Did 400 mg of Modafinil Reduce Methamphetamine Use? Why Urine Tests and Self-Reports Disagreed in a Seven-Person Pilot

Answer: In a small, uncontrolled outpatient study of modafinil for methamphetamine dependence, participants reported less methamphetamine use over time, but the proportion of amphetamine-positive urine screens did not significantly decline. The authors enrolled eight people, seven completed intake, and only four completed the study. Modafinil-related tolerability and mood measures were exploratory, not proof of effective treatment. The study did not measure a quantitative methamphetamine concentration in human venous whole blood, nor did it validate any individual impairment, last-use, or event-causation inference.

The original primary study by McGaugh and colleagues, Open Label Pilot Study of Modafinil for Methamphetamine Dependence (2009), supplies a useful lesson in how different endpoints answer different questions. It also shows why a safety-and-feasibility pilot should not be mistaken for a placebo-controlled effectiveness trial. The primary research is identified by DOI 10.1097/JCP.0b013e3181b591e0.

The scientifically defensible answer is narrower than a treatment claim

Participants receiving prescribed modafinil, behavioral therapy, attendance-related incentives, and repeated clinic assessments reported decreasing methamphetamine use. Yet the urinary screening endpoint remained positive at substantial frequency, without a statistically significant time trend. These observations can coexist. A questionnaire records what an individual says; a urine screening result identifies whether an assay category was positive at the relevant collection. Neither alone measures the number of milligrams of methamphetamine consumed, the exact time of use, or the degree of psychomotor impairment.

The study was designed primarily to explore tolerability of modafinil up to 400 mg per day in people who might continue using methamphetamine. It was not designed to distinguish the medication’s pharmacologic effect from clinic contact, cognitive behavioral therapy, expectancy, natural change, or selective follow-up among participants who continued attending. A statement that participants improved during treatment is therefore substantially different from a statement that modafinil caused improvement.

What kind of experiment was this?

It was a six-week, open-label outpatient pilot without a placebo or untreated comparison arm. All medication-taking participants received modafinil and other supportive services. The first three days used 200 mg daily, followed by an intended increase to 400 mg daily. The protocol allowed reductions by 100 mg increments to no less than 200 mg if troublesome symptoms emerged, although the investigators reported no requested or required dose reductions. The researchers then observed participants after medication discontinuation near the end of the schedule.

Participants also received weekly manual-based cognitive behavioral therapy and contingency management: monetary incentives tied to returning blister packs, providing supervised urine samples, and attending therapy. These components matter scientifically because multiple interventions changed at once. A falling symptom score cannot be attributed to modafinil rather than to treatment engagement or other contemporaneous influences without a suitable comparison.

Why the participant denominator matters

The abstract says eight individuals were enrolled; the Results clarify that one never completed intake or began medication. Thus seven people actually entered the medication-taking group, and four of the eight enrolled completed the entire protocol. Three others contributed partial treatment follow-up, for less than one week, approximately 4.1 weeks, and approximately five weeks, respectively. Attrition is not just an administrative detail. Later-week averages describe the subset still observed and may no longer represent the same mix of people as earlier weeks.

The seven intake completers were three men and four women, all described by the source as Caucasian, with a mean age of 45.3 years and standard deviation of 7.2 years. Three reported injecting methamphetamine and four reported smoking it. They described using methamphetamine on a mean 20.1 of the previous 30 days, with a standard deviation of 8.25 days. These characteristics describe the recruited group; they do not support demographic generalization to a broader population of people with methamphetamine use disorder.

Study structure and disposition: keep the denominators straight

The table intentionally distinguishes people, repeat specimens, prescribed doses, and an assay cutoff. Treating three urine visits per week as three additional independent subjects would inflate the study size. Treating a 300 ng/mL urine screening criterion as a blood impairment threshold would confuse entirely different specimen types and interpretive purposes.

What was administered, and what was actually analyzed?

The administered study medication was modafinil, initially 200 mg orally per day, with a targeted increase to 400 mg. Methamphetamine was not experimentally administered. Instead, participants were recruited because of their existing substance-use histories and were assessed while some continued using methamphetamine outside a controlled dosing schedule. The authors did not establish the stereochemical composition, purity, actual doses, or administration times of illicit substances reportedly used during follow-up.

