Does a High-Fat Meal Triple CBD Exposure? A 32-Adult Powder-versus-Oil Crossover Study

Short answer: In a randomized four-period pharmacokinetic crossover involving 32 healthy adults, a high-fat meal was associated with approximately three- to three-and-a-half-fold greater plasma CBD exposure over 24 hours after an identical 400-mg oral dose. Yet food did not increase every measured CBD metabolite in the same way. Compared with fasting, the powdered emulsion produced higher parent-CBD exposure but lower 7-carboxy-CBD exposure after a meal. Formulation also mattered: fasting adults given the powder reached peak CBD concentrations earlier and had higher metabolite exposure than when given the oil solution, even though the difference in parent-CBD 24-hour exposure was not statistically significant. These are population pharmacokinetic findings—not evidence that any particular person was impaired, took an unlawful substance, or had a particular concentration at a past event.

The underlying original experiment is Bendik and colleagues’ 2026 open-label, randomized crossover comparison of CBtru powder and Epidyolex oil, published in CNS Drugs: doi:10.1007/s40263-026-01280-1. Its analytical endpoints were directly quantified plasma concentrations of cannabidiol (CBD), 7-hydroxy-CBD (7-OH-CBD), and 7-carboxy-CBD (7-COOH-CBD), together with pharmacokinetic quantities calculated from those concentrations. The experiment did not measure tetrahydrocannabinol (THC), whole-blood CBD, road driving, sobriety tests, or individual forensic impairment.

The central forensic question: can the same oral CBD dose create different test results?

Yes. A labeled dose is an amount administered to a person; it is not a biological concentration. Even when the same number of milligrams is swallowed, absorption, first-pass metabolism, food, formulation, sampling time, and biological variability can produce different plasma results. This controlled study isolated two especially important determinants: the type of oral CBD product and whether it was taken after a standardized high-fat meal. It supplies experimental evidence that nominal dose equivalence and measured systemic exposure are separate propositions.

The distinction matters when records mention a CBD capsule or oral solution and a later laboratory result. One cannot translate a stated dose directly into an individual’s blood concentration, and a metabolite cannot automatically be interpreted as proof of a different dose, intoxication, or timing of administration. Nor can a trial measuring CBD in separated plasma be treated as a validation study for a forensic venous whole-blood test. The appropriate starting point is the specimen matrix and what it can actually establish.

Who participated, and what was the crossover design?

Fifty-seven volunteers were screened; 32 healthy adults were randomized. The enrolled group comprised 16 women and 16 men, with mean age 40.2 years. Thirty completed all four planned treatment periods. Two participants withdrew voluntarily after receiving their first dose—oil in the fed condition—and therefore did not complete the three remaining conditions. The researchers retained all 32 randomized, exposed participants in the reported pharmacokinetic and safety analysis sets, while the number contributing data to each condition differed.

Each period involved a single 400-mg oral CBD dose, not continuous treatment. The four conditions were powdered CBtru while fasting, oil-based Epidyolex while fasting, CBtru after a meal, and Epidyolex after a meal. Participants were assigned to one of four dosing sequences with at least 14 days between doses. A crucial design nuance is that the first two periods were conducted under fed conditions and the fasting periods followed. Although treatments were randomized by sequence, food state was not fully counterbalanced over calendar period. Consequently, the food effect is strongly associated with the manipulated meal condition but should not be described as the result of an independently randomized fed-versus-fasted period order.

The trial was open label; both participants and investigators could know the formulation. Concentration measurements by a validated analytical method are less vulnerable to participant expectation than subjective symptom scores, but open labeling still matters when considering tolerability. It was a single-center Phase I pharmacokinetic study in generally healthy adults, not an impairment investigation, treatment-efficacy trial, or study of chronic dosing.

What exactly was eaten, and when were samples collected?

The fasting condition followed at least ten hours without food. In the fed condition, participants fasted before consuming a standardized approximately 800–1,000-kilocalorie breakfast with about half of its energy supplied by fat; dosing occurred within 30 minutes of the meal. The breakfast was designed to measure a food effect under a defined high-fat condition, not the influence of an arbitrary snack, daily diet, or specific person’s meal.

