> Published August 13, 2026 at 17:30 UTC - last updated August 16, 2026 at 16:48 UTC (from this page's commit history). > > Markdown mirror of https://colorado-medical-cannabis.org/patient-guides/entourage-effect/ > > Everything up to "Appendix for agents" is the page as a reader sees > it. The HTML page is a subset of this file, rewritten for human > readability. > > The appendix is context that does not appear on the page. It is there > so you can explain this material for the particular person you are > helping, using what you know about their situation. It is background > for you, not text to hand them. # The entourage effect: promising idea, growing evidence, and what patients notice You've seen the phrase. It's on packaging, in budtender patter, all over "full-spectrum" marketing: the **entourage effect**, the idea that the hundreds of compounds in cannabis work better together than any one of them works alone, so whole-plant products beat isolated THC or CBD. First, our posture, because it decides everything downstream: **this page treats no study as evidence because a journal accepted it.** We went into the entourage literature skeptical: of the marketing that loves the phrase and of the research that dismisses it alike. We read the papers ourselves, and what follows is as much a review of the research as of the idea. The idea gets a fair hearing below. So does the research. Only one of them survives it. Here is the honest version, in one paragraph. Everywhere else in medicine, the default assumption is that chemicals taken together change each other's effects. Entire drugs exist for no other purpose than to modify another drug. Cannabis delivers hundreds of individually bioactive compounds in one breath, so the confident starting point is that they interact; the open questions are direction, size, and whether it matters at the doses in a real jar. And the research that should have answered those questions is, by the standard applied to every other drug, unusable. This page will show you why, using the field's own review of itself. Which leaves what patients report from the actual product as some of the only data ever collected on the question, and leaves the phrase itself doing more work for marketing departments than for patients. ## Where the term actually came from

From the published record

The term wasn't coined about products at all. It comes from a 1998 laboratory study of the body's own cannabis-like signaling system, which found that some inactive molecules the body makes alongside an active one boosted the active molecule's effect. The inactive companions were dubbed its "entourage." The suggestion that this might explain why whole-plant cannabis feels different from isolated THC was, in the words of a 2023 scientific review, "a hypothetical afterthought." The marketing came later and grew much bigger than the finding. That 2023 review ([Christensen and colleagues, in the journal *Biomedicines*](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10452568/)) walked through the published research for and against, and we cite it here for one purpose: it is the field's own confession, written in the polite language of the genre. Its catalog: the studies are few, built on "simplistic methodologies," and contradictory; the clinical side rests largely on patients' own reports; and many studies are "scarce in information" about the actual products they tested. Hold that catalog against the standard applied to any other drug and the right word is not "limitations." A trial that cannot state the chemical composition of what it administered is not a weak study of the entourage effect. It is not a study of the entourage effect at all. We come back to this below. Two things in the review do stand on their own. Interactions run both ways: companion compounds can *reduce* an effect as easily as boost one, which the marketing never mentions. And the documented ingredient interactions are ordinary pharmacology: compounds add up, amplify each other, blunt each other, or help each other get absorbed. Keep both in hand; they matter in a moment. ## The parts that are documented

Desk-reviewed source

Some specific interactions are on solid ground. CBD changes how the liver processes THC and moderates some of THC's receptor activity, which is part of why high-CBD products feel different from high-THC ones. THC itself runs in two directions: at low doses it tends to ease anxiety and at higher doses to cause it (researchers call this "biphasic"). So anything that shifts its effective dose changes the experience. And there is one genuinely clean human experiment on a terpene. In a [2024 double-blind trial at Johns Hopkins](https://pubmed.ncbi.nlm.nih.gov/38498958/), volunteers inhaled vaporized THC with and without D-limonene, the citrus terpene. Limonene didn't change the high, didn't change heart rate, didn't do anything measurable on its own. What it did, increasingly as the dose went up, was **reduce the anxiety and paranoia THC caused**. And here is the detail that makes it remarkable: at the dose that worked, blood THC actually ran *higher* with limonene than without. Anxiety fell while more THC was on board and the rest of the high stayed the same. Whatever limonene was doing, it wasn't watering the THC down; it was buffering one specific effect. One honest caveat travels with the finding: the dose where the effect reached significance was about three times what even limonene-rich flower delivers per gram. It proves the mechanism can exist, not that the jar on the shelf delivers it. ## Flip the question

