Does Lisdexamfetamine Help Adults With ADHD Quit Smoking? A 32-Person Nicotine-Patch Trial

Short answer: In a randomized pilot trial of 32 adult smokers diagnosed with attention-deficit/hyperactivity disorder (ADHD), adding oral lisdexamfetamine to nicotine patches did not establish a higher rate of sustained smoking abstinence than adding placebo. Carbon-monoxide-defined four-week abstinence occurred in 5 of 17 people assigned lisdexamfetamine and 3 of 15 assigned placebo (p = .54). Clinician- and participant-rated ADHD symptoms improved more with lisdexamfetamine, but that clinical symptom benefit did not translate into a statistically established cessation benefit. Exhaled carbon monoxide was used as an indirect smoking marker, not as an assay of nicotine, lisdexamfetamine, dexamfetamine, or a statutory blood-drug concentration. The trial does not determine anyone’s drug concentration, precise last-smoking time, driving impairment, or cause of an event.

The primary evidence is the complete original clinical study by Kollins, English, Itchon-Ramos, Chrisman, Dew, O’Brien, and McClernon, A Pilot Study of Lis-Dexamfetamine Dimesylate (LDX/SPD489) to Facilitate Smoking Cessation in Nicotine-Dependent Adults With ADHD, published online in 2012 and appearing in Journal of Attention Disorders 18(2):158–168 (2014). Read the original study and its DOI. The distinction between its smoking outcome, ADHD ratings, breath carbon monoxide, and unmeasured blood drug concentrations is central to its interpretation.

What was the trial’s actual question?

The investigators asked whether lisdexamfetamine, a prescribed prodrug stimulant used for ADHD, could improve a smoking quit attempt when both randomized groups already received transdermal nicotine replacement therapy. This was not a comparison of stimulant treatment against no smoking treatment. It was an incremental-treatment question: does adding lisdexamfetamine to a shared nicotine-patch program outperform adding a blinded placebo capsule? That distinction matters because an effect shared by the groups cannot be assigned to lisdexamfetamine.

The trial also explored ADHD symptoms, computerized attention and inhibitory-control measures, mood, vital signs, and adverse events. Those were separate outcomes. A medication can change a psychiatric symptom score while leaving the probability of meeting a predefined smoking abstinence endpoint statistically indistinguishable from placebo. Conversely, a nonsignificant cessation comparison does not establish that each individual participant experienced no benefit. The study was described by its authors as a proof-of-concept pilot, not a definitive effectiveness trial across all smokers.

Who participated, and how many independent people were studied?

The independent randomized sample consisted of 32 adults, not 32 at each clinic visit or hundreds of independent observations. Seventeen received lisdexamfetamine, and fifteen received placebo. Participants were between 18 and 50 years old, smoked at least ten cigarettes per day, met the study’s diagnostic criteria for ADHD, and wanted to quit. Entry included breath carbon monoxide of at least 10 parts per million, psychiatric screening, a urine drug screen, and an estimated IQ eligibility threshold. Those requirements yielded a selected clinical group, not a random population sample of all nicotine users.

The study included twelve women, six in each arm; 37.5% of all participants were women. Other psychiatric conditions and illicit drug use were exclusions, and existing ADHD medications were washed out before study participation for certain participants. Subsequent repeated questionnaires, clinic visits, carbon monoxide readings, and reaction-time tasks all came from the same individuals; treating each as a new biological subject would falsely inflate statistical certainty.

Which drugs were actually administered?

Both groups began 21 mg/day transdermal nicotine replacement before the scheduled quit attempt. The initial patch period was approximately ten days, and the protocol later reduced patches to 14 mg/day and then 7 mg/day after quitting. Patch exposure therefore started before any randomized lisdexamfetamine or placebo treatment. During the quit phase, the active group began 30 mg oral lisdexamfetamine once daily, with permitted clinician-guided increases to 50 mg and 70 mg at specified study visits and possible dose reductions for tolerability. The comparison group received visually matched placebo capsules.

Lisdexamfetamine is a prodrug that produces dexamfetamine after absorption and biotransformation; it is not methamphetamine. The doses in milligrams identify administered pharmaceutical amounts, not analytically measured concentrations in a person’s blood. The original investigation did not provide paired human whole-blood, plasma, or serum concentration-time measurements for these agents. It therefore cannot calculate an individualized conversion from prescribed capsule dose to a forensic blood concentration or a driving-performance threshold.

Why did nicotine-patch pretreatment matter?

