Can Oral THC Cause Prolonged Feed Refusal and Elevated Troponin in Sheep? An Eight-Ewe Study

Short answer: In a 2026 primary experiment, eight Merino ewes received two large oral doses of delta-9-tetrahydrocannabinol (THC) through an orogastric tube. Every ewe developed observable neurological or behavioral signs within 24 hours. Average dry-matter intake fell markedly during the following week, and three animals had a plasma cardiac troponin I concentration above a provisional 100 ng/L interpretive threshold at one selected sampling time. The experiment demonstrates adverse findings after this controlled high-dose exposure. It does not determine a generally safe hemp-feed concentration, prove that three animals had myocardial infarction, or establish the cause of signs in an individual livestock case.

The evidence comes from Stevens and colleagues, The toxicology of orally administered Δ9-tetrahydrocannabinol in sheep, published in Toxicology Reports in 2026, DOI 10.1016/j.toxrep.2026.102221. The report presents adverse clinical observations from a larger pharmacokinetic investigation; its cannabinoid concentration-time data were discussed in a separate publication. That separation matters when interpreting what the toxicology report actually measured and disclosed.

What scientific question did the researchers test?

The investigators were evaluating possible animal-health consequences of industrial-hemp biomass as a ruminant feed. Hemp varieties described as low in THC can still supply appreciable cumulative THC if a large amount is eaten. For an adverse-effect experiment, the investigators modeled a high-exposure grazing scenario and administered THC in a concentrated resin solution rather than feeding a conventional hemp ration. The narrower research question was whether a large controlled oral THC challenge would be followed by neurological signs, altered feeding, body-weight change, or a cardiac biomarker response in sheep.

Those are four distinct outcomes. Ataxia is an observed gait sign; feed consumption and body weight are production and nutritional measures; cardiac troponin I is a biochemical marker associated with myocardial cell injury. They are not different measurements of the same variable. A veterinary opinion should identify which endpoint supports which conclusion instead of treating a positive finding in one domain as proof in every other domain.

How many sheep were studied, and what did they receive?

Eight Merino ewes, approximately 24 months old, formed the exposed cohort. According to the Methods and abstract, they received two orogastric THC administrations of 88.5 mg/kg body weight each, twelve hours apart, for a nominal combined amount of 177 mg/kg. The formulation used a measured 73.2% THC resin dissolved in dimethyl sulfoxide (DMSO); the investigators used an orogastric tube and an additional DMSO flush. The table and figure label the exposure as 177 mg/kg. The paper also describes a reduction from a higher planned exposure to reduce potential fatality risk. That design history should not be confused with an independently demonstrated safe dose.

The animals underwent a pre-study adaptation period and a subsequent sample-collection period. They had access to wheaten chaff and commercial sheep pellets, with refusals measured for dry-matter-intake calculations. There was no simultaneously randomized untreated or vehicle-only group in this adverse-effect report. Each ewe’s observations before and after dosing therefore provide temporal evidence, but cannot isolate THC from all possible dosing-procedure, vehicle, housing, and nutrition effects with the rigor of a placebo-controlled trial.

What were the directly observed clinical signs?

Within the first 24 hours, all eight showed ataxia, abnormal posture, somnolence, and wool pulling to varying degrees. The researchers described stumbling or swaying; heads held low or resting against a pen wall; and unusual positioning of the hind legs. Two of eight also had increased salivation. Many conspicuous signs became less apparent over the ensuing 24–48 hours, but wool pulling persisted for roughly a week.

These are important adverse findings precisely because they were observed by animal handlers rather than inferred from a modeled blood concentration. Nevertheless, there was no blinded behavioral rating against an untreated group and no independently validated sheep-specific psychomotor impairment score. The paper does not provide a quantitative correlation between a particular ewe’s measured plasma THC level and her ataxia severity. Observing ataxia is evidence of abnormal function during the experiment; it is not an analytical THC concentration or proof of the identical mechanism in another animal.

