Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Tobacco Cessation Pharmacotherapy and Neuropsychiatric Outcomes

Abstract. This fundamental research monograph synthesizes pharmacological, experimental, clinical, and regulatory evidence on medications for cessation of combustible tobacco use and their neuropsychiatric outcomes. It reconstructs pivotal trials from early bupropion and varenicline comparisons through the EAGLES safety program, reviews contemporary Cochrane syntheses, the WHO guideline, the 2026 VA/DoD recommendations, and a 2026 network meta-analysis in severe mental illness. The work differentiates medication adverse reactions, nicotine withdrawal, baseline psychiatric illness, changes caused by smoking cessation itself, and indirect exposure effects through cytochrome P450 induction. Its central result is an evidence-graded account of sustained abstinence alongside event-specific safety: the major randomized evidence does not show a significant increase in the prespecified composite of moderate-to-severe neuropsychiatric events with varenicline or bupropion relative to patch or placebo in studied stable populations, while important labeled risks, imprecise rare-event estimates, clinical heterogeneity, and monitoring needs persist. The synthesis is a critical narrative monograph with traceable primary sources and regulatory documents, not a registered systematic review, original trial, or patient-specific prescribing guide.

Keywords: tobacco cessation pharmacotherapy; varenicline; bupropion; nicotine replacement therapy; cytisine; nicotine withdrawal; neuropsychiatric adverse events; EAGLES trial; smoking cessation; serious mental illness; suicidality; CYP1A2; clozapine; smoking abstinence; pharmacovigilance; clinical trial methodology

Introduction: Clinical and Scientific Definition

Tobacco cessation pharmacotherapy is the use of medicines to increase the probability of sustained discontinuation of tobacco consumption. The present research object is the joint distribution of cessation benefit and psychological or psychiatric outcomes during treatment, withdrawal, and follow-up. The central outcome is abstinence from combustible tobacco, preferably sustained for at least six months and supported by biochemical verification. The safety outcomes encompass affective deterioration, agitation, anxiety, sleep disturbance, hostility, mania, psychosis, suicidal ideation and behavior, severe adverse events, and treatment discontinuation. These outcomes require separate definitions, observation windows, severity gradings, and ascertainment procedures. A single measure called tolerability cannot capture their clinical differences. Anthenelli et al. (2016) World Health Organization (2024) Myers et al. (2026)

Interpretation requires a causal framework that includes the pharmacological action of the cessation medicine, reduced nicotine exposure, loss of the conditioned cigarette-smoking routine, baseline mental-health status, psychosocial circumstances, and changes in exposure to other medicines after combustion ceases. The tobacco smoke constituent responsible for an important class of drug interactions is not nicotine itself: combustion-derived polycyclic aromatic hydrocarbons induce hepatic enzymes, particularly CYP1A2. Thus an episode of sedation after quitting may reflect altered clozapine clearance rather than a direct adverse effect of nicotine replacement or varenicline. A scientifically adequate safety inquiry follows exposure, chronology, co-medications, and diagnostic history. MHRA (2014) DailyMed (2025)

This work integrates historical randomized trials, the multinational EAGLES program, subsequent subgroup analysis, systematic-review estimates, contemporary guidelines, and current product labels. Primary questions concern comparative abstinence, whether the therapies increase prespecified neuropsychiatric events, how well findings extend to severe or unstable psychiatric conditions, and which methodological uncertainties remain. The method is a structured critical narrative synthesis rather than a prospectively registered systematic review; no patient-level data or new effect estimates are generated. Regulatory cautions are represented according to their actual labels instead of interpreted as estimates of incidence. Anthenelli et al. (2016) Bekkering et al. (2026) World Health Organization (2024)

Scope, Constructs, and Outcome Taxonomy

Nicotine dependence involves pharmacological adaptation and repeatedly reinforced behavior in environmental contexts. Cessation changes exposure to an addictive substance and disrupts routines tied to attention, stress regulation, social interaction, work transitions, and affect regulation. A pharmacological intervention can attenuate receptor-level withdrawal while a person continues to experience cue-induced craving, conditioned automatic behavior, or grief over a perceived coping ritual. These different processes explain why a medicine's effect on cravings does not automatically resolve all cessation difficulties and why counseling remains an important co-intervention. World Health Organization (2024) Myers et al. (2026)

An abstinence outcome needs an explicit reference period. Seven-day point prevalence answers whether smoking occurred in a short interval near follow-up; continuous abstinence from a designated quit date or treatment week is a stricter endpoint. Prolonged abstinence allows specified grace periods under some definitions. Carbon monoxide monitoring is useful for recent combustible exposure but has a short detection window, whereas cotinine reflects nicotine exposure and may remain positive during legitimate replacement therapy. Rates derived from divergent biochemical rules should not be pooled as if they represent identical estimands. Missing visits often receive a smoking classification, which can protect against optimistic attrition assumptions but can also obscure differences between missingness and confirmed relapse. Jorenby et al. (2006) Anthenelli et al. (2016) Courtney et al. (2021)

A neuropsychiatric adverse event is an observed symptom or clinically defined syndrome during exposure; attribution to treatment is an additional inference. EAGLES specified a composite of moderate and severe events meeting particular severity thresholds, while labels additionally warn about rare events reported after authorization. The composite has a different interpretation from insomnia, vivid dreaming, seizure, clinically diagnosed depression, or completed suicide. Psychiatric history defines a key effect-modifier candidate, and stability of illness at entry determines the population to which an experiment applies. Anthenelli et al. (2016) Evins et al. (2019) DailyMed (2026), bupropion

Historical Evolution of Cessation Medicines

Early nicotine-replacement studies investigated whether separating nicotine delivery from combustible tobacco could reduce withdrawal and facilitate abstinence. Transdermal delivery supplied nicotine at a comparatively steady level and permitted placebo-controlled testing alongside behavioral intervention. The design established an enduring experimental distinction: a drug can replace the active addictive compound without carrying smoke combustion exposure, while successful cessation remains dependent on behavior and context. Later trials compared patch, short-acting formulations, and combined regimens to learn whether baseline suppression of withdrawal and episodic control of cravings are complementary. Theodoulou et al. (2023) World Health Organization (2024)

The antidepressant bupropion became an important non-nicotine comparator. The 1997 placebo-controlled trial by Hurt and colleagues studied sustained-release bupropion while excluding people with current depression, a design choice that matters for generalization to active psychiatric disease. A subsequent factorial-style comparison of bupropion, patch, both, and placebo in 1999 expanded the comparative therapeutic question. These trials established clinically meaningful cessation effects and brought monoaminergic mechanisms into the study of tobacco dependence while leaving questions about seizure vulnerability, manic switching, and neuropsychiatric reactions that required dedicated safety assessment. Hurt et al. (1997) Jorenby et al. (1999) DailyMed (2026), bupropion

