Skip to content

Diagnosis and classification

TL;DR — GAD is defined by excessive, difficult-to-control worry lasting ≥6 months with associated arousal symptoms (restlessness, fatigue, poor concentration, irritability, muscle tension, sleep disturbance) (DeMartini 2019, PMID 30934083). Almost every element of that sentence has been challenged in the peer-reviewed literature by the people who built the criteria: the 6-month duration threshold does not separate clinically distinct groups (Lee 2009, PMID 19091158; Angst 2006, PMID 16734945), the excessiveness requirement excludes cases that resemble diagnosed cases on nearly every validator (Ruscio 2005, PMID 16300690; Ruscio 2024, PMID 39364896), and reliability has been the diagnosis's structural weakness since 1983 — the original ADIS field study found good agreement for every anxiety category except GAD (Di Nardo 1983, PMID 6625856), and in the NCS reappraisal lifetime GAD reached κ=0.53, rising to κ=0.78 when the excessiveness criterion was dropped (Wittchen 1995, PMID 7666382). ICD-11 and DSM-5 now converge structurally but not operationally: ICD-11 has no fixed symptom count and no excessiveness requirement, and in the WHO global field study clinicians diagnosed GAD more accurately with ICD-11 guidelines than with ICD-10 (Rebello 2019, PMID 32018071). The largest unresolved question is not how to apply the criteria but whether the current criteria carve the right group — see the diagnostic boundary.

The DSM-5 criterion set, and where each element came from

Element DSM-5 requirement Origin Best evidence against it
Anxiety and worry Excessive, about a number of events or activities DSM-III-R shifted the core from somatic anxiety to worry (Crocq 2017, PMID 28867935) Removing "excessive" raises global lifetime prevalence from 2.6% to 4.0%; non-excessive cases match excessive cases on family history, comorbidity, suicidality (Ruscio 2024, PMID 39364896)
Frequency More days than not DSM-III-R/DSM-IV A targeted PubMed search on 2026-09-02 located no large criterion-level validation isolating frequency from duration and excessiveness
Duration ≥6 months Raised from 1 month in DSM-III-R (Crocq 2017, PMID 28867935) Lifetime prevalence at 1/3/6/12-month thresholds: 7.5%/5.2%/4.1%/3.0% in developed countries; no difference in onset, severity, persistence, comorbidity or impairment between 6-month and shorter GAD (Lee 2009, PMID 19091158). In the Zurich cohort the 6-month rule would deny the diagnosis to about half of those actually treated for generalized anxiety (Angst 2006, PMID 16734945)
Control Worry is difficult to control DSM-IV The one criterion with demonstrated incremental validity: after controlling for excessiveness, uncontrollability still predicted GAD severity, clinician-rated anxiety, comorbidity count, and medication/psychotherapy use; excessiveness predicted nothing over uncontrollability (Hallion 2013, PMID 23713499)
Associated symptoms ≥3 of 6 (restlessness, easy fatigue, concentration difficulty, irritability, muscle tension, sleep disturbance) DSM-IV cut the DSM-III-R list of 18 Relaxing the 3-symptom rule together with duration and excessiveness more than doubles estimated prevalence, and the broadened category still predicts first onset of secondary disorders (Ruscio 2007, PMID 17118626)
Clinical significance Distress or functional impairment DSM-IV Removing it improves DSM-IV/ICD-10 concordance (Andrews 2002, PMID 12145492)
Exclusions Not substance/medical; not better explained by another mental disorder DSM-III onward The exclusion rules are what make GAD partly a residual diagnosis; concordance modelling shows the two systems can only be aligned if all diagnoses are considered together (Andrews 2002, PMID 12145492)

The direction of travel across editions has been consistently toward the cognitive: DSM-III's anxiety neurosis was split into GAD and panic disorder in 1980 partly on differential imipramine response, and from DSM-III-R onward worry — not autonomic arousal — became the defining feature (Crocq 2017, PMID 28867935). One consequence is that the DSM-5 workgroup literature seriously discussed relabelling the condition generalized worry disorder, with a 3-month rather than 6-month duration and worry reframed as a cognitive avoidance strategy (Andrews 2010, PMID 20058241). That relabelling did not happen.

