|Home | About | Journals | Submit | Contact Us | Français|
Excessive worry is required by DSM-IV, but not ICD-10, for a diagnosis of generalized anxiety disorder (GAD). No large-scale epidemiological study has ever examined the implications of this requirement for estimates of prevalence, severity, or correlates of GAD.
Data were analyzed from the US National Comorbidity Survey Replication, a nationally representative, face-to-face survey of adults in the US household population that was fielded in 2001–2003. DSM-IV GAD was assessed with Version 3.0 of the WHO Composite International Diagnostic Interview. Non-excessive worriers meeting all other DSM-IV criteria for GAD were compared with respondents who met full GAD criteria as well as with other survey respondents to consider the implications of removing the excessiveness requirement.
The estimated lifetime prevalence of GAD increases by approximately 40% when the excessiveness requirement is removed. Excessive GAD begins earlier in life, has a more chronic course, and is associated with greater symptom severity and psychiatric comorbidity than non-excessive GAD. However, non-excessive cases nonetheless evidence substantial persistence and impairment of GAD as well as significantly elevated comorbidity compared to respondents without GAD. Non-excessive cases also have socio-demographic characteristics and familial aggregation of GAD comparable to excessive cases.
Although individuals who meet all criteria for GAD other than excessiveness have a somewhat milder presentation than those with excessive worry, their syndromes are sufficiently similar to those with excessive worry to warrant a GAD diagnosis.
Ever since generalized anxiety disorder (GAD) was first defined as a disorder of pathological worry in the revised third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R) (American Psychiatric Association, 1987), researchers have debated the features that make worry pathological and that should be required for a diagnosis of GAD (Borkovec et al., 1991; Weems et al., 2000). The requirement that worry be excessive was one of few worry features introduced in DSM-III-R that was retained in DSM-IV (American Psychiatric Association, 1994). However, in contrast to the close scrutiny received by other GAD symptoms such as uncontrollable worry (Borkovec et al., 1991; Craske et al., 1989) and somatic associated symptoms (Brown et al., 1995; Marten et al., 1993), very little has been written about the excessiveness criterion. This lack of attention is surprising given that the criterion has repeatedly been identified as a controversial aspect of the GAD diagnosis (Kessler & Wittchen, 2002; Rickels & Rynn, 2001) and given that ICD-10 (International Classification of Diseases, 10th ed.) (World Health Organization, 1993), unlike DSM-IV, does not require worry to be excessive for GAD to be diagnosed.
The excessiveness criterion is based on a handful of studies that found excessive worry about minor matters to distinguish individuals with GAD from individuals with other anxiety disorders (Craske et al., 1989; Roemer et al., 1997; Sanderson & Barlow, 1990). The DSM, however, does not give privileged status to minor worries, but more broadly requires excessive worries about multiple events or activities. This broad requirement has been found to be less discriminating than most other symptoms of GAD in distinguishing threshold and subthreshold cases of the disorder (Diefenbach et al., 2003; Hoyer et al., 2002; Ruscio, 2002). Little else is known about the excessiveness requirement because existing studies of GAD, conducted primarily using DSM-III-R criteria that required worry to be “unrealistic or excessive,” rarely evaluated the implications of excessiveness separately from those of the later-discarded requirement of unrealistic worry.
Notably, the excessiveness criterion poses many of the same assessment challenges that led to the removal in DSM-IV of the requirement that worry must be “unrealistic” (Abel & Borkovec, 1995; Barlow & Wincze, 1998). For example, there is considerable confusion over what makes worry “excessive,” with excessiveness variably defined as worry that is frequent and intense (Hoyer et al., 2002), persistent for much of the day (Campbell et al., 2003), focused on unimportant things (Bienvenu et al., 1998), out of proportion to objective life circumstances (Brown, 1997; Kessler & Wittchen, 2002), or occurring more than the individual wishes or more than others feel is warranted (Roemer et al., 2002). There is also uncertainty over who should determine whether worry is excessive, and whose views should take precedence in the common event that the assessor and respondent, or different assessors, disagree (Wittchen et al., 1995). Finally, there is the question of what leads individuals to appraise their worry as excessive (Borkovec et al., 1991), whether this appraisal corresponds to objective characteristics of the worry experience (Ruscio & Borkovec, 2004; Ruscio et al., 2003), and whether the appraisal of excessiveness is influenced by differing cultural standards for describing internal states or for defining worries as excessive (Kessler & Wittchen, 2002). Given these many ambiguities, it is perhaps not surprising that one study of DSM-III-R GAD found excessiveness to be the criterion on which assessors most often disagreed, with its elimination leading to a dramatic rise in interrater reliability of diagnosis (Wittchen et al., 1995).
