The Genetic The

Epidemiology of Phobias in Women

Interrelationship of Agoraphobia, Social Phobia, Situational Phobia, and Simple Phobia Kenneth S. Kendler, MD; Michael C. Neale, PhD; Ronald C. Kessler, PhD; Andrew C. Heath, DPhil; Lindon J. Eaves, PhD, DSc

personally interviewed female twins from a population-based registry, the pattern of age at onset and comorbidity of the simple phobias (animal and situational) early onset and low rates of comorbidity\p=m-\differed significantly from that of agoraphobia\p=m-\later onset and high rates of comorbidity. Consistent with an inherited "phobia proneness" but not a "social learning" model of phobias, the familial aggregation of any phobia, agoraphobia, social phobia, and animal phobia appeared to result from genetic and \s=b\ In 2163

\p=m-\

not from familial-environmental

factors, with estimates of

heritability of liability ranging from 30% to 40%. The bestfitting multivariate genetic model indicated the existence of genetic and individual-specific environmental etiologic factors common to all four phobia subtypes and others specific for each of the individual subtypes. This model suggested that (1) environmental experiences that predisposed to all phobias were most important for agoraphobia and social phobia and relatively unimportant for the simple phobias, (2) environmental experiences that uniquely predisposed to only one phobia subtype had a major impact on simple phobias, had a modest impact on social phobia, and were unimportant for agoraphobia, and (3) genetic factors that predisposed to all phobias were most important for animal Phobia and least important for agoraphobia. Simple phobias appear to arise from the joint effect of a modest genetic vulnerability and phobia-specific traumatic events in childhood, while agoraphobia and, to a somewhat lesser extent, social phobia result from the combined effect of a slightly stronger genetic influence and nonspecific environmental experiences. (Arch Gen Psychiatry. 1992;49:273-281) of the wellamong the Phobias, of particular recognized psychiatric syndromes, interest because several most

common

are

to

Accepted

genetic epidemiologists

com-

for publication July 24, 1991. From the Departments of Psychiatry (Dr Kendler) and Human Genetics (Drs Kendler, Neale, and Eaves), Medical College of Virginia/Virginia Commonwealth University, Richmond; the Institute for Social Research, University of Michigan, Ann Arbor (Dr Kessler); and the Department of Psychiatry, Washington University School of Medicine, St Louis, Mo (Dr Heath). Reprint requests to Department of Psychiatry, Medical College of Virginia, Box 710, Richmond, VA 23298-0710 (Dr Kendler).

peting etiologic

theories have been developed, each of which suggests that phobias arise mainly from one of the three domains of risk factors: individual-specific environ¬ ment, family environment, and genes. One school of thought has suggested that phobias arise as a result of classic conditioning from the accidental pairing of benign and fear-inducing stimuli.1"4 This theory, which postu¬ lates that phobias should be entirely environmental in or¬ igin, would predict that the familial aggregation of pho¬ bias, which would result only from the correlated exposure among family members to such a coincidental pairing of stimuli, is likely to be modest. Social learning theory provides the basis of another be¬ havioral theory of phobias. In humans and nonhuman primates, the fear of objects or situations can be learned from observing the fear responses of others.5 Since most phobias begin in childhood,6 other family members, par¬ ticularly parents, are likely sources for the "vicarious" ac¬ quisition of phobias. The social learning theory, therefore, predicts substantial familial aggregation of phobias from shared environmental experiences. As first observed by Darwin,7 the choice of feared ob¬ jects and situations in phobias is not random.28-9 This has been explained by two related theories of "inherited pho¬ bia proneness": (1) a "preparedness" to develop condi¬ tioned fears of certain stimuli9 or (2) innate fears that re¬ quire no learning.10 Both theories suggest that, through natural selection, humans have evolved an inherited pre¬ disposition to form phobic reactions to certain stimuli, but they differ as to whether conditioning is required to man¬ ifest this predisposition. It is probable that phobia prone¬ ness would evolve through stabilizing selection11 in which a moderate level produced optimal fitness and fitness de¬ clined when an individual became either too phobia prone or insufficiently phobia prone. Since such selection usu¬ ally results in substantial additive genetic variation in populations,1113 the inherited phobia proneness theory predicts significant genetic influences on the predisposi¬ tion to

phobias.

THE GENETICS OF PHOBIAS Information is available to evaluate these three models preliminarily. Several studies, using a twin or twin-family design, have found that genetic factors and individual-

Downloaded From: http://archpsyc.jamanetwork.com/ by a Monash University Library User on 06/18/2015

specific environment, but not family environment, play an important role in the origin of self-reported phobialike

fears.14"18 We are aware of six studies of the familial aggre¬ gation of clinically defined phobias. Four family history studies, three beginning with agoraphobic probands19"21 and one with social phobic probands,22 have found evidence of the familial aggregation of phobias. Using a family study design, Noyes et al23 found, in relatives of agoraphobic vs control probands, significantly increased rates of agoraphobia but not of social or simple phobia.

Beginning with probands with simple phobia, Fyer et al,24 using direct interviews, found significantly higher rates for simple but not social phobia in relatives of affected vs control probands. Two twin but no adoption studies have examined clin¬ ically defined phobias. Carey and Gottesman,25 examin¬ ing 21 twin probands hospitalized with phobia, found low and similar concordance rates in monozygotic (MZ) and dizygotic (DZ) twins for treated phobia. However, using the broad definition of phobic "symptoms or features," they found a much higher concordance rate in MZ (88%)

than in DZ (38%) twins. Torgersen26 studied 12 twin probands in Norway with a primary diagnosis of phobia and found that none of the cotwins were phobic. How¬ ever, two of the four MZ cotwins had another

Agoraphobia Going out of house alone Being in crowds Being in open spaces Social phobia Meeting new people Giving a speech Using public bathrooms Eating in public Animal phobia Insects

Spiders Mice

Snakes Bats

Situational phobia Tunnels Other closed places

Bridges Airplanes Other high places

anxiety

compared with none of the eight DZ cotwins. Although clinically defined phobias appear to be famil¬ ial, we do not know whether this results from genetic disorder