DSM-5 versus ICD-11

DSM-5 moved OCD, PTSD and acute stress disorder out of the anxiety disorders and moved separation anxiety disorder and selective mutism in, producing a more symmetrical adult/child structure; agoraphobia became independent of panic disorder (Park 2020, PMID 32002930). ICD-11 made structurally parallel changes: "anxiety or fear-related disorders" became a free-standing grouping, agoraphobia and panic disorder became separately and comorbidly codable, separation anxiety and selective mutism moved in under a lifespan logic, and mixed anxiety-depressive disorder was moved out to the mood grouping as "mixed depressive and anxiety disorder" (Domschke 2025, PMID 40728738).

Feature DSM-5 ICD-11
Core requirement Excessive worry, difficult to control Apprehension/worry or general apprehensiveness, not required to be "excessive" (Domschke 2025, PMID 40728738)
Symptom count Fixed: ≥3 of 6 No precise minimum number specified (Domschke 2025, PMID 40728738)
Duration ≥6 months Several months
Mixed anxiety–depression Not a diagnosis Present, but reclassified into the mood grouping (Domschke 2025, PMID 40728738)
Field-tested clinician accuracy — GAD is one of three diagnoses clinicians diagnosed significantly more accurately under ICD-11 than ICD-10 in a randomized vignette study of 1,840 clinicians in six languages (Rebello 2019, PMID 32018071)

Empirically the two systems agree well when applied to the same patients by the same instrument: κ=0.86 between DSM-IV and ICD-10-DCR GAD in a consecutive panic/agoraphobia clinic sample (Starcevic 1999, PMID 9885394). They disagree about who is a case in the population, because the identifying criteria, the clinical-significance rule and the exclusion hierarchies differ (Andrews 2002, PMID 12145492). Self-report instruments aligned to ICD-11 rather than DSM now exist — the International Anxiety Questionnaire, validated in a UK-representative sample of 2,058, yielded 7.1% meeting the ICD-11 GAD algorithm (Shevlin 2023, PMID 36215152), several-fold above WMH interview-based DSM-5 12-month estimates of 1.8% (Ruscio 2017, PMID 28297020). Algorithm, instrument and system each move the number.

Reliability: the diagnosis's oldest structural problem

Study Design GAD reliability Note
Di Nardo 1983 (PMID 6625856) 60 outpatients, two ADIS interviewers, DSM-III Good agreement for agoraphobia, panic, social phobia and OCD; not for GAD The problem is as old as the diagnosis
Wittchen 1995 (PMID 7666382) NCS clinical reappraisal, UM-CIDI vs SCID, DSM-III-R Test–retest κ=0.53 lifetime GAD; κ=0.78 with the excessiveness criterion removed; CIDI-vs-SCID concordance κ=0.35 (0.66 without excessiveness) Discordance traced specifically to criteria A2 (excessive/unrealistic) and D (symptoms "often")
Gordon 2011 (PMID 21596519) 129 GAD patients, ADIS-IV feature ratings Inter-rater reliability of individual GAD features ranged good→poor; internal consistency moderate→low; poor discriminant validity against social anxiety measures Feature-level, not just diagnosis-level, instability
Regier 2013 (PMID 23111466) DSM-5 field trials, 11 North American academic centres, independent clinicians Across 15 adult diagnoses: 5 very good (κ 0.60–0.79), 9 good (0.40–0.59), 6 questionable (0.20–0.39), 3 unacceptable (<0.20) The PubMed abstract does not report the GAD-specific κ; no disorder-specific value is inferred here
Chmielewski 2015 (PMID 26098046) 339 patients, SCID-I/P; audio-recording vs independent re-interview Audio method mean κ=0.80; test–retest method mean κ=0.47, in the same clinic Much of the "DSM-5 is less reliable" story is a method artifact; self-reported symptoms were highly stable even when diagnoses flipped

The Chmielewski result matters for how the whole GAD reliability literature should be read: apparent worsening between DSM-IV-era and DSM-5-era estimates is largely a change in measurement method, not a change in the criteria. But it does not rescue GAD, because the GAD-specific weakness — disagreement about whether worry is "excessive" — is visible in every era and reverses when that criterion is removed (Wittchen 1995, PMID 7666382).