In addition to these problems of conceptual confusion which decrease diagnostic reliability, concerns can be raised about the implications of the excessiveness criterion for diagnostic validity. In particular, critics have noted that the excessiveness requirement excludes from the GAD diagnosis individuals who develop clinically significant generalized anxiety in the context of chronic, objectively stressful situations (Kessler & Wittchen, 2002). This is in striking contrast to a diagnosis such as major depression, where there is no requirement that the dysphoria or anhedonia must be excessive in relation to any objective life circumstances with which they may be associated. Importantly, community epidemiological research has shown that a sizable group of individuals exist who are diagnosed with GAD by ICD-10, but not by DSM-IV, solely because they fail to meet the excessiveness criterion (Slade & Andrews, 2001). At present, little is known about these individuals or their likely impact on estimates of the prevalence, severity, or correlates of GAD. Although a prior study found similar demographic and comorbidity patterns for GAD and non-GAD groups differing only in their endorsement of excessive or unrealistic worry (Bienvenu et al., 1998), no research has yet examined the implications of the excessiveness criterion alone for a broad range of epidemiological variables traditionally used to describe mental disorders and to help evaluate their validity (Robins & Guze, 1970). Such research has taken on renewed importance as work begins on DSM-V and ICD-11, providing a unique opportunity to inform revision of the GAD criteria in ways that may enhance the validity of the diagnosis.
The current paper evaluates the implications of the GAD excessiveness criterion for estimates of prevalence and correlates based on data from the recently completed US National Comorbidity Survey Replication (NCS-R) (Kessler & Merikangas, 2004). Non-excessive worriers meeting all other DSM-IV criteria for GAD were compared with GAD excessive worriers and with other NCS-R respondents to consider the consequences of the excessiveness requirement. Given indications that excessiveness and uncontrollability of worry are highly correlated (Brown et al., 2001), we also evaluated the importance of the excessiveness criterion over and above the effect of the requirement that worry be uncontrollable and the effect of other DSM-IV required features of pathological worry.
The NCS-R is a nationally representative survey of the US household population that was conducted from February 2001 to December 2003 (Kessler & Merikangas, 2004). Participants were ages 18 or older and were selected using a multi-stage clustered area probability sampling design. The response rate was 70.9%. Consistent with the recruitment procedures of the baseline NCS a decade earlier (Kessler et al., 1994), participants first received a letter and study fact brochure, then were visited by an interviewer who described the study to them and obtained verbal informed consent before scheduling an interview. Respondents received $50 for their participation.
The interview was carried out in two parts. Part I consisted of the core diagnostic assessment and was administered to all 9,282 respondents. Part II consisted of measures assessing correlates of disorders and was administered to 100% of the respondents who met criteria for any disorder in Part I plus a probability subsample of other respondents. The analyses reported here were performed on this Part II sample, which included 5,692 respondents. The Part II sample was weighted to adjust for differential probabilities of respondent selection within households as a function of household size, more intense recruitment of difficult-to-recruit respondents, the higher selection probabilities of Part I respondents with a lifetime disorder, and residual variation between sample distributions and population distributions on a range of geographic and socio-demographic variables in the 2000 US Census. More detailed descriptions of NCS-R sampling design and weighting procedures are presented elsewhere (Kessler et al., 2004b).