% of Sample % With Subtype of Phobia Endorsing This Fear Accompanied Endorsing This Fear by Interference

factors and/or shared environmental factors. The two twin studies of phobias25,26 had quite small samples,

ascertained only treated cases (which constitute a small and unrepresentative proportion of affected individu¬ als27), and assessed twins with knowledge of the clinical status of the cotwin. SUBTYPES OF PHOBIA: THE SAME OR DIFFERENT? The phobias are a set of syndromes all of which share an irrational and fearful avoidance of objects or situations that cannot be explained as a function of the threat truly posed thereby. However, the phobias also differ in important ways. In simple phobias, the phobic stimulus is well circumscribed, while in social phobia and particu¬ larly agoraphobia, the phobic stimuli are relatively dif¬ fuse. Furthermore, Klein28 argued that the development

of agoraphobia, as a sequela of spontaneous panic attacks, is qualitatively different from that of other phobias. What is the relationship between the genetic and envi¬ ronmental risk factors for the different subtypes of pho¬ bias? Age at onset2-29 and patterns of comorbidity30-31 sug¬ gest meaningful differences between phobic subtypes. Family studies of phobia performed to date have found some evidence of the specificity of familial factors for in¬ dividual phobic subtypes.23,24 In the only study to date using multivariate genetic analysis, Phillips et al,15 in a twin-family design, examined seven categories of fears as assessed by self-report and found evidence of a single common genetic factor for all categories of fears plus ge¬ netic factors specific to each fear. GOALS the By examining for presence of any phobia and of four subtypes of phobia as assessed at personal interview in 2163 adult female twins from a population-based twin register, we sought to clarify the role of genetic and en¬ vironmental factors both in the etiology of phobias and in

the

4.1 6.6 0.9

46.6 75.1 10.1

4.4 9.4 1.9 1.6

37.8 81.9 16.1 13.7

1.0 3.3 2.3 5.5 1.6

9.4 30.2 21.3 50.6 14.8

2.9 3.7 3.1 4.7 7.4

23.8 30.6 25.3 38.1 60.0

interrelationship between the individual phobic sub¬

types.

SUBJECTS AND METHODS

Sample and Assessment Procedures As outlined in detail elsewhere,32 female-female twin pairs from the population-based Virginia Twin Registry were person¬ ally interviewed by social workers who were blind to the clinical status of the cotwin. Zygosity was determined blindly by stan¬ dard questions,33 photographs, and, when necessary, DNA inter¬ testing.34 Of the eligible twins, 92% were viewed, 89% in person and 11% by telephone. Lifetime prevalence was assessed for the following psychiat¬ ric disorders by means of an adapted version of the Structured Clinical Interview for DSM-III-R Diagnosis35: major depression, generalized anxiety disorder, panic disorder, alcoholism, and eating disorders (anorexia nervosa and bulimia nervosa). A life¬ time history of phobias was assessed by an adaptation of the Phobic Disorders section of the Diagnostic Interview Schedule Version III-A,36 which in turn was based on the DSM-III criteria.37 Three significant modifications were made. First, instead of the respondent being asked individually about a series of fears, the fears were grouped into four categories (Table 1), each of which was inquired about in a single question. Second, several mod¬ ifications were made to the list of 17 specific fears: (1) fear of us¬ ing public bathrooms was substituted for fear of writing in pub¬

successfully

lic, (2) fears of water, thunder, lightning, being alone, buses, and blood

or

injections were dropped, (3)

animal

phobias were ex¬

fears of bugs, spiders, mice, snakes, and bats, and (4) situational phobias were expanded to

panded

to include individual

separate fear of tunnels, bridges, airplanes, and other high

places. While DSM-III mentions fear of bridges and tunnels as typical of agoraphobia, in our interview these fears were consid¬ ered part of situational phobia. We also inquired about other phobias, and if a phobia described there belonged with one of the four specific subtypes, it was so treated. Other phobias mentioned in response to this question (eg, fear of water, the dark, driving, and blood) were included in our analysis of "any" phobia but were not counted as belonging to one of the specific subtypes. Third, instead of interference being asked about with each

Downloaded From: http://archpsyc.jamanetwork.com/ by a Monash University Library User on 06/18/2015

specific fear, it was asked about after each of the four specific and one miscellaneous category of unreasonable fears. In the Diag¬ nostic Interview Schedule, which was developed for nonclinician interviewers, the respondent makes the judgment about

interference from an irrational fear; in contrast, in our interview, designed for clinicians, the interviewer made this assessment. Our interviewers were trained to rate whether the unreasonable fear had significant objective behavioral impact on the respon¬ dent's life. Of the "unreasonable" fears expressed in our sample, 40.1% were judged to be accompanied by significant interfer¬ ence.

Phobia was defined in this report, then, as the presence of a fear that the respondent recognized as unreasonable and that, in the judgment of the interviewer, significantly interfered with the respondent's life. For all analyses reported in this article, a twin was considered affected for all subtypes of phobias for which she met criteria. Fifty-three interviews were jointly conducted by two inter¬ viewers. Reliability was assessed by examining the five groups of unreasonable fears in each of these interviews. The two inter¬ viewers agreed on the presence or absence of an unreasonable fear 261 of 265 times (ie, 53 respondents times five subtypes) (agreement rate, 98.5%; + .95± .0238). In the 83 cases in which both raters agreed that an unreasonable fear was present, their rating of interference agreed in 75 cases (agreement rate, 90.3%; =

=+.81±.07).

