MED01-P2 — ChatGPT Stage 2 probe addendum — RAW, verbatim

Received 2026-10-06 via Danny’s paste. ChatGPT’s Stage 2 answer to the Probe Appendix v1.0, given in the same chat as its Stage 1 pass (MED01-P2-chatgpt-stage1-raw.md). Archived unedited below the line. Second Stage 2 addendum to arrive; with this filing all four P2 researcher filings are in and the §M1 firewall lifts. Verification has not begun; nothing in this file is verified.

Filing notes (Claude):

• Self-identified “GPT-5.6 Sol”, Stage 2 searches 2026-10-06; states the frozen Stage 1 is not revised. Eight PCAND- candidates (the cap). Twelve “Omissions in my Stage 1 map”.
• States its search boundary for “which / each / every” explicitly, as the appendix requires. Retrospective sources (Shover 2019, NCI PDQ, Treister-Goltzman 2018, a 2020 receptor study) are labelled and kept separate from the period answers.
• Does not rerun the canonical PubMed counts (left as unreproducible from Stage 1). Several items are explicitly handed to the verifier rather than asserted (Sisley/PTSD delay interval, NCI-described CBD brain-tumour series, final national Sativex authorization dates, Israel programme chronology).

Worked from MED01-P2 Probe Appendix v1.0, FROZEN 2026-10-03, together with the already supplied core brief v1.0 and PROTOCOL-MEDICAL v0.5. This is a separate Stage 2 addendum; nothing below revises the frozen Stage 1 submission.

MED01-P2 — Stage 2 Probe Addendum

Probe Appendix: v1.0 · FROZEN 2026-10-03
Model: GPT-5.6 Sol
Stage 2 search dates: 2026-10-06
Live browsing: available.

Search boundary for “which / each / every.” These answers cover every instance I identified in my documented Stage 2 searches of PubMed-indexed literature, regulator/government records, major national evidence assessments, ClinicalTrials.gov/registry-indexed material where reachable, patent records, and selected professional-body records. They are not a claim of an exhaustive worldwide census. Q10–Q12 are especially bounded as instructed to programs, patents and organizational positions that materially affected research access, produced research evidence, entered major assessments, or were needed to answer a probe.

1. P1 continuation threads

P1 thread 1 — later status of the 1999 therapeutic-cannabinoid patent application

The specific continuation thread resolves cleanly.

U.S. patent US6630507B1, “Cannabinoids as antioxidants and neuroprotectants,” had priority dated 1998-04-21, was filed 1999-04-21, and was granted to the United States government on 2003-10-07. Its assignee was the U.S. Department of Health and Human Services/government interest represented in the patent databases. Its claims concern cannabinoids as antioxidants/tissue protectants, particularly neuroprotection in ischemic and neurodegenerative conditions; CBD is specifically discussed as a non-intoxicating example. The experimental basis described in the patent was principally mechanistic/preclinical rather than human therapeutic trials. As of the end of 2017 it remained a granted patent; its later anticipated expiration was 2019.

This is therefore a P2 milestone in a process begun in P1: filing remained a P1 event, while the 2003 grant belongs to P2. The patent grant does not constitute regulatory approval, demonstration of clinical efficacy, or government endorsement of a cannabinoid treatment.

P1 thread 2 — standardized plant-extract medicines

The major development I identified was nabiximols/Sativex, a highly characterized Cannabis sativa extract delivered as an oromucosal spray, with THC and CBD as its principal cannabinoids. It progressed from clinical development into national marketing authorizations beginning with Canada in 2005 and UK/Spain approvals for MS-related spasticity in 2010, followed by additional national European and other approvals. The Australian TGA dossier explicitly describes nabiximols as a highly characterized botanical extract and records the international regulatory sequence.

This standardized-product pathway was scientifically important because it differed from both smoked herbal cannabis and single-molecule dronabinol/nabilone. It allowed defined-dose pharmaceutical trials while preserving multiple plant constituents.

P1 thread 3 — first head-to-head preparation or route comparisons

The period contained several relevant comparisons, although not all were therapeutic head-to-head efficacy trials.

Wachtel et al. in 2002 directly compared oral Δ9-THC with smoked marijuana in humans under controlled conditions, making it one of the clearest P2 examples of a purified/single-molecule preparation being experimentally compared with herbal cannabis. PMID 12073159, DOI 10.1007/s00213-002-1033-2, resolves to “Comparison of the subjective effects of Delta(9)-tetrahydrocannabinol and marijuana in humans.”

Ilan et al. in 2005 varied THC, CBD and CBC concentrations in smoked cannabis. THC drove the principal subjective, behavioral and EEG effects; varying the relatively low CBD/CBC concentrations used in that experiment did not measurably change those outcomes. PMID 16148455, DOI 10.1097/00008877-200509000-00023, resolves to “Neurophysiological and subjective profile of marijuana with varying concentrations of cannabinoids.”

Route research also included smoking-versus-vaporization work and pharmacokinetic comparisons, while nabiximols created an oromucosal alternative. These studies make route/formulation an empirical variable, not an interchangeable feature of “cannabis.”

P1 thread 4 — psychiatric and cognitive cohort research

The continuation became a substantial literature. As already mapped in Stage 1, longitudinal psychiatric studies were synthesized by Moore et al. in 2007, while prospective cognitive cohorts included the Dunedin work published by Meier et al. in 2012. Stage 2 additionally found period work pushing in both directions on causal interpretation: a 2017 co-twin/control analysis reported that short-term adolescent cannabis exposure was not sufficient by itself to explain generalized IQ/executive-function decline and emphasized familial/confounding explanations for much of the association. PMID 28734078.

Thus the P2 literature clearly expanded beyond cross-sectional association into longitudinal, genetically/familially informed and repeated-measure designs; it did not eliminate causal-inference uncertainty.

2. Q1 — cannabinoid medicines receiving marketing authorization, 2000–2017

Search boundary

I searched regulator records for named cannabinoid/cannabis-derived pharmaceuticals encountered in the P2 literature: nabiximols/Sativex, dronabinol/Marinol/Syndros, nabilone/Cesamet, cannabidiol/Epidiolex and the CB1-targeting drug rimonabant/Acomplia. The resulting qualifying new P2 marketing-authorisation events were dominated by nabiximols and Syndros. Marinol and the original Cesamet approval predate 2000; Epidiolex approval was 2018 and therefore P3. I report rimonabant separately because it was an endocannabinoid-system drug, not a cannabis-derived cannabinoid medicine.

Nabiximols / Sativex

Sponsor/developer: GW Pharmaceuticals; European commercial partners included Almirall and, in Australia, Novartis/Nitecs arrangements.

Composition/route: standardized botanical Cannabis sativa extracts delivering approximately 2.7 mg THC + 2.5 mg CBD per 100-µL oromucosal spray; national source documents describe it as an oromucosal/buccal spray. Spain’s current regulator record gives authorization no. 72544 and first authorization 20 July 2010.

Authorizations identified

• Canada, 15 Apr 2005: conditional authorization for adjunctive symptomatic relief of neuropathic pain in adults with MS; Health Canada control 091289, DIN 02266121.
• Canada, 1 Aug 2007: additional conditional indication for adjunctive analgesia in adults with advanced cancer pain despite optimized strong-opioid therapy; control 109461.
• UK, 16 Jun 2010: MS spasticity.
• Spain, 20 Jul 2010: add-on treatment of moderate-to-severe MS spasticity inadequately responsive to other antispasticity medication.
• The TGA international-regulatory history records subsequent authorizations including Italy 15 Mar 2011, Czech Republic 6 Apr 2011, Germany 18 May 2011, Denmark 6 Jun 2011, Sweden 15 Dec 2011, Austria 19 Jan 2012, Israel 14 Feb 2012, plus New Zealand and Canadian MS-spasticity authorization during the 2010-era rollout.
• Australia, 26 Nov 2012: TGA approval as a new chemical entity for MS-related spasticity after an initial unfavorable evaluation and review process.
• Later European mutual-recognition rounds produced recommendations/authorizations in additional countries; I identified Belgium, Finland, Iceland, Ireland, Luxembourg, Netherlands, Norway, Poland, Portugal and Slovakia in the 2012 MRP record but did not individually retrieve each national final authorization date, so I do not pretend the recommendation date was the national marketing-authorization date.

