Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Addiction, Substance Use Disorders, and Diagnostic Boundaries

Abstract. This fundamental research monograph investigates how addiction and substance use disorders are conceptually defined, operationalized, diagnosed, distinguished from hazardous use and physiological adaptation, and validated against clinically meaningful outcomes. Using a critical narrative synthesis of original diagnostic papers, DSM-5-TR and ICD-11 classification documents, foundational motivational theories, instrument-development studies, historical cohort research, and contemporary US population surveillance, it reconstructs the changing boundaries among exposure, harm, dependence, intoxication, withdrawal, and disorder. It compares the DSM symptom-count architecture with ICD-11's separate harmful-use and dependence categories, evaluates diagnostic thresholds and measurement validity, and examines substance-specific, developmental, social, cultural, ethical, and digitally mediated challenges. The findings support criterion-level, substance-specific, longitudinal and context-sensitive inference and identify unresolved research questions concerning thresholds, prognostic validity, misclassification, and clinical utility.

Keywords: addiction; substance use disorder; dependence; physical dependence; tolerance; withdrawal; DSM-5-TR; ICD-11; diagnostic validity; harmful use; hazardous use; craving; AUDIT; ASSIST; addiction psychology; clinical psychometrics; differential diagnosis; Artificial Era.

Introduction: The Problem of a Valid Diagnostic Boundary

Addiction research confronts a problem of classification before it can explain a mechanism: which patterns of substance-related experience and behavior constitute a disorder, which constitute exposure or risk, and which are pharmacological adaptations to repeated administration? Alcohol, nicotine, opioids, sedatives, cannabis, and stimulants differ in molecular actions, patterns of use, acute dangers, and withdrawal physiology. A scientifically defensible diagnosis must nevertheless identify recognizable disturbances in control, salience, persistence, and functioning. This monograph examines the boundary rules that turn observations about use into justified clinical inferences. It treats diagnostic categories as empirically revisable tools that connect clinical decisions, epidemiological measurement, and research rather than as transparent readings of an underlying essence. The DSM-5 revision process provides a documented example of how empirical evidence can reorganize criteria and thresholds. Hasin et al. (2013) WHO (2024)

The central questions are whether diagnostic systems consistently identify clinically consequential disorders, how they distinguish an enduring disorder from an isolated harmful event, and how their thresholds perform across substances, cultures, ages, and treatment contexts. Four levels must be kept analytically connected: actual exposure, consequences and risk, patterns of behavioral regulation, and formal diagnostic assignment. At each level the evidence changes. Exposure can be measured by interview or biological assay; impairment requires temporal, functional, and contextual assessment; syndrome attribution requires a coherent diagnostic formulation; and diagnostic validity requires prospective or external evidence that classification captures meaningful outcomes. The distinction is especially important where prescribed medications produce tolerance, where intoxication generates acute harm, or where institutional sanctions create consequences independent of psychopathology. NYSDOH AI (2024) Wakefield (1992)

The present work is a critical narrative synthesis of identified original clinical and conceptual papers, classification documents, instrument-development studies, and official population surveillance. It compares diagnostic logic, examines sources of misclassification, and draws methodological conclusions from evidence with different designs. It does not claim a preregistered systematic search or quantitative meta-analysis, nor does it report new patient observations. Its evidentiary core includes the original alcohol-dependence syndrome formulation, the DSM-5 revision rationale, the ICD-11 diagnostic description and commentary, screening-tool development studies, and national survey data. Direct primary references are provided near their uses and collected in the terminal bibliography. Edwards & Gross (1976) Hasin et al. (2013) Saunders et al. (2019) SAMHSA (2026)

Scientific Definitions and Levels of Observation

Substance use is the administration or consumption of a psychoactive substance, described by dose, route, frequency, setting, purpose, and exposure history. Hazardous use describes a pattern that increases the risk of harm, even when injury or diagnostic syndrome has not occurred. Harmful use identifies actual adverse consequences attributable to use under an applicable classification scheme. Substance intoxication denotes a temporally associated syndrome following exposure; withdrawal denotes signs and symptoms following reduction or discontinuation after adaptation. Substance dependence has a clinical syndrome meaning in ICD-11 and a narrower physiological meaning in some pharmacological usage. Addiction is an important clinical and scientific umbrella term, but its colloquial use lacks the precision needed to apply a formal diagnostic code. These distinctions must be explicitly established in every research design. Saunders et al. (2019) WHO (2024)

The term substance use disorder, or SUD, designates the DSM's substance-specific pattern of problematic use meeting a defined minimum of criteria within a specified time. It is a diagnostic construct; its presence is not automatically determined by route, quantity, intoxication, or an observer's disapproval. Severity, temporal course, and the individual's experience of impaired control contribute to interpretation. In ICD-11, the classificatory architecture instead identifies disorders such as harmful pattern of use and dependence, and also separately recognizes certain acute or substance-induced conditions. The same patient presentation may therefore have different formal descriptions under the two systems even when clinicians agree about the observations themselves. Hasin et al. (2013) Saunders et al. (2019)

Operationalization requires distinct research records: substance exposure; each diagnostic criterion and its onset and duration; functional consequences; treatment indication; medically supervised exposure; contextual hazards; and alternate explanations. Researchers should record whether information comes from self-report, informants, clinical observation, toxicology, service records, or longitudinal follow-up. Each method answers a different question. Toxicology can establish recent presence of some substances within assay windows but cannot establish most motivational and social diagnostic criteria. Retrospective interviews can capture experiences and motives but are susceptible to recall error and social-desirability effects. A defensible diagnosis combines observations without converting a convenient measurement into a complete theory. WHO ASSIST Working Group (2002) NYSDOH AI (2024)

Historical Formation of the Dependence Construct

Modern diagnostic reasoning developed through repeated attempts to organize varied patterns of substance-related suffering into syndromes. Edwards and Gross's 1976 provisional description of an alcohol dependence syndrome shifted attention toward an integrated cluster of clinical phenomena rather than any single drinking quantity or consequence. Their formulation was deliberately provisional and grounded in clinical observations, reflecting the difficulty of finding a single behavioral or physiological marker sufficient for all cases. The paper is historically significant for organizing dependence as a syndrome while acknowledging variation across patients. Contemporary classification carries that syndromic insight into more explicit diagnostic requirements, even as the individual criteria and boundaries have changed. Edwards & Gross (1976)

The subsequent separation of substance abuse and dependence in DSM-IV attempted to distinguish harmful consequences from a broader dependence syndrome. Over time, research identified problems in the architecture: a conceptual gap could exist between the two categories, some diagnostic items behaved inconsistently, and dependence was often misunderstood as mere medication-induced physiological adaptation. Hasin and colleagues documented the evidence and reasoning for the DSM-5 revision, including the integration of abuse and dependence criteria into one disorder and the addition of craving. This change represented a substantive reclassification rather than a renaming of a historically fixed disease entity. The research program used reviews and analyses across large datasets, which strengthened the argument for revision while leaving specific measurement and threshold questions open. Hasin et al. (2013)

