DEPRESSION AND ANXIETY 31:756–764 (2014)

Research Article
RISK FACTORS FOR ANXIETY DISORDERS: COMMON
AND SPECIFIC EFFECTS IN A NATIONAL SAMPLE
Carlos Blanco, M.D., Ph.D.,1 ∗ José Rubio, M.D.,1 Melanie Wall, Ph.D.,1,2 Shuai Wang, Ph.D.,1
Chelsea J. Jiu, M.D.,1 and Kenneth S. Kendler, M.D.3

Background: Anxiety disorders and major depressive disorder (MDD) often co-
occur and share a broad range of risk factors. The goal of this study was to examine
whether the co-occurrence of anxiety disorders and MDD could be explained by an
underlying latent factor and whether the risk factors exert their effect exclusively
through this factor, directly on each disorder, or through a combination of effects
at both levels. Methods: Data were drawn from a large, nationally representative
sample. Confirmatory factor analysis was used to identify the latent structure
of anxiety disorders. A multiple indicators multiple causes (MIMIC) approach
was used to assess the common and specific effects of risk factors for anxiety
disorders. Results: A one-factor model provided a good fit to the co-occurrence
of anxiety disorders. Low self-esteem, family history of depression, female sex,
childhood sexual abuse, White race, years of education, number of traumatic
experiences, and disturbed family environment increased the risk of anxiety
disorders and MDD through their effect on the latent factor. There were also
several direct effects of the covariates on the disorders, indicating that the effect of
the covariates differed across disorders. Conclusions: Risk for anxiety disorders
and MDD appears to be mediated partially by a latent variable underlying
anxiety disorders and MDD, and partially by disorder-specific effects. These
findings may contribute to account for the high rates of comorbidity among
disorders, identify commonalities in the etiologies of these disorders, and provide
clues for the development of unified preventive interventions. Depression and
Anxiety 31:756–764, 2014. 
C 2014 Wiley Periodicals, Inc.

Key words: risk factors; anxiety disorders; common effects; specific effects;
conceptual model

A nxiety disorders and major depressive disorder

(MDD) are among the most common mental disor-
ders in the United States,[1] with high individual and
1 Department of Psychiatry, Columbia University, New York societal burden derived from their considerable asso-
State Psychiatric Institute, New York, New York
2 Department of Biostatistics, Mailman School of Public Health,
ciated work and social impairment.[2, 3] Decreasing the
burden of anxiety disorders and MDD is an important
Columbia University, New York, New York
3 Department of Psychiatry, Virginia Institute for Psychiatric public health priority. Anxiety disorders often co-occur
and Behavioral Genetics, Virginia Commonwealth University,
with each other and with MDD[4, 5] suggesting the ex-
Richmond, Virginia istence of a latent structure that may parsimoniously
Contract grant sponsor: NIH; Contract grant numbers: DA019606,
DA023200, MH076051, MH082773. Received for publication 11 July 2013; Revised 12 November 2013;
∗ Correspondence Accepted 18 January 2014
to: Carlos Blanco, Department of Psychiatry,
College of Physicians and Surgeons of Columbia University, DOI 10.1002/da.22247
1051 Riverside Drive, Unit 69, New York, NY 10032. E-mail: Published online 27 February 2014 in Wiley Online Library
[email protected] (wileyonlinelibrary.com).


C 2014 Wiley Periodicals, Inc.
 Research Article: Risk Factors for Anxiety Disorders 757

