Empirical changes in DSM conceptualisations of ADHD

To ground the theoretical discussion, key empirical findings from our prior systematic analysis of DSM descriptions of ADHD ^{Reference Ophir, Shir-Raz and Tikochinski4} are reproduced here in condensed form.

Finding 1: across DSM editions, ADHD has shifted from a primarily behavioural disorder to a neurodevelopmental condition, with an emphasis on biological and cognitive underpinnings.

Finding 2: although initially conceptualised as a disorder of young boys, ADHD has evolved into a lifespan condition spanning ages and genders. Reported prevalence in children increased from 3 to 7.2%, whereas the male-to-female ratio declined from 10:1 to 2:1. DSM-5 recognised adult ADHD (2.5% prevalence), lowered symptom thresholds for older individuals and extended the age-at-onset criterion from 7 to 12 years.

Finding 3: the definition of impairment has broadened beyond academic difficulties to include occupational, social and emotional domains. Later editions also lowered the threshold for impairment, allowing diagnosis even in cases of relatively minor functional difficulties.

Finding 4: diagnostic criteria have become more flexible, moving beyond rigid subtypes to accommodate changing presentations and subthreshold cases (e.g. unspecified ADHD).

Finding 5: the scope of ADHD comorbidities and differential diagnoses has expanded significantly; DSM-5-TR lists 17 differential diagnoses, excluding normal behavioural variations.

Finding 6: alongside the neurodevelopmental framing, later DSM editions also include cultural and environmental factors shaping symptom expression.

Taken together, these patterns reflect a substantial expansion in the conceptual scope of ADHD over time. The interpretation of this expansion, however, remains debated, with some viewing it as having improved diagnostic sensitivity and others raising concerns about the validity and practical usefulness of the diagnosis. ^{Reference Ophir, Shir-Raz and Tikochinski4}

Diagnostic revisions and conceptual tensions in psychiatric classification

Before examining the implications of the patterns outlined above, it will be useful to situate the discussion within a broader philosophy-of-science framework with respect to how scientific theories respond to conflicting observations. A central concern articulated by Karl Popper is that when empirical findings seem to be inconsistent with an established theory, there may be a tendency to introduce auxiliary assumptions that accommodate discordant evidence while preserving the core theory. Such adjustments, in Popper’s formulation, may ‘rescue a theory from refutation only at the price of destroying, or at least lowering, its scientific status’. ^{Reference Popper5}

This perspective does not preclude updating of theories in light of new evidence. However, when theories of categorical psychiatric constructs (e.g. ADHD and autism spectrum disorder) are modified, maintaining their clinical meaning and conceptual coherence remains essential. Psychiatric diagnoses, following revision, are expected to retain adequate construct validity. This includes capturing consistent patterns of experience and behaviour (reflecting convergent validity), being distinguishable from other conditions (discriminant validity), and remaining distinct from normative variation, as reflected in the DSM definition of mental disorder (Box 1). In the case of ADHD, the following analysis examines how successive DSM revisions bear on questions of construct validity in psychiatric diagnosis, thereby helping to clarify broader conceptual tensions in psychiatric nosology.

Prevalence and the threshold of deviance

The first conceptual tension concerns changes in the reported prevalence of ADHD across successive DSM revisions (finding 2), a central focus of ongoing debate. ^{Reference Cortese, Daley, Hollis, Rae, Ani and Asherson6,Reference Pearson7} Notably, ADHD was described as a ‘common’ disorder from its initial formulation in DSM-III (1980). Widely regarded as a turning point in psychiatry, DSM-III sought to address concerns regarding diagnostic inconsistency and limited reliability in earlier childhood labels (e.g. minimal brain dysfunction) by consolidating them into a single diagnostic construct. The manual justified this consolidation by noting that ‘attentional difficulties are prominent and virtually always present among children with these [earlier] diagnoses’. ⁸ It further estimated that the diagnosis ‘may occur in as many as 3% of prepubertal children’.

