CDSI

Library / Papers

PUBLIC HEALTH

Adulterant Amplification

How EVALI could have been caught — a counterfactual public-health analysis.

PaperP-004TypeRetrospective / counterfactualStatusFounding Draft v0.3Length~3,800 words

How Pre-Market Loaded-State Off-Gas Testing Could Have Surfaced the EVALI Signature

Author: Matt Macosko, Founder, Cannabis Device Safety Institute Affiliation: Cannabis Device Safety Institute, Arcata, California Status: Founding draft, v0.3 (April 2026) Suggested citation: Macosko, M. Adulterant Amplification: How Pre-Market Loaded-State Off-Gas Testing Could Have Surfaced the EVALI Signature. CDSI Working Paper P-004. 2026.

A note on tone before we begin

This paper makes a counterfactual claim: that 68 deaths and 2,807 hospitalizations from the 2019 e-cigarette and vaping-associated lung injury (EVALI) outbreak might have been preventable by a routine, pre-market hardware testing protocol that did not exist at the time and still does not exist as a regulatory requirement.

Counterfactual claims about specific harms are uncomfortable. They are also, sometimes, the most honest argument available. From where this author sits, the alternative — declining to make the claim because it is uncomfortable — is the failure mode that lets the next outbreak happen on the same terms.

The paper is written carefully. It does not accuse any individual party of negligence. It does identify a structural absence and propose a remedy. Read it, please, in that spirit.

Abstract

The 2019 EVALI outbreak was attributed primarily to vitamin E acetate (VEA) in illicit-market cannabis cartridges. That attribution is correct but incomplete. VEA at ambient conditions is largely inert; the lung injury was the product of a device-mediated thermal reaction in which VEA, when heated by an active vaporizer at typical operating temperatures, decomposes into ketene gas and a class of acutely lung-toxic carbonyl compounds. We propose this category of harm be called adulterant amplification (CDSI Lexicon §4.3): harm that is neither solely the adulterant’s fault nor solely the device’s, but the product of their interaction. We argue that a pre-market loaded-state off-gas (LSO) testing protocol, applied to consumer cartridges in 2018, would have detected the amplification signature. The paper specifies what that signature looks like, why neither concentrate-side testing nor empty-heater hardware testing would have surfaced it, and what a forward-looking testing protocol would need in order to catch the next adulterant before it scales. The Cannabis Device Safety Institute (CDSI) was constituted, in part, to be the body whose protocol would have caught EVALI.

Keywords: EVALI, vitamin E acetate, ketene, vaporizer, adulterant, loaded-state off-gas, cannabis hardware safety

1. The 2019 outbreak and what was attributed to it

Between June 2019 and February 2020, the Centers for Disease Control and Prevention (CDC) identified 2,807 hospitalizations and 68 confirmed deaths attributable to EVALI, “e-cigarette, or vaping, product use-associated lung injury.” The clinical picture was severe: bilateral pulmonary infiltrates, hypoxemia, acute respiratory distress syndrome, in many cases requiring mechanical ventilation. Median patient age was 24.

By November 2019, CDC investigators had identified vitamin E acetate as a “chemical of concern” present in 94% of bronchoalveolar lavage fluid samples from EVALI patients. The compound’s source was traced to illicit-market THC cartridges, where it had been used as a viscosity-modifying additive — a “cutting agent” that thinned distillate while preserving the appearance and density characteristic of higher-quality concentrate.

The public-health response was rapid and, on the regulatory dimensions available, effective. Several states banned VEA in cannabis products. CDC issued consumer guidance. The illicit market shifted away from VEA, and EVALI cases declined sharply through 2020.

The story, as it is generally told, ends there. Vitamin E acetate was the cause, the cause was identified, and the response was implemented.

This paper argues that the story is incomplete in a way that matters.

2. The incomplete attribution

Vitamin E acetate, at ambient conditions, is not acutely lung-toxic. It is approved as a topical cosmetic ingredient and as a dietary supplement. It is sold over the counter. The compound itself, in the absence of a vaporizer, would not have produced the EVALI cluster.

What produced the cluster was the interaction between vitamin E acetate and a hot vaporizer heating element — typically operating between 350 °F and 700 °F, well within the range at which VEA undergoes pyrolytic decomposition. The decomposition products include ketene gas (CH₂=C=O), an acutely lung-toxic compound with no known safe inhalation threshold, and a class of related carbonyl species with similar toxicity profiles.

The pyrolysis pathway was characterized in the analytical chemistry literature within months of the EVALI peak. By early 2020, papers in the Journal of Inhalation Toxicology and elsewhere had documented the temperature thresholds, the product distribution, and the mechanism. The science was, in the technical sense, settled within a year.