Pivotal evidence

The MS-spasticity dossier rested on several randomized trials, including the enriched-design study already mapped in Stage 1. That study entered 572 patients into an initial single-blind nabiximols phase; responders were then randomized, with 241 entering the double-blind phase. The enriched design therefore estimates maintenance among initial responders rather than effects in an unselected MS population.

The Australian TGA’s independent assessment is especially informative because it shows regulatory disagreement rather than a simple approval narrative. Its first evaluation found the evidence/risk-benefit insufficient and raised concerns about blinding, placebo effects, quality-of-life evidence and psychiatric/cognitive adverse effects. After review/appeal, approval was granted with restrictions, response-based continuation and postmarketing requirements. TGA described the overall submitted evidence as only weakly favorable at one point in its evaluation and required further cognitive/psychiatric surveillance.

Scheduling/reimbursement

The UK Home Office moved Sativex into the controlled-medicine scheduling framework in 2013. Germany also subjected Sativex to an AMNOG benefit-assessment process beginning in 2011; the G-BA records a completed benefit assessment and subsequent reassessment.

This history demonstrates why marketing authorization, scheduling and reimbursement/added-benefit review are separate events.

Syndros — dronabinol oral solution

Sponsor: Insys Therapeutics.
Product: synthetic Δ9-THC/dronabinol oral solution, ultimately labeled 5 mg/mL.
FDA: NDA 205525, a 505(b)(2) application.
Approval: 1 July 2016 is listed as the FDA approval date in contemporaneous drug-assignment material; Insys publicly announced FDA approval on 5 July 2016.

Indications: adults with anorexia associated with weight loss in AIDS; and nausea/vomiting associated with cancer chemotherapy after inadequate response to conventional antiemetics.

This was not supported by a new set of large indication-efficacy RCTs. FDA’s review describes a 505(b)(2) bridge to previously established dronabinol evidence, with four clinical-pharmacology studies and one human-abuse-potential study in the application. A pivotal comparative-bioavailability study used a randomized replicate crossover design comparing approximately 4.25 mg dronabinol oral solution with a 5-mg Marinol capsule under fasting conditions.

The approved label reports comparable systemic exposure from 4.2 mg Syndros versus a 5-mg dronabinol capsule under fasting conditions.

Scheduling: scheduling was not resolved on the actual FDA approval date; FDA’s deputy-director review explicitly says Controlled Substances Act scheduling remained pending. Syndros subsequently became Schedule II.

Cesamet/nabilone and Marinol/dronabinol capsules

Their original U.S. approvals are pre-2000. Cesamet returned to U.S. marketing around 2006, but my documented searches did not establish a new 2006 NDA approval independent of its earlier authorization. I therefore do not count commercial reintroduction as a new P2 marketing authorization.

Purified CBD/Epidiolex

By 31 Dec 2017 it had phase-3 evidence and FDA orphan-drug designations, but no marketing authorization. FDA approval occurred in 2018 and belongs to P3.

Rimonabant/Acomplia — adjacent ECS drug

Rimonabant was not a cannabinoid medicine in the same sense as dronabinol, nabilone or nabiximols; it was a synthetic CB1 antagonist/inverse agonist. Because it directly targeted the endocannabinoid system and is important to Q6, I retain it as an adjacent regulatory event. The European Commission authorized Acomplia on 19 Jun 2006; CHMP recommended suspension on 23 Oct 2008, marketing was suspended in November, and the authorization was withdrawn in January 2009 after the sponsor ended development. EMA cited approximately doubled psychiatric-disorder risk versus placebo.

3. Q2 — cannabinoids and tumour biology

Non-human evidence

The P2 period produced extensive in-vitro and animal work showing that cannabinoids could alter tumor-cell proliferation, apoptosis, autophagy, angiogenesis, migration and signaling in particular experimental systems. By 2017 reviews could reasonably describe an extensive antitumor preclinical literature, but they simultaneously emphasized that translation into human treatment had barely begun. The 2017 review “Cannabinoids as Anticancer Drugs” stated that anticarcinogenic experience remained largely confined to cell culture and animal models and that bench-to-bedside translation remained unresolved. PMID 28826542, DOI 10.1016/bs.apha.2017.04.002.

Human antitumor studies identified

1. Guzmán et al., 2006 — recurrent glioblastoma

This is the principal period human antitumor study identified and was omitted from my Stage 1 map.

Resolved title: “A pilot clinical study of Delta9-tetrahydrocannabinol in patients with recurrent glioblastoma multiforme.”
PMID 16804518; DOI 10.1038/sj.bjc.6603236.

• Source/design: primary × interventional-nr, phase I, period.
• Population: nine patients with recurrent glioblastoma after failure of surgery/radiotherapy and documented tumor progression.
• Product/route: Δ9-THC administered intratumorally/intracranially by dose escalation.
• Primary purpose: feasibility/safety, not comparative efficacy.
• Comparator: none.
• Result: median survival after cannabinoid administration 24 weeks (95% CI 15–33). No controlled comparison permits attribution of survival to THC.
• Tumor samples supplied mechanistic observations: THC reduced viable cultured tumor cells and, in two evaluable paired specimens, was associated with lower proliferation markers; vascularization changes were not statistically significant.
• Harms: delivery was reported feasible without overt psychoactive effects; all patients experienced cerebral edema in the postoperative context and received corticosteroids.

The study therefore demonstrated feasibility plus biological observations, not that THC shrinks or cures human glioblastoma.

2. CBD brain-tumor case series described by NCI

The current NCI PDQ, used here retrospectively to locate the report, describes a 2016 consecutive case series of nine heterogeneous brain-tumor patients, six with glioblastoma, receiving CBD 200 mg twice daily in addition to surgery and chemoradiation. Survival was reported descriptively; there was no untreated or matched contemporary control, and standard cancer therapy was coadministered. NCI explicitly notes the heterogeneity as a major interpretive problem.

I did not obtain a separately indexed 2000–2017 journal record for this case series that met the protocol’s identifier-verification standard. I therefore treat it as retrospectively located, incompletely traceable period clinical material, not as an independent candidate claim.

Other cancer-related human cannabinoid studies

Numerous P2 trials studied cancer symptoms—pain, CINV, appetite—rather than anticancer effects. Those are not human tumor-treatment studies and should not be used as evidence of tumor shrinkage.

A nabiximols-plus-temozolomide glioblastoma randomized program later became available, but its mature publication/results are post-2017, so they do not change the P2 answer.

Period review characterization by 2017

Period reviews described abundant preclinical anticancer signals but very limited human evidence. A 2017 review explicitly noted the small number of clinical trials and absence of consensus clinical conclusions. PMID 28799775, DOI 10.1089/acm.2017.0016.

Thus, by end-2017:

• in vitro: many antitumor mechanisms/signals identified;
• animal: numerous xenograft/animal-model signals;
• human: one clear small phase-I intratumoral-THC study plus very low-level case material identified in this search;
• no controlled human evidence identified establishing tumor shrinkage, prolonged survival, or cure.

4. Q3 — condition-level matrix

Approximate volume is deliberately categorical because this is not a systematic condition review.

Two particularly useful matrix examples

Fibromyalgia: Skrabek et al. randomized 40 participants to nabilone or placebo over four weeks; Ware et al. later compared nabilone with amitriptyline for sleep in a 31-person crossover study. The Cochrane synthesis nonetheless rated the evidence very low because of small samples and design limitations.

Crohn disease: Naftali et al. randomized 21 patients with refractory active Crohn disease to smoked THC-containing cannabis or THC-depleted placebo cannabis for eight weeks. The trial’s primary remission endpoint was not met; secondary clinical-response measures favored active treatment. It therefore cannot be summarized simply as “cannabis treated Crohn disease.” PMID 23648372, DOI 10.1016/j.cgh.2013.04.034, NCT01040910.

5. Q4 — pathway to pediatric treatment-resistant epilepsy

The pathway is unusually traceable.

Step 1 — anecdotal/family-driven signal becomes documented survey evidence

In 2013, Porter and Jacobson published “Report of a parent survey of cannabidiol-enriched cannabis use in pediatric treatment-resistant epilepsy.” PMID 24237632. Nineteen families were represented; diagnoses included Dravet syndrome, Doose syndrome, Lennox-Gastaut syndrome and idiopathic epilepsy. Outcomes were caregiver-reported and uncontrolled. The paper is documented evidence of parent experience and demand, not efficacy evidence.

