RESEARCH DOSSIER
Stimulant and Nonstimulant Pharmacotherapy for ADHD
Abstract. Pharmacotherapy for attention-deficit/hyperactivity disorder (ADHD) consists of stimulant and nonstimulant interventions whose clinical effects arise from interacting pharmacodynamic, pharmacokinetic, developmental, psychological, and care-system processes. This fundamental research monograph examines the historical emergence of ADHD medication, contemporary nosology and treatment targets, catecholaminergic mechanisms, methylphenidate- and amphetamine-based preparations, norepinephrine reuptake inhibitors, alpha-2 adrenergic agonists, comparative efficacy, dose-response relationships, clinical trial design, cardiovascular and psychiatric safety, growth and sleep, substance-use concerns, developmental and reproductive populations, functional outcomes, and longitudinal evidence. It critically compares randomized trials, network meta-analyses, pharmacovigilance, pharmacoepidemiology, official U.S. prescribing information, and professional recommendations, with attention to confounding, measurement validity, duration of follow-up, selective enrollment, and the difference between symptom response and real-life functioning. The synthesis finds substantial short-term symptom benefits for well-studied medications, drug-class differences in clinical tolerability and pharmacological time course, and considerably greater uncertainty about person-specific optimization, uncommon harms, long-term causal effects, and generalization across settings. It frames treatment as a longitudinal and monitored clinical process requiring explicit distinctions among regulatory authorization, trial evidence, observed associations, and patient-centered outcomes.
Keywords: attention-deficit/hyperactivity disorder; ADHD; stimulants; methylphenidate; amphetamine; lisdexamfetamine; dexmethylphenidate; atomoxetine; viloxazine; guanfacine; clonidine; psychopharmacology; norepinephrine; dopamine; clinical trials; network meta-analysis; drug safety; cardiovascular monitoring; suicidal ideation; treatment adherence; functional outcomes.
Scientific Overview and Research Questions
ADHD is a clinically defined neurodevelopmental syndrome involving developmentally inappropriate patterns of inattention and/or hyperactivity–impulsivity associated with impairment across relevant contexts. Pharmacotherapy investigates how medicines modify the symptoms, associated functional restrictions, and treatment burdens of that syndrome. The scientific object of this dossier is the therapeutic use of medications in people evaluated for ADHD, including comparisons across drug classes and the consequences of prescribing, monitoring, continued exposure, and discontinuation. The broader etiology, full diagnostic classification, educational interventions, and every form of executive-function research belong to connected scientific domains. Here those domains enter as determinants of who is studied, what outcome is measured, and how treatment effectiveness can be interpreted.
The central research questions concern the mechanisms through which different pharmacological classes influence attention and behavior; the validity of reported symptom improvements; comparative efficacy and tolerability by age and formulation; causal identification of rare and cumulative adverse events; the relationship between medication response and educational, occupational, interpersonal, and safety outcomes; and the conditions under which findings from clinical trials can inform longitudinal decisions. The scientific evidence has several temporal scales. Some direct effects can be observed within hours of stimulant administration. Most trials assess outcomes over weeks. Growth, cardiovascular disease, developmental trajectories, use patterns, and quality of life require years of observation and substantially different study designs. A single estimate cannot represent all these levels.
This monograph uses a critical narrative synthesis. Original randomized trials, systematic reviews, observational studies, clinical guidance, and regulatory product labeling are attributed according to their distinct evidentiary functions. It is not a preregistered systematic review and contains no newly pooled treatment-effect calculation. Controlled efficacy claims derive primarily from clinical trials and reviews; mechanisms are represented as pharmacological explanations supported to varying degrees by experimental work; observational associations are discussed in the context of confounding and selection. References are traceable through direct DOI, PubMed, official journal, or regulatory links. The resulting review is an academic research resource rather than an individualized medication protocol.
The Clinical Construct and the Meaning of Medication Response
ADHD diagnosis rests on symptom history, impairment, developmental course, cross-setting information, and evaluation of alternative or co-occurring explanations. A high score on an attention questionnaire alone does not establish the disorder. Sleep difficulties, trauma, anxiety, mood disorders, substance exposure, learning disabilities, and other conditions can change attention and activity, while also coexisting with ADHD. The American Academy of Pediatrics clinical practice guideline (Wolraich et al., 2019) (https://publications.aap.org/pediatrics/article/144/4/e20192528/81590/Clinical-Practice-Guideline-for-the-Diagnosis) specifies systematic evaluation and screening for co-occurring conditions in young people. The NICE guideline NG87 National Institute for Health and Care Excellence. (2018, updated 2019). *A similarly anchors treatment in diagnosis, needs assessment, and functional impairment.
Response has several operational meanings. A clinician may describe a reduction on an ADHD Rating Scale, a parent may observe fewer interruptions, a teacher may report increased task completion, and an adult may report less disorganization or time lost to unfinished activities. These outcomes have overlapping but distinct error structures and social meanings. Symptom reduction is often easier to demonstrate than sustained academic gains, occupational stability, or quality-of-life improvement. Ratings may reflect medication efficacy, expectancy, adverse-event-related unblinding, the observation context, or changes in what a rater regards as problematic. Robust research therefore specifies the outcome, rater, observation window, and intended inference.
Medication response is also heterogeneous. Some patients show clinically meaningful improvement with one stimulant class after an unsatisfactory response to another. Others experience limiting adverse effects or prefer nonstimulant treatment. This pattern motivates individualized assessment while leaving open the question of which pretreatment biomarkers reliably forecast the best medicine. The construct of an individual responder is itself sensitive to cutoffs, measurement error, day-to-day symptom variation, and whether the outcome is defined by a clinician, the patient, or an observer in a specific environment.
Historical Development of ADHD Pharmacotherapy
The history of ADHD medication predates modern diagnostic terminology. Early clinical reports examined stimulant treatment for behavioral and school difficulties decades before the DSM categories of attention-deficit disorder and ADHD became established. The meaning of a drug response changed alongside educational norms, pediatric psychiatry, behavioral rating scales, the development of extended-release technology, and regulatory expectations for controlled trials. Retrospective claims about a single discovery event can obscure the parallel histories of compounds, diagnostic classifications, and the institutions that tested their therapeutic use.
Methylphenidate became a major pharmacological approach during the postwar expansion of psychopharmacology. Amphetamine compounds had a longer medical history but acquired specialized research and prescribing traditions for attention and behavioral symptoms. Later, atomoxetine represented a noncontrolled, norepinephrine reuptake inhibitor approach, while extended-release guanfacine and clonidine translated alpha-2 adrenergic pharmacology into specific ADHD treatment products. Viloxazine extended-release capsules entered the U.S. market for ADHD much later, despite the compound's earlier history in other uses. Regulatory labels are product-specific historical records: the atomoxetine label identifies original U.S. approval in 2002, and the Qelbree label Food and Drug Administration / National Library of Medicine. (2025). Qelbre records initial U.S. approval in 2021.
Clinical research changed decisively with coordinated randomized treatment comparisons. The Multimodal Treatment Study of Children with ADHD, or MTA (1999) MTA Cooperative Group. (1999). A 14-month randomized clinical trial of trea randomized 579 children to systematic medication management, intensive behavioral treatment, their combination, or community care for 14 months. This design evaluated strategies of organized care rather than an isolated molecular effect of one tablet. The trial found more improvement in core ADHD symptoms under carefully managed medication and combined treatment than under behavioral treatment alone or community care; some noncore outcomes favored combined care over comparison conditions. Its enduring contribution was methodological: dose titration, repeated clinical assessment, and the comparison of several interventions under the same protocol substantially changed interpretation of the medication literature.
Later research increasingly compared numerous treatments through systematic synthesis and network meta-analysis. Cortese and colleagues (2018) (https://doi.org/10.1016/S2215-0366(18)30269-4) integrated short-term randomized evidence in children, adolescents, and adults. Ostinelli and colleagues (2025) (https://doi.org/10.1016/S2215-0366(24)00360-2) expanded adult comparisons to pharmacological, psychological, and neurostimulation interventions, while Nourredine and colleagues (2026) (https://doi.org/10.1016/S2215-0366(26)00091-X) used dose-effect network models. The contemporary historical trajectory is therefore one of progressively more demanding outcome definitions, comparative designs, and safety surveillance, alongside continuing challenges to trial quality and external validity.
Pharmacological Targets and Catecholamine Signaling
Catecholamines are central to the pharmacological account of ADHD medications. Dopamine contributes to reinforcement learning, behavioral invigoration, cognitive control, and the regulation of actions; norepinephrine contributes to arousal, responsivity, and the modulation of attention. These functions are distributed across brain systems, not isolated to a single anatomical center. Therapeutic drugs alter neurotransmitter availability or receptor signaling in ways whose observable outcomes depend on network state, dose, temporal dynamics, task structure, and individual physiology.
Methylphenidate inhibits dopamine and norepinephrine transporters, changing extracellular catecholamine availability. Amphetamine-related agents affect monoamine transport and release through mechanisms that differ from transporter blockade alone. Atomoxetine inhibits norepinephrine reuptake and may affect dopamine signaling indirectly in prefrontal regions with particular transporter distributions. Viloxazine's clinically relevant pharmacology includes norepinephrine reuptake inhibition, although reducing the medicine to a single receptor-level claim would overstate mechanistic certainty. Guanfacine and clonidine act as alpha-2 adrenergic receptor agonists, with guanfacine commonly characterized by relative alpha-2A selectivity. Their clinical effects are accompanied by distinct autonomic properties.
