Introduction
The diagnostic threshold for a disorder serves a gatekeeping function (Halpin, Reference Halpin2016). Individuals who meet this threshold are eligible for health care, disability income, and other resources that may be unavailable to individuals who fall short of the threshold. In resource-limited systems, categorical decisions are unavoidable, making diagnosis necessary even for forms of psychopathology that are dimensional in nature (Haslam, McGrath, Viechtbauer, & Kuppens, Reference Haslam, McGrath, Viechtbauer and Kuppens2020; Ruscio, Reference Ruscio2019). Given the personal and public health impact of these decisions, where we draw the threshold matters.
Unfortunately, it is not always clear where diagnostic thresholds should be drawn. A compelling example is offered by generalized anxiety disorder (GAD). GAD is centrally defined by worry, yet worry is experienced to some degree by most people. To impose a threshold, the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5; American Psychiatric Association [APA], 2013) requires the worry to focus on multiple topics, occur most days for six months, be excessive and uncontrollable, be accompanied by three associated symptoms of anxiety, and cause clinically significant distress or impairment.
Is this threshold optimal for separating normal from pathological worry? There are indications that it is not. Considerable evidence exists that the GAD diagnosis misses individuals with clinically significant anxiety (Andrews et al., Reference Andrews, Hobbs, Borkovec, Beesdo, Craske, Heimberg and Stanley2010). Subthreshold GAD, variably defined, is associated with a disease burden comparable to the full disorder (Haller, Cramer, Lauche, Gass, & Dobos, Reference Haller, Cramer, Lauche, Gass and Dobos2014). Subthreshold GAD is also common in clinical settings (Olivares et al., Reference Olivares, Carrasco, Álvarez, Perez, López-Gómez and Rejas2013), with one study reporting that 3 in 4 adults seeking treatment for clinically relevant worry fell short of a GAD diagnosis by a single diagnostic criterion (Lawrence & Brown, Reference Lawrence and Brown2009).
A particularly problematic criterion is the requirement that worry must be ‘excessive.’ Like the requirement of ‘unrealistic’ worry, which was discarded in DSM-IV, there is no accepted standard for what constitutes excessive worry (Andrews et al., Reference Andrews, Hobbs, Borkovec, Beesdo, Craske, Heimberg and Stanley2010; Starcevic, Portman, & Beck, Reference Starcevic, Portman and Beck2012). Perhaps for this reason, clinical raters often disagree whether worry is excessive, and removing this criterion improves diagnostic reliability (Wittchen, Kessler, Zhao, & Abelson, Reference Wittchen, Kessler, Zhao and Abelson1995). Importantly, excessiveness is highly correlated with uncontrollability of worry (Gordon & Heimberg, Reference Gordon and Heimberg2011; Rutter & Brown, Reference Rutter and Brown2015), suggesting that it may be unnecessary to require both. Of the two, uncontrollability has superior incremental validity (Hallion & Ruscio, Reference Hallion and Ruscio2013) and is supported by research linking cognitive control deficits with worry (e.g. Gustavson et al., Reference Gustavson, Lurquin, Michaelson, Barker, Carruth, von Bastian and Miyake2020; Hallion, Ruscio, & Jha, Reference Hallion, Ruscio and Jha2014; Zetsche, Bürkner, & Schulze, Reference Zetsche, Bürkner and Schulze2018), arguing indirectly for removing excessiveness.
Based on these considerations, Ruscio et al. (Reference Ruscio, Lane, Roy-Byrne, Stang, Stein, Wittchen and Kessler2005) investigated the impact of removing the excessiveness requirement. Analyzing an epidemiological sample of U.S. adults, they found that lifetime prevalence of DSM-IV GAD increased by approximately 40% when excessiveness was omitted. Although non-excessive GAD was somewhat milder than excessive GAD, non-excessive cases reported substantial persistence, comorbidity, role impairment, and treatment-seeking, supporting their inclusion in the GAD diagnosis. Subsequent studies examining a narrower range of outcomes obtained similar results (Beesdo-Baum et al., Reference Beesdo-Baum, Winkel, Pine, Hoyer, Höfler, Lieb and Wittchen2011; Lee, Ma, Tsang, & Kwok, Reference Lee, Ma, Tsang and Kwok2009; Ruscio et al., Reference Ruscio, Chiu, Roy-Byrne, Stang, Stein, Wittchen and Kessler2007).
These findings, although suggestive, did not lead to the removal of excessiveness in DSM-5. In fact, despite pointed criticism (Shear, Reference Shear2012), excessiveness was added as a qualifier for worry when revising the GAD definition for the 11th edition of the International Classification of Diseases (World Health Organization, 2019). There are likely at least two reasons for this. First, eliminating excessiveness might raise concerns about over-pathologizing, especially given the ubiquity of worry. To address these concerns, there is a need to demonstrate that broadening GAD to encompass non-excessive cases yields improvements in clinical utility that justify the increased prevalence and associated costs. As the primary stated purpose of DSM-5 diagnoses is to facilitate clinical care (APA, 2013), it is especially important to evaluate whether non-excessive GAD is clinically significant.
Second, the impact of eliminating excessiveness has been investigated in only a few high-income countries. Given cross-national differences in GAD prevalence (Marques, Robinaugh, LeBlanc, & Hinton, Reference Marques, Robinaugh, LeBlanc and Hinton2011; Ruscio et al., Reference Ruscio, Hallion, Lim, Aguilar-Gaxiola, Al-Hamzawi, Alonso and Scott2017), there is a need to evaluate impact in other parts of the world (Lewis-Fernández et al., Reference Lewis-Fernández, Hinton, Laria, Patterson, Hofmann, Craske and Liao2010; Stein, Scott, de Jonge, & Kessler, Reference Stein, Scott, de Jonge and Kessler2017). It is plausible that there are differing cultural standards for what is considered excessive worry. Moreover, countries vary in exposure to objectively stressful circumstances that set the stage for worry. Extreme poverty, limited health services, political instability, and other challenges faced disproportionately in lower-income countries may lead to chronic worry that is not ‘excessive’ given the context but may still be disabling and merit treatment (Kessler & Wittchen, Reference Kessler and Wittchen2002; Stein & Seedat, Reference Stein and Seedat2007).