ICD-11 followed its own international classification development, preserving a distinct role for dependence and for harmful use patterns. Saunders and colleagues describe how alcohol dependence and harmful use are differentiated in ICD-11, and how hazardous alcohol use sits outside the mental-disorder category as a health-risk factor. This separation offers a conceptual reminder that public-health interventions may be appropriate before a psychiatric disorder has developed. History thus supplies a methodological lesson: diagnostic nomenclature evolves with the purposes of medicine, observed patterns of harm, and evidence about reliability and validity. Period comparisons must retain the source classification rather than back-translating older diagnoses mechanically into modern ones. Saunders et al. (2019) WHO (2024)

Addiction, Compulsion, Craving, and Self-Regulation

Addiction research investigates repeated substance pursuit despite personally significant harms, changes in the motivational power of drug-related cues, increasing priority of use, and difficulties regulating behavior under changing incentives. These features motivate multiple explanatory traditions. Incentive-sensitization theory proposes that drug-related cues may acquire disproportionate motivational significance, generating wanting that can diverge from the hedonic pleasure of consumption. Robinson and Berridge's original theoretical paper and later reassessment show the explanatory ambition of this framework, while also making clear that a laboratory account of cue-driven motivation does not by itself establish a clinical diagnosis in a particular person. Robinson & Berridge (1993) Robinson & Berridge (2025)

Allostatic and negative-reinforcement accounts analyze the interaction between repeated exposure, stress-related processes, reduced reward responsivity, and relief from negative states. Koob and Le Moal formulated drug abuse in terms of dysregulation of hedonic homeostasis. Their model illuminates transitions in motivational dynamics and has influenced contemporary neurobiological research. It is complementary to questions about social opportunity, available alternatives, drug supply, treatment access, and relationships. Neither an incentive-salience measure nor a stress-system measure is currently a stand-alone diagnostic test for SUD. A scientific account of a disorder needs a bridge from proposed mechanisms to reliably observed clinical manifestations. Koob & Le Moal (1997) Robinson & Berridge (2025)

The word compulsion requires particular care. Persistent substance pursuit may involve habit, cue-induced motivation, withdrawal avoidance, immediate reinforcement, or constraints on alternative actions, with different mixtures across people and circumstances. Clinical language about a loss of control records a consequential subjective and behavioral pattern; it need not imply literal absence of choice or a uniform brain lesion. Conversely, the presence of purposeful action does not invalidate the possibility of a disorder. A constructive assessment documents attempted changes, outcomes, recurrent harmful consequences, and the environment in which actions occur. The classification question is whether a meaningful syndrome is established, while a mechanistic study asks which processes best explain it. Hasin et al. (2013) McLellan et al. (2000) Robins et al. (1975)

DSM-5-TR Substance Use Disorder: Criteria and Severity

The DSM-5-TR retains an eleven-criterion framework for most substance-use disorder diagnoses, organized around impaired control, social impairment, risky use, and pharmacological indicators. Clinically relevant manifestations include using more or longer than intended, unsuccessful efforts to reduce use, considerable time devoted to use-related activity, craving, role disruption, interpersonal problems, reduction of important activities, recurrent use in dangerous situations, continued use despite related health or psychological problems, tolerance, and withdrawal. The eleven are summarized here for analytical purposes rather than reproduced as a clinical instrument. The applicable substance chapter, temporal specification, exclusions, and professional diagnostic judgment remain essential. American Psychiatric Association (2022) NIAAA (n.d.)

The familiar DSM-5-TR SUD threshold requires at least two qualifying criteria during a twelve-month period, with severity classified as mild for two or three, moderate for four or five, and severe for six or more. Counts are useful for common coding and research comparability, but equal totals may involve different symptom configurations. One patient's two criteria might comprise repeated unsuccessful cutbacks and role impairment; another's might reflect distinct consequences. Counts do not provide the full clinical history, and a criterion's meaning depends on its cause and context. The shift to a dimensional severity scheme drew on a substantial empirical assessment of the former abuse/dependence split. NIAAA (n.d.) Hasin et al. (2013)

An especially important safeguard concerns prescribed medication. Tolerance and withdrawal that occur solely in the course of taking a medication appropriately under medical supervision are not counted as substance use disorder criteria under the relevant DSM provisions. The clinician must still consider other indications of problematic control or harm where present. This prevents a predictable pharmacological response to an indicated opioid, sedative, or another medication from being treated as sufficient proof of an addictive disorder. Dependence terminology is particularly prone to confusion because physiology, clinical syndrome, and everyday usage attach different meanings to the same word. A careful clinical assessment states exactly which meaning applies. NYSDOH AI (2024) American Psychiatric Association (2022)

ICD-11: Dependence, Harmful Patterns, and Risk Factors

ICD-11 organizes disorders due to substance use through categories that include dependence, harmful pattern of use, harmful episode, intoxication, withdrawal, and specified substance-induced syndromes, with substance- and presentation-specific requirements. Its dependence diagnosis emphasizes a strong internal drive to use, impaired control, increasing priority assigned to use, and continued use in spite of harm, while relevant accompanying features may include pharmacological adaptation. The classification also allows clinically meaningful harmful use to be recognized without requiring a dependence syndrome. The authorized 2024 WHO clinical descriptions are the contemporary primary document for formal implementation. WHO (2024) Saunders et al. (2019)

A key conceptual division separates health risk from current disorder. Hazardous use denotes exposure patterns that raise the likelihood of future harmful outcomes; ICD-11 places hazardous alcohol use within a risk-factor framework rather than necessarily classifying it as a mental disorder. Harmful use requires harm that can be attributed to the person's pattern or episode of substance use under the specific rules of the category. Dependence describes a syndrome with a distinctive priority-and-control organization. These distinctions make it possible for clinicians and public-health programs to respond proportionately: preventive screening can be warranted even where a dependence diagnosis would be unsupported. Saunders et al. (2019) WHO (2024)

Differences between ICD-11 and DSM-5-TR have methodological consequences. A study using a DSM SUD threshold and a study using an ICD harmful-pattern or dependence code may recruit overlapping but nonidentical groups. Rates and effect estimates cannot be compared without a crosswalk that specifies assessment items, time windows, exclusion rules, and diagnostic algorithms. The same applies to clinical service data, where coding practices, clinician training, and health-system incentives affect recorded prevalence. International comparison also calls for linguistic and cultural adaptation, because the meaning of impairment, responsibilities, and social problems varies with opportunity and circumstances. Saunders et al. (2019) Hasin et al. (2013) WHO (2024)

Comparing Diagnostic Architectures Without Erasing Differences

DSM and ICD classification systems can be compared at three levels: their definition of a clinically relevant disturbance, the indicator patterns used to recognize it, and the institutional use of the resulting code. DSM-5-TR operationalizes a graded syndrome largely through a criterion count. ICD-11 preserves diagnostically distinct harmful-use and dependence constructs, alongside acute manifestations and risk factors. This difference is scientifically useful. It permits testing whether a broad symptom-count dimension, a narrower dependence syndrome, or an explicitly harmful-use category better forecasts outcomes relevant to patients and health systems. Hasin et al. (2013) Saunders et al. (2019)

