OpenSubstance cites primary sources wherever possible. Quantitative fields and brain-mechanism sections are annotated with confidence tiers: measured (direct from peer-reviewed data), derived (calculated from primary sources), estimated (informed approximation), or editorial (team judgment).
Citing a source isn't the same as confirming it says what we claim. Every sourced fact is audited by several independent AI models — Claude, ChatGPT, Gemini, and an open-weight model (Qwen) — which each check the claim against its citation. Their verdicts are combined into a conservative consensus, and any disagreement is flagged for human review.
Harm scores for 14 of our substances come directly from this landmark MCDA. Four more (fentanyl, nitrous, DMT, poppers) now have MCDA scores from subsequent studies (Broman 2025, Ferreira 2022). The remaining substances (tramadol, kratom, DXM, 2C-x, PCP, mescaline, caffeine) are OpenSubstance estimates marked with dashed bars. Note: nicotine is scored as 'tobacco' (26) - includes smoking harms, not nicotine alone.
US expert MCDA scoring 19 drugs. Fentanyl scored highest at 90/100. Also scored DMT/ayahuasca (5), providing the first MCDA data for these substances. Used for our fentanyl and DMT harm scores.
International Delphi MCDA study scoring additional substances not in Nutt 2010, including nitrous oxide (6/100) and poppers (5/100). Used for our nitrous and poppers harm scores.
The base drug combination chart is derived from TripSit v4.0, with ratings adjusted where peer-reviewed data supports a different assessment. Key differences: cocaine+opioids rated 'Dangerous' (CDC SUDORS), stimulant+opioid combos rated 'Dangerous' (unmasking mechanism), MDMA+SSRIs rated 'Decrease' (Liechti 2000). Mechanism explanations cite individual studies.
47% of all drug overdose deaths in 2023 involved both opioids and stimulants. 79% of cocaine-involved deaths co-involved opioids. Source for cocaine+opioid 'Dangerous' rating.
Liechti 2000: citalopram attenuated 30-80% of MDMA's subjective effects - SSRIs block MDMA rather than creating dangerous synergy. Liechti 2006: reviewed GHB+MDMA co-use, finding that stimulants do NOT protect against GHB overdose due to temporal mismatch in duration.
Definitive review establishing that serotonin syndrome risk is highest when drugs affect serotonin through different mechanisms (reuptake inhibition + MAO inhibition + direct release). Used for mechanism explanations across serotonergic combinations.
CDC surveillance of 152 kratom-positive overdose deaths. Only 4.6% had kratom as sole substance. 65% involved fentanyl, 33% heroin, 22% benzodiazepines. Used for kratom interaction ratings.
Cocaethylene (formed from cocaine + alcohol in the liver) carries an 18-25x increase in sudden death risk vs. cocaine alone. Used for cocaine+alcohol mechanism explanation.
Xylazine-involved deaths increased nearly 20-fold from 2015 to 2020. Philadelphia saw xylazine in up to 95% of fentanyl cases at peak. Xylazine and fentanyl suppress breathing through different mechanisms - naloxone alone is insufficient.
Retrospective analysis of 80 online experience reports of lithium + psychedelic use. 47% involved seizures and 39% required emergency medical attention. Used for our lithium warning on all psychedelic substance pages. Lamotrigine did not carry this risk.
FDA required new warnings about risk of respiratory depression with gabapentinoids, especially when combined with opioids. 49 cases reviewed, 12 deaths, 92% involved CNS depressant co-use. Source for gabapentin interaction ratings.
First documented fatal case of ritonavir potentiating MDMA. Post-mortem MDMA blood levels equivalent to 18-tablet overdose from a 2.5-tablet dose. Source for HIV antiretroviral interaction warnings.
FAERS database analysis finding bupropion has the highest odds ratio (2.82) for death when combined with MDMA of any antidepressant examined. Source for bupropion interaction rating.
University of Liverpool's comprehensive drug interaction database for HIV medications, including recreational drug interactions. The gold standard reference for antiretroviral interaction data.
Comprehensive review: CYP3A4 handles ~50% of all drug metabolism. Foundation for understanding why CYP inhibitors (cancer drugs, antifungals, antibiotics) affect so many recreational substances.
Identified 85 drugs with grapefruit interactions via CYP3A4 inhibition. Basis for the 'grapefruit warning = CYP interaction with recreational drugs' heuristic used on the site.