We carried out the first cross-national study to explore the implications of removing excessiveness from the GAD diagnosis. We updated prior research on this topic by examining DSM-5 GAD, which is more severe than DSM-IV GAD because it includes individuals whose symptoms co-occur with major mood episodes (Ruscio, Hallion, Demyttenaere, Lee, & Lim, Reference Ruscio, Hallion, Demyttenaere, Lee, Lim, Stein, Scott, de Jonge and Kessler2018), and we expanded on prior research in several ways. First, to evaluate the clinical significance of non-excessive GAD, we considered an expanded set of outcomes of interest to clinicians. This included, for the first time, a systematic comparison of the content of excessive and non-excessive worry. Second, we supplemented inspection of correlates with formal models in which GAD meeting full DSM-5 criteria was compared to non-excessive GAD in predicting the subsequent onset and persistence of temporally secondary outcomes. This allowed us to explore whether the prognostic value of the GAD diagnosis might be improved by capturing non-excessive cases. Third, we utilized a much larger and more diverse sample than previous studies, including respondents from 25 countries representing all regions of the globe. This increased statistical power to detect differences between excessive and non-excessive GAD, while also providing a first look at the global implications of the excessiveness requirement.
Method
Sample
The World Mental Health (WMH) surveys are a coordinated set of community epidemiological surveys administered to probability samples of the non-institutionalized population in countries throughout the world (https://www.hcp.med.harvard.edu/wmh/). Data for the current report came from 28 WMH surveys, 12 carried out in countries classified by the World Bank as Low- or Middle-Income (LMICs; Brazil, Bulgaria, Colombia, Iraq, Lebanon, Mexico, Peru, Romania, South Africa, Ukraine) and 16 in countries classified as High-Income (HICs; Argentina, Australia, Belgium, France, Germany, Israel, Italy, Japan, Netherlands, New Zealand, Northern Ireland, Poland, Portugal, Spain, United States). There were two surveys each in Bulgaria, Colombia, and Spain. Twenty surveys were based on nationally representative household samples, whereas one was representative of a selected region (Murcia, Spain), three of selected metropolitan areas (São Paulo, Brazil; Medellin, Colombia; 11 metropolitan areas in Japan), and four of all urbanized areas in the country (Argentina, Colombia, Mexico, Peru). Response rates ranged from 45.9% (France) to 97.2% (Medellin) and averaged 71.4% across surveys (online Supplementary Table S1).
The interview schedule was developed in English and translated into other languages using a standardized WHO translation, back-translation, expert review, and harmonization protocol (Kessler & Üstün, Reference Kessler and Üstün2004). Interviews were administered face-to-face in respondents' homes after obtaining informed consent using procedures approved by local Institutional Review Boards. Interviews had two parts. Part I was administered to one, or in a few surveys two, randomly-selected adults in each sampled household. Part I contained assessments of core mental disorders, including GAD. A Part I weight adjusted for differential probabilities of selection within and between households. Part II, which included questions about other mental disorders and correlates, was then administered to 100% of Part I respondents who met lifetime criteria for any Part I disorder plus a probability subsample of other Part I respondents. A Part II weight equal to the inverse of the probability of selection into Part II was used to restore the representativeness of the Part II sample. A third weight was then applied to the Part II sample to calibrate discrepancies between sample and population distributions on the cross-classification of Census socio-demographic and geographic variables.
Measures
Interview schedule
All surveys used the WHO Composite International Diagnostic Interview (CIDI; Kessler & Üstün, Reference Kessler, Üstün, Kessler and Üstün2011), a fully-structured diagnostic interview administered by trained lay interviewers. Consistent interviewer training and quality control monitoring procedures were used across surveys (Pennell et al., Reference Pennell, Mneimneh, Bowers, Chardoul, Wells, Viana, Saiz, Kessler and Üstün2008). In addition to GAD, 16 potentially comorbid DSM-IV disorders were assessed, including anxiety (panic disorder, agoraphobia, social phobia, specific phobia, posttraumatic stress disorder, separation anxiety disorder), mood (major depressive disorder, bipolar spectrum disorder), substance-related (alcohol abuse or dependence, drug abuse or dependence), and disruptive behavior disorders (attention-deficit/hyperactivity disorder, oppositional defiant disorder, conduct disorder, intermittent explosive disorder, bulimia nervosa, binge eating disorder). Diagnoses based on the CIDI have been shown to have good concordance with diagnoses based on the clinician-administered Structured Clinical Interview for DSM-IV (SCID; First, Spitzer, Gibbon, & Williams, Reference First, Spitzer, Gibbon and Williams2002) in blinded clinical reappraisal interviews (Ghimire, Chardoul, Kessler, Axinn, & Adhikari, Reference Ghimire, Chardoul, Kessler, Axinn and Adhikari2013; Haro et al., Reference Haro, Arbabzadeh-Bouchez, Brugha, de Girolamo, Guyer, Jin, Kessler, Kessler and Üstün2008; Kessler et al., Reference Kessler, Al-Desouki, King, Sampson, Al-Subaie, Al-Habeeb and Altwaijri2020; Kimerling et al., Reference Kimerling, Serpi, Weathers, Kilbourne, Kang, Collins and Magruder2014; Montoya Gonzalez et al., Reference Montoya Gonzalez, Restrepo Bernal, Mejía-Montoya, Bareño-Silva, Sierra-Hincapié, Torres de Galvis and Gaviria-Arbeláez2016).