Varenicline introduced direct targeting of nicotinic acetylcholine receptors through partial agonism. Large 2006 randomized comparisons found greater smoking cessation efficacy than placebo and generally better outcomes than bupropion at relevant trial endpoints. The 2010s brought heightened scrutiny of spontaneous psychiatric reports, a randomized EAGLES trial designed around safety, and revised U.S. labeling in December 2016. The later evidence base also expanded to cytisine and maintenance strategies in psychiatric cohorts. This trajectory illustrates how drug innovation, signal detection, experimental safety adjudication, and regulatory review inform one another. Gonzales et al. (2006) Jorenby et al. (2006) Anthenelli et al. (2016) FDA (2016)

Mechanistic Pathways: Nicotinic Signaling and Reward

Nicotine activates nicotinic acetylcholine receptors distributed across central and peripheral circuits, including networks involved in reinforcement, attention, arousal, and autonomic regulation. Repeated intermittent delivery through smoking produces adaptations in responsiveness and learned associations with sensory cues and situational contingencies. After cessation, the decline in nicotine stimulation can be experienced as irritability, concentration difficulty, restlessness, negative affect, and craving. The association between withdrawal and particular clinical symptoms is probabilistic: shared phenomenology does not establish that every mood symptom has the same cause. A longitudinal assessment should register baseline levels, quit timing, cigarette exposure, and co-occurring conditions. Livingstone-Banks et al. (2023) Taylor et al. (2021)

Varenicline acts as a partial agonist at α4β2-containing nicotinic receptors, supplying receptor activity while competing with nicotine, a pharmacological rationale for both withdrawal attenuation and reduced reward from cigarettes smoked during a quit attempt. Its clinical benefit was tested directly in comparative randomized trials rather than inferred solely from receptor pharmacology. Receptor occupancy, cigarette satisfaction, craving scores, and sustained abstinence represent different links in a causal sequence. Effectiveness depends on exposure, adherence, counseling, and the broader social and motivational environment. Gonzales et al. (2006) Jorenby et al. (2006) Livingstone-Banks et al. (2023)

Cytisine is another nicotinic receptor partial agonist with a distinct history of access and treatment schedules. Its pharmacological kinship with varenicline makes it a plausible cessation medicine, but comparable receptor activity does not demonstrate equal efficacy under every regimen. The 2021 randomized noninferiority study compared a 25-day cytisine course with an 84-day varenicline course and did not establish the prespecified noninferiority criterion. This result applies to those schedules and that trial design, and should be interpreted together with broader placebo-controlled evidence and international differences in formulation availability. Courtney et al. (2021) Livingstone-Banks et al. (2023) World Health Organization (2024)

Bupropion and the Limits of a Simple Antidepressant Explanation

Bupropion has pharmacological activity affecting catecholaminergic signaling and other pathways relevant to nicotine dependence. Its cessation effect is supported by randomized comparisons, yet the effect cannot be reduced to treating depression. The original placebo-controlled cessation trial excluded current depression; later clinical evidence includes patients with varied histories and psychiatric diagnoses. The measurable benefits involve abstinence, craving, and treatment continuation within a multifactorial smoking-cessation process. The Cochrane review of antidepressants for smoking cessation identifies bupropion as an efficacious agent while evaluating both adverse events and uncertainty about rare serious outcomes. Hurt et al. (1997) Hajizadeh et al. (2023)

Bupropion is pharmacologically and clinically distinguished by a dose-related seizure risk, activating adverse effects such as insomnia, and the need to consider bipolar disorder, eating-disorder history, other seizure-threshold-lowering medicines, and abrupt discontinuation of alcohol or sedative drugs. The safety profile belongs to a specific formulation and clinical context; the fact that the same active ingredient appears in antidepressant products does not make all branded indications, dosing instructions, or regulatory warnings interchangeable. Drug interactions and individual risk assessment are central parts of interpretation. DailyMed (2026), bupropion

Some adverse events occurring during a bupropion-assisted quit attempt may be associated with withdrawal, underlying illness, coincident treatment changes, or direct drug action. The existence of postmarketing cases warrants vigilance while randomized comparison supplies a more controlled estimate of common and prespecified events in eligible participants. These two forms of evidence address different questions. The FDA's 2016 reassessment changed specific smoking-cessation neuropsychiatric boxed-warning language while maintaining relevant other cautions, including the antidepressant-class suicidality warning applicable to bupropion products. FDA (2016) Anthenelli et al. (2016) DailyMed (2026), bupropion

Nicotine Replacement as a Family of Interventions

Nicotine replacement therapy (NRT) includes transdermal patches and faster-acting oral or other delivery systems. The patch offers relatively continuous exposure; a lozenge or gum can be used to respond to situational craving. These pharmacokinetic profiles are clinically meaningful because withdrawal has both persistent and episodic components. Treatment comparisons should identify whether the control is placebo, a single form of nicotine replacement, combined NRT, or another pharmacotherapy. Studies of delivery mode cannot be transferred mechanically to all products and use contexts. Theodoulou et al. (2023) World Health Organization (2024), evidence

The 2023 Cochrane review of NRT delivery found that using a fast-acting form alongside a patch improves smoking cessation compared with a single NRT form. The WHO guideline reports a high-certainty relative risk of 1.25 with a 95% confidence interval of 1.15–1.36 across fourteen studies and 11,356 participants for combination versus single NRT. This is a comparison within nicotine replacement, not a comparison with varenicline. Differences in adherence, oral formulation tolerability, baseline consumption, and access can alter the absolute benefit in particular populations. Theodoulou et al. (2023) World Health Organization (2024), evidence

Safety evaluation includes sleep disturbance, vivid dreams, nausea, palpitations, skin reactions, accidental poisoning, and contraindication or caution language for particular products. These are qualitatively different events from a prespecified severe neuropsychiatric composite. Current U.S. patch labeling instructs users to seek help for features of nicotine overdose and warns about potentially dangerous residual nicotine in used patches. Product information is useful for risk characterization even though its existence does not by itself yield comparative event incidence. DailyMed (2026)

Behavioral Co-Intervention and the Meaning of Medication Effect

Pharmacotherapy is typically delivered in an environment that includes brief counseling, quit planning, monitoring, or structured behavioral support. EAGLES participants received brief cessation counseling at visits in all randomized arms. Therefore the estimated medication effect is an incremental difference within a background of shared support, not an estimate of medication operating in complete social isolation. Historical placebo-controlled trials similarly embedded follow-up and behavioral contact. This design promotes comparability between treatments but sets a context for translating effects to settings with scarce clinician time and uneven adherence. Anthenelli et al. (2016) Jorenby et al. (1999) World Health Organization (2024)

