agorphobia, simple phobia, and social phobia in the national comorbidity survey

10
Agoraphobia, Simple Phobia, and Social Phobia in the National Comorbidity Survey William J. Magee, PhD; William W. Eaton, PhD; Hans-Ulrich Wittchen, PhD; Katherine A. McGonagle, PhD; Ronald C. Kessler, PhD Background: Data are presented on the general popu- lation prevalences, correlates, comorbidities, and im- pairments associated with DSM-III-R phobias. Methods: Analysis is based on the National Comorbid- ity Survey. Phobias were assessed with a revised version of the Composite International Diagnostic Interview. Results: Lifetime (and 30-day) prevalence estimates are 6.7% (and 2.3%) for agoraphobia, 11.3% (and 5.5%) for simple phobia, and 13.3% (and 4.5%) for social phobia. Increasing lifetime prevalences are found in recent co- horts. Earlier median ages at illness onset are found for simple (15 years of age) and social (16 years of age) pho- bias than for agoraphobia (29 years of age). Phobias are highly comorbid. Most comorbid simple and social pho- bias are temporally primary, while most comorbid agora- phobia is temporally secondary. Comorbid phobias are gen- erally more severe than pure phobias. Despite evidence of role impairment in phobia, only a minority of individuals with phobia ever seek professional treatment. Conclusions: Phobias are common, increasingly preva- lent, often associated with serious role impairment, and usually go untreated. Focused research is needed to in- vestigate barriers to help seeking. (Arch Gen Psychiatry. 1996;53:159-168) THIS ARTICLE presents nation¬ ally representative data on the general population de¬ scriptive epidemiology of DSM-ÍÍÍ-R agoraphobia (with or without panic), social phobia, and simple phobia from the National Comor¬ bidity Survey (NCS).1 The focus is on prevalence, sociodemographic corre¬ lates, comorbidity, impairment, and pro¬ fessional help seeking. RESULTS PREVALENCE Lifetime and 30-day prevalence esti¬ mates are presented in Table I. As re¬ ported previously,1 lifetime prevalence es¬ timates are 6.7% for agoraphobia, 11.3% for simple phobia, and 13.3% for social phobia. The prevalence estimates for ago¬ raphobia and simple phobia are over twice as high among women as among men, with female-male prevalence ratios of 2.2:1.0 for agoraphobia (z = 5.7, P-C05) and 2.3:1.0 for simple phobia (z = 7.4, P<.05). The female-male prevalence ratio for social phobia is considerably lower (1.4:1.0), al- though still statistically significant (z = 3.4, P<.05). Lifetime prevalence estimates gener¬ ally decline with age, as shown in Table 1, suggesting the possible existence of co¬ hort effects. Analysis of age-at-onset curves, shown in Figure 1, Figure 2, and Figure 3, documents that these cohort differences are significant for all three pho¬ bias and most pronounced for agorapho¬ bia. Despite the apparent sex difference in the age effect in Table 1, none of the co¬ hort effects in the Figures 1 through 3 dif¬ fers significantly by sex. Intercohort dif¬ ferences emerge quite early in life for agoraphobia and, while most pro¬ nounced for the difference between the most recent (ages 15 to 24 years at inter¬ view) vs earlier (ages 25 to 54 years at in¬ terview) cohorts, are also shown in Fig¬ ure 1 to be statistically significant within the three later cohorts. Intercohort differ¬ ences in simple and social phobias are con- From the Department of Psychiatry, The University of Wisconsin, Madison (Dr Magee); the Department of Mental Hygiene, The Johns Hopkins University, Baltimore, Md (Dr Eaton); the Department of Clinical Psychology, The Max Planck Institute of Psychiatry\p=m-\Clinical Institute, Munich, Germany (Dr Wittchen); and the Institute for Social Research (Drs McGonagle and Kessler) and the Department of Sociology (Dr Kessler), The University of Michigan, Ann Arbor.

Upload: muhammad-zaka

Post on 11-Apr-2015

41 views

Category:

Documents


3 download

DESCRIPTION

paper

TRANSCRIPT

Page 1: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

Agoraphobia, Simple Phobia, and Social Phobiain the National Comorbidity SurveyWilliam J. Magee, PhD; William W. Eaton, PhD; Hans-Ulrich Wittchen, PhD;Katherine A. McGonagle, PhD; Ronald C. Kessler, PhD

Background: Data are presented on the general popu-lation prevalences, correlates, comorbidities, and im-pairments associated with DSM-III-R phobias.Methods: Analysis is based on the National Comorbid-ity Survey. Phobias were assessed with a revised versionof the Composite International Diagnostic Interview.

Results: Lifetime (and 30-day) prevalence estimates are

6.7% (and 2.3%) for agoraphobia, 11.3% (and 5.5%) forsimple phobia, and 13.3% (and 4.5%) for social phobia.Increasing lifetime prevalences are found in recent co-

horts. Earlier median ages at illness onset are found forsimple (15 years of age) and social (16 years of age) pho-

bias than for agoraphobia (29 years of age). Phobias are

highly comorbid. Most comorbid simple and social pho-bias are temporally primary, while most comorbid agora-phobia is temporally secondary. Comorbid phobias are gen-erally more severe than pure phobias. Despite evidence ofrole impairment in phobia, only a minority of individualswith phobia ever seek professional treatment.

Conclusions: Phobias are common, increasingly preva-lent, often associated with serious role impairment, andusually go untreated. Focused research is needed to in-vestigate barriers to help seeking.(Arch Gen Psychiatry. 1996;53:159-168)

THIS ARTICLE presents nation¬ally representative data on

the general population de¬scriptive epidemiology ofDSM-ÍÍÍ-R agoraphobia

(with or without panic), social phobia, andsimple phobia from the National Comor¬bidity Survey (NCS).1 The focus is on

prevalence, sociodemographic corre¬

lates, comorbidity, impairment, and pro¬fessional help seeking.