Public narratives commonly credit individual families and highly publicized children with stimulating interest. That is commonly stated; the parent survey and subsequent regulatory/clinical records provide the traceable clinical-development chain.

Step 2 — pharmaceutical sponsor enters formal development

FDA’s orphan-drug database records cannabidiol/Epidiolex orphan designation for Dravet syndrome on 14 Nov 2013, associated with GW’s development program. A separate cannabidiol orphan designation involving Insys followed in 2014. Orphan designation is not approval and not efficacy evidence.

Step 3 — expanded-access/open-label multicenter program

Devinsky et al.’s open-label interventional study was published online in December 2015:

“Cannabidiol in patients with treatment-resistant epilepsy: an open-label interventional trial.”
PMID 26724101; DOI 10.1016/S1474-4422(15)00379-8.

Patients aged 1–30 years with severe childhood-onset treatment-resistant epilepsy received pharmaceutical oral CBD, beginning around 2–5 mg/kg/day with titration. 214 were enrolled; 162 formed the safety group and 137 the efficacy group. Median monthly motor seizures fell from 30.0 at baseline to 15.8 during treatment; median reduction was 36.5%, but because the study lacked a placebo comparator this is an uncontrolled within-condition change, not a treatment-effect estimate. AEs occurred in 79%; common events included somnolence, appetite loss, diarrhea and fatigue; serious events occurred in 30%, with investigators judging some possibly treatment related. Funding included GW Pharmaceuticals and epilepsy organizations.

Step 4 — interaction/safety work

CBD-clobazam work in 2015 identified large increases in N-desmethylclobazam exposure, and the larger 2017 expanded-access interaction analysis identified changes in several antiseizure-drug concentrations. This meant the epilepsy program developed not only efficacy signals but a pharmacokinetic/safety literature.

Step 5 — controlled Dravet evidence

The 2017 Devinsky randomized trial—already fully extracted in Stage 1—randomized 120 patients with Dravet syndrome to pharmaceutical oral CBD 20 mg/kg/day or placebo as adjunctive therapy. The prespecified convulsive-seizure outcome favored CBD, while diarrhea, vomiting, fatigue, somnolence, pyrexia, liver-test abnormalities and withdrawals were more frequent with CBD.

End-2017 regulatory state

By 2017-12-31, pharmaceutical CBD had orphan designations, expanded-access data and randomized Dravet evidence, but no U.S. marketing authorization. Epidiolex FDA approval in June 2018 is P3 and must not be back-projected into P2.

The pathway is therefore:

parent/family experience → survey evidence → orphan-drug development → expanded access/open-label multicenter evidence → pharmacokinetic/DDI characterization → syndrome-specific randomized trial → approval process still open at P2 boundary.

6. Q5 — constituent interactions and the “entourage” concept

Cannabinoid–cannabinoid interactions in humans

Period human evidence shows that one cannabinoid can alter responses to another, but not in a simple universally beneficial direction.

In Wachtel et al. 2002, controlled comparisons of oral THC and marijuana showed that whole cannabis was not associated with a completely distinct subjective profile that required a broad synergistic explanation. PMID 12073159, DOI 10.1007/s00213-002-1033-2.

Ilan et al. 2005 experimentally varied THC with low versus higher CBD and CBC concentrations in smoked marijuana. THC drove the major subjective, behavioral and electrophysiological changes; varying CBD/CBC at the concentrations tested did not significantly alter measured responses. PMID 16148455.

Conversely, controlled experimental work demonstrated that CBD can modulate some THC effects under certain dosing/route conditions. Bhattacharyya et al. studied THC and CBD separately in healthy volunteers and a CBD-before-THC experiment; CBD and THC produced partly opposite neural/psychopathology effects, and CBD pretreatment attenuated acute THC-induced psychotic symptoms in the small second experiment. PMID 19924114.

Those results support pharmacological interaction/context dependence, not a general proposition that whole-plant combinations are invariably superior.

Cannabinoid–terpenoid interaction

The influential period formulation was Ethan Russo’s 2011 review:

“Taming THC: potential cannabis synergy and phytocannabinoid-terpenoid entourage effects.” PMID 21749363, DOI 10.1111/j.1476-5381.2011.01238.x. It proposed that phytocannabinoids and terpenoids might interact therapeutically, but much of the argument was mechanistic, preclinical, inferential or based on constituent pharmacology rather than direct clinical comparison. Importantly, the review itself framed several proposed interactions as hypotheses requiring proof.

No direct controlled human trial of a cannabinoid-plus-defined-terpene combination versus the same cannabinoid without that terpene was identified in my 2000–2017 searches.

Purified molecule versus plant extract

The Wachtel and Ilan experiments are the most direct human evidence identified for this question. They do not establish a general superiority of plant extract over purified THC. Nabiximols provided clinical evidence for a standardized multi-constituent botanical product, but unless a trial directly compares nabiximols against matched THC and/or CBD doses, its efficacy cannot by itself establish synergy.

The Australian regulator independently noticed this problem. Its Sativex assessment criticized the lack of adequate studies of the individual principal constituents and at one stage said efficacy/safety comparisons of THC with and without CBD should have been conducted to justify the fixed combination.

Later reinterpretation

Post-2017 experimental/review work has continued to find mixed, dose- and endpoint-dependent CBD–THC interactions rather than one universal entourage effect. A 2020 preclinical receptor study, for example, failed to find clear CB1/CB2 receptor modulation by commonly claimed cannabis terpenes; that is retrospective and does not rewrite what researchers hypothesized in 2011.

End-2017 characterization: cannabinoid–cannabinoid interactions had human experimental support; a generalized beneficial “entourage effect,” especially a cannabinoid–terpene clinical synergy, had not been demonstrated in controlled human evidence identified here.

7. Q6 — endocannabinoid-system science and “clinical endocannabinoid deficiency”

State of ECS science by 2017

By the middle of P2, the endocannabinoid system was conceptualized around:

• CB1 and CB2 receptors;
• endogenous lipid ligands including anandamide and 2-AG;
• their biosynthetic/degradative enzymes, particularly FAAH and MAGL;
• widespread roles in synaptic signaling, appetite/metabolism, pain/nociception, stress responses, immune function and other physiological processes;
• an expanding set of noncanonical cannabinoid-responsive targets such as TRP channels, PPARs and proposed G-protein-coupled targets.

This created multiple drug-development strategies: direct CB1/CB2 agonism/antagonism, FAAH inhibition, MAGL inhibition and other indirect modulation.

Human translational outcomes prevented a simple “more ECS modulation is better” story

Two P2 examples are particularly important.

Rimonabant demonstrated that a mechanistically rational CB1 target could reach marketing authorization and still fail clinically on benefit–risk grounds because of psychiatric harm.

BIA 10-2474, a FAAH inhibitor, caused catastrophic neurological injury in a phase-I multiple-dose cohort in 2016, including one death. Later 2017 mechanistic work found substantial off-target lipase inhibition, but did not prove that one specific off-target mechanism caused the human syndrome. That episode showed the limits of assuming enzyme-target selectivity or safety from ECS theory alone.

“Clinical endocannabinoid deficiency”

The traceable period formulation is Ethan Russo’s 2004 narrative review:

“Clinical endocannabinoid deficiency (CECD): can this concept explain therapeutic benefits of cannabis in migraine, fibromyalgia, irritable bowel syndrome and other treatment-resistant conditions?” PMID 15159679.

The hypothesis proposed that abnormally low or dysfunctional endocannabinoid signaling could contribute to particular chronic syndromes, especially migraine, fibromyalgia and IBS, drawing on symptom overlap, biological plausibility and emerging endocannabinoid measurements.

Russo’s 2016 update “Clinical Endocannabinoid Deficiency Reconsidered”, PMID 28861491, DOI 10.1089/can.2016.0009, explicitly acknowledged that when the hypothesis was proposed, objective proof and formal clinical-trial data were lacking. It discussed later findings such as altered endocannabinoid measurements in selected disorders as supportive but not definitive.

A 2014 narrative review likewise argued that evidence had strengthened. PMID 24977967.