Although a prevalence estimate of 3% may appear modest by contemporary standards, the characterisation of ADHD as ‘common’ raises a central conceptual challenge in psychiatric diagnosis, namely, distinguishing clinically significant conditions from normative variation in behaviour and life difficulties – what may be described as the ‘threshold of deviance’. ^{Reference Davis9} This challenge becomes more pronounced as prevalence increases, reaching ‘about 7.2% of children’ in DSM-5-TR (finding 2) and extending into double-digit prevalence estimates in various national surveys. ^{Reference Danielson, Bitsko, Ghandour, Holbrook, Kogan and Blumberg10,Reference Visser, Danielson, Bitsko, Holbrook, Kogan and Ghandour11} Indeed, some scholars have argued that the true prevalence of ADHD remains around 5% (in the UK) when carefully assessed using standardised procedures. ^{Reference Cortese, Daley, Hollis, Rae, Ani and Asherson6} However, this position leaves open the question of how substantially higher rates (reported in other settings and populations) should be interpreted. In the USA, for example, Centers for Disease Control and Prevention (CDC) data indicate prevalence estimates of approximately 20% among young adolescent boys. ^{Reference Danielson, Bitsko, Ghandour, Holbrook, Kogan and Blumberg10,Reference Visser, Danielson, Bitsko, Holbrook, Kogan and Ghandour11} In Israel, data from a large-scale health maintenance organisation, originally collected to examine associations between ADHD and COVID-19, indicated that more than 20% of individuals aged 5–20 years had received an ADHD diagnosis. ^{Reference Merzon, Manor, Rotem, Schneider, Vinker and Golan Cohen12} Notably, these figures were based on formal medical records rather than self-report and were consistent with earlier Israeli findings showing that application of DSM-5 criteria increased prevalence estimates from 12.7% under DSM-IV to approximately 21%. ^{Reference Rigler, Manor, Kalansky, Shorer, Noyman and Sadaka13}

These elevated prevalence rates have become an explicit topic of discussion within contemporary clinical and scientific discourse. A recent Nature news feature, titled ‘ADHD diagnoses are growing. What’s going on?’, summarised national surveillance data indicating a sustained increase in diagnostic rates over recent decades, with rates approaching 20% among American adolescent boys. ^{Reference Pearson7} Similarly, a recent New York Times Magazine feature cited leading experts who questioned the expansion of the diagnosis and the underlying assumptions. ^{Reference Tough14} Comparable concerns are reflected in findings from an Israeli survey indicating that most physicians specialising in neurology, psychiatry and child development believe that a substantial proportion of ADHD diagnoses are unjustified. ^{Reference Davidovitch, Shmueli, Rotem and Bloch15} Furthermore, diagnostic prevalence may be higher than reported in light of the contemporary conceptualisation of ADHD as a lifelong neurodevelopmental condition (finding 1). Within this framework, individuals diagnosed later in life are assumed to have experienced the condition earlier, consistent with DSM criteria requiring symptom onset before the age of 12 years (and previously before age 7 years). This implies that ADHD prevalence is cumulative across the lifespan, and that cross-sectional estimates may underestimate the proportion of individuals who eventually meet diagnostic criteria. Retrospective studies from Israel support this view, showing that when young adults are asked whether they have ever received an ADHD diagnosis, reported rates rise to approximately 23–30%, reflecting delayed diagnosis in many cases. ^{Reference Ophir16}

This epidemiological pattern carries important validity implications. If rising ADHD prevalence reflects a genuine increase in the underlying condition over a short period of four decades, this creates two conceptual problems. First, such a rapid population-level increase is difficult to reconcile with biological and evolutionary expectations – a difficulty that has led some theorists to propose the ‘evolutionary mismatch’ hypothesis. ^{Reference Hartmann17} According to this account, ADHD-like traits have not suddenly become more prevalent; rather, they may have long existed without being framed as pathological because they were neutral or even advantageous in ancestral environments. Traits such as motor restlessness, rapid attentional shifting, heightened responsiveness to novelty and intense focus on immediately salient tasks may have supported exploration, vigilance and flexible responses to changing conditions in hunter-gatherer societies. Under modern educational and occupational demands, however, these same traits may be more readily experienced and classified as impairing. From this perspective, rising diagnostic rates reflect less a genuine increase in the underlying traits than a change in the environmental demands against which they are judged and classified. Second, prevalence estimates approaching double-digit proportions challenge the fundamental threshold of deviance described above. After all, children’s behaviours vary naturally, and increasingly broad diagnostic boundaries risk blurring the distinction between disorder and normative variation.