explain these patterns of comorbidity[6–9] . Several stud- MEASURES

ies have examined the genetic liability of anxiety disor- DSM-IV Anxiety Disorders and MDD. All psychiatric diag-
ders and MDD, and consistently concluded that most noses were made according to DSM-IV-TR criteria[35] using the Al-
of this genetic liability is exerted at the level of the la- cohol Use Disorder and Associated Disabilities Interview Schedule—
tent variable,[8, 10–13] rather than at the disorder-specific DSM-IV Version (AUDADIS-IV), Wave 2 version.[36, 37] The lifetime
level. Because anxiety disorders and MDD share a broad DSM-IV anxiety disorders included panic disorder, social anxiety dis-
range of other risk factors,[14–24] an essential step in the order (SAD), specific phobia, PTSD and generalized anxiety disorder
(GAD). Obsessive-compulsive disorder was not assessed in the NE-
development of more effective treatment and prevention
SARC, and thus was not included in this study. By contrast, based
interventions is to determine whether those factors exert
on its high comorbidity with anxiety disorders and to be consistent
their effect through the common latent variable, directly with and build on Kendler’s original model, we included MDD in our
on each disorder, or through a combination of effects at study. AUDADIS-IV has shown fair to good test–retest reliability in
both levels.[14, 16, 25] the general population for anxiety disorders and MDD.[33, 38]
We sought tsreo address this gap in knowledge by Conceptual Model. In accord with prior genetic and epidemi-
drawing on data from a large, nationally representative ologic research, we conceptualized the individual anxiety disorders
survey to compare the common and the specific con- as indicators of an underlying latent variable and sought to examine
tribution of a set of observable risk factors to risk of whether the risk factors exerted their effect through this latent vari-
anxiety disorders and MDD. We followed a concep- able or through direct effects on the disorders. Also consistent with
tual model of risk factors, which was initially developed previous research,[26–28] we selected our risk factors based on a con-
ceptual model for risk factors, which addresses the etiologic complexity
by Kendler and colleagues to examine the prevalence of internalizing disorders.[14–24] The risk factors included were fam-
of MDD,[26, 27] but seems to be applicable to other in- ily history of depression,[33, 38] low parental warmth (assessed with the
ternalizing disorders.[28] Because anxiety disorders often neglect items of the Child Trauma Questionnaire[39] ), parental loss be-
co-occur with each other and with MDD and seem to fore age 18, disturbed family environment (operationalized, as in pre-
have a partially shared liability,[8, 13, 14] prior to our anal- vious studies, as parental absence or separation from a biological parent
yses, we hypothesized that the comorbidity structure of before age 18),[40, 41] childhood sex abuse (also measured with items
anxiety disorders and MDD would be well described by from the Child Trauma Questionnaire), history of conduct disorder
a single latent variable and that the risk factors postu- (assessed with the AUDADIS),[15] low self-esteem (a binary variable
lated in Kendler’s model would exert part of their ef- considered present if probands believed they were not as good, smart,
or attractive as most other people), number of traumatic experiences
fect through this latent variable. However, because the
prior to age 21, history of substance use disorder (SUD) prior to age
prevalence and course of the different anxiety disorder 21 (assessed with the AUDADIS, using methods previously reported