Psilocybin cleared >80% by UGT glucuronidation and MAO-A, not CYP enzymes. Basis for rating psilocybin as low-risk with CYP inhibitors - important for cancer patients considering psychedelic therapy.
Documents the challenge of managing CYP-mediated interactions in patients receiving both HIV and cancer treatment. Source for stacked CYP inhibitor warnings.
Preclinical mouse study establishing an order-of-administration effect for ketamine + cannabis. THC followed by ketamine worsened motor coordination, blunted ketamine's anxiolytic effect, and roughly doubled THC-induced hypothermia; the reverse sequence showed no such interaction. Used for the first-class ketamine + cannabis combo page.
Observational analysis: regular cannabis use during ketamine antidepressant treatment did not significantly affect PHQ-9 response. Used in the mechanism section of the ketamine + cannabis combo page to establish that chronic background cannabis exposure does not blunt ketamine's antidepressant pathway.
Validated scale establishing that ketamine and cannabis independently produce schizophrenia-like acute effects on the same psychotomimetic dimensions (separate, not combined dosing). Used for the psychotomimetic-stacking risk note on the ketamine + cannabis combo page.
Case report suggesting synergistic analgesia at sub-anesthetic ketamine doses with oral cannabis. Tier-3 evidence on the ketamine + cannabis combo page.
Placebo-controlled human trial of cannabis and tobacco, alone and combined - the strongest direct human evidence on the acute pair. Tier-1 evidence on the cannabis + nicotine combo page.
Systematic review of cannabis-nicotine co-use and adolescent/young-adult brain development. Source for the shared CB1/nAChR reward-region framing on the cannabis + nicotine combo page.
Joints produced higher plasma THC and greater subjective effects than tobacco-wrapped blunts. One side of the unsettled question of whether tobacco raises blood THC; contrasts with the co-inhalation finding (Chaychi 2025).
High-resolution PET study finding reduced dopamine transporter binding in reward regions among cannabis and tobacco co-users. Tier-2 evidence on the cannabis + nicotine combo page.
Observational evidence linking co-use of tobacco with cannabis to more cannabis dependence symptoms. Supports the dependence-coupling framing on the cannabis + nicotine combo page.
Preclinical review of endocannabinoid-nicotinic crosstalk. Basis for the mutual-modulation and anandamide/2-AG mechanism statements on the cannabis + nicotine combo page.
Preclinical: brief nicotine exposure induces long-term potentiation of excitatory inputs to reward areas. Basis for the nicotine-priming/head-rush hypothesis on the cannabis + nicotine combo page.
The primary source for our addiction data. Analyzed 43,093 participants from the NESARC study using survival analysis to estimate the probability of transitioning from first use to dependence for nicotine, alcohol, cannabis, and cocaine. Found that 67.5% of nicotine users, 22.7% of alcohol users, 20.9% of cocaine users, and 8.9% of cannabis users develop dependence at some point in their lives.
The original landmark study on addiction capture rates from the National Comorbidity Survey (n=8,098). Established that 32% of tobacco users, 23% of heroin users, 17% of cocaine users, 15% of alcohol users, 9% of cannabis users, and ~4.9% of hallucinogen users develop dependence. Note: 'hallucinogens' includes PCP and possibly MDMA, not limited to classical psychedelics.
Using NHSDA 2000-2001 data (n=114,241), found 2-3% of recent-onset hallucinogen users developed dependence within 24 months under DSM-IV criteria. The best available incidence-window estimate for hallucinogen dependence.
Comprehensive review comparing addiction capture rates across studies, including both Anthony (1994) and SAMHSA (2016) data. Confirmed that the rank ordering of substance dependence rates has remained consistent across decades: nicotine > heroin > cocaine > alcohol > amphetamine > cannabis > psychedelics.
Review confirming LSD produces no physical dependence or withdrawal. Tolerance builds rapidly via 5-HT2A downregulation but dissipates within days without craving. Basis for rating classical psychedelics at 0% dependence.
Dual-tracer PET ([11C]McN5652 and [11C]DASB) finding reduced serotonin transporter (SERT) binding in MDMA users versus controls. Foundational human imaging evidence for MDMA's serotonergic changes; cross-sectional and polydrug-confounded.
[11C]DASB PET plus structural imaging in ecstasy users. Reduced cortical SERT binding correlated with use; also the basis for MDMA's modest, confounded memory/attention findings.
