Does Acute Cold Stress Release Stored THC Into Plasma? A 15-Person Study

Answer: A ten-minute cold-water stress experiment increased biochemical indicators of fat breakdown in regular cannabis users, but did not significantly increase directly measured plasma delta-9-tetrahydrocannabinol (THC). In 15 analyzed participants, the statistical test for changes in plasma parent THC across baseline, immediately after immersion, and two hours afterward returned p = 0.950. Plasma glycerol and free fatty acids increased significantly, demonstrating why evidence of lipolysis should not be treated as evidence of THC mobilization, renewed intoxication, or impaired driving.

The experiment also found no statistically significant overall change in several laboratory attention, information-processing, or arithmetic tasks. These observations are limited to a brief, moderate cold-water challenge in a small, uncontrolled human sample. They cannot establish what happens after prolonged physical exertion, severe trauma, a collision, or more intense stress. Importantly, the study measured separated plasma, not the forensic whole-blood matrix used in many driving-under-the-influence investigations.

This analysis examines the original primary experiment by McCartney and colleagues in Psychopharmacology, DOI: 10.1007/s00213-025-06833-8. Its practical question is not whether THC can enter circulation after cannabis use—that is established—but whether this specific stressor demonstrably increases circulating THC when indicators of lipolysis change.

What was the central scientific hypothesis?

THC is highly lipophilic, meaning that some drug and related compounds distribute into lipid-rich tissues after cannabis exposure. A proposed mechanism of so-called “reintoxication” is that sympathetic stress stimulates the breakdown of stored triglycerides, potentially releasing THC from adipose tissue and producing higher circulating concentrations without new cannabis consumption. The mechanism is plausible enough to test, but several separate propositions must be established before it can support a forensic conclusion: stress must provoke measurable lipolysis; lipolysis must mobilize stored parent THC in a manner that measurably increases a specified biological matrix; the increase must be large and sufficiently timed to matter; and any behavioral consequence must be measured independently.

The authors subjected experienced users to short cold-water immersion as a practical sympathetic stressor. They collected serial biological samples and performed cognitive tasks to test whether biochemical stress accompanied a rise in plasma parent THC, its inactive carboxy metabolite, or objective task deterioration. The design thus permits a measured comparison between a physiological response and circulating analytes, rather than assuming the first establishes the second.

Who was studied, and what was the exposure?

Although 18 people initially consented, three were excluded for recent exposure or screening concerns. The final analytic sample consisted of 15 regular cannabis users, nine women and six men. Participants reported cannabis use on an average of about five days per week. The median self-reported time since last cannabis use was 23 hours, with an interquartile range of 14–40 hours. These are self-reports, not proof from continuous observation, residential sequestration, or a validated individual last-use assay.

Participants underwent approximately 10 minutes of immersion in water near 10–11 °C, up to the clavicle. They had been instructed to fast overnight and abstain from cannabis and caffeine for more than 12 hours. The intervention was neither a cannabis-administration trial nor an experimentally randomized comparison of high versus low THC doses. There was no sham-immersion group, blinding, or untreated time-control group. Consequently, even a significant change after immersion cannot automatically be attributed entirely to the cold exposure rather than elapsed time, fasting, or other accompanying factors.

Blood sampling occurred before immersion, approximately five minutes afterward, and again two hours afterward. Not every collection yielded the planned specimen: the paper reports 42 plasma samples from 45 intended sampling occasions. Multiple laboratory analyses of one specimen are technical observations, not additional independent participants. Interpretation of the repeated-measures statistics must respect both the small biological sample and missing scheduled specimens.

What did the investigators actually measure?

Venous blood was collected in EDTA tubes and centrifuged to separate plasma. The investigators used liquid chromatography–tandem mass spectrometry to assay parent delta-9-THC and its 11-nor-9-carboxy-THC (THC-COOH) metabolite. THC-COOH is analytically distinct from parent THC and is not an interchangeable measure of contemporaneous psychoactive effect. The published plasma concentrations are in ng/mL; they are not venous whole-blood THC concentrations and should not be substituted into a whole-blood per se cutoff or multiplied by an assumed fixed matrix conversion factor.