DSM-IV disorders were assessed using Version 3.0 of the WHO Composite International Diagnostic Interview (CIDI 3.0), a fully-structured, lay-administered diagnostic interview. Blind clinical re-interviews of a probability subsample of NCS-R respondents using the Structured Clinical Interview for DSM-IV (SCID) (First et al., 2002) found generally good concordance between SCID and CIDI 3.0 diagnoses (Kessler et al., in press-a). DSM-IV diagnostic hierarchy and organic exclusion rules were used in making diagnoses. Respondents who met all other criteria for GAD were separated into those defined as having excessive worry and those not having excessive worry based on their responses to the question, “Do you think your (worry or anxiety) was ever excessive or unreasonable or a lot stronger than it should have been?” Importantly, CIDI 3.0 is constructed so that all respondents continue the GAD section regardless of their response to this question, allowing us to compare respondents who meet all DSM-IV GAD criteria except excessiveness with those who meet full GAD criteria.
In addition to comorbid DSM-IV disorders, other correlates of GAD considered here include socio-demographic variables, parental GAD, and functional impairment. The socio-demographic variables include respondent age, sex, race-ethnicity, education, employment status, and marital status. Parental history of GAD was reported by respondents for their biological father and mother (or the adult male and female who raised them) in response to questions designed to expand the Family History RDC interview (Andreasen et al., 1977) to include GAD (Kendler et al., 1997). Functional impairment was assessed in two ways, both focused on 12-month GAD. First, the Sheehan Disability Scales (Leon et al., 1997) assessed the degree to which GAD interfered with home management, work, close relationships, and social life during the month in the past year when the respondent’s GAD was reported to be most severe. Each domain was rated on a 0–10 scale that was divided into the categories of none (0), mild (1–3), moderate (4–6), and severe (7–10) for the present analyses. Second, role impairment in the past year was assessed by two variables: (a) days out of role, reflecting the number of days in the past 12 months during which the respondent was “totally unable” to work or carry out daily activities because of GAD, and (b) role impairment in episode, reflecting the percent of all days in the GAD episode that were spent out of role.
The excessive and non-excessive GAD subgroups were compared in terms of prevalence, age of onset, persistence, severity, socio-demographic correlates, comorbidity, impairment, and parental GAD. Age-of-onset curves were estimated using the actuarial method (Halli & Rao, 1992). Z-tests were used to compare means on measures of persistence, whereas chi-square analyses were used to compare the groups on severity. Logistic regression was used to examine associations with socio-demographics and comorbid disorders as well as with parental history of GAD. To determine whether excessiveness uniquely contributes to the prediction of these variables, over and above other features of worry required for a DSM-IV GAD diagnosis, all analyses yielding statistically significant group differences were rerun controlling for five dichotomous variables representing uncontrollability of worry, distress associated with worry, and impairment caused by worry.
Standard errors and significance tests were estimated using the Taylor series linearization method (Wolter, 1985) implemented in the SUDAAN (2002) software system to adjust for the fact that the NCS-R sample design included clustering and for the fact that the Part II NCS-R data are weighted. Multivariate significance was estimated in logistic regression analysis using Wald χ2 tests based on design-corrected coefficient variance-covariance matrices. Statistical significance was evaluated using two-tailed .05-level tests.
Broadly-defined GAD includes more than twice as many excessive (n = 411, 70.1%) as non-excessive (n = 172, 29.9%) cases. When non-excessive cases meeting all other criteria are diagnosed with GAD, the prevalence of the disorder increases by approximately 40%, with comparable increases in lifetime (from 3.8 to 5.5%), 12-month (from 2.0 to 2.8%), and 1-month (from 1.2 to 1.7%) prevalence estimates.
Although cumulative age-of-onset curves differ significantly for excessive and non-excessive GAD, the two distributions have similar slopes. (Figure 1) The major difference between the two distributions is that excessive GAD begins significantly earlier in life (M = 26.5) than non-excessive GAD (M = 32.0; z = 3.6, p = .001). Excessive GAD also has a more persistent course than non-excessive GAD. Excessive cases experience GAD during more years of their lives (M = 9.8) than non-excessive cases (M = 7.8; z = 2.4, p = .022). Excessive cases also have greater annual persistence of GAD than non-excessive cases, calculated as the number of years with GAD divided by the total years between the first and last episode (0.6 vs. 0.5; z = 2.6, p = .011). These differences remain significant even after uncontrollability, distress, and impairment are controlled (all z > 2.0, all p < .05). By contrast, there is no difference between the two subgroups in number of months in episode in the past year among 12-month cases (7.9 vs. 8.1, z = 0.3, p = .784).