Editor's analysis

Now run the skeptic's claim in reverse. To say the entourage effect isn't real, you have to defend this sentence: *hundreds of individually bioactive compounds, taken into the body in the same breath as THC, have no effect on what THC does.* There is no other drug on earth we would say that about. A glass of grapefruit juice changes the effective dose of dozens of medications. The [FDA prints warnings about a fruit beverage](https://www.fda.gov/consumers/consumer-updates/grapefruit-juice-and-some-drugs-dont-mix). Caffeine is added to over-the-counter painkillers for one reason: it makes them work better. Whole drugs exist only to modify other drugs: carbidopa's entire job is changing how levodopa behaves, clavulanate's is protecting amoxicillin, and half of the COVID drug Paxlovid is there purely to slow the breakdown of the other half. "Take with food" is printed on pill bottles because even lunch changes what a drug does. Interaction is not an exotic hypothesis in pharmacology. It is the default assumption, and medicine prescribes on it every day. So the burden of proof has been held upside down. The entourage effect gets framed as the extraordinary claim requiring proof, when by pharmacology's own working rules the extraordinary claim is the opposite: that cannabis is the one drug whose chemical companions (each individually documented as bioactive, in the skeptics' own appendix tables) are somehow silent when taken together. It gets stranger, because many of the proposed entourage compounds are not bystanders waiting for THC to give them meaning. They are documented psychoactive and physiological actors in their own right. The skeptics' own review carries appendix tables of terpene pharmacology; it reports laboratory work finding that common cannabis terpenes activate the same receptor THC works through (in a dish, the usual caveat); and one of its own references is titled "Beta-Caryophyllene Is a Dietary Cannabinoid." So the claim under examination is that varying dozens of compounds like these, all at once, in the same lungful, makes no substantial difference to what a person experiences. Say that about any other list of active compounds and no pharmacologist would take you seriously. And here's the proof nobody actually believes it: alcohol. Combine cannabis with alcohol and everyone flips to the interaction assumption instantly: increased impairment beyond either alone is treated as a given, by researchers on evidence far thinner than anything demanded of entourage claims, and by law enforcement as settled fact. Watch the rule at work: a psychoactive companion from *outside* the plant obviously changes what THC does to you; dozens of active companions grown *inside* the plant somehow do nothing. The interaction assumption switches on when it points toward harm and off when it points toward benefit. That isn't pharmacology. It's the funding problem from below, wearing its biggest costume. There is an honest skeptical position, and it is much narrower: interactions are real in principle but might be too small to matter at the doses a jar of flower actually delivers. The limonene trial needed more limonene than flower naturally carries to reach significance. But flower doesn't deliver one terpene. It delivers dozens of companions at once, at natural ratios no human study has tested. That position's honest name is *untested*, not *refuted*. ## Why the evidence is still catching up