Respondents who met lifetime criteria for a given disorder in the CIDI were asked about age of onset using a question series designed to encourage accurate dating (McGrath et al., Reference McGrath, Al-Hamzawi, Alonso, Altwaijri, Andrade and Bromet2023). Subsequent disorder-specific questions asked about course of illness, current prevalence, and treatment-seeking. In addition, separate sections considered in this report assessed socio-demographics, suicidality, and family history of GAD. A more detailed description of the CIDI is presented elsewhere (Kessler, Heeringa, Pennell, Sampson, & Zaslavsky, Reference Kessler, Heeringa, Pennell, Sampson, Zaslavsky, Stein, Scott, de Jonge and Kessler2017).
GAD assessment
The CIDI assessment of GAD begins with retrospective questions about lifetime episodes of worrying. Respondents who report such episodes are shown a list of possible worry topics (including spaces for ‘Other’ self-nominated topics) and asked to describe the focus of their worries. Interviewers probe for up to three examples and record all topics mentioned. Respondents reporting multiple topics are then asked whether their worry or anxiety ‘was ever excessive or unreasonable or a lot stronger than it should have been?’ We classified respondents as having excessive or non-excessive worry based on this question. Subsequent questions were used to determine whether other GAD diagnostic criteria were met.
In the U.S. WMH survey, GAD diagnoses based on the CIDI had good concordance with diagnoses based on blinded clinical reassessments using the SCID (Ruscio et al., Reference Ruscio, Lane, Roy-Byrne, Stang, Stein, Wittchen and Kessler2005). Clinical reappraisal studies in other WMH surveys, although not evaluating GAD in isolation, found good concordance between CIDI and SCID diagnoses for any anxiety disorder including GAD (Haro et al., Reference Haro, Arbabzadeh-Bouchez, Brugha, de Girolamo, Guyer, Jin, Kessler, Kessler and Üstün2008; Kessler et al., Reference Kessler, Al-Desouki, King, Sampson, Al-Subaie, Al-Habeeb and Altwaijri2020; Montoya Gonzalez et al., Reference Montoya Gonzalez, Restrepo Bernal, Mejía-Montoya, Bareño-Silva, Sierra-Hincapié, Torres de Galvis and Gaviria-Arbeláez2016). Consistent with our previous work (Ruscio et al., Reference Ruscio, Hallion, Lim, Aguilar-Gaxiola, Al-Hamzawi, Alonso and Scott2017, Reference Ruscio, Hallion, Demyttenaere, Lee, Lim, Stein, Scott, de Jonge and Kessler2018), we generated DSM-5 GAD diagnoses by removing the DSM-IV hierarchical exclusion of a GAD diagnosis when symptoms occur exclusively during a mood disorder, as this was the only diagnostic difference between DSM-IV and DSM-5 GAD. Age of onset of GAD was assessed using probing methods demonstrated to improve dating accuracy (Knäuper, Cannell, Schwarz, Bruce, & Kessler, Reference Knäuper, Cannell, Schwarz, Bruce and Kessler1999). Respondents with 12-month GAD were asked additional questions about functioning, including an expanded Sheehan Disability Scale (Leon, Olfson, Portera, Farber, & Sheehan, Reference Leon, Olfson, Portera, Farber and Sheehan1997) that assessed role impairment caused by GAD during the most severe month in the year before interview.
Analysis methods
Analyses compared three mutually exclusive groups of respondents: those without GAD, those with non-excessive GAD (i.e. meeting all DSM-5 criteria other than excessiveness), and those with excessive GAD (i.e. meeting all DSM-5 criteria including excessiveness). We examined the data using simple cross-tabulations, regression analyses, and – in the case of models predicting the associations of lifetime GAD with subsequent first onset of temporally secondary comorbidity and suicidality – discrete-time survival models with person-year as the unit of analysis (Singer & Willett, Reference Singer and Willett1993). All analyses controlled for basic socio-demographic variables. All regression analyses and survival analyses of dichotomous outcomes used modified Poisson models with robust error variances to estimate risk ratios (RRs) directly (Mittinty & Lynch, Reference Mittinty and Lynch2023). Given that WMH data are weighted and geographically clustered, the Taylor series method was used to calculate design-based standard errors of proportions and 95% confidence intervals of RRs (Wolter, Reference Wolter2007). All analyses were carried out using Stata Statistical Software, Release 16.1 (StataCorp LLC, 2020).
Results
Prevalence and persistence
Globally, broadening the diagnosis to include non-excessive cases increases CIDI GAD prevalence estimates by approximately 50% over prevalence based on full DSM-5 criteria, with 30-day prevalence rising from 0.5% to 0.8%, 12-month prevalence rising from 1.3% to 1.9%, and lifetime prevalence rising from 2.6% to 4.0% (online Supplementary Table S2). Excessive lifetime cases are somewhat more likely to persist to the interview year (50.9%) than non-excessive cases (43.4%; χ21 = 26.9, p < 0.001). However, among 12-month cases, excessive GAD is no more likely to persist throughout the year (41.1%) than non-excessive GAD (43.7%; χ21 < 0.1, p = 0.977).
As hypothesized, non-excessive GAD is found disproportionately in LMICs (χ21 = 27.9, p < 0.001), with non-excessive cases making up 44.3% of broadly-defined lifetime GAD in LMICs compared to 33.6% in HICs (online Supplementary Table S2). Despite this, GAD is more persistent in LMICs, with more lifetime cases persisting to the interview year in LMICs than HICs for non-excessive (52.0% v. 40.1%; χ21 = 15.5, p < 0.001) as well as excessive (61.4% v. 48.4%; χ21 = 23.8, p < 0.001) GAD.