A similarly high proportion of excessive (91.8%) and non-excessive (86.5%) GAD cases have at least one other DSM-IV disorder assessed in the NCS-R (χ12 = X.X, p = .XXX). Both excessive and non-excessive GAD have significant lifetime comorbidity with the vast majority of these disorders after adjusting for respondent age, sex, and race. (Table 1) However, excessive GAD has consistently higher ORs than non-excessive GAD, significantly so for 8 of the 15 mental disorders assessed. The median OR (inter-quartile range) is 5.5 (3.6–8.4) for excessive GAD compared to 2.8 (1.8–4.2) for non-excessive GAD. Seventy-five percent of the statistically significant excessive vs. non-excessive differences in ORs remain significant after worry uncontrollability, distress, and impairment are controlled.
The greater chronicity and comorbidity of excessive than non-excessive GAD suggest that excessiveness might be a generalized severity marker. To test this hypothesis directly, a measure of GAD severity was constructed from a set of 11 nested dichotomous variables representing uncontrollability, distress, and impairment associated with worry and generalized anxiety. These variables were submitted to an IRT analysis and the resulting dimensional score was trichotomized into mild, moderate, and severe categories. A significantly higher percentage of non-excessive (37.3%) than excessive (26.0%; χ12 = 6.7, p = .010) cases had mild GAD according to this classification, whereas a significantly higher percentage of excessive (42.7%) than non-excessive (32.7%; χ12 = 6.7, p = .010) cases had severe GAD. Notably, although these differences are statistically significant, they are not substantial in substantive terms. Over 60% of non-excessive cases were classified as moderate or severe. Moreover, non-excessive cases were no more likely than excessive cases to report minimally diagnostic levels of uncontrollability, distress, or impairment.
There is no reliable association between excessiveness and impairment among 12-month cases. (Table 2) Of the 12 nested severity responses on the Sheehan Disability Scales, only one (severe interference with close relationships) significantly distinguishes excessive and non-excessive GAD. Furthermore, linear regression analysis found no difference between the GAD subgroups either in days out of role (t = 0.3, p = .565) or in overall role impairment while in episode (t = 0.6, p = .442). In an absolute sense, non-excessive cases reported considerable functional impairment, with the vast majority (69.7–84.4%) reporting at least some impairment in household, occupational, interpersonal, or social functioning and the majority (52.2–68.2%) reporting moderate or severe impairment in these domains.
Excessive cases are no more likely than non-excessive cases to receive mental health treatment
Excessive and non-excessive GAD cases have very similar socio-demographic profiles. (Table 3) Both are significantly more prevalent among females than males, less prevalent among Non-Hispanic Blacks and Hispanics than other respondents, positively related to years of education, more prevalent among the unemployed than the employed, more prevalent among the previously married than the married, and more prevalent among the married than the never married. Among individuals with broadly-defined GAD, excessiveness is not significantly related to sex, race, marital status, educational attainment, or employment status. Excessiveness is related, though, to age, consistent with the age-of-onset differences reported earlier. Importantly, non-excessive cases are far more similar to excessive cases (significantly different on only 1 of 6 socio-demographic variables) than to non-GAD cases (significantly different on 5 of 6 socio-demographic variables).
The association of parental GAD with respondent GAD does not vary as a function of excessiveness. Having at least one parent with a history of GAD is associated with elevated risk for both excessive (OR = 4.5) and non-excessive (OR = 4.1) GAD, with no significant difference in the ORs across the two subgroups (χ12 = 0.1, p = .744).
Our interpretation of the results reported here must be tempered by the recognition of several study limitations, including the use of fully-structured lay interviews, the assessment of excessive worry using a single interview question, and the retrospective reporting of lifetime symptoms and course of disorder. These aspects of the study may have reduced our ability to detect reliable differences between excessive and non-excessive subgroups. However, diagnostic classifications of GAD based on the fully-structured interview used here have good individual-level concordance with independent clinician diagnoses based on blinded SCID reappraisal interviews (AUC = .83) and do not differ significantly from clinical diagnoses in estimated prevalence (McNemar χ12 = 1.7, p = .192) (Kessler et al., 2004a), increasing confidence in the assessment of GAD and, more specifically, in the assessment of excessiveness. It is less clear how the experience of excessive worry may have affected recollections or reports of past anxiety experiences. Additional research will be needed, ideally with a prospective design and more detailed assessment of excessiveness appraisals, to further clarify the relation of excessiveness to the characteristics and correlates of GAD.