Pharmacological target engagement establishes a plausible chain between dose and signaling; it does not itself prove a specific improvement in academic functioning or sustained psychological development. Different tasks rely on partially overlapping neural systems. The inverted-U concept used in catecholaminergic cognitive-control research captures the possibility that too little and too much neural modulation can each impair some functions; individual differences and task demands complicate direct translation into bedside dose decisions. Research must distinguish receptor or transporter effects observed experimentally from mediators of patient-reported symptom changes demonstrated in clinical populations.
Pharmacokinetics, Formulation, and Duration of Action
Pharmacokinetics describes absorption, distribution, metabolism, and elimination. In ADHD treatment it shapes the temporal pattern of symptom control and adverse effects. An immediate-release formulation may produce a different rise, peak, and decline in drug exposure than an extended-release tablet or capsule containing related active ingredients. Modified-release systems differ in their release mechanisms; their marketed durations cannot be assumed interchangeable across products. A long-acting clinical effect may arise from engineered delivery, a slowly converted prodrug, or the intrinsic persistence of the active compound.
Methylphenidate preparations include immediate-release and multiple extended-release technologies. Dexmethylphenidate is the pharmacologically active d-enantiomer of methylphenidate, and product-specific dosing cannot be inferred simply from an assumed generic equivalence. Lisdexamfetamine is a prodrug converted to dextroamphetamine after absorption, while mixed amphetamine salts and other formulations have their own labeled properties. The scientific consequence is that the same broad stimulant category contains meaningfully different exposure profiles, even when products target similar clinical constructs.
Nonstimulant agents also differ in exposure and dosing conditions. Atomoxetine metabolism is strongly influenced by CYP2D6 activity, including genetic variation and interactions with inhibitors. Guanfacine is affected by CYP3A4 interactions, while viloxazine inhibits CYP1A2 and may alter exposure to interacting drugs; product instructions must be checked rather than translated from class stereotypes. The Intuniv prescribing information Food and Drug Administration / National Library of Medicine. (2025). INTUNI specifically warns against equating extended-release tablets with other guanfacine products and describes clinically important changes under CYP3A4 inhibition or induction. The Qelbree prescribing information Food and Drug Administration. (2025). Qelbree (viloxazine extended-release) identifies contraindicated interactions involving sensitive CYP1A2 substrates. Formulation and metabolic exposure are therefore part of the evidence base, not minor dispensary details.
Methylphenidate and Dexmethylphenidate
The methylphenidate group is one of the best-studied pharmacological approaches to ADHD. Across placebo-controlled trials, methylphenidate generally reduces observer-rated core symptoms during active treatment. Its size of benefit varies by comparator, design, measurement, participant group, and assumed blinding. Storebø and colleagues' updated Cochrane review (2025) Storebø, O. J., Storm, M. R. O., Pereira Ribeiro, J., Skoog, M., Groth, C., concluded that teacher-rated symptoms may improve, while judging the certainty of the evidence very low because of methodological shortcomings, adverse-event-related loss of masking, and other limitations. That conclusion concerns confidence in precise estimated effects; it does not convert observed short-term differences into proof of no benefit.
The 2018 network meta-analysis by Cortese and colleagues (https://doi.org/10.1016/S2215-0366(18)30269-4) recommended methylphenidate as a preferred initial short-term medicine for children and adolescents at the group level, based on the combined comparison of efficacy and tolerability. This is a population-level interpretation of trials, not a universal ranking for every patient. A clinically plausible response to one preparation can coexist with poor tolerability of another because the formulation, meal conditions, absorption profile, context of assessment, and timing all affect the experience of treatment.
The principal ordinary burdens of stimulant treatment include reduced appetite, insomnia, abdominal discomfort, headache, and changes in mood or irritability, with marked variation among people. Blood pressure and pulse may increase; psychiatric and cardiovascular histories can alter the risk evaluation. The FDA's class-wide labeling action of May 2023 Food and Drug Administration. (2023, May 19). FDA requires updates to clari reinforced warnings about misuse, addiction, and overdose and called for repeated risk assessment. This regulatory action is a safety communication applicable to stimulant prescribing, while its frequency and causal estimates need evidence beyond the presence of a warning.
Amphetamine Preparations and Lisdexamfetamine
Amphetamine products form a distinct stimulant group. Their actions on monoamine transport and release account for differences from methylphenidate at the pharmacological level. Trials and network comparisons often find robust short-term symptom changes, particularly in adults. Cortese et al. (2018) (https://pubmed.ncbi.nlm.nih.gov/30097390/) reported that amphetamines had strong comparative short-term efficacy across age groups and judged amphetamines a preferred first-choice medicine in adults when considering overall trial evidence. The estimate should be read alongside its tolerability profile, the limitations of cross-trial comparison, and the distinctions among individual products.
Lisdexamfetamine has a prodrug exposure profile that can affect onset, duration, and susceptibility to some forms of manipulation. These properties alone do not guarantee absence of misuse or superior clinical outcomes under every circumstance. The relevant comparisons include active-drug response, discontinuation because of adverse events, daytime coverage, storage and diversion risk, and adherence in ordinary environments. A compound can score strongly in efficacy comparisons while being difficult to sustain for people with appetite suppression, sleep disruption, cardiovascular concerns, or other intolerable reactions.
A particularly important low-frequency safety question concerns new psychotic symptoms. In a large comparative observational investigation of young people starting stimulant treatment, Moran and colleagues (2019) Moran, L. V., Ongur, D., Hsu, J., Castro, V. M., Perlis, R. H., & Schneewei observed a higher risk of a documented psychosis episode with amphetamine than with methylphenidate. Because that was not a randomized head-to-head trial, channeling, clinical selection, and outcome ascertainment remain important. Nevertheless, it provides evidence relevant to monitoring unusual psychiatric reactions and to the limitations of treating all stimulant formulations as one homogeneous intervention.
Why Stimulants and Nonstimulants Differ in Time Course
Stimulants typically have an observable pharmacological effect during the same day that a dose is taken, although optimization over successive visits may be needed to identify a useful formulation and pattern of coverage. Some nonstimulants commonly require several weeks of treatment before their effect is assessed adequately; the response may accumulate as dose and exposure stabilize. These differences matter for trial length. A brief study ending before an agent can be adequately titrated may understate its practical effect, while a short study that records immediate stimulant response may overstate certainty about continuation over years.
The NICE NG87 evidence explanation (https://www.nice.org.uk/guidance/ng87/resources/attention-deficit-hyperactivity-disorderdiagnosis-and-management-pdf-1837699732933) explicitly recognizes that stimulants act more quickly than medicines such as atomoxetine or guanfacine. Duration should be measured as effective coverage of clinically relevant tasks, not solely plasma elimination or manufacturer-stated release. The scientifically meaningful time course includes morning activities, school or work, evening interpersonal functioning, sleep, and rebound or withdrawal-related changes. Interpretation must also take account of missed doses, timing of food, variable working hours, and the expectation that behavior may differ by setting.
Atomoxetine and Selective Norepinephrine Reuptake Inhibition
Atomoxetine is an ADHD medicine without stimulant controlled-substance scheduling in the United States. It inhibits norepinephrine reuptake and is used in children, adolescents, and adults within applicable product labeling. Its therapeutic effects are established in randomized trials, with typical evaluation over several weeks rather than solely immediately after administration. It is one of the comparatively well-supported nonstimulants in the Cortese et al. (2018) (https://doi.org/10.1016/S2215-0366(18)30269-4) and Ostinelli et al. (2025) (https://doi.org/10.1016/S2215-0366(24)00360-2) syntheses.
The clinical safety profile differs substantially from that of stimulants. Atomoxetine can be associated with gastrointestinal complaints, appetite changes, sleep or arousal changes, and increases in pulse or blood pressure. CYP2D6 metabolic variability can change exposure, and interacting medications may matter. The U.S. DailyMed atomoxetine prescribing information updated in 2026 Food and Drug Administration / National Library of Medicine. (2025–2026). A includes a boxed warning about increased suicidal ideation in children and adolescents, with an explicit monitoring requirement at initiation and during relevant clinical change. The existence of that warning calls for active vigilance; it does not establish that every patient has the same absolute risk.
Comparative trials must also distinguish efficacy from acceptability. The adult Ostinelli et al. (2025) (https://pubmed.ncbi.nlm.nih.gov/39701638/) network meta-analysis found improvements in core symptoms for atomoxetine and stimulants on both self- and clinician-rated scales, but atomoxetine showed more all-cause discontinuation than placebo in its analysis. Efficacy can therefore be demonstrated at the same time that trial participants find continued use less acceptable. A decision framework that considers symptom scores alone would miss this interaction.
Viloxazine Extended Release and the Expansion of Nonstimulant Options
Viloxazine extended release broadens the contemporary nonstimulant class. Its earlier pharmacological history does not substitute for trials of the currently marketed formulation and indication. The U.S. Qelbree prescribing information Food and Drug Administration / National Library of Medicine. (2025). Qelbre identifies approved use in adults and patients aged six years and older. It describes evidence from controlled clinical studies and assigns prominent warnings to treatment-emergent suicidal thoughts and behaviors, increases in blood pressure or heart rate, and activation of mania or hypomania. The label's drug-interaction profile, especially strong CYP1A2 inhibition, is clinically distinct from atomoxetine and the alpha-2 agonists.