Counseling addresses the associative and practical components of tobacco use: identifying environmental triggers, managing high-risk situations, planning for lapses, negotiating social support, and sustaining motivation across repeated attempts. A medication may reduce withdrawal intensity yet a person may resume smoking in response to social exposure or stress. Conversely, strong counseling may support abstinence even when a preferred medication is contraindicated or unavailable. Clinical recommendations favor offering both evidence-based medication and behavioral assistance while adapting intensity to patient circumstances. World Health Organization (2024) Myers et al. (2026)

The integration of psychological and pharmacological approaches is especially relevant when cessation alters mood, sleep, and psychiatric medication exposure. A scheduled review of symptoms provides a temporal account of drug exposure, smoking status, early withdrawal, and life events. It also allows a clinician to distinguish expected transient experiences from red flags, and to modify support if the person has another attempt after a lapse. This is a model for research and clinical interpretation, not an individual prescription protocol. Myers et al. (2026) Anthenelli et al. (2019)

Comparative Efficacy: The Experimental Evidence

Comparative cessation trials have progressively moved from testing whether one medicine improves quitting over placebo to head-to-head assessments of different treatment classes and combinations. In the 2006 varenicline registration-era trials, the main prespecified efficacy outcome was a carbon-monoxide-verified period of continuous abstinence late in treatment, with longer follow-up to assess persistence. In the Jorenby trial, 1,027 participants were randomized and week 9–12 continuous abstinence was 43.9% with varenicline, 29.8% with bupropion, and 17.6% with placebo; these are trial-specific short-term outcomes, not general lifetime probabilities of quitting. The subsequent follow-up narrowed absolute differences because relapse remained common. Jorenby et al. (2006) Gonzales et al. (2006)

An evidence hierarchy matters for comparing drugs. A single favorable trial may depend on selection, adherence, intensity of support, outcome definition, and sponsor; a pooled synthesis asks whether a pattern persists across settings. Livingstone-Banks and colleagues evaluated nicotine receptor partial agonists in randomized trials with at least six months of cessation follow-up. Hajizadeh and colleagues reviewed antidepressant agents, including bupropion, using separate efficacy and safety inclusion criteria. The WHO guideline integrated these reviews and assessed certainty of benefit, harms, acceptability, cost, and feasibility, recommending varenicline, NRT, bupropion, and cytisine as pharmacological options. Livingstone-Banks et al. (2023) Hajizadeh et al. (2023) World Health Organization (2024), evidence

One practical interpretation of the comparative evidence is that varenicline generally yields greater long-term quit success than bupropion or single-product NRT under studied conditions. Combination NRT improves upon a single NRT form and can serve as an important comparator rather than an inferior default. Treatment selection cannot be inferred from ranking alone: histories of seizures, product tolerability, interaction risks, previous attempts, cost, preference, and concurrent medicines affect realized benefits. The 2026 VA/DoD guideline endorses approved cessation pharmacotherapies, favors varenicline over other single agents, and favors combination over single-product NRT. World Health Organization (2024), evidence Myers et al. (2026)

Duration, Relapse Prevention, and Adaptive Treatment

An initial quit attempt is a dynamic course with changing risk. Physiological withdrawal is most salient near cessation, but learned craving and habitual exposure to tobacco-related contexts can extend far beyond the initial treatment interval. Clinical trials frequently distinguish treatment-phase continuous abstinence from abstinence during drug-free follow-up. A medicine that improves early abstinence may still require sustained support because relapse represents a recurring feature of dependence, not a single isolated decision. Jorenby et al. (2006) Evins et al. (2014)

The randomized maintenance study by Evins and colleagues examined people with schizophrenia or bipolar disorder who had already achieved initial abstinence during an open-treatment phase. Randomization to maintenance varenicline versus placebo occurred only after that selection. The finding of improved prolonged abstinence therefore speaks most directly to maintenance among initially responsive participants with serious mental illness, not to all unselected smokers with active psychosis or mania. The design is particularly informative because it addresses treatment continuation in a high-risk population, with psychiatric monitoring and accompanying behavioral treatment. Evins et al. (2014)

An alternative strategy is to support reduction before a person is ready to attempt full abstinence. The randomized varenicline reduction-to-quit trial examined people who were willing to reduce smoking with the eventual goal of quitting, expanding the range of behavioral states that can be studied. The 2026 VA/DoD approach likewise considers medication as a tool for making changes when a person is not prepared to set an imminent quit date. These approaches require careful specification of intention, baseline smoking, eventual abstinence, and assessment time; simply reducing daily cigarette counts is not interchangeable with long-term cessation. Ebbert et al. (2015) Myers et al. (2026)

Combination Pharmacotherapy: Incremental Benefit and Added Exposure

Combination regimens attempt to address more than one aspect of nicotine dependence or increase the strength of treatment when a single intervention is insufficient. The strongest routinely recommended example is pairing a nicotine patch with a fast-acting NRT product. A patch may reduce sustained withdrawal while a second formulation addresses breakthrough craving. Evidence supporting the combination comes from comparisons with single-form NRT; it does not establish that every pair of cessation medicines has equivalent safety or incremental efficacy. Theodoulou et al. (2023) World Health Organization (2024), evidence

The randomized 2014 trial by Ebbert and colleagues compared varenicline plus sustained-release bupropion with varenicline plus placebo. Evidence of benefit depended on outcome interval and participant characteristics, while some adverse symptoms occurred more often with combined exposure. A combination adds pharmacological complexity, prescribing cautions, and often cost, so a favorable estimate for a selected group cannot be generalized as the preferred first regimen for everyone. WHO's 2024 evidence profiles describe potential added efficacy for bupropion plus varenicline but imprecise serious-adverse-event estimates. Ebbert et al. (2014) World Health Organization (2024), evidence

Older combinations of bupropion and nicotine patch illustrate the need to examine the actual comparator rather than assume pharmacological additivity. If combination treatment exceeds placebo, that result alone does not establish that it exceeds the more effective monotherapy. Precision, trial population, definitions of adherence, and safety exclusions determine whether the interaction itself is estimable. Combination interventions may also be accompanied by particularly intensive counseling, making the incremental value of each component dependent on the design. Jorenby et al. (1999) Hajizadeh et al. (2023)

Cytisine, International Treatment Traditions, and Regulation

Cytisine has a longer history of use in several countries than its recent prominence in international guidelines suggests. The 2014 randomized comparison with nicotine replacement in New Zealand and the 2021 Australian comparison with varenicline brought the treatment into modern evidence discussions. Differences in dosing schedules, follow-up, control conditions, and adverse-event ascertainment limit a simple declaration of class equivalence among nicotinic partial agonists. A drug's international use and inclusion in a WHO guideline do not automatically mean that a particular formulation is licensed in every national jurisdiction. Walker et al. (2014) Courtney et al. (2021) World Health Organization (2024)