With these limitations in mind, we found several differences between excessive and non-excessive worriers who met all other DSM-IV criteria for GAD. Results revealed that excessive worry begins earlier in life, has a more persistent course and greater comorbidity, and is associated with greater symptom severity. At the same time, non-excessive cases have a severity distribution not markedly different from that of excessive cases, significant comorbidity with the vast majority of the other DSM-IV disorders assessed in the survey, and similar odds of parental GAD as excessive cases. In addition, the socio-demographic profile of non-excessive cases is quite similar to that of excessive cases. Finally, non-excessive cases with episodes in the past 12 months reported disability comparable to that of excessive cases. Taken together, these results suggest that non-excessive worry is associated with a somewhat milder symptom presentation than excessive worry but is still sufficiently severe and impairing to warrant a diagnosis of GAD.
The obvious implication of this conclusion is that individuals need not view their worry as excessive to experience clinically significant impairment, at least when worry co-occurs with the other symptoms of GAD. This is consistent with the suggestion in the DSM-IV text that worry can be inferred to be excessive, even if excessiveness is denied by the worrier, so long as other features of pathological worry (uncontrollability, distress, functional impairment) are evident (American Psychiatric Association, 1994, p. 433). It has been observed that no other DSM disorder, including other disorders involving severe levels of normal emotional processes (e.g., depressed mood in major depressive disorder), requires an explicit judgment of excessiveness as part of its primary diagnostic criteria (Rickels & Rynn, 2001). The discovery of a group of significantly impaired worriers who are excluded from diagnosis by the excessiveness criterion challenges its appropriateness for GAD as well.
One might ask, given its limited conceptual and evidentiary base, why the excessiveness criterion was originally adopted in DSM-III-R. One reason may have been the desire to avoid pathologizing normative reactions to stressful life events. Because worry is experienced to some degree by most psychologically healthy individuals (Borkovec, 1994; Muris et al., 1998), especially in times of stress, the GAD diagnostic criteria must be able to distinguish normal, transient stress reactions from clinically significant anxiety. What is more contentious, however, is how the criteria should classify severe, persistent anxiety that is experienced in response to severe or chronic stressors. Although such anxiety may not be considered excessive, and so would not merit a GAD diagnosis by current DSM criteria, it could still cause considerable suffering and impairment that might benefit from treatment, perhaps especially treatments that focus on enhancing effective coping. For example, Ballenger et al. (2001) noted that the GAD syndrome is common among patients with chronic physical conditions, but that physicians are often reluctant to diagnose GAD because they consider the presence of anxiety to be normal and justified by the physical illness. Yet this co-occurring GAD often worsens prognosis for the physical illness, which means that failure to detect, diagnose, and treat the GAD could lead to substantially poorer health outcomes for such patients. The considerable symptom severity, psychiatric comorbidity, and functional impairment reported by generally anxious individuals in our sample—even when the anxiety was judged not to be excessive—similarly underscore the potential clinical and public health importance of diagnosing and treating these anxious individuals.
Despite their many similarities, excessive and non-excessive GAD cases were distinguished by several robust differences that remained significant even after worry uncontrollability, distress, and impairment were controlled. These findings raise at least two intriguing questions for future investigation. First, might excessive GAD, with its earlier onset and more chronic, comorbid course, represent a different form of the disorder than non-excessive GAD? It has been suggested that earlier-onset GAD represents a more severe disorder stemming from temperament factors or from extreme early stressors that predispose the individual to a range of emotional disorders, whereas late-onset GAD represents a more circumscribed, less characterological condition precipitated by moderate life stress (Brown et al., 1994; Campbell et al., 2003). It is possible that the concepts of excessive GAD and early-onset GAD converge on an overlapping set of individuals who are especially vulnerable to developing severe, chronic emotional disturbance (see Hudson & Rapee, 2004). At the same time, our finding that excessive GAD is no more familial than non-excessive GAD suggests that the relations among the relevant vulnerability factors may be quite complex. Additional research is needed to determine whether there are distinct variants of GAD for which excessive worry may be a marker.