RESULTS

PREVALENCE

Lifetime and 30-day prevalence esti¬mates are presented in Table I. As re¬

ported previously,1 lifetime prevalence es¬

timates are 6.7% for agoraphobia, 11.3%for simple phobia, and 13.3% for socialphobia. The prevalence estimates for ago¬raphobia and simple phobia are over twiceas high among women as among men, withfemale-male prevalence ratios of 2.2:1.0 foragoraphobia (z = 5.7, P-C05) and 2.3:1.0for simple phobia (z = 7.4, P<.05). Thefemale-male prevalence ratio for socialphobia is considerably lower (1.4:1.0), al-

though still statistically significant (z = 3.4,P<.05).

Lifetime prevalence estimates gener¬ally decline with age, as shown in Table1, suggesting the possible existence of co¬

hort effects. Analysis ofage-at-onset curves,shown in Figure 1, Figure 2, andFigure 3, documents that these cohortdifferences are significant for all three pho¬bias and most pronounced for agorapho¬bia. Despite the apparent sex difference inthe age effect in Table 1, none of the co¬

hort effects in the Figures 1 through 3 dif¬fers significantly by sex. Intercohort dif¬ferences emerge quite early in life foragoraphobia and, while most pro¬nounced for the difference between themost recent (ages 15 to 24 years at inter¬view) vs earlier (ages 25 to 54 years at in¬terview) cohorts, are also shown in Fig¬ure 1 to be statistically significant withinthe three later cohorts. Intercohort differ¬ences in simple and social phobias are con-

From the Department ofPsychiatry, The Universityof Wisconsin, Madison(Dr Magee); the Departmentof Mental Hygiene, The JohnsHopkins University, Baltimore,Md (Dr Eaton); theDepartment of ClinicalPsychology, The Max PlanckInstitute of Psychiatry\p=m-\ClinicalInstitute, Munich, Germany(Dr Wittchen); and theInstitute for Social Research(Drs McGonagle and Kessler)and the Department ofSociology (Dr Kessler), TheUniversity of Michigan,Ann Arbor.

Page 2: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

METHODS

SAMPLE

As described in more detail in previous articles publishedin the Archives,1"' the NCS was administered to a strati¬fied, multistage area probability sample of 8098 respon¬dents (age range, 15 to 54 years) selected from the nonin-stitutionalized household population of the coterminousUnited States and to a supplemental sample of students incampus group housing. The survey was fielded between Sep¬tember 1990 and March 1992. The response rate was 82.4%.The data were weighted for differential probabilities of se¬lection and differential nonresponse and to adjust the sampleto approximate the cross-classification of the population dis¬tribution on a range of sociodemographic characteristics.These weights are described in more detail elsewhere. '

DIAGNOSTIC ASSESSMENT

The NCS diagnoses are based on a modified version of theWorld Health Organization's Composite International Di¬agnostic Interview (WHO-CIDI),4 a structured interviewdesigned to be administered by trained interviewers whoare not clinicians. Phobias were assessed by asking respon¬dents whether there was ever a time in their life when anyof 19 potentially phobic situations always made them soafraid that they either tried to avoid it or felt very uncom¬fortable in the situation. These situational descriptors were

presented in three lists (five situations for agoraphobia, eightfor simple phobia, and six for social phobia). The mainchange in the version of the CIDI developed at The Uni¬versity of Michigan, Ann Arbor, for the NCS (UM-CIDI)was that respondents were asked to review these lists vi¬sually as the interviewer read the stem questions in an ef¬fort to focus memory search and reduce chances of induc¬ing the "no" response set that has been documented to occurwhen "yes-no" questions are presented in lists of this sort.5

Respondents who endorsed one or more situationaldescriptors were then administered the structured CIDIquestions that assessed DSM-ÍII-R criteria for agoraphobia

(with or without panic), simple phobia, or social phobia.Short-term test-retest reliabilities of the diagnoses gener¬ated from these structured questions were assessed in theWHO-CIDI Field Trials. Values of were .68 for agora¬phobia, .59 for simple phobia, and .64 for social phobia.6Although disorder-specific validity estimates have not beenreported, a comparison of CIDI diagnoses and clinical di¬agnoses made by one of two psychiatrists in the CIDI FieldTrials documented good concordance ( = .76) for the anxi¬ety and phobic disorders as a whole.6 The psychiatrists inthis concordance study were either observers of the CIDIinterviews or, on a small number of occasions, adminis¬tered the CIDI. The psychiatrists were then allowed to askwhatever additional questions they wished to complete aDSM-ÍÍÍ-R criteria checklist as the basis for clinical diag-

ANALYSIS PROCEDURES

Estimates of disorder prevalences, professional help seek¬ing, and proportions of cases with impairment were ob¬tained by calculating means for dichotomous outcomes us¬

ing the PSRATIO program in the OSIRIS software package.7Age-at-onset curves were calculated using the SURVIVALprocedure in the SPSS software package.8 Cross-sectionalestimates of sociodemographic correlates and bivariate co-morbidities were obtained by estimating odds ratios (ORs)based on logistic regression models with dichotomous pre¬dictor variables using the LOGISTIC procedure in the SASsoftware package.9

Owing to the complex sample design and weightingof the NCS, special analysis procedures were required toobtain unbiased estimates of SEs of parameter estimates.The Taylor series linearization method10 was used to ad¬just SEMs. The method of Balanced Repeated Replica¬tions,11 operationalized in a SAS MACRO,9 was used to ad¬just SEs of ORs. Odds ratios for bivariate models are reportedbelow, with 95% confidence intervals (CIs) adjusted for de¬sign effects (ie, the effects of weighting and clustering ofobservations). All statistical tests were evaluated at the .05level of significance, using two-tailed tests and design-based SEs.

fined to cases with disease onsets after age 16 years. Whilethe cohort effect for social phobia in Figure 3 is largelyowing to relatively early age at onset among the most re¬

cent cohort, intercohort differences are statistically sig¬nificant both between the most recent and earlier co¬

horts and within the earlier cohorts. It is also noteworthythat simple and social phobias generally have earlier es¬timated ages at onset than agoraphobia, based on retro¬spective self-reports. Median ages at onset in the oldestcohort, where years of risk are greatest, are 15 years forsimple phobia, 16 years for social phobia, and 29 yearsfor agoraphobia.

