Can THC in Breath Aerosols Prove Recent Cannabis Use? An 18-Person Impaction-Filter Study

Quick answer: No single positive or negative breath-aerosol THC result from this pilot study can reliably establish whether a particular person recently used cannabis, much less whether that person was impaired. Jeerage and colleagues physically collected 42 exhaled-breath aerosol samples from 18 cannabis-using volunteers using an impaction-filter device. Laboratory mass spectrometry identified THC in 4 of 13 baseline-intake extracts (31%), 5 of 14 baseline-experimental extracts (36%), and 12 of 15 extracts approximately one to one-and-a-half hours after reported product use (80%). The post-use group showed more frequent detection, but baseline positives, post-use negatives, incomplete sample recovery and discordance with separately collected plasma THC all limit individual classification.

These are actual human breath-aerosol filter extracts, not THC vapor concentrations, a validated roadside breathalyzer, or a blood-THC conversion formula. The investigators reported analyte mass in nanograms per collection device after laboratory extraction. They did not obtain a calibrated volume of exhaled air, establish population diagnostic sensitivity and specificity, or test a validated impairment classification system. The original experimental report is Jeerage et al., THC in breath aerosols collected with an impaction filter device before and after legal-market product inhalation—a pilot study (2023), DOI 10.1088/1752-7163/acd410.

What forensic question did this experiment actually test?

The operational question was whether exhaled aerosol particles retained by a new impaction-filter collection device could yield identifiable cannabinoids under conditions approaching real-world cannabis consumption. The study brought trained participants to a mobile laboratory for scheduled breath collection, allowed them to use a specified legal-market flower product at home, and subsequently extracted the filters for laboratory analysis. A demonstration that a substance can be collected from breath is important analytical feasibility evidence. It is not identical to validating a cutoff that separates recent use from residual detection or reliably separates an impaired driver from an unimpaired driver.

THC has limited volatility compared with ethanol. The researchers investigated a particle-collection approach because cannabinoids may reside in aerosols formed from airway lining fluid. This collection hypothesis matters: the device was designed to trap particles, whereas conventional ethanol breath testing assesses an entirely different volatile analyte in a different physical phase. One should not infer a legal whole-blood THC concentration by multiplying an aerosol-filter mass by an alcohol blood-to-breath conversion ratio.

Who participated, and why are 42 breath samples not 42 people?

The relevant biological denominator was 18 independent adult volunteers, not the 42 laboratory breath specimens and not the number of filters or mass-spectrometer injections. Participants came from a naturalistic research program examining use of cannabis and anxiety. Each could provide an intake baseline, another baseline at a later experimental visit, and a post-use sample at that experimental visit. The experimental visit followed approximately four weeks during which volunteers were instructed to become familiar with the specified product. Some visits or samples were unavailable, producing 13 intake-baseline, 14 experimental-baseline and 15 post-use breath collections rather than three complete sets of 18.

Repeated collections from the same participant are related observations. Their percentages cannot be treated as independent probabilities from 42 different people. The original paper also did not establish how a future population of infrequent users, daily users, nonusers, people exposed to environmental smoke, or persons tested at different times would perform against a predeclared diagnostic rule. Most participants did not provide a complete sequence of all three breath measurements, which constrains paired comparisons.

What cannabis product was used, and how precisely was exposure controlled?

Participants were directed to obtain and use a particular licensed-dispensary, THC-dominant flower. The source reports laboratory characterization of the flower at approximately 25% tetrahydrocannabinolic acid (THC-A) and approximately 1.5% Δ9-THC, with less than 1% certain other cannabinoids. THC-A is a plant precursor that can be converted to THC by heating; a percentage printed for THC-A is not a directly measured dose of THC absorbed by any individual. Cannabis was consumed at participants’ homes, out of the investigators’ view. The investigators did not standardize the amount of flower actually smoked, number or depth of inhalations, combustion losses, or delivered systemic milligrams.

Participants were asked not to use cannabis on the day preceding scheduled sessions. That abstinence request was not verified by continuous observation. Intake interviews indicated a wide range of preceding 14-day use histories. Consequently a baseline THC-positive breath specimen does not necessarily demonstrate that the analytical device was contaminated or that a new unreported dose was taken immediately before collection. It may reflect residual material, differing use histories or other factors that this pilot cannot separate conclusively.