What was not settled — what has still not been settled, six years later — is whether the standard cannabis hardware safety testing landscape is structured to catch the next compound that behaves like vitamin E acetate did. From where this author sits, the answer is plainly no.

The reason: VEA’s harm did not appear in:

  • Concentrate-side testing, because VEA is not a pesticide, not a solvent residue, not a microbial contaminant, and not a heavy metal. It does not appear on any state cannabis lab’s required analyte list, and would not appear under a generalized “anything bad” requirement either, because at ambient temperature VEA is not detectably bad.
  • Empty-heater hardware testing, because the harm is not a property of the heater alone. The heater operating empty produces no ketene, no VEA decomposition products. A device that passed a comprehensive dry-fire test could and did contribute to fatal injury when loaded with VEA-containing concentrate.

The harm lives in the interaction. Concentrate-side testing cannot see it. Empty-heater testing cannot see it. The category of testing that can see it — loaded-state off-gas analysis under realistic operating conditions — was not, and is still not, required of any cannabis hardware before market.

This is what we mean by adulterant amplification: the harm that emerges when a not-particularly-toxic adulterant in concentrate meets a hot heating element under operating conditions and produces a different, dangerous compound class. The adulterant alone does not explain it. The device alone does not explain it. The interaction is the harm.

3. What pre-market LSO testing would have shown

Imagine, counterfactually, that in 2018 a body resembling CDSI had existed and had been performing CDSI-001-style loaded-state off-gas testing on consumer cannabis cartridges entering the regulated market. What would the analyte tables have looked like for VEA-containing units?

The protocol’s target analyte list (CDSI-001 §6.1) includes carbonyls — specifically formaldehyde, acetaldehyde, acrolein, and glyoxal — and a category called “cutting agent indicators” that explicitly includes vitamin E acetate, propylene glycol, and polyethylene glycol breakdown products.

Under the CDSI-001 operating profile (three temperatures, three matrices, twenty draws per cycle, with sorbent and particulate-filter capture), a VEA-loaded cartridge would have shown:

  • Vitamin E acetate detected on the sorbent tube above the protocol detection limit, by source-fingerprint: this finding alone would have triggered an immediate flag on the public report
  • Ketene gas detected as a transient signature, with quantitative assignment difficult due to ketene’s reactivity but with reliable detection of its conjugate stable products (acetic anhydride decomposition signatures, specific carbonyls)
  • Carbonyl concentrations 5–20× above baseline for an unadulterated cartridge of similar architecture
  • Distinctive thermal-speciation behavior — the concentration of VEA decomposition products rising sharply with temperature in a pattern that does not match any non-VEA matrix

These signatures are not subtle. They do not require novel instrumentation. They are detectable by routine GC-MS using sorbent capture, the same analytical platform used in occupational hygiene laboratories worldwide for the past four decades.

The CDSI-001 protocol calls for every test report to be public within 90 days of intake, regardless of outcome. A 2018 cartridge cohort that included VEA-adulterated units would have produced public reports flagging the VEA presence, the ketene signature, the carbonyl elevation, and the matrix-temperature profile — all by the time many EVALI patients were first encountering VEA-containing products.

The reports would not have been ignored. They would have flagged a public-health pattern across multiple manufacturers, multiple states, multiple supply chains. They would have given public-health officials, regulators, and clinicians something to point to before the cluster scaled.

This is not a claim that a single CDSI report would have prevented EVALI. It is a claim that the existence of a body whose job is to look for amplification signatures would have shifted the timeline. Sixty-eight deaths happened over roughly nine months of compounding case identification. A six-month earlier surfacing of the amplification signature could have averted a meaningful fraction of those deaths.

4. What we still don’t know — and what comes next

EVALI is the public-health-scale case study. It is not the only adulterant amplification event possible, only the largest documented one. The structural conditions that produced it — adulterants entering an unmonitored hardware-concentrate interaction at scale — are unchanged.

Active areas of concern for adulterant amplification, as of 2026:

  • Botanical terpene additives sourced outside the cannabis supply chain (e.g., orange-peel-derived limonene, conifer-derived alpha-pinene). These are generally regarded as safe for ingestion but their thermal decomposition behavior at vaporizer temperatures has not been systematically characterized.
  • Polyethylene glycol (PEG) carriers used in some cartridge formulations. PEG decomposition products at vaporizer temperatures include formaldehyde and acetaldehyde at concentrations that exceed routine indoor-air guidelines.
  • Untested cannabinoid analogs (delta-8-THC, HHC, THCP) entering markets faster than analytical methods can characterize their thermal behavior. Some of these compounds are isomerized from CBD using acid catalysts whose residues are not on any state cannabis lab’s analyte list.
  • Microemulsion stabilizers used in water-soluble vape formulations. The interaction between surfactants and hot heaters is well-studied in food-processing contexts and poorly studied in vape contexts.