Thirty-day prevalence estimates are 2.3% for agora¬phobia, 5.5% for simple phobia, and 4.5% for social pho¬bia. The estimates for agoraphobia and simple phobia are

substantially higher among women than among men, withprevalence ratios of 2.2:1.0 for agoraphobia (z = 3.4, P<.05)and 3.8:1.0 for simple phobia (z = 7.2, P<.05). The female-male prevalence ratio for social phobia is much smaller

(1.4:1.0), but still statistically significant (2=2.2, P<.05).Age differences (which could also be interpreted as co¬

hort differences) are generally not significant, althoughthere is a fairly consistent trend for 30-day prevalence es¬timates to be highest in the youngest cohort.

SOCIODEMOGRAPHIC CORRELATES

Bivariate associations between sociodemographic vari¬ables and 30-day phobias are presented in Table 2 as a

way of providing descriptive information on the socialdistribution of phobias. Thirty-day agoraphobia is nega¬tively related to income and education and is signifi¬cantly elevated among blacks; homemakers; persons whoare neither working, nor students, nor homemakers; andthose who live with someone other than a spouse. Thirty-day simple phobia is negatively related to education butnot income, and it is significantly elevated among His-panics, all persons who are not employed (including stu-

Page 3: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

Ages 15-24 y at Interview. Born 1966-1975

Ages 35-44 y at Interview, Born 1946-1955

Ages 25-34 y at Interview, Born 1956-1965

Ages 45-54 y at Interview, Born 1936-1945

2(3)=59.1, <.05 2 (2)=5.2, P<.05 for the Most Recent vs Earlier Cohorts 2 (2)=23.3, P<.05 Excluding the Most Recent Cohort

m-f- | | | | | | |-1-1-1-1-1-1-1-1-1-1-1-1-1-1-1-10 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54

at Onset, y

Figure 1. Cumulative lifetime prevalence of agoraphobia by cohort. Numbers in parentheses indicate degrees of freedom.

dents, homemakers, and others), and those who live withtheir parents. Thirty-day social phobia is negatively re¬lated to education and income and significantly el¬evated among the never married; students; persons whoare neither working, nor students, nor homemakers; and

those who live with their parents. None of the phobiasis meaningfully associated with either region of the coun¬

try or urbanicity. Multivariate relationships were also ex¬amined and found to be quite similar to those reportedin Table 2. (The results of these and other analyses that

Page 4: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

Ages 15-24 y at Interview, Born 1966-1975

Ages 35-44 y at Interview, Born 1946-1955

Ages 25-34 y at Interview, Born 1956-1965

Ages 45-54 y at Interview, Born 1936-1945

£ .10-

.08

04

02

2 (3)=5.8, NS, for Total Sample 2(3)=1.2, NS, for Ages 0-15 y 2(3)=29.3, P<.05,forAges16+y 2 (1 )=0.7, NS, for the Most Recent vs Earlier Cohorts, for Ages 16+ y 2 (2)=13.6. P<.05, Excluding the Most Recent Cohort, tor Ages 16+ y

12 14 16 22 34 36 38 42 44~~r~

46 50~1~

52 54

Age at Onset, y

Figure 2. Cumulative lifetime prevalence of simple phobia by cohort. NS indicates not significant; numbers in parentheses, degrees of freedom.

Ages 15-24 y at Interview, Born 1966-1975

A Ages 35-44 y at Interview, Born 1946-1955

Ages 25-34 y at Interview, Born 1956-1965

Ages 45-54 y at Interview, Born 1936-1945

E<3 ·'

.02

2 (3)=21.8. P<.05, for Total Sample 2 (3)=5.8, NS, for Ages 0-15 y 2(3)=12.6, P<.05, for Ages 16+y 2 (1 )=6.7, P<.05, for the Most Recent Cohort vs Earlier Cohorts, for Ages 16+ y 2 (2)=9.9, P<.05, Excluding the Most Recent Cohort, for Ages 16+ y

18 20"130

—I32

~134

~138

~142 46 50

"I-52

Age at Onset, y

Figure 3. Cumulative lifetime prevalence of social phobia by cohort. NS indicates not significant; numbers in parentheses, degrees of freedom.

Page 5: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

*0R indicates odds ratio; CI, confidence interval.-[Significant at the .05 level, two-tailed test.{students excluded from education and income analyses.

are described in the text without the presentation of re¬

sults can be obtained from the public-access Internet filedescribed in the acknowledgments.) Correlates of 30-day prevalence among lifetime cases controlling for ageat onset and time since onset were also examined andfound to be largely nonsignificant, suggesting indirectlythat the persistence of phobias is unrelated to these so-

ciodemographic variables.

COMORBIDITY

As shown in Table 3, the vast majority of persons withphobia (87.6% of persons with agoraphobia, 83.4% of per¬sons with simple phobia, and 81.0% of persons with so¬

cial phobia) reported at least one other lifetime DSM-ÍÍÍ-R/NCS disorder. These proportions are significantly higherthan those found among the subsample of NCS respon¬dents who reported never having a phobia, with bivari¬ate ORs of 6.8 (agoraphobia), 5.3 (simple phobia), and

4.8 (social phobia). Disorder-specific ORs show that life¬time phobias are strongly comorbid with each other (ORrange, 7.1 to 8.7), with other anxiety disorders (OR range,2.7 to 11.9), and with affective disorders (OR range, 3.1to 10.0). Phobias are more weakly comorbid with alco¬hol and drug dependence (OR range, 1.7 to 2.9), and theyare not significantly comorbid with either alcohol or drugabuse without dependence (OR range, 0.9 to 1.2). Thesingle largest OR in the data is between panic disorderand agoraphobia (11.9), a result observed because ago¬raphobia with panic and agoraphobia without panic were

not distinguished in defining agoraphobia. There is alsoa very large OR between mania and simple phobia ( 10.0).It is noteworthy that only a minority (35.8%) of the re¬

spondents diagnosed in the UM-CIDI as having lifetimeagoraphobia reported ever having an unexpected panicattack (including 21.6% with panic disorder and an ad¬ditional 14.6% with panic attacks but not panic disor¬der) despite the high panic-agoraphobia OR.