Former (abstinent) MDMA users showed SERT binding comparable to controls, evidence that reductions largely normalize after extended abstinence. Basis for the 'partially reversible' framing.
SPECT study reporting larger serotonergic reductions in women and recovery of binding with sustained abstinence. Basis for the sex-difference and recovery statements on the MDMA long-term entry.
The definitive source for safety ratio data (lethal dose ÷ effective dose). The 2004 Addiction paper covers 20 substances including GHB (ratio 8), DXM (10), poppers (8), and caffeine (100). The 2006 American Scientist article adapted the same data. Cannabis, LSD, and psilocybin exceed 1,000:1.
Found a 23.7-fold increase in MI risk (95% CI: 8.5-66.3) during the 60 minutes after cocaine use. This is a relative risk, not absolute per-use incidence. Used for our cocaine cardiac risk data.
MDMA is a mechanism-based inhibitor of CYP2D6 - it permanently inactivates the enzyme that metabolizes it. This means a second dose hits harder than the first because the body can't clear it as fast. Explains why redosing is disproportionately dangerous.
Analysis of 151 overdose deaths in London. Informed our understanding of polydrug fatality patterns and tolerance-related overdose risk.
Structure-activity relationships and forensic case series for 2-benzylbenzimidazole 'nitazene' opioids. In vitro potency data establishing that some variants are 40-90x more potent than fentanyl.
In vivo SAR study in mice. Etonitazene most potent at ED50 3-12 µg/kg. Alkoxy chain length at one position governs potency across the class.
Synthesis, characterization, and µ-opioid receptor activity of emerging nitazene opioids. All tested variants are full MOR agonists with nanomolar binding affinities.
Adulteration and purity data, especially for MDMA, is informed by DrugsData's 25+ year database of anonymously submitted drug samples analyzed by DEA-licensed laboratories. Their finding that only ~74% of 'MDMA' samples contain only MDMA (as of 2023) is referenced in our supply risk ratings.
The oldest drug checking kit manufacturer in the US. Their #TestIt alerts and on-site drug checking data inform our adulteration warnings. DanceSafe provides reagent test kits and fentanyl test strips.
Fentanyl co-occurrence rates in other drug supplies from forensic laboratory analysis of law enforcement seizures across the US. DEA CY2024 Annual Cocaine Report: levamisole now ~5% (down from 87% in 2017), average purity 88%.
Found fentanyl co-occurs in over 10% of cocaine and methamphetamine samples in several Northeast US states. 2.7% nationally across 11.9 million NFLIS samples.
GC/MS-confirmed concordance rates: LSD 100%, cocaine 92%, amphetamine 91%, heroin 88%, MDMA 84%, ketamine 78%. One of the few studies with full laboratory confirmation of field drug checking results.
Only 22% of 792 samples were concordant with what buyers intended. Only 1.9% of samples intended as heroin tested positive for heroin. 57.3% contained xylazine. Demonstrates how dramatically the opioid supply has shifted.
One of the most transparent drug checking programs, publishing detailed monthly reports with substance-by-substance concordance rates. Tests thousands of samples per year from harm reduction sites across British Columbia using FTIR and DART-MS confirmation.
New Zealand's national drug checking program tested 3,200+ samples in 2024. Found 68% of drugs were what people expected, up from 64% in 2023. Detailed breakdowns by substance type.
Public alerts on emerging novel psychoactive substances including nitazene variants and synthetic cathinones. First to identify N-isopropyl butylone (Aug 2024) and phenazolam in the US supply.
Annual report on novel psychoactive substance detections in US clinical and forensic samples. N-isopropyl butylone accounted for ~45% of synthetic stimulant detections in 2025.
February 2024 advisory on the emergence of nitazene-class opioids. At least 26 unique variants identified across Asia, Europe, North America, Oceania, and South America.
Gladden et al. 2022: eutylone (synthetic cathinone) associated with 343 deaths in 2020, 75% in Florida and Maryland. Documents the real-world mortality impact of cathinone substitution in the MDMA supply.
Comprehensive structure-activity review of cathinone pharmacology. Explains the two subcategories: methylenedioxy variants (MDMA-like, serotonergic) vs pyrrolidine variants (pure DAT blockers, psychosis-prone).
Carfentanil (~100x fentanyl potency) is resurgent in the US supply. Deaths increased ~7x from Jan-Jun 2023 to Jan-Jun 2024. Detected in 37 states, increasingly pressed into counterfeit pills.