A meaningful empirical crosswalk would apply both algorithms to the same participants, document discordant cases, and follow outcomes such as persistent impairment, morbidity, help seeking, emergency presentations, and quality of life. Agreement alone is insufficient: two rules can agree because they share source items while both fail to predict an important future outcome. Conversely, discordance may reflect useful differentiation rather than error. Clinical validity depends on the aim of classification, including treatment planning, prevention, clinical communication, population measurement, or research selection. Comparisons should report sensitivity and specificity against an independently justified criterion rather than declare one system universally superior. Wakefield (1992) Hasin et al. (2013)

An additional source of discordance is the identity of the substance and the relevant syndrome. Withdrawal is prominent for some drug classes, differently characterized for others, and in certain contexts clinically hazardous; individual DSM and ICD chapters therefore matter. The research category cannot be transferred from alcohol to stimulants or from nicotine to sedatives by substituting a word in the heading. A correct cross-system study identifies the actual drug class, diagnoses in use, assessment version, duration, and clinical context. The resulting crosswalk is a reproducible research product, not a loose equivalence between familiar labels. American Psychiatric Association (2022) WHO (2024)

The Concept of Disorder: Harm, Function, and Context

The diagnostic boundary is also a question in the philosophy and methodology of medicine. Wakefield's harmful-dysfunction account distinguishes a claim that an internal process is dysfunctional from a judgment that the condition is harmful within a social context. This is one influential conceptual proposal, rather than a universally agreed biological assay for mental disorder. It clarifies why moral dislike, illegality, and statistical rarity cannot independently establish clinical pathology. Substance-use diagnosis demands observable patterns and consequences anchored to defensible criteria, while their evaluation inevitably engages the person's goals, environment, relationships, and opportunities. Wakefield (1992)

A diagnostic system may misclassify in more than one direction. A low threshold with poorly discriminating criteria can classify transient, situational, or medically expected experiences as disorder. A high threshold centered on an idealized syndrome can overlook severe harm in someone without a full dependence pattern. A person may incur legal or employment consequences from a drug policy that differs across jurisdictions; counting those consequences as evidence of internal disorder requires particular justification. The DSM-5 revision removed recurrent legal problems as a criterion and introduced craving, reflecting empirical and conceptual assessment of what each indicator contributed. Hasin et al. (2013)

Clinical significance is best established through triangulation. A criterion count identifies a possible category; examination of functional deterioration, distress, repeated choices, symptom chronology, treatment circumstances, and independent information tests its interpretation. Longitudinal observation offers evidence about persistence and change. A clinician should distinguish a harmful exposure from a syndrome, recognize the urgency of acute toxicological danger independently of chronic diagnostic status, and preserve the patient's own account of what matters. These principles support proportionate health care and improve the quality of scientific datasets derived from clinical assessments. NYSDOH AI (2024) McLellan et al. (2000)

Operational Assessment: From Reports to Criterion-Level Evidence

A diagnostic interview translates experiences into time-bounded observations. Its quality depends on the interviewer asking what occurred, when it occurred, and why it is attributable to the substance, rather than accepting a general label such as addiction or dependence. Repeated unsuccessful efforts to cut back require an account of intentions and attempts; role impairment requires the nature and chronology of missed obligations; hazardous use requires a specific situation and associated risk. Craving is a motivational experience whose intensity and frequency can vary without necessarily determining the severity category on their own. Assessment of tolerance and withdrawal requires attention to prescribed medication and the substance-specific rules. A transparent record keeps qualifying criteria and disqualifying contexts separate. Hasin et al. (2013) NYSDOH AI (2024)

Diagnostic inference should be reproducible. The same case material can be evaluated under DSM-5-TR and ICD-11 if the history contains sufficiently precise observations, even though the algorithms and terminology differ. Interrater reliability asks whether two qualified assessors using the same rules reach similar judgments; test-retest reliability asks whether the diagnosis remains sufficiently stable over a short interval when the clinical state is relatively unchanged. Agreement can deteriorate when ambiguous descriptions, retrospective periods, or language differences are left unresolved. Validity is a further question: consistent classification may still fail to distinguish clinically important patterns or to predict future harm. Evaluation therefore needs reliability studies and external outcome evidence. Hasin et al. (2013) WHO (2024)

A strong assessment addresses social and medical context without treating it as an excuse to discard symptoms. Housing instability, criminalization, discrimination, medication access, pain management, workplace sanctions, and family obligations can influence both exposure and the consequences counted as impairment. A causal history considers whether difficulties preceded substance use, arose during intoxication, developed with withdrawal, or persisted during periods of reduced exposure. Collateral information may clarify chronology but introduces confidentiality and consent requirements. In applied research, these contextual variables should be encoded rather than relegated to a free-text comment that cannot be analyzed. Wakefield (1992) NYSDOH AI (2024)

Screening Instruments, Diagnostic Interviews, and Their Different Purposes

Screening identifies people who may benefit from further discussion or assessment; diagnosis assigns a clinical category using applicable requirements. The Alcohol Use Disorders Identification Test, developed in a multinational WHO collaborative project, is a screening instrument addressing patterns of drinking and consequences, not a stand-alone implementation of every DSM or ICD disorder code. The WHO Alcohol, Smoking and Substance Involvement Screening Test assesses involvement and associated concerns across substance classes for primary-care contexts. Their original development papers describe the intended use and early measurement evidence. A positive screen should be followed by a suitable assessment rather than converted mechanically into a final diagnosis. Saunders et al. (1993) WHO ASSIST Working Group (2002)

A screen's threshold changes the balance between sensitivity and specificity, while predictive values also depend on the prevalence of the target condition in the population tested. An instrument evaluated in a particular clinical sample may perform differently in adolescents, older adults, pregnant patients, people receiving prescribed opioids, or populations speaking another language. Validation requires more than a high internal-consistency coefficient. Investigators should report the reference assessment, recruitment setting, handling of missing responses, interval between assessments, and the exact clinical target. For longitudinal use, they should also test whether score changes track meaningful improvements in functioning or reduced harm. WHO ASSIST Working Group (2002) Saunders et al. (1993)

Research designs should preserve a criterion-level record alongside an instrument total whenever feasible. This allows examination of which symptoms drive scores, whether severity thresholds show discontinuities, and how the result maps to different classifications. A generic self-rating that asks whether someone feels addicted can have value as a subjective report but cannot substitute for standardized operational criteria. For medically supervised exposure, the risk of confusion between pharmacological adaptation and clinical disorder is especially high. Screening can guide questions, but the interpretive burden remains on an informed, context-sensitive diagnostic assessment. NYSDOH AI (2024) American Psychiatric Association (2022)

Diagnostic Validity, Severity, and Dimensional Measurement

The validity of a diagnostic construct can be investigated by testing how its indicators covary, whether there is a defensible common dimension, and whether diagnostic assignment forecasts outcomes independently of the measurements used to construct it. Hasin and colleagues describe evidence supporting the integration of former DSM-IV abuse and dependence symptoms into a single DSM-5 disorder dimension. Such evidence is consequential for classification design, but a statistical dimension does not establish that every criterion has identical causal meaning or that a given minimum count is a natural biological threshold. A severity index is a useful summary of clinically heterogeneous information. Hasin et al. (2013)

