: A Health Perspective and Research Agenda (WHO/MSA/PSA/97.4) reviewed health effects across organ systems and included a therapeutic-use section and research recommendations.

I. Disease-modification claims and absences

End-of-era statement (brief §3 I-4) — draft, for the interim synthesis

Held for primary-document checks before v0.2

Part 2 — Modern Clinical Development, 2000–2017 · ledger v1.0 — SIGNED OFF by Danny 2026-10-10 · P2 LOCK 2026-10-14

A. Shape of the literature

B. Science that redirected the field (non-human unless stated — mechanism/plausibility only, §M3)

MED01-C045 — In 2015, cannabidiol was reported to act as a negative allosteric modulator of the CB1 receptor in cell systems (Laprairie et al.), one of several mechanisms proposed in the period for CBD’s distinct pharmacology.

C. What was studied in humans (findings of the period, by condition; not present-day verdicts)

MED01-C057 — In Crohn’s disease, a 21-patient trial of smoked cannabis did not meet its primary outcome (remission 5 of 11 vs 1 of 10, P=.43) while reporting clinical response in 10 of 11 vs 4 of 10 (Naftali 2013), and a later low-dose CBD trial found no benefit (Naftali 2017).

MED01-C060 — The one large human disease-modification trial identified in the window, CUPID (Zajicek 2013: 498 patients with progressive MS randomized to oral THC or placebo for three years), found no effect on progression (HR 0.92, 95% CI 0.68–1.23; p=0.57).

D. Products and regulatory decisions (approval ≠ efficacy evidence, §M8)

E. Study quality and reporting

F. Conditions of research

G. Harms research

H. What the syntheses said (attributed and dated — never this programme’s finding)

I. Opioids, absences and public claims

MED01-C085 — Public-claim rulings (probe labels, not §M9; each describes what period evidence supported by end-2017):

End-of-era statement, P2 (brief §3 I-4) — draft, describes 2017 not today

Held for full-text or primary-document checks — unresolved at P2 ledger v1.0; carried forward to the interim P2 synthesis and §M10a cross-audit

B. Discoveries that redirected the field (all non-human evidence — mechanism/plausibility only, §M3). Design tag for pure chemistry (isolation, structure, synthesis) = `mechanistic (chemical/structural; no biological outcome)`, not `in-vitro`.