and MDD does not fully overlap[29–31] and they also have by our group),[42, 43] and years of education (measured by self-report).
nonshared variance, we also hypothesized that the risk The model also controlled for race/ethnicity and sex.
factors would also partially exert their effect directly on To minimize the risk that the results were due to reverse causality
the individual disorders. that could arise if the onset of the disorders preceded the occurrence
of the risk factors, we repeated our analyses restricting our sample to
individuals whose onset of anxiety disorders was after age 21, to ensure
that all risk factors were present before the onset of any anxiety disor-
METHODS der. Because the results of both analyses are very similar, we present
SAMPLE the results of the full sample and indicate differences with the results of
the restricted sample. Full results of the restricted sample are available
The National Epidemiologic Survey on Alcohol and Related Con-
upon request.
ditions (NESARC) sample is a nationally representative sample of the
adult population of the United States. The NESARC target population
was the civilian noninstitutionalized population consisting of 18 years STATISTICAL ANALYSIS
and older residing in households and group living quarters, includ- To assess the association between all risk factors and the five anx-
ing residents of the continental United States, Columbia, Alaska, and iety disorders and MDD, separate logistic regressions were run for
Hawaii. The 2004–2005 Wave 2 NESARC is the second wave longi- each disorder including all risk factors simultaneously, and odds ratios
tudinal follow-up of the Wave 1 NESARC, conducted in 2001–2002. (ORs) and 95% confidence intervals are presented. Given previous
The first wave included face-to-face interviews with 43,093 respon- results indicating the five anxiety disorders and MDD load on a sin-
dents, yielding an overall response rate of 81.0%.[32, 33] The Wave 2 gle factor, we then performed a confirmatory factor analysis (CFA) to
interview was conducted approximately 3 years later. Excluding indi- assess the fit of a one-factor model. Goodness-of-fit measures includ-
viduals who were ineligible (e.g., deceased), the response rate in Wave ing the comparative fit index (CFI), Tucker–Lewis index (TLI), and
2 was 86.7% (n = 34,653).[32] Wave 2 NESARC data were adjusted for root mean squared error of approximation (RMSEA) were used to as-
nonresponse based on sociodemographic characteristics and presence sess fit. Hu and Bentler recommended CFI and TLI values above .95,
of any lifetime Wave 1 NESARC psychiatric disorder. The adjusted and RMSEA values below .06, as representing good model fit.[44] We
data are representative of the civilian population of the United States also examined the eigenvalues of the tetrachoric correlation matrix to
based on the 2000 Decennial Census.[34] The research protocol, in- quantify what percentage of the variability in the anxiety disorders was
cluding informed consent procedures, received full human subjects explained by a single common factor.
review and approval from the U.S. Census Bureau and the U.S. Of- We used a multiple indicators multiple causes (MIMIC)
fice of Management and Budget. Because several of the risk factors approach[45] to assess the common and specific effects of the different
included in our conceptual model were only measured in Wave 2, the risk factors. The MIMIC model is a special case of structural equa-
sample for this study was composed of all individuals who participated tion modeling (SEM) that includes three sets of relationships: (1) the
in both waves (n = 34,653). measurement model, which describes the relationships between the