How were exhaled breath aerosols physically collected?

The team used the BreathExplor impaction collection device, a mouthpiece and three parallel internal filters. Participants were coached to exhale deeply, briefly hold their breath after a low-volume maneuver, inhale fully and exhale through the device, repeating the maneuver for 12 exhalations. This protocol attempts to promote generation and capture of exhaled respiratory particles. The three collection filters from a device were processed into a combined analytical extract, so they were not three independent human test results.

The researchers had considered measuring breath volume and flow with spirometry but omitted that measurement because the closer participant–researcher contact needed for the setup conflicted with pandemic precautions. Thus the laboratory reported the amount recovered per device, not a volume-normalized gas concentration. Two people could produce different aerosol yields from the same breathing instructions, and the same person might execute those instructions differently before and after cannabis use. Without contemporaneous exhaled volume, flow, and particle-number measurements, a low filter mass cannot be distinguished cleanly from inefficient aerosol formation or collection.

The devices were not taken into the rooms where the product was smoked. That reduced direct exposure of the devices to smoke at the consumption site, but it does not itself quantify potential mouth or saliva contribution during subsequent collection. The investigators did not measure salivary markers such as alpha-amylase to rule out oral-fluid contamination; their experimental goal was recovery of breath aerosols from the device, not anatomical proof of the location in the respiratory tract from which each molecule originated.

What exactly was measured in the laboratory?

The physical sample submitted for chemical testing was material eluted from three breath impaction filters, not a sealed venous blood tube and not whole exhaled gas. Each filter was soaked in methanol with ethylene glycol, residual liquid was recovered, the eluates were combined and concentrated, and the residue was analyzed by liquid chromatography–tandem mass spectrometry (LC–MS/MS). Multiple-reaction monitoring used both quantitative and qualifying ion transitions, retention-time criteria, and deuterated internal standards for compound identification.

The targeted compounds were parent Δ9-THC, cannabidiol (CBD), cannabinol (CBN), 11-hydroxy-THC (called THC-OH in the source), and 11-nor-9-carboxy-THC (THC-COOH). Parent THC is not the same substance as inactive THC-COOH. Likewise, detection of CBD or CBN cannot substitute for a measured parent-THC value. THC-OH and THC-COOH were not identified in the collected breath extracts at the source’s analytical limits; this does not prove they were absent in the participants’ blood or bodies.

Chromatographic evidence of an analyte in a prepared filter extract is more specific than a broad unconfirmed immunoassay reaction. Nevertheless, selective laboratory identification does not solve all the prior collection questions: the analyte must first reach the device, remain stable during storage, survive extraction, and pass the assay’s identification and reporting thresholds. The researchers quantified mass per device in ng/device. It would be erroneous to present those results as ng/mL whole blood, ng/L exhaled gas, or a driver impairment index.

What were the observed THC-positive breath results at each sampling stage?

The aggregate pattern supports analytical feasibility but not reliable individual use classification. THC was reported in about one-third of baseline extracts from each of the two baseline visits and four-fifths of post-use extracts. Because the denominators are small, these percentages can change markedly if only one or two results differ. The two baselines are distinct visits, not a validated negative-control set of cannabis-naïve people. The post-use measurements follow reported cannabis use, not a standardized smoked dose.

These percentages describe selected observed samples, not sensitivity for a known time-since-last-use population. The post-use fraction of 12/15 cannot establish a validated 80% sensitivity, because use occurred unobserved, timing and dose were uncertain, participants were not verified abstinent at baseline, and samples were not enrolled to establish a clinical gold-standard classification. A roadside sensitivity estimate would require a defined target condition, independently verified recent-dose timing, representative users and controls, and prespecified cutoff with reproducible collection conditions.

Why did some positive baseline and negative post-use results occur?

Baseline specimens contained detectable parent THC in both visits even though the participants were asked to abstain on the preceding day. The source documents a wide spectrum of prior cannabis use frequencies rather than a verified drug-free baseline. Detectable aerosol-associated THC at baseline could therefore have multiple explanations. This dataset does not determine which exposure, retention, airway or collection mechanism was responsible for a given baseline positive, and the timing of the last cannabis event cannot be reconstructed from that one result.