Page 6: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

*NSC indicates National Comorbidity Survey; ORs, odds ratios; CI, confidence interval; and PTSD, posttraumatic stress disorder. [Significant at the .05 level, two-tailed test. With or without panic disorder.^Excluding the relevant phobia(s).

As shown in Table 4, comorbid agoraphobia is usu¬

ally secondary (55.4% of comorbid cases) in the senseof agoraphobia onset occurring at a later age than at leastone other comorbid condition. Comorbid simple pho¬bia and social phobia, in comparison, are usually tem¬porally primary (56.8% for simple phobia and 52.6% forsocial phobia). These results are consistent with the ear¬

lier observation that the retrospectively reported me¬dian ages at onset of simple and social phobias are dur¬ing the adolescent years (15 to 16 years) compared witha median age at onset of 29 years for agoraphobia.

IMPAIRMENT AND PROFESSIONALHELP SEEKING

The proportions of persons with lifetime phobia who per¬ceived themselves to ever have significant role impairment(as operationalized by a report that the phobia interfered"a lot," rather than "some," "a little," or "not at all," withtheir life and activities), who sought professional help, or

who took medication more than once for their phobias are

presented in Table 5. All three of these severity indica¬tors are consistently more prevalent among comorbid thanpure cases and, within the subsample of comorbid cases,more prevalent among those with a lifetime history ofpanic.

Perceived role impairment is significantly less preva¬lent overall among persons with agoraphobia (26.5%) thanamong persons with either simple (34.2%, = 2.4, P<.05)

or social (33.5%, = 2.4, P<.05) phobias. However, thereis no significant difference across types of phobia in theprevalence of perceived role impairment in the sub-sample of those with panic attacks. A higher proportionof persons with agoraphobia (41.0%) sought profes¬sional help for their phobias at some time in their life thanthose with either simple (30.2%, = 3.1, P< .05) or social(19.0%, z = 4, P<.05) phobias. Furthermore, a higher pro¬portion of persons with agoraphobia reported lifetime useof medications for their phobias (21.6%) than those witheither simple (8.0%, = 5.9, P.05) or social (6.2%, = 7.5,P<.05) phobias. Overall, approximately half of personswith agoraphobia, simple phobia, and social phobia re¬

ported at least one outcome indicative of severity at some

time in their life (either a lot of interference, professionalhelp seeking, or use of medications more than once).

The opposite-sign pattern seen in Table 5, ie, a lowerproportion ofpersons with agoraphobia who reported im¬pairment but higher proportions of those who reportedhelp seeking and use of medication, implies that the re¬

lationship between perceived impairment and treat¬ment differs by type of phobia. This is examined di¬rectly in Table 6, where it is shown that the probabilitiesof professional help seeking and use of medication amongthe persons with agoraphobia at a given level of per¬ceived role impairment are generally higher than amongthe persons with simple or social phobias who reportedthe same levels of role impairment.

Page 7: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

* This phobia is the respondent's only lifetime National ComorbiditySurvey (NCS) DSM-III-R disorder.

\This phobia was at an earlier age of onset (based on retrospectiveself-reports) than any other disorder and was followed by the onset of atleast one other «CS/DSM-III-R disorder.

$This phobia and at least one other WCS/DSM-III-R disorder were at thesame ages of onset, which were earlier than those of any other disorder.

§At least one other /CS/DSM-III-R disorder occurred at an earlier agethan this phobia.

Given that the finding of less role impairment amongpersons with agoraphobia than among those with otherphobias is inconsistent with prior research,1217 an analy¬sis of more objective role impairment was conducted. Thiswas done by estimating logistic regression equations, inwhich measures of30-day agoraphobia, simple phobia, andsocial phobia were used to predict various measures of cur¬

rent role impairment, controlling for sociodemographicvariables (age, sex, education, marital status, number ofpreschool-aged children in the home, and interactions be¬tween sex and marital status and between sex and pre¬schoolers), and comorbid disorders. The analyses showedthat agoraphobia and simple phobia, but not social pho¬bia, are associated with significantly increased work ab¬sence among employed men (equivalent to 1.1 days permonth for persons with agoraphobia and 0.7 day per monthfor persons with simple phobias [z = 6.7 and 6.1, respec¬tively] , compared with a nonsignificant 0.05 days permonth for persons with social phobia [z = 0.1]) but not

among employed women. In addition, simple and socialphobias, but not agoraphobia, are associated with low so¬

cial support (defined as the lowest decile in the sample),with ORs of 1.92 (95% CI, 1.11 to 3.35) for simple pho¬bia and 1.74 (95% CI, 1.03 to 2.95) for social phobia, com¬

pared with a nonsignificant OR of 1.08 (95% Cl, 0.23 to5.04) for agoraphobia. None of the phobias was signifi¬cantly related to financial adversity, as defined by eitherlow family income in the total sample or low earnings inthe subsample of the employed.