Depression and Anxiety

758 Blanco et al.

indicators (in this case, the five anxiety disorders and MDD) and the cated the presence of one common factor that explained
latent factors; (2) the structural model, which describes the relation- 3.1/6 = 52% of the shared variance in the anxiety dis-
ships between the latent factors and the covariates (in this case, the orders and MDD. GAD had the largest loading (0.78),
risk factors); and (3) the relationships between the indicators and the followed by panic disorder (0.70), SAD (0.69), specific
covariates (i.e., risk factors), called the direct effects, because they in-
phobia (0.63), PTSD (0.57), and MDD (0.51; Fig. 1).
dicate effects of the covariates on the indicators that are not mediated
through the factors (i.e., are still present after adjusting for the effect
of the covariates on the latent factors).[46] In this approach, the struc-
tural model represents the effect of the covariates (i.e., risk factors) on COMMON AND SPECIFIC EFFECTS OF
the latent factor (i.e., the effect of the risk factors on the shared latent INDICATORS ON ANXIETY DISORDERS
liability), whereas the direct effects represent the disorder-specific ef-
Overall, the risk factors explained 46% of the variance
fects of the risk factors. For example, if the coefficient of a predictor
(e.g., family history of MDD) on the latent factor is positive, but the
in the common factor for anxiety disorders and MDD.
coefficient of the direct effect on a disorder is negative (e.g., specific After adjusting for other covariates, risk factors that had
phobia), the interpretation would be that predictor increases the over- a statistically significant independent effect on the la-
all risk of disorders represented by the latent factor, but this increase tent variable included low self-esteem, family history of
in risk is smaller for the specific disorder with the negative direct effect MDD, female sex, childhood sexual abuse, White race,
(in this case, specific phobia) than for the other disorders in the factor. years of education, number of traumatic experiences by
By contrast, if the coefficient of the direct effect is also positive (e.g., age 21, and disturbed family environment (Fig. 1).
in the case of MDD), this would mean that family history of MDD Consistent with the varying magnitudes of the ORs
increases the risk of MDD even more than for the other disorders in of the risk factors with different anxiety disorders, there
the factor.
were several significant direct effects, indicating that the
Modification indices were used to identify whether direct (i.e., spe-
cific) effects from risk factors to specific disorders were warranted us-
effect of the risk factors was not completely mediated
ing a conservative cutoff for determining statistical significance fol- through the latent variable (Table 2 and Fig. 1). The
lowing previously established procedures.[47] Specifically, we used a total effects of the covariates on each disorder, which
modification index cutoff of 10 corresponding to a chi-square test with are calculated by combining the indirect (i.e., mediated
1 degree of freedom and a critical value of .001 to identify predictors through the factor) and direct (i.e., without mediation
with significant direct effects. through the factor) effects, are presented in Table 2.
All analyses were conducted in Mplus Version 6.1,[46] which takes For example, the total effect of female gender on SAD
into account the NESARC sampling weights and design effects in all (–0.14) was the result of adding the indirect effect (0.31)
analyses, including parameter as well as standard error estimation and to the direct effect (–0.44), whereas the effect of low self-
model fit calculations. The default estimator for the analysis was the
esteem on SAD was 0.82, resulted from combining the
variance-adjusted weighted least squares (WLSMV), a robust estima-
tor that does not assume normally distributed variables and provides
indirect (0.48) and direct (0.34) effects. In most cases,
the best option for modeling categorical or ordered data. after adjusting for the effect of the other covariates in
the model, the total effects were positive, indicating that
the presence of the risk factor increased the likelihood
RESULTS of having the disorder.
BIVARIATE ASSOCIATIONS BETWEEN RISK
FACTORS AND ANXIETY DISORDERS
ADDITIONAL ANALYSES
In the bivariate analyses, female gender, family his-
tory of MDD, disturbed family environment, childhood The CFA of the subsample with onset of the first anx-
sexual abuse, low self-esteem, and lower educational at- iety disorder after age 21 also indicated that a one-factor
tainment all increased the risk of all anxiety disorders and model had excellent fit χ 2 = 119.22, df = 9, P < .001;
MDD. White race increased the risk of all disorders ex- CFI = .98; RMSEA = .02; SRMR = .05 with all loadings
cept PTSD, conduct disorder increased the risk of SAD over 0.53.
and specific phobia, and the number of traumas before After adjusting for other covariates, all of the risk fac-
age 21 increased the risk of panic disorder, PTSD, and tors that had a significant effect on the latent factor in
GAD. Early parental loss increased the risk of PTSD, the whole sample remained significant in the subsample
but decreased the risk of panic disorder, whereas SUD with onset of the first anxiety disorder after age 21. How-
before age 21 decreased the risk of SAD, specific phobia, ever, modification indices for this model indicated that
and PTSD (Table 1). only 8 of the 19 direct effects that were significant in
the whole sample remained significant in this subsam-
ple: after adjusting for the effects of the covariates on
FACTOR ANALYSIS the common factor, female gender decreased the prob-
Fit indices for the CFA model with one factor were ability of SAD but increased the probability of specific
χ 2 = 130.11, df = 9, P < .001; CFI = .99; RMSEA = phobia and MDD, White race decreased the probability
.02; standardized root mean square residual (SRMR) = of PTSD, family history of MDD decreased the proba-
.03, indicating an excellent fit. All loadings were at least bility of SAD and specific phobia but increased the prob-
0.60, further indicating the adequacy of a single-factor ability of MDD, and low self-esteem increased the prob-
solution. Eigenvalues (3.1, 0.8, 0.6, 0.6, 0.5, 0.4) indi- ability of SAD (data available upon request).
Depression and Anxiety
 Research Article: Risk Factors for Anxiety Disorders 759

TABLE 1. Lifetime anxiety disorders by risk factors. NESARC Wave 2 (n = 34,653)