The significance of viloxazine for research extends beyond adding another product to a table. Network comparisons of established treatments are constrained by the date through which studies were searched, the trials available for each drug, and the connection of interventions within a statistical network. A newer medicine may have fewer head-to-head trials and relatively limited longitudinal exposure histories. The results of an adult open-label phase 4 study by Adler and colleagues (2026) Adler, L. A., Lieberman, V. R., Brijbasi, L., Mattingly, G. W., Yarullina, described changes in symptom scores among adults with ADHD and depression or anxiety symptoms over 14 weeks. Without an active comparator or blinded placebo arm, those changes cannot by themselves isolate a causal effect of viloxazine, and concomitant medications complicate the inference. Such studies provide useful evidence about feasibility and clinical representation of comorbidity, while definitive comparative claims require controlled designs.
The current clinical evidence should not be collapsed into a contrast between an allegedly immediate “strong” stimulant and a uniformly delayed “weak” nonstimulant. Response distributions overlap, rater measures differ, long-term persistence varies, and side effects can become determining factors in usefulness. The scientific task is to understand which outcomes and populations each trial can identify and which remain untested.
Guanfacine Extended Release
Guanfacine is a central alpha-2 adrenergic agonist, with preferential activity at the alpha-2A receptor. Extended-release guanfacine has evidence as monotherapy and as an adjunct to stimulant medication in pediatric ADHD. Its effects may be particularly relevant where stimulant treatment has produced partial symptomatic improvement, although evidence supporting adjunctive treatment is not a license for unsupervised polypharmacy. The Intuniv U.S. label Food and Drug Administration / National Library of Medicine. (2025). INTUNI includes evidence of monotherapy and adjunctive efficacy and cautions about somnolence, hypotension, bradycardia, syncope, and cardiovascular conduction effects.
An important original trial enrolled 461 children and adolescents aged 6–17 years who had a suboptimal response to an existing long-acting psychostimulant. Participants continued that stimulant and were randomized to extended-release guanfacine in the morning, guanfacine in the evening, or placebo for nine weeks. Wilens and colleagues (2012) (https://pubmed.ncbi.nlm.nih.gov/22176941/) found greater improvement on ADHD symptom scores in both adjunctive guanfacine groups than in the adjunctive placebo group. The causal interpretation applies to the incremental effect of adjunctive treatment under that protocol, not to the effects of combining any two medicines in any clinical context.
An important distinction is physiological direction: whereas stimulants and some reuptake inhibitors can increase heart rate or blood pressure, alpha-2 agonists can reduce them. The two patterns are not mechanically interchangeable, nor does combining treatments cancel their separate risks. Guanfacine exposure is modified by CYP3A4 interactions, and abrupt discontinuation can produce rebound hypertension. In contrast to a simple “stimulant holiday,” interruption of an alpha-2 agonist involves its own cessation hazards; product-specific discontinuation guidance exists for that reason. More evidence is needed about durability, optimal combined-care selection, and long-term functional gains beyond short rating-scale trials.
Clonidine and Extended-Release Alpha-2 Agonist Formulations
Clonidine is another centrally acting alpha-2 adrenergic agonist. Its pharmacological effects on autonomic tone account for clinically relevant decreases in blood pressure and pulse, as well as sedation. Extended-release clonidine products are used for ADHD within their respective regulatory authorizations. Such products cannot be substituted dose-for-dose with unrelated clonidine formulations on the assumption that the active ingredient alone determines the exposure curve.
The 2024 FDA-approved ONYDA XR prescribing information Food and Drug Administration. (2024). ONYDA XR (clonidine hydrochloride) ex identifies an extended-release oral suspension indicated as monotherapy or adjunctive to stimulants in children aged six years and older. It emphasizes formulation-specific administration and rebound hypertension upon abrupt cessation. This illustrates the difference between the scientific ingredient category, the proprietary delivery system, and a particular label's approved ages and conditions. The presence of an approved pediatric product does not establish comparable adult evidence or broad interchangeability with other alpha-2 agonists.
Clonidine has sometimes been considered in clinical contexts involving sleep disturbance, impulsivity, or tics, but evidence for an individual co-occurring symptom should be separated from evidence for core ADHD and from the labeled indication. NICE NG87 recommends specialist input before particular off-label uses, including some uses of clonidine for children with ADHD and sleep disturbance or other complicated symptoms. Research on alpha-2 agonists must record sedation and alertness as independent outcomes, especially when classroom or work performance is the therapeutic goal. Quiet behavior produced by sedation does not necessarily represent an improvement in active cognitive control.
Comparative Short-Term Efficacy and Tolerability
Comparative evidence is informative only when its intervention definitions, time frames, and measurement procedures are preserved. The large 2018 network meta-analysis (https://pubmed.ncbi.nlm.nih.gov/30097390/) compared medications for ADHD in children, adolescents, and adults across double-blind randomized studies. The authors reported that multiple medicines were superior to placebo on short-term symptom outcomes, with differences in the magnitude of average effects and in tolerability. Their clinical interpretation favored methylphenidate for children and adolescents and amphetamines for adults in the short term. Network ranking is a statistical summary across particular trial networks, not a mechanistic law governing every new preparation or patient.
By contrast, Ostinelli and colleagues (2025) (https://pubmed.ncbi.nlm.nih.gov/39701638/) analyzed 113 adult randomized trials with 14,887 participants, spanning medication, psychological interventions, neurostimulation, and controls. Its pharmacological component represented 63 trials with 6,875 participants. On both self-rated and clinician-rated core ADHD symptom outcomes near 12 weeks, stimulants and atomoxetine were effective compared with placebo. Estimates of quality-of-life benefits were more uncertain, and the investigators explicitly noted limited investigation of long-term outcomes. Confidence in treatment effects varied from very low to moderate across comparisons. The component network framework changed the research question: what are the effects of treatment components when measured across a wider intervention ecosystem?
The 2025 Cochrane methylphenidate review Storebø, O. J., Storm, M. R. O., Pereira Ribeiro, J., Skoog, M., Groth, C., provides a methodological counterweight to simple league tables. It judged the certainty of all assessed outcomes very low, highlighting unblinding and other trial limitations. The correct scientific synthesis preserves both observations: repeated placebo-controlled contrasts generally support short-term improvement in core symptoms, and the magnitude and stability of estimated effects require caution because study quality and outcome assessment limit precision. Review heterogeneity cannot be resolved by selectively quoting the highest drug-ranking result.
Dose–Effect Relationships and the Limits of Escalation
Dose–response research distinguishes an intervention's assigned dose from the actual exposure received, its mean efficacy curve, and its adverse-effect curve. Many pivotal trials use flexible titration in which clinicians adjust doses in response to efficacy and tolerability. Observed outcomes in such studies cannot be naively attributed to a single fixed exposure, because treatment assignment, interim response, tolerability, and discontinuation interact. Randomized fixed-dose studies and carefully modeled dose–effect syntheses can address different aspects of this problem.
The 2026 dose–effect network meta-analysis by Nourredine and colleagues (https://pubmed.ncbi.nlm.nih.gov/42134365/) used double-blind randomized trials of oral monotherapy and modeled dose–effect curves separately in children/adolescents and adults. It identified 164 eligible reports for systematic review and included 113 trials in the dose–effect network analysis: 68 pediatric trials involving 14,138 participants and 45 adult trials involving 11,016 participants. The investigators found drug- and age-specific dose–effect patterns, including regions of diminishing additional benefit and dose-related increases in discontinuation due to adverse events. The authors appropriately discussed sparse data and uncertainty at higher exposures.
This analysis does not establish a universally optimal numeric daily dose. Milligram amounts are not interchangeable between methylphenidate, amphetamines, guanfacine, or differing formulations; their pharmacological exposures and labeled ranges differ. A network-derived apparent plateau is an estimate conditioned on drugs, ages, doses, and trial populations represented in the model. It is scientifically appropriate to infer that dose escalation can encounter diminishing returns and greater harms. It would be inappropriate to turn a cross-drug graph into instructions for an individual patient without clinical assessment and the specific product label.
The resulting research program should model benefits and harms jointly, including individuals who discontinue early because of adverse events. An efficacy curve calculated only among those who tolerate treatment may exaggerate the expected outcome for all patients who begin therapy. A study that reports a dose effect but omits the competing probability of discontinuation answers only half of the treatment question.
Randomized Trial Design and Sources of Bias
The randomized, double-blind, controlled trial remains the principal design for identifying short-term efficacy. Its usefulness depends on adequate allocation concealment, comparable follow-up, prespecified outcomes, and analysis accounting for missing observations. ADHD trials confront particular measurement challenges: stimulant effects and adverse events may make active allocation recognizable; parents and teachers can observe different contexts; and symptom ratings can change with expectations about treatment. An active placebo could sometimes help assess whether recognizable adverse effects compromise masking, as discussed by Storebø et al. (2025) Storebø, O. J., Storm, M. R. O., Pereira Ribeiro, J., Skoog, M., Groth, C.,.
An open-label extension can estimate longer-period tolerability and retention among those who choose and are able to continue. It is substantially less informative for causal efficacy because placebo effects, regression to the mean, changes in clinical support, and selection by early response remain. Withdrawal designs, in which responders are randomized to continuation or cessation, answer a different question: whether continued exposure maintains an established response. They may exaggerate general applicability because only initial responders or tolerators enter the randomized phase.
Network meta-analysis synthesizes direct and indirect comparisons. It depends on transitivity: important treatment effect modifiers should be sufficiently similar across the set of trials being connected. Differences in average age, baseline severity, comorbidity, product formulation, dosing, outcome rater, or trial duration can undermine that assumption. Statistical consistency checks are useful but cannot prove that all unmeasured effect modifiers are balanced. Interpretation should acknowledge the difference between a direct head-to-head estimate and a model-derived indirect comparison, especially when new nonstimulants enter a sparse evidence network.