Canadian Centre on Substance Use and Addiction: fentanyl analog detection rose from 33% of samples in 2017 to 65% in 2024. Ortho-methylfentanyl surpassing para-fluorofentanyl in some regions.
5F-ADB detected in 80% and MDMB-4en-PINACA in 31% of synthetic cannabinoid death cases (2023-2025). Increasingly found in vapes and edibles.
Detection, CB1 activity (EC50 = 2.47 nM, ~100x THC), and toxicology in 25 forensic death cases for the currently dominant synthetic cannabinoid.
NFLIS-Drug data showing NBOMe reports peaked at 1,968 in 2014, declining to single digits by 2024 after scheduling. Linked to at least 19 US overdose deaths.
Comprehensive review of NBOMe pharmacology, toxicology, and deaths. Narrow safety margin compared to LSD despite similar subjective effects.
Emerging global synthetic opioid threats: benzimidazol-2-ones (orphines). At least six analogs confirmed in drug markets. N-propionitrile chlorphine rising since mid-2024.
Dosage ranges, onset times, duration, and subjective effect profiles. One of the most detailed and regularly updated substance databases.
Detailed pharmacological profiles for 16 substances including dosage ranges, pharmacokinetics, metabolism, and drug interactions.
Effect descriptions, experience reports, and pharmacological reference data. The foundational online drug information resource since 1995.
Prevalence data and national trends in substance use. The most comprehensive annual survey of drug use in the United States.
The world's largest annual drug survey (85,000+ respondents in GDS2019). The GDS2020 psychedelics self-treatment study (n=3,364) found 22.5% of people self-treating with LSD or psilocybin reported negative psychological effects (note: combined both substances in a self-treatment context, not general recreational use).
Comprehensive review of detection windows in blood, urine, and oral fluid for all major drug classes. The primary reference for detection window data across 25 substances.
Oral LSD PK in 16 healthy subjects. t½ 3.6±0.9 hr. The definitive modern LSD pharmacokinetic study.
PK of 15/25/30mg psilocybin in 28 subjects. Psilocin t½ 1.4-1.8 hr. Head-to-head comparison with LSD PK.
Demonstrated MDMA's nonlinear PK via CYP2D6 autoinhibition - a second dose hits harder because the enzyme that clears it is already disabled.
Definitive cannabis PK review. THC fat accumulation, 77-day detection in chronic users, metabolite pathways.
Evidence-based review of alcohol's zero-order elimination kinetics (~15-20 mg/dL/hr).
Controlled study of oral meth excretion duration. Mean detection 87±51 hr. Meth:amphetamine ratio data.
First-pass metabolism, bioavailability definitions, route comparison.
Route characteristics, onset times, clinical considerations.
Per-substance bioavailability data, onset/duration by route, harm reduction notes.
Safer injection guide, route-specific risk profiles, abscess prevention.
Within-subject PK study: oral cocaine bioavailability 32-45% vs IV.
Ketamine bioavailability: oral 17%, intranasal 45%, rectal 25%.
Oral LSD bioavailability ~71% in healthy volunteers.
Oral THC bioavailability 6-20%. Extensive first-pass metabolism.
Pharmacokinetic profiles including bioavailability by route for 16 substances.
Route transitioning (injection → smoking/oral) as public health harm reduction strategy.
Definitive controlled study testing temperature, container type, light wavelength, pH, and metal ions on LSD stability. Key findings: no loss at 25°C/4 weeks in dark; 30% loss at 37°C/4 weeks; amber glass/opaque containers fully protect from light.
21-week controlled study at three temperatures (-20°C, 4°C, 20°C). No significant degradation of MDMA or analogs in any matrix at any temperature.
Most rigorous academic psilocybin stability study. Dried biomass in dark at room temperature showed ~13% psilocybin loss over 15 months. Fresh frozen at -80°C showed highest degradation (counterintuitive - ice crystal damage releases enzymes).
31 seized samples, two storage conditions, measured at 9/12/36/60 months. Most degradation in first 12 months (~27%), then levels off. Low-purity samples degraded fastest.
4-year UNODC study: THC content decreased 16.6% after 1 year, 26.8% after 2 years, 34.5% after 3 years, 41.4% after 4 years at room temperature.