Statistical

Analysis

Our approach to the analysis of twin data, has been outlined in detail elsewhere,32,39,40 and consists of inferring the action of genetic and environmental risk factors from the way in which affected and unaffected twins are distributed in twin pairs. In this article, we expand on the previously described univariate analyses to conduct a multivariate genetic analysis of the subtypes of phobias. While the goal of univariate genetic anal¬ ysis is the decomposition of the variance of a trait into genetic and environmental components, in multivariate genetic analysis, the focus shifts to decomposing sources of covariance between traits. To illustrate the difference between univariate and multivariate twin analysis, the concept of latent or unobserved factors is in¬ troduced. In a traditional or "phenorypic" factor analysis, latent factors are postulated to cause the resemblance or, more techni¬ cally, covariation among items. The goal of factor analysis is to explain the correlations between a large number of variables as resulting from the effects of a small number of latent factors. Multivariate genetic analysis is also a method of explaining cor¬ relations between multiple items. However, it goes beyond tra¬ ditional factor analysis in that it provides insight into the causes of resemblance among variables. In univariate analysis, information regarding the causes of variation is obtained by comparing the resemblance of MZ and DZ twin pairs for a single variable. In the multivariate case, the correlation between two or more variables is the primary unit of analysis. By comparing the cross-twin, cross-variable correlation in MZ and DZ twins, and contrasting that to the cross-twin within-variable and within-twin cross-variable correlations, the covariation of two or more variables can be partitioned into its genetic and environmental components. Two alternative models are tested to describe how genetic and environmental factors may influence covariation. In the common pathway model,41 genetic and environmental factors influence covariation through a single common pathway. Such a model contains no separate genetic and environmental latent factors. Rather, genetic and environmental variables act conjointly through one or more latent phenotypes. By contrast, in the in¬ dependent pathway model, genes and the environment can con¬ tribute to covariation through separate genetic and environmen¬ tal latent factors. Because we had only four subtypes of phobias to examine, only single-factor models were identified. It was not possible to test multifactor models, such as, for example, that animal and

social phobias arise from an distinct evolutionary and neurobi¬ ologie mechanisms.8 The form of data for our multivariate genetic analysis is two 8x8 tetrachoric correlation matrixes, calculated by PRELIS,42 giving the tetrachoric correlations within and across twins for the four phobia subtypes, separately for MZ and DZ twins. To de¬

scribe best how genes and environment influence the resem¬ blance among the subtypes of phobias, a series of multivariate models were fitted to these matrixes by means of LISREL43 by the method of weighted least squares. As in the univariate analysis of individual subtypes of phobias, the fit of the various models was compared by a 2 difference test, and the model that best combined the features of parsimony and goodness of fit was se¬ lected by Akaike's44 information criterion (AIC). The impact of mode of interview and zygosity (MZ vs DZ) on the risk of phobias was assessed by logistic regression45 in which the dependent variable was the presence vs absence of a diag¬ nosis of phobia. In all such analyses, age was included as a con¬ trol variable and the statistical significance of the independent variable was determined by a 2 test with 1 df. To test the "equal environment assumption" (ie, that MZ and Dz twins are equally correlated for exposure to the environmental experiences that are etiologically relevant for phobias), the impact of the similar¬ ity of childhood and adult environment on the similarity for phobias was assessed by logistic regression in which twin-pair similarity for phobias (concordant for affection or nonaffection vs discordant) was the dependent variable. The difference in comorbidity in the different phobic subtypes was assessed by a log-linear model in CATMOD.45 Standardized life table curves (in which data for only affected individuals were analyzed, so that 100% of the sample eventually experienced onset) were obtained with the use of the LIFETEST procedure.45 Our analyses of age at onset, which was not available for all phobia subtypes in some twins, were based on the following sample size: agoraphobia, 33 twins; social phobia, 81 twins; an¬ imal phobia, 118 twins; and situational phobia, 96 twins.

RESULTS

Prevalence and Mode of Interview Of the 2163 interviewed twins, 722 (33.4%) gave a lifetime his¬ tory of any phobia, 654 of whom met criteria for one or more of

the four specific phobic subtypes. Of the four subtypes, situa¬ tional phobia was the most common (12.3%), followed by social phobia (11.5%), animal phobia (10.9%), and agoraphobia (8.7%). The lifetime prevalence of any simple phobia (animal or situa¬ tional) was 20.2%. Of the twins suffering from one or more of the four subtypes, 443 (67.7%) met criteria for only one subtype, 149 (22.8%) met criteria for two, 51 (7.8%) met criteria for three, and 11 (1.7%) met criteria for all four specific phobias. The propor¬ tion of the total sample and the proportion of individuals meet¬ ing criteria for each subtype of phobia that endorsed the assessed specific unreasonable fears are seen in Table 1. The mode of interview (phone vs face to face) was not signif¬ icantly related to risk for any phobia or any specific subtype of phobia. Whether or not the cotwin agreed to be interviewed was also unrelated to risk for agoraphobia, social phobia, or situa¬ tional phobia. The results for any phobia ( 2 4.12, P= .04) and animal phobia ( 2 5.35, .02), while significant, were in the opposite direction from that predicted (refusal in cotwin pre¬ dicting lower risk for phobia). =

=

=

Age at Onset, Structure of Phobias, and Comorbidity The standardized onset curves for the four forms of phobia (Fig 1) were significantly different from one another (log-rank 2 30.10, d/=3, PSi Ag>S, An, Si;

Alcoholism

2.79t

3.84+

2.62+

2.86+

2.35+ 3.31+

2.02+

Generalized disorder

Eating disorders *Ag indicates agoraphobia; An,

+PAn, Si

Ag>Si An>Ag, S,

Si

Lifetime

Probandwise

Prevalence^

Concordance^

%

Phobia

Any

Tetrachoric Correlation

%

(±SD)

MZ

DZ

MZ

DZ

MZ

DZ

30.0

31.2

41.9

37.0

.32+ .06

.15 ±.08

Agoraphobia Social phobia

7.3

10.3

23.2

15.3

.41

±

.11

.15±.13

11.7

11.6

24.4

15.3

.31

±

.09

.12 + .12

Animal

10.5

12.1

25.9

11.0

.38±.10

11.6

12.9

22.2

23.7

.27±.10

.04±.13 .27±.11

phobia Situational phobia *MZ indicates

monozygotic; DZ, dizygotic.