In the Courtney randomized trial, 1,452 adult daily smokers were assigned to standard short-course cytisine or standard-course varenicline with standard telephone support offered to both groups. Biochemically verified six-month continuous abstinence at seven months was 11.7% versus 13.3%, respectively. The prespecified noninferiority margin was not met, and the comparison also involved unequal treatment duration. The trial is valuable as a direct contemporary head-to-head experiment, but regulatory and clinical interpretation depends on how those regimen differences are judged. Courtney et al. (2021)

The Cochrane nicotine-receptor partial-agonist review found persuasive efficacy evidence for varenicline and an important, growing evidence base for cytisine while identifying uncertainty in some direct comparisons. WHO recommends cytisine as an option based on moderate certainty and emphasizes implementation context, national formulary availability, and behavioral support. For U.S. readers, eligibility and product availability must always be verified against current FDA authorization at the time of actual care; a general monograph can distinguish international science from local permission without implying approval that has not been verified. Livingstone-Banks et al. (2023) World Health Organization (2024), evidence

Heterogeneity of Benefit Across Participants

People entering cessation studies differ in duration of smoking, dependence severity, psychiatric symptoms, previous treatment experience, motivation, concurrent substance use, age, and social circumstances. Randomization balances measured and unmeasured factors on average within a sufficiently large trial; it does not erase this heterogeneity. A treatment mean from a trial of motivated volunteers may not represent outcomes among patients experiencing acute housing instability, active suicidality, severe uncontrolled psychosis, or fragmented access to follow-up. Evidence translation should distinguish statistical interaction, plausible clinical modifiers, and untested subgroup narratives. Anthenelli et al. (2016) Bekkering et al. (2026)

Trial eligibility is especially important in pharmacotherapy safety. EAGLES enrolled people with and without psychiatric disorders but required psychiatric stability in its psychiatric cohort, limiting extrapolation to severe acute episodes. Clinical trial exclusion of seizure disorders and other relevant contraindications may be necessary for participant protection but leaves the highest-risk patients outside the denominator used for safety estimates. Observational registries can broaden representation, while their nonrandom treatment allocation introduces confounding by illness severity and prescribing judgment. Anthenelli et al. (2016) DailyMed (2026), bupropion

Variation in smoking context also alters pharmacokinetic effects unrelated to cessation medication. Someone taking clozapine, theophylline, or another sensitive CYP1A2 substrate may undergo a measurable exposure shift when combustible smoking stops. People taking no such medicines do not share the same interaction hazard. Stratifying by current medicines, psychiatric stability, and smoking history is therefore more meaningful than treating a psychiatric diagnosis as the only determinant of monitoring intensity. MHRA (2014) Myers et al. (2026)

Interpreting Systematic Review and Network Meta-Analysis

A conventional pairwise meta-analysis combines studies of a specified treatment contrast. A network meta-analysis additionally uses indirect comparisons connected through a network of shared comparators, under assumptions about similarity and consistency. These methods may sharpen the evidence base for long-term abstinence, but heterogeneity of counseling intensity, smoking definitions, participant severity, and drug formulation can threaten transitivity. A ranking probability is not a clinical effect size, and a high-certainty result for abstinence should not be transferred to a rare safety endpoint without independent evaluation. World Health Organization (2024), evidence Bekkering et al. (2026)

The 2026 BMJ Medicine review by Bekkering and colleagues included 74 randomized trials involving 11,023 participants with severe mental illness. The review reported an absolute increase of about ten additional long-term quitters per hundred receiving varenicline versus placebo or minimal care (95% confidence interval five to sixteen) and about five additional quitters per hundred for bupropion (95% confidence interval one to ten). This population-specific evidence materially extends confidence that treatment benefit is not restricted to people without psychiatric disorders. At the same time, the authors described serious-adverse-event estimates as highly uncertain, and identified risk of bias in most included trials. Bekkering et al. (2026)

The most defensible clinical synthesis retains outcome-specific grading: substantial certainty for comparative cessation efficacy; moderate or limited certainty for some subgroup benefits; and weaker evidence for uncommon, heterogeneous, or long-latency neuropsychiatric harms. A nonsignificant drug-placebo comparison with a wide confidence interval supports a different statement from proven equivalence. Review tables should report uncertainty and reference populations, rather than convert statistical nonsignificance into the absence of possible harm. Bekkering et al. (2026) Thomas et al. (2015)

Evaluating Absolute and Relative Benefit

Relative risks, odds ratios, risk differences, and numbers needed to treat answer related but distinct questions. A relative measure can appear impressive while the absolute benefit changes substantially with baseline quit probability. An odds ratio is not numerically interchangeable with a risk ratio when outcomes are common, especially near end-of-treatment periods. Numbers needed to treat depend on follow-up time and the underlying comparison group. A clinical interpretation should preserve the exact source population and outcome duration before presenting an NNT as if it were a universal property of a medicine. World Health Organization (2024), evidence Jorenby et al. (2006)

WHO's 2024 evidence chapter estimates an NNT around 7.6 for an additional person attaining abstinence at six months or later with varenicline compared with placebo, and approximate NNTs of sixteen to twenty-two when comparing varenicline with bupropion or single-form NRT. These estimates are useful summaries of evidence in the referenced comparisons, rather than deterministic expectations for individuals. The same chapter notes higher rates of nausea, abnormal dreams, insomnia, and other tolerability outcomes that need to be assessed alongside benefit. World Health Organization (2024), evidence

Clinical net benefit is a multidimensional calculation involving the consequences of continuing tobacco use, medication-specific risks, the probability of adherence, potential psychiatric vulnerability, and exposure to other medicines. A favorable population average supports offering treatment, while the individual selection process remains sensitive to contraindications and preferences. Research should measure treatment acceptance and persistence as outcomes in their own right, because an efficacious medicine cannot produce its trial effect when its regimen is inaccessible or abandoned early. Myers et al. (2026) World Health Organization (2024)

The EAGLES Trial: Design and Population

The EAGLES study was conceived to investigate a specific unresolved safety question after reports of mood, behavior, and thinking changes during smoking-cessation treatment. Anthenelli and colleagues conducted a randomized, double-blind, triple-dummy, active- and placebo-controlled comparison of varenicline, bupropion, nicotine patch, and placebo, combining a twelve-week treatment course with additional follow-up. The design allowed the comparison of two non-nicotine medications with nicotine replacement and with absence of active pharmacotherapy under similar clinical contact. Both psychiatric and nonpsychiatric cohorts were included, making it unusually informative about a population frequently absent from earlier efficacy trials. Anthenelli et al. (2016) FDA (2016)