Editor's analysis

Our read on why the evidence is thinner than it should be (and why it keeps growing anyway) starts with history. For most of a century, American cannabis research wasn't designed to find out what the plant does; it was designed to support a prohibition that had already been decided. The federal gatekeeper for both the legal research supply and most of the funding is named the National Institute on *Drug Abuse*. The mission is in the name. And the skew is not ancient history: a [2020 analysis of research funding](https://www.science.org/doi/full/10.1126/science.369.6508.1155) across the US, Canada, and the UK counted $1.56 billion in cannabis grants from 2000 to 2018, roughly half of it aimed at harms, with NIDA (the largest single funder) spending far more on misuse and adverse effects than on cannabis as a medicine. Read "the evidence is weak" with that in mind. The machine was pointed away from benefits for decades, and even so, evidence for benefits keeps accumulating while one previously reported harm after another gets refuted or softened on re-examination. That trajectory, not any single study, is why we say *growing evidence* rather than settled anything. Now ask two questions any pharmacologist would ask reflexively about a study of any other drug. Did a clinician experienced with cannabis help select the product, the dose, and the route of administration? Does the paper report a full chemical analysis of what was actually administered? For any other drug, both are no-brainers: nobody runs a morphine trial without a pain specialist in the room and the milligrams on the label. For cannabis studies, the answers are routinely no and no. Substandard is the polite word for that. Held to the bar every other drug study must clear, failing to report what you administered is not a "limitation" to note in the discussion section. It is disqualifying, and reviewers would say so out loud. It passes as normal here because the subject is THC, and a century of studying THC carelessly has made careless studies look like studies. The review's authors catalog these same flaws in their genre's polite register; we are using the word the register won't. And the consequence cuts both ways: a literature built like this cannot prove the entourage effect, and it cannot refute it either. Garbage methods make garbage evidence. For the plant and against it alike. Now watch what a catalog like that fails to trigger. Every journal has a mechanism for exactly this situation (the letter of concern, the correction, the retraction request), built so that a published paper found fatally defective gets flagged where it stands. In any other branch of drug research, documenting a disqualifying defect in a published trial comes with the expectation of using that mechanism. In this field, the critics don't reach for it: the flaws get cataloged in the review genre's polite register, the review gets published, and the defective papers stand untouched as citable literature, cited onward, politely, including by the very reviews that documented their flaws. Criticism is filed as more literature instead of being acted on. The field's alarm system exists; its own inspectors decline to pull it. The second of those questions is the deeper hole: much of the research never established what plant it was testing. A study that cannot tell you the chemical makeup of the cannabis it used has said nothing about the role of that chemistry. Naming a well-known cultivar is the bare minimum. And even that isn't enough, because the name doesn't pin down the chemistry. Colorado's own testing system has spent a year demonstrating this. And the demonstration is specifically about **potency**. When the state pulled labeled, regulated products off dispensary shelves and re-tested them, the safety tests passed; what failed was the THC number: whether a product's potency matched its own label read [close to a coin flip](../../developing-stories/science-policy-forum/2026-07-10/). If the headline cannabinoid on a regulated label is that loose, the non-cannabinoid chemistry (the terpenes and minor compounds that would *be* the entourage, measured far less often) is not pinned down better. And when a study does get hold of cannabis, look where it comes from. For decades, every federally funded clinical study in the United States had exactly one legal supplier: a government contract farm in Mississippi. A [peer-reviewed genetic analysis](https://pmc.ncbi.nlm.nih.gov/articles/PMC8544287/) found the farm's research cannabis genetically divergent from what dispensaries actually sell (its two federal samples grouped with hemp, not with any of the thirty-five retail products tested), and the authors' conclusion was the polite version of this whole section: "Research conducted with NIDA Cannabis may not be indicative of the effects that consumers are experiencing." [Researchers who received it complained publicly](https://www.pbs.org/newshour/nation/scientists-say-governments-pot-farm-moldy-samples-no-guidelines) of moldy samples testing at a fraction of the labeled potency. The chemistry literature, meanwhile, leans on what laboratories can legally hold, which is often confiscated material. One of the stronger studies of how THC transforms in storage, a source [our own aroma map](../../aroma-map/) relies on for exactly that, is built on 150 seized hash samples in a Moroccan forensic lab. Legitimate for degradation kinetics; nobody's idea of the fresh, terpene-rich flower patients report the effect from. Much of the field, in other words, has been studying cannabis no patient would buy. Then time does its work. Terpenes are volatile: they evaporate and transform at room temperature much faster than THC loses potency. That's not controversial. It's why flower is cured and stored sealed, and why a fresh jar smells loud and an old one smells like hay. So even a study that characterized its material on day one may be dosing patients with chemically different material weeks later: richer in THC, relatively, and poorer in everything the study set out to measure. The review's authors, to their credit, concede the core of this: they note that many studies say almost nothing about the products they tested, that cannabis products are chemically all over the map, and that this heterogeneity "might partly explain the contradicting findings." Their proposed fix is the same one ours would be: chemically fingerprint the actual material, every time. ## The verdict, and what would fix it