What did the study actually measure? Evidence table one

The table keeps the cohort denominator at eight and separates direct observation, longitudinal nutritional measurements, and biochemical testing. It does not turn repeated daily measurements into hundreds of independent sheep. Nor does it transform the proportion three of eight into a population incidence rate for hemp-fed livestock.

How much did feed intake decrease, and was the difference statistically supported?

The strongest longitudinal quantitative comparison in this particular paper involves dry-matter intake (DMI). Before dosing, mean DMI was 3.4% of body weight. During the first week afterward, it was 1.2%, followed by 2.3% in the second week and 3.3% in the third. Each weekly estimate in the authors’ Table 1 has a reported least-squares standard error of 0.138 percentage points of body weight. The repeated-measures mixed model found an overall week effect at p < 0.0001.

The original table’s letter groupings distinguish the first post-dose week from the second, and both from the pre-dose and third-week values at the authors’ pairwise significance criterion. This is not a claim that the entire first week’s decline was unchanged every day. Their Figure 1 shows a transient rise shortly after exposure, a substantial fall beginning around the dosing period, and a progressive recovery. Reporting only the lowest point without its recovery trajectory would omit important toxicological context.

Because the outcome is expressed as a percentage of body weight, it describes ration consumption normalized for animal size—not the THC percentage of feed. A statement such as “intake decreased to 1.2% THC” would therefore be a serious unit error. The result is a change in food consumption following dosing, not a test of a 1.2%-THC crop.

Did the ewes lose weight, and how well was weight change established?

Descriptive mean body weight was 47.75 kg (SD 3.49) on dosing day zero, 43.13 kg (SD 3.40) at day 14, and 49.50 kg (SD 3.23) on day 21. The mean difference between day zero and day 14 is 4.62 kg, but that arithmetic comparison is not itself an independently reported significance test. The researchers attributed the temporary reduction in weight to reduced feed intake and described recovery as feeding resumed.

Weight trajectories provide a plausible production-relevant effect, particularly when paired with prospective measurements of uneaten feed. Yet body weight also responds to hydration, gastrointestinal contents, and management conditions. The published toxicology report did not provide a randomized comparison or complete individual body-weight trajectories needed for a fully adjusted independent causal estimate. In a livestock-loss dispute, date-specific weights, feed logs, husbandry records, and independent illness assessment would be necessary before assigning the same mechanism to a particular loss.

What does an elevated plasma troponin I result establish?

The investigators collected jugular venous blood and analyzed plasma cardiac troponin I by a Siemens Centaur chemiluminescent immunoassay at a veterinary pathology laboratory. They measured troponin I on one selected post-exposure time point per ewe, chosen around that animal’s observed peak plasma THC concentration from the larger pharmacokinetic work. Those selected times ranged from about 12 to 168 hours after dosing. The article did not publish paired serial troponin curves for each animal.

Three measured concentrations—144, 146, and 275 ng/L—exceeded a provisional 100 ng/L interpretive threshold; five did not. Overall, the eight results had a mean of 96.3 ng/L, an SD of 87.57 ng/L, and a range of 13–275 ng/L. The authors explained that the assay was not sheep-specific and that a firm validated ovine normal reference interval was unavailable. They used a roughly 0–100 ng/L contextual range from existing ovine literature and earlier samples.

Troponin supports the possibility of myocardial cell injury, but it does not identify the source of the injury, distinguish every competing veterinary diagnosis, or establish an infarction from one isolated result. The study does not describe a matched untreated troponin comparison, serial ECG-confirmed ischemia in each elevated ewe, or histopathologic verification of myocardial necrosis. The paper’s phrase suggesting a myocardial effect is appropriately read as an interpretation of a limited biomarker signal, not proof that exactly three animals had a defined cardiac event.

How should biological specimens and analytical methods be distinguished?

Several scientific variables occur in the same project. The administered dose was calculated from a weighed THC resin. The clinical signs were observations of the animals. The dry-matter intake was inferred from feed offered and refusals. The troponin I result was a direct immunochemical measurement in processed plasma. Separately, blood samples for cannabinoid pharmacokinetics were collected in the overarching investigation, but the concentration-time results were reported in another publication.