Severity
Reflecting the seriousness of their condition, fully half (50.5%) of non-excessive cases meet at least one of three indicators indexing severe GAD: frequent difficulty controlling worry (as opposed to occasional difficulty, which meets the diagnostic requirement), severe distress (as opposed to moderate distress, which meets the diagnostic requirement), or extreme impairment (as opposed to moderate impairment, which meets the diagnostic requirement) (online Supplementary Table S3). By comparison, two-thirds (66.1%) of excessive cases meet at least one of these indicators, and the number of indicators met is higher for excessive than non-excessive cases (χ23 = 28.9, p < 0.001). To account for this difference in severity, we repeated each subsequent analysis adjusting for the number of severity indicators met (0–3). This allowed us to determine whether differences observed between non-excessive and excessive GAD were attributable to other GAD severity features.
Worry content
Non-excessive cases worry about many of the same topics as excessive cases (Table 1). The most common worry topic is finances, reported by nearly half of non-excessive (49.4%) and excessive (47.8%) cases. The two groups also report comparable worry about family relationships and about a wide range of societal problems. However, the groups diverge in their worry about other topics. Non-excessive cases (32.5%) focus more on the health and welfare of loved ones than excessive cases (25.9%) – a group difference that is highly specific and does not extend to other concerns about one's social network. By contrast, excessive cases focus more on personal problems (excluding finances) than non-excessive cases. Excessive worry is also more likely to be diffuse in content (e.g. worrying about ‘everything’) than non-excessive worry. These differences persist after adjusting for GAD severity, suggesting that variation in worry content relates specifically to excessiveness.
Table 1. Worry contenta of non-excessive and excessive DSM-5 GAD
a Values represent the proportion of respondents with non-excessive (n = 2096) and excessive (n = 3595) lifetime GAD, respectively, who endorsed each worry topic. Within each problem domain, the final (‘any’) category reflects the proportion of respondents who endorsed at least one worry topic within that domain. All χ2 tests controlled for age at interview, sex, and country. Adjusted tests additionally controlled for severity, which was a 0–3 score summing across the three dichotomous indicators of uncontrollability, distress, and impairment.
b Includes worry about physical health, mental health, or substance use.
c Includes worry about any specific problem other than those listed above.
Socio-demographic characteristics
Non-excessive and excessive GAD have very similar socio-demographic correlates (Table 2). Both are elevated among females, those previously married (divorced, separated, or widowed), and those with an ‘Other’ employment status (mainly unemployed or disabled). Both groups also have similar patterns of associations with lower educational attainment and low household income compared to non-cases, although the associations are modest in magnitude. The only significant difference between GAD groups is that excessive cases are younger on average than non-excessive cases (χ23 = 30.6, p < 0.001), a difference that remains after adjusting for severity. This converges with hazard rates showing that excessive GAD typically begins at an earlier age than non-excessive GAD (online Supplementary Fig. S1). Whereas excessive GAD most often begins in adolescence or early adulthood, non-excessive GAD has a relatively consistent risk of onset across the lifespan.
Table 2. Socio-demographic correlatesa of non-excessive and excessive DSM-5 GAD
a The risk ratios (RRs) were estimated in modified Poisson regression models, with each sociodemographic variable predicting each GAD group. All models controlled for age at interview, sex, and country. Adjusted models additionally controlled for severity, which was a 0–3 score summing across the three dichotomous indicators of uncontrollability, distress, and impairment.
b Analyses were performed in the Part I sample. Mutually exclusive groups with non-excessive (n = 2096) and excessive (n = 3595) GAD were each compared with respondents having no lifetime GAD (n = 127 923), and then compared with each other.
c Analyses were performed in the Part II sample, which included 1983 respondents with non-excessive GAD, 3418 respondents with excessive GAD, and 70 749 respondents with no lifetime GAD.
* Significant at the 0.05 level.
Family history of GAD
The likelihood of having a parent with GAD is much higher for non-excessive (5.2%; RR = 2.5) and excessive (9.7%; RR = 3.6) cases than for individuals without GAD (1.9%; both χ21 > 24.3, both p < 0.001). The difference between excessive and non-excessive cases is also significant, but far smaller (RR = 1.2; χ21 = 10.2, p = 0.002). Similarly, excessive cases have more close relatives who are very worried or anxious compared to non-excessive cases (χ22 = 5.6, p = 0.004), but the difference is small and significant only for the risk of having two or more anxious relatives (RR = 1.1). All group differences hold when adjusting for severity, suggesting that excessive GAD is modestly more familial than non-excessive GAD.
Temporally secondary comorbidity
The vast majority of respondents with a lifetime history of non-excessive (73.7%) and excessive (87.1%) GAD had at least one other lifetime mental disorder occur subsequent to the onset of GAD, compared to just 28.2% of respondents without GAD (online Supplementary Table S4). Strikingly, individuals with non-excessive GAD are more than eight times as likely as those without GAD to develop a later mental disorder (Table 3). While there are some disorders for which excessive GAD is a stronger predictor than non-excessive GAD, the two GAD groups do not differ overall in their risk of subsequent psychopathology (RR = 1.2, χ21 = 3.0, p = 0.081). GAD also predicts the course of subsequent disorders, operationalized as 12-month persistence among lifetime cases (online Supplementary Table S5). There is no difference in this regard between excessive and non-excessive GAD (RR = 1.0; χ21 < 0.1, p = 0.989): Both are associated with a 30% increase in risk of experiencing a persistent comorbid disorder.