Plasma glycerol, expressed in mg/L, and free fatty acids (FFA), expressed in mEq/L, were analyzed as markers consistent with lipolysis. The authors also tracked heart rate, blood pressure, tympanic temperature, subjective stress and intoxication ratings, and three cognitive or psychomotor task families. Separate oral-fluid samples were analyzed for THC, and a point-of-collection drug screen was used. A baseline urine sample provided additional cannabinoid information, but the urine was not a serial poststress kinetic investigation.

For a useful overview of why the specimen matrix matters in forensic interpretation, see the published biological matrices guide. A concentration in one physical matrix must remain identified as a concentration in that matrix throughout the analysis.

Did plasma THC rise during or after cold-water immersion?

No statistically significant overall time effect was found for measured parent THC. The primary analysis reported χ²(2) = 0.103, p = 0.950. Original observed plasma THC medians were 0.70 ng/mL before immersion, 0.63 ng/mL just after immersion, and 0.59 ng/mL two hours later. The interquartile ranges overlapped substantially. These descriptive values are medians of available sample data; they are not modeled estimates or individual continuous concentration curves.

A nonsignificant finding is not proof that an increase could never occur. With only 15 analyzed people, the study may fail to detect small, brief, variable, or uncommon changes. Nor does the experiment contain direct human adipose-tissue THC measurements showing that no molecule left fat stores. The narrower supported conclusion is that this protocol did not demonstrate a detectable group-level increase in plasma parent THC over the measured sampling times.

What happened to the THC-COOH metabolite?

THC-COOH did exhibit an overall time effect: F(2, 25.0) = 12.1, p < 0.001. That result should not be described as a stress-induced increase in active THC. The raw median plasma THC-COOH values were 8.2 ng/mL at baseline, 7.8 ng/mL immediately postimmersion, and 6.5 ng/mL at two hours. The investigators reported that the immediate pre/post difference was not significant (p = 0.215); the two-hour measurements were lower than earlier measurements in subsequent comparisons.

These results are consistent with temporal changes in metabolism and elimination, but the absence of a no-immersion control group prevents confidently assigning the pattern to cold exposure. Equally, a significant metabolite time effect cannot rescue a failed hypothesis about increasing parent THC. Parent THC, 11-hydroxy-THC, and THC-COOH have different forensic meanings, and substituting one analyte for another would change the scientific question.

Evidence table: measured cannabinoids and lipolysis indicators

Interpretive note: The source reports medians [interquartile ranges] for the cannabinoid data and arithmetic means ± standard deviations for glycerol and free fatty acids. The time-effect tests use outcome-specific statistical models; the raw descriptors are not necessarily the model-adjusted estimated marginal means. A change in FFA or glycerol measures a physiological process, not directly a parent-THC release rate.

Did cold exposure produce a measurable stress response?

Yes, according to several independent physiological and subjective indicators. Heart rate varied significantly across the immersion and postimmersion measurements (overall p < 0.001), and tympanic temperature also changed (p < 0.001). Systolic blood pressure rose from a raw preimmersion mean of approximately 117 ± 9 mmHg to 122 ± 15 mmHg immediately afterward. The model reported an overall systolic effect of p = 0.021; the immediate comparison had p = 0.014. By contrast, diastolic blood pressure had p = 0.208 for its overall time test, a genuine null finding.

Participants generally rated themselves as less calm immediately after the cold-water challenge, while excitement and nervousness also changed with time. Importantly, increases in glycerol and free fatty acids are the direct evidence that the cold-water manipulation was accompanied by a response consistent with lipolysis. The principal empirical separation is therefore clear: measurable markers of lipolysis rose even though plasma THC did not show a significant group-level increase. This is an informative negative result for a specific stress-to-THC inference, not evidence that stress was ineffective.

Were objective attention and information-processing tasks affected?

The study used a Digit Symbol Substitution Task (DSST), Divided Attention Task (DAT), and Paced Serial Addition Task (PSAT), with several outputs for each. The original analyses found no significant time effect for the DSST number correct (p = 0.760), attempts (p = 0.758), or percentage accuracy (p = 0.335). DAT hits (p = 0.767), response time (p = 0.413), and tracking error (p = 0.182) likewise lacked significant overall time effects. PSAT correct responses (p = 0.565) and response time (p = 0.461) were also nonsignificant.

These are observed laboratory cognitive or psychomotor task outcomes, not on-road driving measures. Participants were not operating vehicles, undertaking divided attention while steering through traffic, or responding to unexpected traffic hazards. A limited battery with repeated testing cannot establish that every participant was safe to drive. Conversely, an unsupported claim that cold stress renewed THC intoxication should not be imposed on the null findings without individual behavioral or pharmacological evidence.