Measurement: Symptoms, Cognition, Function, and Quality of Life
The core symptom construct is often assessed with clinician, parent, teacher, or self-report scales. Measurement instruments must distinguish attention problems from broader behavioral difficulty and track whether scores represent total severity, inattentive symptoms, hyperactivity–impulsivity, or a response threshold. A rating measured at the medication's expected peak can differ from a full-day functional assessment. Patient and caregiver perceptions also vary with burden and treatment expectations.
Laboratory tests of sustained attention, response inhibition, reaction-time variability, working memory, and executive control provide other measurement levels. Their association with everyday functioning is imperfect. A drug may change performance on a computerized task without producing a proportionate shift in school completion, relationships, or safe driving, and the reverse may be possible. Laboratory outcomes can clarify mechanistic hypotheses; they require independent validation to serve as surrogates for meaningful clinical benefit.
Functioning includes academic attainment, organization of daily activity, participation in relationships, occupational reliability, driving safety, and quality of life. These are influenced by environmental accommodations, educational instruction, socioeconomic resources, family stress, coexisting conditions, and development. The 2025 adult synthesis by Ostinelli et al. (https://doi.org/10.1016/S2215-0366(24)00360-2) found that core-symptom effects did not straightforwardly establish additional patient-centered benefits such as quality of life. Scientific reporting therefore should never equate a reduction in a symptom score with demonstrated restoration of every important life outcome.
Developmental Stages and Treatment Choice
Age changes diagnosis, clinical priorities, pharmacokinetics, consent procedures, and the balance of available evidence. For preschool-aged children, the AAP guideline (2019) (https://publications.aap.org/pediatrics/article/144/4/e20192528/81590/Clinical-Practice-Guideline-for-the-Diagnosis) recommends evidence-based parent training in behavior management and/or behavioral classroom interventions as first-line treatment; methylphenidate may be considered under specified circumstances of persistent moderate-to-severe impairment. For school-aged children, the guideline recommends approved medication together with behavioral and educational intervention, while adolescent care includes the adolescent's assent and attention to school and transition needs. These are professional care pathways, not claims that every young person should receive the same medication.
The NICE NG87 guideline National Institute for Health and Care Excellence. (2018, updated 2019). *A recommends methylphenidate first-line for children aged five years and older and young people, then provides conditional sequencing involving lisdexamfetamine, atomoxetine, or guanfacine. For adults, it recommends methylphenidate or lisdexamfetamine initially, with further choices conditioned on adequate trials and tolerability. These recommendations reflect evidence reviews, the regulatory and clinical context of the United Kingdom, and local authorization details. They must not be silently transposed to U.S. labeling or substitute for evaluation by qualified professionals.
Adult ADHD often coexists with anxiety, depressive symptoms, sleep problems, substance-use histories, and medical conditions. Trials can exclude precisely those patients who create the greatest difficulty in ordinary care. In later adulthood, polypharmacy, cardiovascular risk, sleep architecture, and changing functional expectations create additional questions for which direct randomized evidence is comparatively limited. Age-specific estimates must be interpreted within the populations actually studied, not extended automatically across the lifespan.
Coexisting Conditions and Diagnostic Complexity
ADHD pharmacotherapy takes place in patients whose psychological and physical conditions may involve overlapping symptoms. Anxiety may impair concentration; depression can alter motivation and psychomotor speed; sleep disorders can produce inattention; bipolar episodes can be associated with impulsivity; and substance exposure can change executive performance. Successful medication research measures these coexisting phenomena without treating all distress as a direct consequence of ADHD.
The NICE guideline National Institute for Health and Care Excellence. (2018, updated 2019). *A addresses treatment choices for ADHD with anxiety disorders, tic disorders, or autism spectrum disorder and requires particular caution during acute psychotic or manic episodes. A positive short-term response in a narrowly selected trial cannot prove that the same regimen is appropriate in mania, unstable cardiovascular illness, or complicated substance-use circumstances. Clinical judgment integrates the direction of evidence with the person's actual condition and possible drug interactions.
Coexisting tics, irritability, and emotional dysregulation have often appeared as reasons to consider a nonstimulant or adjunctive treatment. Research should distinguish improvement of the ADHD syndrome from independently established benefit for a tic or affective symptom. The guanfacine adjunctive trial demonstrates changes in core ADHD scales in a partially responsive pediatric population; secondary analyses of oppositional symptoms address additional questions with their own multiplicity and generalizability limits. Nonstimulant selection should be evaluated as an evidence-based clinical choice, not as an assumed treatment for all comorbidity.
Cardiovascular Physiology and Short-Term Hemodynamic Safety
ADHD medications produce different physiological responses depending on their mechanism, formulation, and the patient’s baseline cardiovascular state. Methylphenidate, amphetamine derivatives, atomoxetine, and viloxazine can increase pulse or blood pressure in some patients. Guanfacine and clonidine can reduce pulse or blood pressure and may produce orthostatic symptoms. A change in a measured hemodynamic parameter is a physiological outcome; it is not automatically equivalent to a myocardial infarction, stroke, arrhythmia, or later cardiovascular disease. Research must preserve this distinction.
A contemporary comparative analysis, Farhat and colleagues (2025) (https://pubmed.ncbi.nlm.nih.gov/40203844/), evaluated randomized trials of several ADHD medication classes and cardiovascular measurements across pediatric and adult populations. Randomized evidence can estimate near-term changes under treatment assignment, but trials are usually too small and brief to establish the incidence of rare major cardiovascular events. The necessary interpretation is bidirectional: short-term physiological changes support the biological plausibility of concern about long-term exposure, while their clinical meaning must be evaluated through appropriately designed event-outcome research.
Product labeling and professional guidelines require attention to prior cardiac disease, family and personal cardiovascular history, current medication, blood pressure, and pulse. The NICE guideline National Institute for Health and Care Excellence. (2018, updated 2019). *A includes pretreatment physical and cardiovascular assessment and continuing monitoring. The Intuniv label Food and Drug Administration / National Library of Medicine. (2025). INTUNI particularly emphasizes slow heart rate, low blood pressure, syncope, and conduction concerns. Monitoring measures the individual trajectory over time rather than assuming that any one group-average effect can predict a specific event.
Cumulative Exposure and Cardiovascular Disease: Observational Evidence
Long-term event risks require methods beyond short randomized trials. Zhang and colleagues (2024) Zhang, L., Li, L., Andell, P., Garcia-Argibay, M., Quinn, P. D., D'Onofrio, conducted a Swedish nested case-control study within a population of 278,027 people aged 6–64 years with diagnosed ADHD or ADHD medication exposure. Over up to 14 years of follow-up, greater cumulative medication exposure was associated with greater odds of cardiovascular disease, especially hypertension and arterial disease. The authors reported an adjusted odds ratio of 1.27 (95% confidence interval 1.17–1.39) for three to five years of cumulative use compared with the reference category, and 1.23 (1.12–1.36) for more than five years. These are association estimates from the study's defined comparison, not direct proof that treatment caused each observed disease.
The dose–duration relationship is important but complex. People treated longer may differ in ADHD severity, persistence of care, access, health surveillance, comorbid conditions, and clinician selection. Some measured differences can be adjusted statistically, while unmeasured confounding and time-varying treatment decisions may remain. The investigators used a three-month lag for exposure classification to reduce reverse causation in prescribing decisions. This is a meaningful design improvement without being a substitute for randomization.
A public account of this research should avoid two equally misleading simplifications: claiming that cardiovascular harms have been excluded by reassuring short-term trials, and claiming that a register association determines a universal cardiovascular outcome for every person who continues a medicine. Evidence instead supports continued assessment of blood pressure and pulse, consideration of the evolving risk context, and the need for stronger causal studies of exposure histories, formulations, and subgroups. Longitudinal effects should be evaluated alongside potential benefits of ADHD treatment on injury and other outcomes rather than in isolation.
Sleep, Appetite, Weight, and Growth
Appetite and sleep changes are among the frequent treatment burdens reported in stimulant trials, and they can have developmental and functional consequences. Medication-associated reduced appetite may lead to lower weight gain; changes in sleep onset, duration, or subjective sleep quality can influence the very daytime attention and self-regulation that treatment intends to improve. The Cochrane review of methylphenidate (Storebø et al., 2025) Storebø, O. J., Storm, M. R. O., Pereira Ribeiro, J., Skoog, M., Groth, C., identified more nonserious adverse events, including sleep problems and appetite reduction, while also stressing the uncertainty of longer-term effects.
Growth research requires longitudinal designs that distinguish constitutional differences from treatment-associated changes. Height and weight are developmentally patterned and affected by nutrition, maturation, illness, and changing dose exposure. A treated-versus-untreated comparison can be influenced by the reasons a particular child continues or stops medication. Measures should include growth trajectories, changes in appetite, timing of exposure, and the circumstances of clinical follow-up. A single short-term change in weight cannot answer a question about adult stature.
Nonstimulant medicines are also relevant to sleep and wakefulness. Alpha-2 agonists can produce sedation and somnolence; viloxazine or atomoxetine may have different sleep-related effects. Sedation may interfere with classroom participation or safety-critical tasks, even if some behavioral ratings appear to improve. The NICE monitoring recommendations National Institute for Health and Care Excellence. (2018, updated 2019). *A provide age-related measurement intervals for height and weight and recognize the need to document adverse effects in ongoing care. An ethically complete outcome model includes sleep and nutritional functioning as primary dimensions of quality of treatment, not merely secondary tolerability notes.