Second, what leads individuals to describe their worries as excessive? Our finding that excessive GAD is associated with greater symptom severity could indicate that excessive cases actually experience more frequent or intense worry than non-excessive cases. An alternative explanation is that the worry experiences of the two groups are objectively the same, but that the former appraise the experiences in a more negative light than the latter. A growing body of research suggests that severe worriers with and without GAD are distinguished more by their subjective interpretations of worry and anxiety than by the actual frequency, severity, or disruptiveness of their anxiety experiences (Ruscio & Borkovec, 2004; Ruscio et al., 2003). There is a need to examine the correspondence of excessiveness appraisals to actual anxiety symptoms, to identify factors other than anxiety symptoms that may lead worry to be appraised as excessive, and to consider the relative weight that should be given to anxiety symptoms versus appraisals of these symptoms when GAD is diagnosed.
Despite group differences, the bulk of the present findings argue for eliminating the excessiveness requirement from the GAD criterion set. Removing this vaguely worded, poorly defined requirement could have the added benefit of enhancing the reliability of the GAD diagnosis, which—despite significant improvements in DSM-IV—continues to be the lowest of any principal anxiety disorder (Brown et al., 2001). Nevertheless, decisions about excessiveness will need to be considered alongside other modifications to the GAD criteria proposed for DSM-V and ICD-11, especially others that would result in a more liberal diagnostic threshold (e.g., Kessler et al., in press-b), to ensure that the revised diagnostic criteria do not substantially increase the number of false-positive diagnoses. If the decision ultimately is made to retain the excessiveness criterion, research might evaluate whether diagnostic reliability and validity could be improved by replacing the ambiguous wording of “excessive worry” with a more explicit descriptor (e.g., intense, frequent) or with the more narrow specifier of worry about minor matters from which the criterion was originally derived (e.g., Craske et al., 1989). Whether excessiveness is discarded or retained, there will be a need for additional research into the features that distinguish pathological worry and anxiety from normal variants of these processes (Borkovec et al., 1991; Weems et al., 2000), leading to further refinement of the diagnostic criteria and a more valid and useful GAD diagnosis.
The National Comorbidity Survey Replication (NCS-R) is supported by NIMH (U01-MH60220) with supplemental support from NIDA, the Substance Abuse and Mental Health Services Administration (SAMHSA), the Robert Wood Johnson Foundation (RWJF; Grant 044708), and the John W. Alden Trust. Collaborating investigators include Ronald C. Kessler (Principal Investigator, Harvard Medical School), Kathleen Merikangas (Co-Principal Investigator, NIMH), James Anthony (Michigan State University), William Eaton (The Johns Hopkins University), Meyer Glantz (NIDA), Doreen Koretz (Harvard University), Jane McLeod (Indiana University), Mark Olfson (Columbia University College of Physicians and Surgeons), Harold Pincus (University of Pittsburgh), Greg Simon (Group Health Cooperative), Michael Von Korff (Group Health Cooperative), Philip Wang (Harvard Medical School), Kenneth Wells (UCLA), Elaine Wethington (Cornell University), and Hans-Ulrich Wittchen (Max Planck Institute of Psychiatry). The views and opinions expressed in this report are those of the authors and should not be construed to represent the views of any of the sponsoring organizations, agencies, or U.S. Government. A complete list of NCS publications and the full text of all NCS-R instruments can be found at http://www.hcp.med.harvard.edu/ncs. Send correspondence to NCS/at/hcp.med.harvard.edu. The NCS-R is carried out in conjunction with the World Health Organization World Mental Health (WMH) Survey Initiative. We thank the staff of the WMH Data Collection and Data Analysis Coordination Centres for assistance with instrumentation, fieldwork, and consultation on data analysis. A complete list of WMH publications and instruments can be found at (http://www.hcp.med.harvard.edu/wmhcidi).