Venous blood was collected by catheter or venipuncture and centrifuged; the separated plasma was stored frozen before analysis. Each treatment period included a predose specimen and samples at 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 5, 6, 8, 12, and 24 hours after dosing. Across the trial, 1,708 plasma samples were analyzed, with 35 noted as potentially hemolyzed. These are repeated measurements from no more than 32 independently enrolled people, not 1,708 independent participants.

That distinction is essential when interpreting concentration-time graphs. The study depicts repeated profiles within people and then summarizes them; the apparent abundance of sample observations does not remove uncertainty about person-to-person variation. Likewise, a sampled peak time is constrained by the study’s sampling schedule, and a 24-hour follow-up is not a complete examination of all later elimination.

How were CBD and its metabolites measured?

The laboratory used liquid chromatography–tandem mass spectrometry (LC–MS/MS), with isotope-labeled internal standards and a liquid-liquid extraction step. It measured parent CBD, 7-OH-CBD, and 7-COOH-CBD directly in plasma. The reported lower limits of quantification were 1.00 ng/mL for CBD, 0.50 ng/mL for 7-OH-CBD, and 5.00 ng/mL for 7-COOH-CBD. Samples were analyzed without a preliminary beta-glucuronidase hydrolysis step; thus the reported method should not casually be equated with a total-after-hydrolysis assay or a method covering every conjugated cannabinoid metabolite.

The parent compound and metabolites have different biological meanings. CBD documents exposure to cannabidiol in the sampled plasma compartment. 7-OH-CBD is a hydroxylated metabolite, often described in drug development as pharmacologically active. 7-COOH-CBD is a more extensively represented carboxylated metabolite in circulation. Its relatively large concentration does not make it a direct measurement of the same quantity as CBD, nor prove that its pharmacological effects are interchangeable with those of the parent. None of these three results, by itself, identifies THC consumption or proves an individual’s driving fitness.

The assay is a chemical measurement. Subsequent estimates—peak concentration, 24-hour exposure, apparent clearance, half-life, and metabolite-to-parent ratio—come from the concentration-time observations and pharmacokinetic calculations. The distinction between what the instrument assayed and what the model summarized should remain explicit in reports and testimony. For a broader introduction, see confirmatory drug testing and its interpretive limits.

Evidence table: the four observed dosing conditions

The table reproduces selected values from the original primary trial’s Table 2. AUC from 0 to 24 hours is a time-integrated concentration exposure, expressed in h·ng/mL; C_max is a geometric mean concentration in ng/mL. Median t_max represents time to peak concentration. Group-level figures cannot be substituted for any one person’s result.

The AUC and C_max entries are geometric means, while t_max is a median. Different column Ns reflect two withdrawals, not different independently randomized cohorts. The numerical differences here must be interpreted with the statistical comparisons below; a larger number in one column does not automatically prove a significant formulation effect.

Did the powder outperform the oil while fasting?

Not for the trial’s primary CBD exposure comparison at conventional statistical significance. The powder-to-oil geometric least-squares mean ratio for CBD AUC_0–24 was 115.3% (90% confidence interval 95.8%–138.6%; p=0.204). For CBD C_max, the corresponding ratio was 134.6% (p=0.065). Both were numerically greater for powder but did not meet the study’s specified 0.05 threshold. It would therefore overstate this experiment to claim that the powder definitively delivered more parent CBD exposure while fasting.

The metabolite findings were different. Compared with oil under fasting, powder produced approximately 49% greater 7-OH-CBD AUC_0–24 (ratio 149.2%; p<0.001) and approximately 58% greater 7-COOH-CBD AUC_0–24 (ratio 158.4%; p<0.001). The 7-OH-CBD C_max ratio was 184.5% and the 7-COOH-CBD C_max ratio was 168.9%, also with p<0.001. Those significant metabolite contrasts cannot be rewritten as a significant result for parent CBD.

The median time to parent-CBD maximum was 3.0 hours with powder versus 3.5 hours with oil (p=0.015). A half-hour difference in the median under these laboratory conditions may be scientifically interesting, but it does not establish the precise time of an unknown person’s earlier dose. An observed peak may be broad or multiple, and individual curves can differ markedly from the group median.