Of 8,144 randomized participants, 4,116 belonged to the psychiatric cohort and 4,028 to the cohort without psychiatric disorders. The analysis sets differed by outcome: the safety analysis included participants who received treatment, while the efficacy analysis included randomized participants. The primary safety outcome was a prespecified composite of moderate and severe neuropsychiatric adverse events, not an unrestricted count of every psychological symptom. The psychiatric cohort consisted of people meeting trial stability criteria, a feature that strengthens internal consistency while limiting extension to acute, poorly controlled, or unstable psychiatric illness. Anthenelli et al. (2016) Evins et al. (2019)

The trial also captured abstinence efficacy, including carbon-monoxide-confirmed continuous abstinence at weeks nine through twelve. It thereby linked safety and benefit within a common randomized framework. The study was industry-sponsored, with investigators from the medication manufacturers and academic institutions. Conflict-of-interest disclosure, endpoint choice, access to raw adverse-event classifications, and trial-registration transparency matter when interpreting its conclusions. Regulatory decisions should be traced to both the experiment and the FDA's independent assessment rather than to either source alone. Anthenelli et al. (2016) FDA (2016)

EAGLES Neuropsychiatric Findings and Their Interpretation

In the cohort without psychiatric disorders, moderate or severe composite events occurred in 1.3% of varenicline recipients, 2.2% of bupropion recipients, 2.5% of nicotine-patch recipients, and 2.4% of placebo recipients in the safety analysis. The corresponding varenicline-minus-placebo risk difference was −1.28 percentage points (95% confidence interval −2.40 to −0.15), and the bupropion-minus-placebo difference was −0.08 points (95% confidence interval −1.37 to 1.21). These source-specific estimates provide evidence against a large generalized increase in the composite with either drug among participants satisfying the trial criteria. They do not prove an absolute absence of risk for every constituent symptom or a very rare adverse outcome. Anthenelli et al. (2016)

The psychiatric cohort had higher event rates overall across treatment groups than the nonpsychiatric cohort. The varenicline-minus-placebo risk difference for the composite in the psychiatric cohort was about 1.59 percentage points (95% confidence interval −0.42 to 3.59) and the bupropion-minus-placebo difference about 1.78 points (95% confidence interval −0.24 to 3.81). These intervals include both little or no increase and clinically relevant positive differences. Reporting only that a treatment difference failed a significance test would discard important information about precision. The authors found no statistically significant increase in the primary safety outcome attributable to either agent relative to patch or placebo in the studied cohorts. Anthenelli et al. (2016) DailyMed (2025)

EAGLES also observed greater cessation efficacy for active medications than placebo. Varenicline generally yielded the most favorable abstinence outcomes, and participants in both psychiatric and nonpsychiatric cohorts could benefit. The study thus transformed a debate formerly dominated by spontaneous safety reports into a quantitative comparison of benefit and harm. Its interpretive boundaries remain important: the prespecified composite, treatment period, eligible psychiatric stability, trial monitoring, and number of rare severe outcomes determine the strength of the conclusion. Anthenelli et al. (2016) FDA (2016)

Psychiatric Subgroups: Psychotic, Anxiety, and Mood Disorders

The secondary EAGLES analysis by Evins and colleagues examined 4,092 participants with psychotic, anxiety, or mood disorders, including 390 in the psychotic-disorder group, 792 in the anxiety-disorder group, and 2,910 in the mood-disorder group. This stratification is scientifically useful because pooling all psychiatric diagnoses into a single risk class may hide very different symptom trajectories, drug regimens, and baseline vulnerabilities. However, smaller subgroups have reduced ability to estimate uncommon severe reactions and can generate unstable estimates when events are sparse. Evins et al. (2019)

In these diagnostic groups, the observed incidence of the prespecified neuropsychiatric composite varied, but the secondary analysis found no statistically significant increase with varenicline or bupropion relative to nicotine patch or placebo. The absence of a detected interaction should be evaluated alongside smaller subgroup sizes and selection for psychiatric stability. The diagnosis-stratified analysis is valuable convergent evidence, but it is not a new independent randomized trial: its participants are drawn from the parent EAGLES population and must not be counted again as additional independent participants in a quantitative synthesis. Evins et al. (2019) Anthenelli et al. (2016)

An important translation question is whether treatment can be offered within psychiatric care without automatically assuming that smoking is a necessary stabilizing behavior. The evidence supports offering effective cessation assistance to many people with stable psychotic, anxiety, and mood conditions while monitoring symptom trajectories and interactions with established treatment. This inference coexists with legitimate concerns about recent suicidality, mania, severe substance withdrawal, and treatment changes that were not adequately represented by the trial's enrollment. Evins et al. (2019) Myers et al. (2026)

Independent Safety Syntheses and the Problem of Rare Events

The 2015 systematic review by Thomas and colleagues examined randomized comparisons of varenicline with placebo for neuropsychiatric outcomes. Its analysis preceded the large EAGLES safety trial and helped clarify which outcomes had convincing evidence for elevated incidence and which conclusions were limited by rarity or inconsistent ascertainment. Postmarketing reports are designed to raise alarms about possible adverse reactions; randomized trials with systematic outcome collection improve estimates of comparative incidence but may still be too small to measure events with very low baseline frequency. Thomas et al. (2015) Anthenelli et al. (2016)

One difficulty is that safety outcomes often differ in their operational definitions. A report of abnormal dreams, a visit for anxiety, a validated suicidal ideation score, an emergency admission, and completed suicide should not be compressed into one interchangeable event class. Composite outcomes may improve power while sacrificing interpretability if common mild elements dominate or if threshold rules shift across trials. Retrospective data extraction from efficacy studies can also generate systematic underrecognition of adverse phenomena that were not actively solicited. Thomas et al. (2015) DailyMed (2025)

The 2026 systematic review in severe mental illness included seventy-four randomized trials yet described the serious adverse event evidence as highly uncertain. Most included studies had some judged risk of bias, and numerous trials lacked enough serious events for precise estimates. This demonstrates a persistent asymmetry: common cessation benefits can be quantified relatively well, whereas rare safety harms and particular vulnerable subgroups may remain uncertain despite many total participants. The responsible synthesis reports both findings simultaneously. Bekkering et al. (2026)

Nicotine Withdrawal, Baseline Disorders, and Temporal Confounding

Nicotine withdrawal can include irritability, anxiety, dysphoria, restlessness, difficulty concentrating, appetite changes, and sleep disruption. Some of these manifestations overlap with depressive, anxiety, and other psychiatric symptoms. Consequently, a new complaint appearing immediately after stopping smoking may reflect nicotine withdrawal, an adverse medicine effect, destabilization of existing illness, or several mechanisms simultaneously. Time series of smoking exposure, medication initiation, symptom onset, and response to continued abstinence are needed to distinguish these possibilities; symptom occurrence alone rarely establishes causal assignment. Anthenelli et al. (2019) Taylor et al. (2021)