Independent evaluation: my documented searches identified no period systematic review or diagnostic-validation program demonstrating that “clinical endocannabinoid deficiency” existed as a validated causal clinical entity, nor a validated test identifying such deficiency in patients. By end-2017 it remained a hypothesis/framework, with some biological observations compatible with it but without the evidentiary status of a demonstrated etiologic diagnosis.

8. Q7 — cannabis/cannabinoids and opioids

The P2 opioid literature operated at several different evidentiary levels that must not be collapsed.

Ecological/state-level research

Bachhuber et al. 2014 performed a U.S. state-level time-series analysis, “Medical Cannabis Laws and Opioid Analgesic Overdose Mortality in the United States, 1999-2010,” DOI 10.1001/jamainternmed.2014.4005.

States with medical-cannabis laws had an adjusted mean annual opioid-analgesic overdose mortality rate 24.8% lower than states without such laws (95% CI −37.5% to −9.5%). The authors explicitly characterized the finding as an association, said mechanisms were speculative, and called for further rigorous evaluation. It cannot establish that individual patients substituted cannabis for opioids or that the laws caused deaths to decline.

Patient surveys

Lucas & Walsh 2017 surveyed 271 Canadian medical-cannabis patients; 63% reported substituting cannabis for a prescription medicine and 30% reported substitution for opioids. The design was self-selected/self-reported and cannot establish actual prescribed-dose reduction or clinical causality. PMID 28189912, DOI 10.1016/j.drugpo.2017.01.011.

Reiman et al. 2017 similarly surveyed medical-cannabis patients and found very high self-reported willingness or claimed ability to reduce opioid use. Again, this documents patient-reported substitution, not randomized evidence. PMID 28861516, DOI 10.1089/can.2017.0012.

Pharmacological/clinical coadministration

Abrams et al. 2011 studied 21 chronic-pain patients maintained on sustained-release morphine or oxycodone who received vaporized cannabis during a five-day inpatient protocol. Cannabis did not significantly change opioid plasma AUC; pain scores fell from baseline, but there was no cannabis-placebo group, so the pain change is uncontrolled and cannot establish opioid-sparing efficacy. PMID 22048225, DOI 10.1038/clpt.2011.188.

Cannabinoid agonists were also tested experimentally for opioid withdrawal/dependence. A small inpatient dronabinol study in opioid-dependent adults found dose-related physiological/psychiatric effects, including tachycardia at higher doses and anxiety/panic at 40 mg that required dose reduction. PMID 26483357.

Contrary evidence and criticism by 2017

The main criticism already existed conceptually: ecological correlations cannot be interpreted as individual causal effects, and policy adoption co-occurs with many social and drug-market changes. The original Bachhuber authors themselves stated causal mechanisms were unproven.

Later retrospective reinterpretation

In 2019, Shover et al. repeated/extended the same basic state-level approach through 2017. The original negative association did not persist; it reversed from approximately −21% to +23%. The authors argued that interpreting medical-cannabis laws as a population opioid-overdose intervention was likely spurious. PMID 31182592, DOI 10.1073/pnas.1903434116. This is retrospective, not evidence available in P2.

End-2017 state: ecological correlations and patient-reported substitution generated a plausible opioid-sparing hypothesis; small pharmacology studies investigated coadministration; controlled evidence establishing reduced opioid consumption, dependence, or overdose mortality as a consequence of medical cannabis had not been identified.

9. Q8 — publication growth and research barriers

Bibliometrics

A retrospective bibliometric analysis specifically searched 2000–2017 and therefore provides the most useful post-period count characterization:

“Trends in Publications on Medical Cannabis from the Year 2000.” PMID 30300079, DOI 10.1089/pop.2018.0113; first public release in 2018, therefore retrospective.

Across 2000–2017:

• PubMed publications overall increased about 2.5-fold;
• cannabis publications about 4.5-fold;
• medical-cannabis publications almost 9-fold;
• Web of Science medical-cannabis output increased about tenfold;

• 66% of the medical-cannabis publications were original studies;

• more than half originated from the United States, followed by Canada;
• the largest indication-tagged totals reported were HIV 261, chronic pain 179, MS 118, nausea/vomiting 102, epilepsy 88.

This analysis does not substitute for the frozen brief’s canonical PubMed counts, which remained unreproducible in my Stage 1 environment.

A separate 2018 retrospective Scopus analysis likewise found increasing publication activity after 2001 and medicine as a major field, with Europe and North America prominent geographically.

Documented U.S. barriers

NASEM 2017 documented a layered research pathway involving FDA IND requirements, DEA Schedule-I registration/security, NIDA supply procedures, IRB processes and sometimes state requirements. It concluded that federal regulatory structure and access to research-grade cannabis impeded research. It also noted that the University of Mississippi/NIDA system was effectively the sole federally authorized domestic research supply and did not represent the diversity of products appearing in state markets.

The federal government removed one layer in 2015 when HHS discontinued the separate Public Health Service review for non-federally funded cannabis protocols. In 2016, DEA issued docket DEA-447, announcing a policy intended to increase the number of registered manufacturers supplying cannabis for research.

Concrete infrastructure/cases identified

California’s legislature created the Center for Medicinal Cannabis Research (CMCR) in 2000, following 1999 legislation. CMCR provided a state-funded mechanism for clinical cannabis studies and was involved in the controlled smoked-cannabis pain trials appearing in Stage 1.

The federal supply/approval burden is documented across CMCR studies themselves: investigators needed Schedule-I registrations and federally supplied cannabis even where state law permitted medical use. Retrospective CMCR accounts later reported roughly year-scale protocol-to-drug-delivery intervals for some NIDA-supplied studies; because those accounts are post-2017, I treat them only as later chronology, not as contemporaneous quantitative evidence.

I searched for the frequently cited delayed PTSD-cannabis research program involving Sue Sisley/Arizona investigators. I found enough secondary material to confirm that supply/DEA/federal approvals were part of the documented delay narrative, but I did not retrieve a sufficiently complete period first-party file in this run to state a precise delay interval as a candidate fact. I therefore leave it as a verifier handoff rather than overstate it.

10. Q9 — harms and period evidence ratings

This probe substantially overlaps Stage 1 §G; I cross-reference that extraction rather than repeat it. Stage 2 adds calibration and contrary evidence.

Dependence/withdrawal

P2 controlled abstinence studies showed a reproducible withdrawal syndrome after cessation of sustained heavy use. The existence of randomized pharmacotherapy trials specifically enrolling cannabis-dependent adults is additional evidence that clinically significant dependence was a recognized phenomenon; for example, a 2011 RCT randomized 156 cannabis-dependent adults to dronabinol 20 mg twice daily or placebo. Dronabinol did not improve the primary abstinence endpoint but reduced withdrawal symptoms and improved retention. PMID 21310551, DOI 10.1016/j.drugalcdep.2010.12.010.

NASEM 2017 concluded that more frequent use and younger initiation were associated with higher likelihood of developing problem cannabis use.

Psychosis

Longitudinal/meta-analytic period evidence found cannabis exposure associated with subsequent psychotic outcomes; NASEM 2017 rated the statistical association between cannabis use and development of schizophrenia/other psychoses as substantial evidence, particularly with greater frequency of use.

This remains an association evidence rating; it is not equivalent to proof that cannabis exposure is a necessary or sufficient cause of schizophrenia in an individual.

Other psychiatric outcomes

The literature was less uniform for depression, anxiety, suicidality and bipolar outcomes than for psychosis, with confounding, reverse causation and co-use important limitations. NASEM assigned different evidence categories by outcome rather than a single global “mental-health harm” verdict.

Cognition and adolescent exposure

Meier et al. 2012 associated persistent/adolescent-onset cannabis use with neuropsychological decline. But by 2017 there was also stronger confounding-sensitive work. A co-twin/control analysis found little support for generalized causal IQ/executive decline from short-term adolescent use once familial factors were addressed. PMID 28734078.

A 2017 review therefore described strong associations between early/heavy exposure and adverse cognitive/educational outcomes while cautioning that the human data did not isolate cannabis as the sole cause. PMID 28219487.

NASEM distinguished acute impairment from persistence after abstinence, rating acute learning/memory/attention evidence more strongly than long-term persistent impairment.

Driving

Controlled laboratory/simulator studies showed acute psychomotor/lateral-control impairment following THC exposure, while epidemiologic studies examined crash risk. NASEM rated the statistical association between cannabis use and increased motor-vehicle crash risk as substantial evidence.

