Can Atomoxetine Reduce Dextroamphetamine’s Effects? What a 10-Person Crossover Actually Measured

Short answer: In a small blinded crossover experiment, four days of atomoxetine pretreatment reduced some blood-pressure, plasma-cortisol and self-reported responses to one oral dose of dextroamphetamine. The researchers studied ten healthy adults, not patients with an intoxication diagnosis or impaired drivers. They did not chemically measure dextroamphetamine or atomoxetine concentrations in forensic whole blood. A lower rating of feeling “high,” or a smaller physiological response, cannot establish that a person had a lower stimulant concentration or was unimpaired.

The primary study was Sofuoglu, Hill, Kosten and Poling, Atomoxetine Attenuates Dextroamphetamine Effects in Humans, published in 2009 in The American Journal of Drug and Alcohol Abuse, 35(6):412–416. Its original research report is available at the primary study DOI. The findings bear on noradrenergic modulation of a controlled stimulant challenge, not on retrospective reconstruction of a particular person’s conduct.

What is the answer a forensic reader should take away?

Atomoxetine pretreatment changed several responses measured after dextroamphetamine exposure. The clearest reported comparisons concerned systolic and diastolic blood pressure, the participants’ ratings of stimulation, drug-related pleasure and feeling high, and cortisol measured from separated plasma. The article reports statistical differences between pretreatment periods, but does not supply a quantitative whole-blood stimulant concentration series or a validated personal impairment classification. Those omissions limit what a toxicologist can infer from the experiment.

Reduced response is not synonymous with absence of pharmacological activity. Both laboratory periods included an active oral dextroamphetamine challenge. The question tested was whether pretreatment changed the response, not whether atomoxetine prevented intoxication, improved safe driving, lowered a drug assay result or protected against every cardiovascular event. A legal opinion should preserve those distinctions even when the observed group differences are statistically significant.

Who was actually studied?

Ten healthy adult volunteers completed both study conditions and formed the analysis sample. Three were men and seven were women; the reported mean age was 29.1 years, with standard deviation 6.4 years. Three additional recruits did not complete participation—one because an intravenous line could not be maintained and two because cocaine was detected on screening during the study. Accordingly, the physiological and questionnaire statistics should not be described as coming from thirteen completed participants.

The researchers screened for major medical and psychiatric conditions, a substance-use-disorder history and substantial daily smoking or caffeine consumption. Excluding people with high nicotine or caffeine intake aimed to reduce withdrawal-related confounding. It also narrows generalizability: this was not an observational sample of people using multiple street drugs, individuals with cardiovascular disease, or persons receiving emergency care after a collision. Ten people each experienced two laboratory periods; twenty periods are not twenty independent adults.

How did the crossover design work?

Each participant received two four-day pretreatment periods in a counterbalanced sequence. In one, the participant took atomoxetine at 40 mg once daily; in the other, a matching placebo was given. The investigators separated periods by at least four days. On the fourth day of each period they conducted a roughly six-hour laboratory visit with supervised medication administration, serial measurements and a standardized dextroamphetamine challenge.

The oral stimulant dose was specified as 20 mg per 70 kg body weight, a weight-adjusted nominal dose. Atomoxetine treatment was a fixed daily dose, not an individually titrated chronic regimen. These are administered amounts, not concentrations that the investigators later established in venous whole blood. The crossover design helps control stable between-person differences, but it does not eliminate order effects, expectation, incomplete washout or all consequences of the experiment’s other tasks.

What exactly happened before the stimulant challenge?

Participants received the day’s atomoxetine or placebo dose and a light meal. A psychological stress procedure followed about an hour after pretreatment, then a physical stress procedure approximately thirty minutes later. Roughly two hours after the pretreatment dose, baseline subjective, cardiovascular and cognitive measures were collected; then the participants took oral dextroamphetamine. The article places the detailed analyses of those stress challenges elsewhere. Thus, the experiment’s outcomes reflect a specific laboratory sequence rather than a clean test of stimulant exposure in a stress-free environment.

The participant’s prior stress responses could affect later cortisol or cardiovascular reactivity, even though the crossover intended to hold the procedural sequence constant across treatments. Readers should not silently remove those procedures when describing the science. Likewise, a measured response during the following three to four hours is not evidence that the same participant would show an identical response at another time of day, after repeated chronic use, in illness, or following a different route of stimulant exposure.