How the diagnosis is actually made

Almost nothing in the GAD literature rests on unstructured clinical judgement; the number that a study reports is a property of its instrument.

Instrument Type Used for GAD-relevant performance
ADIS / ADIS-IV-L Semi-structured clinician interview, anxiety-specialised Clinic samples, most CBT trials Reliability good→excellent for most DSM-IV categories, but GAD–MDD is named as a specific boundary problem, and disagreements concentrate on whether symptoms are sufficient in number, severity or duration (Brown 2001, PMID 11261399)
CIDI (UM-CIDI, CIDI 3.0) Fully structured lay-administered Every population survey cited on this condition — NCS-R, WMH, Singapore, Zurich Lifetime GAD κ=0.53; concordance with SCID κ=0.35, both driven by the excessiveness item (Wittchen 1995, PMID 7666382)
SCID Semi-structured clinician interview Reference standard in accuracy studies Same-clinic reliability is method-dependent: κ=0.80 by audio re-rating vs κ=0.47 by independent re-interview (Chmielewski 2015, PMID 26098046)
MINI Brief structured interview National surveys (e.g. India's NMHS) and trial screening Used as the diagnostic standard in the Indian National Mental Health Survey, which produced a current GAD prevalence of 0.57% (Jayasankar 2023, PMID 38298878)
GAD-7 / GAD-2 Self-report screen — not a diagnostic instrument Primary care, research triage Sensitivity 0.64 (95% CI 0.56–0.72), specificity 0.91 (0.87–0.93) at ≥10 for GAD across 35 studies (Aktürk 2025, PMID 40130828)

Two implications follow. First, the population estimates and the clinic estimates are not measuring the same construct with the same error: a fully structured lay interview and a semi-structured clinician interview disagree at κ≈0.35 on lifetime GAD (Wittchen 1995, PMID 7666382). Second, the near-universal use of GAD-7 as a proxy for "GAD" in service data and digital-intervention trials imports a screen with 64% sensitivity into places where a diagnosis is assumed (screening and measurement).

Instability is not confined to adults: in a 311-child sample assessed with a DSM-5 structured interview, inter-rater reliability was in the questionable range specifically for GAD and MDD while other diagnoses ranged good→very good (Tolin 2023, PMID 36632642) — the same two diagnoses that Brown 2001 flagged as a boundary problem two decades earlier (special populations).

The residual-diagnosis problem

GAD carries exclusion criteria that other anxiety disorders do not carry to the same degree: worry occurring exclusively within another disorder's content (fear of panic attacks, fear of scrutiny, trauma-related fear) is assigned elsewhere. Three measurable consequences:

  • Prevalence is hierarchy-dependent. Modelling DSM-IV against ICD-10 shows that concordance cannot be fixed diagnosis-by-diagnosis, because each system is a set of interdependent diagnoses with different exclusion rules (Andrews 2002, PMID 12145492).
  • "Pure" GAD is not rare. The counter-claim that GAD is essentially always comorbid does not hold in primary care: in a 20,000-patient German study, 3.8% had pure GAD, 4.4% pure MDE and only 1.6% comorbid GAD/MDE — explicitly described by the authors as inconsistent with the "GAD is usually comorbid" claim (Wittchen 2002, PMID 12044105). In the WMH surveys, by contrast, lifetime comorbidity reached 81.9% (Ruscio 2017, PMID 28297020). Lifetime versus current comorbidity, and clinic versus community sampling, explain most of that gap — a distinction routinely lost when the "GAD is never pure" claim is repeated.
  • Somatic presentation hides it. GAD is described as the most prevalent anxiety disorder in primary care, at ~8% of attenders, with patients presenting somatic complaints rather than worry (Wittchen 2002, PMID 12497648) — see comorbidity and primary care.

Differential diagnosis

GAD is diagnosed partly by exclusion, so the differential is the diagnosis.