What happened after a high-fat breakfast?

Food was the strongest quantitative influence on parent-CBD exposure in the study. Relative to the same product taken fasting, CBD AUC_0–24 was approximately 301.4% for powder (three-fold; p<0.001) and 350.1% for oil (three-and-a-half-fold; p<0.001). CBD C_max was also higher with food: 209.2% of fasting for powder and 224.8% for oil, both p<0.001. These are model-adjusted ratios of geometric means, not changes observed identically in every person.

Under fed conditions the powder-to-oil CBD AUC_0–24 ratio was 99.3% (p=0.946), a result consistent with similar 24-hour exposure in the studied group. The powder-to-oil CBD C_max ratio was 125.3% (p=0.155), also nonsignificant. In contrast, the median CBD t_max was 5.0 hours after powder and 8.0 hours after oil (p<0.001). Thus an earlier median peak did not translate into statistically proven greater total parent-CBD 24-hour exposure.

When evaluating a claimed administration time, food status changes the interpretation of any hypothetical time-concentration relationship. An eight-hour median time to peak should not be misconstrued as a mandatory eight-hour delay for every oil user: the original observed oil-fed range extended from 1.5 to 23.6 hours. That breadth is itself a warning against reconstructing an individual ingestion time from a single result or from a simplified textbook curve.

Why did food increase CBD but reduce a metabolite in one formulation?

For the powdered product, fed-versus-fasted CBD AUC_0–24 increased about three-fold while 7-COOH-CBD AUC_0–24 fell to 73.6% of its fasting value (p<0.001). The powder’s 7-COOH-CBD peak was 61.4% of its fasting peak (p<0.001). Oil behaved differently: the corresponding fed-versus-fasted 7-COOH-CBD AUC ratio was 102.1% (p=0.781), with no significant difference. This is an instructive counterexample to the assumption that a parent drug and every downstream metabolite must rise or fall together after a change in absorption.

The primary authors discussed potential mechanisms, including altered first-pass metabolism, saturation or redirection of metabolic pathways, and delayed metabolite profiles not captured completely by 24-hour sampling. None was separately demonstrated as the unique causal mechanism in this human experiment. There was no randomized liver-enzyme intervention or direct isotope tracing to assign the observed shift to a particular enzyme. Likewise, a lower metabolite AUC does not by itself show that less parent CBD was consumed or that an exposure history was false.

The trial also defined summary combinations called “active drug exposure” (CBD plus 7-OH-CBD AUCs) and “total drug exposure” (CBD plus 7-OH-CBD plus 7-COOH-CBD AUCs). These are author-defined mathematical sums of molecular concentrations integrated over time; they are not independently measured biological effect scores and do not establish clinical potency equivalence across metabolites. In the fasted product comparison the powder’s total-sum AUC was greater, but it would be scientifically unsound to convert the numerical sum into a claim about sedation, seizure control, or driver impairment.

Evidence table: which differences were statistically supported?

The table summarizes prespecified or reported statistical comparisons from the original Table 3. Ratios compare geometric least-squares means unless noted. “Not significant” refers to the tested contrast, not proof that products or conditions are identical.

How large was the between-person variation?

Variation was substantial. Under fasting, the geometric coefficient of variation for parent CBD peak concentration was 73.5% with powder versus 117.1% with oil. Under fed conditions it was 61.4% with powder and 73.3% with oil. The 24-hour exposure geometric CVs were 70.0% versus 80.8% fasting and 35.1% versus 54.4% fed, powder versus oil respectively. These descriptive statistics suggest somewhat more consistent concentrations after powder, but the trial did not independently validate an individual prediction rule with a known error distribution for forensic reconstruction.

High interindividual variation means that a population average should not become a substitute for a verified person’s actual test history. The differences can also be affected by body composition, enzyme activity, gastrointestinal conditions, coadministered medications, and deviations from the studied schedule; this particular experiment did not determine all such contributions. Its food conditions were controlled and its participants screened, so real-world variability may differ in either direction.