The chemical test involved supervised urine samples analyzed for an amphetamine screening category using a Hitachi 717 automated analyzer. The stated screening cutoff was 300 ng/mL. The study described classifying each week’s urine findings as no positive tests, some positive tests, or all positive tests. Those are categorical urine-screen outcomes. The report did not provide a contemporaneous quantitative, chromatographically confirmed series of parent methamphetamine concentrations for each person; nor did it establish that a positive screen quantified methamphetamine alone as opposed to a broader tested category.

Blood pressure and heart rate were physiologic measurements, not quantitative blood concentrations of modafinil or methamphetamine. The Hamilton scales and side-effect checklist were interviews or symptom ratings. Self-reported days of methamphetamine use came from participant reports. These differing measurement processes have different error structures and cannot be exchanged as if they were one common toxicological variable.

Urine positivity and participant reports point in different directions

The paper reports that every participant produced amphetamine-positive urine results during follow-up, with at least half of samples positive per week. The time-trend model for the proportion of positive urine screens was nonsignificant, t = −0.52, 23 degrees of freedom, p = .61. By contrast, self-reported methamphetamine use decreased significantly over time: t = −2.86, 259 degrees of freedom, p < .005. This is the central finding that should remain visible whenever the study is summarized.

The contrast does not prove either measure was fraudulent or useless. Urine screening and self-report answer separate questions over imperfectly aligned intervals. One can report fewer episodes yet continue to have positive specimens. Assay detection windows, specimen timing, reporting patterns, and categorical classification may influence observed patterns. This paper did not provide enough individual chemical concentration and timing data to allocate those mechanisms quantitatively or to adjudicate individual truthfulness.

The model’s reported degrees of freedom also must not be mistaken for additional enrolled participants. The same small group was repeatedly observed and modeled over time. A large model degree-of-freedom count cannot repair the absence of an independent placebo group, nor does it change the number of biological persons exposed to the regimen.

Which findings were statistically significant and which were not?

The table reports the investigators’ original numerical analyses, not a fresh pooled meta-analysis or a modern independent replication. Many endpoints were assessed in a pilot with few people and no untreated counterfactual. Nominally significant longitudinal scores support a description of within-study change; they do not identify the component responsible. Nonsignificant trends likewise do not demonstrate equivalence or prove absence of cardiovascular risk.

Why unchanged blood pressure is not a declaration of safety

The investigators found no statistically significant time trend for systolic pressure, diastolic pressure, or heart rate. They reported individual mean systolic readings spanning approximately 102–150 mmHg and diastolic readings spanning approximately 62–102 mmHg during follow-up. Heart rates were generally in a broad range of roughly 65–95 beats per minute. These are descriptive physiologic values during clinical encounters, not continuous cardiovascular monitoring or a randomized comparison.

A study with seven medication-taking participants and incomplete follow-up can miss uncommon, delayed, or context-dependent harms. Moreover, the Results document an event involving tachycardia, dry mouth, and nervousness after a participant doubled a dose the day after a missed dose; the symptoms resolved within an hour according to the investigators. This observation is inconsistent with turning the nonsignificant mean vital-sign analyses into a categorical assertion that modafinil carries no acute cardiovascular concern when combined with unsupervised drug use.

What adverse events actually occurred?

Five adverse events were recorded, three judged by the investigators to be at least probably related to study participation. Reported experiences included tachycardia and dry mouth, with the double-dose incident also involving nervousness. The authors categorized events overall as mild to moderate, with no required or requested modafinil dose reduction. They also described practical attendance barriers, including transportation and employment demands, in connection with withdrawal from the pilot.

The number of events is not the number of people harmed, and it does not quantify incidence in a population. One person can contribute more than one symptom or event. Similarly, absence of a serious reported study event is not proof of a risk-free medication or of safety in a different individual with cardiovascular disease, multiple substances, an unknown illicit-drug dose, or different co-medications.

Why did self-reported side effects and mood scores decrease?