Competing explanation Discriminating feature Source
Major depressive disorder Worry is future-oriented and anticipatory rather than ruminative about past loss; but genetic risk factors are shared essentially completely in women (rg ≈ +1.00) Hettema 2008 (PMID 18412057); Kendler 2007 (PMID 17121688)
Panic disorder Paroxysmal autonomic surges vs continuous apprehension; the PHQ panic items outperform GAD instruments for panic (LR+ 78) Herr 2014 (PMID 25058220)
Social anxiety disorder GAD feature ratings show poor discriminant validity against social anxiety measures — this is a documented failure point, not a theoretical one Gordon 2011 (PMID 21596519)
PTSD Trauma-anchored intrusion and avoidance; a neighbouring condition curated separately, not a subset see the PTSD condition
Substance/medication effect and medical mimics Excluded by criterion; see red flags and safety concerns —
Subthreshold GAD Not a DSM diagnosis, but roughly twice as prevalent as full GAD, persistent, and associated with greater impairment, benzodiazepine use and primary-care use than in non-anxious individuals Haller 2014 (PMID 24886240)

Subthreshold GAD is the single most important entry in that table, because it is where the criteria's arbitrariness becomes a clinical population. In Canadian national data, 2.6% met threshold GAD and a further 2.3% subthreshold GAD, and both were independently associated with 12-month suicidal ideation after adjustment for sociodemographics and psychiatric comorbidity (Gilmour 2016, PMID 27849314).

What this means for reading the rest of this condition

Three practical consequences run through every other page here:

  1. Prevalence figures are criterion artefacts. Any GAD prevalence number is uninterpretable without its diagnostic system, instrument and duration rule (epidemiology and burden).
  2. Trial populations are DSM-defined. Essentially the entire treatment evidence base recruits DSM-III-R/IV/5 GAD, so it says nothing about the non-excessive, short-duration or subthreshold groups that the criteria exclude — a group at least as large as the diagnosed one (Haller 2014, PMID 24886240; Ruscio 2024, PMID 39364896).
  3. Screening instruments and diagnostic criteria are different objects. The GAD-7 was constructed against DSM-IV GAD (Spitzer 2006, PMID 16717171) and is now widely used as if it were a severity measure for an ICD-11 world (screening and measurement).

Open questions

  • Which κ did GAD actually achieve in the DSM-5 field trials, and how does it compare with the ICD-11 field study's accuracy gain? The two literatures have never been placed side by side (Regier 2013, PMID 23111466; Rebello 2019, PMID 32018071).
  • Does the ICD-11 definition, without an excessiveness requirement and without a fixed symptom count, identify the same people as DSM-5 in a general population sample? Only clinic-sample concordance exists (Starcevic 1999, PMID 9885394), and the one ICD-11-aligned population instrument gives a much higher rate (Shevlin 2023, PMID 36215152).
  • If uncontrollability outperforms excessiveness on incremental validity (Hallion 2013, PMID 23713499), why has no revision proposed keeping the former and dropping the latter as a package?
  • What is the treatment response of criteria-excluded cases—non-excessive worry, duration <6 months, or two associated symptoms? A targeted PubMed search rerun on 2026-09-02 found epidemiological reclassification studies but no treatment trial prospectively stratified by those exclusions; this is a dated evidence gap.