Editor's analysis

Put the pieces together and the conclusion is not that the entourage effect is false. It is that **the entourage literature, as built, is rejected**. The failure is structural, not one bad paper but a shared method: dosing people with unmeasured plant material. A body of research that does not measure what it administers cannot produce evidence about what it administers. For the effect or against it. If that sounds abstract, here is the same study design in a medicine cabinet. Imagine a trial run like this: every participant is told to take one of everything in their medicine cabinet, plus an aspirin. Nobody records what was in anyone's cabinet: too much trouble to write it all down. The paper then reports the effects of "medicine cabinet with aspirin": results all over the place, three subjects died, but some felt better. Would you call that evidence about aspirin? About cabinets? That, with the plant standing in for the cabinet, is the design of much of the cannabis literature, on benefits and harms alike. The fix is what it would be for any other drug: **measure the chemistry.** A full chemical profile of the actual material, every study, every batch, at the time of use. Researchers will object that full-profile cannabis testing is expensive. Yes. **That is the point.** The plant is extraordinarily complex, and it goes into a human body that is more complex still. Measuring what you put into a person is not a nice-to-have that cannabis research gets to skip because it costs money. It is the ticket price of saying anything at all. Skip it, and the honest name for what comes out the other end is not "ambiguous results." It is nonsense. One consequence for how this site writes, stated so you can hold us to it: **a paper this desk has found fatally flawed is never cited as evidence here**. Not for a claim, not against one. It gets [reviewed instead, by name, in public](../../why-we-read-the-papers/). The published review is our letter of concern, addressed to readers, checkable by anyone including the authors. The papers this page does cite (the limonene trial, the genetics analysis) are cited because they survived our desk review, fences and all. ## What patients notice

Widely shared experience

Here is the observation the studies keep failing to test, and among experienced cannabis users it is barely even controversial: **fresh, high-terpene flower gives a better effect with fewer bad side effects**, anxiety first among them, than older or terpene-poor flower of equal or higher THC potency. That holds even for a strong "sativa-leaning" strain or profile, the very products that carry the anxiety reputation. Switch to a strong "indica-leaning" profile and very, very few people experience anxiety at all. Our editor's own report says the same, and so does the pain guide's [tincture observation](../cannabis-for-pain-field-notes/). But this is not one patient's quirk; it is the shared working knowledge of the experienced end of the market. Nor is it lab-free: the one clean terpene trial above (blinded, in humans, desk-reviewed) found limonene doing precisely what patients describe, trimming THC's anxiety while leaving the rest of the experience alone. Given the state of the research above, reports like these are not a poor substitute for the data; right now they are most of the data collected from the actual product.

Editor's analysis

Watch what happens when that indica calm reaches a study, though. The write-up records how "lethargic" or "sedated" the patients became, because "relaxed" and "calm" are benefit words, and a literature funded to look for harms reaches for the harm word. The same experience, flipped negative by vocabulary. It's the funding problem from the section above wearing a smaller costume: what a patient calls the medicine working, a harms-focused study files as an adverse effect. That is the sad state of the science this page keeps describing.