The mean modafinil side-effect checklist score was highest before modafinil began and decreased during follow-up. Both Hamilton anxiety and depression scores likewise showed statistically significant declining time trends. The study does not isolate the mechanism. Participants knew they were receiving modafinil; weekly psychotherapy and visit structure could influence distress and reporting, while regression toward typical symptom levels and attrition can change group summaries.

That an adverse-effects checklist declines across visits does not imply that all adverse experiences disappeared. The investigators simultaneously reported discrete events such as tachycardia and dry mouth. Scale trends aggregate subjective ratings; incident reporting records selected episodes. In expert analysis, both findings belong together rather than selecting one and suppressing the other.

How the original statistical model shapes interpretation

The investigators used hierarchical linear or random-coefficient regression with time in weeks as a continuous predictor. Most outcomes were modeled with normal errors; skewed counts of reported methamphetamine use were analyzed with a Poisson distribution and log link. Categorical urine-screen results were modeled as logits. This allowed the authors to characterize change in repeated observations, although some participants provided markedly fewer observations than others.

A longitudinal model is an efficient way to use the observations collected, but its assumptions deserve scrutiny when only seven people reached intake. Estimated week effects may depend on distributional assumptions, irregular follow-up, variation among individuals, and unmodeled aspects of who returned to clinic. The study’s report does not supply a comparison arm that experienced the same contact and incentives without active medication. It therefore cannot separately estimate a causal medication effect from time and clinic engagement.

Why the urine assay is important but limited

A positive supervised urine screen can be evidence that a testing category exceeded the laboratory’s screening decision level in that specimen. It is not an automatic measurement of contemporaneous brain effect, the precise amount ingested, or the exact timing of all previous use. Urine analyte excretion depends on multiple biologic and analytic factors and can extend beyond the period of observable effects. The paper did not produce case-level chromatograms, calibrator data, quantified parent/metabolite series, or laboratory validation records that would allow an independent person-specific reconstruction.

The reporting level of 300 ng/mL was a urine screen cutoff, not a legal human whole-blood amphetamine concentration. Do not move that numerical threshold into blood, oral fluid, sweat, hair, or breath. For a general discussion of why different specimens carry distinct interpretive limits, see the independently published guide to biological matrices. For distinctions between presumptive screening and compound-specific analytical confirmation, see the published confirmatory testing guide.

Which scientific question can the study answer?

It supports a bounded report that modafinil at the studied outpatient regimen could be administered and followed in a very small group of methamphetamine-dependent adults, and that the investigators observed few clinically documented adverse events and no significant average vital-sign time trends. It describes improvements over time in selected subjective scales. At the same time, it shows no significant decline in the proportion of amphetamine-positive urine specimens despite declining self-reported use. These are useful observations for designing larger, controlled studies.

It does not answer whether modafinil independently reduces methamphetamine consumption, whether 400 mg/day is safe in a broader medically heterogeneous population, whether any one participant abstained during an individual interval, or how much methamphetamine was present in a participant’s blood at a particular moment. The authors themselves identified the absence of blinding and placebo control and the very small sample as central limitations.

What a controlled replication would need to measure

A larger trial would need an adequate randomized control condition, clearer handling of dropouts, consistent drug-use measurement intervals, prespecified primary outcomes, and verification of how concomitant counseling or incentives affect both groups. To compare self-report with objective toxicology, it would need transparent specimen collection schedules, confirmatory compound-specific methods, explicit analytes and reporting limits, and a prespecified way to reconcile repeated urine samples within persons. If the research question involved drug pharmacokinetics, it would require appropriately timed quantitative specimens and validated drug assays rather than relying on clinical vital signs.

Even such an improved trial would not automatically establish legal driving impairment. Translating a result to a particular driver would additionally require correctly identified case samples, matrix-specific validated measurement, pharmacologic context, relevant functional testing, and the particular facts of the event. This pilot provides none of those individual-level case inputs.

Forensic interpretation: four questions must stay separate

Identification: what chemical or screening category was detected? Here, the primary urine measure is an amphetamine-positive screening classification. Quantity: how much parent compound was present in a validated specimen? That patient-specific quantitative concentration was not provided. Timing: when was the compound taken relative to the sample or event? Individual methamphetamine administration was not controlled or reconstructed. Effect: was the person demonstrably impaired or behaviorally altered at a legally relevant moment? The trial collected neither case-specific driver performance nor individual intoxication determinations.