References

  1. Crocq MA. The history of generalized anxiety disorder as a diagnostic category. Dialogues Clin Neurosci. 2017;19:107-116. PMID 28867935
  2. 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
  3. DeMartini J, et al. Generalized Anxiety Disorder. Ann Intern Med. 2019;170:ITC49-ITC64. PMID 30934083
  4. Park SC, et al. Anxiety Disorders in the DSM-5: Changes, Controversies, and Future Directions. Adv Exp Med Biol. 2020;1191:187-196. PMID 32002930
  5. Domschke K, et al. Taxonomy of anxiety disorders — a comparison of ICD-10 and ICD-11. Nervenarzt. 2025;96:1-5. PMID 40728738
  6. Rebello TJ, et al. Anxiety and Fear-Related Disorders in the ICD-11: Results from a Global Case-controlled Field Study. Arch Med Res. 2019;50:490-501. PMID 32018071
  7. Shevlin M, et al. The development and initial validation of self-report measures of ICD-11 depressive episode and generalized anxiety disorder. J Clin Psychol. 2023;79:854-870. PMID 36215152
  8. Andrews G, et al. The classification of anxiety disorders in ICD-10 and DSM-IV: a concordance analysis. Psychopathology. 2002;35:100-6. PMID 12145492
  9. 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
  10. 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
  11. Di Nardo PA, et al. Reliability of DSM-III anxiety disorder categories using a new structured interview. Arch Gen Psychiatry. 1983;40:1070-4. PMID 6625856
  12. Gordon D, et al. Reliability and validity of DSM-IV generalized anxiety disorder features. J Anxiety Disord. 2011;25:813-21. PMID 21596519
  13. Chmielewski M, et al. Method matters: Understanding diagnostic reliability in DSM-IV and DSM-5. J Abnorm Psychol. 2015;124:764-9. PMID 26098046
  14. Regier DA, et al. DSM-5 field trials in the United States and Canada, Part II: test-retest reliability of selected categorical diagnoses. Am J Psychiatry. 2013;170:59-70. PMID 23111466
  15. Ruscio AM, et al. Should excessive worry be required for a diagnosis of generalized anxiety disorder? Results from the US National Comorbidity Survey Replication. Psychol Med. 2005;35:1761-72. PMID 16300690
  16. Ruscio AM, et al. Broadening the definition of generalized anxiety disorder: effects on prevalence and associations with other disorders in the National Comorbidity Survey Replication. J Anxiety Disord. 2007;21:662-76. PMID 17118626
  17. Hallion LS, Ruscio AM. Should uncontrollable worry be removed from the definition of GAD? A test of incremental validity. J Abnorm Psychol. 2013;122:369-75. PMID 23713499
  18. 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
  19. 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
  20. Angst J, et al. Varying temporal criteria for generalized anxiety disorder: prevalence and clinical characteristics in a young age cohort. Psychol Med. 2006;36:1283-92. PMID 16734945
  21. Haller H, et al. The prevalence and burden of subthreshold generalized anxiety disorder: a systematic review. BMC Psychiatry. 2014;14:128. PMID 24886240
  22. 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
  23. Herr NR, et al. Does this patient have generalized anxiety or panic disorder? The Rational Clinical Examination systematic review. JAMA. 2014;312:78-84. PMID 25058220
  24. Hettema JM. The nosologic relationship between generalized anxiety disorder and major depression. Depress Anxiety. 2008;25:300-16. PMID 18412057
  25. 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
  26. Gilmour H. Threshold and subthreshold Generalized Anxiety Disorder (GAD) and suicide ideation. Health Rep. 2016;27:13-21. PMID 27849314
  27. Spitzer RL, et al. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med. 2006;166:1092-7. PMID 16717171
  28. Brown TA, et al. Reliability of DSM-IV anxiety and mood disorders: implications for the classification of emotional disorders. J Abnorm Psychol. 2001;110:49-58. PMID 11261399
  29. Tolin DF, et al. The DIAMOND-KID: Psychometric Properties of a Structured Diagnostic Interview for DSM-5 Anxiety, Mood, and Obsessive-Compulsive and Related Disorders in Children and Adolescents. Assessment. 2023;30:2351-2363. PMID 36632642
  30. Jayasankar P, et al. Current prevalence and determinants of generalized anxiety disorder from a nationally representative, population-based survey of India. Indian J Psychiatry. 2023;65:1244-1248. PMID 38298878
  31. 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
  32. Wittchen HU. Generalized anxiety disorder: prevalence, burden, and cost to society. Depress Anxiety. 2002;16:162-71. PMID 12497648
  33. Aktürk Z, et al. Generalized Anxiety Disorder 7-item (GAD-7) and 2-item (GAD-2) scales for detecting anxiety disorders in adults. Cochrane Database Syst Rev. 2025;3:CD015455. PMID 40130828