Therefore, a plasma troponin concentration of 275 ng/L is not a plasma THC concentration of 275 ng/L. Nor does a 177 mg/kg nominal administered oral dose equal a measured whole-blood THC concentration, brain exposure, or a validated individual intoxication threshold. Neither exhaled breath THC nor human forensic venous whole-blood THC was an endpoint of this eight-ewe toxicology paper.

For a broader primer on why physically different specimens and molecular targets must be kept separate, see the already published biological matrices guide. The related published pharmacokinetics guide explains why administered dose, distribution, measured concentration, and modeled clearance are distinct concepts.

Does this experiment establish that ordinary industrial-hemp feed harms sheep?

No universal conclusion about ordinary hemp feed follows from the dosing method. The investigators used a purified THC-rich resin in DMSO administered through a tube. Sheep grazing actual hemp biomass ingest a chemically variable mixture of plant material, cannabinoids, acidic cannabinoid precursors, fiber, other nutrients, and potentially different doses over time. The oral bioavailability and tolerability of a resin challenge cannot automatically be equated with grazing or eating a particular finished ration.

The authors considered Australian regulatory and worst-case grazing assumptions when designing the experiment and discussed a hypothetical feed-equivalence estimate. These were exposure scenarios, not measurements of THC in feed that the eight ewes actually consumed. The paper also summarized other feeding studies with different rations and concentrations, including studies without the same conspicuous adverse effects. Those other publications are contextual comparisons; they were not additional animals randomized within Stevens and colleagues’ experiment.

Importantly, the authors suggested that a possible upper safe hemp concentration might fall between approximately 0.05% and 0.1% THC, based on their study and the literature. This is an exploratory hypothesis, not a directly determined safe threshold or a dose-response no-observed-adverse-effect level. There was only one exposed cohort at the high experimental schedule, and no series of graded oral doses allowing a threshold to be estimated from this experiment alone.

What do the principal numerical outcomes show? Evidence table two

The intake table’s uncertainty term is least-squares standard error, whereas body weight and troponin variability are reported as standard deviations. They should not be presented as interchangeable. Statistical significance attaches to the authors’ repeated-measures weekly intake model, not automatically to every animal sign, the selected troponin threshold, or a predicted grazing concentration.

Could wool pulling and anorexia have other explanations?

Yes. The authors offered more than one possible explanation for wool pulling: a THC-associated sensory or behavioral disturbance, adaptation stress, or a secondary nutritional process as eating became irregular. None of these alternatives was experimentally isolated. Their discussion drew a comparison with certain symptoms described in humans, but a sheep’s wool-pulling behavior cannot establish a human psychiatric diagnosis or the animal’s subjective experience.

Anorexia likewise may reflect neurological depression, gastrointestinal effects, diet characteristics, husbandry, competing illness, or combinations of causes. The timing after THC exposure gives the observation clinical significance, and the recovery pattern is consistent with a transient exposure-associated process. But without an untreated randomized group or a vehicle-only control, the individual causal pathway remains incompletely identified. The distinction is particularly important when someone seeks to attribute a particular animal’s injury, death, productivity loss, or abnormal behavior to a specific batch of feed.

How would an expert evaluate a real livestock-exposure dispute?

Begin with the actual material rather than the paper’s modeled worst-case dose. Obtain the feed or plant batch, chain of custody, cannabinoid sampling plan, total-THC calculation method, delta-9 THC and THCA concentrations, and any evidence of degradation or processing. Obtain each animal’s estimated ingestion history, body weight, timing, and whether exposure was continuous or a single large challenge. A chemical assay of the feed does not itself establish how much one ewe swallowed or absorbed.

Next, separate the observations by source: handler narratives, video, gait examinations, ration refusals, weights, jugular blood collection, plasma troponin, and any confirmed cannabinoid analysis. Determine whether troponin was measured with a validated method and an appropriate species-specific interpretive interval. A negative or equivocal biomarker does not erase witnessed ataxia, and a positive biomarker does not uniquely diagnose THC-caused cardiac disease.