Psychiatric Adverse Events, Suicidality, Mania, and Psychosis
Rare psychiatric reactions have large clinical significance despite the limited precision of short trials. Stimulant labels warn about new psychotic or manic symptoms, and the Moran et al. (2019) (https://pubmed.ncbi.nlm.nih.gov/30893533/) comparative study found higher observed new-onset psychosis risk with amphetamine relative to methylphenidate among adolescents and young adults initiating treatment. The study's design supports a clinically meaningful association with drug choice but remains sensitive to confounding and diagnostic recording. Identifying an adverse psychiatric outcome is distinct from inferring that every hallucination during treatment was caused by that medicine.
The atomoxetine U.S. prescribing information Food and Drug Administration / National Library of Medicine. (2025–2026). A includes a boxed warning about suicidal ideation in children and adolescents. Its cited pooled short-term pediatric trial data reported suicidal ideation in 5 of 1,357 treated patients, compared with none of 851 assigned placebo. Such figures describe trial observations in the populations and interval studied; they do not define the risk for every age group, person, or duration. Both the size of the observed difference and the limits of precision matter because the event is uncommon and the outcome serious. The viloxazine extended-release label Food and Drug Administration / National Library of Medicine. (2025). Qelbre similarly contains a boxed warning concerning suicidal thoughts and behaviors and warns about possible activation of mania or hypomania.
Research must distinguish adverse-event signals, causal clinical judgments, and changes in the underlying disorder. ADHD itself is associated with mental-health comorbidity and elevated risks in some populations. Observational studies of medication initiation and subsequent suicidal behavior answer different questions than randomized comparisons of medication-associated suicidal thinking over short trial intervals. Both lines of evidence may be valid within their designs. Ongoing psychiatric monitoring and rapid professional evaluation of serious new symptoms are therefore integral to safety evaluation without requiring a misleading assertion that all treatment causes or prevents suicidality.
Misuse, Diversion, Addiction, and Controlled-Substance Policy
Stimulants used for ADHD are controlled medications in the United States. Their therapeutic uses coexist with potential for misuse, diversion, and addiction, and the frequency of each outcome depends on definitions and context. Taking a prescribed medication as directed under clinical supervision, sharing it with someone else, using it without a prescription, and administering it by an unapproved route represent different exposures. Studies that collapse these categories risk confounding pharmacological risk with patterns of access, dose, and social behavior.
In May 2023, the U.S. Food and Drug Administration Food and Drug Administration. (2023, May 19). FDA requires updates to clari required standardized prescription-stimulant labeling on misuse, abuse, addiction, overdose, and the need for secure storage and monitoring. The dexmethylphenidate labeling (https://www.dailymed.nlm.nih.gov/dailymed/fda/fdaDrugXsl.cfm?setid=638f614c-bf91-4bf6-876f-bf25fa273bc4&type=display) illustrates this class warning. The presence of the warning correctly directs attention to drug-related hazards while leaving empirical research to quantify rates under particular clinical and nonmedical-use conditions.
Treatment of ADHD in a person who has a substance-use disorder requires an integrated clinical assessment. Studies need to distinguish the effect of appropriate ADHD treatment from the effects of unregulated stimulant use and the severity of comorbid addiction. Nonstimulant options may be relevant when risks or preferences weigh against a controlled medicine, but “nonstimulant” does not mean absent risk, absent interactions, or universal efficacy. A scientifically useful prescribing framework also considers storage, diversion prevention, adherence, access, and continuity of follow-up.
Drug Interactions and Pharmacogenomic Variation
Pharmacokinetic heterogeneity is a major reason nominally identical doses can yield different exposures. Atomoxetine is particularly influenced by CYP2D6 genetic metabolizer status and inhibitory comedications. Guanfacine is affected by CYP3A4 activity. Viloxazine is a strong CYP1A2 inhibitor under the Qelbree label Food and Drug Administration. (2025). Qelbree (viloxazine extended-release), creating important implications for interacting medications. These pathways can influence concentration-dependent benefits and adverse events, although the presence of a genotype or metabolic marker does not automatically validate a complete treatment-selection algorithm.
Potential drug interactions span more than enzymes. Concurrent agents that elevate blood pressure, change heart rhythm, induce sedation, or alter arousal may produce additive physiological effects. Existing cardiac and renal conditions can modify the relevant risk. Drug–drug interaction studies, labeling, and clinical outcome data provide complementary evidence: measured plasma concentrations characterize exposure, whereas actual adverse-event rates determine whether that change has clinical significance.
Pharmacogenomics can be important where a validated relationship between metabolic phenotype and exposure is established. The general claim that DNA testing can identify the ideal ADHD medication for every patient exceeds the demonstrated precision of current research. Such a claim would require replicated prediction of comparative clinical benefit and harms across diverse populations, with favorable decision outcomes in prospective trials. The correct scientific use of metabolic data is specific to the measured pathway and the question for which its validity has been established.
Pregnancy, Lactation, and Developmental Exposure
Reproductive pharmacotherapy raises questions about both maternal ADHD-related impairment and fetal or infant exposure. It also presents strong confounding by indication: people who continue treatment during pregnancy may differ systematically from those who stop in illness severity, other medicine use, access to care, and social conditions. Comparisons based only on exposed versus unexposed births can conflate the drug with these factors. Evidence must distinguish congenital and obstetric outcomes, neurodevelopmental endpoints, lactation exposure, and the effects of untreated maternal symptoms.
In a large U.S. cohort, Suarez and colleagues (2024) Suarez, E. A., Bateman, B. T., Hernandez-Diaz, S., Straub, L., McDougle, C. evaluated prescription-stimulant exposure during the second half of pregnancy and offspring neurodevelopment. After applying approaches intended to reduce confounding, the investigators found no meaningful association with the specific composite and neurodevelopmental outcomes examined for amphetamine/dextroamphetamine or methylphenidate. The result concerns those exposures and outcomes within the study design; it does not establish comprehensive safety for every obstetric consequence or every medication class.
The viloxazine label Food and Drug Administration. (2025). Qelbree (viloxazine extended-release) contains specific pregnancy warnings, while other medicines have different and changing evidence. Clinical decisions in pregnancy involve risks of both treatment and untreated impairment and require individualized professional assessment. Research priorities include better active-comparator cohorts, patient-reported functioning, prospective pregnancy registries, breastfeeding exposure data, and explicit analysis of missing outcomes. Conclusions should not be generalized from one stimulant cohort to all ADHD drugs.
Longitudinal Functioning, Injuries, and Real-World Outcomes
Longitudinal studies move beyond symptom scores to outcomes relevant to the everyday consequences of ADHD. The U.S. motor vehicle crash study by Chang and colleagues (2017) Chang, Z., Quinn, P. D., Hur, K., Gibbons, R. D., Sjölander, A., Larsson, H used within-individual comparisons of medicated and unmedicated months among insured people with ADHD. This design reduces confounding by stable personal characteristics, although changing work demands, care engagement, and other time-varying factors can remain. Its research significance is the demonstration that policy-relevant outcomes can be studied outside short symptom trials.
A Swedish observational study, Li and colleagues (2024) Li, L., Zhu, N., Zhang, L., Kuja-Halkola, R., D'Onofrio, B. M., Brikell, I., investigated ADHD pharmacotherapy initiation and mortality in people diagnosed with ADHD. It reported an association between medication initiation and lower mortality, particularly from unnatural causes. As with other observational comparisons, apparent benefits can reflect a mixture of pharmacological effects, clinical contact, differences in severity, and contextual variables. The result is informative for public health but should not be converted into an individualized survival prediction.
A later Swedish target-trial emulation (2025) (https://pubmed.ncbi.nlm.nih.gov/40803836/) examined 148,581 people aged 6–64 years with newly diagnosed ADHD, comparing medication initiation within a defined interval with noninitiation. Initiation was associated with lower rates of first suicidal-behavior events, substance misuse, transport accidents, and criminality across two years; the difference in first accidental injuries was not statistically significant. These estimates are conditional on eligibility definitions and causal modeling assumptions. They extend scientific understanding beyond symptom reduction while requiring careful interpretation of residual confounding, treatment crossover, and administrative outcome definitions.
The apparent coexistence of observational benefits for accidents or some mortality endpoints with cardiovascular risk signals is scientifically coherent. Different causal pathways and different outcome windows may operate simultaneously. A comprehensive longitudinal assessment must examine absolute baseline risks, the distribution of competing harms and benefits, adherence, time-varying exposures, and the distinction between starting treatment and remaining exposed for years.
Long-Term Follow-Up and the MTA Research Program
The MTA trial established stronger short-term effects for closely managed medication on core symptoms than certain comparison strategies, but long-term interpretation involves subsequent treatment choices and changes in participants' lives. Once the randomized treatment phase ends, treatment exposure is no longer assigned in the same way. Families and clinicians choose continuation, change drugs, modify educational supports, or discontinue according to response, preference, and changing circumstances. Later follow-up comparisons are therefore subject to selection and time-varying confounding that the original randomization does not automatically control.
The deeper methodological lesson is that a difference between groups after 14 randomized months and an association between years of later medication use and later outcomes are separate estimates. Neither alone answers whether continuous medication from childhood changes adult educational attainment or occupational trajectories. Such conclusions require explicit longitudinal causal models, rich measurements of changing exposures, and controls for the nonrandom decisions that follow the original protocol.
