Skip to content

The diagnostic boundary

TL;DR — Whether GAD is a distinct disorder is an open, actively argued question, and the arguments come from inside the field rather than from its critics. Four separate lines of evidence pull toward not distinct: taxometric analyses in both community (n=2,061) and clinical (n=1,175) samples find no taxon — worry and generalized anxiety behave as continua, making any threshold arbitrary (Marcus 2014, PMID 24377439; Kertz 2014, PMID 24334160); bivariate twin models put the genetic correlation between GAD and major depression at +1.00 in women and +0.74 in men (Kendler 2007, PMID 17121688); symptom-specificity analysis in a community sample found that of the DSM-IV GAD symptoms only muscle tension distinguished GAD from non-GAD, with fatigue, restlessness and concentration difficulty more prevalent in major depression (Faravelli 2012, PMID 22578985); and the diagnosis's own architects have published a case for deleting its defining criterion (Ruscio 2024, PMID 39364896). Pulling the other way: GAD is separable prognostically and practically — pure GAD is common in primary care (3.8% vs 1.6% comorbid GAD/MDE in a 20,000-patient study; Wittchen 2002, PMID 12044105), impairment in pure GAD matches pure MDD (Hoffman 2008, PMID 17146763), and uncontrollability of worry has demonstrated incremental validity over every other candidate criterion (Hallion 2013, PMID 23713499). This page keeps the dispute open. It is also where the condition's central methodological hazard lives: a pooled anxiety-disorder effect size is not a GAD effect size.

Four criterion disputes, and what the data say

Dispute Proposal Evidence for Evidence against Status
Excessiveness Delete it Removing it raises global lifetime prevalence 2.6%→4.0%, with larger increases in LMICs; non-excessive cases match excessive cases on sociodemographics, family history of GAD, risk of secondary comorbidity and suicidality, and often seek treatment (Ruscio 2024, PMID 39364896). In the US NCS-R the same removal raised lifetime prevalence ~40% (Ruscio 2005, PMID 16300690). Reliability rises from κ=0.53 to κ=0.78 (Wittchen 1995, PMID 7666382) Excessive cases are more severe, earlier-onset and more chronic on average (Ruscio 2005, PMID 16300690); excessiveness does predict self-reported anxiety, depression and stress (Rutter 2015, PMID 25465882) Unresolved. No revision has adopted it; ICD-11 already does not require it (Domschke 2025, PMID 40728738)
Uncontrollability Delete it (proposed pre-DSM-5) Conceptual overlap with excessiveness (Hallion 2013, PMID 23713499) Uncontrollability predicted GAD severity, clinician-rated anxiety, comorbidity count and treatment use over and above excessiveness; the reverse did not hold (Hallion 2013, PMID 23713499). Independently, uncontrollability predicted clinical severity and number of diagnoses while excessiveness predicted self-report scales (Rutter 2015, PMID 25465882) Retained, and the better-supported of the two worry criteria
6-month duration Shorten to 1–3 months No difference between 6-month GAD and 1–5-month GAD in age of onset, severity, persistence, comorbidity or impairment across 17 countries, n=85,052 (Lee 2009, PMID 19091158); in the Zurich cohort the rule excludes about half of those actually in treatment (Angst 2006, PMID 16734945) ≥12-month GAD is the most severe and persistent subgroup, so duration is not information-free — it just does not cut at six months (Lee 2009, PMID 19091158) Unresolved; DSM-5 kept 6 months, the DSM-5 workgroup literature proposed 3 (Andrews 2010, PMID 20058241)
≥3 of 6 associated symptoms Relax or restructure Relaxing duration + excessiveness + symptom count more than doubles prevalence, and the broadened category still predicts first onset of secondary disorders (Ruscio 2007, PMID 17118626). ICD-11 specifies no minimum count (Domschke 2025, PMID 40728738) An alternative proposal goes the other way — require ≥4 of 7 tension/vigilance symptoms plus ≥1 hyperarousal symptom to sharpen the GAD–depression border (Starcevic 1999, PMID 9885394) Unresolved and bidirectional: proposals exist to both loosen and tighten it

The pattern is diagnostic in itself. Every criterion that has been tested for incremental validity except uncontrollability has failed the test, and the failures have been published for twenty years without changing the manual.