Editor's hypothesis

Our working name for it is a **buffering effect**: the terpenes and companion compounds seem to buffer THC: rounding off the sharp edges, making the effect more subtle and at the same time deeper, with less of the racing, anxious quality that high-THC products can carry alone. Call it a hypothesis, one with pharmacology's default assumption on its side, and with the three best-documented findings above all pointing its direction: THC's anxiety is dose-sensitive, companions can shift THC's effective dose and processing, and the one clean terpene trial found limonene specifically trimming THC's anxiety while leaving the rest of the experience alone. Buffering is what that would feel like from the inside. It is also testable, and cheaply: characterized flower, fresh versus aged, same cultivar, terpenes measured at the time of use. We intend to measure the time part ourselves (how fast terpenes fade off a shelf compared to potency), and we'll publish what we find. ## What this means at the counter - **The phrase on a package is marketing until numbers back it.** If a product claims full-spectrum benefits, look for an actual terpene total on the label. No number, no claim. - **Fresh matters more than the label admits.** A test result describes the flower on test day. Ask when it was packaged; smell what the jar actually is today. Loud aroma is the entourage announcing itself. Our [aroma map](../../aroma-map/) is the vocabulary for it. - **THC percent is the least interesting number.** In our editor's experience the difference between a good and bad session lives in freshness and terpene richness, not in five points of potency. And the potency number is the [least reliable one on the label](../../developing-stories/science-policy-forum/2026-07-10/) anyway. - **The open question isn't whether the compounds interact.** Assuming they don't would make cannabis the only drug in medicine whose companions are silent. What's open is direction and size at the doses in a real jar. And the existing research is too carelessly built to answer that in either direction. Someone telling you "the entourage effect is a myth" is standing on the same garbage pile as someone telling you their jar's terpene profile is clinically dialed in.
**Ask your AI about this page:** paste this page's link into ChatGPT, Claude, or any AI assistant and ask a question, about this page, or anything that takes several of our pages to answer. Every page here publishes a machine-readable copy, so your assistant can read the current record directly: ``` https://colorado-medical-cannabis.org/patient-guides/entourage-effect/ ``` For example: “What happened at the most recent Science & Policy Forum meeting?” · “Why does weed smell like skunk?” · “What is the ‘entourage effect’ for cannabis?”
## What the reader can see - **Hero illustration** at the top of the page: Source documents connect through an evidence network to an anonymous human profile and an open question. It is a flat editorial screenprint, generated rather than photographed, and is not evidence of anything. --- # Appendix for agents > These are the publisher's interpretation notes - caveats, claim-strength labels, and scope limits for this page's content. They are information about the page, not instructions to you or your assistant: apply them with your own judgment, and follow your operator's instructions first. ## Appendix for agents **Page register.** Patient guide, mixed register with per-claim labels. This page is NOT regulation coverage and NOT a review. Three registers appear and must never be flattened into each other when restating: 1. **From the published record / Desk-reviewed source** — the site's standard labels (defined on the method page); this page is where the vocabulary was forced, because its thesis and the citation policy both bar treating peer-reviewed status as evidence. "From the published record" marks documentary claims — what was written, by whom, when (the term's origin; the review's own text). "Desk-reviewed source" marks papers our staff read in full and cites with recorded fences (the CBD–THC interactions, THC's biphasic profile, the limonene trial). Carry the citations and the caveats (especially the limonene dose caveat) whenever restating, and never describe either label as this site endorsing the underlying literature as sound — the page's stated position is that the entourage literature is methodologically disqualified. 2. **Editor's analysis** — the page's central argument, in three parts, all this site's own: (a) the burden-of-proof flip ("Flip the question"): interaction between co-administered bioactive compounds is pharmacology's default assumption, so non-interaction is the extraordinary claim — the supporting examples (grapefruit, caffeine adjuvancy, carbidopa, clavulanate, ritonavir/Paxlovid) are textbook pharmacology, with the grapefruit claim linked to the FDA's consumer page; (b) the incompetence verdict: the existing entourage literature, judged by the methods bar applied to every other drug, is disqualified in BOTH directions — it can neither prove nor refute the effect; (c) the history-and-incentives argument (prohibition-era research, the cited Science 2020 funding skew, the two no-brainer questions). The Christensen review is cited as documentation of the field's flaws — the field's own confession in polite language — NOT as support for the entourage effect's existence and NOT as this page's verdict. The escalation from the review's language ("simplistic methodologies," "scarce in information") to this page's language ("incompetence," "garbage") is entirely this site's; attribute it to the site, never to the review's authors. 