FDA Shelf Life Extension Program data: 88% of 122 drugs could be extended more than 1 year past expiration, average extension 66 months (5.5 years), maximum 278 months (23+ years).
Tested 8 medications (including codeine and diazepam) found in original sealed containers 28-40 years past expiration. 12 of 14 active ingredients still at >90% of labeled concentration.
360-day study: total THC decreased ~12% in first 30 days across all conditions. Amber jars retained 2.56% more THC than clear jars at 360 days. Samples at 4°C retained 14.1% more than those at 30°C.
Preliminary data: ~40% total potency loss in 6 months in stored samples. Psilocybin→psilocin conversion observed with further psilocin degradation. Not yet peer-reviewed.
12-month ayahuasca stability study. DMT was perfectly stable refrigerated at 4°C. Harmala alkaloids (especially THH) showed significant degradation after 4 months. First controlled study of ayahuasca component stability.
Firsthand history from one of ketamine's developers. Documents the synthesis at Parke-Davis (1962), first human trials (1964), FDA approval for human anesthesia (1970), and adoption as a battlefield anesthetic in the Vietnam War. Used for the historical claims in the ketamine blurb.
Reviews ketamine's profile in armed-conflict and prehospital settings: preserved respiratory drive and airway reflexes versus opioids and earlier anesthetics. Source for the 'safer in the field than opioids' claim in the ketamine blurb.
World Health Organization's official listing of medicines deemed essential for a basic health system. Ketamine is listed as an essential general anesthetic for surgery and emergency care. Source for the 'WHO Essential Medicines' claim in the ketamine blurb.
Searchable map of 600+ syringe services programs across the US, US Virgin Islands, and Puerto Rico. SSPs are the most common access point for fentanyl/xylazine test strips and FTIR drug checking. Directory is opt-in, not exhaustive.
Curated list of drug checking programs in the US and Canada with technology used (FTS, FTIR, GC/MS, reagent) and submission method (drop-off, collection box, mail, outreach).
Mail-in GC/MS service for harm reduction programs. Tested ~20,000 samples since 2022; serves 192+ harm reduction programs across 296 counties in 43 states. Free for NC groups and drug user unions, $20/sample for other organizations. Director Nabarun Dasgupta received a 2025 MacArthur Fellowship for this work.
Mail-in lab analysis for individuals using FTIR, LC-MS, NMR, and NIR. As of 2026, the active mail-in service for individual harm reduction users in the US and internationally.
Smartphone app and web portal originally built by MADDS at Brandeis University. Aggregates community drug checking results from labs nationwide, including UNC Street Drug Lab data. Useful for seeing what's circulating in your local supply.
NHRC's searchable map for naloxone, sterile syringes, and harm reduction services across California.
Federal government's official locator for substance use treatment facilities, opioid treatment programs, and buprenorphine prescribers. Not a drug checking directory — included because users seeking testing often also want treatment options.
January 2026 review of state laws governing drug checking. As of publication, three states (CA, VT, MA) protect both providers and clients; WA and OR protect providers only. Other states operate through SSP paraphernalia exemptions.
Peer-operated phone line (1-800-484-3731). Caller provides location; operator stays on the line during substance use and contacts 911 if the caller becomes unresponsive. Available 24/7.
Free, confidential psychedelic peer support by phone and text. Call or text 62-FIRESIDE (623-473-7433), daily 11am-11pm PT. Published research: 65.9% of callers de-escalated from distress (Pleet et al. 2023, Psychedelic Medicine).
Kariisa et al. 2023. Xylazine-involved overdose deaths increased nearly 20-fold from 2015 to 2020. Xylazine deepens fentanyl respiratory depression through non-opioid pathways not reversed by naloxone.
NIDA overview of xylazine as an emerging adulterant in the illicit drug supply. Covers mechanism, naloxone limitations, and skin wound complications.
Every quantitative claim on this site - harm scores, addiction rates, safety margins, purity percentages - and every brain-mechanism explainer has a per-section or per-field citation in the source code. Each value or explainer is tagged with its source study, a confidence level (measured, derived, estimated, or editorial), and a note explaining how it was obtained. Tap the ? buttons in the Rankings charts to see sourcing for individual numbers, and open a substance detail to see the source badge on the brain section.
No single source is perfect. We cross-reference multiple sources and err on the side of caution - when sources disagree, we use the more conservative estimate. If you find an error, please email corrections@opensubstance.org.