Is GAD categorical at all?

Taxometrics asks whether indicators of a construct have a latent categorical (taxonic) or continuous structure. For GAD, three independent applications converge:

Study Sample Indicators Result
Marcus 2014 (PMID 24377439) 2,061 midlife US adults (MIDUS), plus 1,228 re-interviewed 10 years later DSM-IV-TR GAD criterion items Multiple taxometric procedures gave no evidence of a taxon; results consistent with a continuum. Authors note dichotomising loses statistical power and that thresholds are "likely to be arbitrary"
Kertz 2014 (PMID 24334160) 1,175 (PSWQ-A) and 638 (GAD-7) clinical outpatients Worry and GAD symptom items MAXCOV, MAMBAC and L-Mode all support dimensional structure — the first clinical-sample test
Faravelli 2012 (PMID 22578985) Sesto Fiorentino community study; 105 DSM-IV GAD cases Symptom-level specificity Only 17.1% of GAD cases had no comorbid DSM-IV disorder; only excessive worry and muscle tension were specific to GAD; restlessness, concentration problems and fatigue were more common in MDD

There is a rejoinder, and it is not statistical. Dimensionality of a construct does not by itself invalidate a diagnostic threshold — hypertension and hyperlipidaemia are dimensional and clinically useful. The GAD-specific problem is that the threshold is both arbitrary and unaccompanied by any criterion that reliably separates cases from the neighbouring disorder.

Separability from depression

This is the oldest form of the dispute and the least likely to be settled by more of the same evidence.

Against separability. Bivariate twin modelling in 37,296 Swedish twins put the GAD–MDD genetic correlation at +1.00 (women) and +0.74 (men); individual-specific environment was also shared (r=+0.59 men, +0.36 women); neuroticism-indexed genetic factors contributed only ~25% of the GAD–MDD genetic covariance (Kendler 2007, PMID 17121688). The earlier Virginia study reached the same conclusion in 1,033 female twin pairs and framed it memorably: same genes, partly different environments (Kendler 1992, PMID 1514877). Multivariate modelling of six disorders found the second genetic factor loading on major depression and GAD together, separate from the phobia/panic/bulimia factor (Kendler 1995, PMID 7726718). Molecular work is consistent: a GWAS meta-analysis of 122,341 anxiety cases found substantial genetic correlation with depression, neuroticism and other internalizing phenotypes (Strom 2026, PMID 41634414), and a dedicated fear-vs-GAD GWAS comparison found the fear–GAD genetic correlation differed from unity only under restrictive assumptions (rg=0.87, p=9.32×10⁻³), concluding that the stronger signal is an overarching internalizing liability (Ter Kuile 2026, PMID 42374129).

For separability. GAD's own review of the validator classes concluded that although GAD and MDD overlap substantially in genetics, childhood environment, demographics and personality, this is also true of other anxiety disorders — so the overlap gives "little more reason" to question GAD's nosological validity than panic disorder's (Hettema 2008, PMID 18412057). Pure GAD is not a rarity: 3.8% of 20,000 German primary-care patients had pure GAD versus 1.6% comorbid GAD/MDE, a result the authors explicitly framed as inconsistent with the claim that GAD is usually comorbid (Wittchen 2002, PMID 12044105). Role impairment in pure GAD is similar in magnitude to pure MDD in large representative samples (Hoffman 2008, PMID 17146763). And the GAD-7's development paper found by factor analysis that GAD and depression symptoms load as distinct dimensions with differing, independent effects on functional impairment and disability days (Spitzer 2006, PMID 16717171).