Longitudinal research should also address whether initial optimization translates into reliable treatment adherence and whether functional change persists when the environment changes. Transition from school to employment, changing sleep patterns, shared decision-making, or disruptions in access can alter the perceived benefit of a given medicine without any direct change in its pharmacodynamic mechanism. The long-term scientific object is a treatment trajectory embedded in a changing life course.
Adherence, Persistence, Access, and Treatment Burden
Adherence reflects the practical and psychological conditions of sustained treatment. Some individuals experience difficulty managing prescriptions, remembering doses, attending follow-up, or absorbing medication costs. Others intentionally stop because of adverse effects, preference, perceived lack of benefit, or fear of dependence. These behaviors cannot all be assigned to “noncompliance” as if they were one construct. Different patterns require different explanations, and changes in adherence affect observed treatment effectiveness.
Modified-release formulations may reduce the burden of medication administration in school or at work and alter diversion opportunities. NICE NG87 discusses these factors in its recommendation to consider once-daily modified-release stimulants. Yet a longer-acting product may be poorly suited to an individual who experiences prolonged appetite suppression or sleep disturbance. Comparative effectiveness research must therefore evaluate convenience and acceptability jointly with symptom control and adverse effects.
Access restrictions, controlled-substance refill practices, shortages, and continuity of clinical support create health-system exposures that can determine whether a pharmacologically efficacious medicine produces stable real-world benefit. These circumstances differ among jurisdictions and change over time. A review that extrapolates trial efficacy into health-system success without measuring affordability, equity, monitoring, and retention neglects a central part of the intervention as actually experienced.
Monitoring and Medication Management as Longitudinal Research
The administration of a medication is only one component of pharmacotherapy. Before and during treatment, the clinical process encompasses accurate diagnosis, explanation of expected benefits and risks, assessment of current medicines and physical health, selection of a suitable formulation, documentation of personally meaningful treatment targets, evaluation of symptom changes, and monitoring of adverse effects. Each part creates information relevant to the interpretation of later outcomes. A treatment failure after inconsistent access and insufficient monitoring cannot be equated with pharmacological inefficacy under a well-executed trial protocol.
Professional guidelines operationalize aspects of this process. The American Academy of Pediatrics (Wolraich et al., 2019) Wolraich, M. L., Hagan, J. F., Jr., Allan, C., Chan, E., Davison, D., Earls recommends systematic chronic-care follow-up and measures addressing both medication effects and educational or behavioral supports. The NICE guideline NG87 National Institute for Health and Care Excellence. (2018, updated 2019). *A requires baseline assessment of physical health and cardiovascular history before starting medication, monitoring of growth in young people, and documentation of heart rate and blood pressure before and after relevant treatment changes and at specified intervals. These are clinical recommendations based on particular reviews and service settings. The research task is to test whether implementing such monitoring reliably improves safety, adherence, functioning, or satisfaction.
Titration itself creates an adaptive intervention. Dose, timing, and formulation are altered in response to new information. A clinical trial may therefore compare a structured titration policy against another policy rather than isolated doses. Proper causal analysis would treat treatment adjustment as time-varying and partially determined by earlier outcomes. Recording only the final successful dose discards people who did not tolerate previous doses and can produce selection bias. A scientifically complete treatment record retains initial assignment, attempted modifications, perceived gains, adverse effects, discontinuations, and clinically important outcomes after stopping.
Shared decision-making is especially consequential in ADHD, where the medication is experienced during school, work, sleep, travel, and family life. The patient may prioritize reduced distractibility, sustained task completion, fewer emotional confrontations, improved driving, or less effort expended on routine organization. These priorities are not identical, and a narrow symptom scale may fail to capture them. Adolescents' participation in decisions, questions about diversion, reproductive plans, and the realities of follow-up are part of the clinical effectiveness problem. A study that does not specify the decision context cannot establish the preferred regimen for all patients.
Regulatory Authorization, Clinical Guidelines, and Evidence Boundaries
A product's U.S. Food and Drug Administration labeling states specific indications, age ranges, approved formulations, contraindications, warnings, interactions, and evidence evaluated for that product. Regulatory labels evolve when new safety or efficacy information becomes available. They should be consulted at the time of a clinical decision and do not automatically determine how another country regulates the same ingredient. A clinical guideline performs a different function: it integrates evidence, resource considerations, patient values, and a healthcare system's intended care pathway.
The distinction is visible across nonstimulants. The original U.S. atomoxetine approval and subsequent changes in boxed warning and other label information are product-specific. Intuniv is an extended-release guanfacine product with indications and interactions that differ from those of other guanfacine forms. Qelbree, an extended-release viloxazine formulation, has a boxed warning concerning suicidal thoughts and behaviors and clinically significant CYP1A2 inhibition. ONYDA XR is an extended-release clonidine suspension with its own approved administration and cessation warnings. These products cannot be treated as interchangeable on the basis of a common descriptor such as “ADHD medicine” or “nonstimulant.”
The FDA's 2023 prescription stimulant class action Food and Drug Administration. (2023, May 19). FDA requires updates to clari represents a regulatory response to safety risks involving abuse, misuse, addiction, and overdose; it is neither a randomized treatment comparison nor an estimate of overall population-level benefit. NICE NG87 National Institute for Health and Care Excellence. (2018, updated 2019). *A recommends choices and switching sequences within the United Kingdom. The AAP guideline develops pediatric management principles relevant to U.S. practice. Comparing these sources requires explicit documentation of jurisdiction, date, therapeutic purpose, and evidence generation. A review of current treatment must distinguish each source's evidentiary authority.
Confounding by Indication and the Untreated-ADHD Counterfactual
The causal question “What happens with medication?” requires a defined alternative. That alternative might be untreated ADHD, behavioral treatment alone, an alternative medicine, a modified dose, or a program combining medication with other support. The correct counterfactual depends on the research objective. A placebo-controlled study estimates a short-term treatment contrast under its trial conditions. A population study comparing patients who do and do not initiate medication estimates a different contrast and must account for reasons for treatment selection.
Confounding by indication is an especially substantial challenge: patients with more severe or persistent ADHD may be more likely to receive medication, while those with better healthcare access and consistent clinical engagement may also be more likely to receive it. Both sets of differences can affect later outcomes. Ordinary regression adjustment helps only to the extent that the relevant variables are observed and adequately modeled. Within-person designs reduce confounding by stable personal attributes but can still be affected by changing symptoms, occupation, school demands, comorbid episodes, and disruptions in care.
Target-trial emulation attempts to specify eligibility, start of follow-up, treatment strategies, outcomes, and analysis before drawing conclusions from observational data. Zhang and colleagues (2025) Zhang, L., Zhu, N., Sjölander, A., Nourredine, M., Li, L., Garcia-Argibay, used this approach to study psychiatric, accidental, and social outcomes in a large Swedish cohort. Its results offer clinically meaningful observational evidence associated with initiation, while remaining conditional on assumptions about exchangeability, measurement, and sustained treatment strategies. The subsequent published erratum The BMJ. (2025). Correction: ADHD drug treatment and risk of suicidal behav belongs to the bibliographic record and illustrates why research read-back and correction tracking matter. Neither a large sample nor a formal causal model can independently guarantee that all confounding has been eliminated.
An ethically complete causal question also accounts for the potential harms of leaving severe symptoms and impairment untreated. This consideration is essential when interpreting medication-associated cardiac risks, depressive symptoms, motor vehicle crashes, or substance use. The risk associated with treatment should be compared to the risk under a specific feasible alternative, with comparable measurement of outcomes and selection processes. Pharmacotherapy research is strongest when benefits and harms share a common causal frame rather than being presented in disconnected evidence streams.
The Difference Between Average Effects and Individual Optimization
Treatment guidelines and meta-analyses provide estimates for specified populations. An individual clinical outcome depends on baseline symptom patterns, comorbidity, formulation, adherence, exposure, metabolism, contextual demands, expectations, and cointerventions. An apparent subgroup difference may be a genuine treatment-effect modifier or the result of sampling error, multiple testing, or regression toward the mean. A model that selects a medicine from pretreatment characteristics needs external validation and ideally a prospective test of whether its recommendations improve patient outcomes.
Cross-trial medication rankings are particularly sensitive to differences in ages, baseline severity, comparison durations, dose policies, and which outcome rater is used. A medication with a higher average standardized symptom effect may carry a less favorable tolerability or retention profile for some patients. A drug with a smaller average effect may be clinically suitable when sustained exposure, cardiovascular history, preference, contraindications, or misuse risk are central. There is no coherent individual “best drug” independent of the objective function and the alternative treatments available.
The scientific path toward precision treatment requires better baseline and repeated measures, transparent prediction models, evaluation of incremental usefulness beyond routine assessment, and tests of whether personalization actually changes decisions and outcomes. Pharmacogenetic pathways can be relevant for drug exposure, particularly atomoxetine and CYP2D6, but selecting an optimal treatment from genotype alone remains a substantially broader claim than its current evidence establishes.
Evidence Hierarchy, Reproducibility, and the Quality of Conclusions
The scientific literature on ADHD pharmacotherapy contains randomized trials, systematic reviews, network comparisons, register studies, product labels, and clinical guidelines. Each answers a particular question. Randomized placebo comparisons are strong evidence for efficacy over the measured interval if allocation, concealment, follow-up, and masking are credible. Network meta-analyses add indirect comparisons but inherit biases and structural inconsistencies from their component trials. Observational studies can follow rare events and multi-year exposure, yet require assumptions about confounding and selection. Regulatory safety warnings identify hazards meriting attention but do not supply an absolute risk estimate for all exposed individuals.