Criterion-response patterns matter because people meeting the same threshold may enter the category through distinct routes. Tolerance may reflect biological adaptation; repeated hazardous use may reflect exposure in dangerous contexts; abandonment of activities may reflect persistent motivational narrowing; and continued use in the face of problems may capture behavior with many psychological and social determinants. A simple total imposes an ordinal ordering while discarding symptom composition. Researchers can address this loss through item-level response models, latent-class analyses, and prospective validation, provided assumptions about independence, sampling, and measurement invariance are examined rather than assumed. Hasin et al. (2013) Wakefield (1992)

Severity is also multidimensional outside the diagnostic manual. Medical risk may be high with a relatively low syndrome count, including a single dangerous intoxication or withdrawal episode. Conversely, a formal disorder can be present without the most dramatic acute harm. Researchers should separately measure syndrome severity, intensity of exposure, physical morbidity, psychiatric symptoms, social impairment, and mortality. This enables meaningful comparisons of change and prevents a single clinical scale from being interpreted as a comprehensive measure of health burden. A classification threshold is one decision rule within a much larger explanatory and clinical framework. Saunders et al. (2019) NYSDOH AI (2024)

Epidemiology: What National Surveys Actually Measure

A national prevalence estimate is the outcome of a diagnostic algorithm applied to eligible respondents under defined interview and sampling procedures. The United States 2025 National Survey on Drug Use and Health estimated that 44.6 million people aged twelve or older, approximately 15.3 percent of that population, met its past-year SUD definition. The same report estimated 25.7 million with alcohol use disorder, 26.0 million with a drug use disorder, and 7.1 million with both; overlapping conditions mean those subtotals should not be added as if they represented mutually exclusive groups. These are U.S. survey estimates for 2025, published in 2026, rather than clinical counts or global prevalence. SAMHSA (2026)

The same source reports differences by age group and severity. Such estimates require careful handling of eligibility, interviews, nonresponse, survey weights, criterion definitions, and uncertainty; an apparent year-to-year change can reflect both epidemiological developments and methodological factors. The 2025 report describes a decrease in the estimated past-year SUD percentage from 16.7 percent in 2021 to 15.3 percent in 2025 and notes revised earlier weights for comparability. This is a statement about the measured national indicator, not a causal attribution to a particular policy or intervention. The survey's coverage frame also differs from clinical registries and from some populations at particularly high risk. SAMHSA (2026)

A historically different population survey, the National Epidemiologic Survey on Alcohol and Related Conditions III, was used by Grant and colleagues to examine DSM-5 alcohol use disorder epidemiology. Their analysis belongs to a different data collection and observation period, demonstrating how the diagnostic scheme can be used in a large representative household study. Direct numerical comparison with NSDUH requires attention to survey modes, periods, and case definitions. The wider principle is that prevalence belongs to a defined measurement system. A count of people meeting a criterion threshold is meaningful only with a documented denominator and a reproducible classification procedure. Grant et al. (2015) SAMHSA (2026)

Substance-Specific Distinctions: Alcohol and Nicotine

Alcohol illustrates why the boundaries between risky consumption, harmful pattern, dependence, intoxication, and withdrawal are scientifically important. Consumption may produce acute injuries without a long-standing dependence syndrome. Longitudinal heavy exposure can produce multiple physiological and behavioral manifestations, yet one such manifestation does not automatically determine the whole diagnosis. ICD-11 and DSM-5-TR frame these cases differently, particularly in their separation of harmful patterns from dependence and their rules for severity. A research article that reports alcohol dependence must therefore disclose which diagnostic definition was used and the period of assessment. Saunders et al. (2019) Hasin et al. (2013)

Alcohol screening instruments are designed for early identification of risky and problematic consumption and may be deployed in settings where most individuals lack a formal disorder. The AUDIT development project is evidence that public-health detection can be organized around risk before a clinical syndrome is established. The instrument's performance does not transform the risk categories into DSM diagnosis codes. Researchers should distinguish positive drinking screens, self-reported consumption, clinician-assigned disorders, and treatment episodes in any aggregated dataset. This also permits appropriate study of prevention, where the primary endpoint may be decreased hazardous drinking rather than diagnostic remission. Saunders et al. (1993) NIAAA (n.d.)

Nicotine reveals another important boundary. Regular nicotine exposure and physiological adaptation may coexist with differing levels of craving, unsuccessful attempts to stop, priority given to consumption, and impairment. The historical language of tobacco dependence and modern DSM tobacco use disorder must be translated with explicit criteria, not inferred from product count. Different delivery systems also create different toxic exposures, which need to be evaluated separately from psychological symptoms. For assessment, the clinically meaningful question is what the person experiences and how their behavior is organized, along with the medical hazards associated with the substance and delivery method. Hasin et al. (2013) American Psychiatric Association (2022)

Substance-Specific Distinctions: Opioids and Prescribed Medication

Opioids make the distinction between physiological dependence and an opioid use disorder particularly consequential. Long-term use of an opioid under appropriate medical supervision can result in tolerance or withdrawal after discontinuation, and those features by themselves do not establish opioid use disorder under the DSM provisions for such treatment. A comprehensive evaluation examines additional behavior and consequences, medical indication, adherence, attempted changes, diversion where present, and substance exposure outside treatment. This separation protects diagnostic validity and prevents the medical fact of adaptation from functioning as an unsupported accusation of addiction. American Psychiatric Association (2022) NYSDOH AI (2024)

Clinical risks are not exhausted by the presence or absence of a use disorder. Overdose, intoxication, medication interactions, withdrawal management, and infectious or injury-related exposures may require medical assessment independently of a chronic disorder diagnosis. Conversely, a clear disorder may exist in an individual whose current physical examination or recent toxicology findings are unremarkable. Researchers must resist both reducing the disorder to a biological assay and treating every drug-related emergency as proof of a sustained addictive syndrome. A complete epidemiological or clinical study can represent acute harm, long-term disorder, treatment exposure, and functional outcome as distinct variables. NYSDOH AI (2024) WHO (2024)

The terms withdrawal, dependence, and addiction also affect professional communication. A patient who anticipates withdrawal when tapering a prescribed medication may interpret a casual statement of dependency as a psychiatric diagnosis, while a clinician may have intended a physiological description. Clear recording uses precise terms: pharmacological adaptation; withdrawal syndrome if applicable; substance use disorder if diagnostic conditions are met; and the specific clinical risk that motivated evaluation. This precision improves conversations about treatment and avoids falsely inflating disorder prevalence in administrative records. NYSDOH AI (2024) McLellan et al. (2000)