The reconciling position is that the disagreement is about levels of analysis, not facts: shared genetic liability at the etiological level is compatible with separable syndromes at the phenotypic and prognostic level. The tripartite model formalised this in 1991 — general distress shared, physiological hyperarousal specific to anxiety, anhedonia specific to depression (Clark 1991, PMID 1918611). Every dimensional system since has been a variation on it.

The comorbid state as its own object

ICD-11 kept a mixed category (mixed depressive and anxiety disorder) that DSM-5 declined to include because proposed criteria were insufficiently reliable (Möller 2016, PMID 27002521). A general-population factor-mixture test of the ICD-11 criteria found something the criteria did not predict: the mixed class is not subsyndromal to either disorder but a comorbid class with higher symptom levels on both, and it was the largest symptomatic group, with high functional impairment and somatisation (Shevlin 2022, PMID 34273110). WMH data agree that comorbidity is the norm at lifetime scale — 81.9% lifetime comorbidity in DSM-5 GAD, 63.0% with mood disorders (Ruscio 2017, PMID 28297020).

This condition treats comorbid depression as a cross-linked neighbouring state rather than absorbing it: see comorbidity and primary care, and the separately curated depression condition. The point that belongs here is nosological — if the largest, most impaired symptomatic group in the population is the mixed one, then a classification that has no place for it is describing the tails of a distribution and calling them the disorders.

The network view: which symptoms bridge GAD and depression

Network psychometrics asks a different question from taxometrics: not is there a boundary but which symptoms hold the two syndromes together.

Source Scope Finding
Cai 2024 (PMID 38238548) Systematic review and statistical evaluation of 33 network-analysis studies, 78,721 participants; 23 cross-sectional networks based on PHQ and GAD-7 items examined for centrality, 12 for bridge centrality Most central symptoms: sad mood, uncontrollable worry, worrying too much. Most frequent bridge symptoms: sad mood, restlessness, motor disturbance. Most frequent robust edge: sleep–fatigue. Substantial variability between samples and network features
Dobson 2021 (PMID 33221716) 52 adolescents with GAD who had not yet developed major depression; community analysis plus cortical thickness in 39 Ten symptoms clustered as anxious, five as depressive, and five bridged them: impaired schoolwork, excessive weeping, low self-esteem, disturbed appetite and physical symptoms of depression. Greater depressive-cluster burden was associated with altered prefrontal and parietal cortical thickness and with weaker cortical-thickness–age relationships

Two implications for the boundary. First, the symptoms that bridge GAD and depression in the largest synthesis — sad mood, restlessness, motor disturbance, and the sleep–fatigue edge — are precisely the associated-symptom criteria that Faravelli found were more prevalent in depression than in GAD (Faravelli 2012, PMID 22578985). The criteria that fail to discriminate are the same ones that structurally connect the two syndromes. Second, "uncontrollable worry" and "worrying too much" are among the most central symptoms and are not among the bridge symptoms (Cai 2024, PMID 38238548) — consistent with uncontrollability being the criterion with demonstrated incremental validity (Hallion 2013, PMID 23713499) and with the metacognitive finding that beliefs about the uncontrollability and danger of thoughts are most prevalent in GAD specifically (Sun 2017, PMID 28763680).

The convergence is worth stating plainly: four methodologically independent literatures — criterion incremental validity, symptom specificity, network bridge analysis and metacognition — all point to uncontrollability of worry as GAD's most defensible discriminating feature, and to the somatic/arousal criteria as the shared ones. No revision proposal has been built on that convergence.