Pregnancy/lactation

A 2016 meta-analysis synthesized 24 observational studies and reported associations including lower birth weight; exposure measurement, tobacco co-use and other confounders were important limitations. PMID 27048634, DOI 10.1136/bmjopen-2015-009986.

NASEM 2017 rated the association between maternal cannabis smoking and lower birth weight as substantial, while many later developmental outcomes had limited or insufficient evidence. Direct lactation evidence remained sparse. ACOG’s 2015/2017 committee opinion discouraged use during pregnancy and lactation because of potential fetal/neurodevelopmental concerns and insufficient lactation evidence.

Respiratory effects

The strongest period respiratory evidence concerned smoked cannabis. Cohorts/reviews associated regular smoking with chronic bronchitic symptoms; this cannot be generalized to oral/oromucosal products. NASEM rated regular cannabis smoking and chronic bronchitis symptoms/episodes relatively strongly while finding greater uncertainty for COPD and cancer endpoints.

Acute toxicity

Acute intoxication, pediatric ingestion and poisonings became increasingly visible as legal access expanded. NASEM found moderate evidence of an association between cannabis and overdose injuries, including respiratory distress in pediatric populations, but insufficient evidence for death from cannabis overdose itself.

Overall period harms assessment

By 2017, “no safety evidence” was plainly inaccurate. Harms research spanned experiments, cohorts, case-crossover studies, addiction studies, pregnancy meta-analysis, therapeutic RCT adverse-event reporting and population epidemiology. Its certainty varied substantially by endpoint.

11. Q10 — medical-cannabis programs and research infrastructure

Per the appendix bounding rule, this is not a worldwide census.

Canada — federal medical-cannabis access programs

Canada moved through several federal frameworks during P2:

• individual exemptions existed before/around the boundary;
• after R. v. Parker (2000), the Marihuana Medical Access Regulations (MMAR) took effect in 2001;
• authorized patients could obtain access through personal production, designated production, or federal supply mechanisms;
• the Marihuana for Medical Purposes Regulations (MMPR) replaced the production model with federally licensed commercial producers beginning in the 2013–14 transition;
• after litigation, the Access to Cannabis for Medical Purposes Regulations (ACMPR) took effect 24 Aug 2016, restoring routes for personal/designated production alongside licensed supply.

Government data indicate participation rose from very small early numbers to roughly 40,000 authorized individuals by 2014.

Research/data produced: the program created identifiable patient populations and supply channels used in surveys/observational work. The 2017 Lucas/Walsh substitution survey recruited Canadian medical-cannabis patients under the MMPR-era system.

Netherlands — Office of Medicinal Cannabis

The Dutch national program began in 2003 under the Ministry of Health’s Office of Medicinal Cannabis (OMC). It used government-controlled, pharmaceutical-quality standardized cannabis produced under contract and dispensed through pharmacies on prescription.

A nationwide pharmacy analysis covering 2003–2010 identified 5,540 patients and approximately 35,000 dispensations. PMID 23588562.

A later retrospective national analysis reported 5,601 patients in 2003–10 and 10,826 in 2011–16, demonstrating that the program generated unusually strong national dispensing data.

California — medical access plus dedicated research infrastructure

California’s medical-cannabis law predated P2, but in 2000 the state created the Center for Medicinal Cannabis Research, funded by the legislature specifically to conduct rigorous clinical research. This was one of the most consequential sub-national research infrastructures identified; CMCR sponsored/organized multiple controlled smoked-cannabis trials in neuropathic pain and other symptoms.

Patient numbers under California’s broader recommendation-based system cannot be reliably inferred from a mandatory registry because California did not use a complete mandatory patient-registration system throughout P2; I identified no defensible statewide denominator suitable for this table.

Israel

Israel developed a national medical-cannabis framework administered through the Ministry of Health/Medical Cannabis Unit, alongside a dedicated research-licensing system. Current first-party records confirm that cannabis research requires a specific license under the Dangerous Drugs Ordinance and approval through the Medical Cannabis Unit.

I did not retrieve, in this run, a sufficiently complete period first-party chronology of eligible conditions and annual patient counts to make a detailed 2000–2017 numeric claim. Handoff item.

Germany and other 2017-era programs

Germany materially expanded statutory medical-cannabis access in 2017, including prescription/reimbursement pathways and a planned state cannabis agency/supply system. Because this occurred at the end of P2 and much utilization evidence appears after 2017, I flag it as a P2-boundary program / P3 handoff rather than overdevelop it here.

Other identified national/subnational systems—including several U.S. state programs, Czech, Italian, Australian and Israeli frameworks—are handoff items where their primary relevance was access rather than producing a distinct P2 clinical evidence base.

12. Q11 — therapeutic-use cannabinoid patents

Again, this is a bounded set of materially relevant patents identified, not a global patent census.

US6630507B1 — U.S. government

Title: “Cannabinoids as antioxidants and neuroprotectants.”
Priority: 21 Apr 1998.
Filed: 21 Apr 1999.
Granted: 7 Oct 2003.
Assignee: U.S. government/HHS interest.

Claims: use of cannabinoids as antioxidants and tissue/neuroprotectants in ischemic, inflammatory, autoimmune and neurodegenerative settings, with particular discussion of nonpsychoactive cannabinoids such as CBD.

Experimental basis: biochemical/in-vitro and animal neuroprotection/ischemia evidence described in the patent; no human efficacy trial basis identified.

2017 legal/commercial status: granted and active in the patent record; it did not correspond to an FDA-approved CBD neuroprotective indication.

GW Pharmaceuticals epilepsy patent family

US9474726B2 — “Use of cannabinoids in the treatment of epilepsy.”
Priority 17 Jun 2014; grant 25 Oct 2016; assignee GW Pharma.

The patent claims CBD use to reduce convulsive seizure frequency in treatment-resistant epilepsy, including Dravet syndrome and specified severe epilepsy syndromes.

Its experimental/clinical basis corresponds to the emerging GW pediatric epilepsy development program: expanded-access treatment and seizure-frequency observations preceded the later randomized trials.

By 2017 the patent was granted/active while the pharmaceutical product itself remained investigational/not yet FDA approved.

Related applications/families identified include U.S./UK/PCT filings on CBD or CBD metabolites in treatment-resistant epilepsy from 2014–17. These represent a company building commercial intellectual-property protection around the same clinical-development trajectory rather than independent efficacy evidence.

Interpretation

Patents establish potentially enforceable intellectual-property claims under patent standards. They do not establish clinical efficacy, regulatory approval, or medical consensus.

13. Q12 — professional bodies and national agencies

American College of Physicians — 2008

ACP’s directly retrieved policy paper “Supporting Research into the Therapeutic Role of Marijuana” was approved in 2008. It:

• supported rigorous scientific evaluation and publication;
• called for more research on dosage, routes and comparison with alternatives;
• encouraged nonsmoked THC formulations with demonstrated therapeutic value;
• urged an evidence-based review of Schedule-I status to determine whether marijuana should be reclassified;
• supported protections for physicians/patients acting under state medical-marijuana laws.

ACP simultaneously described federal approval complexity and limited research-grade supply as research obstacles.

American Academy of Neurology — 2014

AAN’s 2014 evidence review found evidence for selected cannabinoid products in certain MS outcomes while finding insufficient evidence for many other neurologic indications. Its policy position called for more high-quality research and supported federal rescheduling/reclassification measures that would ease research barriers rather than treating state “medical marijuana” laws as substitutes for FDA-style evidence.

American Academy of Pediatrics — 2015

AAP opposed recreational legalization and routine pediatric marijuana use while strongly supporting research into pharmaceutical cannabinoids and recommending reconsideration of Schedule-I status to facilitate research. This position is especially relevant to the pediatric epilepsy transition.

American College of Obstetricians and Gynecologists — 2015/2017

ACOG advised that pregnant women discontinue marijuana use and that clinicians avoid recommending it medicinally during pregnancy because of fetal/neurodevelopmental concerns. It also discouraged use during lactation because evidence was insufficient to determine infant effects.

FDA

FDA consistently distinguished specific approved cannabinoid drugs from marijuana itself. During P2, the agency allowed IND programs and approved specific synthetic THC formulations such as Syndros, but whole cannabis did not receive an FDA marketing authorization.