Cannabis, Stimulants, Sedatives, and Polysubstance Presentations

Cannabis-related presentations include frequency and route of use, functional concerns, intoxication effects, craving, and withdrawal where the relevant diagnostic criteria are satisfied. Risk is influenced by potency, age of onset, co-occurring conditions, and context, but no single exposure characteristic independently defines DSM cannabis use disorder. Differences between study samples complicate the interpretation of diagnostic thresholds. A sample recruited from treatment services may have more severe impairment than a household sample with the same nominal category, and a prospective study may operationalize consequences differently from a retrospective survey. A valid comparison states the criteria applied and the population actually examined. Hasin et al. (2013) SAMHSA (2026)

Stimulant use disorders present heterogeneous patterns of binge episodes, craving, sleep disruption, work and relationship consequences, and psychiatric symptoms. An assessment must distinguish intoxication-related behavioral and perceptual changes from persisting primary psychiatric conditions. Sedative, hypnotic, and anxiolytic drugs bring strong issues of medical exposure and withdrawal physiology, requiring careful interpretation of tolerance under supervised treatment. Across these categories, substance-specific diagnostic exclusions and conditions remain essential. Broad statements that every psychoactive substance has the same dependence or withdrawal syndrome are biologically and diagnostically inaccurate. American Psychiatric Association (2022) WHO (2024)

Polysubstance presentations expose limitations of assessments that assign a single dominant substance. People may meet separate disorder criteria for alcohol and another drug, may show interacting acute toxicities, or may experience symptoms difficult to attribute to one exposure. The 2025 NSDUH separately reports overlapping alcohol and drug use disorders, illustrating the population-level significance of multiple classifications. A scientific case formulation records every implicated substance, dates and routes of use, symptom attribution, and diagnostic uncertainty. This is essential for evaluating interactions and for preventing duplicate counting across conditions when calculating overall SUD prevalence. SAMHSA (2026) NYSDOH AI (2024)

Acute Syndromes and the Boundary With Enduring Disorder

Intoxication and withdrawal should be investigated as clinical phenomena with temporal proximity to a substance and with presentation-specific diagnostic requirements. They can occur with or without a long-standing substance use disorder, depending on exposure and context. Their urgent clinical significance can be greater than the severity of a chronic classification at a particular time. ICD-11 distinguishes these syndromes in its substance-related categories, and DSM-5-TR also has substance-specific intoxication, withdrawal, and substance-induced disorder categories. The distinction is clinically actionable because immediate evaluation and long-term diagnostic formulation answer separate questions. WHO (2024) American Psychiatric Association (2022)

A substance-induced psychiatric syndrome introduces an attribution problem: the symptom may be temporally connected to exposure, withdrawal, independent psychiatric illness, or a combination. A careful history establishes whether the syndrome predates use, persists beyond an expected pharmacological period, or recurs during abstinence, along with medical history and other possible causes. No isolated symptom can ordinarily settle these questions. Diagnostic confidence should change as follow-up information accumulates. Epidemiological studies that lack detailed chronology must acknowledge the limits of causal inference from co-occurrence alone. NYSDOH AI (2024) WHO (2024)

The general methodological principle is to separate time scales. Acute physiological hazard unfolds over minutes or days; adaptation and withdrawal may unfold over longer intervals; the development of a use disorder and its remission may extend across months or years. Cross-sectional information collapses these processes and can make causal sequences appear simpler than they are. Repeated assessment, service histories, and prospective outcomes can distinguish persistent disorder, intermittent harmful episodes, and isolated exposure. Such designs are indispensable to research on diagnostic boundaries because classification is fundamentally concerned with trajectories as well as present symptoms. McLellan et al. (2000) Robins et al. (1975)

Behavioral Addictions and the Limits of Category Extension

The translation of addiction terminology beyond psychoactive substances is a major test of diagnostic boundaries. Gambling disorder has a formal place among addictive disorders in contemporary classifications, while gaming disorder is recognized in ICD-11 and Internet gaming disorder was identified in the DSM-5 research framework for further study. These developments do not establish that every repetitive, rewarding, highly preferred, or commercially designed activity is a psychiatric disorder. The decisive clinical questions concern impaired control, priority over other activities, continuation despite significant consequences, and the requirements established for the particular classification. The research standards for substance-use disorders cannot be transferred to digital activity without evaluating mechanisms, impairment, and potential measurement artifacts. Petry et al. (2014) WHO (n.d.)

Petry and colleagues' international consensus article documented methodological disagreement about Internet gaming disorder and offered a common way to interpret proposed diagnostic criteria. Its significance lies in standardizing research questions across countries, rather than supplying proof that any high-frequency player has a disorder. The WHO's explanation of gaming disorder similarly emphasizes impaired control, increasing priority, persistence despite consequences, and significant impairment. A defining problem is distinguishing sustained psychological dysfunction from immersion, enthusiasm, professional engagement, and conflicts about leisure choices. These distinctions become especially important when questionnaires equate ordinary engagement with addictive symptoms. Petry et al. (2014) WHO (n.d.)

Behavioral-addiction claims require evidence that the proposed syndrome is coherent, sufficiently reliable, impairing, clinically discriminable, and not simply a relabeling of existing psychopathology or social judgment. Clinical theory should accommodate incentives and learning mechanisms while retaining discriminant validity. A general similarity in reward processing is a hypothesis about a mechanism, not a diagnostic license. For the present dossier, behavioral addictions serve as an illuminating boundary comparison. Detailed gambling and gaming phenomenology belongs to the separate ADD-17 and ADD-18 scientific objects already defined in the research architecture; no second canonical monograph is created here. Wakefield (1992) Petry et al. (2014)

Development, Age, and Life-Course Transitions

Age shapes exposure, neural development, opportunity, social sanction, and the interpretive meaning of diagnostic criteria. An adolescent's roles and responsibilities differ from those of a midlife adult; the same item wording can therefore capture different degrees of functional disturbance. Substance experimentation during a particular developmental period cannot be assumed to lead inevitably to chronic disorder, and a period of remission cannot be interpreted as absence of previously established impairment. Studies need age-appropriate assessments of risk, persistent symptoms, and social functioning. The 2025 NSDUH publishes distinct youth, young-adult, and older-adult estimates, underscoring that population denominators and developmental context belong alongside prevalence claims. SAMHSA (2026)

The Vietnam veteran research associated with Robins and colleagues demonstrates why a clinical trajectory must be situated in its environment. The 1975 study interviewed 898 returnees concerning narcotic use during service in Southeast Asia and afterward, providing a historically important basis for examining how changes in setting accompany changes in drug use. The study does not prove that the environment is the sole cause of addiction or that every person follows a similar course. It establishes a demanding empirical question: why do apparently similar exposure histories generate different trajectories under changed social conditions? Proper assessment therefore records availability, uncertainty, distress, roles, and access to alternative rewards. Robins et al. (1975)

Longitudinal research can observe onset, persistence, remission, recurrence, and changes in functioning, but diagnostic transitions also reflect the measurement rules used at each wave. A participant may cross the two-criterion threshold with a small change in reported symptoms while an underlying risk continuum changes gradually. Conversely, a clinically important change may fail to alter a diagnosis. Investigators should therefore retain raw criteria and interval-specific exposures, not only yearly yes/no categories. In developmental research, differential attrition and retrospective reconstruction can systematically distort estimates of chronicity. A strong life-course account combines epidemiology with carefully interpreted clinical histories rather than treating any single calendar snapshot as a final outcome. Hasin et al. (2013) Robins et al. (1975)

