Can Medicinal THC Slow Complex Trail Making While Reaction Time Remains Unchanged? Two Studies of 78 Patients

Answer: Yes, in two within-person studies of adults prescribed cannabis containing delta-9-tetrahydrocannabinol (THC), performance became slower on the more complex Trail Making Test B after a usual prescribed dose, even though computerized simple and five-choice reaction-time tests did not show statistically significant deterioration. The oil study included 41 people and the vaporized-flower study included 37, for 78 distinct participants. This is evidence of selective changes on particular laboratory tasks—not proof that every patient was impaired while driving, or that a negative result on one task establishes unimpaired performance on another.

Stefanidis and colleagues reported these findings in an original 2025 study in Drug and Alcohol Review, DOI: 10.1111/dar.14060. Their data are especially relevant when a forensic opinion relies too heavily on either a single normal reaction-time result or a single positive THC screening result. Neither measurement answers every functional question. The source is a study of cognitive testing under specified research conditions; it is not a validation study of a roadside impairment test.

The important result is a difference between tasks, not a universal cognitive deficit

Both participant groups completed testing in a baseline condition and again after their prescribed medication. The investigators counterbalanced the order of those conditions and generally scheduled the appointments about one week apart. Study 1 assessed 41 patients 90 minutes after one dose of prescribed cannabis oil. Study 2 assessed 37 patients 15 minutes after vaporizing prescribed flower. Thus, the comparison reflects separate medication routes, distinct testing delays and two different groups—not a randomized head-to-head test proving that one route is more impairing than the other.

Mean Trail Making B completion time increased from 54.96 to 71.15 seconds after oil (p < 0.001) and from 63.57 to 86.15 seconds after vaporized flower (p < 0.001). Larger completion times signify slower performance. By contrast, the corresponding computerized simple reaction-time means changed from 328.99 to 326.05 milliseconds (oil, p = 0.461) and from 340.15 to 332.28 milliseconds (flower, p = 0.197). Those latter comparisons did not reach statistical significance; they must not be characterized as demonstrated improvement or proof of preserved driving ability.

Who actually participated in these two studies?

Participants were adults living in Queensland, Australia, who held prescriptions for either oil containing THC or cannabis flower. They had current driver licenses and drove at least weekly. Study 2 did not enroll anyone who participated in Study 1. The study was therefore composed of two nonoverlapping samples, not 78 people randomized between oil and vaporization. The oil group averaged approximately 46 years of age; the flower group averaged approximately 43. Most participants reported frequent or longstanding cannabis use, making this primarily evidence about a regular-use medicinal population rather than cannabis-naive patients.

Study 1 analyzed 41 participants. Study 2 recruited 38 but removed one extreme statistical outlier whose inclusion affected the reported analysis, leaving 37. Three possible extreme outliers in Study 1 were examined in a sensitivity analysis and retained because excluding them did not change the key findings. The difference in handling outliers matters when applying group-level results to an unrelated defendant, employee or patient.

Medical conditions, prescription drugs, prior illicit use and symptom burden also varied. Those real-world features improve relevance to treated patients but limit how narrowly the findings can be attributed to THC alone. There was no untreated placebo arm and no blind administration of an inactive product. Counterbalanced sessions mitigate some order effects without providing the experimental control of a blinded placebo crossover trial.

What exactly was administered—and was exposure measured in blood?

In the oil condition, participants used their own prescribed product, mainly sublingually: 38 of 41 placed it under the tongue, while three used an oral method. The mean documented THC dose was 10.80 milligrams, with a standard deviation of 11.95 milligrams and a broad 0.06–50 mg range. The mean cannabidiol amount was 16.05 mg, but values varied widely, including products with no CBD. Oil dosing was visually observed against prescriptions; the study did not independently assay each administered dose or characterize its bioavailability by serial blood sampling.

In the flower condition, participants weighed a mean 0.22 grams of prescribed cannabis and inhaled vapor generated in a vaporizer, using between one-half and four balloons. Based on the flower’s stated strength and loaded mass, investigators estimated a mean 48.49 mg THC in the weighed material (SD 33.33 mg; range 10.20–190 mg). That number is not a measured inhaled or systemically absorbed THC dose. Vaporizer temperature, balloon consumption, inhalation technique and losses mean the material’s nominal THC content cannot be equated with the delivered amount.