The Cochrane review by Storebø et al. (2025) Storebø, O. J., Storm, M. R. O., Pereira Ribeiro, J., Skoog, M., Groth, C., and the Cortese et al. (2018) (https://doi.org/10.1016/S2215-0366(18)30269-4) network analysis are sometimes used to make apparently contradictory claims: one reports uncertain evidence quality for methylphenidate estimates, while the other finds positive relative efficacy and makes a conditional group-level ranking. Their claims address different inferential layers. A treatment can show repeated measured advantages over placebo while the exact size, functional transfer, and durability of benefit remain uncertain because masking and reporting have limitations.
The publication of the updated Cochrane review in December 2025 Storebø, O. J., Storm, M. R. O., Pereira Ribeiro, J., Skoog, M., Groth, C.,, involving 212 included trials and 16,302 randomized participants, adds a critical contemporaneous benchmark. The investigators continued to classify certainty for measured outcomes as very low, in part because adverse effects might disclose treatment assignment and many trials were short. They reported a teacher-rated symptom effect but identified inadequate evidence for durable effects and quality-of-life improvements. Separately, Gosling and colleagues' 2025 umbrella review in *The BMJ* Gosling, C. J., Garcia-Argibay, M., De Prisco, M., Arrondo, G., Ayrolles, A brought together 221 intervention-comparison-outcome meta-analytic estimates and used its own standardized evidence-grading framework; it reported moderate- or high-certainty short-term evidence for several drug groups in some populations while finding no high-certainty long-term effects. Different evidence-grading procedures, study eligibility criteria, and assessment targets explain why their strength-of-evidence statements need to be examined rather than blended into an undocumented consensus.
Reproducibility should be assessed at several levels. Independent trials can replicate a symptom contrast. Different clinical networks can reproduce the comparative ordering of groups. Register studies can reproduce a longer-term association in another healthcare system. Neither form necessarily replicates the same causal proposition. A more rigorous agenda would prespecify outcomes, publish protocols and deviations, improve reporting of adverse events and discontinuation, include active-comparator designs, use independent data sources for rare events, and support privacy-preserving access to deidentified evidence where ethically permissible.
Financial and organizational interests may influence research questions, selection of endpoints, publication, and interpretation. Disclosure of conflicts is useful but cannot replace independent replication. Likewise, independent authorship alone cannot remedy insufficient masking or a poorly measured outcome. An appropriate evidence assessment evaluates the method itself, not simply the affiliation of the researchers or the reputation of the journal.
Comparative Evidence Matrix
The following matrix summarizes the distinct functions of important sources. Its rows represent study families rather than pooled estimates generated for this monograph.
| Evidence source | Principal inferential question | Contribution | Central limit | | --- | --- | --- | --- | | MTA Cooperative Group (1999) | Comparative effects of organized pediatric treatment strategies | Randomized 14-month evidence on medication management, behavioral therapy, and combined care | Later follow-up is no longer an unchanged randomized exposure contrast | | Cortese et al. (2018) | Short-term symptom efficacy and acceptability across drug classes and age groups | Medication network meta-analysis and conditional group-level treatment ranking | Indirect comparisons, trial heterogeneity, and short duration | | Storebø et al. (2025); Gosling et al. (2025) | Quality of methylphenidate trial evidence in children and adolescents | Systematic review highlighting risk of bias and very low certainty | Does not itself identify a better alternative for each individual | | Ostinelli et al. (2025) | Adult ADHD treatment effects across intervention types | Larger, multicomponent evidence network; patient-reported and clinician-rated outcomes | Few reliable long-term, functional, or quality-of-life endpoints | | Nourredine et al. (2026) | How efficacy and adverse effects vary with dose | Dose–effect network models across pediatric and adult trials | Sparse observations and uncertain extrapolation at some doses | | Farhat et al. (2025) | Short-term cardiovascular physiological differences | Randomized network evidence on hemodynamic measures | Underpowered for rare clinical cardiovascular events | | Zhang et al. (2024) | Association of cumulative ADHD medicine use with cardiovascular disease | Large longitudinal nested case-control study | Residual confounding and nonrandom treatment exposure | | Zhang et al. (2025) | Observational associations of initiation with consequential outcomes | Emulated target-trial framework in Swedish registers | Assumptions about confounding, treatment transitions, and measurement | | U.S. FDA product labels | Product-specific authorization and safety obligations | Indication, warnings, interactions, and clinical risk information | Labels are not independent comparative-effect meta-analyses | | NICE NG87 and AAP guideline | Clinically organized management pathways | Structured diagnosis, choice, and monitoring recommendations | Recommendations reflect particular age groups and jurisdictions |
Artificial Era: Medication Information, Digital Monitoring, and Epistemic Responsibility
The Artificial Era introduces increasingly extensive artificial systems into information retrieval, appointment support, symptom recording, and shared clinical communication. For ADHD pharmacotherapy, computational systems may help detect missing outcome data, compare product-specific regulatory documents, organize longitudinal symptom measurements, and make relevant research available to clinicians and patients. These are concrete information-processing functions that can be evaluated through accuracy, completeness, timeliness, clinical usability, bias, and safety.
An artificial system that summarizes ADHD pharmacotherapy must preserve the inferential status of every claim. A randomized symptom effect should be identified as a trial estimate; an association from register data should retain its observational design; a product warning should not be translated into an invented adverse-event frequency; and a hypothetical personalization model should not be presented as validated clinical guidance. The same exactness is necessary for identifying a patient's age, formulation, concurrent medicines, and jurisdiction whenever medication information is presented in a clinical workflow.
Automated symptom tracking may increase the density of repeated observations but creates new measurement issues. A digital task result may depend on engagement, device familiarity, time of day, selective reporting, and alert fatigue. The presence of more data points does not guarantee construct validity or justify automated dose adjustment. Research should evaluate whether artificial information systems improve actual patient-centered outcomes, while maintaining accountable clinical governance, privacy, appropriate human oversight, and explicit documentation of the evidence used.
In the philosophical horizon From Homo to Artificial, the relevant scientific transition is the possibility of new nonbiological systems participating in psychological knowledge work. This does not alter the empirical object of ADHD medication trials in human populations. It creates a distinct research agenda concerning evidence integrity, clinical information representation, and accountable human–artificial collaboration.
Unresolved Scientific Problems and Research Priorities
A first priority is the longer-term causal balance of medication benefits and harms. Extended follow-up in randomized cohorts becomes increasingly complicated by crossover, changing doses, discontinuation, and other care. Large observational datasets create sufficient statistical power for uncommon outcomes but require careful control of time-varying confounding. Future designs should specify meaningful alternative treatment strategies, clinically relevant time horizons, and joint benefit–risk outcomes.
A second priority is functional transfer. Sustained attention on a controlled task, a lower symptom scale, and lasting success in educational or occupational participation are related but different observations. Trials should include outcomes that patients, caregivers, teachers, and clinicians consider meaningful and distinguish pharmacological effects from accommodations, instruction, psychotherapy, and environmental change.
A third priority is adequate representation of people often excluded from trials: adults with multimorbidity, older people, patients with complicated psychiatric conditions, pregnant patients, and diverse populations experiencing inequitable access to healthcare. Stratified analyses should avoid presenting noisy exploratory findings as validated treatment-selection rules. Culturally and linguistically valid measures are necessary to interpret symptoms and functioning across contexts.
A fourth priority concerns the safety of specific formulations and combinations over time. Few short randomized studies can reliably quantify rare cardiovascular events, serious psychiatric outcomes, the developmental consequences of exposure, or long-term outcomes after discontinuation. Independent surveillance, transparent adverse-event reporting, observational triangulation, and pharmacoepidemiological replication are essential.
A fifth priority is personalization. Predictive models should be judged by calibrated predictions of clinically important outcomes, prospective validation, fairness across populations, and whether using predictions improves actual decisions compared with high-quality usual care. Mechanistically appealing biomarkers need evidence of clinical incremental validity before becoming treatment-selection authorities.
Conclusion
Stimulant and nonstimulant pharmacotherapy for ADHD is an established field supported by a substantial clinical-trial, pharmacological, regulatory, and observational evidence base. Methylphenidate and amphetamine preparations show important short-term effects on core symptoms in well-studied populations. Atomoxetine, guanfacine extended release, clonidine extended release, and viloxazine extended release provide other evidence-based approaches with distinct exposure patterns, clinical authorization, tolerability, and monitoring needs. Differences between products are meaningful at the levels of molecular mechanism, dosing formulation, developmental evidence, physiological effects, and observed continuation.
The evidence is strongest for short-term changes in defined ADHD symptom outcomes and considerably more heterogeneous for long-term functioning, person-specific optimization, rare adverse events, and comparative durability. Clinical trials, network meta-analyses, register studies, and product warnings are complementary forms of evidence whose conclusions depend on different design assumptions. The most scientifically coherent interpretation retains both the documented benefits of symptom treatment and the safety information needed to assess treatment as an evolving clinical trajectory.
Future progress depends on more complete outcome measurement, rigorous comparative methods, long-term causal inference, independent reproducibility, and continuous attention to the daily contexts in which medication is used. The research object extends from synaptic signaling to sustained psychological functioning, physical health, patient autonomy, and healthcare systems. Accurate pharmacotherapy scholarship makes each of these levels explicit and maintains a verifiable boundary between established results, plausible mechanisms, and questions awaiting decisive study.