Study denominators, administered doses and sampling windows

The sampling schedule matters because the authors compared maximum response above a prechallenge baseline for most subjective and cardiovascular outcomes. For cortisol, the statistical analysis expressly incorporated serial time points. These are not interchangeable statistical summaries: a peak-minus-baseline number discards some temporal information, while a time-by-treatment interaction asks whether the profiles differed over the sampled interval.

What changed in blood pressure?

The report describes significantly smaller systolic and diastolic blood-pressure increases in the atomoxetine condition after the same oral dextroamphetamine challenge. The systolic comparison was reported as F(1,18)=8.8 with p<.01. The diastolic difference was likewise reported p<.01. The full original text prints an unusually large diastolic F statistic; the direction and p threshold should be distinguished from certainty about that numerical transcription, especially without the original graphic data table available in the supplied XML.

These observations concern changes in measured blood pressure. A cuff reading is neither a chemical drug assay nor an independent proof of behavioral impairment. A blunted increase might coexist with clinically relevant tachycardia, anxiety, preexisting hypertension or other symptoms not addressed by the reported treatment contrast. The article was too small and short to evaluate uncommon adverse events. It would overstate the research to call atomoxetine universally protective simply because average blood-pressure response was smaller under one four-day pretreatment protocol.

Which subjective drug experiences differed?

The Drug Effects Questionnaire asked participants to rate sensations on a 0–100 scale. Compared with placebo pretreatment, atomoxetine lowered changes in the ratings of feeling stimulated, feeling high and experiencing good drug effects. The authors’ reported test statistics for those three comparisons were F(1,9)=5.9, 5.4 and 5.3, respectively, each with p<.05. Those labels are responses to a questionnaire, not measured blood concentrations or objective assessments of whether a motorist can maintain lane position.

Other questionnaire items included anxiety, sedation, negative feelings, drug strength, liking and wanting more. The source did not establish a comparably significant treatment reduction for every possible sensation. Moreover, the Profile of Mood States did not show a statistically significant pretreatment effect in the reported analysis. One should not collapse all drug-related sensations into a single claim that the drug had “no effect” or that atomoxetine blocked euphoria for every participant.

What did the plasma cortisol result establish?

The investigators collected blood through an indwelling intravenous catheter and assessed cortisol concentration in plasma. Their repeated-measures analysis reported a treatment effect, a time effect, and a treatment-by-time interaction, each at p<.05. Plasma cortisol increased more under placebo pretreatment than under atomoxetine pretreatment at the sampled postchallenge times. This is an endocrine response measured in the plasma compartment, not a direct measurement of dextroamphetamine parent compound.

A plasma cortisol assay can respond to stress, circadian physiology, medication effects and the design of the laboratory visit. It cannot be converted into an amphetamine concentration by labeling both as “blood tests.” Nor does a cortisol difference itself identify norepinephrine release at a particular synapse. The interpretation offered by the authors is consistent with a noradrenergic role but is not a complete molecular-mechanism experiment.

What were the original numerical statistical contrasts?

The reported p thresholds are the original publication’s inferential summaries, not independently recomputed statistics. No exact p-value should be invented from an F statistic when the original’s analysis assumptions or numerator and denominator degrees of freedom are not independently reproducible from participant-level records. These multiple questionnaire, cardiovascular and endocrine comparisons were tested within a small sample; the paper does not establish a modern familywise multiplicity-corrected discovery set. A forensic explanation should say “reported p<.05” or “reported no significant difference,” not imply a universal biological rule.

Did the investigators chemically measure dextroamphetamine in whole blood?

No. The study measured plasma cortisol, repeated cuff blood pressure, pulse and self-reported psychological effects. The intravenous catheter was used for blood sampling, but the publication does not report a validated quantitative assay of dextroamphetamine or atomoxetine parent drug, an analytical calibration range for either drug in a forensic blood specimen, or contemporaneous drug concentrations in separated serum or plasma. Clinical urine toxicology during screening served eligibility and exclusion purposes, not a serial confirmatory challenge-day dextroamphetamine assay.

The biologically material difference is that cortisol is a hormone quantified in separated plasma. A legally collected forensic venous whole-blood specimen contains erythrocytes as well as liquid components and may require distinct specimen handling, validation and reporting conventions. One cannot translate a hormone observation into ng/mL of amphetamine. For a broader explanation, see the site’s independently published guide to biological matrices in toxicology.