Finally, review differential diagnoses, exposure timing, other feed and medications, hydration, infectious or metabolic illness, clinical treatment, follow-up samples, and any necropsy or pathology. The published confirmatory testing guide explains why testing method, target analyte, and question must align. Expert conclusions should separate support for exposure, support for a toxic syndrome, evidence of a causal mechanism, and the proof required for a specific damages or liability opinion.

What are the most important study limitations?

Only eight ewes were exposed, and all received a large THC-rich resin challenge with the same broad dosing schedule. No randomized untreated or DMSO-only group was reported in this toxicology analysis. The study was not a long-term feeding trial across commercial hemp rations and did not map a graded dose-response curve. The figures provide a useful repeated-measures trajectory, but animal-to-animal heterogeneity and possible methodological confounders remain substantial.

Cardiac inference is even narrower: the troponin assay was not species-specific, a validated ovine reference interval was unavailable, and troponin was tested once per animal at differing selected times. No individual myocardial infarction endpoint or mortality from this oral dosing experiment was demonstrated. Pharmacokinetic THC concentration-time values belong to a companion paper; an individual exposure–effect model cannot be reconstructed from the adverse-effect report alone. These limitations affect the strength of interpretation, not whether the observed clinical and intake findings deserve attention.

Frequently asked questions

Did every sheep become clinically abnormal after the oral THC challenge?

Yes. The researchers reported ataxia, abnormal posture, somnolence, and wool pulling in all eight ewes within 24 hours. The observations were not obtained from a blinded comparison with untreated sheep.

Was 177 mg/kg the THC concentration measured in sheep blood?

No. It was the nominal total orally administered THC amount per kilogram of body weight across two 88.5 mg/kg administrations. It is not a plasma or whole-blood concentration; blood cannabinoid pharmacokinetics were reported separately.

How large was the drop in average feed consumption?

Weekly mean dry-matter intake changed from 3.4% of body weight before dosing to 1.2% in week one, recovering to 2.3% in week two and 3.3% in week three. The authors’ repeated-measures analysis reported p < 0.0001 for the week effect.

Did the investigators diagnose heart attacks in three ewes?

No. Three of eight had a single plasma troponin I result above a provisional 100 ng/L level. The non-sheep-specific assay and absence of confirmatory cardiac pathology do not support labeling those three animals as confirmed myocardial-infarction cases.

Did the experiment directly test normal hemp grazing?

No. The ewes received concentrated THC resin dissolved in DMSO by orogastric tube. A commercial hemp ration contains other ingredients and may involve very different dose and exposure patterns.

Does the study establish a legally or medically safe percentage of THC in feed?

No. The authors discussed a possible range using this and other studies, but this high-exposure cohort was not a multi-dose controlled safety-threshold trial. A proposed feed specification requires evidence on actual rations, dose, animal health, and applicable regulation.

Can these sheep findings prove THC impairment or accident causation in a human?

No. The outcomes concern a small ovine high-dose toxicology cohort. They contain no human driving task, measured human blood THC impairment threshold, individual road behavior, or case-specific causation evidence.

Primary study and interpretive conclusion

Stevens SA, Edwards SH, Noble GK, Scrivener CJ, Petzel CE, May CD, Tai ZX, Blake BL, Dods KC, Warne LN, Krebs GL. The toxicology of orally administered Δ9-tetrahydrocannabinol in sheep. Toxicology Reports. 2026;16:102221. DOI: 10.1016/j.toxrep.2026.102221.

Forensic bottom line: A controlled large oral THC challenge was followed by conspicuous neurological signs, a statistically supported prolonged decline in feed intake, weight change, and three elevated single-time-point plasma troponin I readings among eight ewes. These are real experimental observations, but their value in a livestock investigation depends on an accurate exposure reconstruction, species- and assay-specific interpretation, and consideration of competing explanations. They do not independently define a safe hemp crop, a confirmed cardiac lesion, or a human impairment rule.