Neither cohort contributed quantitative venous whole-blood THC concentrations, serum THC concentrations or a validated blood-versus-performance regression in this paper. The authors used a qualitative oral-fluid screening instrument before sessions, not a reported confirmatory blood assay. The administered mass of THC, the presence of an oral-fluid screen signal and the seconds needed to complete a cognitive test are three separate quantities.

Evidence table 1: design, administered products and screening context

These baseline oral-fluid results are easy to misread. “Positive at baseline” did not mean a laboratory-confirmed current blood concentration, and it did not mean the person had consumed cannabis just before arriving. Participants were instructed to abstain for at least 11.5 hours before their visits; a screening indication may persist in frequent users after the immediate dosing episode. The paper records the screening result but does not use it to derive individual BAC-equivalent or THC impairment thresholds.

Why Trail Making Test B matters more than an ordinary reaction-time number here

Trail Making A asks the participant to connect numbered targets in ascending order. Trail Making B requires alternating between numerical and alphabetical sequences. Completing B entails visual search, motor coordination, sequencing and switching between rules, and is timed in seconds. An increase in B time can therefore arise from several interacting cognitive and visuomotor demands. The test does not isolate one molecular mechanism or identify a particular drug concentration.

Study 1 observed an average B-time increase of 16.19 seconds after oil. Study 2 observed an average increase of 22.58 seconds after vaporized flower. Both had p < 0.001 and reported large effect sizes (Cohen’s d 0.99 and 0.91 respectively). The authors also calculated a normalized Trail Making ratio, (B time minus A time) divided by A time, to examine B performance relative to the simpler component. That ratio increased in the oil group from 1.43 to 1.96 with p < 0.001 and d 0.61.

Although the flower group’s ratio increased from 1.81 to 2.30 with a displayed p = 0.017, the original Table 6 did not mark that comparison significant after the authors’ correction for multiple testing. A responsible article must preserve that distinction. The two groups should not be described as having identical significant changes across every ratio and test.

Evidence table 2: actual cognitive observations and study-specific nulls

The Rapid Visual Processing task also failed to show a statistically significant change in either sample on its reported mean latency and sensitivity outcomes. Study 1’s rapid visual processing data were based on 40 rather than all 41 oil participants. This detail matters when a report treats all tests as if they share a single universal denominator. The table emphasizes effect-specific outcomes instead of assigning a blanket label of “impaired” or “not impaired” to all 78 people.

Did the effects survive checks for other drug screening indications?

Yes for the central Trail Making B result. The investigators conducted sensitivity analyses excluding participants who screened positive for another drug class at either assessment: seven in the oil study and eight in the flower study. In the oil group, the key Trail Making B and ratio findings remained; the paper did not identify significant changes on its other measures. In the flower group, Trail Making A and B remained significant, but the other comparisons did not survive that exclusion. This is particularly important when citing the full-sample Stroop C result: the stronger claim that Stroop C consistently worsened after addressing other detected drugs is not supported by the sensitivity analyses.

These exclusions did not transform screening results into a quantitative toxicology investigation. A positive screen may not establish a drug’s dose, timing or clinically meaningful activity, and a negative screen cannot rule out every other medication or physiologic cause of variation. The investigators also observed that some participants had prescriptions for opioids, antidepressants or anxiolytics, while histories of mood symptoms and chronic conditions were common. These factors warrant explicit review in any medicolegal use of the findings.

Does a positive THC oral-fluid result predict which participant slowed?

Not in the way an individualized impairment threshold would require. The investigators divided participants by their THC screening status at baseline. Among oil users, Trail Making B remained significantly slower after dosing both for baseline-screen-negative participants (22 people; d 1.3, p < 0.001) and for screen-positive participants (19 people; d 0.71, p = 0.009). Among flower users, the same test remained significant in those initially screen-positive (25 people; p < 0.001) and screen-negative (12 people; p = 0.002), with large effect sizes reported for both subgroups.