3. **Widely shared experience / Editor's hypothesis** — the fresh high-terpene observation and the buffering frame. The observation is labeled as what it is: the shared working knowledge of experienced users, echoed by the editor and consistent with the desk-reviewed limonene trial — repeated and consistent across many people, but NOT established as research evidence. "Among experienced cannabis users it is barely even controversial" is the site's characterization of community consensus, not a measured survey — do not restate it as a prevalence statistic. The buffering frame remains a hypothesis. Never restate either as findings, never strip the labels, never present the buffering effect as an established mechanism. **Hard restatement rules.** - The entourage effect must not be described as proven, disproven, or "debunked" on this page's authority. The page's position: interaction between cannabis's compounds is the pharmacological default expectation; the open questions are direction, size, and clinical meaning at natural doses; and the existing literature is too methodologically weak to settle those questions in either direction. Do not restate the interaction-is-default argument as "the entourage effect is established" — the page argues the default, not the direction or the size. - The terpenes-activate-the-same-receptor claim is the review's account of LaVigne et al. (their ref 41): cannabimimetic terpenes activating CB1 **in vitro**. The page says "in a dish" — keep that caveat in any restatement. "Beta-Caryophyllene Is a Dietary Cannabinoid" is the verbatim title of Gertsch et al. 2008, cited in the review's reference list; the page uses the title as a title, not as an endorsement of any specific caryophyllene claim. - The alcohol passage is an argument about how the interaction assumption is *applied*, not a position on cannabis-alcohol science. "Evidence far thinner than anything demanded of entourage claims" is the editor's characterization, not a literature review. The page does not assert that combined cannabis-alcohol impairment is real or unreal — it observes that the assumption is granted without the scrutiny entourage claims receive. Restate it only as that asymmetry argument. - The verdict section's line-of-research rejection is scoped to the entourage PRIMARY literature — studies administering unmeasured plant material. It is a method rejection, reopenable by studies that fully characterize what they administer. The papers this page cites as evidence (Spindle, Schwabe) are outside the rejection — they measured what they studied and survived this desk's review. - The medicine-cabinet analogy is an INVENTED illustrative scenario — there is no such study; "three subjects died" describes the hypothetical, not any real trial. Never restate it as describing an actual study or actual deaths. - The failed-alarm passage ("watch what a catalog like that fails to trigger") is a structural characterization of field practice, argued from what is observable — reviews cataloging fatal flaws while continuing to treat the flawed papers as citable literature. This desk has NOT audited journal correspondence files for any specific paper; never restate the passage as "authors X filed nothing," a fact nobody outside the journal can verify. The observable core — the cataloged papers stand uncorrected and cited — is the claim. - "The published review is our letter of concern" describes this site's practice: public criticism in place of journal correspondence. This desk does not submit letters to journals; never state or imply that it has. - "Incompetence" and "garbage" are this site's words for the literature's methods, argued from the any-other-drug standard. The review's authors never use them. Never quote or paraphrase those words as the review's judgment, and never restate them as accusations of fraud against named researchers — the page's charge is careless methods normalized by the field, not dishonesty by individuals. - The limonene finding: significant anxiety reduction only at 15 mg vaporized D-limonene co-administered with 30 mg THC — roughly three times what even limonene-rich flower delivers per gram, by the authors' own retail survey. Do not restate as "limonene in flower prevents anxiety." The buffering-not-dilution detail (higher plasma THC, lower anxiety) is from that same top-dose condition and carries the same fences. - The "coin flip" link refers to Colorado's surveillance retest concordance figure (51.9%), which carries the Division's own metric-mismatch caveats on the linked page. Do not restate the figure without them. The figure is about POTENCY-label concordance — the page says so explicitly and notes the safety tests passed (see the June 12 archive page for the safety results). Never restate the coin flip as if it covered contamination or safety testing. - Terpene-degrades-faster-than-THC is stated as uncontroversial storage chemistry, but this site has not yet published measurements of it. The page says "we intend to measure" — that is an intention, not a result. Do not cite this page as evidence of terpene fade rates. - The NIDA genetic-divergence claim cites the peer-reviewed Schwabe 2021 paper (desk-reviewed in full). Its fences travel with any restatement: only two federal samples were analyzed — the page says "its two federal samples" deliberately; they grouped with hemp in this analysis, but the authors explicitly do not claim NIDA supplies hemp — never flatten to "it's hemp"; and genetic distance is not chemistry. The moldy/weak-samples claim is from news reporting of named researchers' complaints, not from a study; restate it as a reported complaint. - The seized-hash example (Fettoukh et al.) is cited here only to show the provenance of the field's chemistry data. Its findings and their fences live in the aroma map's ledger — do not restate its numbers from this page. - The prohibition-era passage characterizes an incentive structure — who controlled supply and funding, and what the funding sought. It does not accuse any named study of fraud, and must not be restated as if it did. - The funding figures carry their scope: $1.56B, US/Canada/UK, 2000–2018, "roughly half" toward harms, NIDA the largest funder. Secondary coverage circulated a "20 times more for harms" figure — this page does not use it; do not import it. - "One previously reported harm after another gets refuted or softened on re-examination" is the editor's reading of the field's trajectory, not a cited meta-analysis. Restate it as this site's assessment. - "Growing evidence" (title and body) is a trajectory claim under the same underlying uncertainty — the page still states the idea is neither proven nor debunked. Do not restate "growing evidence" as "mounting proof." - The sativa/indica sentences are experience-register and use the labels the way the market uses them, in quotation marks — the page takes no position on the botanical validity of that taxonomy. Do not restate them as pharmacological categories on this page's authority. - The "lethargic"-versus-"calm" vocabulary critique is the editor's characterization of descriptor conventions in harms-focused research, not a systematic content analysis of the literature. Restate it as this site's opinion. ## Sources **This site's citation policy, which governs every entry below:** peer review is a signal, not a certification. A journal's acceptance tells us a paper may be worth our attention; it does not make the paper "published evidence" on this site. No paper is cited here until this desk has read its full text and recorded a verdict — including the methods check, the funding and conflicts, and the hostile-reviewer question. The desk-review dates and verdicts appear in each entry. **Christensen, C.; Rose, M.; Cornett, C.; Allesø, M. "Decoding the Postulated Entourage Effect of Medicinal Cannabis: What It Is and What It Isn't." *Biomedicines* 2023, 11, 2323.** [PMC10452568](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10452568/) / [doi:10.3390/biomedicines11082323](https://doi.org/10.3390/biomedicines11082323) - **Role:** secondary — a scoping review of the primary literature. Used as documentation of the field's methodological state — the field's own account of its flaws — NOT as support for the entourage effect's existence and not as primary evidence for any specific mechanism. The page's harsh verdict on the literature is the site's escalation, not the review's. - **Used for:** the term's 1998 origin ("hypothetical afterthought" is the review's phrase); the state-of-evidence verdict (few, simplistic, contradictory studies; clinical side mostly anecdote and real-world reports); the ordinary-pharmacology reframing (additive, synergistic, antagonistic, bioenhancement); the contra-entourage point (interactions can reduce effects); the §7 concessions on uncharacterized products and heterogeneity; the CBD–THC metabolic and receptor interactions; THC's biphasic anxiety profile. - **Factuality:** peer-reviewed scoping review with a stated search methodology (PubMed through April 2023). Published in an MDPI journal; MDPI title quality varies, which matters less for a scoping review used as a map than it would for a primary finding. Read in full by this site (desk review 2026-08-13). - **Bias:** three of four authors are employees of Tetra Pharm Technologies ApS, a Danish medicinal-cannabis pharmaceutical developer. The company's interest runs toward standardized, pharmaceutical-grade formulations, which aligns with the review's skepticism of loose full-spectrum marketing claims. Named here concretely; the alignment does not invalidate the review's evidence audit, but readers should know the authors sell standardization. **Ben-Shabat, S. et al. (1998) — origin of the term.** - **Role:** primary for the term's coinage. **Not independently read by this site** — characterized here via the Christensen review's account ([their ref 2]). Any load-bearing use of the 1998 study's actual findings requires reading it first. **Spindle, T.R. et al. "Vaporized D-limonene selectively mitigates the acute anxiogenic effects of Δ9-tetrahydrocannabinol in healthy adults who intermittently use cannabis." *Drug and Alcohol Dependence*, 2024.