One commonly proposed response to this concern is overdiagnosis – that a proportion of diagnoses may be clinically unjustified. ^{Reference Kazda, Bell, Thomas, McGeechan, Sims and Barratt18} However, this concept is most appropriate in situations in which only a relatively small proportion of diagnoses represent false positives within an otherwise reliable diagnostic system, that is, a system with a limited margin of error. In the case of ADHD, elevated prevalence rates across many settings often reach 2–3 times the already high estimate reported in DSM-5-TR (7.2%). Such a discrepancy is difficult to attribute solely to clinician error. When diagnostic guidelines are applied in good faith yet consistently yield prevalence rates far exceeding those anticipated by the nosological framework, the source of the discrepancy may lie less in the clinicians who apply the criteria than in the criteria themselves. ADHD may represent a particularly clear illustration of this issue, but other highly prevalent diagnostic constructs, including anxiety and depressive disorders, post-traumatic stress disorder and, to some extent, autism spectrum disorder, may be vulnerable to similar validity concerns.

Gender and diagnostic heterogeneity

A related structural tension emerges in the evolving gender distribution of ADHD (finding 2). When first introduced, ADHD was diagnosed far more frequently in boys than in girls, with the ratio at times approaching 10:1 (e.g. DSM-III). Such a pronounced disparity posed a conceptual challenge, particularly as ADHD came to be framed as a neurodevelopmental condition with a presumed biological basis (finding 1). For conditions presumed to have a biological basis, large sex differences in prevalence typically invite explicit aetiological explanation; where no disorder-specific explanation has been established, such disparities may also reflect, at least in part, normative sex differences.

Recent DSM editions seem to address this discrepancy by reporting a narrower male-to-female ratio (2:1) and offering a phenomenological account, according to which ‘females are more likely than males to present primarily with inattentive features’ (finding 2). This revision is often understood as an attempt to address underrecognition of girls and women with less disruptive and more internally oriented presentations. At the same time, the broader inclusion of distinct symptom profiles introduces a different challenge: predominantly inattentive, internally oriented patterns and overtly hyperactive or disruptive patterns may be subsumed under a single neurodevelopmental condition, even when they show little or no overlap. Although such variability is not inherently implausible, it increases the heterogeneity of the diagnostic construct and further expands its conceptual boundaries (finding 4).

From a philosophy-of-science perspective, this tension echoes the broader concern noted above that modifications introduced to reconcile conflicting observations may come at the cost of reduced conceptual precision. ^{Reference Popper5} From a clinical perspective, broadening the range of symptom presentations may increase the number of individuals meeting diagnostic criteria, further contributing to rising prevalence, as discussed in the previous section.

Aetiological framing and explanatory coherence

The inclusion of two distinct symptom profiles within a single diagnostic category invites closer examination of the explanatory assumptions underlying the diagnosis. DSM-5 (2013), the first edition to include a separate category for neurodevelopmental disorders, further consolidated the framing of ADHD as a neurodevelopmental condition with intrinsic biological origins (finding 1). Accordingly, ADHD is often described in the literature as a ‘disorder of the brain’. ^{Reference Hoogman, Bralten, Hibar, Mennes, Zwiers and Schweren19} At the same time, the DSM increasingly acknowledges cultural, environmental and contextual influences on symptom expression and diagnostic patterns (finding 6), consistent with growing evidence of variation in ADHD prevalence across countries and communities (e.g. Ophir ^{Reference Ophir16} and Wong ^{Reference Wong and Landes20} ). For example, a large CDC survey reported that ADHD rates differed across nearly all major sociodemographic variables (e.g. age, gender and ethnicity). ^{Reference Visser, Danielson, Bitsko, Holbrook, Kogan and Ghandour11} Likewise, an expanding literature has identified environmental correlates of attentional difficulties, including educational contexts and patterns of digital media exposure. ^{Reference Ra, Cho, Stone, De La Cerda, Goldenson and Moroney21} This aetiological tension is manifested in a formal diagnostic caveat introduced in recent DSM editions (Box 1). According to DSM-5-TR, signs of ADHD may be ‘minimal or absent’ during highly stimulating activities or in ‘one-on-one situations (e.g. the clinician’s office)’ (italics added). ²² The implications of this caveat are twofold: clinically, it implies that diagnosis may rely on retrospective or third-party reports even when observable symptoms are minimal. Theoretically, it complicates the neurodevelopmental framing (finding 1), which locates the disorder in intrinsic biological mechanisms, by acknowledging that symptom expression may vary substantially, and at times disappear, across contexts.