These analyses show that the direction of the observed group-level Trail Making result did not depend entirely on whether the screening instrument already indicated THC before dosing. They do not provide a validated sensitivity, specificity, positive predictive value, or cutoff for predicting an individual’s performance from a screen. The study lacks confirmatory oral-fluid concentrations, controlled blood concentrations and incident-level observations necessary for that extension.

Why the nonsignificant tests cannot be treated as proof of no effect

A nonsignificant paired comparison means this experiment did not establish a reliable mean difference on that outcome under its analysis. It is not automatically an equivalence test demonstrating that the test was insensitive to THC or that each participant was unaffected. A finite group, variability between people, measurement precision, practice effects and timing all influence the ability to identify changes. The paper did not define equivalence margins or perform a clinical noninferiority trial for driving fitness.

Moreover, the cognitive measures do not assess identical abilities. A computer-based simple reaction-time task measures performance on a relatively constrained response to a stimulus; Trail Making B requires alternating sequences under motor and visual search demands. Normal group results on the simpler task may coexist with slower mean performance on the more complex task without contradiction. Conversely, a poorer Trail Making score cannot by itself reveal whether the change resulted primarily from visual scanning, motor speed, set-shifting, fatigue, anxiety, illness or THC pharmacodynamics.

Can this work validate an officer’s roadside sobriety test or a driving simulator?

No. The researchers administered a short neuropsychological battery in laboratory sessions; they did not directly evaluate on-road lane control, collision avoidance, hazard response, vehicle speed regulation or standardized field sobriety test classification accuracy in this paper. A cognitive test could be associated with driving-relevant functions without becoming a validated predictor of unsafe driving. Before translating a Trail Making time into a legal conclusion, the method would need appropriately designed validation against the actual legal or functional criterion of interest.

The study likewise does not establish a permissible THC blood concentration, zero-tolerance policy threshold, roadside test cutoff or universal post-dose waiting period. Its two testing delays—90 minutes for oil and 15 minutes for flower—were prespecified points informed by prior pharmacology, not a time-course proving recovery at any later hour. The approximate 20-minute battery itself captures a window of observation rather than the instant at which blood or oral-fluid samples would have been taken during an actual traffic stop.

How much does long-term medicinal use change the interpretation?

The participants’ longstanding, often daily cannabis exposure is essential context. Study 1 participants reported roughly 15 years of cannabis-use history on average, and the flower sample approximately 23 years. The investigators considered possible tolerance and symptom relief when discussing why some reaction-time, Stroop and sustained-attention measures did not show robust changes. The observed pattern is compatible with selective tolerance, but the protocol did not randomly assign new versus chronic users or directly measure a person’s pharmacodynamic tolerance threshold.

Regular exposure should therefore neither be used as a categorical guarantee of unimpaired functioning nor automatically ignored. A case review should separate self-reported consumption history from actual dose, product composition, prescribed schedule, other drug use, cognitive baseline, clinical symptoms and behavior at the relevant time. This study is especially instructive precisely because its volunteers were prescribed users rather than an exclusively occasional-use sample. Its population specificity is a strength and an external-validity boundary at the same time.

What measurement limits matter in forensic toxicology?

First, identify the biological matrix and the analytical question. The Dräger DrugCheck 3000 was used as an oral-fluid screening tool. An oral-fluid screening signal is a categorical outcome at that device’s operating conditions; it is not a laboratory-confirmed whole-blood THC concentration. This paper contains no reported quantitative THC assay in blood, plasma or serum, so it cannot supply a person-specific relationship between those concentrations and Trail Making time. A guide to confirmatory drug testing explains why initial indications and definitive analyte identification answer different questions.

Second, distinguish the amount of drug placed in a dosing device from exposure reaching tissues. Milligrams in oil or weighed flower are administered-product descriptions, not proof of the bioavailable mass delivered to the brain. Third, distinguish directly observed seconds or milliseconds in a cognitive test from physiological or legal categories. A measured increase in completion time is an observed task outcome; it does not directly measure arterial THC, synaptic activity, intent, accident causation or inability to drive.