+Corrected, by regression analysis, for small

mean

differences in age in MZ and DZ twins.

that

were of etiologic importance only for specific subtypes of phobia. As with all our analyses, our goal was to find the most parsimonious model that provided a good explanation of the observed data. This was done by determining the model that produced the most negative value of the AIC. We began with a full independent pathway model (see above and Kendler et al,41 especially their Figure, for further informa¬ tion about independent vs common pathway models) that included additive genetic, familial environmental, and individual-specific environmental factors both unique to each phobia and common to all four phobias. This model (model 1) fit very well ( 2 33.3, d/=48, P= .60, AIC= -38.7). We then fit a common pathway model of the same type (ACE for both pho¬ bia specific factors and phobia common factors) (model 2), which produced a slightly less negative AIC than model 1 ( 2 45.9, d/=54, AIC =-38.1). =

=

In model 3, we eliminated from model 1 the familial environ¬ mental common factor and produced a modestly more negative AIC ( 2 41.1, d/=52, AIC= -38.9). However, eliminating the additive genetic common factor from model 1 (model 4) pro¬ duced a slight deterioration in the AIC ( 2 41.7, df =52, AIC =-38.3). Eliminating phobia specific familial environmental factors from model 3 (model 5) substantially improved the fit ( 2 42.4, df= 56, AIC 45.6). Since the individual environmental factors specific to agoraphobia and the additive genetic factors specific to animal phobias were estimated at zero in model 5, in model 6 these were set to zero, producing the same fit, but two fewer d/and the best obtainable AIC value (-49.6). The parameter estimates for the best-fitting model 6 are depicted in Fig 2. The additive genetic common phobia factor had the highest loading on animal phobias and the lowest on agoraphobia. The individual-specific environmental common phobia factor had the reverse pattern, with the highest load¬ ing on agoraphobia and the lowest on animal phobia. Com¬ paring these loadings with those found in the standard (phenotypic) factor analysis (Table 2) suggested that most of the observed pattern of comorbidity between the subtypes of phobias results from the operation of environmental risk fac¬ =

=

=

=

-

tors.

The sources of variance in liability to the individual phobias as determined by model 6 vary widely across the four phobia sub¬ types (Table 7). Specific genetic factors were more important than common genetic factors for all subtypes except animal phobia. Environmental risk factors specific for individual phobia subtypes were more important than environmental risk factors common to all phobias for all subtypes except agoraphobia. So¬ cial and situational phobias were similar in that about two thirds of the genetic variation was due to genes specific to the individ¬ ual phobias. However, they differed in that for social phobia, common and specific environmental factors were about equally important, while for situational phobia, environmental factors

Fit of Twin

Model, 2 Units ACE+

CE

AE

Parameter Estimates

for ADE+

Best-Fitting Model

(df=0) (eff=1) (df=1) (tff=0)

Phobia

Any phobia Agoraphobia

0.03

3.07

0.03+-

0.16

2.39

0.16

.

.

.

0.00

0.68

0.32 ...

0.39

.

.

Social

phobia

0.61

.

0.10

1.64

0.10+-

0.00

0.30

2.95

7.13

2.95

1.13

0.32

0.00*

1.41

.

.

.

0.70

Animal

phobia

...

0.68

Situational

phobia

0.00

*A indicates additive

0.27

.

genetic effects; C,

0.73

familial envi¬ ronmental effects; E, individual-specific environmental effects; and D, dominance genetic effects. Thus, an ACE model includes additive genetic, common environmental, and individual-specific environ¬ mental risk factors; a2 indicates proportion of variance in liability due to additive genetic factors (narrow heritability); c2, proportion of vari¬ ance in liability due to common or familial environmental effects; and e2, proportion of variance in liability due to individual-specific envi¬ ronmental effects. tThe fit can be greater than zero with these models because all pa¬ rameters are constrained to be greater than or equal to zero. ÉBest-fitting model by Akaike's44 information criterion.

Phobia Model

Specific

Phobia Common Factors

common or

2

ACE ACE

Common

45.9 42 -38.1

5

AE

AE

6+

AE

AE

Independent Independent Independent Independent

41.1

4

ACE ACE ACE ACE

Pathway Independent

1 2 3

Factors

AE CE

33.3

df

AIC

48 -38.7 52 -38.9

41.7 52 -38.3

42.4 56 -45.6 42.4 58 -49.6

*A indicates additive

genetic effects; C, common or familial en¬ vironmental effects; E, individual-specific environmental effects; and AIC, Akaike's44 information criterion. Specific factors influence only each individual phobia subtype, while common factors influ¬ ence all the phobia subtypes. See text and Kendler et al41 for an ex¬ planation of common vs independent pathway models. tBest-fitting model by AIC.44

Downloaded From: http://archpsyc.jamanetwork.com/ by a Monash University Library User on 06/18/2015

earlier in the animal than in the situational subtype.29,46 As in the Epidemiologie Catchment Area study,6 we found that age at onset of phobias was usually in childhood and adolescence. Our results also replicated previous findings that the rate of onset of social phobia is highest in the

teenage years.47

In accord with previous research,30,31 we found that the pattern of comorbidity differed substantially across the

Agoraphobia

0.54 A

0.46 0.60 A

0.77

0.45

E

,

0.73

four subtypes of phobias; agoraphobia consistently had the highest rates of comorbidity, and animal and situa¬ tional phobias had the lowest. As with age at onset, social phobias appeared, from the perspective of comorbidity, to fall between agoraphobia and animal and situational

phobias.

A,5!

Univariate Twin

Fig 2.—Results from the best-fitting multivariate genetic model for the subtypes of phobias (model 6). A indicates additive genetic, and E, individual-specific environmental effects. The subscript C refers to the phobia's common factors, while the other subscripts referto the additive genetic and individual-specific environmental effects specific to each phobic subtype: AC, agoraphobia; SP, social pho¬ bia; AP, animal phobia; and SI, situational phobia. Thus, while Ac refers to the additive genetic common factor (additive genes that influence liability to all phobia subtypes), EAP refers to individualspecific environmental experiences that uniquely influence liabil¬ ity only to animal phobia, as, for example, being bitten by a dog as a young child. The magnitude of each path, as estimated by model 6, is shown in the figure. These path or standardized partial regres¬ sion coefficients must be squared to equal the proportion of vari¬ ance in the dependent variable accounted for by the latent predic¬ tor variable.

Common

Phobia

Agoraphobia Social phobia

MZ and DZ twins.