Psychiatric vulnerability is also related to clinical context. People with higher baseline anxiety or more recent psychiatric symptoms may have an elevated background probability of experiencing a clinically important event during any quit attempt. EAGLES predictor analyses considered characteristics associated with events across treatment groups, drawing attention to the predictive value of patient history rather than a simple one-drug explanation. Prediction is not itself proof of mechanism: the presence of risk factors could reflect a shared upstream vulnerability. Anthenelli et al. (2019)

A research design capable of separating these pathways would repeatedly measure craving, sleep, medication exposure, alcohol and other substance use, acute life stress, psychiatric symptoms, and biochemical confirmation of smoking. It would compare trajectories rather than only binary endpoint counts, predefine adverse-event thresholds, and document competing causes. Such designs could estimate whether the temporal pattern of a symptom is linked to pharmacological exposure, reduction in smoking, or preexisting illness, while retaining uncertainty when multiple pathways converge. Anthenelli et al. (2016) Thomas et al. (2015)

Mental Health After Smoking Cessation

The effect of quitting smoking on longer-term mental health is a distinct question from immediate adverse drug reactions. The Cochrane review by Taylor and colleagues synthesized longitudinal evidence linking cessation with subsequent psychological outcomes, reporting that cessation was not associated with deterioration in mental health and that small to moderate improvements were observed in several outcomes. These studies are not all randomized trials assigning participants to quit or continue smoking, so temporal confounding, selection, and reverse causation require careful interpretation. Taylor et al. (2021)

The review found associations with improvements in anxiety, stress, positive affect, and aspects of psychological quality of life, including analyses among people with mental health conditions. The certainty varied across outcomes: low or very low for some symptom domains and moderate for others. These findings challenge a simple assumption that long-term mental stability depends on continued smoking, but they do not guarantee immediate mood improvement for each quitter. Withdrawal can be difficult and persistent life stress can coexist with later improvement. Taylor et al. (2021)

A rigorous longitudinal account distinguishes the early transition from nicotine dependence, the medium-term stabilization of routines, and longer-term mental-health trajectories. Interventions may have different effects at each interval. A patient experiencing transient irritability during initial abstinence can subsequently report better well-being, while another requires active psychiatric assessment because symptoms are severe or persist. The evidence warrants a temporal rather than binary conception of psychological outcome. Taylor et al. (2021) World Health Organization (2024)

Severe Mental Illness: Efficacy, Safety, and Equity

People with schizophrenia-spectrum disorders, bipolar disorder, severe recurrent depression, and other serious psychiatric conditions have substantial tobacco-related disease burdens and often face barriers to cessation services. The research question includes both whether the medicines work in these groups and whether the health system offers them equitably. Exclusion on the basis of diagnostic labels alone would make efficacy evidence for the general population inaccessible to patients who may experience major benefit, while careful attention to clinical stability and drug interactions remains necessary. Evins et al. (2014) Bekkering et al. (2026) Myers et al. (2026)

The 2026 Bekkering network meta-analysis provides population-specific support for varenicline and bupropion in severe mental illness and less certain evidence for some other interventions. Its authors found strong evidence for a meaningful absolute improvement in long-term abstinence with varenicline. Nevertheless, the review's estimates of serious adverse events were highly uncertain and most contributing trials had risks of bias. The findings thus substantiate efficacy while leaving important safety and implementation questions open. Bekkering et al. (2026)

The maintenance trial by Evins and colleagues supplies complementary evidence that selected patients with schizophrenia or bipolar disorder who have attained early abstinence may sustain quitting with continued varenicline and behavioral treatment. This trial's relapse-prevention design differs from trials that initiate pharmacotherapy in everyone who currently smokes. Together, the evidence favors integrated tobacco treatment in psychiatric services, active follow-up, and preservation of continuity with antipsychotic and mood-stabilizing treatment, rather than a static claim that psychiatric diagnosis determines a single medication choice. Evins et al. (2014) Myers et al. (2026)

Regulatory History: Safety Signals, Boxed Warnings, and Controlled Evidence

Regulatory safety evaluation proceeds through successive forms of evidence. Spontaneous reports can identify unexpected patterns and prompt warnings, but reporting rates lack a reliable denominator and are sensitive to publicity, concomitant medicines, underlying disease, and temporal coincidence. Randomized safety trials can provide comparative incidence for prespecified endpoints, while postauthorization surveillance can detect uncommon events in populations too heterogeneous or large for a registration trial. The historical sequence for varenicline and bupropion illustrates the interaction of these approaches. Earlier behavioral warnings were followed by the dedicated EAGLES investigation, then by revised FDA labeling that incorporated the trial findings. FDA (2016) Anthenelli et al. (2016)

In December 2016 the FDA removed the boxed warning regarding serious mental health side effects from varenicline and removed the corresponding neuropsychiatric language from the boxed warning of smoking-cessation bupropion. The antidepressant-class boxed warning concerning suicidal thinking and behavior in younger people remained applicable to bupropion products. This specific regulatory action should be represented accurately: it changed the placement and evidential characterization of certain warnings, rather than declaring every neuropsychiatric event impossible. Existing prescribing information continues to describe neuropsychiatric reactions, advise appropriate observation, and specify other serious risks. FDA (2016) DailyMed (2025) DailyMed (2026), bupropion

The decisive regulatory question is always formulation- and jurisdiction-specific. A clinical guideline may endorse a drug class after evaluating scientific evidence, while the local regulator may authorize a particular product for a narrower population or with different warnings. Nicotine replacement, bupropion, varenicline, and cytisine have different legal histories across countries, and the relevant label can change over time. Guidelines from WHO, NICE, and the VA/DoD therefore complement the actual U.S. prescribing information rather than replacing it. Scientific synthesis should document the version and date of each regulatory instrument used. World Health Organization (2024) NICE (2025) Myers et al. (2026)

Adverse Effects Beyond the Prespecified Psychiatric Composite

Neuropsychiatric safety is one dimension of the total treatment burden. Varenicline trials and labeling describe nausea, abnormal dreams, sleep disturbance, and other common adverse reactions. Such experiences may undermine persistence even when they fall short of the moderate-to-severe composite outcome selected for EAGLES. A patient who stops treatment because of nausea and someone who requires emergency psychiatric care have experienced fundamentally different kinds of harm, and both outcomes belong in a complete evidence account. Trial reports should distinguish adverse-event frequency, severity, seriousness, causality assessment, and discontinuation. DailyMed (2025) Anthenelli et al. (2016) Livingstone-Banks et al. (2023)