COMMENT

PREVALENCE

Previous studies of phobias in North American commu¬

nity samples have largely been based on DSM-ÍÍÍ crite-

ria, as defined in the Diagnostic Interview Schedule.18 TheNCS agoraphobia lifetime prevalence estimate of 6.7%is at the upper end of the range of these earlier surveys(2.9% to 6.9%),19M while the NCS simple phobia life¬time prevalence estimate of 11.3% is in the middle of therange in earlier surveys (6.5% to 21.6%).21 The NCS so¬

cial phobia lifetime prevalence estimate of 13.3%, in com¬

parison, is far above the range in earlier surveys (1.6%to 3.2%),20·22 although lower than in a recent Europeancommunity epidemiologie survey (16.0%) that used DSM-ÍÍÍ-R criteria and made diagnoses with the CIDI.23 Fur¬thermore, a recent epidemiologie survey of a Canadiancommunity, based on DSM-ÍÍÍ-R criteria, reported a pe¬riod prevalence of social phobia (7.1%)24 that is quite simi¬lar to the NCS 12-month prevalence estimate of 7.9%.]

A blind clinical reappraisal of a subsample of NCSrespondents who endorsed the stem question for socialphobia found that 100% of CIDI cases were confirmedas cases and 84% of noncases were confirmed as non-

cases,25 documenting that the higher prevalence in theNCS than in earlier surveys is not owing to invalidity.But how can this high prevalence be explained? At leastthree factors appear to be involved.

First, the phobia stem questions in the NCS are basedon DSM-ÍÍÍ-R, which requires either avoidance or dis¬tress on exposure, rather than the narrower DSM-ÍÍI re¬

quirement of avoidance. This has a much larger effect on

the prevalence of social phobia than other phobias, be¬cause it is much more difficult to avoid social situationsthan the situations associated with simple phobia or ago¬raphobia.

Second, while the NCS descriptors for simple pho¬bia and agoraphobia were similar in number and word-

Page 8: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

* Seeking help for the fear from a nonpsychiatric medical doctor, amental health specialist, or any other professional (eg, priest, minister,rabbi, counselor, or nurse).

t Taking medication more than once because of the fear.{.Percentage of subjects with lifetime agoraphobia who sought any helpin the subsample of those who reported "a lot" of role impairment due tothe phobia.

ing to those used in the Epidemiologie Catchment Area(ECA) there was an expansion from three social phobiadescriptors in the earlier surveys to six in the NCS. TheECA asked about low-prevalence social-phobic fears: ofeating in public (endorsed by 2.7% of ECA respondentsand 2.7% of NCS respondents), of speaking to new ac¬

quaintances (endorsed by 4.7% of ECA respondents andnot included in the NCS), and of speaking in front of a

small group of people you know (endorsed by 6.5% ofECA respondents). The third of these questions was modi¬fied in tbe NCS to delete the words "you know," with anincrease in the prevalence of endorsement to 14.6%. Thenew questions added to the NCS all tapped high-prevalence social-phobic fears: of having to use the toi¬let when away from home (endorsed by 6.3%), of writ¬ing when someone watches (endorsed by 6.1%), of talkingto people and not having anything to say or sounding fool¬ish (endorsed by 13.5%), and of speaking in public (en¬dorsed by 29.1%).

Third, the estimated prevalence of social phobia ismore sensitive than those of simple phobia or agorapho¬bia to changes in diagnostic criteria, as documented bythe fact that the prevalence estimates for agoraphobia andsimple phobia did not change very much when the di¬agnostic criteria were changed from DSM-ÍÍÍ-R to DSM-111 (from 6.7% to 5.5% for agoraphobia and from 11.3%to 10.9% for simple phobia). The prevalence estimate forsocial phobia, in comparison, changes substantially, from13.3% to 8.3%, with this change in diagnostic criteria.When diagnosis was further restricted to the NCS re¬

spondents who endorsed one of the situational descrip¬tors most comparable to those used in the Diagnostic In¬terview Schedule-DSM-III surveys, the lifetime socialphobia prevalence dropped to 4.8%, which is close to theestimates in the earlier surveys.

AGE AT ONSET

A limitation of the age-at-onset analysis is that the NCSasked respondents to date their age at first onset of theirunrealistic fear, not the age when they first developed a

full phobic syndrome. Nonetheless, the estimated me¬dian ages at onset of 15 years for simple phobia, 16 yearsfor social phobia, and 29 years for agoraphobia are quitesimilar to those obtained by Ost26 in a review of clinicalstudies (between 7 and 20 years for separate simple pho¬bias, 16 years for social phobia, and 28 years for agora¬phobia).

COHORT EFFECTS

The significant intercohort difference in lifetime phobiaprevalence was found to be more pronounced for ago¬raphobia than for the other phobias and only to appearfor later-onset cases of simple and social phobias. Thesespecifications could explain why Burke et al,27 who ex¬

amined cohort differences in onset distributions of totalphobia in the ECA (not distinguishing among agorapho¬bia, simple phobia, and social phobia) up to age 29 years,failed to document consistent evidence for increasingprevalences in more recent cohorts.

Although it is conceivable that evidence for increas¬ing prevalences of phobias in more recent cohorts couldbe owing to a methodological artifact involving eithergreater recall failure among older respondents or greaterexclusion from the sample of older phobies, neither ofthese possibilities can explain the fact that the signifi¬cant cohort differences in lifetime simple and social pho¬bias are confined to later-onset cases. This specificationraises the possibility that there is heterogeneity in thecauses of simple and social phobias depending on age atonset. More precise assessment of this possibility, how¬ever, will require analysis of longitudinal data to distin¬guish a true cohort effect from a methodological arti¬fact.