Did reduced subjective effects mean reduced drug exposure?

Not as an experimentally established conclusion. The same nominal oral challenge was administered during each pretreatment phase, but absorption, systemic exposure, tissue distribution and elimination were not independently quantified as parent-drug concentration–time curves. A change in receptor signaling, autonomic adaptation or expectations could influence a response despite similar concentrations. Conversely, a pharmacokinetic interaction might theoretically affect exposure; this source simply does not contain the assays necessary to evaluate that mechanism.

That distinction matters when an expert is asked whether a coadministered medication made a blood finding harmless. This study does not supply paired whole-blood amphetamine values and objective impairment outcomes from which to establish such a rule. Lower “high” ratings are not a negative confirmatory drug test. The interpretive gap also runs the other direction: a positive blood amphetamine assay is not by itself a validated measure of exactly how stimulated a particular person felt.

What can be said about norepinephrine mechanisms?

Atomoxetine primarily inhibits the norepinephrine transporter, while dextroamphetamine affects catecholamine signaling through other processes. The authors used atomoxetine pretreatment as a pharmacological probe to ask whether changing noradrenergic tone would modify responses to a stimulant challenge. The observed attenuation was consistent with such involvement, particularly for autonomic, endocrine and some subjective endpoints.

It does not follow that the experiment directly measured norepinephrine released within the brain, receptor occupancy, dopamine transporter function or a complete cause-and-effect chain from transporter to behavior. Participants did not undergo a brain microdialysis experiment, and the source contained no cerebrospinal-fluid catecholamine assay. Proposed adaptation involving inhibitory autoreceptors or downstream receptors is a mechanistic explanation, not something directly confirmed in each volunteer by this protocol.

Why are the negative and unmeasured outcomes important?

A study that reports three significant subjective questionnaire items is not automatically positive on all other items. The source specifically reported no treatment effect on its broad Profile of Mood States outcome. It did not report a definitive significant result for every heart-rate or DEQ item, and omitted comprehensive exact p-values for many comparisons. That does not prove precise equivalence for those outcomes: with only ten participants, estimates may be imprecise, and absent published tests cannot be converted into numerical null p-values.

More critically for legal applications, the investigators did not administer an on-road driving test, a standardized alcohol or drug recognition examination, a reaction-time battery validated for individual impairment, or a longitudinal substance-use-disorder treatment trial. No individual classifier, specificity, sensitivity, blood-drug legal threshold or forensic causation model was validated. The missing measurements are not defects that can be cured by an attractive pharmacological narrative.

What study-design limits affect generalizability?

The analysis includes ten healthy adults, two four-day pretreatment conditions and one controlled stimulant challenge per condition. Results might differ with age, repeated long-term atomoxetine, established stimulant tolerance, other psychoactive substances, illness or different amphetamine formulations. The washout was at least four days, and crossover order was included in the statistical model, but neither feature creates evidence about every possible carryover or drug interaction. A short laboratory challenge also cannot evaluate whether atomoxetine meaningfully treats methamphetamine-use disorder.

The absence of a dextroamphetamine-placebo challenge is particularly consequential. Because dextroamphetamine was given in both phases, investigators compared modulation of its responses rather than measuring the full interaction against an unstimulated control. The intervening stress tasks likewise limit clean separation of medication and laboratory-context effects on cortisol. These issues should temper clinical extrapolation without erasing the observed within-person statistical contrasts.

How should an expert separate subjective, cardiovascular and analytical evidence?

Begin by identifying what produced each number. A subjective questionnaire records the participant’s rating. A sphygmomanometer reports cardiovascular physiology. A cortisol assay quantifies a hormone in plasma. A specific chromatographic drug assay would quantify an identified chemical in a validated matrix, but this study does not provide that analysis. The scientific meaning of a positive result depends on the method, specimen, comparator and question, not merely on the fact that all results arise in the context of the same administered drug.

For comparison with actual drug confirmation in a disputed specimen, the site’s published confirmatory drug-testing guide explains why identity, specimen and analytical validation matter. In an individual case, an expert would need the collection record, quantitative method, calibration and control data, relevant medication history, time intervals, signs and behavior. None of those person-specific evidentiary items is generated automatically by citing an experimental mean from ten volunteers.

Can these results be applied to impaired-driving litigation?