Instrumenting the dimensional alternative

If GAD were to be replaced by dimensions, the measures would need to exist. The HiTOP internalizing self-report scales were built from a 430-item pool covering 57 target constructs, reduced by item-level factor analysis to 39 scales / 213 items, replicated across a development sample (N=1,870) and an independent validation sample (N=496). Structural analysis revealed a strong general factor and, across the 35 non-sexual-dysfunction scales, a replicable four-factor solution labelled Distress, Fear, Body Dysmorphia and Mania (Watson 2022, PMID 33794667). GAD's constructs fall in Distress, alongside depression — the same placement the twin and molecular genetics give it (Kendler 1995, PMID 7726718; Ter Kuile 2026, PMID 42374129).

So the dimensional alternative is no longer only a critique: it has an instrument, a factor structure and a place for GAD's content. What it does not yet have is evidence that planning treatment by dimension rather than diagnosis improves outcomes (Cicero 2024, PMID 39443056).

Alternative frameworks

Framework What it does with GAD Status
HiTOP Dissolves GAD into the internalizing spectrum, with distress/fear subfactors and lower-level symptom components; treatment is planned by symptom severity rather than by diagnosis (Cicero 2024, PMID 39443056) Assessment instrument (HiTOP-SR) exists; clinical-utility evidence is early
HiTOP for genetics Argues genes operate at different hierarchy levels, so diagnosis-level phenotypes are underpowered by construction (Waszczuk 2020, PMID 31804095) Consistent with the pleiotropy seen in the 58-locus anxiety GWAS (Strom 2026, PMID 41634414)
RDoC Reframes generalized anxiety as anxious apprehension within negative valence systems / potential threat, decoupled from the DSM category (Crocq 2017, PMID 28867935) Research framework; not a clinical nosology
Dimensional severity indicators "Percent of the day worried" had moderate test–retest reliability where the DSM-IV criteria themselves had low reliability (Niles 2012, PMID 22245699) Never adopted as the primary criterion

The pooling hazard

This is the practical corollary of an unsettled boundary, and it contaminates the treatment literature more than the nosology literature.

  • Pooled anxiety-disorder effects are not GAD effects. The reference placebo-controlled CBT meta-analysis covers acute stress disorder, GAD, OCD, panic, PTSD and social anxiety together, reporting an overall Hedges' g=0.56 on target-disorder symptoms — but with large effects for OCD, GAD and acute stress disorder and only small-to-moderate effects for PTSD, social anxiety and panic (Carpenter 2018, PMID 29451967). Quoting 0.56 as "the CBT effect in GAD" would understate it; quoting the GAD stratum as "the CBT effect in anxiety" would overstate it. Both errors appear in the secondary literature.
  • Mixed-anxiety trials are not GAD trials. Screening-review evidence pools "anxiety" outcomes across disorders when computing effects such as SMD −0.41 for psychological interventions in primary care (O'Connor 2023, PMID 37338868).
  • Comorbid-enriched trials are not pure-GAD trials. The largest GAD network meta-analysis of psychological and pharmacological interventions explicitly "allowed for all comorbidities" across its 91 trials and 14,812 participants (Chen 2019, PMID 31494377).
  • PTSD is a neighbour, not a subset. Where a source pools GAD with PTSD, this knowledge base says so and takes any GAD figure from that source's own stratum.

Every treatment page in this condition states, for each headline effect size, whether it comes from a GAD-only source or a pooled one.

What would settle it

  1. Re-derive the evidence base under the alternative definition. Prevalence, comorbidity, course and treatment response, recomputed with excessiveness removed. The first two have been done (Ruscio 2024, PMID 39364896); the last two have not [unverified as of 2026-09-02].
  2. A treatment-response discriminator. If GAD and MDD respond differentially to an agent at equal baseline severity, that is validator evidence no twin study can provide. Existing network meta-analyses cannot answer it because they enrol comorbid samples (Chen 2019, PMID 31494377).
  3. A prospective test of the mixed class. Does the comorbid class identified by Shevlin 2022 (PMID 34273110) have a distinct course, or is it just severity?