Specific

-1

Genetic Environmental Genetic Environmental 0.07

0.64

0.29

0.00

0.10

0.32

0.21

0.36

0.35

0.05

0.00

0.59

0.09

0.17

0.20

0.53

Animal

phobia Situational

phobia that

specifically predisposed only

more ences

than three times that predisposed

phobia were experi¬ generally to all phobia subtypes. as

to situational

important

as

environmental

COMMENT In this article, we sought to understand the role of ge¬ netic and environmental influences in the etiology of phobias and, in particular, to clarify the interrelationships among the genetic and environmental risk factors for the

subtypes of phobias.

Age at Onset and Comorbidity Consistent with results from treated samples,2,29 age at our community sample differed significantly across subtypes of phobias, being earliest in animal pho¬ bias, intermediate in social phobias, and latest in agora¬ phobia. Age at onset also differed meaningfully between the two subtypes of simple phobia, being substantially onset in

Prior to our twin analyses, we tested for potential con¬ founding biases. The risk for any phobia and for three of the four specific phobias was unrelated to zygosity. However, MZ twins had significantly lower rates of ago¬ raphobia than DZ twins, as was found, in this sample, for one of six definitions of generalized anxiety disorder.48 This difference might reflect a protective effect of the close emotional bond between MZ twins, an impact of agora¬ phobia on cooperation rates,49 or sibling interaction in which the predisposition to agoraphobia in one twin re¬ duces the liability to agoraphobia in the cotwin.33 It is, however, difficult to see why these processes would affect only agoraphobia and not other phobias, alcohol depen¬ dence, or nearly all definitions of major depression, none of which have significant differences in prevalence in MZ vs DZ twins. Differences in base rates in the two zygosity groups have been incorporated into the analysis, as the

tetrachoric correlations allow for different thresholds in

Proportion of Variance -1

Analyses

Similarity of childhood or adult environment was unrelated to twin resemblance for all except animal pho¬ bias. The relationship between similarity of childhood environment and concordance for animal phobia was, given the size of the sample, at a modest level of statisti¬ cal significance ( 2 4.27, P= .04) and could plausibly be the result of chance fluctuations. However, it could be a real effect of exposure to environmental stimuli in child¬ hood, such as spiders, snakes, or frightening dogs, which would be more highly correlated in MZ than DZ twin girls because the former more often slept in the same room and played together. If real, this effect of environmental sim¬ ilarity could upwardly bias our estimates of the heritabil¬ ity of animal phobias. To test this, we matched MZ and DZ twins for similarity of childhood environment and then reapplied our twin models. The new estimate of the heritability of liability to animal phobias (30.5%) was only slightly less than that obtained above (32.5%), indicating that, if real, the greater similarity of childhood environ¬ ment for MZ vs DZ twins did not substantially bias our results. In general, then, our findings support the valid¬ ity of the "equal environment assumption" of the twin method with respect to the risk for phobias. Consistent with previous family studies, we found that the liability to phobias was significantly correlated in twin pairs. Model fitting suggested that for any phobia, and for three of the four phobia subtypes, this familial aggrega¬ tion was best explained solely by additive genetic factors. Estimates for the heritability, the proportion of total vari=

Downloaded From: http://archpsyc.jamanetwork.com/ by a Monash University Library User on 06/18/2015

in disease liability due to genetic factors, ranged from 30% for social phobia to 39% for agoraphobia. Contrary to the two previous small-sample twin studies of phobia, where evidence of genetic effects only occurred when phobialike fears or nonphobic anxiety disorder were con¬ sidered,25,26 we found evidence of the genetic transmis¬ sion of clinically defined phobias. Although not trivial, the estimated heritabilities of liability to phobias were substantially lower than those previously reported from twin studies of bipolar illness and schizophrenia, where they have usually ranged from 60% to 90% ,50,51 However, for situational phobias, a qualitatively differ¬ ent result was obtained. Model fitting suggested that fa¬ milial resemblance was the result solely of familial envi¬ ronmental influences. These results can be interpreted in two plausible ways. Either the causes of familial resem¬ blance really differ for situational vs other forms of pho¬ bias and this is a true finding, or they do not differ and this finding results from random statistical fluctuation. We evaluated these two hypotheses in three ways. First, in the multivariate analysis, which, unlike the univariate analyses, includes the cross-twin cross-phobia correla¬ tions, the best-fitting model predicted that familial resem¬ blance for situational phobia resulted only from genetic factors. Second, when we expanded our analyses to include, in a multiple threshold model, unreasonable sit¬ uational fears without interference, the tetrachoric corre¬ lations in MZ twins substantially exceeded those in DZ twins, and model fitting predicted that familial resem¬ blance for the liability to situational fears/phobias resulted solely from genetic factors. Third, we expanded the def¬ inition of situational phobia to include other simple, nonanimal phobias, such as fear of water and thunder¬ storms. This change of definition modestly increased the number of twins with situational phobia, but the MZ and DZ twin tetrachoric correlations remained similar for both zygosity groups, and model fitting indicated that familial resemblance was still due entirely to familial environmen¬ tal influences. Only a new sample will definitively ad¬ dress this question, although parsimony and our clinical intuition suggest that the second hypothesis is more ance

plausible.

In summary, the results of univariate genetic analysis consistent with predictions of the inherited phobia proneness model. If evolutionary fitness is maximized by a moderate degree of phobia proneness, our findings of modest additive genetic variance for the liability to pho¬ bias are as predicted by evolutionary theory. The esti¬ mated heritability of liability for phobias found in this sample indicates that genetic factors play a significant but by no means overwhelming role in the etiology of phobias. Individual-specific environment appears to ac¬ count for approximately twice as much variance in liabil¬ ity to phobias as do genetic factors. Although some of this environmental variance may result from unreliability of measurement, our findings are also consistent with con¬ ditioning models of phobias in suggesting that individualspecific traumatic environmental experiences play a ma¬ jor role in the etiology of phobias. With the possible exception of situational phobias, our results provide little support for the "social learning" model of phobias. If such a mechanism is etiologically important, it appears to occur in a small minority of cases,52 or phobias are rarely "learned" from other family members. However, even with the large sample of twins are

examined in this report, our ability to discriminate be¬ tween different models of familial transmission for the li¬ ability to phobias is limited. In the presence of moderate genetic influences, we have almost no power to detect additional modest familial environmental effects.53 All that we can safely conclude is that, except perhaps for situational phobias, large familial environmental influ¬ ences on individual phobias are unlikely.