Table 3. Associationsa of non-excessive and excessive DSM-5 GAD with the onset of subsequent DSM-IV disorders
a The risk ratios (RRs) come from discrete-time survival analysis, with person-year as the unit of analysis. Each model estimated the association between variably-defined GAD and the subsequent onset of a temporally secondary disorder, controlling for age at interview, age-squared, sex, and country. Adjusted models additionally controlled for severity, which was a 0–3 score summing across the three dichotomous indicators of uncontrollability, distress, and impairment. The models for aggregate disorder categories (‘any’ disorder) based associations on the earliest-occurring disorder within the category.
b Analyses were performed in the Part I sample. Mutually exclusive groups with non-excessive (n = 2096) and excessive (n = 3595) GAD were each compared with respondents having no lifetime GAD (n = 127 923), and then compared with each other.
c Not assessed in Israel.
d Not assessed in Australia, Israel, South Africa, or Ukraine.
e Analyses were performed in the Part II sample, which included 1983 respondents with non-excessive GAD, 3418 respondents with excessive GAD, and 70 749 respondents with no lifetime GAD. The Part II sample for childhood disorders included 845 respondents with non-excessive GAD, 1823 respondents with excessive GAD, and 38 776 respondents with no lifetime GAD.
f Not assessed in Australia, Israel, Japan, New Zealand, Poland, South Africa, or Ukraine.
g Includes bipolar I disorder, bipolar II disorder, or subthreshold bipolar disorder as defined by Merikangas et al. (Reference Merikangas, Akiskal, Angst, Greenberg, Hirschfeld, Petukhova and Kessler2007). Not assessed in Belgium, France, Germany, Italy, Netherlands, South Africa, Spain, or Ukraine.
h Not assessed in Portugal.
i Not assessed in Australia, Bulgaria, Israel, Japan, New Zealand, South Africa, or Ukraine.
j Recall of childhood disorders was restricted to respondents < 45 years of age at interview (with the exception of Ukraine, which was restricted to age < 40 at interview).
k Not assessed in Australia, Bulgaria, Bulgaria 2, Israel, Japan, Lebanon, New Zealand, South Africa, or Ukraine.
l Not assessed in Australia, Israel, Japan, Lebanon, New Zealand, South Africa, or Ukraine.
m Not assessed in Australia, Belgium, Bulgaria 2, France, Germany, Israel, Italy, Medellin, Mexico, Murcia, Netherlands, New Zealand, or Spain.
n Eating disorders were assessed in a random 50% of the Part II sample. Not assessed in Australia, Bulgaria, Bulgaria 2, Israel, Japan, Lebanon, Romania, South Africa, or Ukraine.
* Significant at the 0.05 level.
Subsequent suicidality
Compared to those without GAD, individuals with non-excessive GAD are six times as likely to subsequently develop suicidal ideation, two times as likely to make a suicide plan, and nearly two times as likely to make a suicide attempt (Table 4). Excessive and non-excessive GAD differ only in their prediction of suicidal ideation (χ21 = 10.0, p = 0.002); the predictive advantage for excessive GAD is modest (RR = 1.3) and disappears when controlling for severity (RR = 1.1, χ21 = 2.0, p = 0.159). GAD also predicts the persistence of suicidal ideation once it develops (online Supplementary Table S6), with no difference in prediction by excessiveness (RR = 1.1; χ21 = 0.7, p = 0.397).
Table 4. Associationsa of non-excessive and excessive DSM-5 GAD with the onset of subsequent suicide-related outcomes
a The risk ratios (RRs) come from discrete-time survival analysis with person-year as the unit of analysis. Analyses were performed in the Part II sample, which included 1983 respondents with non-excessive GAD, 3418 respondents with excessive GAD, and 70 749 respondents with no lifetime GAD. Each model estimated the association between variably-defined GAD and the subsequent onset of a temporally secondary suicide-related outcome. The models for suicide plan were estimated among respondents with lifetime suicidal ideation. The models for suicide attempt were estimated among respondents with lifetime suicidal ideation and also controlled for person-year lifetime suicide plan status. All models controlled for age at interview, age-squared, sex, and country. Adjusted models additionally controlled for severity, which was a 0–3 score summing across the three dichotomous indicators of uncontrollability, distress, and impairment.
* Significant at the 0.05 level.
Treatment-seeking for GAD
Nearly half (43.4%) of non-excessive cases seek treatment specifically for GAD in their lifetime (online Supplementary Table S7). Treatment-seeking is somewhat higher among excessive cases (51.6%; χ21 = 6.8, p = 0.010), but this difference declines to non-significance after adjusting for severity (χ21 = 2.7, p = 0.101). In the year before interview, 18.9% of non-excessive and 26.9% of excessive 12-month cases received treatment for GAD symptoms (χ21 = 19.6, p < 0.001). This group difference is attenuated, though not eliminated, after adjusting for severity (χ21 = 12.6, p < 0.001).
Impairment due to GAD
Excessive and non-excessive 12-month cases report similar functional impairment due to GAD in the past year (Table 5). Approximately 65–75% of respondents in both groups report some impairment, and approximately 50–60% report moderate or severe impairment, in at least one of the four domains assessed. Only 1 of the 12 nested responses distinguishes the two groups; this difference disappears after adjusting for GAD severity. Remarkably, non-excessive 12-month cases report an average of 30 days (s.d. = 79.6) in the past year when they were totally unable to work or carry out their normal activities because of GAD. This is comparable to the average of 33 days (s.d. = 90.9) reported by excessive 12-month cases, whether analyzed across the entire year (χ21 = 1.5, p = 0.213) or as a fraction of the time spent in GAD episodes during the year (χ21 = 2.6, p = 0.109).