Psychiatric Comorbidity and Differential Attribution

Anxiety, depressive symptoms, trauma-related difficulties, psychotic experiences, and attention-related problems may co-occur with substance use disorders, yet their overlap supports multiple possible temporal and causal explanations. Substance exposure may exacerbate psychiatric symptoms; psychiatric symptoms may alter motivations for use; shared vulnerabilities may influence both; and service selection can inflate observed associations in clinical populations. A diagnostic evaluation should document independent symptoms, longitudinal course, substance-induced episodes, and functional consequences. It should avoid making an individual's psychiatric history an automatic explanation for every later substance-related behavior. Contemporary classification provides separate rules for many substance-induced psychiatric syndromes, highlighting the need for attribution rather than casual grouping. WHO (2024) American Psychiatric Association (2022)

The phrase dual diagnosis is useful in service organization but can hide important diversity. A co-occurring disorder may be established before substance exposure, identified during withdrawal, or suspected because of symptoms in a period of intoxication. Reliable diagnosis depends on when observations were made and how alternative causes were addressed. Studies using one-time questionnaires often describe associations between symptom scales and substance use, while structured longitudinal assessments can better clarify temporal ordering. Even temporality alone does not settle causation because shared environmental conditions or preexisting vulnerability may generate both processes. Diagnostic research requires transparent separation of measured association, plausible mechanism, and verified clinical finding. NYSDOH AI (2024) Wakefield (1992)

A clinically responsible formulation preserves the possibility of more than one condition, but also guards against double counting. Role impairment may be associated with both depressive symptoms and problematic drinking; determining which disorder criteria are justified requires a careful narrative and explicit coding rules. The evidence for an SUD does not come solely from the presence of psychiatric comorbidity. Conversely, a person's ability to maintain employment or education does not categorically exclude disorder if other qualifying criteria are established. These examples show why classification needs full symptom information and professional clinical reasoning rather than shorthand assumptions about social identity, treatment status, or diagnostic labels. Hasin et al. (2013) American Psychiatric Association (2022)

Neurobiology and the Inferential Gap Between Mechanism and Diagnosis

Substance-related behaviors engage learning, motivation, stress regulation, attention, and physiological adaptation. Incentive-sensitization theory offers an account of enhanced cue-triggered wanting; the allostatic perspective emphasizes adaptations in reward and stress systems. These theories can be used to form testable hypotheses about craving, relapse vulnerability, and the changing effects of contexts across time. They operate at a different explanatory level from formal diagnostic thresholds. A response to a drug-associated cue cannot be translated directly into a DSM symptom count, and a brain-imaging difference at group level does not identify the diagnostic category of an individual participant. Robinson & Berridge (1993) Robinson & Berridge (2025) Koob & Le Moal (1997)

Group-level mechanistic findings require interpretation across scales. Experimental tasks often isolate a component process under highly controlled conditions; clinical interviews integrate behaviors and experiences over months in variable settings. Between these scales lie ecological exposure, life events, sleep, housing, relationships, co-occurring symptoms, and medically indicated treatment. A model can explain part of the mechanism of repeated seeking without explaining when a specific person experiences material impairment or how a classification distinguishes disorder from risk. The empirical task is to connect a proposed process to independently measured outcomes and to examine whether the relationship generalizes beyond the sample and setting in which it was discovered. Koob & Le Moal (1997) Hasin et al. (2013)

The clinical narrative of addiction as a chronic illness has influenced insurance, service organization, and expectations for continuing care. McLellan and colleagues advanced an influential argument for treating drug dependence with models comparable to other chronic medical illnesses. That account highlights relapse and long-term care needs, while leaving room for variation in course, context, and response to support. The historical Vietnam follow-up illustrates environmental change as one dimension of such variation. A useful synthesis recognizes an important neurobehavioral substrate while measuring recovery, remission, function, and social circumstances empirically. It avoids treating either a biological theory or an environmental observation as a total explanation. McLellan et al. (2000) Robins et al. (1975)

Misclassification, Stigma, and Ethical Research Practice

False-positive diagnosis can produce stigma, inappropriate service categorization, and the misinterpretation of expected medication withdrawal as addiction. False-negative diagnosis may obscure persistent impairment or delay needed evaluation. The costs differ by institution and population, so the decision threshold should be examined in relation to the purpose of assessment. A screening program designed to identify preventable harm may intentionally favor sensitivity, while a formal diagnosis requires a more specific evidentiary process. The conceptual distinction between harm and dysfunction gives this problem a deeper scientific basis: social judgments must not silently replace observations of disorder. Wakefield (1992) NYSDOH AI (2024)

Research communication should use language that distinguishes the person from a diagnosis and describes observed behavior precisely. A label such as addict can collapse dissimilar trajectories into a stereotype, while dependence can ambiguously refer to physiological adaptation or a clinical syndrome. Language used in a research instrument is itself part of measurement: ambiguous terms increase response variation and may undermine comparability across settings. In clinical writing, the purpose is clear documentation of criteria and consequences; in public communication, the same clarity reduces unwarranted moral inference. Classification systems remain accountable to the practical impact of their terminology. Hasin et al. (2013) Edwards & Gross (1976)

Consent, confidentiality, and risk communication are particularly important in research involving potentially illegal behaviors or stigmatized diagnoses. Disclosure of substance exposure may influence employment, custody, or legal proceedings, so study design should describe how sensitive data are protected and why a particular measurement is necessary. The ethics of research also require accurate representation of uncertainty. A preliminary screen should not be disclosed as a confirmed psychiatric disorder, and a statistical risk factor should not be recast as a personal diagnosis. Honest distinction among these claims is an essential part of scientific credibility. NYSDOH AI (2024) Saunders et al. (1993)

Reliability, Reproducibility, and Longitudinal Validation

A classification system should be evaluated by independently replicable research, not only by a consensus among its designers. Reliability studies ask whether different trained interviewers apply criteria consistently; latent-structure studies ask whether symptom patterns support an underlying model; prospective cohorts examine relationships between diagnostic assignments and later outcomes; and interventional studies may investigate whether a category usefully predicts response to an intervention. None of these designs alone completes a validation program. A credible diagnostic framework accumulates convergent evidence under openly stated measurement assumptions. Hasin et al. (2013) WHO ASSIST Working Group (2002)

The original DSM-5 review included analyses supporting a combined abuse-and-dependence dimension and a revised criterion set. Those analyses explain a historical decision; they do not settle questions about every population, time period, or substance. A replication program should compare new samples with original ones, use independently developed instruments where appropriate, and examine the behavior of individual items in more than one demographic group. In the presence of selective publication or researcher flexibility, apparently robust factor structures can be misleading. Registration of hypotheses, transparent code, and reporting of alternative specifications improve credibility. These methodological recommendations follow from general measurement logic and the specific kinds of evidence described in the original revision rationale. Hasin et al. (2013)