Panic Social anxiety Specific phobia PTSD GAD MDD

OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI

White race 1.4 (1.2−1.5) 1.4 (1.2−1.5) 1.1 (1.0−1.2) 0.9 (0.8−1.1) 1.4 (1.2−1.5) 1.4 (1.3−1.5)
Family history MDD 3.1 (2.8−3.4) 3.2 (2.9−3.5) 2.2 (2.0−2.4) 2.3 (2.0−2.5) 3.5 (3.1−3.9) 3.3 (3.0−3.5)
Female 2.1 (1.9−2.3) 1.4 (1.3−1.5) 2.1 (2.0−2.3) 2.2 (2.0−2.4) 2.1 (1.9−2.4) 2.1 (2.0−2.3)
Disturbed family 1.1 (1.1−1.1) 1.1 (1.1−1.1) 1.0 (1.0−1.0) 1.1 (1.1−1.1) 1.1 (1.1−1.1) 1.1 (1.1−1.1)
environment
Early parental loss 0.9 (0.7−1.0) 1.0 (0.9−1.2) 1.1 (1.0−1.2) 1.2 (1.0−1.4) 1.1 (0.9−1.3) 1.0 (0.9−1.1)
Childhood sexual abuse 3.4 (3.0−3.8) 2.9 (2.6−3.3) 2.5 (2.2−2.7) 3.9 (3.5−4.4) 3.4 (3.0−3.9) 2.6 (2.3−2.8)
Conduct disorder 1.3 (0.9−2.0) 1.6 (1.1−2.4) 1.5 (1.1−2.0) 1.3 (0.9−2.1) 1.3 (0.8−2.0) 1.5 (1.1−2.1)
SUD before 21 1.1 (1.0−1.3) 0.5 (0.4−0.6) 0.6 (0.5−0.6) 0.7 (0.6−0.8) 1.0 (0.8−1.1) 1.2 (1.1−1.3)
Number of traumas 1.2 (1.2−1.2) 1.0 (1.0−1.1) 1.0 (1.0−1.0) 1.3 (1.3−1.3) 1.2 (1.2−1.2) 1.1 (1.1−1.2)
before 21
Years of education 0.9 (0.9−0.9) 0.8 (0.8−0.8) 0.8 (0.8−0.8) 0.9 (0.9−0.9) 0.9 (0.9−0.9) 0.9 (0.9−1.0)
before 21
Low self-esteem 3.6 (3.2−4.1) 6.3 (5.6−7.0) 2.9 (2.6−3.2) 2.7 (2.4−3.1) 4.2 (3.7−4.7) 2.4 (2.2−2.6)

PTSD, posttraumatic stress disorder; GAD, generalized anxiety disorder; MDD, major depressive disorder. Significance at P < .05. Significant
odds ratios (ORs) are bolded.

DISCUSSION variable; (2) a comprehensive model initially developed

to explain the etiology of MDD could also contribute to
In a large, nationally representative sample, we ex- explain the risk of anxiety disorders; and (3) the risk fac-
amined whether (1) the comorbidity structure of anxiety tors of that model exerted their effect through a common
disorders and MDD was well explained by a single latent latent variable or through disorder-specific effects. We

Figure 1. Model including significant effects of the risk factors on the common factor and direct effects on the disorders.

Depression and Anxiety

 760

Depression and Anxiety

TABLE 2. Direct, indirect and total effects of risk factors on the anxiety disorders

Panic disorder SAD Specific phobia PTSD GAD MDD

Indirect Direct Total Indirect Direct Total Indirect Direct Total Indirect Direct Total Indirect Direct Total Indirect Direct Total

White race 0.15 0.15 0.16 0.16 0.16 0.16 0.12 − 0.14 − 0.02 0.16 0.16 0.11 0.11
Family history 0.35 0.35 0.39 0.39 0.39 − 0.23 0.16 0.30 − 0.13 0.16 0.39 0.39 0.27 0.32 0.59
MDD
Female 0.28 0.28 0.31 − 0.44 − 0.14 0.31 0.31 0.24 0.24 0.31 0.31 0.21 0.20 0.42
Disturbed family 0.04 0.04 0.04 0.04 0.04 − 0.09 − 0.05 0.03 0.03 0.04 0.04 0.03 0.03
environment
Early parental loss
Childhood sexual 0.22 0.22 0.24 0.24 0.24 0.24 0.18 0.18 0.24 0.24 0.17 0.17
abuse
Conduct disordera
Blanco et al.