They can inform a narrowly framed explanation that coadministered medications may alter physiological or subjective responses to a stimulant. They cannot determine whether an identified driver was intoxicated, prove that a lawfully collected sample contained less amphetamine, or rule out clinically important effects at a specific time. The laboratory volunteers were not observed driving, and there is no reported collision or legally controlled roadside examination. Even objective blood-pressure observations were not validated as diagnostic indicators of driving impairment.

One should also resist an opposite overclaim: a significant group difference does not prove a given driver on atomoxetine would feel less impaired than that same driver without it. Individual pharmacokinetic differences, tolerance, other medications, task demand and elapsed time were not resolved into a validated predictive formula. The appropriate forensic product is a bounded interpretation of the source, not an individualized verdict made from the paper alone.

Does the experiment show that atomoxetine treats stimulant addiction?

No clinical treatment claim can be established from this experiment. These were healthy volunteers without substance-use-disorder histories, exposed to a short pretreatment and a controlled dextroamphetamine dose. Lower self-rated pleasant effects were considered potentially relevant to future treatment development, but the investigators did not test retention in therapy, relapse prevention, abstinence, real-world craving reduction or long-term clinical outcomes. The absence of those endpoints is especially important when a reader sees the phrase “therapeutic utility” in a research discussion.

Similarly, the blood-pressure comparison is not a clinical safety trial of concurrent atomoxetine and illicit stimulant use. A short reduction in a mean laboratory response should not substitute for individualized prescribing, medical monitoring or emergency evaluation. The claim warranted by the original design is narrower: under this protocol, atomoxetine pretreatment attenuated selected responses following a common oral dextroamphetamine challenge.

Frequently asked questions

1. Did the researchers administer methamphetamine?

No. The controlled challenge was oral dextroamphetamine, the dextrorotatory isomer of amphetamine, at 20 mg per 70 kg of body weight. Methamphetamine is a distinct chemical, and the results should not be silently relabeled as controlled methamphetamine administration. The study’s references to methamphetamine in its broader rationale do not change which drug the participants actually received.

2. How many people produced the measured responses?

Ten healthy adults completed the crossover and were analyzed. Three additional recruits were not part of the complete analysis. Each participant supplied observations during two pretreatment periods and repeated measurements after a challenge; those observations are correlated within the same biological person.

3. Was the drug concentration lower with atomoxetine?

The paper did not provide quantitative dextroamphetamine blood, plasma or serum concentrations, so it cannot answer that chemical-exposure question. The reported significant differences concerned specific cardiovascular, hormone and subjective outcomes, not a measured lower parent-drug concentration.

4. Was cortisol measured in whole blood?

No. Cortisol was measured in plasma collected from blood samples. That matrix and analyte differ from a forensic whole-blood amphetamine assay. The cortisol result is useful for understanding a stress-hormone response, not for calculating a personal amphetamine concentration.

5. Did every questionnaire response show an effect?

No. The study identified lower “stimulated,” “high” and “good drug effects” ratings with atomoxetine pretreatment, while it did not find a significant Profile of Mood States treatment effect. Results for other individual questionnaire items were not established as universally significant; one should not manufacture missing p-values.

6. Does this paper prove that a driver taking atomoxetine is unimpaired after amphetamine?

No. The trial involved no actual driving performance, independently validated individual impairment threshold or quantitative paired whole-blood drug results. A small group difference in selected physiological or subjective responses cannot decide fitness to drive or causation in an individual incident.

7. What additional evidence would matter in an actual forensic case?

The precise identified drug and formulation, dose history, timing, validated specimen type, collection and storage details, analytical confirmation and quantitative uncertainty would matter, as would documented behavior and plausible alternative explanations. If the question concerns a motor-vehicle event, independent contemporaneous performance and case facts are required; a hormone or questionnaire result from another study cannot replace them.

Primary scientific source and the bottom line

Sofuoglu M, Hill K, Kosten T, Poling J. Atomoxetine Attenuates Dextroamphetamine Effects in Humans. American Journal of Drug and Alcohol Abuse. 2009;35(6):412–416. Read the original research publication. This explanation distinguishes what the researchers measured from clinical and forensic claims that require other data.

Bottom line: Four-day atomoxetine pretreatment was associated with smaller selected cardiovascular, subjective and plasma-cortisol responses to a common oral d-amphetamine challenge in ten healthy adults. It was not a study of forensic whole-blood stimulant concentration, individual legal impairment, driver safety, or case-event causation.