For bupropion, seizure risk, insomnia, contraindicated conditions, and relevant drug interactions require attention independent of its efficacy in cessation trials. Its effects on noradrenergic and dopaminergic systems do not make it an equivalent substitute for every antidepressant; specific prescribing restrictions remain applicable. Nicotine replacement has a distinct adverse-event pattern influenced by route of administration, with patch skin reactions, oral irritation, sleep disturbance, and risks from nicotine overdose or accidental exposure. These treatment-specific profiles are more useful for comparative decision-making than a single aggregate adverse-event number. DailyMed (2026), bupropion DailyMed (2026) Hajizadeh et al. (2023)

Some harms are difficult to measure precisely in short-duration cessation trials. Sleep patterns, persistent affective change, changes in alcohol intake, relapse-associated distress, and medication discontinuation may evolve across several stages of quitting. Future trials should use prespecified event definitions, systematic solicitation, patient-reported measures, and longitudinal outcomes beyond the active treatment phase. Safety claims should state the follow-up horizon, because absence of detected short-term harm does not automatically establish long-term safety for every population and co-medication regimen. Thomas et al. (2015) Bekkering et al. (2026)

The Pharmacokinetic Consequences of Stopping Combustible Tobacco

Combustible cigarette smoke contains polycyclic aromatic hydrocarbons that induce hepatic enzymes, including CYP1A2. Nicotine, whether delivered by a cigarette or replacement product, is not the principal cause of this induction. When a person substantially reduces or stops smoking, induction declines and concentrations of some concurrently prescribed drugs can increase without any change in the prescribed dose. This creates an important causal distinction: a neuropsychiatric or neurological symptom during cessation may reflect rising exposure to another medicine rather than an intrinsic toxic action of the cessation aid. The magnitude depends on pretreatment smoking exposure, metabolic variation, concurrent illness, and the pharmacokinetics of the affected drug. MHRA (2014)

Clozapine and olanzapine are prominent examples of psychotropic medicines whose exposure can be altered by smoking status. Theophylline and caffeine provide additional clinically relevant examples from outside or adjacent to psychiatric treatment. For clozapine, a change in smoke exposure can affect sedation, cardiovascular or neurological adverse effects, and the interpretation of drug concentrations; in a susceptible patient, a clinically significant change requires professional assessment. The same principle applies when smoking resumes after discharge or a period of enforced abstinence, because enzyme induction may return and drug exposure may decline. MHRA (2014)

Research on cessation outcomes should collect smoking intensity, smoke-free periods, vaping or nicotine-replacement use, and co-medication data with sufficient timing. Clinical teams should coordinate monitoring for narrow-therapeutic-index drugs during changes in combustible tobacco exposure. This does not justify a universal dose-adjustment formula: the precise response depends on measured concentrations, symptoms, treatment context, and the pharmacokinetic profile of each medicine. The conceptual lesson is that behavioral exposure to combustion belongs inside psychopharmacological causal models. MHRA (2014) Myers et al. (2026)

Comorbidity, Contraindications, and Generalizability

Cessation treatment commonly occurs amid cardiovascular disease, chronic respiratory illness, substance use, psychiatric disorders, pregnancy, adolescence, older age, and extensive polypharmacy. These circumstances differ in their risk–benefit balance and frequently determine eligibility for clinical trials. The conclusions of an adult trial of psychiatrically stable volunteers should not be asserted without qualification for an adolescent, a pregnant patient, a person with acute mania, or an individual experiencing recent severe suicidal thoughts. Evidence assessment therefore begins by identifying the actual participants and exclusion criteria. Anthenelli et al. (2016) World Health Organization (2024)

Seizure vulnerability is particularly relevant for bupropion, including specified histories and interacting medicines. For all regimens, treatment selection must account for hypersensitivity, previous response, severe adverse events, organ function, and concurrent prescriptions described in current labels. Local regulatory approval also matters for newer or internationally established products such as cytisine. The fact that a medicine is recommended by an international body does not eliminate the need to consult the applicable national authorization and product information. DailyMed (2026), bupropion Courtney et al. (2021) World Health Organization (2024)

The evidentiary gaps are ethically important because people with active psychiatric symptoms or complex social circumstances may have high tobacco-related mortality and limited access to treatment. Better evidence requires trials that expand inclusion with appropriate safeguards, transparent subgroup reporting, and outcome systems capable of documenting both psychiatric stability and cessation benefit. Excluding complex cases from research and then treating their uncertain evidence as proof of inevitable harm would obstruct the development of effective integrated care. The alternative is deliberate, monitored inclusion and accurate communication of existing uncertainty. Bekkering et al. (2026) Evins et al. (2019)

Clinical Guidance and Integrated Behavioral Health Care

The 2024 WHO clinical treatment guideline recommends effective cessation pharmacotherapies and behavioral support, accounting for evidence certainty, acceptability, feasibility, resource conditions, and access. NICE guidance similarly integrates licensed pharmacotherapies within a broader pathway that addresses dependence and behavioral assistance. The 2026 U.S. VA/DoD guideline supplies a contemporary American implementation context, including preferred medication strategies and treatment for individuals who are not yet prepared to quit immediately. These documents differ in population and health-system setting but converge in recognizing tobacco dependence as a treatable, recurrent condition. World Health Organization (2024) NICE (2025) Myers et al. (2026)

A comprehensive care pathway includes recognition of current smoking status, assessment of dependence and previous quit attempts, review of concurrent medicines, discussion of patient preferences, evidence-based treatment, and follow-up of abstinence and adverse symptoms. In people receiving psychiatric care, the same process also requires communication between tobacco treatment providers and the clinicians managing mood stabilizers, antidepressants, or antipsychotics. Early withdrawal symptoms, changes in drug exposure, and worsening of an independent psychiatric disorder may all warrant different responses. Therefore the temporal relationship between quitting, medicines, and symptoms must be documented rather than interpreted from a diagnosis label alone. MHRA (2014) Myers et al. (2026)

These principles are a synthesis of research and guideline-level practice, not a prescription for a particular reader. Treatment choice, eligibility, and monitoring need professional judgment informed by current local product labeling. People experiencing severe mood changes, suicidal thinking, seizures, or acute psychiatric symptoms require prompt clinical assessment. This cautious clinical translation follows directly from the evidence: average efficacy and comparative safety are strong enough to support access to treatment, while rare harms and individual contraindications retain practical importance. FDA (2016) DailyMed (2025) DailyMed (2026), bupropion

Methodological Priorities: Safety Estimands and Reproducibility

A major methodological challenge is the definition of the safety estimand. Investigators must specify the target population, treatment contrast, outcome, event-severity threshold, observation interval, and handling of treatment discontinuation and smoking relapse. The intention-to-treat effect of assignment to a pharmacotherapy differs from the effect of sustained use, and events occurring after treatment cessation may be attributed differently across analyses. Without an explicit estimand, studies can appear inconsistent when they are in fact measuring different causal questions. The EAGLES design improves interpretability by prespecifying its moderate-to-severe composite while leaving separate event-specific questions for additional investigation. Anthenelli et al. (2016) Evins et al. (2019)