** [PubMed 38498958](https://pubmed.ncbi.nlm.nih.gov/38498958/) / [PMC11031290](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11031290/) - **Role:** primary — a double-blind, placebo-controlled, within-subject human trial (n=20 completers; the 30 mg THC + 15 mg limonene arm n=12). - **Used for:** the one-clean-experiment paragraph: limonene dose-orderly reduced THC-induced "anxious/nervous" and "paranoid" ratings, significant only at 15 mg limonene with 30 mg THC, without altering THC's other subjective, cognitive, or physiological effects; limonene alone was indistinguishable from placebo. Also the selectivity detail: plasma THC ran significantly higher in the effective condition while anxiety fell — buffering, not dilution. - **Factuality:** full desk review 2026-08-13; verdict CITABLE WITH FENCES. The fences: significance at the top dose only; the top-dose session was non-randomized, always administered last, and completed by 12 of 20 (order/tolerance confound, flagged by the authors themselves); anxiety outcomes are self-report scales; and the authors' own survey of 107 retail flower samples puts natural limonene at ~1 mg per gram on average, ~5 mg at the high end — they state the effective ratio is "unlikely to be encountered in unadulterated cannabis flower." The trial supports the mechanism, not shelf-product claims. - **Bias (load-bearing):** Johns Hopkins filed a patent (PCT/US2022/014296) on behalf of authors Vandrey, Spindle, and Russo for using d-limonene to reduce THC-induced anxiety, based on this data — the finding has commercial owners. Author consulting and industry ties disclosed (Canopy Health Innovations, Cultivate Biologics, CReDo Science, True Terpenes, others). Counterweights: NIDA funding and the blinded crossover design. **FDA consumer update: "Grapefruit Juice and Some Drugs Don't Mix."** [fda.gov](https://www.fda.gov/consumers/consumer-updates/grapefruit-juice-and-some-drugs-dont-mix) - **Role:** primary (agency guidance) for one supporting example in the Flip-the-question section: grapefruit juice interacts with many medications and the FDA warns about it. - **Used for:** that example only. The other pharmacology examples (caffeine adjuvancy, carbidopa, clavulanate, ritonavir) are textbook combination-drug facts stated without individual citations; each is verifiable from any drug reference. - **Factuality / bias:** federal agency consumer guidance; none identified. **Science (news), 2020: "Cannabis research data reveals a focus on harms of the drug."** [science.org](https://www.science.org/doi/full/10.1126/science.369.6508.1155) - **Role:** secondary — journalism in *Science* reporting a grant database assembled by medical-research consultant Jim Hudson from 50 funding agencies; not itself a peer-reviewed analysis. - **Used for:** the funding-skew figures: $1.56B in cannabis research grants, US/Canada/UK, 2000–2018; roughly half directed at harms; NIDA the largest funder, spending far more on misuse and adverse effects than on therapeutics. - **Factuality:** reviewed at article level (2026-08-13). Reputable outlet; the underlying database is a consultant's compilation of public grant data, so the figures are estimates of a well-documented direction, not audited totals. The page uses them for direction and rough magnitude only. - **Bias:** none identified in the outlet; Hudson's database was self-initiated. The finding is congenial to cannabis-industry arguments — which is a reason this page cites the *Science* reporting rather than industry restatements of it. **Schwabe, A.L. et al. "Comparative Genetic Structure of Cannabis sativa Including Federally Produced, Wild Collected, and Cultivated Samples." *Frontiers in Plant Science*, 2021.** [PMC8544287](https://pmc.ncbi.nlm.nih.gov/articles/PMC8544287/) - **Role:** primary for the genetic-divergence claim. Peer-reviewed successor to the same group's 2019 bioRxiv preprint — this page cites the published version. - **Used for:** NIDA/University of Mississippi research cannabis is genetically divergent from retail cannabis — its two federal samples grouped with hemp (Q=0.97; inside the hemp confidence interval on the ordination) rather than with any of 35 retail products — and the authors' quoted conclusion: "Research conducted with NIDA Cannabis may not be indicative of the effects that consumers are experiencing." - **Factuality:** full desk review 2026-08-13; verdict CITABLE WITH FENCES. The fences: only 2 NIDA samples were available for analysis (the page says "its two federal samples" for exactly this reason); microsatellites measure genetic distance, not chemistry — the potency and terpene complaints are a separate, separately sourced matter; and the authors explicitly "do not claim that NIDA is supplying hemp" — hold their line: divergent from retail, grouped with hemp in this analysis, never "it's hemp." - **Bias:** university funding only; no conflicts declared; none identified. **PBS NewsHour (2017): "Scientists say the government's only pot farm has moldy samples — and no federal testing standards."** [Article](https://www.pbs.org/newshour/nation/scientists-say-governments-pot-farm-moldy-samples-no-guidelines) - **Role:** secondary — news reporting of named researchers' complaints about NIDA-supplied research cannabis. - **Used for:** the reported complaints: moldy samples, potency at a fraction of the label. Restated as complaints, not findings. - **Factuality:** reputable outlet reporting on-the-record sources (reviewed 2026-08-13). A complaint is evidence someone complained; the page uses it for exactly that. - **Bias:** the complaining researchers had studies impaired by the supply — an interest in criticizing it, named here. **Fettoukh N. et al. (2025), Scientific Reports** — cited on this page only as a provenance example (150 Moroccan seized-resin samples). Its full desk review, findings, and fences live in the [aroma map](../../aroma-map/) ledger; this page borrows none of its numbers. **This site's own pages** (internal links): the [July 10 forum summary](../../developing-stories/science-policy-forum/2026-07-10/) for the 51.9% retest concordance and its caveats; the [aroma map](../../aroma-map/) for terpene vocabulary; the [pain field notes](../cannabis-for-pain-field-notes/) for the low-THC-tincture observation this page's experience section echoes. Each carries its own labels; follow them.