Temporal validity is a separate issue. A criterion count may remain stable while harm and functioning change, or may change due to an altered opportunity for substance use rather than a change in vulnerability. Repeated interviews, ecological assessments, and documented clinical outcomes can examine such discrepancies. Prospective prediction should be tested outside the sample used to optimize a diagnostic threshold. When evidence is thin, researchers should report model uncertainty rather than claim a universal or permanent boundary. A diagnosis earns scientific value by reliably organizing information and supporting clinically meaningful decisions across observable circumstances. McLellan et al. (2000) Robins et al. (1975)

Classification, Clinical Utility, and the Artificial Era

Digital assessment tools can increase the availability of repeated self-report, passive behavioral observations, and decision support, but the inferential distinctions in this monograph apply to them in full. An automated system may classify text mentioning drinking, craving, or withdrawal; that capacity does not make the classification equivalent to a verified clinical diagnosis. Training data can contain unexamined assumptions about age, culture, access to care, and moral judgments about use. Algorithms also learn the coding practices of source institutions, which may reflect prior diagnostic misclassification. The appropriate research object is the relationship between model output, evidence provenance, clinical interpretation, and independently observed outcomes. Wakefield (1992) Hasin et al. (2013)

The Artificial Era establishes new contexts for cognitive partnership and research organization. Repeated digital monitoring could support more precise temporal models of craving and harmful use, provided consent, privacy, measurement validity, and the right to revise interpretations are preserved. Computational systems can assist in comparing DSM and ICD rules or tracing evidence to diagnostic criteria, yet the scientific status of their output depends on transparent data, validated methods, and human clinical governance where health decisions are involved. A prediction of risk and a diagnosis answer different questions. Their difference should be encoded explicitly in any system architecture rather than left for the reader to infer. WHO (2024) WHO ASSIST Working Group (2002)

Research on artificial agents and human–artificial relations also invites careful conceptual comparison of habit, reward, incentive, control, and persistence. Similar formal descriptions may apply across biological and artificial systems, but they do not establish an ICD or DSM disorder in an artificial entity. Diagnostic constructs are defined within specific clinical and evidentiary contexts. The scientifically constructive path is to specify operational analogies, identify which processes can be observed, and test their predictive scope. This maintains the wider horizon From Homo to Artificial while protecting the precision of mental-health classification and respecting the distinct history of human clinical research. Robinson & Berridge (2025) Wakefield (1992)

Clinical Reasoning Through Illustrative Boundary Cases

Consider an adult who uses a prescribed opioid over an extended period, experiences expected physiological adaptation, and develops withdrawal symptoms when doses are reduced. This hypothetical presentation requires medical evaluation of treatment and tapering concerns. It does not, on those observations alone, satisfy the DSM-5-TR definition of opioid use disorder when tolerance and withdrawal occur solely during appropriate medical supervision. If additional qualifying features emerge, they must be assessed individually rather than presumed from the prescription history. The case illustrates the importance of documenting the cause of each criterion. It also demonstrates why opioid safety and possible use disorder require distinct clinical assessments. American Psychiatric Association (2022) NYSDOH AI (2024)

Consider a different hypothetical adult who experiences a serious alcohol-related injury during a single episode but has no established pattern meeting a dependence or DSM SUD threshold. The immediate harm is real, and preventive or medical intervention may be warranted. The classification question remains whether the episode fits a specific acute or harmful-use designation, whether a broader pattern exists, and whether there is evidence for sustained impaired regulation. The ICD-11 distinction between harmful episode, harmful pattern, and dependence provides an instructive framework; DSM criteria must be applied according to their own requirements. An episode of severe harm should never be minimized merely because evidence of a chronic disorder is incomplete. Saunders et al. (2019) WHO (2024)

A third hypothetical person repeatedly misses major obligations, reports unsuccessful attempts to reduce stimulant use, and continues despite awareness of serious related problems. Such observations warrant systematic criteria-based evaluation, including chronology, associated mental symptoms, and alternative explanations. The result cannot be inferred from the label stimulant user or from a positive urine test alone. These cases are conceptual demonstrations, not clinical records or validation datasets. Their value lies in showing where observation ends and diagnostic inference begins, and how precise definitions improve the reliability of both decisions and future research. Hasin et al. (2013) NYSDOH AI (2024)

Practical Uses of Classification in Prevention, Treatment, and Recovery Research

Classification serves several different practical functions. Screening identifies candidates for further assessment; diagnosis organizes a clinical formulation; severity coding supports communication and research stratification; and repeated measurement describes change. Prevention studies may primarily evaluate reductions in harmful exposure or acute injury, while treatment studies may evaluate symptom remission, sustained functioning, reduced mortality, patient-defined recovery, or engagement with services. These are distinct endpoints. Investigators should state whether their primary outcome concerns fewer symptoms, a transition below a threshold, reduced harm, or improved quality of life. A positive change in one domain need not imply the same degree of improvement in another. Saunders et al. (1993) McLellan et al. (2000)

Diagnostic thresholds are especially relevant to the interpretation of remission. A participant can move from three criteria to one and cease meeting a current DSM disorder threshold, while still reporting craving, vulnerability, or significant life consequences. Another person may continue to meet a threshold but achieve major functional gains. Both observations can be clinically meaningful. Studies should report the distribution of individual symptoms, the time since last use, relevant treatments, and separately measured health outcomes. The chronic-illness perspective advanced by McLellan and colleagues highlights the importance of continuing care, but a full scientific outcome model must also accommodate diverse trajectories and recovery experiences. McLellan et al. (2000) Hasin et al. (2013)

Population data should guide equitable resource planning without turning a survey estimate into a treatment directive for every respondent. The 2025 NSDUH measures past-year diagnostic conditions and service contacts using specific definitions; its estimates of unmet or received care need to be interpreted within those definitions. People with similar survey diagnoses may have very different preferences, acute risks, co-occurring conditions, and opportunities for care. Public-health planning can use prevalence, service access, and harm indicators together, while individual clinical decisions require professional evaluation and the person's goals. This distinction preserves both the social value of surveillance and the dignity of the person whose experiences are classified. SAMHSA (2026) NYSDOH AI (2024)

Critical Evidence Assessment and Competing Interpretations

The reviewed evidence has several strengths. Foundational clinical accounts identify historically influential patterns of dependence; the DSM-5 revision rationale documents a large empirical and conceptual classification program; WHO's ICD-11 clinical descriptions provide an internationally coordinated diagnostic standard; the original AUDIT and ASSIST papers establish the provenance of widely used screening approaches; and the federal NSDUH report presents defined, recent population estimates. These sources support different kinds of claims. The original diagnostic papers and manual descriptions establish what a rule states, whereas prospective validation must establish how it performs in a particular population and toward a particular outcome. Edwards & Gross (1976) Hasin et al. (2013) WHO (2024) Saunders et al. (1993) WHO ASSIST Working Group (2002) SAMHSA (2026)