Smoking changed before the randomized stimulant comparison began. Across the study groups, average cigarettes per day fell from 14.9 before nicotine-patch initiation to 9.8 by the randomization visit; the reported paired comparison was p < .0001. Mean expired-air carbon monoxide also declined, from 24.5 ppm to 15.5 ppm (p < .001). These are observed before-versus-after changes during shared patch pretreatment. They cannot establish that lisdexamfetamine caused a decline, because the group-specific randomized capsules had not yet begun.

Another boundary is important: the pretreatment comparison was not a separately randomized nicotine-patch versus no-patch experiment. The reductions occurred alongside instruction, repeated contact, monitoring, participant motivation and time. They are compatible with benefit during patch treatment but do not isolate nicotine replacement as a uniquely causal agent. The trial’s randomized inference begins with the later lisdexamfetamine-versus-placebo comparison in participants who already had nicotine replacement.

Evidence table 1: primary abstinence and pre-randomization smoking data

The first three rows compare randomized treatment arms; the last two describe the shared phase before allocation to lisdexamfetamine or placebo. The different columns for those phases should not be treated as equivalent experimental groups. The randomized treatment produced a numerically higher CO-defined continuous-abstinence proportion in the lisdexamfetamine arm, but the sample and statistical comparison do not support a conclusion that it reliably improved smoking cessation. The self-report definition yielded a different numerical ordering, also without statistical significance.

Why are carbon-monoxide-defined abstinence and self-report not interchangeable?

The primary outcome required sustained four-week abstinence defined by expired breath carbon monoxide at or below 4 ppm at every postquit study visit, with anyone leaving the trial counted as having lapsed. The instrument measured carbon monoxide in exhaled air, a combustion-related exposure marker. It did not directly assay nicotine, cotinine, the ADHD medication, or illicit drugs. Its cutoff was a protocol definition for a smoking outcome, not a statutory threshold for driver impairment or an independent proof of a particular inhalation time.

Participants also reported smoking in diaries and at visits. Only two placebo participants, compared with none in the lisdexamfetamine arm, reported uninterrupted abstinence under the self-report definition; that difference too was nonsignificant. The contrast with five versus three under CO verification is not a reason to declare one measurement fabricated or to combine the numerators. The source describes different assessment rules and potential reporting behaviors. Casework should preserve each measure’s timing, cutoff, missing-data assumptions and purpose, rather than replacing one with the other.

Did the trial establish lower relapse risk at any later time?

No treatment-related separation in the reported survival analyses was established. The CO-defined lapse curves had a log-rank p of .50, and the self-reported lapse analysis had p of .15. Both are nonsignificant comparisons. These results are consistent with the primary conclusion that adding lisdexamfetamine did not show a demonstrable cessation advantage over nicotine patches with placebo in this small selected trial.

A survival curve uses the participant’s course over time, not merely the final count, but it remains based on 32 randomized people. Moreover, missing participants were treated as lapses for the primary outcome. That rule is defensible as a conservative trial convention, but it is not direct observation of cigarette consumption in every absent participant. No systematic extended follow-up after treatment termination was collected, so long-term cessation and relapse rates beyond the trial cannot be inferred from these visits.

Did lisdexamfetamine improve ADHD symptoms?

Yes, within this trial the authors reported significantly greater improvement in ADHD symptom ratings for the active medication. For clinician-rated symptoms on the Conners Adult ADHD Rating Scale, the treatment-by-visit interaction was p = .01. The corresponding participant-rated interaction was p < .001. Treatment-arm separation in clinician ratings appeared from roughly two weeks after the quit date, whereas the self-report comparisons separated after about three weeks.

The strength of this finding is that it concerns outcomes actually assessed in the randomized groups. Its limits are equally specific: these were clinical rating instruments, not chemically assayed dexamfetamine concentrations, psychomotor driving tests, objective verification that a particular person had stopped smoking, or proof that attention improved in an emergency-driving situation. The symptom result should not be substituted for the nonsignificant cessation endpoint. A single medicine may have a favorable effect on its intended disorder without improving every concurrent behavioral goal.

What happened to attention, inhibition, and negative mood after quitting?

Investigators examined computerized Continuous Performance Test (CPT) commission errors as one index of inhibitory control, a reaction-time variability measure as an index of consistency of attention, and the negative-affect component of the Positive and Negative Affect Schedule (PANAS). Across visits, they did not find significant treatment main effects or treatment-by-visit interactions for those task and mood measures. The reported visit effect for reaction-time standard error was marginal (p = .051), which should not be relabeled statistically significant at a conventional .05 cutoff.