Table 5. Severity of role impairmenta associated with 12-month non-excessive and excessive GAD
a Values represent the proportion of respondents with non-excessive (n = 935) and excessive (n = 1822) 12-month GAD, respectively, reporting severe (score of 7–10), severe or moderate (score of 4–10), or any (score of 1–10) impairment in each of the four Sheehan Disability Scale domains of functioning. All χ2 tests controlled for age at interview, sex, and country. Adjusted tests additionally controlled for severity, which was a 0–3 score summing across the three dichotomous indicators of uncontrollability, distress, and impairment.
Discussion
This study undertook the first cross-national test of the GAD excessiveness requirement. We found that removing this requirement increases the prevalence of DSM-5 GAD by about 50% worldwide. While the increase in prevalence may raise concerns about pathologizing normal anxiety or diagnosing the so-called ‘worried well,’ our data do not bear out these concerns. Instead, they show that non-excessive worriers who meet all other GAD criteria are very similar to excessive worriers with GAD, and very different from persons without GAD, on a wide array of clinically significant features. Where differences emerge between non-excessive and excessive GAD, they are modest in magnitude, and many disappear when other disorder features are controlled. Most important, on almost any metric of seriousness (e.g. help-seeking, functional impairment), non-excessive GAD appears to be clinically significant.
Removing excessiveness would bring GAD in line with other disorders that often emerge in the context of significant stressors. There is no requirement, for example, that sadness must be excessive for major depression to be diagnosed. Instead, major depression is distinguished from normal sadness through other criteria specifying intensity, frequency, and duration of symptoms that are judged so substantial and impairing that they warrant clinical attention. At the same time, removing excessiveness could open GAD up to criticism, previously levied at depression, that it is inappropriate to diagnose symptoms that are severe yet proportional to their context (Horwitz, Reference Horwitz2015). Our results inform this larger debate by showing that non-excessive GAD is highly familial, persistent, disabling, and predictive of a wide range of outcomes that matter for clinical care. Whether it is appropriate to diagnose proportional yet disabling reactions to stress remains an important question for the field to resolve (Stein, Palk, & Kendler, Reference Stein, Palk and Kendler2021), but there seems no reason why the answer should be different for GAD than for other disorders.
Our finding that non-excessive GAD is similar to but less severe than excessive GAD is consistent with mounting evidence for the dimensional nature of anxiety (Haslam, Holland, & Kuppens, Reference Haslam, Holland and Kuppens2012; Ruscio, Reference Ruscio2019). Given this dimensionality, any reduction in diagnostic requirements will result in new cases of lower severity, just as any increase in diagnostic criteria beyond those currently required will result in fewer cases of greater severity. The open question is where along this dimension the diagnostic threshold should be drawn for categorical decision making. Following DSM's own directive that the threshold be placed at the point of clinical significance (APA, 2013), our results suggest that the DSM-5 threshold for GAD is set too high. Non-excessive cases are just as disabled by their symptoms as excessive cases, report losing a full month in the past year due to those symptoms, and often seek treatment for those symptoms.
Why might individuals whose symptoms are so severe that they otherwise qualify for a GAD diagnosis regard their worry as non-excessive? We found that non-excessive worriers often focus on topics that might be perceived as normative (e.g. finances, well-being of loved ones), and focus less on non-specific concerns (e.g. ‘everything’) that may be easier to recognize as excessive. This may reduce insight, perhaps especially among lifelong worriers who have a limited basis for comparison. Alternatively, non-excessive worry might be a justifiable reaction to genuine deprivation or stress. We observed higher rates of non-excessive GAD in LMICs than HICs, perhaps reflecting heightened stress and uncertainty in parts of the world where caring for oneself and one's family can be a daily struggle. Greater persistence of symptoms in LMICs may also reflect a greater likelihood of ongoing stressors in those countries. Critically, an unintended consequence of the excessiveness requirement may be that individuals with low socioeconomic status or stigmatized identities – whose worries may be viewed as an understandable reaction to the stressors they face – may be less likely to receive a GAD diagnosis and gain access to quality treatment than more-advantaged patients with the same presenting complaint. We did not find a stronger association of household income with non-excessive than excessive GAD, but were unable to test whether other forms of disadvantage previously linked to chronic stress were concentrated among non-excessive cases (Hatzenbuehler, Reference Hatzenbuehler2009; Zvolensky, Garey, & Bakhshaie, Reference Zvolensky, Garey and Bakhshaie2017). Such tests are needed to explore whether eliminating excessiveness reduces disparities in GAD diagnosis and treatment.
Our results imply that clinicians should not dismiss symptoms for being consistent with (worrisome) life circumstances, as intervention offers opportunities for symptom relief and perhaps even prevention of subsequent comorbidity and suicidality. Notably, nearly half of non-excessive cases seek help for GAD symptoms, though whether clinicians recognize their difficulties as GAD is unclear. Diagnosing these cases with an ‘other specified’ anxiety disorder is an imperfect solution, as this does not promote the use of treatments that are empirically supported for GAD. Removing the excessiveness requirement may also have implications for early intervention. Recently, there has been growing attention to anxiety screening in primary healthcare, with the U.S. Preventive Services Task Force (2023) recommending that all adults under age 65 be screened for anxiety. As GAD is the most prevalent anxiety disorder in primary care (DeMartini, Patel, & Fancher, Reference DeMartini, Patel and Fancher2019; Lieb, Becker, & Altamura, Reference Lieb, Becker and Altamura2005) and is the focus of recommended screening tools (U.S. Preventive Services Task Force, 2023), it is likely to be among the most common anxiety disorders detected through screening. Knowing where to set the cutoff for GAD is crucial, as it will determine who will be referred for further services, which will have direct implications for access and cost.