For practical interpretation, compare results only when the laboratory has confirmed the same substance, specimen matrix, preparation method, reporting units, and sampling context. A measured metabolite-to-parent ratio is not a clock absent an appropriately validated model and individual facts. See the more general pharmacokinetics and ADME framework.

Why not simply calculate elimination half-life or time of ingestion?

The original authors calculated half-life and clearance parameters, but the 24-hour sampling window did not support a reliable terminal-phase estimate in many conditions. Their Table 2 footnotes show how few participant profiles allowed extrapolation of AUC to infinity: only three in the oil-fed condition contributed usable CBD AUC_inf estimates, despite 32 people receiving that treatment. The oil-fed 7-COOH-CBD half-life estimate rested on just one participant. A precise-looking table number can therefore have a substantially smaller analytical denominator than the study’s total enrollment.

Moreover, apparent oral clearance (CL/F) and apparent volume (Vz/F) combine disposition with unknown absolute bioavailability F. They are not directly measured organ clearances or distribution volumes independent of absorption. Using these composite variables to back-calculate a person’s dose from a lone plasma sample would demand assumptions the trial did not test. AUC_0–24 is better supported here than extensive conclusions about the terminal elimination phase.

Timing of food intake, formulation, interindividual variation, and sparse late measurements further frustrate reverse inference. The study does not provide a reliable person-specific rule for reconstructing a lost or undocumented administration time, even in a healthy adult who took CBD alone. Extending it to co-use of alcohol or THC, chronic administration, medical illness, or a collision scene adds further unsupported assumptions.

What did the safety observations actually establish?

The study recorded 32 treatment-emergent adverse events among 14 of the 32 exposed participants (44%). Most were mild. One moderate episode of drowsiness occurred about 15 minutes after powder under fed conditions and resolved within two hours. Headache was reported by six people; nausea and loose stools were also observed. No severe or serious adverse events, deaths, or withdrawals because of an adverse event were reported in this small study.

Those observations support a narrow short-term tolerability description at a single 400-mg CBD dose under supervised trial conditions. They do not demonstrate that CBD can never produce drowsiness, that either product is safe for every patient, or that participants performed tasks without impairment. Formal vehicle operation, divided attention, braking, and driving fitness were not outcomes. The absence of a clinically significant average ECG or vital-sign signal is also not a validated psychomotor safety finding.

The experiment was funded and sponsored by the developer of the tested powder formulation, and some authors had commercial affiliations. That does not invalidate laboratory results, but it makes the independent analytical details, prespecified endpoints, null comparisons, and participant denominators particularly important for critical evaluation. Investigators’ commercial interests are not a substitute for reviewing the original data and methods.

What does this study not prove in a forensic or litigation setting?

It does not prove that any detectable CBD concentration identifies a particular commercially sold product. It does not establish that CBD or 7-COOH-CBD in plasma equals THC in whole blood, or that higher CBD exposure necessarily produces stronger intoxication. It does not validate conversion from plasma ng/mL to whole-blood ng/mL or from group AUC to an individual time of administration. It does not quantify impairment probability, driving risk, pain relief, work fitness, mental state, witness reliability, or accident causation.

Its participants were healthy adults selected for a pharmacokinetic study. The investigated single dose was known, meal conditions were tightly standardized, and specimen collection times were predefined. A real forensic case commonly presents precisely the opposite: unknown product composition, uncertain dose, uncertain meal history, unknown chronic exposure, delayed specimen collection, and other substances. Those missing facts cannot be filled by assigning the trial’s group geometric mean to the individual.

Although plasma parent CBD and metabolites were chemically identified by LC–MS/MS, the corresponding behavioral effects were not determined. Chemical exposure is one evidentiary layer; actual performance and legal impairment require their own data. Careful forensic opinions should explicitly identify which layer was measured and which is hypothesized.

How should a laboratory report or expert review use the results?

A defensible interpretation begins with the precise question. If the issue is why two documented 400-mg CBD administrations produced different plasma CBD profiles, this original study provides experimentally relevant evidence that formulation and meal status can alter exposure. If the issue is whether a high plasma CBD or 7-COOH-CBD concentration independently establishes misconduct, THC use, incapacity, or a prior time of ingestion, this experiment does not answer it.