Each of these categories is a future EVALI candidate. The CDSI position is not that any of them will produce a comparable event; it is that the structural absence of a body whose job is to characterize them in advance is what makes another EVALI-class event possible.

5. What CDSI proposes

To prevent the next EVALI-class event, four operational practices are needed, and CDSI is constituted to provide them:

  1. Routine pre-market LSO testing of new cartridge formulations and adulterant chemistries entering regulated markets, performed under a published protocol (CDSI-001 or its descendant), reported publicly, and submitted as a baseline reference into the analytical chemistry literature.
  2. Open analyte lists — published, version-controlled, and updated quarterly to incorporate new adulterant chemistries surfacing in either licensed or illicit markets.
  3. Cross-laboratory validation of the signature library, so that a finding from any participating laboratory can be replicated and confirmed by another. CDSI proposes the Humboldt model (CDSI Lexicon §5.4) — region-anchored institutes coordinating through shared methodology — as the operational form.
  4. Direct line to public-health officials when an amplification signature is detected. The CDSI-001 protocol’s 90-day public-report window is appropriate for routine findings; an amplification signature detected in a market-circulating product warrants accelerated notification (proposed: 7-day window to relevant state public-health bodies, public report within 30 days).

The CDSI-001 protocol satisfies the first practice. The CDSI Lexicon and Papers Pipeline are early instantiations of the second. The Humboldt model and the proposed CDSI-HIIMR partnership are early instantiations of the third. The fourth is a governance practice that CDSI is in the process of formalizing in its bylaws.

None of this is exotic. All of it could have existed in 2018. None of it was structurally available, because the body whose job is this work did not exist.

6. Conclusion

Sixty-eight deaths. Two thousand eight hundred and seven hospitalizations. A median patient age of 24.

The story we tell about EVALI matters. The version where vitamin E acetate is “the cause” implies that once VEA was identified, the problem was understood. That version is comforting and incomplete. It does not explain why no testing infrastructure caught it. It does not explain what infrastructure would catch the next one.

The version this paper proposes — that EVALI was a public-health-scale case of adulterant amplification, an interaction between adulterant and device that lives in a category neither concentrate-side nor empty-heater testing surfaces, that a pre-market LSO protocol could have flagged six months earlier than the clinical case identification did — is harder. It implicates a structural absence. It points to a body that does not yet exist.

It also points to what to build.

CDSI is one such body. The Humboldt model invites others. From where we sit, the work is overdue, and the cost of waiting is measured in people. We are starting it.

Acknowledgments

The author is grateful to the CDC investigators whose 2019–2020 work made this paper possible, and to the analytical chemists whose subsequent characterization of vitamin E acetate’s pyrolysis pathway provided the technical foundation for the adulterant amplification concept developed here. Any errors of interpretation in this paper are the author’s alone.

Disclosures

The author is the founder of Divine Tribe and Nice Dreamz LLC. Divine Tribe manufactures cannabis consumption hardware; the author has commercial interest in the regulatory environment for that hardware. These disclosures are placed on the cover page of every CDSI document per pay-the-lab discipline.

The author has no financial interest in the testing or banning of vitamin E acetate.

References (preliminary)

  1. Centers for Disease Control and Prevention. Outbreak of Lung Injury Associated with the Use of E-Cigarette, or Vaping, Products: Final Report. 2020.
  2. Blount, B.C., et al. Vitamin E Acetate in Bronchoalveolar Lavage Fluid Associated with EVALI. NEJM 2020.
  3. Wu, D., O’Shea, D.F. Potential for Release of Pulmonary Toxic Ketene from Vaping Pyrolysis of Vitamin E Acetate. PNAS 2020.
  4. Cannabis Device Safety Institute. CDSI-001 Protocol: Off-Gas Analysis of Cannabis Concentrate Vaporizers, v1.0. 2026.
  5. Cannabis Device Safety Institute. The Hardware Vacuum. CDSI Working Paper P-001. 2026.
  6. Cannabis Device Safety Institute. Loaded-State Off-Gas Analysis of Cannabis Concentrate Vaporizers. CDSI Working Paper P-002. 2026.
  7. Cannabis Device Safety Institute. The CDSI Lexicon, v. 2026-04. 2026.

Comments and corrections to: matt@ineedhemp.com (until papers@cdsi.io is provisioned).