Multivariate Genetic

Analysis

We applied the method of multivariate genetic analysis to clarify the relationship of two kinds of genetic and en¬ vironmental risk factors for the subtypes of phobia: those that are general, that influence risk for all phobia subtypes, and those that are specific, that influence risk to only one subtype of phobia. From a genetic perspective, this anal¬ ysis examines the extent to which genes influence a gen¬ eral "phobia-proneness set-point" as opposed to specifi¬ cally predisposing to only one phobia subtype. From an evolutionary perspective, we can ask: has evolution selected for a general predisposition to all phobic fears or only for the vulnerability to individual phobias? From an environmental perspective, these analyses will distin¬ guish between environmental experiences that influence only the general phobia-proneness set-point vs those that influence risk to a single phobic subtype. Learning theory models of phobias would generally predict that most en¬ vironmental risk factors for phobias would be subtype specific, as, for example, a traumatic exposure to a ening animal would be unlikely to increase the risk for a

fright¬

public-speaking phobia. In the initial

stages of the multivariate analyses, the in¬

dependent pathway model fit the data somewhat better than the common pathway model, suggesting that the genetic and environmental risk factors do not influence the pattern of the subtypes of phobias in the same way. Consistent with previous results from twin studies using self-report measures of fears,15,18 we found strong evi¬

dence of the existence both of genetic and environmental risk factors unique to each kind of phobia and for genetic and environmental risk factors that influenced all phobia subtypes. Our results were midway between the two ex¬ treme hypotheses regarding the interrelationship of the subtypes of phobias: (1) the subtypes of phobias are dis¬ tinct, unrelated syndromes and (2) the subtypes of pho¬ bias represent minor variations of a single disorder. Klein28 emphasized the major differences between ag¬ oraphobia and other phobic syndromes. A plausible pre¬ diction of this perspective is that agoraphobia should have the lowest loading on the phobia common factors. By con¬ trast, in our phenotypic factor analysis (Table 3), agora¬ phobia had the highest loading of any phobia on the pho¬ bia common factor. Our multivariate genetic analysis indicated that the phenotypic structure was due almost entirely to the operation of environmental factors. While agoraphobia was most influenced by the environmental risk factors common to all phobias, it was least influenced by the genetic factors common to all phobias.

General Implications Our results suggest that the subtypes of phobias can be placed along an etiologic continuum. At one end of this continuum lies agoraphobia, which has (1) the latest age at onset, (2) the highest rates of comorbidity, (3) the highest heritability, (4) the highest loading on the envi-

Downloaded From: http://archpsyc.jamanetwork.com/ by a Monash University Library User on 06/18/2015

ronmental common factor, (5) the least specific environ¬ mental influences, and (6) the lowest loading on the ge¬ netic common factor. At the other end of this continuum lie the simple phobias (animal and situational), which have (1) the earliest age at onset, (2) the lowest rates of comorbidity, (3) the lowest heritability, (4) the lowest loading on the environmental common factor, (5) the highest specific environmental influences, and (6) the (at least for animal phobias) highest loading on the common genetic factor. On nearly all of these measures, social phobia lies intermediate between agoraphobia and the

simple phobias. Contrary

to some

previous research,46

results suggest that situational phobias are more closely related to animal phobias than to agoraphobia. Our results confirm the utility of the twin design for clarifying the nature of environmental risk factors. Con¬ sistent with previous research,52 we found that in the simple phobias, pathogenic environmental experiences are usually highly specific (eg, being locked in a dark closet, bitten by a snake, nearly falling out of a window, etc). The role of the environment in agoraphobia, by con¬ trast, appears to be quite different, consisting of experi¬ ences that only influence the general level of phobia proneness. our

Limitations Our results should be interpreted in the context of six potentially significant limitations. First, these results ap¬ ply only to women. Although no study has examined the sex-specific transmission of phobias, the substantial gen¬ der difference in prevalence54 suggests that genetic and environmental risk factors may differ in their impact on phobias in men and women. Second, while based on a complete search of birth cer¬ tificates, our final study sample is unlikely to be entirely representative of the total twin population. Twins who moved out of state or did not return earlier questionnaires were unlikely to be included in our sample. Third, the lifetime prevalence rates for phobias in this sample are toward the high end of the range of those pre¬ viously reported in women. Whether this is due to differ¬ ences in population or method is not clear. Our interviewer-based assessment of impairment, while more objective, may produce a lower "threshold" for interfer¬ ence than the self-report measure used in the Diagnostic Interview Schedule. This might explain the higher rates of phobias found in our survey compared with those ob¬ tained with the Diagnostic Interview Schedule in New Haven, Conn; St Louis, Mo55; Puerto Rico56; Edmonton, Alberta57; and New Zealand.58 However, a lower thresh¬ old for interference should predict higher rates for all phobic subtypes. Compared with the national population estimates for white women from the Epidemiologie Catchment Area studies,54 however, our lifetime preva¬ lence rate for agoraphobia was similar, whereas our rates for social and simple phobia were considerably higher. Furthermore, the lifetime prevalence rates of phobia were quite variable in the four Epidemiologie Catchment Area sites with comparable information.54 The rates found for any phobia in white women in Baltimore, Md, and the Piedmont (25.0% and 22.8%, respectively; W. Eaton, PhD, written communication, October 1990) are, in fact, only modestly lower than those reported here. Our prev¬ alence rates of phobia may accurately reflect those found in the south-central Atlantic coast of the United States.