The case-specific analysis should establish the exact medication or product, dispensing and administration record, formulation, meal timing, specimen matrix and collection time, laboratory target analytes and quantification limits, calibration validity, sample integrity, and evidence of coexposures. A measured concentration should be reported with the appropriate physical units and analyte identity. Comparing a person’s result with trial averages requires a scientifically justified bridge; without one, qualitative discussion of plausible variability is more defensible than a calculated individual conclusion.

Where later exposure or clinical performance is disputed, the expert should distinguish original experimental findings, case records, patient report, assumptions, and inference. A product-name difference by itself is not a scientifically established difference in individual function. Equally, no statistically significant average formulation difference does not prove every participant’s profiles were identical.

Frequently asked questions about food, formulations, and CBD testing

1. Did the same 400-mg dose produce different CBD plasma exposure after food?

Yes, at the group level. Compared with fasting, fed CBD AUC_0–24 was approximately 3.01-fold after the powder and 3.50-fold after the oil, with both contrasts statistically significant in the trial. It was an 800–1,000-kcal high-fat breakfast, not a study of every possible meal.

2. Was powdered CBD significantly better absorbed than the oil while fasting?

For the primary parent CBD AUC_0–24 comparison, no. The powder-to-oil ratio was 115.3%, p=0.204. However, both measured metabolites had significantly higher 24-hour exposure after powder while fasting, and the median parent-CBD peak occurred half an hour earlier.

3. Were CBD and its metabolites assayed in whole blood?

No. Venous blood was collected and centrifuged, and the study’s reported chemical values were for separated plasma. The results should not be labeled as directly measured forensic whole-blood cannabinoid concentrations or numerically converted across matrices without a validated procedure.

4. Can 7-COOH-CBD concentration identify the amount or time of CBD taken?

Not from this trial alone. The metabolite changed differently from parent CBD depending on food and formulation, and elimination characterization was constrained by 24-hour sampling. No individually validated concentration-to-dose or metabolite-to-time inference rule was established.

5. Why could more CBD accompany less 7-COOH-CBD after a meal?

For the powdered product, fed CBD AUC increased while 7-COOH-CBD AUC decreased. The paper considers changes in absorption, metabolism, and delayed metabolite appearance, but it did not prove any one mechanism. It demonstrates why a parent molecule and a downstream metabolite need not track each other proportionally.

6. Did the trial establish that higher CBD exposure causes driver impairment?

No. Its outcomes were plasma pharmacokinetic parameters and general adverse-event monitoring, not actual road driving, sobriety testing, or a validated impairment classifier. One brief drowsiness episode was observed, but that finding cannot be transformed into an impairment threshold or a guarantee of safety.

7. Does a nonsignificant formulation comparison mean the products are interchangeable for every person?

No. Lack of significance for a group comparison is not proof of individual equivalence. Absorption timing, metabolite exposure, and variability differed, the enrolled population was small, and some extrapolated PK metrics rested on very few participants. Clinical interchangeability, therapeutic efficacy, and forensic case interpretation are separate questions.

Bottom line for evidence-based CBD interpretation

This 32-person controlled pharmacokinetic study shows that food can markedly increase measured plasma parent-CBD exposure while formulation can alter absorption timing and metabolite profiles. Its strongest forensic lesson is the separation of an administered milligram dose, a directly assayed plasma concentration, an estimated pharmacokinetic metric, and an individual’s observed behavior. The first three cannot, without additional validated evidence, establish the fourth or determine what caused a particular event.

Primary study: Bendik I, Beck M, Manderna A, et al. Comparative Pharmacokinetics and Safety of Cannabidiol in a Powder Formulation, CBtru, vs an Oil-Based Formulation, Epidyolex, Under Fasted and Fed Conditions in Healthy Participants: A Randomized Open-Label Cross-Over Phase I Study. CNS Drugs. 2026;40:709–723. https://doi.org/10.1007/s40263-026-01280-1.