SOCIODEMOGRAPHIC CORRELATES

The finding that agoraphobia and simple phobia are ap¬proximately twice as prevalent among women as amongmen while there is a much smaller sex difference in so¬cial phobia is consistent with the results of several pre¬vious studies.19·2028"32 The strong and negative associa¬tion of at least one indicator of socioeconomic status(education or income) with all three phobias is also con¬sistent with previous research.33 34 The inconsistency ofthe associations between phobias and other sociodemo-graphic correlates mirrors inconsistencies found in ear¬lier epidemiologie studies.14·20·21·23·26"28·35·36

COMORBIDITY

The NCS finding of significant comorbidities betweenphobias and other disorders are consistent with prior clini¬cal studies37"39 and community epidemiologie sur¬

veys.1222·40 The strongest of the NCS comorbidities is thatbetween panic and agoraphobia (OR = 11.9), a result con¬sistent with prior research41"43 and owing to the fact that

Page 9: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

we did not distinguish agoraphobia with panic from ago¬raphobia without panic in the definition of agorapho¬bia. Despite this very high OR, however, only about onethird of NCS agoraphobic respondents reported ever hav¬ing an unexpected panic attack, a result also found in theECA Study.44 Subsequent clinical reanalysis of ECA cases

found that many of those classified as agoraphobic with¬out panic had, in fact, misdiagnosed simple or social pho¬bias,43 a result that is consistent with the fact that ago¬raphobia without a history of panic is very rare in clinicalsamples.46,47 More detailed analysis of the NCS data is cur¬

rently under way to discriminate persons with agora¬phobia who deny ever having simple or social phobic fearsfrom those with such fears (who could have misdiag¬nosed simple or social phobias) in an effort to evaluatethe possibility that a similar diagnostic confusion under¬lies the appearance that agoraphobia without panic is morecommon than agoraphobia with panic in the NCS.

Another important result concerning comorbidity isthat all the phobias were found to be more weakly comor¬bid with substance use disorders than with other anxietydisorders or affective disorders. This result is consistent witha comprehensive review of the literature by Schuckit andHesselbrock, who concluded that "the available data, whileimperfect, do not prove a close relationship between life¬long anxiety disorders and alcohol dependence."48

Finally, the NCS result that comorbid simple andsocial phobias are usually temporally primary while co¬morbid agoraphobia is usually temporally secondary isconsistent with previous results of both clinical and epi¬demiologie studies.14·22·49·50 However, it is important toremember that both the NCS and many of the previousstudies that examined temporal priority between pho¬bias and other disorders dated the age at onset of the firstfear of phobic situations rather than of first meeting fulldiagnostic criteria. It is possible that an assessment of thelatter would lead to different conclusions concerning tem¬poral priorities between phobias and comorbid disor¬ders. Although clinical studies might be expected to pro¬vide valid data of this sort, such studies often rely on

retrospective patient reports that are subject to the same

sort of bias.35 The only rigorous way to obtain unequivo¬cal data on this issue is to use longitudinal general popu¬lation data. Data of this sort document that anxiety dis¬orders generally occur before comorbid affective disorders,but are less clear that anxiety disorders predict the sub¬sequent onset of substance use disorders.51"33

IMPAIRMENT

The NCS results concerning perceived role impair¬ments are at odds with most other clinical and commu¬

nity studies in finding that a somewhat smaller propor¬tion ofpersons with agoraphobia (26.5%) than those witheither simple (34.2%) or social (33.5%) phobias re¬

ported that their phobia "interfered a lot" with their "lifeand activities." This might be caused by imprecision inthe single self-reported question used to assess per¬ceived role impairment, which could miss the fact thatavoidance of public places impairs ability to work muchmore than does avoidance of harmless animals or avoid¬ance of social situations. More detailed and objective as-

sessments of role functioning have generally found thatthe rank ordering of the different phobias in terms of im¬pairment varies depending on the way impairment is as¬

sessed,14·15 a result that is consistent with our analysis ofthe impairments associated with current phobias. Moredetailed assessment of functional impairments is neededin future epidemiologie research on this issue.

PROFESSIONAL HELP SEEKING

The NCS results are consistent with earlier community sur¬

veys in finding that only a minority of individuals with pho¬bia ever seek professional help,15·24·54 as well as in findingthat persons with agoraphobia are more likely than are thosewith other phobias to obtain treatment.54 While it has pre¬viously been assumed that the higher prevalence of helpseeking among persons with agoraphobia is an effect of theirgreater impairment, the NCS results suggest that this mightnot be the case, as this association persists after control¬ling for perceived impairment. One plausible interpreta¬tion of this result is that subjects with agoraphobia are more

likely than are those with other phobias to interpret theirsymptoms in mental health terms. This could occur be¬cause the symptoms of agoraphobia first occur much laterthan the symptoms of other phobias and, because of this,are more readily seen as deviations from some prior stan¬dard of behavior, rather than the way things have alwaysbeen. Or it could be that fears of agoraphobic situationsare more readily interpreted as pathological than fears ofsimple phobic situations (which might be interpreted ascaused by cautiousness rather than mental disorder) or so¬

cial phobic situations (which might be interpreted as causedby shyness rather than mental disorder). Whatever the case

may be, the objective role impairments, barriers associ¬ated with problem recognition, and subsequent help seek¬ing among individuals with simple and social phobias needto be studied more seriously in future research.

Accepted for publication September 7, 1995.The National Comorbidity Survey (NCS) is a col¬

laborative epidemiologie investigation of the prevalence,causes, and consequences of psychiatric morbidity and co-

morbidity in the United States, supported by the NationalInstitute of Mental Health (grants ROI 46376 andROI MH49098) and the National Institute of Drug Abuse(through a supplement to ROÍ MH/DA46376), Rockville,Md, and the W. T. Grant Foundation, New York, NY (grant90135190; Dr Kessler, Principal Investigator). Prepara¬tion of this report was also supported by Research Scien¬tist Award K05 MH00507 and by Training grants T32MH14641 and T32 MH16806 from the National Instituteof Mental Health. Collaborating NCS sites and investiga¬tors are The Addiction Research Foundation, Toronto, On¬tario (Robin Room, PhD); Duke University Medical Cen¬ter, Durham, NC (Dan Blazer, MD, PhD, and MarvinSwartz, MD); The fohns Hopkins University, Baltimore,Md (fames Anthony, MD, William Eaton, PhD, and PhilipLeaf, PhD); the Max Planck Institute of Psychiatry-Clinical Institute, Munich, Germany (Hans-Ulrich Wittchen,PhD); the Medical College of Virginia, Richmond (Ken¬neth Kendler, MD); The University of Michigan, Ann Ar¬bor (LloydJohnston, PhD, and Ronald Kessler, PhD); New

Page 10: Agorphobia, Simple Phobia, And Social Phobia in the National Comorbidity Survey

York University, New York (Patrick Shrout, PhD); SUNYStony Brook (Evelyn Bromet, PhD); the University of Mi¬ami, Miami, Fla (R. Jay Turner, PhD); and WashingtonUniversity School of Medicine, St Louis, Mo (Linda Cot¬tier, PhD).