Fourth, a lifetime history of phobia was assessed in this

study at a single point in time. Although the interrater reliability of our assessment was high, the test-retest re¬ liability may be considerably lower.59 If uncorrelated in twin pairs, unreliability of measurement is indistinguish¬ able from the effects of individual-specific environment in

twin models. If our twin models were "corrected" for the effect of unreliability, the estimates of the heritability of liability to phobias might be considerably higher than those reported here. Fifth, several times in our multivariate genetic analysis, one model was chosen over another on the basis of only slight statistical superiority. Therefore, some skepticism is indicated in acceptance of our final solution. Two alter¬ native multivariate models fit the data nearly as well: a common pathway model, with a phobia general factor influenced by genes and individual-specific environment, and an independent pathway model, with familial and individual-specific environmental common factors. Finally, the results presented herein are in several ways incomplete. We plan in the future to examine the follow¬ ing, from a genetic and epidemiologie perspective: (1) the relationship between phobias and unreasonable fears, (2) the interrelationship of phobias, panic, and generalized anxiety disorder, (3) the relationship between bloodinjury phobias (being assessed at a follow-up interview) and the other phobic subtypes, (4) possible precursors for phobic disorders, including personality, social class, rear¬ ing experiences, and premature parental loss, and (5) the nature of parent-offspring transmission of the predispo¬ sition to phobias. our

This work was supported by grant MH-40828 from the National Institute of Mental Health, Bethesda, Md. The Virginia Twin Regis¬ try, established and maintained by W. Nance, MD, PhD, and L. Co¬ rey, PhD, is supported by the National Institutes of Health grants HD-26746 and NS-25630. Ellen Walters, MS, assisted in the data analysis, and Becky Gan¬ der, MA, assisted in the manuscript preparation. Data were collected under the direction of Patsy Waring, assisted by Lou Hopkins. In¬ terviews were conducted by J. Barwick, MSW, M. J. Beaman, MSW, B. Berry, MSW, C. Brankley, C. Browder, E. Carpenter, PhD, E. Fay, MSW, S. Hall, MSW, K. Hough, MSW, J. Johnson, V. Johnson, MSW, M. Kim, R. Lemon, M. Peery, H. Smith, MSW, W. Suddith, MSW, E. Tille«, and J. Young, MSW. References

JB, Rayner B. Conditioned emotional reactions. J Exp Psychol. 1920;3:1-14. 2. Marks IM. Fears, Phobias, and Rituals. New York, NY: Oxford University Press; 1987. 1. Watson

3. Rachman S. The conditioning theory of fear-acquisition: a critical examination. Behav Res Ther. 1977;15:375-387. 4. Eysenck HJ. The conditioning model of neurosis. Behav Brain Sci.

1979;2:155-166.

5. Bandura A. Social Foundations of

Cognitive Theory. Englewood Cliffs, NJ:

Thought and Action: A Social Prentice-Hall International Inc;

1986. 6. Burke

CB, Burke JD, Regier DA, Rae DS. Age at onset of selected mental disorders in five community populations. Arch Gen Psychiatry.

1990;47:511-518.

7. Darwin C. A biographical sketch of an infant. Mind. 1877;2:285\x=req-\ 294. 8. Ohman A. Face the beast and fear the face: animal and social fears as prototypes for evolutionary analyses of emotion. Psychophysiology.

1986;23:123-145.

9. Seligman MEP. Phobias and preparedness. Behav Ther. 1971;2:307\x=req-\ 320. 10. Gray JA. The Neuropsychology of Anxiety. New York, NY: Oxford University Press; 1982. 11. Hartl DL. Principles of Population Genetics. Sunderland, Mass: Sinauer Associates Inc Publishers; 1980. 12. Mather K. Variability and selection. Proc R Soc Ser B. 1966;164:328\x=req-\ 340.

Downloaded From: http://archpsyc.jamanetwork.com/ by a Monash University Library User on 06/18/2015

13. Lande R. The maintenance of genetic variability by mutation in a polygenic character with linked loci. Genet Res. 1976;26:221-235. 14. Torgersen S. The nature and origin of common phobic fears. Br J Psychiatry. 1979;134:343-351. 15. Phillips K, Fulker DW, Rose RJ. Path analysis of seven fear factors in adult twin and sibling pairs and their parents. Genet Epidemiol.

1987;4:345-355. 16. Rose RJ, Ditto WB. A developmental-genetic analysis of common

fears from

early

adolescence to

early

adulthood. Child Dev.

1983;54:361-368. 17. Rose RJ, Miller JZ, Pogue-Geile MF, Cardwell GF. Twin-family studies of common fears and phobias. In: Gedda L, Parisi P, Nance WE, eds. Twin Research, 3: Intelligence, Personality, and Development. New York, NY: Alan R Liss Inc; 1981:169-174. 18. Neale MC, Fulker DW. A bivariate path analysis of fear data on twins and their parents. Acta Genet Med Gemellol. 1984;33:273-286. 19. Solyom L, Beck P, Solyom C, Hugel R. Some etiological factors in phobic neurosis. Can Psychiatr Assoc J. 1974;19:69-78. 20. Burglass D, Clarke J, Henderson AS, Kreitman M, Presley AS. A study of agoraphobic housewives. Psychol Med. 1977;7:73-86. 21. Moran C, Andrews G. The familial occurrence of agoraphobia. Br J Psychiatry. 1985;146:262-267. 22. Reich J, Yates W. Family history of psychiatric disorders in social phobia. Compr Psychiatry. 1988;29:72-75. 23. Noyes R, Crowe RR, Harris EL, Hamra BJ, McChesney CM,

Chaudhry DR. Relationship between panic disorder and agoraphobia. Psychiatry. 1986;43:227-232. 24. Fyer AJ, Mannuzza S, Gallops MS, Martin LY, Aaronson C, Gorman JM, Liebowitz MR, Klein DF. Familial transmission of simple phobias and fears: a preliminary report. Arch Gen Psychiatry. 1990;47:252\x=req-\

Arch Gen

256. 25. Carey G, Gottesman II. Twin and family studies of anxiety, phobic and obsessive disorders. In: Klein DF, Rabkin JG, eds. Anxiety: New Research and Changing Concepts. New York, NY: Raven Press; 1981. 26. Torgersen S. Genetic factors in anxiety disorders. Arch Gen Psy-

chiatry. 1983;40:1085-1089. 27. Barlow DH. Anxiety and Its Disorders. New York, NY: Guilford Press; 1988. 28. Klein DF. Anxiety reconceptualized. Compr Psychiatry. 1980;21: 411-427. 29. Ost LG. Age of onset in different phobias. J Abnorm Psychol. 1987;96:223-229. 30. Schapira K, Kerr TA, Roth M. Phobias and affective illness. Br J Psychiatry. 1970;117:25-32. 31. Boyd JH, Burke JD, Gruenberg E, Holzer CE III, Rae DS, George