The limitations are equally important. Classification consensus is not equivalent to a laboratory-confirmed disease marker. Large factor-analytic datasets can support a dimension but may leave unresolved the best threshold and the distinct implications of individual criteria. Historical follow-up studies provide valuable contextual observations but cannot isolate every social mechanism. Population surveys describe observed prevalence under their algorithms and sampling schemes; their findings do not themselves demonstrate treatment efficacy or the biological cause of disorder. Clinical narratives and philosophical arguments sharpen definitions yet must not be reported as independent randomized tests. These differences in evidence type govern the strength of every conclusion. Hasin et al. (2013) Robins et al. (1975) SAMHSA (2026) Wakefield (1992)

Two important competing interpretations can be productively tested. One treats substance-related problems mainly as a severity continuum, with categorical thresholds imposed for clinical use; another assigns distinctive significance to dependence and harmful-use syndromes with partly different trajectories. Similarly, neurobehavioral models emphasize sensitized wanting or stress-system adaptations, while environmental research highlights the effect of availability, roles, and opportunity on drug use. The research literature supports useful mechanisms and observations within each tradition, but the claims do not collapse into one global explanatory variable. Discriminating tests require multiple measures and well-defined outcomes; rhetorical assertions that addiction is only a brain process or only a social condition are scientifically insufficient. Hasin et al. (2013) Saunders et al. (2019) Robinson & Berridge (2025) Koob & Le Moal (1997) Robins et al. (1975)

Unresolved Questions and a Research Program for Diagnostic Validity

A first open problem concerns thresholds. Researchers should investigate whether the clinical costs of false positives and false negatives vary systematically across substance classes and settings. This requires studies that apply the same criterion-level assessment to community and clinical samples, compare formal algorithms, and follow subsequent impairment and harms. A second concerns heterogeneity: equivalent DSM criterion counts can arise from different symptom combinations, some of which may carry distinct prognostic significance. Item-level models and individual longitudinal histories can address the issue, provided predictive tests use independent samples. A third concerns cultural and developmental invariance, where superficially similar questions may function differently because environments and obligations differ. Hasin et al. (2013) Wakefield (1992)

Another research priority is accurate identification of medication-related physiological adaptation. Studies should record indication, supervision, treatment course, documented misuse where present, withdrawal circumstances, and each qualifying nonpharmacological symptom. This would allow estimates of diagnostic misclassification and improve the interpretation of administrative health data. Digital tools may support such structured records, but data security and validation requirements are substantial. Researchers should separate risk prediction from diagnostic assignment, establish clinician agreement, and publish error analyses that include cases in which automated classification fails. These tasks will become more important as Artificial Era systems assist with clinical documentation and public-health research. NYSDOH AI (2024) WHO (2024)

Further work is needed on the relation among clinical syndrome, acute medical danger, and recovery. Prospective studies could jointly measure changing exposure, harms, symptoms, treatment engagement, social circumstances, and self-defined functioning rather than treating one diagnostic transition as the entire outcome. Natural experiments and carefully controlled interventions can help distinguish causal effects from shared vulnerabilities when the design justifies such inference. Evidence from the original Vietnam follow-up encourages attention to context, while the long-term-care literature encourages attention to persistence and recurrence. Both point toward a research program in which the diagnostic category is an evolving summary of a better-measured trajectory. Robins et al. (1975) McLellan et al. (2000)

Scientific Synthesis: An Integrated Boundary Framework

The evidence supports an integrated explanatory structure with four observational domains and two inferential decisions. The first domain is exposure: substance, dose, route, timing, pattern, and context. The second is acute consequence and adaptation: intoxication, injury, tolerance, and withdrawal, assessed by substance-specific rules. The third is behavioral and psychological organization: craving, control, priority, persistence, role functioning, and attempts to change. The fourth is longitudinal setting: development, social opportunities, treatment circumstances, co-occurring conditions, and recovery. The first inferential decision asks which observations qualify under a selected diagnostic classification; the second asks how valid and clinically useful that classification is for a defined purpose. This arrangement synthesizes existing evidence and classification practice rather than claiming a newly discovered disorder or a novel experimental mechanism. Hasin et al. (2013) WHO (2024) Saunders et al. (2019)

The model clarifies several persistent ambiguities. Frequent exposure is an observation, not an automatic diagnosis. Pharmacological adaptation describes a process that may occur during appropriate medical care. Acute harm is significant even where a chronic syndrome is unestablished. Craving and impaired control are clinically meaningful indicators, but require context, chronology, and applicable diagnostic rules. DSM-5-TR and ICD-11 organize these indicators into distinct architectures, and both must be evaluated for reliability and external validity. A good classification helps clinicians communicate and researchers compare evidence while remaining corrigible when new data expose limitations. American Psychiatric Association (2022) NYSDOH AI (2024) Hasin et al. (2013)

Artificial systems can contribute to this research by enabling explicit provenance, consistent coding of observations, cross-system rule comparison, and longitudinal modeling, provided their output is critically validated. The most durable scientific insight is that diagnostic boundaries are established through accountable inference from evidence. A reader should be able to reconstruct how a claimed disorder arose from observed criteria and why an alternate explanation was rejected or retained. That standard binds clinical care, epidemiological reporting, and the future study of psychological organization in the Artificial Era. The relevant discipline remains empirical psychology informed by clinical science, medical classification, and the ethics of meaningful human outcomes. Wakefield (1992) WHO ASSIST Working Group (2002)

Conclusion

Addiction and substance use disorders are best investigated through a principled separation of exposure, risk, actual harm, pharmacological adaptation, behavioral dysregulation, and formal diagnostic classification. The historical dependence concept developed into distinct modern classification systems, with DSM-5-TR using a graded substance use disorder criterion count and ICD-11 retaining separate dependence and harmful-use categories. The systems share significant clinical concerns but organize them differently. Their differences should be stated precisely in scientific work, especially when study populations, symptom windows, and national diagnostic practices vary. Edwards & Gross (1976) Hasin et al. (2013) Saunders et al. (2019) WHO (2024)

The empirical record supports the value of syndrome-based clinical assessment, screening tools with explicit purposes, large population surveys with defined denominators, and multiple neurobehavioral and contextual explanations of problematic substance use. It also shows why these forms of evidence must retain their respective inferential limits. The DSM-5 revision rationale supports a measurable disorder dimension, yet the question of a clinically optimal threshold remains open to validation. The ICD-11 categories permit differentiated attention to hazardous exposure, harmful use, and dependence. Neither classification removes the need to distinguish physiological tolerance from a disorder, acute intoxication from a chronic pattern, or social consequences from clinical dysfunction. Hasin et al. (2013) Saunders et al. (2019) Wakefield (1992)

A rigorous scientific and clinical account reports observable criteria, the rules used to interpret them, the populations to which estimates apply, and the outcomes through which diagnostic validity is evaluated. The continued development of addiction psychology depends on longitudinal measurement, cross-cultural assessment, transparent classification crosswalks, individual symptom trajectories, and research designs capable of examining competing explanations. Within the Artificial Era, the same evidentiary discipline will govern the use of digital classification and decision-support systems. Diagnostic precision is the foundation on which meaningful prevention, recovery research, and responsible psychological science can be built. SAMHSA (2026) McLellan et al. (2000) WHO (2024)

References