Evidence table: task performance, symptoms, and specimen-testing signals

Why did subjective “stoned” ratings change without a plasma THC increase?

“Stoned” and “drug effect” ratings both showed statistically significant time effects (p = 0.002), yet their magnitudes were very small. Raw “stoned” medians on a 100-mm visual analog scale were 0 mm before immersion, 1 mm afterward, and 1 mm two hours afterward. The article itself recognizes that participants were unblinded, knew the purpose of the cold-water procedure, and might respond to expectancy or ordinary mood effects of immersion.

Statistical significance on a near-zero subjective scale is not equivalent to a clinically meaningful intoxication threshold. Equally, one should not claim that all subjective experiences were unchanged: two ratings were significant. The defensible characterization preserves both details—small but detectable self-reported differences in this cohort, without a corresponding demonstrated parent-THC plasma increase or objective task deterioration.

What did oral-fluid testing show, and was the roadside-style screen reliable?

Oral fluid is not plasma, and it is not venous whole blood. Mass-spectrometric THC was detected above the reported 0.5 ng/mL limit of detection in 17 of 45 oral-fluid samples collected from seven participants. Those 17 are repeated specimens, not 17 distinct people. Detections were sufficiently sparse that the authors did not perform the main longitudinal analysis on oral-fluid THC concentrations. A single participant might contribute several oral-fluid results at different times.

A Dräger DrugTest 5000 point-of-collection device yielded two qualitative THC-positive screens across the 45 test occasions: one before cold immersion and one at two hours. Corresponding quantified oral-fluid THC concentrations were reported as 1.9 and 4.1 ng/mL, below the device’s nominal 25-ng/mL detection setting described by the authors. The study did not validate an individualized device false-positive rate from these two observations or establish precisely why the screening result differed from separately measured THC concentrations.

Oral-fluid parent THC may reflect oral mucosal contamination from previous inhalation or oral administration. It is not necessarily an equilibration readout of venous plasma THC, nor can a positive qualitative screen alone establish time of ingestion, statutory impairment, or incident cause. For the analytical distinction between screening and compound-specific assays, the site’s published confirmatory drug testing guide provides context.

Can a stress-related plasma result be converted into legal whole-blood THC?

No fixed conversion is established by this experiment. The investigators separated plasma from venous blood before measuring the cannabinoids. Some forensic legal frameworks instead specify concentrations in directly analyzed whole blood. Because the partitioning of THC between plasma and cellular components is variable, taking 0.70 ng/mL of plasma THC and describing it as 0.70 ng/mL of statutory whole-blood THC would mislabel the specimen. Assuming an invariant plasma-to-blood ratio for a person at a disputed time would add an unsupported calculation.

Even a perfectly identified whole-blood THC value is analytically distinct from demonstrated behavioral impairment and may not reliably identify the exact time of consumption. This experiment did not calibrate a human pharmacokinetic back-extrapolation model, validate a collision-time concentration, establish a per se statutory cutoff, or compare actual crash causation against a control group. The published pharmacokinetics guide addresses why dose, matrix, distribution, and sampling time should be kept separate.

How does this experiment relate to exercise, severe trauma, or a collision?

The brief immersion appeared to impose a more moderate physiological challenge than earlier supervised exercise studies. The authors compare their results with reports after cycling or running, in which stronger or longer exertion sometimes accompanied increases in circulating THC. Those protocols differed in exercise intensity, participant-use history, abstention interval, sample size, and sometimes matrix: serum and plasma are not interchangeable. Consequently, the negative cold-water observation neither establishes that vigorous exercise cannot mobilize THC nor proves that it does.

A motor vehicle collision, physical assault, prolonged panic episode, severe injury, or resuscitation has a different physiological context from ten minutes of cold-water exposure in screened, fasting volunteers. The study did not measure cortisol-mediated injury responses, major hemorrhage, traumatic hypoperfusion, acute administered intravenous fluids, postmortem redistribution, or contemporaneous real-world vehicle control. No experimental trauma group was used. Extrapolating its p = 0.950 finding to every possible stressor is as unwarranted as using glycerol’s p = 0.008 to claim that a given person’s THC must rise after a collision.