The source itself cautioned that both arms had little measured disturbance after the quit attempt on these particular instruments. Prequit nicotine replacement in both arms could have blunted differences, but that is a plausible interpretation rather than a demonstrated isolated mechanism. A null CPT or PANAS comparison also does not mean that no withdrawal symptom occurred in any participant, or that a computerized task validates an individual legal assessment of intoxication, attention during a collision, or workplace fitness for duty.

Evidence table 2: clinical improvement, genuine nulls and tolerability

Numerically higher adverse-event reporting is compatible with a tolerability burden, but the original study did not establish a statistically significant group difference in any-adverse-event occurrence in this small sample. Specific symptoms are descriptive counts rather than controlled pharmacovigilance incidence estimates. Conversely, the nonsignificant vital-sign group comparisons do not prove that stimulant exposure has no cardiovascular effects in all people.

What adverse events and withdrawals were reported?

Across the entire study, fifteen of seventeen participants assigned lisdexamfetamine and twelve of fifteen assigned placebo described at least one adverse event; the group difference was not significant. When restricted to after randomized capsules began, adverse events were reported by thirteen of seventeen versus eight of fifteen, again without a statistically established group difference. The original table lists postbaseline reduced appetite in 35.3% of the active arm versus 0% of placebo, and irritability in 17.6% versus 6.7%, but it does not provide a defensible adjusted causal comparison for each symptom.

One participant receiving lisdexamfetamine discontinued because of marked anxiety considered severe and possibly drug-related. There were no serious adverse events reported in the trial. The small sample, limited treatment duration, concurrent nicotine replacement and eligibility screening constrain how far these observations generalize. An individual’s treatment history, contemporaneous clinical observations, medication records and validated chemical measurements would still be required for an incident-specific toxicological opinion.

Exactly what did the investigators measure in biological specimens?

The principal biochemical smoking marker was carbon monoxide in exhaled breath, reported in parts per million. The study also obtained urine drug screens for enrollment and during postquit monitoring, but these were eligibility or surveillance tests; they were not a validated quantitative series of lisdexamfetamine, dexamfetamine, nicotine, or cotinine concentrations. No assay calibration, compound-specific quantitative result, time-concentration curve or validated individual blood pharmacokinetic threshold for those drugs was reported as a study endpoint.

In particular, the study did not measure venous forensic whole blood dexamfetamine or nicotine, separately centrifuged plasma or serum parent concentrations, saliva/oral-fluid drug levels, hair concentrations, sweat, vitreous, or postmortem tissues. A urine screening outcome should not be treated as equivalent to a confirmed parent-drug blood result. Expired carbon monoxide is not a blood-alcohol concentration, a breath-ethanol result, or a surrogate lisdexamfetamine concentration. These analytical distinctions are foundational to sound forensic interpretation.

Why is a trial smoking cutoff not a forensic impairment cutoff?

The investigators’ exhaled-CO threshold of 4 ppm served their longitudinal operational definition of smoking abstinence at observed visits. It did not purport to determine neurological performance at that moment, diagnose recent use of a particular nicotine device, or establish a jurisdiction’s impaired-driving limit. CO can vary with time, exposure and measurement conditions; the report did not establish an individualized clock for exact last cigarette use. Smoking behavior, nicotine replacement, medication administration and actual measured drug concentration are analytically separate questions.

If a case instead involves suspected stimulant impairment, a defensible analysis needs the actual biological matrix, sample-collection time, method validation, analyte identity, concentration, chain of custody and separate functional observations. The existing guide to biological matrices explains why urine, plasma, serum and whole blood must not be casually interchanged. Even analytically valid concentration results generally do not, alone, establish individual impairment or causation. This clinical cessation study provides no substitute for that case-specific evidence.

Does the clinical result show that lisdexamfetamine is ineffective for every smoker?

No. It supports a narrower conclusion: the researchers did not detect an incremental four-week cessation benefit in their randomized sample, in which all participants received prequit and postquit nicotine patches. The observed primary proportions differ numerically, but the statistical test did not establish a reliable treatment-group difference. The pilot size makes its estimates imprecise, and the study did not track long-term smoking outcomes after treatment ceased. It did not enroll smokers without ADHD or compare lisdexamfetamine against an otherwise identical group receiving no nicotine patch.

It would be equally unwarranted to claim that treating ADHD must promote cigarette cessation. Clinical improvement and sustained abstinence were directly separated by the trial’s results. The correct inference is outcome-specific and population-specific, with the limitations of the original design expressly retained. Personal treatment choices require individualized clinical assessment rather than extrapolation from this limited pilot trial.