Future safety research should combine sufficiently large randomized studies with active surveillance, validated psychiatric measures, and carefully designed observational analyses. Repeated measurements of smoking status, withdrawal, sleep, psychiatric symptoms, adherence, and co-medication concentrations would help separate direct drug effects from withdrawal and pharmacokinetic mechanisms. Observational emulation of target trials can improve temporal framing but must control for confounding by indication, healthcare access, psychiatric stability, and clinician decisions. Blinded event adjudication, open analysis plans, and transparent adverse-event definitions would strengthen reproducibility and reduce outcome-selective interpretation. Thomas et al. (2015) Anthenelli et al. (2019)

Meta-analytic methods also require vigilance about overlapping study populations: EAGLES primary analyses, predictor reports, and psychiatric subgroup studies are multiple publications from related data, not three independent experiments. Likewise, network comparisons among medications depend on transitivity across trial eligibility, counseling intensity, and abstinence definitions. Sensitivity analyses should test these assumptions, while the certainty of rare-event conclusions should be reported independently of abstinence efficacy. Clinical reviewers should scrutinize the precise source of every quantitative claim and DOI before public release. Anthenelli et al. (2016) Evins et al. (2019) Bekkering et al. (2026)

Artificial Era: Digital Support, Behavioral Measurement, and Human Oversight

The Artificial Era creates new contexts for cessation research through digital self-monitoring, adaptive reminders, conversational systems, and data-driven behavioral support. These interventions can record craving episodes, smoking-related cues, adherence, mood changes, and lapse patterns at greater temporal resolution than conventional clinic visits. Their scientific value depends on validity and measurement reliability: a text description of a craving episode is a reported observation, whereas verified abstinence remains a distinct endpoint. Digital intensity may also differ by access, literacy, privacy expectations, and willingness to disclose psychiatric symptoms. World Health Organization (2024) Myers et al. (2026)

Artificial systems used in tobacco cessation can help organize information and provide consistent prompts, while human clinical responsibility remains essential for medical selection, serious adverse-event triage, and management of concurrent prescriptions. Adaptive systems may personalize behavioral timing or identify needs for follow-up, but algorithmic confidence cannot substitute for validated risk prediction. Research should examine effectiveness against transparent comparators and test for differential failure among people with serious mental illness, limited connectivity, or other underrepresented contexts. Claims of effectiveness require empirical trials and clearly defined safety outcomes. Myers et al. (2026) World Health Organization (2024)

Within a broader psychology of human–artificial relations, cessation offers an instructive setting for studying how a nonbiological system can participate in monitoring, explanation, and supportive communication. The relevant theoretical questions include reliance, trust calibration, agency, autonomy, adherence, and distributed decision support. These are research directions for future empirical testing rather than evidence that any specific artificial assistant has established therapeutic benefit. Maintaining this distinction allows interdisciplinary inquiry to advance without transforming a pharmacological evidence review into an unsupported claim about artificial cognition. World Health Organization (2024)

Open Questions and Future Research Priorities

Several major questions remain: the magnitude of rare neuropsychiatric harms among patients with acutely unstable mental illness; comparative efficacy in young, pregnant, and medically complex populations; optimal sequencing after failed treatment; safety of extended and combined regimens; and the behavioral mechanisms that explain relapse after initial pharmacological success. Long-term studies should distinguish continuous abstinence, intermittent smoking, alternative nicotine delivery, and complete nicotine discontinuation, because these are clinically and mechanistically different trajectories. Research also needs absolute risk estimates that remain intelligible to patients, rather than rankings alone. Bekkering et al. (2026) World Health Organization (2024), evidence Myers et al. (2026)

Medication–smoking interactions merit greater integration into randomized and observational research. Many studies report the assigned cessation medicine while collecting incomplete information about clozapine, olanzapine, theophylline, caffeine, and other exposure-sensitive agents. Systematic recording of combustion exposure, pharmacokinetic monitoring where appropriate, and outcome adjudication would clarify apparently paradoxical cases of psychiatric or neurological deterioration after successful quitting. These questions connect cessation psychopharmacology with pharmacovigilance and behavioral medicine. MHRA (2014) Anthenelli et al. (2019)

A stronger future evidence architecture would treat benefit, symptom experience, serious harm, access, adherence, and sustainability as coordinated but separate domains. Broader trial populations, patient-centered outcomes, transparent protocols, independent replication, and comparable definitions across jurisdictions are particularly valuable. Research should continue to investigate how behavioral interventions and pharmacotherapy interact while acknowledging that head-to-head comparisons under equal counseling conditions identify an incremental medication effect rather than the entire causal process of behavior change. Anthenelli et al. (2016) Bekkering et al. (2026) World Health Organization (2024)

Conclusion

Tobacco cessation pharmacotherapy has an established evidence base for increasing the probability of sustained abstinence, with varenicline, bupropion, nicotine replacement, and cytisine supported to differing degrees by randomized trials, systematic reviews, and international guidelines. The strongest comparative evidence generally favors varenicline over placebo, bupropion, and single-form nicotine replacement, while combined nicotine replacement improves outcomes compared with single-form replacement. These results describe population averages under particular trial conditions, and individual treatment benefit also depends on contraindications, adherence, clinical context, and access to behavioral support. Livingstone-Banks et al. (2023) Hajizadeh et al. (2023) Theodoulou et al. (2023) World Health Organization (2024), evidence

The EAGLES trial is a central randomized contribution to the analysis of neuropsychiatric safety. Among eligible, predominantly psychiatrically stable participants, it did not identify a statistically significant increase in its prespecified moderate-to-severe composite with varenicline or bupropion relative to patch or placebo. It was not designed to establish the impossibility of every rare adverse event, and event-specific labeling, baseline psychiatric vulnerability, nicotine withdrawal, and limitations of subgroup precision remain relevant. Later review evidence strengthens the case for offering cessation treatment to people with serious mental illness while preserving uncertainty about uncommon serious harms. Anthenelli et al. (2016) Evins et al. (2019) Bekkering et al. (2026)

A complete interpretation also recognizes that cessation changes the organism's pharmacological environment. Withdrawal alters subjective experience and stopping combustible smoke reverses enzyme induction that can influence concentrations of concomitant medicines. In this setting the scientifically central task is causal separation: identify what was assigned, what was consumed, what changed, and what was measured over time. Integrated behavioral and clinical care, rigorous surveillance, and evidence-grading by distinct outcomes provide the strongest foundation for future research and responsible clinical translation. MHRA (2014) Taylor et al. (2021) Myers et al. (2026)

References