Open questions

  • Does removing the excessiveness requirement change treatment response, not just prevalence? Nobody has re-analysed a trial dataset under the alternative criterion (Ruscio 2024, PMID 39364896).
  • If worry is dimensional (Marcus 2014, PMID 24377439; Kertz 2014, PMID 24334160), what is the defensible basis for any threshold — impairment, cost-effectiveness of treatment, or predictive validity for secondary disorders?
  • Is muscle tension really the only GAD-specific symptom (Faravelli 2012, PMID 22578985), and if so should it be weighted rather than counted equally with five non-specific ones?
  • Should a criterion revision be built on the four-way convergence around uncontrollability (Hallion 2013, PMID 23713499; Faravelli 2012, PMID 22578985; Cai 2024, PMID 38238548; Sun 2017, PMID 28763680)? No proposal has assembled these literatures.
  • Why do genetic correlations between GAD and MDD approach unity while phenotypic impairment profiles remain separable (Kendler 2007, PMID 17121688; Hoffman 2008, PMID 17146763)?

References

  1. Marcus DK, et al. The latent structure of generalized anxiety disorder in midlife adults. Psychiatry Res. 2014;215:366-71. PMID 24377439
  2. Kertz SJ, et al. Examining the latent structure of worry and generalized anxiety in a clinical sample. J Anxiety Disord. 2014;28:8-15. PMID 24334160
  3. Faravelli C, et al. Generalized anxiety disorder: is there any specific symptom? Compr Psychiatry. 2012;53:1056-62. PMID 22578985
  4. Kendler KS, et al. The sources of co-morbidity between major depression and generalized anxiety disorder in a Swedish national twin sample. Psychol Med. 2007;37:453-62. PMID 17121688
  5. Kendler KS, et al. Major depression and generalized anxiety disorder. Same genes, (partly) different environments? Arch Gen Psychiatry. 1992;49:716-22. PMID 1514877
  6. Kendler KS, et al. The structure of the genetic and environmental risk factors for six major psychiatric disorders in women. Arch Gen Psychiatry. 1995;52:374-83. PMID 7726718
  7. Ruscio AM, et al. The case for eliminating excessive worry as a requirement for generalized anxiety disorder: a cross-national investigation. Psychol Med. 2024;54:3447-3458. PMID 39364896
  8. Ruscio AM, et al. Should excessive worry be required for a diagnosis of generalized anxiety disorder? Psychol Med. 2005;35:1761-72. PMID 16300690
  9. Ruscio AM, et al. Broadening the definition of generalized anxiety disorder. J Anxiety Disord. 2007;21:662-76. PMID 17118626
  10. Hallion LS, Ruscio AM. Should uncontrollable worry be removed from the definition of GAD? J Abnorm Psychol. 2013;122:369-75. PMID 23713499
  11. Rutter LA, Brown TA. Reliability and validity of the dimensional features of generalized anxiety disorder. J Anxiety Disord. 2015;29:1-6. PMID 25465882
  12. Wittchen HU, et al. Reliability and clinical validity of UM-CIDI DSM-III-R generalized anxiety disorder. J Psychiatr Res. 1995;29:95-110. PMID 7666382
  13. Lee S, et al. Implications of modifying the duration requirement of generalized anxiety disorder in developed and developing countries. Psychol Med. 2009;39:1163-76. PMID 19091158
  14. Angst J, et al. Varying temporal criteria for generalized anxiety disorder. Psychol Med. 2006;36:1283-92. PMID 16734945
  15. Andrews G, et al. Generalized worry disorder: a review of DSM-IV generalized anxiety disorder and options for DSM-V. Depress Anxiety. 2010;27:134-47. PMID 20058241
  16. Starcevic V, et al. The concept of generalized anxiety disorder: between the too narrow and too wide diagnostic criteria. Psychopathology. 1999;32:5-11. PMID 9885394