A complete list of NCS publications can be obtainedfrom the NCS Study Coordinator, Room 1006, Institutefor Social Research, The University of Michigan, Box 1248,Ann Arbor, MI 48106-1248. The text of this and other NCSpublications, working papers, and the raw data from theNCS can also be obtained from the NCS home page byusing the URL: http://www.umich.edu/~ncsum/.

We appreciate the thoughtful comments of MurrayStein and the anonymous reviewers.

Reprint requests to Survey Research Center, The Uni¬versity of Michigan, 426 Thompson St, PO Box 1248, AnnArbor, Ml 48106-1248 (Dr Kessler).

REFERENCES

1. Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshleman S, WittchenH-U, Kendler KS. Lifetime and 12-month prevalence of DSM-III-R psychiatricdisorders in the United States: results from the National Comorbidity Survey.Arch Gen Psychiatry. 1994;51:8-19.

2. Wittchen H-U, Zhao S, Kessler RC, Eaton WW. DSM-III-R generalized anxietydisorder in the National Comorbidity Survey. Arch Gen Psychiatry. 1994;51:355-364.

3. Warner LA, Kessler RC, Hughes M, Anthony JC, Nelson CB. Prevalence andcorrelates of drug use and dependence in the United States: results from theNational Comorbidity Survey. Arch Gen Psychiatry. 1995;52:219-229.

4. World Health Organization. Composite International Diagnostic Interview (Ver-sion 1.0). Geneva, Switzerland: WHO; 1990.

5. Kessler RC, Wethington E. The reliability of life events reports in a communitysurvey. Psychol Med. 1991;21:1-16.

6. Wittchen, H-U. Reliability and validity studies of the WHO-Composite InternationalDiagnostic Interview (CIDI): a critical review. J Psychiatr Res. 1994;28:57-84.

7. University of Michigan. OSIRIS VII. Ann Arbor, Mich: Institute for Social Re-search, The University of Michigan; 1981.

8. Norusis MJ. SPSS Advanced Statistics User's Guide. Chicago, III: SPSS Inc;1990.

9. SAS Institute. SAS 6.03. Cary, NC: SAS Institute; 1988.10. Woodruff RS, Causey BD. Computerized method for approximating the vari-

ance of a complicated estimate. J Am Stat Assoc. 1976;71:315-321.11. Koch GG, Lemeshow S. An application of multivariate analysis to complex sample

survey data. J Am Stat Assoc. 1972;67:780-782.12. Davidson JRT, Hughes DL, George LK, Blazer DG. The boundary of social pho-

bia: exploring the threshold. Arch Gen Psychiatry. 1994;51:975-983.13. Agras S, Sylvester D, Oliveau D. The epidemiology of common fears and pho-

bias. Compr Psychiatry. 1969;10:151-156.14. Wittchen H-U, Essau CA, von Zerssen D, Krieg CJ, Hecht H. Lifetime and six$

month prevalence of mental disorders in the Munich follow-up study. Eur ArchPsychiatry Clin Neurosci. 1992;241:247-258.

15. Angst J, Dobler-Mikola A. The Zurich Study, V: anxiety and phobia in youngadults. Eur Arch Psychiatry Clin Neurol Sci. 1985;235:171-178.

16. Costello CG. Fears and phobia in women: a community study. J Abnorm Psy-chol. 1982;91:280-286.

17. Uhlenhuth EH, Baiter MB, Mellinger GG, Cisin IH, Clinthorpe J. Symptom check-list syndromes in the general population: correlations with psychotherapeuticdrug use. Arch Gen Psychiatry. 1983;40:1167-1173.

18. Robins LN, Helzer JE, Croughan JL, Ratcliff KS. National Institute of MentalHealth Diagnostic Interview Schedule: its history, characteristics and validity.Arch Gen Psychiatry. 1981;38:381-389.

19. Bland RC, Orn H, Newman SC. Lifetime prevalence of psychiatric disorders inEdmonton. Acta Psychiatr Scand. 1988;77:24-32.

20. Dagonda M, Angst J. The Zurich Study, XX: social phobia and agoraphobia.Eur Arch Psychiatry Clin Neurosci. 1993;243:95-102.

21. Robins LN, Helzer J, Weissman MM, Orvaschel H, Gruenberg E, Burke JD Jr,Regier DA. Lifetime prevalence of specific psychiatric disorders in three sites.Arch Gen Psychiatry. 1984;41:949-958.

22. Schneier FR, Johnson J, Hornig C, Liebowitz M, Weissman M. Social phobia:comorbidity and morbidity in an epidemiologic sample. Arch Gen Psychiatry.1992;49:282-288.

23. Wacker HR, Mullejans R, Klein KH, Battegay R. Identification of cases of anxi-ety disorders and affective disorders in the community according to ICD-10and DSM-III-R using the Composite International Diagnostic Interview (CIDI).Int J Meth Psychiatr Res. 1992;2:91-100.

24. Stein MB, Walker JR, Forde DR. Setting diagnostic thresholds for social pho-bia: considerations from a community survey of social anxiety. Am J Psychia-try. 1994;151:408-412.

25. Wittchen, H-U, Zhao, S, Abelson, JM, Abelson, JL, Kessler, RC. Reliability andprocedural validity of UM-CIDI DSM-III-R phobic disorders. Psychol Med. Inpress.