SUD before 21 − 0.21 − 0.21 0.12 0.12

Number of traumas 0.13 0.13 0.14 − 0.08 0.06 0.14 0.14 0.11 0.16 0.27 0.14 0.14 0.10 0.10
before 21
Years of education − 0.28 0.18 − 0.10 − 0.30 − 0.18 − 0.48 − 0.31 − 0.39 − 0.69 − 0.23 − 0.23 − 0.30 0.20 − 0.11 − 0.21 0.18 − 0.03
before 21
Low self-esteem 0.44 0.44 0.48 0.34 0.82 0.49 − 0.30 0.19 0.37 − 0.16 0.21 0.49 0.49 0.33 0.33

SAD, social anxiety disorder; PTSD, posttraumatic stress disorder; GAD, generalized anxiety disorder. Only significant values (P < .05) are shown. Indirect effects represent the effect of the
predictor on the disorder occurring through the factor. It is obtained by multiplying the loading of the disorder on the factor by the effect of the covariate on the factor. Direct effects represent
the effect of the predictor on the disorder above and beyond the effect of the covariate on the factor. Total effects are the sum of the direct and the indirect effects.
a Conduct disorder did not have any significant direct or indirect effect on any anxiety disorder. For that reason, all cells are left empty.
 Research Article: Risk Factors for Anxiety Disorders 761