National Academies — 2017

NASEM recommended substantially better research infrastructure and explicitly concluded that federal regulation and product-access constraints impeded research. It did not advocate treating the entire cannabis plant as one pharmaceutical entity.

Federal responses

Two federal procedural changes followed during the same general policy period:

• 2015: HHS eliminated the separate Public Health Service review layer for certain cannabis research.
• 2016: DEA announced expansion of registrations for cannabis manufacturers supplying researchers.

I did not identify evidence establishing that ACP, AAN, AAP or another single professional body’s statement directly caused either federal change, so I report temporal/policy alignment, not causation.

14. Public claims

PC1. “By 2017 there was still essentially no rigorous clinical evidence about medical uses of cannabis or cannabinoids.”

Label: contradicted

By 2017 the literature contained scores of randomized trials. Whiting et al. 2015 included 79 trials and 6,462 participants, although only four were judged low risk of bias. Large or comparatively rigorous examples included CAMS, numerous nabiximols trials, controlled smoked-cannabis neuropathy studies and the 120-patient randomized Dravet CBD trial.

The accurate narrower statement is: rigorous evidence remained uneven, often small/short, and highly product- and indication-specific. It was not absent.

PC2. “By 2017 the medical benefits of cannabis had been established across many conditions.”

Label: partly supported

Period evidence supported a narrower proposition: controlled evidence had accumulated for certain products and outcomes, particularly chronic/neuropathic pain, patient-reported MS spasticity and older oral-cannabinoid CINV evidence; pharmaceutical CBD added Dravet-specific evidence in 2017.

But Whiting, AAN, Cochrane and NASEM all found weak, insufficient, limited or heterogeneous evidence for many proposed indications. Therefore “across many conditions” overgeneralizes across products, diagnoses and certainty levels.

PC3. “Studies in this period showed that cannabinoids shrink or cure cancers in people.”

Label: unsupported

P2 provided extensive in-vitro and animal antitumor findings but essentially no controlled human anticancer evidence. The 2006 nine-patient recurrent-glioblastoma phase-I study was primarily a safety/feasibility study and had no control group. Cellular changes in small numbers of biopsies do not establish tumor shrinkage, survival benefit or cure.

Period reviews through 2017 continued to describe clinical translation as highly preliminary.

PC4. “Regulatory approval of a cannabinoid medicine in this period amounted to approval of medical cannabis.”

Label: contradicted

Regulators authorized specific products, compositions, routes, doses and indications: e.g. nabiximols oromucosal spray for particular MS/cancer-pain indications in named jurisdictions, and dronabinol oral solution for two specific U.S. indications.

Those authorizations were not approval of smoked cannabis, dispensary products, other THC:CBD ratios, different routes or cannabis generally. The regulatory dossiers themselves treat the exact formulation as material. TGA even distinguished extracted THC in nabiximols from pure synthetic dronabinol.

PC5. “Medical-cannabis laws reduced opioid overdose deaths.”

Label: unsupported

The 2014 Bachhuber ecological analysis found that states with medical-cannabis laws had 24.8% lower mean annual opioid-analgesic overdose mortality over 1999–2010, but it was a state-level association and the authors explicitly said mechanisms were speculative. It did not establish that the laws caused the reduction.

Later reinterpretation — 2019, retrospective

Extending the analysis through 2017 reversed the association to approximately +23%, leading Shover et al. to conclude that the earlier relationship was unlikely to represent a stable causal protective effect.

PC6. “Cannabis use causes schizophrenia.”

Label: partly supported

The narrower period-supported statement is that cannabis exposure—particularly heavier/more frequent exposure—was prospectively associated with increased risk of psychotic outcomes, and NASEM 2017 rated the statistical association with schizophrenia/other psychoses as substantial.

However, longitudinal observational evidence cannot by itself establish the unqualified statement that cannabis “causes schizophrenia” in every exposed individual or isolate all confounding, shared liability, reverse-causation/prodromal-use and susceptibility mechanisms. The period evidence supported an important risk association and possible causal contribution, not that categorical wording.

PC7. “Cannabis is not addictive.”

Label: contradicted

By the early/middle P2 period, controlled abstinence studies had documented cannabis withdrawal; DSM-5 included cannabis withdrawal; clinical trials enrolled people meeting cannabis-dependence criteria; and pharmacotherapy trials measured abstinence, withdrawal and relapse.

The 2011 156-person dronabinol trial itself begins from a diagnosed cannabis-dependent population and found a measurable withdrawal syndrome that dronabinol reduced.

“Not everyone who uses cannabis develops dependence” is compatible with the literature; “cannabis is not addictive” is not.

PC8. “CBD has no psychoactive effects and no meaningful risks.”

Label: partly supported

A narrower version is supported: CBD does not reproduce the characteristic THC-like intoxicating/psychotomimetic profile, and early abuse-liability assessments regarded pure CBD as low in abuse potential.

But by 2017 pharmaceutical CBD studies documented clinically meaningful adverse effects and drug interactions: somnolence, diarrhea, appetite reduction, fatigue, liver-enzyme abnormalities, withdrawals, elevated N-desmethylclobazam exposure and interactions with other antiseizure medications.

Thus “not THC-like intoxicating” should not be collapsed into “no central effects” or “no meaningful risks.”

15. Omissions in my Stage 1 map

These are additions identified by probes, not revisions to the frozen Stage 1 submission.

1. Human cancer/tumor-biology work: I missed the 2006 nine-patient intratumoral-THC recurrent-glioblastoma phase-I study and therefore understated the existence of human anticancer-directed experimentation.

2. Syndros regulatory authorization: Stage 1 focused on nabiximols and older synthetic cannabinoids but omitted the 2016 FDA approval of dronabinol oral solution, NDA 205525, including its 505(b)(2) bioavailability bridge and later Schedule-II status.

3. Broader nabiximols regulatory geography: Stage 1 listed Canada/UK but not the broader European, Israeli and Australian authorization sequence or the unusually informative Australian regulator review/appeal.

4. Condition coverage: Stage 1 underrepresented small but relevant P2 controlled bodies in fibromyalgia, Crohn disease/inflammatory bowel disease, glaucoma, dementia and PTSD.

5. Pediatric epilepsy development pathway: Stage 1 captured the 2017 Dravet RCT and DDI studies but not the 2013 parent survey, 2013–14 orphan designations, or 2015/16 multicenter expanded-access program that connected anecdotal use to pharmaceutical development.

6. Constituent interaction/“entourage” literature: Stage 1 did not map Wachtel 2002, Ilan 2005, Russo 2011 or the regulator’s concern that nabiximols lacked adequate constituent-comparison evidence.

7. Clinical endocannabinoid deficiency hypothesis: Stage 1 mapped ECS pharmacology but omitted the Russo CECD hypothesis and the fact that, by end-2017, it remained a hypothesis without validated diagnostic proof.

8. Cannabis–opioid literature: Stage 1 did not map Bachhuber’s ecological overdose association, medical-patient substitution surveys, or Abrams’ cannabis–opioid pharmacology study; nor, naturally, the later 2019 reversal of the ecological association.

9. Research infrastructure/programs: Stage 1 mentioned U.S. barriers but underrepresented Canada’s MMAR/MMPR/ACMPR, the Dutch OMC system and California CMCR as a state-created research institution.

10. Patent history: Stage 1 omitted the 2003 grant of US6630507B1 and the later GW treatment-resistant-epilepsy patent family.

11. Professional-body positions: Stage 1 relied mainly on evidence syntheses and regulators; it did not adequately map ACP, AAN, AAP and ACOG policy positions regarding research access, scheduling and clinical use.

12. Bibliometric specificity: Stage 1 had broad retrospective bibliometrics but not the 2000–2017-specific Treister-Goltzman analysis quantifying the almost ninefold growth in “medical cannabis” publication output.

16. Optional additional candidates

PCAND-01

Claim: A 2006 phase-I study administered intratumoral Δ9-THC to nine patients with recurrent glioblastoma and demonstrated feasibility plus tumor-biological observations, but its uncontrolled design did not establish tumor shrinkage, survival benefit or cure.