26. Ost L-G. Age of onset of different phobias. J Abnorm Psychol. 1987;96:223\x=req-\229.

27. Burke KC, Burke JD, Rae DS, Regier DA. Comparing age at onset of majordepression and other psychiatric disorders by birth cohorts in five US com-munity populations. Arch Gen Psychiatry. 1991;48:789-795.

28. Boyd JH, Rea DS, Thompson JW, Burns BJ, Bourdon K, Locke BZ, Regier DA.Phobia: prevalence and risk factors. Soc Psychiatry Psychiatr Epidemiol. 1990;25:314-323.

29. Chen C, Wong J, Lee N, Chan-Ho M, Lau JT, Fung M. The Shatin CommunityMental Health Survey in Hong Kong, II: major findings. Arch Gen Psychiatry.1993;50:125-133.

30. Hwu HG, Yeh EK, Chang LY. Prevalence of psychiatric disorders in Taiwan asdefined by the Chinese Diagnostic Interview Schedule. Acta Psychiatr Scand.1989;79:136-147.

31. Joyce PR, Bushnell JA, Oakley-Browne MA, Wells JE, Hornblow AR. The epi-demiology of panic symptomatology and agoraphobic avoidance. Compr Psy-chiatry. 1989;30:303-312.

32. Bourdon KH, Boyd JH, Rae DS, Burns BJ, Thompson JW, Locke BZ. Genderdifferences in phobias: results of the ECA community survey. J Anxiety Dis-ord. 1988;2:227-241.

33. Canino GJ, Bird HR, Shrout PE, Rubio-Stipec M, Bravo M, Marinez R, SesmanM, Guevara LM. The prevalence of specific psychiatric disorders in Puerto Rico.Arch Gen Psychiatry. 1987;44:727-735.

34. Eaton WW, Dryman A, Weissman MM. Panic and phobia. In: Robins LN, Re-gier DA, eds. Psychiatric Disorders in America: The ECA Study. New York, NY:The Free Press; 1991:155-179.

35. Brown DR, Eaton WW, Sussman L. Racial differences in prevalence of phobicdisorders. J Nerv Ment Dis. 1990;178:434-441.

36. Escobar JI, Karno M, Burnam AH, RL, Golding J. Distribution on major mentaldisorders in an US metropolis. Acta Psychiatr Scand. 1988;78(suppl):45-53.

37. Wittchen H-U, Essau CA, Kreig J-C. Anxiety disorders: similarities and differ-ences of comorbidity in treated and untreated groups. Br J Psychiatry. 1991;159(suppl):23-33.

38. De Ruiter C, Rijken H, Garssen B, van Schaik A, Kraaimaat F. Comorbidityamong the anxiety disorders. J Anxiety Disord. 1989;3:57-68.

39. Sanderson WC, DiNardo PA, Rapee RM, Barlow DH. Syndrome comorbidity inpatients diagnosed with a DSM-III-R anxiety disorder. J Abnorm Psychol. 1990;99:308-312.

40. Dick CL, Sowa B, Bland RC, Newman SC. Phobic disorders. Acta PsychiatrScand. 1994;89(suppl):36-44.

41. Weissman MM, Myers JK, Harding PS. Psychiatric disorders in a US urbancommunity. Am J Psychiatry. 1978;135:459-462.

42. Weissman MM, Merikangas KR. The epidemiology of anxiety and panic dis-orders: an update. J Clin Psychiatry. 1986;47:11-17.

43. Wittchen H-U. Zum Spontanverlauf unbehandelter Falle mit Angst-storungenund Depressionen. In: Wittchen H-U, von Zerssen D, eds. Verlaufe Behandel-ter und Unbehandelter Depressionen und Angst-storungen. Heidelberg, Ger-many: Springer Verlag; 1988.

44. Weissman MM, Leaf PJ, Blazer DG, Boyd JH, Florio L. The relationship be-tween panic disorder and agoraphobia: an epidemiologic perspective. Psycho-pharmacol Bull. 1986;22:787-791.

45. Horwarth E, Lish JD, Johnson J, Hornig CD, Weissman M. Agoraphobia with-out panic: clinical reappraisal of an epidemiologic finding. Am J Psychiatry.1993;150:1496-1501.

46. Klein DF, Gorman JM. A model of panic and agoraphobia development. ActaPsychiatr Scand. 1987;76(suppl):87-95.

47. Klein DF, Klein HM. The nosology, genetics, and theory of spontaneous panicand phobias. In: Tyrer PJ, ed. Psychopharmacology of Anxiety. New York, NY:Oxford University Press; 1989.

48. Schuckit MA, Hesselbrock V. Alcohol dependence and anxiety disorders: whatis the relationship? Am J Psychiatry. 1994;151:1723-1734.

49. Angst J, Vollrath M, Merikangas KR, Ernst C. Comorbidity of anxiety and de-pression in the Zurich cohort study of young adults. In: Maser JD, CloningerCR, eds. Comorbidity of Mood and Anxiety Disorders. Washington, DC: Ameri-can Psychiatric Press; 1990:123-138.

50. Lesser IM, Rubin RT, Pecknold JC, Rifkin A, Swinson RP, Lydiard RB, Bur-rows GD, Noyes R Jr, DuPont RL Jr. Secondary depression in panic disorderand agoraphobia. Arch Gen Psychiatry. 1988;45:437-443.

51. Murphy JM. Diagnostic comorbidity and symptom co-occurrence: The StirlingCounty Study. In: Maser JD, Cloninger CR, eds. Comorbidity of Mood and Anxi-ety Disorders. Washington, DC: American Psychiatric Press; 1990:153-176.

52. Hagnell O, Tunvia K. Prevalence and nature of alcoholism in a total population.Soc Psychiatry. 1972;7:190-201.

53. Vaillant G. The Natural History of Alcoholism. Cambridge, Mass: Harvard Uni-versity Press; 1983.

54. Thompson JW, Burns BJ, Bartko J, Boyd JH, Taube CA, Bourdon KH. The useof ambulatory service by persons with and without phobia. Med Care. 1988;26:183-198.