References
- Adler, L. A., Lieberman, V. R., Brijbasi, L., Mattingly, G. W., Yarullina, I., Li, Y., Earnest, J., & Rubin, J. (2026). Viloxazine extended release in adults with ADHD and depression and/or anxiety symptoms: Results from a decentralized, open-label, phase 4 trial. *The Journal of Clinical Psychiatry, 87*(2), 25m16234. https://doi.org/10.4088/JCP.25m16234
- Chang, Z., Quinn, P. D., Hur, K., Gibbons, R. D., Sjölander, A., Larsson, H., & D'Onofrio, B. M. (2017). Association between medication use for attention-deficit/hyperactivity disorder and risk of motor vehicle crashes. *JAMA Psychiatry, 74*(6), 597–603. https://doi.org/10.1001/jamapsychiatry.2017.0659
- Cortese, S., Adamo, N., Del Giovane, C., Mohr-Jensen, C., Hayes, A. J., Carucci, S., Atkinson, L. Z., Tessari, L., Banaschewski, T., Coghill, D., Hollis, C., Simonoff, E., Zuddas, A., Barbui, C., Purgato, M., Steinhausen, H.-C., Shokraneh, F., Xia, J., & Cipriani, A. (2018). Comparative efficacy and tolerability of medications for attention-deficit hyperactivity disorder in children, adolescents, and adults: A systematic review and network meta-analysis. *The Lancet Psychiatry, 5*(9), 727–738. https://doi.org/10.1016/S2215-0366(18)30269-4
- Farhat, L. C., Lannes, A., Del Giovane, C., Parlatini, V., Garcia-Argibay, M., Ostinelli, E. G., Tomlinson, A., Chang, Z., Larsson, H., Fava, C., Montastruc, F., Cipriani, A., Revet, A., & Cortese, S. (2025). Comparative cardiovascular safety of medications for attention-deficit hyperactivity disorder in children, adolescents, and adults: A systematic review and network meta-analysis. *The Lancet Psychiatry, 12*(5), 355–365. https://doi.org/10.1016/S2215-0366(25)00062-8
- Food and Drug Administration / National Library of Medicine. (2025–2026). Atomoxetine hydrochloride: U.S. prescribing information. *DailyMed*. https://www.dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=f266ab7b-5a68-42b5-b204-e3249bea0aed
- Food and Drug Administration / National Library of Medicine. (2025). INTUNIV (guanfacine) extended-release tablets: U.S. prescribing information. *DailyMed*. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=b972af81-3a37-40be-9fe1-3ddf59852528
- Food and Drug Administration / National Library of Medicine. (2025). Qelbree (viloxazine extended-release): U.S. prescribing information. *DailyMed*. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=aedf408d-0f84-418d-9416-7c39ddb0d29a
- Food and Drug Administration. (2023, May 19). FDA requires updates to clarify labeling of prescription stimulants used to treat ADHD and other conditions. https://www.fda.gov/drugs/drug-alerts-and-statements/fda-requires-updates-clarify-labeling-prescription-stimulants-used-treat-adhd-and-other-conditions
- Food and Drug Administration. (2024). ONYDA XR (clonidine hydrochloride) extended-release oral suspension: Prescribing information. https://www.accessdata.fda.gov/drugsatfda_docs/label/2024/217645s001lbl.pdf
- Food and Drug Administration. (2025). Qelbree (viloxazine extended-release) capsules: Prescribing information. https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/211964s013lbl.pdf
- Gosling, C. J., Garcia-Argibay, M., De Prisco, M., Arrondo, G., Ayrolles, A., Antoun, S., Caparos, S., Catalán, A., Ellul, P., Dobrosavljevic, M., Farhat, L. C., Fico, G., Eudave, L., Groenman, A. P., Højlund, M., Jurek, L., Nourredine, M., Oliva, V., Parlatini, V., . . . Cortese, S. (2025). Benefits and harms of ADHD interventions: Umbrella review and platform for shared decision making. *BMJ, 391*, e085875. https://doi.org/10.1136/bmj-2025-085875
- Li, L., Zhu, N., Zhang, L., Kuja-Halkola, R., D'Onofrio, B. M., Brikell, I., Lichtenstein, P., Cortese, S., Larsson, H., & Chang, Z. (2024). ADHD pharmacotherapy and mortality in individuals with ADHD. *JAMA, 331*(10), 850–860. https://doi.org/10.1001/jama.2024.0851
- Moran, L. V., Ongur, D., Hsu, J., Castro, V. M., Perlis, R. H., & Schneeweiss, S. (2019). Psychosis with methylphenidate or amphetamine in patients with ADHD. *The New England Journal of Medicine, 380*(12), 1128–1138. https://doi.org/10.1056/NEJMoa1813751
- MTA Cooperative Group. (1999). A 14-month randomized clinical trial of treatment strategies for attention-deficit/hyperactivity disorder. *Archives of General Psychiatry, 56*(12), 1073–1086. https://doi.org/10.1001/archpsyc.56.12.1073
- National Institute for Health and Care Excellence. (2018, updated 2019). *Attention deficit hyperactivity disorder: Diagnosis and management (NICE Guideline NG87).* https://www.nice.org.uk/guidance/ng87/chapter/Recommendations
- Nourredine, M., Jurek, L., Hamza, T., Cipriani, A., Subtil, F., Parlatini, V., Farhat, L. C., Fusetto Veronesi, G., Efthimiou, O., Salanti, G., & Cortese, S. (2026). Pharmacological interventions for ADHD: A systematic review and dose-effect network meta-analysis. *The Lancet Psychiatry, 13*(6), 485–495. https://doi.org/10.1016/S2215-0366(26)00091-X
- Ostinelli, E. G., Schulze, M., Zangani, C., Farhat, L. C., Tomlinson, A., Del Giovane, C., Chamberlain, S. R., Philipsen, A., Young, S., Cowen, P. J., Bilbow, A., Cipriani, A., & Cortese, S. (2025). Comparative efficacy and acceptability of pharmacological, psychological, and neurostimulatory interventions for ADHD in adults: A systematic review and component network meta-analysis. *The Lancet Psychiatry, 12*(1), 32–43. https://doi.org/10.1016/S2215-0366(24)00360-2
- Storebø, O. J., Storm, M. R. O., Pereira Ribeiro, J., Skoog, M., Groth, C., Callesen, H. E., Schaug, J. P., Darling, P., Huus, C.-M. L., Zwi, M., Kirubakaran, R., Simonsen, E., & Gluud, C. (2025). Methylphenidate for children and adolescents with attention deficit hyperactivity disorder (ADHD). *Cochrane Database of Systematic Reviews, 2025*(12), CD009885. https://doi.org/10.1002/14651858.CD009885.pub4
- Suarez, E. A., Bateman, B. T., Hernandez-Diaz, S., Straub, L., McDougle, C. J., Wisner, K. L., Gray, K. J., Pennell, P. B., Lester, B., Zhu, Y., Mogun, H., & Huybrechts, K. F. (2024). Prescription stimulant use during pregnancy and risk of neurodevelopmental disorders in children. *JAMA Psychiatry, 81*(5), 477–488. https://doi.org/10.1001/jamapsychiatry.2023.5073
- The BMJ. (2025). Correction: ADHD drug treatment and risk of suicidal behaviours, substance misuse, accidental injuries, transport accidents, and criminality: Emulation of target trials. *BMJ, 390*, r1827. https://doi.org/10.1136/bmj.r1827
- Wilens, T. E., Bukstein, O., Brams, M., Cutler, A. J., Childress, A., Rugino, T., Lyne, A., Grannis, K., & Youcha, S. (2012). A controlled trial of extended-release guanfacine and psychostimulants for attention-deficit/hyperactivity disorder. *Journal of the American Academy of Child & Adolescent Psychiatry, 51*(1), 74–85.e2. https://doi.org/10.1016/j.jaac.2011.10.012
- Wolraich, M. L., Hagan, J. F., Jr., Allan, C., Chan, E., Davison, D., Earls, M., Evans, S. W., Flinn, S. K., Froehlich, T., Frost, J., Holbrook, J. R., Lehmann, C. U., Lessin, H. R., Okechukwu, K., Pierce, K. L., Winner, J. D., & Zurhellen, W. (2019). Clinical practice guideline for the diagnosis, evaluation, and treatment of attention-deficit/hyperactivity disorder in children and adolescents. *Pediatrics, 144*(4), e20192528. https://doi.org/10.1542/peds.2019-2528
- Zhang, L., Zhu, N., Sjölander, A., Nourredine, M., Li, L., Garcia-Argibay, M., Kuja-Halkola, R., Brikell, I., Lichtenstein, P., D'Onofrio, B. M., Larsson, H., Cortese, S., & Chang, Z. (2025). ADHD drug treatment and risk of suicidal behaviours, substance misuse, accidental injuries, transport accidents, and criminality: Emulation of target trials. *BMJ, 390*, e083658. https://doi.org/10.1136/bmj-2024-083658
- Zhang, L., Li, L., Andell, P., Garcia-Argibay, M., Quinn, P. D., D'Onofrio, B. M., Brikell, I., Kuja-Halkola, R., Lichtenstein, P., Johnell, K., Larsson, H., & Chang, Z. (2024). Attention-deficit/hyperactivity disorder medications and long-term risk of cardiovascular diseases. *JAMA Psychiatry, 81*(2), 178–187. https://doi.org/10.1001/jamapsychiatry.2023.4294