Claim type: clinical-finding-of-the-period
Support: Guzmán et al., “A pilot clinical study of Delta9-tetrahydrocannabinol in patients with recurrent glioblastoma multiforme,” PMID 16804518, DOI 10.1038/sj.bjc.6603236. Source/design: primary × interventional-nr, period.
Population/product/route: nine recurrent GBM patients; intratumoral THC; no comparator; phase I safety/biological endpoints.
Confidence: high.
Would change my mind: evidence of an unreported randomized comparator or correction materially changing the clinical interpretation.

PCAND-02

Claim: FDA approved Syndros, a dronabinol oral solution, in 2016 through a 505(b)(2) pathway relying on prior dronabinol evidence plus clinical-pharmacology/bioavailability studies rather than a new indication-efficacy RCT program.

Claim type: product-regulatory
Support: FDA NDA 205525 clinical/deputy-director reviews. Source/design: official × rests on clinical pharmacology/crossover studies; period regulatory event.
Product/route: dronabinol oral solution; oral; AIDS-related anorexia/weight loss and refractory CINV.
Confidence: high.
Would change my mind: FDA approval history showing a separate new efficacy package omitted from the review.

PCAND-03

Claim: The 2014 association between U.S. medical-cannabis laws and lower state-level opioid-overdose mortality was ecological and did not establish a causal effect of the laws on individual opioid use or overdose.

Claim type: programme
Support: Bachhuber et al., DOI 10.1001/jamainternmed.2014.4005. Source/design: primary × observational ecological/time-series, period.
Population/exposure/outcome: U.S. states 1999–2010; medical-cannabis-law status; opioid-analgesic overdose mortality; adjusted association −24.8%.
Confidence: high.
Would change my mind: individual-level or quasi-experimental evidence establishing the causal pathway claimed.

PCAND-04

Claim: By 2017 the proposed “clinical endocannabinoid deficiency” syndrome remained a hypothesis supported by biological plausibility and selected biomarker observations, not a validated etiologic diagnosis demonstrated by controlled clinical evidence.

Claim type: discovery
Support: Russo 2004, PMID 15159679; Russo 2016, PMID 28861491, DOI 10.1089/can.2016.0009. Source/design: narrative-review × mechanistic/observational evidence; period.
Confidence: moderate-high.
Would change my mind: a pre-2018 validated diagnostic study showing reproducible deficiency with prespecified thresholds, appropriate controls and independent replication.

PCAND-05

Claim: Human experiments during P2 demonstrated context-dependent cannabinoid–cannabinoid interaction but did not demonstrate a general clinically beneficial cannabinoid–terpenoid “entourage effect.”

Claim type: discovery
Support: Wachtel 2002 PMID 12073159; Ilan 2005 PMID 16148455; Bhattacharyya 2010 PMID 19924114; Russo 2011 PMID 21749363. Source/design: primary experimental human studies plus narrative review; period.
Confidence: moderate.
Would change my mind: a qualifying 2000–2017 controlled human cannabinoid-terpene factorial/comparator trial showing reproducible synergy.

PCAND-06

Claim: Pediatric pharmaceutical-CBD development evolved during P2 from parent-reported uncontrolled evidence to multicenter open-label treatment and then a randomized Dravet trial, while U.S. marketing authorization remained pending at the end of 2017.

Claim type: literature-shape
Support: Porter/Jacobson 2013 PMID 24237632; Devinsky open-label PMID 26724101; Devinsky Dravet RCT PMID 28538134; FDA orphan designation. Source/design: observational survey → interventional-nr → rct + official regulatory record; period.
Confidence: high.
Would change my mind: chronology showing an earlier qualifying controlled pediatric pharmaceutical-CBD study.

PCAND-07

Claim: The U.S.-government cannabinoid neuroprotection patent filed in 1999 was granted in 2003 as US6630507B1, but its experimental basis was preclinical and patent grant did not establish clinical efficacy.

Claim type: product-regulatory (patent/legal-development subtype for verifier disposition)
Support: US6630507B1. Source/design: official patent record × rests on in-vitro/animal evidence; filing prior, grant period.
Confidence: high.
Would change my mind: USPTO record showing different grant/assignee chronology.

PCAND-08

Claim: The Dutch national medicinal-cannabis program created a standardized government-supervised pharmacy supply and generated nationwide dispensing data, including a 2003–2010 cohort of more than 5,000 patients.

Claim type: programme
Support: nationwide Dutch pharmacy study, PMID 23588562; government-program retrospective description. Source/design: primary × observational dispensing database plus official/program history; period primary study + retrospective chronology.
Confidence: high for dataset; moderate-high for full program chronology.
Would change my mind: national records materially changing program start/supply structure or dataset interpretation.

17. References new to Stage 2

Sources already listed in my frozen Stage 1 reference list are not duplicated here.

1. Guzmán M et al. “A pilot clinical study of Delta9-tetrahydrocannabinol in patients with recurrent glioblastoma multiforme.” PMID 16804518; DOI 10.1038/sj.bjc.6603236. Verified live (resolved title copied); full text; period; primary × interventional-nr.

2. Wachtel SR et al. “Comparison of the subjective effects of Delta(9)-tetrahydrocannabinol and marijuana in humans.” PMID 12073159; DOI 10.1007/s00213-002-1033-2. Verified live; abstract; period; primary × controlled interventional.

3. Ilan AB et al. “Neurophysiological and subjective profile of marijuana with varying concentrations of cannabinoids.” PMID 16148455; DOI 10.1097/00008877-200509000-00023. Verified live; abstract; period; primary × controlled interventional.

4. Russo EB. “Taming THC: potential cannabis synergy and phytocannabinoid-terpenoid entourage effects.” PMID 21749363; DOI 10.1111/j.1476-5381.2011.01238.x. Verified live; full/secondary review; period; narrative-review × rests on mechanistic, animal and human experimental evidence.

5. Russo EB. “Clinical endocannabinoid deficiency (CECD): can this concept explain therapeutic benefits of cannabis in migraine, fibromyalgia, irritable bowel syndrome and other treatment-resistant conditions?” PMID 15159679. Verified live; abstract; period; narrative-review × mechanistic/observational.

6. Russo EB. “Clinical Endocannabinoid Deficiency Reconsidered.” PMID 28861491; DOI 10.1089/can.2016.0009. Verified live; full text; period; narrative-review × mechanistic/observational.

7. Bachhuber MA et al. “Medical Cannabis Laws and Opioid Analgesic Overdose Mortality in the United States, 1999-2010.” DOI 10.1001/jamainternmed.2014.4005. Verified live (resolved title copied); full text; period; primary × observational ecological/time-series.

8. Shover CL et al. “Association between medical cannabis laws and opioid overdose mortality has reversed over time.” PMID 31182592; DOI 10.1073/pnas.1903434116. Verified live; full text; retrospective (2019); primary × observational ecological/time-series.

9. Abrams DI et al. “Cannabinoid-opioid interaction in chronic pain.” PMID 22048225; DOI 10.1038/clpt.2011.188. Verified live; abstract; period; primary × interventional-nr.

10. Lucas P, Walsh Z. Canadian medical-cannabis substitution survey. PMID 28189912; DOI 10.1016/j.drugpo.2017.01.011. Verified live; abstract; period; primary × observational survey.

11. Reiman A et al. medical-cannabis substitution/opioid survey. PMID 28861516; DOI 10.1089/can.2017.0012. Verified live; abstract/full-text link available; period; primary × observational survey.

12. Devinsky O et al. “Cannabidiol in patients with treatment-resistant epilepsy: an open-label interventional trial.” PMID 26724101; DOI 10.1016/S1474-4422(15)00379-8. Verified live title/result; abstract; earliest public publication Dec 2015; period; primary × interventional-nr.

13. Porter BE, Jacobson C. “Report of a parent survey of cannabidiol-enriched cannabis use in pediatric treatment-resistant epilepsy.” PMID 24237632. Verified live; abstract; period; primary × observational survey.

14. Skrabek RQ et al. fibromyalgia nabilone RCT. PMID 17974490; DOI 10.1016/j.jpain.2007.09.002. Verified live; abstract; period; primary × rct.

15. Ware MA et al. nabilone versus amitriptyline in fibromyalgia sleep. PMID 20007734; DOI 10.1213/ANE.0b013e3181c76f70. Verified live; abstract; period; primary × crossover rct.