A potential resolution to this tension may be found in the framing in DSM-5-TR of contextual influences as factors that shape symptom presentation rather than as causal mechanisms. However, in the absence of clear criteria for distinguishing between these categories, the boundary between ‘influences’ and ‘causes’ remains difficult to delineate. This difficulty reflects the inherently integrated nature of human behaviour, which is shaped by both nature and nurture, alongside the continued absence of validated biological markers and consistent neuroimaging differences in ADHD, as acknowledged in DSM-5-TR (Box 1). Although the APA’s Future DSM Strategic Committee has proposed integration of biological factors into the DSM, it has also cautioned that such advances may have broader implications for the field (see ‘Synthesis and implications’). Furthermore, the growing recognition of environmental and contextual influences brings the diagnosis of ADHD closer to the broader DSM definitional caveat, according to which an expectable ‘response to a common stressor’ or ‘conflicts that are primarily between the individual and society’ do not qualify as mental disorders unless they reflect an underlying ‘dysfunction in the individual’ (Box 1). ²² Specifically in children, this caveat makes it difficult to distinguish between neurodevelopmentally driven symptoms and conflicts with social expectations (e.g. school demands) or expectable responses to common stressors (e.g. educational pressures).

From a philosophy-of-science perspective, this tension reflects a broader challenge: how to integrate emerging aetiological insights into an existing conceptual framework without compromising its explanatory coherence or blurring the boundary between psychopathology and normative variation. From a clinical perspective, this tension has implications for intervention. When ADHD is conceptualised primarily as a biologically based condition, treatment approaches tend to emphasise pharmacological interventions aimed at compensating for presumed underlying neurobiological deficits. Conversely, greater emphasis on environmental and contextual influences may support complementary approaches focused on educational environments, behavioural supports and social contexts. ^{Reference Timimi and Leo23,Reference Ophir, Rosenberg, Tikochinski and Efrati24}

Medical framing and diagnostic boundaries

A central question emerging from the previous discussion is how to account for the increasing prominence of a neurodevelopmental framework (finding 1), despite the continued absence of clear neurobiological markers and the growing recognition of contextual influences (finding 6). Although financial and institutional factors have been discussed in this context, one possible explanation lies in the social process of medicalisation, whereby aspects of human behaviour and life challenges come to be viewed through a medical lens. ^{Reference Searight and McLaren25} This process may be reflected in the expansion of the diagnostic scope of ADHD (findings 2–4); its increasingly neurodevelopmental framing (finding 1); and the growing emphasis on associated adverse outcomes and comorbidities, such as substance use, suicidal behaviour and premature mortality (findings 3 and 5). Where robust evidence supports causal relationships with such outcomes, their inclusion serves an important clinical function. However, when these relationships remain uncertain and open to multiple interpretations, including possible confounding related to treatment effects, ^{Reference Ophir26} their prominence in diagnostic descriptions may contribute to further medicalisation and heightened perceptions of disorder severity. These considerations may be particularly consequential in milder and less specified cases of ADHD introduced in recent DSM editions (findings 3 and 4), given that medical interventions also carry potential risks. At a broader conceptual level, they may further contribute to the blurring of boundaries between normative human behaviour and formal medical conditions.

Comorbidity, differential diagnosis and discriminant validity

In addition to the clinical implications discussed above, the expanding list of adverse outcomes and comorbid disorders (findings 3 and 5) challenges the discriminant validity of ADHD. DSM-5 and DSM-5-TR associate ADHD with a wide range of conditions, including autism spectrum disorder, anxiety disorders and major depressive disorder. They further note that ‘approximately half of children with the combined presentation’ meet criteria for oppositional defiant disorder. This pattern is consistent with findings from the longitudinal Milwaukee study, which reported that in more than 80% of clinic-referred ADHD cases, the individual presented with at least one comorbid disorder, and more than 50% met criteria for two or more. ^{Reference Barkley and Fischer27}

Although Berkson’s bias may partly contribute to the high comorbidity rates observed in clinical samples, as individuals with multiple difficulties are more likely to seek treatment, comorbidity remains central to the diagnosis itself, as reflected in the growing number of differential diagnoses (finding 5). Whereas DSM-III and DSM-III-R listed only 5 or 6 conditions (including ‘age-appropriate overactivity’), DSM-5 and DSM-5-TR expanded this list to 16 and 17, respectively. These figures are unusually high compared with those for other neurodevelopmental conditions, and some analyses suggest that they could be even higher, as a range of non-psychiatric states and even positive factors (e.g. giftedness) can produce behavioural presentations resembling ADHD symptoms. Despite this expansion, recent DSM editions no longer include the earlier consideration of ‘age-appropriate overactivity’ as part of differential diagnosis (finding 5), nor do they refer to related concepts such as ‘normal development’ or ‘normal variations’. This shift, together with the reduced threshold for dysfunction (finding 3), leaves the boundary between disorder and typical developmental behaviour less explicitly defined and may further contribute to the rising prevalence of ADHD (finding 2). This difficulty in distinguishing normal from pathological behaviour may be particularly pronounced in ADHD but is likely to be relevant to other diagnostic constructs, particularly ‘in the absence of clear biological markers’, as acknowledged in DSM-5-TR (Box 1).