Limitations and future directions
Our study had several limitations. First, we asked respondents directly about excessiveness. DSM-5 does not specify who should decide whether worry is excessive; we prioritized self-reports, but other reporters may have answered differently. Second, we were unable to explore respondents' understanding of ‘excessive’ worry. Given ambiguity around the meaning of excessiveness (Shear, Reference Shear2012), there is a need to better understand what factors people consider when appraising their worry as excessive. Third, we did not probe for culturally specific worry topics (Hinton, Hsia, Park, Rasmussen, & Pollack, Reference Hinton, Hsia, Park, Rasmussen, Pollack, McKay, Abramowitz, Taylor and Asmundson2009). However, past research suggests that GAD worries are more similar than dissimilar across cultures (Marques et al., Reference Marques, Robinaugh, LeBlanc and Hinton2011), and our interview offered flexibility by using broad categories and by recording up to three ‘Other’ topics beyond the standard list. Fourth, we tested excessiveness as a predictor of subsequent outcomes, but the temporal order of predictors and outcomes was determined via retrospective reports. Prospective longitudinal studies are needed to replicate these results.
Our analyses did not establish whether these results are unique to excessiveness or whether similar results might be observed for other contested features of GAD (Andrews et al., Reference Andrews, Hobbs, Borkovec, Beesdo, Craske, Heimberg and Stanley2010). Future studies could use the analytic framework employed here to evaluate the population-level impact of revising other diagnostic criteria. All criteria that require revision could then be studied together (cf. Beesdo-Baum et al., Reference Beesdo-Baum, Winkel, Pine, Hoyer, Höfler, Lieb and Wittchen2011; Ruscio et al., Reference Ruscio, Chiu, Roy-Byrne, Stang, Stein, Wittchen and Kessler2007) to determine which combination optimizes the distinction between clinically significant GAD and normal-range anxiety. Finally, we focused on clinically important outcomes that any useful diagnosis should predict. Future work should test an equally wide range of risk factors, to ensure that removing excessiveness preserves the validity of the diagnosis as well as its clinical utility. The main risk factor we considered, family history of GAD, was heightened in excessive compared to non-excessive cases, although the difference was small. Other risk factors, such as life stress, might plausibly show the opposite pattern, given the more widely distributed risk of onset over the lifespan for non-excessive than excessive GAD.
Conclusions
Individuals who deny that their worry is excessive – but meet all other GAD criteria – experience nontrivial mental health symptoms associated with substantial functional impairment, help-seeking, and risk of adverse mental health outcomes. Although somewhat milder in severity than excessive GAD cases, non-excessive cases are affected in ways that unambiguously meet the DSM-5 requirement of a ‘clinically significant disturbance’ that is ‘associated with significant distress or disability’ (APA, 2013, p. 20). Lowering the GAD threshold to include these worriers would improve clinical coverage, promote recognition of symptoms that are important for effective prognosis and clinical management, and yield a more defensible GAD diagnosis.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S003329172400182X
Acknowledgments
Collaborators*: The WHO World Mental Health Survey collaborators are Yasmin A. Altwaijri, PhD; Laura Helena Andrade, MD, PhD; Corina Benjet, PhD; Jose Miguel Caldas-de-Almeida, MD, PhD; Alfredo H. Cía, MD; Louisa Degenhardt, PhD; Meredith G. Harris, PhD; Hristo Hinkov, MD, PhD; Chi-yi Hu, MD, PhD; Georges Karam, MD; Alan E. Kazdin, PhD; Norito Kawakami, MD, DMSc; Salma Khaled, PhD; John J. McGrath, MD, PhD; Maria Elena Medina-Mora, PhD; Jacek Moskalewicz, PhD; Margreet ten Have, PhD; Daniel V. Vigo, MD, DrPH; David R. Williams, MPH, PhD; Bogdan Wojtyniak, ScD; Peter Woodruff, MBBS, PhD, FRCPsych; Miguel Xavier, MD, PhD; Alan M. Zaslavsky, PhD.
Funding statement
The World Health Organization World Mental Health (WMH) Survey Initiative is supported by the United States National Institute of Mental Health (NIMH; R01 MH070884), the John D. and Catherine T. MacArthur Foundation, the Pfizer Foundation, the United States Public Health Service (R13-MH066849, R01-MH069864, and R01 DA016558), the Fogarty International Center (FIRCA R03-TW006481), the Pan American Health Organization, Eli Lilly and Company, Ortho-McNeil Pharmaceutical Inc., GlaxoSmithKline, and Bristol-Myers Squibb. We thank the staff of the WMH Data Collection and Data Analysis Coordination Centers for assistance with instrumentation, fieldwork, and consultation on data analysis. None of the funders had any role in the design, analysis, interpretation of results, or preparation of this paper.
The Argentina survey – Estudio Argentino de Epidemiología en Salud Mental (EASM) – was supported by a grant from the Argentinian Ministry of Health (Ministerio de Salud de la Nación) – (Grant Number 2002-17270/13-5).
The 2007 Australian National Survey of Mental Health and Wellbeing was funded by the Australian Government Department of Health and Ageing.
The São Paulo Megacity Mental Health Survey is supported by the State of São Paulo Research Foundation (FAPESP) Thematic Project Grant 03/00204-3.
The Bulgarian Epidemiological Study of common mental disorders -EPIBUL – is supported by the Ministry of Health and the National Center for Public Health Protection. (For papers that include the Bulgaria 2 survey, add: EPIBUL 2, conducted in 2016–17, is supported by the Ministry of Health and European Economic Area Grants.
The Colombian National Study of Mental Health (NSMH) is supported by the Ministry of Social Protection.
The Mental Health Study Medellín – Colombia was carried out and supported jointly by the Center for Excellence on Research in Mental Health (CES University) and the Secretary of Health of Medellín.