  17. Domschke K, et al. Taxonomy of anxiety disorders — a comparison of ICD-10 and ICD-11. Nervenarzt. 2025;96:1-5. PMID 40728738
  18. Hettema JM. The nosologic relationship between generalized anxiety disorder and major depression. Depress Anxiety. 2008;25:300-16. PMID 18412057
  19. Wittchen HU, et al. Generalized anxiety and depression in primary care: prevalence, recognition, and management. J Clin Psychiatry. 2002;63 Suppl 8:24-34. PMID 12044105
  20. Hoffman DL, et al. Human and economic burden of generalized anxiety disorder. Depress Anxiety. 2008;25:72-90. PMID 17146763
  21. Spitzer RL, et al. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med. 2006;166:1092-7. PMID 16717171
  22. Clark LA, Watson D. Tripartite model of anxiety and depression: psychometric evidence and taxonomic implications. J Abnorm Psychol. 1991;100:316-36. PMID 1918611
  23. Möller HJ, et al. The relevance of 'mixed anxiety and depression' as a diagnostic category in clinical practice. Eur Arch Psychiatry Clin Neurosci. 2016;266:725-736. PMID 27002521
  24. Shevlin M, et al. ICD-11 'mixed depressive and anxiety disorder' is clinical rather than sub-clinical and more common than anxiety and depression in the general population. Br J Clin Psychol. 2022;61:18-36. PMID 34273110
  25. Ruscio AM, et al. Cross-sectional Comparison of the Epidemiology of DSM-5 Generalized Anxiety Disorder Across the Globe. JAMA Psychiatry. 2017;74:465-475. PMID 28297020
  26. Cicero DC, et al. State of the Science: The Hierarchical Taxonomy of Psychopathology (HiTOP). Behav Ther. 2024;55:1114-1129. PMID 39443056
  27. Waszczuk MA, et al. Redefining phenotypes to advance psychiatric genetics: Implications from hierarchical taxonomy of psychopathology. J Abnorm Psychol. 2020;129:143-161. PMID 31804095
  28. Crocq MA. The history of generalized anxiety disorder as a diagnostic category. Dialogues Clin Neurosci. 2017;19:107-116. PMID 28867935
  29. Niles AN, et al. Dimensional indicators of generalized anxiety disorder severity for DSM-V. J Anxiety Disord. 2012;26:279-86. PMID 22245699
  30. Strom NI, et al. Genome-wide association study of major anxiety disorders in 122,341 European-ancestry cases identifies 58 loci and highlights GABAergic signaling. Nat Genet. 2026;58:275-288. PMID 41634414
  31. Ter Kuile AR, et al. Genome-wide genetic overlap between fear-based disorders and generalised anxiety disorder. Mol Psychiatry. 2026. PMID 42374129
  32. Carpenter JK, et al. Cognitive behavioral therapy for anxiety and related disorders: A meta-analysis of randomized placebo-controlled trials. Depress Anxiety. 2018;35:502-514. PMID 29451967
  33. Chen TR, et al. Pharmacological and psychological interventions for generalized anxiety disorder in adults: A network meta-analysis. J Psychiatr Res. 2019;118:73-83. PMID 31494377
  34. O'Connor EA, et al. Anxiety Screening: Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA. 2023;329:2171-2184. PMID 37338868
  35. Cai H, et al. A network model of depressive and anxiety symptoms: a statistical evaluation. Mol Psychiatry. 2024;29:767-781. PMID 38238548
  36. Dobson ET, et al. Bridging Anxiety and Depression: A Network Approach in Anxious Adolescents. J Affect Disord. 2021;280:305-314. PMID 33221716
  37. Watson D, et al. The Development of Preliminary HiTOP Internalizing Spectrum Scales. Assessment. 2022;29:17-33. PMID 33794667
  38. Sun X, et al. Dysfunctional metacognition across psychopathologies: A meta-analytic review. Eur Psychiatry. 2017;45:139-153. PMID 28763680