What should a forensic or litigation review verify before citing this paper?

Start with the original study protocol and clarify which question is being offered as evidence. For cessation, identify the randomized denominator, missing-data convention, exact four-week CO rule, self-reported smoking definition and nonsignificant p values. For ADHD efficacy, cite symptom-rating instruments and treatment-by-visit differences without transforming clinical scales into driver-fitness tests. For medications, distinguish oral lisdexamfetamine dose from measured active dexamfetamine and from chemically different methamphetamine. For nicotine exposure, distinguish delivered transdermal patch dose from any laboratory-determined nicotine or metabolite concentration that was not measured here.

Where case testing exists, compare the clinical paper with an actual validated laboratory method and its uncertainty and reporting limits. The guide to confirmatory drug testing describes the distinction between a nonspecific screen and target-analyte confirmation, while the guide to therapeutic drug monitoring addresses interpretation of properly timed measured concentrations. Those concepts are relevant background, not methods used to measure lisdexamfetamine in this trial.

Seven focused questions about the original evidence

1. Did the medication group have more people with CO-defined abstinence?

Numerically yes: 5/17 compared with 3/15. The difference was not statistically significant (p = .54), and the counts use the trial’s stringent repeated exhaled-CO endpoint plus a missing-equals-lapse rule. Reporting only the percentages while suppressing that null result would materially misstate the research. It was a preliminary sample rather than a powered demonstration of a clinical cessation advantage.

2. Why did self-reported continuous abstinence look different?

The original source reported zero continuously abstinent lisdexamfetamine participants versus two placebo participants under a separate self-report definition, with p = .11. Reports and exhaled-CO visit thresholds address overlapping but nonidentical constructs and rules. The paper does not justify mathematically blending these numerators or replacing one endpoint with the other to produce a desired inference.

3. Did the study administer methamphetamine?

No. It administered oral lisdexamfetamine, an ADHD prodrug medication that yields dexamfetamine, against placebo while both arms received transdermal nicotine replacement. Methamphetamine is a different chemical substance. Neither a positive urine amphetamine-class screen nor an oral prescription amount becomes proof of methamphetamine administration without substance-specific analytical and clinical context.

4. Was expired breath carbon monoxide a measurement of blood nicotine?

No. CO was measured in exhaled breath in ppm and used to support smoking-outcome classification. It was not whole-blood nicotine, serum cotinine, urine nicotine metabolite quantitation or dexamfetamine testing. The 4-ppm research criterion is not an individual intoxication or impaired-driving standard. Analytical matrix and outcome definitions must remain separate.

5. Was computerized attention significantly better with lisdexamfetamine?

The strongest randomized improvement was in self- and clinician-rated ADHD symptoms. For the particular CPT commission-error and reaction-time variability comparisons, the study did not report significant treatment-by-visit benefits, and a reaction-time variability visit effect had p = .051. That is not a conventional positive result. These computer tasks do not independently validate on-road driving competence.

6. Do nonsignificant cardiovascular findings establish that the medicine is risk-free?

No. Between-group blood pressure and heart rate comparisons were nonsignificant in this small short-duration sample, but one lisdexamfetamine participant discontinued for marked anxiety and postbaseline adverse events were descriptively more frequent in the active arm. A lack of statistical separation on a safety variable is not proof of universal safety, especially with limited sample size.

7. Can the study determine when someone last smoked or whether someone drove impaired?

No. A scheduled visit’s exhaled-CO reading and diary history cannot identify a precise last cigarette time without additional evidence, and the investigators conducted neither actual driving tests nor quantitative forensic drug blood analysis. Case-specific opinions require validated analytical specimens and separate functional and temporal evidence. This study speaks to randomized group smoking outcomes and ADHD symptom ratings, not individual incident reconstruction.

Primary-source conclusion

Kollins and colleagues reported a small, blinded, randomized test of lisdexamfetamine added to an existing nicotine-patch quit program. The principal smoking-abstinence and lapse comparisons were nonsignificant, although clinician- and self-rated ADHD symptom improvements favored active treatment. Pooled smoking and expired-CO readings improved during pretreatment before randomized capsules began. The evidence therefore supports separate statements about a null incremental smoking-cessation result, an ADHD symptom benefit, and unmeasured quantitative drug and driving outcomes. Consult the original primary trial for the full experimental report. These conclusions are scientific interpretation of group data, not an individual toxicology test result or an opinion on a particular incident.