Combining these questions into a single assertion creates avoidable error. A urine positive does not prove concurrent impairment; a weekly reduction in reported use does not prove urine conversion to negative; a nonsignificant blood-pressure trend does not prove absence of a dangerous episode; and an association over treatment weeks does not prove that modafinil produced the observed change. Each stronger proposition would require distinct supporting evidence.

How to describe the study accurately in litigation or scientific education

An appropriate summary would identify the study as an eight-enrollment, seven-intake, four-completer open-label modafinil feasibility pilot, with weekly therapy and contingency incentives. It would report the source-defined 300 ng/mL urine-screen decision level, the persistent qualitative screen positivity, the nonsignificant urine time trend at p = .61, and the contrasting significant downward trend in self-reported use at p < .005. The study’s anxiety, depression, side-effect, cardiovascular and adverse-event findings should be reported alongside—not substituted for—those drug-use endpoints.

Any expert relying on the paper should distinguish information reported directly by its investigators from case-specific inference, and should not extrapolate findings from seven observed medication takers into a legal concentration threshold, validated last-use estimator, or opinion about any individual’s physical or mental state. The scientifically useful message is that the denominator, assay matrix, longitudinal measurement structure, and control condition determine what an apparently simple treatment-result sentence means.

Seven frequently asked scientific questions

1. Did modafinil cause the reported reduction in methamphetamine use?

No causal attribution was established. Participants reported lower use over time, but this was an uncontrolled, unblinded pilot combining modafinil with psychotherapy, incentives, and repeated clinical visits. The amphetamine-positive urine proportion did not show the same statistically significant improvement.

2. Were there eight, seven, or four participants?

All three numbers are valid only with their denominators attached: eight enrolled, seven completed intake and apparently began the study regimen, and four completed the full protocol. Repeated urine collections and weeks of ratings were not additional independently recruited individuals.

3. Was methamphetamine administered under controlled study conditions?

No. The controlled treatment was oral modafinil with an intended maintenance dose of 400 mg/day after three days at 200 mg/day. Methamphetamine use arose from participants’ behavior outside an experimental dosing schedule, so actual illicit doses, times, and isomer compositions were not known.

4. What does the 300 ng/mL value represent?

It is the study’s stated urine amphetamine-screen cutoff, not a measured serum or human whole-blood methamphetamine concentration. A screening cutoff is not a pharmacodynamic impairment threshold, and the study did not establish a quantitative blood–urine conversion.

5. Why could participants report less use while still testing positive?

The endpoints differed in time scale and measurement method. A qualitative urine category can remain positive even as reported use frequency declines, and screening does not measure every individual consumption event. This paper cannot assign the discrepancy to a particular person, analytical limitation, or reporting behavior with certainty.

6. Did the nonsignificant vital-sign findings prove modafinil was cardiovascularly safe?

No. The study did not find significant mean longitudinal trends in pressure or heart rate, but it was very small and included reports of tachycardia, particularly around a double-dose event. A nonsignificant p value is not an equivalence test or a comprehensive exclusion of risk.

7. Can this study support an opinion about a particular driver, last use, or criminal incident?

No. It did not report quantitative legal venous whole-blood drug levels, calibrated parent-drug pharmacokinetics, a validated individual last-use model, standardized driving performance, or evidence from a specific incident. Those conclusions require independent case-specific evidence and validated methods.

Original research source and scope

The findings, original numerical values, and experimental design summarized here derive from McGaugh J, Mancino MJ, Feldman Z, Chopra MP, Gentry WB, Cargile C, Oliveto A. Open Label Pilot Study of Modafinil for Methamphetamine Dependence. Journal of Clinical Psychopharmacology. 2009;29(5):488–491. Primary original research DOI 10.1097/JCP.0b013e3181b591e0. This independently written educational interpretation is not a replication, clinical recommendation, substitute laboratory report, or opinion on any person’s case.