The ESEMeD surveys (Belgium; France; Germany; Italy; Netherlands; Spain) were funded by the European Commission (Contracts QLG5-1999-01042; SANCO 2004123, and EAHC 20081308), the Piedmont Region (Italy), Fondo de Investigación Sanitaria, Instituto de Salud Carlos III, Spain (FIS 00/0028), Ministerio de Ciencia y Tecnología, Spain (SAF 2000-158-CE), Generalitat de Catalunya (2017 SGR 452; 2014 SGR 748), Instituto de Salud Carlos III (CIBER CB06/02/0046, RETICS RD06/0011 REM-TAP), and other local agencies and by an unrestricted educational grant from GlaxoSmithKline.
Implementation of the Iraq Mental Health Survey (IMHS) and data entry were carried out by the staff of the Iraqi Ministry of Health and Ministry of Planning with direct support from the Iraqi IMHS team with funding from both the Japanese and European Funds through United Nations Development Group Iraq Trust Fund (UNDG ITF).
The Israel National Health Survey is funded by the Ministry of Health with support from the Israel National Institute for Health Policy and Health Services Research and the National Insurance Institute of Israel.
The World Mental Health Japan (WMHJ) Survey is supported by the Grant for Research on Psychiatric and Neurological Diseases and Mental Health (H13-SHOGAI-023, H14-TOKUBETSU-026, H16-KOKORO-013, H25-SEISHIN-IPPAN-006) from the Japan Ministry of Health, Labour and Welfare.
The Lebanese Evaluation of the Burden of Ailments and Needs of the Nation (L.E.B.A.N.O.N.) is supported by the Lebanese Ministry of Public Health, the WHO (Lebanon), National Institute of Health/Fogarty International Center (R03 TW006481-01), anonymous private donations to IDRAAC, Lebanon, and unrestricted grants from Algorithm, AstraZeneca, Benta, Bella Pharma, Eli Lilly, Glaxo Smith Kline, Lundbeck, Novartis, OmniPharma, Pfizer, Phenicia, Servier, and UPO.
The Mexican National Comorbidity Survey (MNCS) is supported by the National Institute of Psychiatry Ramon de la Fuente (INPRFMDIES 4280) and by the National Council on Science and Technology (CONACyT-G30544-H), with supplemental support from the Pan American Health Organization (PAHO).
Te Rau Hinengaro: The New Zealand Mental Health Survey (NZMHS) is supported by the New Zealand Ministry of Health, Alcohol Advisory Council, and the Health Research Council.
The Northern Ireland Study of Mental Health was funded by the Health & Social Care Research & Development Division of the Public Health Agency.
The Peruvian World Mental Health Study was funded by the National Institute of Health of the Ministry of Health of Peru.
The Polish project Epidemiology of Mental Health and Access to Care –EZOP Project (PL 0256) was carried out by the Institute of Psychiatry and Neurology, Warsaw in consortium with the Department of Psychiatry, Medical University, Wroclaw, the National Institute of Public Health – National Institute of Hygiene, Warsaw and in partnership with Psykiatrist Institut, Vinderen, Universitet Oslo. The project was funded by the European Economic Area Financial Mechanism and the Norwegian Financial Mechanism. EZOP project was co-financed by the Polish Ministry of Health.
The Portuguese Mental Health Study was carried out by the Department of Mental Health, Faculty of Medical Sciences, NOVA University of Lisbon, with collaboration of the Portuguese Catholic University, and was funded by Champalimaud Foundation, Gulbenkian Foundation, Foundation for Science and Technology (FCT), and the Portuguese Ministry of Health.
The Romania WMH study projects ‘Policies in Mental Health Area’ and ‘National Study regarding Mental Health and Services Use’ were carried out by National School of Public Health & Health Services Management (present National School of Public Health Management & Professional Development, Bucharest), with technical support of Metro Media Transilvania, the National Institute of Statistics, SC Cheyenne Services SRL, and Statistics Netherlands. Funding was provided by the Ministry of Public Health with supplemental support of Eli Lilly Romania SRL.
The South Africa Stress and Health Study (SASH) is supported by the US National Institute of Mental Health (R01-MH059575) and National Institute of Drug Abuse with supplemental funding from the South African Department of Health and the University of Michigan.
The Psychiatric Enquiry to General Population in Southeast Spain – Murcia (PEGASUS-Murcia) Project has been financed by the Regional Health Authorities of Murcia (Servicio Murciano de Salud and Consejería de Sanidad y Política Social) and Fundación para la Formación e Investigación Sanitarias (FFIS) of Murcia.
The Ukraine Comorbid Mental Disorders during Periods of Social Disruption (CMDPSD) study is funded by the US National Institute of Mental Health (NIMH; R01-MH61905).
The US National Comorbidity Survey Replication (NCS-R) is supported by the National Institute of Mental Health (NIMH; U01-MH60220) with supplemental support from the National Institute of Drug Abuse (NIDA), the Substance Abuse and Mental Health Services Administration (SAMHSA), the Robert Wood Johnson Foundation (RWJF; Grant 044708), and the John W. Alden Trust.
DJS is supported by the Medical Research Council of South Africa (MRC).
MBS has received research support from NIH, Department of Veterans Affairs, and the Department of Defense.
A complete list of all within-country and cross-national WMH publications can be found at http://www.hcp.med.harvard.edu/wmh/
Competing interests
In the past 3 years, Dr Kessler was a consultant for Cambridge Health Alliance, Canandaigua VA Medical Center, Child Mind Institute, Holmusk, Massachusetts General Hospital, Partners Healthcare, Inc., RallyPoint Networks, Inc., Sage Therapeutics and University of North Carolina. He has stock options in Cerebral Inc., Mirah, PYM (Prepare Your Mind), Roga Sciences and Verisense Health. The other authors declare that they have no conflicts of interest.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