Three post-use breath extracts lacked identifiable THC. That fact is analytically important because those specimens were collected after a scheduled cannabis-use opportunity. Yet it also needs context: consumption was not directly witnessed and some plasma findings implied little or no THC exposure in particular participants. Other participants did show a sizeable plasma change without a corresponding increase in breath THC. Accordingly, the data do not justify the claim that all three negatives are proven collection-system false negatives after verified equal doses. The defensible conclusion is narrower: an isolated negative filter result did not consistently track reported use or independently assayed plasma evidence in this pilot setting.

How large were the measured THC masses, and why does one specimen matter so much?

Most positive filter extracts contained low absolute THC masses, often a few hundredths or tenths of a nanogram per device. One post-use specimen labeled I-1 contained approximately 21 ng/device, about 40 times the next highest reported post-use extract. When that exceptional specimen was included, the authors described a mean post-use result of about 1.5 ng/device; without it, about 0.15 ng/device. The roughly tenfold change in the group average when one sample is removed illustrates why averages from small pilot samples can be unstable.

The original paper did not establish that this high value was fraudulent or necessarily caused by contamination. It identified it as a potential outlier and discussed interparticipant and interstudy variability. An expert evaluation should therefore neither erase the 21-ng value silently nor treat it as representative of everyone. It should present the full range, review collection and analysis quality for its batch, and distinguish a descriptive arithmetic average from a validated prospective diagnostic cutoff.

How low were the instrument detection and quantification limits?

The laboratory analyzed breath extracts over six analytical batches distributed across the research period. The numerical lower limits varied by batch. The limit of detection (LOD) concerns the ability to distinguish a signal from noise under specified criteria; the limit of quantification (LOQ) concerns reliable numerical reporting. The source used signal-to-noise guidelines of at least 3 and 10, respectively, together with chromatographic identification criteria. A trace above detection but below quantification should not be reported as a reliable exact mass.

The table describes laboratory analytical thresholds for extracts. It is not a diagnostic cutoff for defining a recently intoxicated person, nor a direct measurement of exhaled air concentration. Researchers might achieve a very low instrument LOD while a collection device still captures only a fraction of the analyte generated in a person’s respiratory aerosol. The collection and sample-preparation recoveries must be evaluated separately.

What did recovery experiments reveal about the impaction filters?

The study included supplementary recovery testing by artificially fortifying used filter materials with known THC amounts. The team reported extraction recovery of approximately 23 ± 5% for filters eluted promptly and approximately 18 ± 6% after two weeks of storage at −20 °C. These percentages are analytical recovery measurements from prepared filter experiments; they are not the percentage of genuine cannabis users who tested positive and not a clinical sensitivity or specificity estimate. They also do not determine what fraction of inhaled THC would enter breath aerosols in the first place.

The study’s solvent recovery was considerably higher than the total analyte recovery. Thus loss cannot be attributed solely to the mass spectrometer’s nominal threshold; capture, adsorption, elution, concentration and transfers can each matter. In the participant study, extracts were processed at different times because recruitment and COVID-related logistics stretched over more than a year. Some breath samples were held for as long as 30 weeks before analysis. Storage history should therefore be treated as part of sample-specific interpretation, not ignored because a sample eventually passed a chromatographic identification criterion.

Quality-control accuracy in Batch I was outside typical acceptance limits, attributed in the paper to an internal-standard discrepancy for controls. The authors noted that this discrepancy did not affect the breath extracts in the same fashion. Two later batches also showed a possible interferent in blank-filter testing, although it was not observed in breath extracts or solvent blanks. These points warrant analytical scrutiny, but the source does not support simply reclassifying every batch result as invalid. A thorough assessment considers individual standards, controls, blank chromatograms, ion transitions and recovery data rather than imposing a blanket conclusion.

What is the distinction between aerosol THC and THC in exhaled vapor?

The device was an impaction aerosol collector. It trapped particulate material carried in exhaled air; the investigators chemically tested what they could elute from the filters. The analysis did not independently quantify THC in the gas phase, nor did it establish how much material passed through the filters without capture. Similarly, a physically measured high-temperature saturated vapor pressure from a pure cannabinoid reference sample would be a different type of result. A model predicting gas–particle partitioning at hypothetical breath temperatures cannot be substituted for the human breath-device recovery measured here.