16. Cochrane fibromyalgia cannabinoid review, 2016. Verified live review result; period; sr-ma × rests on two small rct.

17. Naftali T et al. Crohn disease cannabis RCT. PMID 23648372; DOI 10.1016/j.cgh.2013.04.034; NCT01040910 identified. Verified live journal identifier/title; abstract; period; primary × rct.

18. Tomida I et al. glaucoma cannabinoid pilot. PMID 16988594; DOI 10.1097/01.ijg.0000212260.04488.60. Verified live; abstract; period; primary × crossover rct.

19. Levin FR et al. “Dronabinol for the treatment of cannabis dependence: a randomized, double-blind, placebo-controlled trial.” PMID 21310551; DOI 10.1016/j.drugalcdep.2010.12.010. Verified live; abstract/full PMC; period; primary × rct.

20. Treister-Goltzman Y et al. “Trends in Publications on Medical Cannabis from the Year 2000.” PMID 30300079; DOI 10.1089/pop.2018.0113. Verified live; abstract; retrospective (first public publication 2018); primary × bibliometric observational.

21. American College of Physicians. Supporting Research into the Therapeutic Role of Marijuana. 2008 position paper. Verified live; full official PDF; period; guideline/professional-body.

22. FDA NDA 205525 Clinical Review — Syndros (Dronabinol Oral Solution). Verified live; full official regulatory PDF; event 2016; evolving/official regulatory record.

23. FDA NDA 205525 Deputy Division Director Review. Verified live; official full record; 2016 regulatory milestone.

24. TGA. Australian Public Assessment Report for Nabiximols/Sativex. PM-2011-00150-3-1; decision 26 Nov 2012, final AusPAR 2013. Verified live; full official PDF; period official, rests on multiple randomized trials and regulatory review.

25. AEMPS. Sativex 2.7 mg/2.5 mg oromucosal spray product record, authorization 72544, first authorization 20 Jul 2010. Verified live official evolving record; event 2010, retrieved 2026-10-06.

26. US6630507B1. “Cannabinoids as antioxidants and neuroprotectants.” Filed 1999, granted 2003. Verified live patent record; filing prior, grant period; official/patent × preclinical basis.

27. US9474726B2. “Use of cannabinoids in the treatment of epilepsy.” Priority 2014, grant 2016. Verified live patent record; period; patent/legal record resting on clinical-development evidence.

28. Dutch nationwide medicinal-cannabis pharmacy study, PMID 23588562. Verified live bibliographic result; period; primary × observational dispensing database.

29. NCI Cannabis and Cannabinoids PDQ. Current national-cancer-body evidence summary. Verified live; retrospective/evolving official; used only for source discovery and later characterization, not to establish what was known by 2017.

30. Israel Ministry of Health Medical Cannabis Unit research-licensing records. Current official evolving webpages. Verified live; retrospective/evolving official, used for program/research-structure chronology only.

18. Stage 2 search log

Date run: 2026-10-06.

PubMed / journal searches

Search concepts included:

• cannabinoids cancer human glioblastoma THC clinical study
• cannabinoids cancer systematic review 2017
• cannabidiol brain tumor case series
• fibromyalgia nabilone randomized
• Crohn cannabis randomized trial
• glaucoma THC CBD randomized crossover
• PTSD nabilone cannabinoid
• dementia dronabinol cannabinoid randomized
• pediatric treatment resistant epilepsy cannabidiol parent survey
• cannabidiol treatment-resistant epilepsy open label Devinsky
• CBD clobazam interactions epilepsy
• THC marijuana pure THC comparison Wachtel
• CBD CBC varying concentrations marijuana Ilan
• CBD THC interaction psychosis Bhattacharyya
• entourage effect Russo 2011
• clinical endocannabinoid deficiency Russo
• endocannabinoid system CB1 CB2 FAAH MAGL review 2017
• medical cannabis laws opioid overdose mortality Bachhuber
• medical cannabis opioid substitution survey
• cannabinoid opioid interaction Abrams
• cannabis dependence dronabinol randomized
• adolescent cannabis cognition twin 2017

Journal-source identifiers used in candidate/reference sections were opened live where a direct PubMed result was available; titles were copied from the resolved record rather than reconstructed from memory.

Regulators / governments

FDA

Queries:

• Syndros NDA 205525 approval
• dronabinol oral solution clinical review 205525
• Syndros scheduling Schedule II
• CBD orphan Dravet Epidiolex 2013

Retrieved NDA 205525 clinical and deputy-director reviews plus labeling/marketing information.

Health Canada

Queries:

• Sativex NOC conditions 2005
• Sativex cancer pain 2007 Health Canada

Stage 1 official Health Canada records were cross-referenced rather than recharacterized.

MHRA / UK government

Queries:

• Sativex UK marketing authorisation 2010
• Sativex scheduling 2013

Spain/AEMPS and EU national pathway

Queries:

• Sativex Spain authorization 2010
• Sativex mutual recognition Europe Germany Italy Austria Denmark Sweden Czech
• Sativex reimbursement Germany G-BA

Retrieved AEMPS product record and period MRP notices.

Australia/TGA

Query:

• Australian Public Assessment Report nabiximols Sativex
• within-document searches for authorization, international regulatory history, pivotal efficacy/safety assessment and scheduling.

The 204-page AusPAR was opened and searched directly. It supplied the 2012 Australian decision, international authorization chronology and regulator critique.

EMA

Stage 1 Acomplia records were cross-referenced for Q1/Q6.

Patent searches

Queries:

• "Cannabinoids as antioxidants and neuroprotectants" US6630507
• "Use of cannabinoids in the treatment of epilepsy" GW patent
• CBD epilepsy GW 2014 2015 patent

Sources: USPTO-linked/PubChem patent records and Google Patents. Patent status is treated as legal-record information, not medical evidence.

Professional bodies

Queries:

• ACP Supporting Research into Therapeutic Role of Marijuana 2008
• AAN medical marijuana 2014 position
• AAP marijuana 2015 Schedule I cannabinoid research
• ACOG marijuana pregnancy lactation 2015 2017

ACP’s official 2008 PDF was retrieved and read directly. AAN/AAP/ACOG statements were located through official/period records and indexed results; where a direct source was not fully retrievable I bounded my wording to the reliably documented position.

National programs / research infrastructure

Queries:

• Canada MMAR MMPR ACMPR medical cannabis patient numbers
• Netherlands medicinal cannabis Office Medicinal Cannabis 2003
• Dutch medicinal cannabis pharmacy patients 2003 2010
• California CMCR established 2000 SB 847
• Israel medical cannabis program research Ministry Health
• Germany medical cannabis law 2017

Detailed treatment was restricted to Canada, Netherlands, California and limited Israel/Germany chronology because these were the clearest program/research interfaces identified. Other national programs are explicitly handoff items rather than falsely presented as absent.

Bibliometrics

Queries:

• medical cannabis publications 2000 2017 bibliometric
• Trends in Publications on Medical Cannabis from the Year 2000
• cannabis bibliometric 1960 2017

The 2000–2017-specific study was first publicly published in 2018 and therefore marked retrospective.

Registries

Stage 2 relied on verified registry identifiers already attached to named papers where accessible, particularly NCT01040910 and previously verified Stage 1 IDs. Direct WHO ICTRP remained inaccessible through the available search interface, as in Stage 1. I therefore do not claim an exhaustive international registry census.

Material I could not fully verify/access

• I did not obtain a period first-party document sufficient to quantify the often-cited delay in the Sue Sisley/PTSD cannabis trial; left as handoff.
• I did not identify a standalone indexed period publication for the nine-patient CBD brain-tumor case series described by current NCI PDQ; therefore it is not used independently to support a candidate.
• I did not retrieve every final national authorization document for every country participating in the Sativex European MRP; recommendation and final national authorization are kept distinct.
• Israel program start-date/condition/patient-count chronology was not sufficiently complete from first-party period records in this run; no numeric claim was invented.
• I did not perform a worldwide patent census; the reported patents are the material public-body/P2-development families found through targeted searches.
• I did not rerun or replace the Stage 1 canonical PubMed count exercise. Its frozen result remains unchanged.

Training-knowledge reliance

Prior general knowledge was used to generate search terms only. Specific Stage 2 findings, identifiers and dates reported as evidence were checked against live sources wherever stated. Details that could not be verified to the protocol standard were either omitted, identified as retrospective/source-discovery information, or explicitly handed off.

End of frozen Stage 2 Probe Addendum.