LK, Karno M, Stoltzman R, McEvoy L, Nestadt G. Exclusion criteria of DSM-III: a study of co-occurrence of hierarchy-free syndromes. Arch Gen Psychiatry. 1984;41:983-989. 32. Kendler KS, Neale MC, Kessler RC, Heath AC, Eaves LJ. A population based twin study of major depression in women: the impact of varying definitions of illness. Arch Gen Psychiatry. In press. 33. Eaves LJ, Eysenck HJ, Martin NG, Jardine R, Heath AC, Feingold L, Young PA, Kendler KS. Genes, Culture and Personality: An Empirical Approach. London, England: Oxford University Press; 1989. 34. Spence JE, Corey LA, Nance WE, Marazita ML, Kendler KS, Schieken RM. Molecular analysis of twin zygosity using VNTR DNA probes. Am J Hum Genet. 1988;43:159. 35. Spitzer RL, Williams JB, Gibbon M. Structured Clinical Interview for DSM-III-R. New York, NY: Biometrics Research Division, New York State Psychiatric Institute; 1987. 36. Robins LN, Helzer JE. Diagnostic Interview Schedule (DIS): Ver-

sion III-A St Louis, Mo: Washington University School of Medicine; 1985. 37. American Psychiatric Association, Committee on Nomenclature and Statistics. Diagnostic and Statistical Manual of Mental Disorders, Third Edition. Washington, DC: American Psychiatric Association; 1980. 38. Cohen J. A coefficient of agreement for nominal scales. Ed Psychol Meas. 1960;20:37-46. 39. Heath AC, Neale MC, Hewitt JK, Eaves LJ, Fulker DW. Testing structural equation models for twin data using LISREL. Behav Genet.

1989;19:9-35. 40. Neale MC, Heath AC, Hewitt JK, Eaves LJ, Fulker DW. Fitting genetic models with LISREL: hypothesis testing. Behav Genet. 1989;19:37\x=req-\

49. 41. Kendler

KS, Heath AC, Martin NG, Eaves LJ. Symptoms of anxiety and symptoms of depression: same genes, different environments? Arch Gen Psychiatry. 1987;44:451-457. 42. Joreskog KG, Sorbom D. PRELIS: A Preprocessor for LISREL. Mooresville, Ind: Scientific Software; 1986. 43. Joreskog KG, Sorbom D. LISREL 7: A Guide to the Program and Applications. 2nd ed. Chicago, III: SPSS Inc; 1989. 44. Akaike H. Factor analysis and AIC. Psychometrika. 1987;52:317\x=req-\

332. 45. SAS Institute. SAS User's Guide: Statistics, Version 5. Cary, NC: SAS Institute Inc; 1985. 46. Himle JA, McPhee K, Cameron OG, Curtis GC. Simple phobia: evidence for heterogeneity. Psychiatr Res. 1989;28:25-30. 47. Amies PL, Gelder MG, Shaw PM. Social phobia: a comparative clinical study. Br J Psychiatry. 1983;142:174-179. 48. Kendler KS, Neale MC, Kessler RC, Heath AC, Eaves LJ. Generalized anxiety disorder in women: a population-based twin study. Arch Gen Psychiatry. 1992;49:267-272. 49. Kendler KS, Holm NV. Differential enrollment in twin registries: its effect on prevalence and concordance rates and estimates of genetic parameters. Acta Genet Med Gemellol. 1985;34:125-140. 50. McGuffin P, Katz R. The genetics of depression and manic\x=req-\ depressive illness. Br J Psychiatry. 1989;155:294-304. 51. Kendler KS. Overview: a current perspective on twin studies of schizophrenia. Am J Psychiatry. 1983;140:1413-1425. 52. Ost LG, Hugdahl K. Acquisition of phobias and anxiety response patterns in clinical patients. Behav Res Ther. 1981;19:439-447. 53. Martin NG, Eaves LJ, Kearsey MJ, Davies P. The power of the classical twin study. Heredity. 1978;40:97-116. 54. Eaton WW, Dryman A, Weissman MM. Panic and phobia. In: Robins LN, Regier DA, eds. Psychiatric Disorders in America. New York, NY: Free Press; 1990. 55. Robins LN, Helzer JE, Weissman MM, Orvaschel H, Gruenberg E, Burke JD, Regier DA. Lifetime prevalence of specific psychiatric disorders in three sites. Arch Gen Psychiatry. 1984;41:949-958. 56. Canino GJ, Bird HR, Shrout PE, Rubio-Stipec M, Bravo M, Martinez R, Sesman M, Guevara LM. The prevalence of specific psychiatric disorders in Puerto Rico. Arch Gen Psychiatry. 1987;44:727-735. 57. Bland RC, Orn H, Newman SC. Lifetime prevalence of psychiatric disorders in Edmonton. Acta Psychiatr Scand. 1988;77(suppl 338):24\x=req32. 58. Wells JS, Bushnell JA, Hornblow AR, Joyce PR, Oakley-Browne MA. Christchurch psychiatric epidemiology study, part I: methodology and lifetime prevalence for specific psychiatric disorders. Aust NZ J

Psychiatry. 1989;23:315-326.

59. Mannuzza S, Fyer AJ, Martin LY, Gallops MS, Endicott J, Gorman J, Liebowitz MR, Klein DF. Reliability of anxiety assessment, I: diagnostic agreement. Arch Gen Psychiatry. 1989;46:1093-1101.

Downloaded From: http://archpsyc.jamanetwork.com/ by a Monash University Library User on 06/18/2015

The genetic epidemiology of phobias in women. The interrelationship of agoraphobia, social phobia, situational phobia, and simple phobia.

In 2163 personally interviewed female twins from a population-based registry, the pattern of age at onset and comorbidity of the simple phobias (anima...
2MB Sizes 0 Downloads 0 Views