Synthesis and implications

Taken together, the results of the conceptual examination of the evolving DSM descriptions of ADHD (findings 1–6) highlight five interrelated domains through which successive diagnostic revisions give rise to broader tensions in psychiatric classification:

1. (a) prevalence and the threshold of deviance;

2. (b) gender and diagnostic heterogeneity;

3. (c) aetiological framing and explanatory coherence;

4. (d) medical framing and diagnostic boundaries;

5. (e) comorbidity, differential diagnosis and discriminant validity.

Considered collectively, these domains capture key challenges in maintaining construct validity in psychiatric nosology. The analysis of ADHD presented here provides a particularly instructive illustration of these challenges, given its high prevalence in certain populations (finding 2) and the tension between its neurodevelopmental framing and the recognition of environmental influences (findings 1 and 6). The patterns of comorbidity and differential diagnosis for ADHD further complicate its distinction from other psychiatric and medical conditions, as well as from typical developmental variation (finding 5), particularly following the inclusion of milder and subthreshold presentations (findings 3 and 4). In this respect, ADHD exemplifies a central challenge in psychiatric classification: how to refine diagnostic constructs without progressively compromising their ability to capture consistent patterns of experience (convergent validity); how to distinguish the disorder from other conditions (discriminant validity); and how to maintain meaningful boundaries between disorder and normative variation, in line with the DSM definition of mental disorder (Box 1).

One of the solutions proposed by the APA’s Future DSM Strategic Committee to address such classification challenges was a shift towards more biologically grounded diagnostic categories. ¹ However, as the committee acknowledged, progress on this front may create an institutional paradox: the identification of robust biological mechanisms could, in some cases, shift conditions beyond the scope of psychiatry into other areas of medicine. The committee further notes that precedents of this kind (e.g. Rett syndrome) are ‘concerning, as [they] might lead to an erosion, if not erasure, of the field as neuroscience advances’ (italics added). ^{Reference Oquendo, Abi-Dargham, Alpert, Benton, Clarke and Compton2} This formulation echoes Popper’s observation that attempts to accommodate empirical complexity may ‘rescue a theory from refutation only at the price of destroying, or at least lowering, its scientific status’. ^{Reference Popper5}

Whether efforts to identify biological markers ultimately succeed or not, the epistemic status of psychiatric diagnosis remains largely unchanged. In the continued absence of such markers (Box 1), psychiatric diagnosis relies primarily on clinical judgement and is best understood at the level of behavioural syndromes, that is, clusters of co-occurring experiences and behaviours. This entails important epistemic limits for psychiatric diagnosis: diagnostic labels describe patterns of distress but do not, in themselves, explain them (e.g. a diagnosis of major depressive disorder does not explain depressive symptoms). In this context, DSM categories may be best understood as shared classificatory tools that enable clinicians and patients to identify and communicate recurring patterns of human experience. Challenges arise when these descriptive constructs are interpreted as concrete biological entities, a process often described as diagnostic reification. ^{Reference Fried28} In contrast to some other medical fields, in psychiatry, language can contribute to the construction of subjective reality. ^{Reference Crowe29} As illustrated in the present analysis, revisions of diagnostic language may reshape how behaviours are perceived, labelled and treated. Moreover, diagnostic categories can influence how individuals understand themselves, as diagnostic labels often carry implicit psychosocial scripts. This dynamic may be especially visible in childhood conditions such as ADHD, ^{Reference Kazda, Bell, Thomas, Hardiman, Heath and Barratt30} but similar processes are likely to operate across a wide range of psychiatric conditions.

In light of these considerations, future diagnostic frameworks would benefit from attention to the epistemic foundations of psychiatric diagnosis and the conceptual tensions discussed in this article. Until robust biological markers become available, psychiatric diagnoses are best understood in relation to the DSM’s own principled definition of mental disorder (Box 1). This definition is familiar to most clinicians, yet, when viewed in light of the issues examined here, several of its elements take on renewed significance.