That distinction has practical consequences. A future collection system could be optimized to trap aerosols, gas, or both, but each design would require its own validated recovery study and a defined clinical use case. A negative aerosol-filter result could reflect low true aerosol abundance, imperfect collection, loss on storage or extraction, low administered dose, no recent exposure, or some combination. Conversely, a positive aerosol THC mass indicates the targeted molecule was present in the processed collection material but does not alone identify the biological compartment of origin, precise use time, or impairment.

Did the study also measure blood THC, and was it whole blood?

The researchers referred to separately collected venous plasma THC from the linked naturalistic program. Plasma is the liquid portion obtained after processing anticoagulated blood. It is not the same matrix as unseparated forensic whole blood, in which red cells and other cellular elements remain present. Neither plasma concentrations nor filter-extract mass may be converted into a whole-blood statutory THC value without appropriate validated matrix-partition evidence.

Plasma blood sampling occurred after participants returned from their at-home cannabis-use opportunity, while breath sampling occurred later after additional assessments, around one to one-and-a-half hours following use. The sampling times were not identical. Instead of reporting precise per-participant plasma concentrations, the source table grouped results into below quantification, below 1 ng/mL, between 1 and 10, between 10 and 50, and above 50 ng/mL categories. The authors describe 12 participants with an expected post-use plasma increase and several with very low or below-quantification plasma results. This provides useful compliance context, not a validated calibration between plasma THC and breath aerosol mass.

Two participants had evidence of a substantial, at least fivefold, increase in plasma THC without a corresponding clear increase in breath THC. That discordance is central to interpretation. Breath and plasma involve distinct biological compartments, different collection times, and very different physical sampling and recovery processes. The study did not derive a clinically validated conversion factor, an individual diagnostic score, or a threshold for impaired driving. A case expert must not replace missing whole-blood THC analysis by a per-device breath number or replace missing breath measurements by plasma extrapolation.

Were THC metabolites detected in breath, and what does a nondetection mean?

Besides parent THC, the investigators looked for CBD, CBN, THC-OH and THC-COOH. CBD was identified in three breath extracts and CBN in five. THC-OH and THC-COOH were not identified in the tested breath extracts. The exact limits for those metabolites differed from the parent-THC limits and are described in the source’s supplementary analytical material. Nonidentification at a specific assay’s LOD does not establish absence from plasma, serum, whole blood, urine, or bodily tissues.

This matters because THC-COOH is an inactive major metabolite often encountered in urine testing, whereas the parent THC measured on a breath filter is a different analyte. Urinary THC-COOH positivity cannot be equated to a positive parent-THC aerosol sample. Likewise, a cannabinoid result below the device’s reporting threshold is not a scientifically secure statement that an individual did not use cannabis. Sound interpretation identifies the molecule and specimen first, then the validated method and timing, before advancing toward any broader conclusion.

Can the study establish diagnostic sensitivity, specificity, or a cutoff for recent use?

Not on the evidence presented. A validated classification study must prospectively identify an independent truth standard for what constitutes recent use, choose sampling times and cutoffs before looking at outcomes, include appropriate nonuser and abstinent-user controls, and report false-positive and false-negative performance with uncertainty. This pilot had a small selected cannabis-using cohort, home exposure that was not witnessed, self-reported preceding abstinence, variable collection success, and no predeclared forensic binary classifier. It offered a valuable descriptive distribution, not a clinically or legally adopted decision rule.

Importantly, the source did not provide a significance test proving that breath THC classification was unrelated to post-use status or proving that the method performs identically across user groups. Absence of a diagnostic validation calculation is not a nonsignificant trial result. The truthful statistical presentation is the observed counts and sources of uncertainty. A single breath result cannot be promoted into a precisely timed dosing event simply because one group showed a higher descriptive detection percentage than another.

What would a stronger validation study need to establish?

A next-generation controlled study would monitor actual inhaled THC mass or administered dose, record a well-defined use time, measure full breath flow and particle production, and collect repeat aerosol and gas-phase breath specimens under standardized protocols. It would include synchronized sampling of correctly labeled venous whole blood and plasma, without treating one as the other; quantify parent THC and relevant metabolites using suitable reference methods; and systematically examine device-to-device, day-to-day and operator variability.