At most, the experiment offers a controlled observational counterexample to a simplistic inference that any demonstrable lipolysis necessarily produces a measurable contemporaneous circulating THC increase. Its value is in separating mechanistic plausibility from observed quantitative drug data, not in deciding the facts of an individual case.

Which limitations matter most for forensic interpretation?

Small and selected cohort: Fifteen regular users with a restricted range of health and use patterns were analyzed. This does not represent all cannabis consumers or infrequent first-time users. Recent-use exclusions and self-reported abstinence are useful screening but do not establish a known zero baseline or a precise residual body burden.

Uncontrolled sequence: Every participant underwent immersion, without a parallel randomized sham or no-immersion control. Fasting and elapsed time may have contributed to glycerol, FFA, or metabolite movement. Three planned plasma collection points were missing, and the study’s inferences are about the available repeated observations rather than 45 independent people.

Limited temporal and tissue coverage: Plasma was sampled at the reported times rather than continuously throughout the intervention. No direct adipose THC release was quantified; brief concentration transients between draws could be missed. The oral-fluid and urine measurements do not remedy that gap, because they address other specimen types and, for urine, a different sampling schedule.

No individual diagnostic or driving validation: The authors did not determine sensitivity, specificity, or a reliable personal intoxication cutoff. None of the cognitive tasks is a demonstrated on-road collision-causation assay. A null group test does not prove biological equivalence or that each participant had no change.

What is the defensible bottom line?

In this original experiment, ten minutes of cold-water immersion was accompanied by significant changes in lipolysis-related biomarkers and several markers of stress, but no significant change in directly measured plasma parent THC (p = 0.950). A significant overall change in plasma THC-COOH reflected a later decline rather than an immediate poststress rise. The measured cognitive battery had nonsignificant overall time effects, while minute subjective “stoned” ratings varied statistically. Oral-fluid screening generated a small number of positives that should not be translated into fresh blood THC exposure or a validated impairment diagnosis.

Forensic analysis must maintain the distinctions among physiological mechanism, separately measured plasma THC, oral-fluid screening, urinary metabolites, actual venous whole-blood concentrations, observed task performance, and individual driving or incident causation. This paper supports none of the last two as a direct finding. Original source: McCartney et al., “Does acute stress induced via cold water immersion increase blood THC concentrations in regular cannabis users?”

Frequently asked questions

1. Did cold-water stress increase THC in the bloodstream?

Not significantly in the plasma measured in this 15-person experiment: the original overall parent-THC time effect had p = 0.950. “Bloodstream” should not be restated as legally assayed whole blood, which was not the study’s quantitative outcome matrix.

2. If glycerol and free fatty acids rose, why did THC not rise?

The markers establish a lipolysis-associated physiological response, not a measured THC release flux from adipose stores. THC partitioning, residual exposure, stress intensity, kinetics, and sampling time can separate these events. The experiment did not directly assay adipose THC release.

3. Did the THC-COOH metabolite increase after the cold-water exposure?

No significant immediate increase was demonstrated: the baseline-versus-postimmersion THC-COOH comparison had p = 0.215. Its overall time effect was significant because the two-hour levels were lower. Parent THC and THC-COOH cannot be used interchangeably.

4. Did participants perform worse on divided-attention or processing-speed tests?

No significant overall time effect was found for the specified DSST, DAT, or PSAT outputs. This means the tested comparisons were not statistically significant in this small cohort, not that every person is always unimpaired or fit to drive.

5. Did oral-fluid tests become positive because stress released THC?

The device generated two positives in 45 occasions—one before immersion and one two hours later. That is not evidence of a newly mobilized plasma THC concentration, and a qualitative saliva screen cannot establish individualized impairment or time of use.

6. Can the reported plasma THC values be used as statutory whole-blood THC results?

No. Plasma was physically separated before quantitative analysis. Statutory whole-blood concentrations, where applicable, require interpreting the actual specified matrix and analytical method; this experiment supplied no fixed person-specific conversion or validated driving cutoff.

7. Does this prove that a crash or a severe stressful event cannot release THC?

No. A ten-minute cold-water exposure in 15 screened volunteers is not a model of every form of exertion, trauma, or emergency response. The result is a useful protocol-specific null finding; evaluating a particular incident would require contemporaneous reliable measurements and separate evidence of behavioral function and causation.