Evidence from different specimen types also needs its own framework. The site’s specimens and biological matrices guide explains why physical matrices are not interchangeable. The specimen-collection guide highlights timing, handling and chain-of-custody questions relevant to any actual case evidence. Neither guide substitutes for the laboratory record of the person at issue.

What would an opposing expert need to disclose?

An expert relying on the Stefanidis paper should identify which cohort is being cited, its N, the drug administration route, the elapsed testing delay, the actual end point and whether the statistic came from the full sample or a sensitivity subgroup. Where discussing Trail Making B, the expert should report both group means and the nature of the paired test. Where discussing the flower Stroop result, the expert should acknowledge its failure to persist after excluding other-drug-screen-positive participants. Where discussing the flower Trail Making ratio, the original footnote about multiplicity correction should be preserved.

A reliable opinion also has to disclose the missing bridge between the research and any case: whether there is a comparable cognitive assessment, a documented patient’s baseline, a quantitative and properly identified physical specimen, known dosing history, other substances, injury or illness, and actual behavior relevant to the event. Without those data, the experiment supports a research-level proposition about differential test sensitivity, not a definitive retrospective diagnosis of impairment. The question is whether the scientific evidence addresses the precise inference being offered, rather than whether a source contains the word “cannabis.”

Seven frequently asked questions

1. Does a normal simple reaction-time result mean medicinal THC had no cognitive effect?

No. The two groups showed no statistically significant mean deterioration on simple reaction time, while both had significantly slower mean Trail Making B completion times. The tasks measure different combinations of processing and response demands.

2. Did the researchers show that oil impairs less than vaporized flower?

No. Separate patient groups, different product formulations, distinct dosing methods, different testing delays and uncertain absorbed vapor dose prevent a causal comparison of routes based on these group means.

3. Is the 48.49 mg flower value a measured inhaled THC dose?

No. It is an estimate based on the amount and labeled THC strength of weighed flower. The study expressly recognized that vaporization and inhalation introduce substantial variation in actual delivery and absorption.

4. Did everyone with a positive oral-fluid THC screen perform poorly?

No such person-by-person classification can be drawn. The study examined paired mean test changes and baseline-screen-status subgroup comparisons; it did not validate oral-fluid screening as an individualized predictor of impairment.

5. Was the flower group’s Trail Making ratio significantly changed after correction?

The table reports a nominal p = 0.017, but does not identify that flower-ratio result as significant after the authors’ Bonferroni–Holm correction. The oil group’s ratio did meet the corrected criterion.

6. Does the study establish how long a medicinal-cannabis patient should wait before driving?

No. Each study tested at one specified post-administration interval and did not determine a recovery trajectory or a validated driving-safety endpoint. No universal waiting-time rule follows from the experiment.

7. Can slower Trail Making B alone prove THC caused impairment in a particular collision?

No. It can be relevant laboratory evidence, but causation in a specific case requires evidence of individual exposure, medical and cognitive baseline, potentially confounding factors, test administration conditions and observed event-related behavior.

Bottom line for scientific testimony

The defensible answer is narrower and more useful than either “tolerant patients cannot be impaired” or “any THC-positive screen proves impairment.” Among 41 prescribed oil users and 37 prescribed flower users, the strongest reproducible study-specific finding was slower performance on a complex Trail Making B task after dosing, while several simpler cognitive outcomes were not significantly altered. The source does not supply an individual impairment cutoff, a forensic whole-blood concentration, proof of driving incapacity or a case-causation rule. It demonstrates why forensic opinions must distinguish the substance tested, the biological matrix, the outcome measured and the inference actually supported.

Original primary research: Stefanidis KB, Schiemer C, Mieran T, Hill A, Horswill MS, Summers MJ. “Elucidating the acute effects of medically prescribed oral and vaporised delta-9-tetrahydrocannabinol on cognitive functions important for driving.” Drug and Alcohol Review. 2025;44(4):1010–1023. https://doi.org/10.1111/dar.14060.