Analytical validation would require measured collection efficiencies, extraction and storage recovery, limits of detection and quantification, carryover, environmental smoke controls, oral contamination checks, and documented reference-material traceability. Diagnostic validation would require a separate, representative participant cohort, clear target conditions (for example, a defined verified interval after dosing rather than an undefined subjective state), prespecified decision thresholds and independent evaluation of false classifications. If impairment is the target question, actual validated psychomotor or driving outcomes must be measured independently. A THC signal by itself does not establish that additional behavior or causal evidence.

How should a forensic reviewer interpret an individual breath-filter result?

First identify the sample as exhaled aerosol retained on a specific collection device, with extraction followed by laboratory LC–MS/MS. Preserve the chain of collection and handling: participant identification, exact collection times, number and nature of exhalations, filter lot, transport and storage durations, and whether any collection defects were recorded. Determine whether the reported mass is identified trace below quantification or a numeric value above the applicable batch LOQ. Obtain calibration, blanks, chromatograms, qualifier-ion ratios, standards, batch controls and extraction-recovery validation before assigning analytical weight.

Second keep the legal and biological questions separate. Parent THC on an aerosol filter is not an ethanol BrAC, a urinary THC-COOH ratio, a venous whole-blood THC concentration, or a proof of an exact drug-consumption time. The reported batch sensitivity must not be mistaken for sensitivity to intoxication. The 18-person study did not observe actual roadway crashes or determine whether cannabis caused a particular event. If a case turns on impairment or causation, independent contemporaneous behavioral, medical, scene and analytical evidence must be evaluated without claiming the breath aerosol result itself answers those questions.

Frequently asked questions

Does THC in breath aerosols prove cannabis was used within the last hour?

No. The investigators measured THC in baseline samples as well as in post-use samples, and exposure timing and amount were not fully observed. A positive filter extract did not establish an individual one-hour window. The study did not validate a diagnostic cutoff linking a measured mass to a specific last-use interval.

Does a negative post-use breath result prove that no THC was consumed?

No. Three of 15 post-use extracts lacked detected THC, and the study had variable exposure compliance, aerosol formation, analytical limits and recovery. Some people with plasma evidence of exposure did not show a clear breath increase. No single negative result established absence of all recent use.

Is an 80% post-use detection fraction an 80% validated sensitivity?

No. Twelve of fifteen available post-use specimens were THC-positive in a selected, naturalistic pilot. That fraction lacks the verified target condition, representative controls, prespecified cutoff and independent diagnostic-validation procedures necessary to claim generalizable clinical sensitivity.

Does the device directly measure gaseous THC in exhaled breath?

No. BreathExplor collects exhaled aerosol particles on impaction filters. Subsequent laboratory extraction and LC–MS/MS measured cannabinoid masses in the combined filter extracts. It did not independently sample and quantify gas-phase THC or report a validated gas concentration per liter of exhaled air.

Are ng/device breath results interchangeable with ng/mL whole-blood THC?

No. They are different units representing different physical matrices, with different recovery and timing. The related source blood measurements were plasma THC ranges, not legally validated whole-blood values. Neither a direct conversion factor nor an impairment threshold was established.

Why does a trace signal differ from a quantified positive result?

The instrument may meet identification criteria above a batch-specific LOD while still falling below the LOQ required for reliable numerical reporting. The source marked those detected-but-unquantified signals as traces. Their presence should not be presented as a precise concentration or evidence of a particular dose.

Can this paper prove that cannabis impaired someone who was driving?

No. This was an analytical feasibility and naturalistic collection pilot, not a controlled driving-impairment or crash-causation study. No personal blood THC cutoff, validated sensitivity to impairment, or individual legal-impairment algorithm was established by the reported breath findings.

Original primary study and related forensic methods

Primary evidence: Jeerage KM, Beuning CN, Friss AJ, Bidwell LC, Lovestead TM (2023). THC in breath aerosols collected with an impaction filter device before and after legal-market product inhalation—a pilot study. Journal of Breath Research 17:037103. https://doi.org/10.1088/1752-7163/acd410. This source reports one experimental pilot; it does not certify a roadside device or govern forensic laboratory practice.

For related but distinct interpretive issues, see the existing published explanations of biological specimen matrices, confirmatory drug testing and specimen collection and handling. Their general methodological framework supplements, rather than replaces, review of the complete original study and case-specific laboratory record.