found that a single factor described well the comorbidity dition to its effect on the latent variable, indicating that
structure of anxiety disorders and MDD, and that both this risk factor may contribute to the risk of MDD be-
common and disorder-specific effects explained the re- yond the risk it confers to anxiety disorders. Similarly,
lationship between risk factors and anxiety disorders and we found that low self-esteem, which has been previ-
MDD. ously documented among individuals with SAD[55] but
In accord with previous research,[8, 10–13] a single fac- also associated with a broad range of mental disorders,[56]
tor described the latent structure of five anxiety disor- has a stronger association with SAD than with other
ders and MDD well,[48] supporting the existence of a anxiety disorders or MDD. The anxiety about one’s
latent shared risk for these disorders. Previous studies performance and fear of negative evaluation by others
have shown that part of this shared risk is due to shared may lead individuals to internalize some of those per-
genetic and environmental liability.[8, 13, 14] Our study ex- ceptions leading to low self-esteem. Alternatively, low
tends these prior findings by identifying some of the risk self-esteem might make individuals more reliant on the
factors that may underlie those shared liabilities. The risk opinions of others, which may lead to anxiety about their
factors examined in this study were based on a concep- evaluation.[57]
tual model for risk of MDD, and had previously shown Traumatic experiences had a large specific effect on
promising results in also predicting the risk of GAD.[28] PTSD above and beyond their effect on the latent fac-
Our study suggests that the model may have broader tor, whereas being White constituted a specific protec-
applicability than previously thought and may help un- tive factor (i.e., after adjusting for the effect of race on
derstand the etiology of a wider range of internalizing the latent factor), probably as a result of lower exposure
disorders and their high rates of co-occurrence. Further- to trauma of White individuals.[58] A surprising finding
more, the finding that the disorders differed in their load- was that history of SUD before age 21 decreased the
ings on the latent factor indicates that the disorders have risk of PTSD. It is possible that substance use may mask
shared but also specific risk factors. In particular, MDD or decrease the intensity of some anxiety symptoms.
had the lowest loading, suggesting that MDD may stand Alternatively, early-onset SUD may indicate a greater
a bit apart from the anxiety disorders, as previously sug- propensity for externalizing disorders than for some in-
gested. ternalizing disorders. Despite these isolated findings,
In accord with previous studies,[15, 28, 49] when exam- overall, the specific effects that were significant in our
ined separately in the bivariate analyses, most risk factors analyses are consistent with specific risk factors identi-
increased the odds of all the anxiety disorders and MDD, fied in previous studies,[56–60] and are informative about
suggesting that they might be common risk factors for how risk factors may shape individual disorders by in-
these disorders. Consistent with this view, the results of volving mechanisms that may be less salient for other
the MIMIC model indicated that most of the effects of disorders.
the risk factors were exerted through the latent variable Our findings have etiological, treatment, and preven-
representing the shared liability for all the disorders. In tive implications. From the etiological point of view,
fact, in some cases, such as childhood sexual abuse, the our findings suggest that risk factors may exert their
effect was exclusively through the latent variable, in line effects, at least in part, through broad dimensions of
with prior findings on the effect of child maltreatment psychopathology, consistent with the views of the Re-
on the structure of psychiatric disorders.[50] For exam- search Domain Criteria (RDoC) project,[61, 62] and that
ple, through its effect on the latent factor, a history of those effects may differ across disorders. Future research
childhood sexual abuse increased by 0.24 standard devi- should seek to uncover the genetic, cellular, and circuit-
ations the risk of having SAD, GAD, or specific phobia, level underpinnings of those psychopathological dimen-
by 0.22 standard deviations the risk of having panic dis- sions and their overlapping and distinct contribution to
order, 0.18 standard deviations the risk of PSTD, and anxiety disorders and MDD. Furthermore, the large ef-
0.17 standard deviations the risk of MDD. There is a fects of risk factors on the latent variable underlying
need to better understand the psychological and biolog- anxiety disorders and MDD suggest that their shared
ical mechanisms underlying the generalized effects of liability[63] may contribute to explain the high comor-
these risk factors.[51–54] bidity among them.[13, 14, 64] From the interventions per-
At the same time, as could be expected from the dif- spective, the large impact of the risk factors on the la-
ferent magnitudes of the association between the risk tent structure of anxiety disorders and MDD suggests
factors and the different disorders observed in the bi- that preventive interventions on a single risk factor with
variate analyses, not all the effects of the risk factors oc- large effect on the latent structure could simultaneously
curred through the latent variable. Nineteen of the pos- decrease the risk for all the disorders. At the same time,
sible 66 disorder-specific effects were significant, and for since several risk factors affect the latent structure jointly,
each disorder, there was at least one significant direct ef- interventions that address several risk factors simultane-
fect. Our results indicate that, in addition to the shared ously are likely to be the most effective. The presence
etiology that pulls them together, all anxiety disorders of disorder-specific effects suggests that supplementary
and MDD have specific effects of risk factors that con- modules may be necessary to provide intervention flexi-
tribute to their individuality. For example, family history bility, promote personalized medicine, and optimize the
of MDD had a substantial specific effect on MDD in ad- outcome of all individuals.

Depression and Anxiety

762 Blanco et al.

Several limitations should be considered in the inter- anxiety disorders: results from the National Epidemiologic Sur-
pretation of these results. First, because of the cross- vey on Alcohol and Related Conditions. Arch Gen Psychiatry
sectional nature of the study, although associations iden- 2004;61:807–816.
tified in this study could be informative regarding the 5. Rodriguez BF, Bruce SE, Pagano ME, Keller MB. Relationships
among psychosocial functioning, diagnostic comorbidity, and the
etiology of anxiety disorders and MDD, causality can-
recurrence of generalized anxiety disorder, panic disorder, and
not be inferred. However, because our results remain
major depression. J Anxiety Disord 2005;19:752–766.
almost unchanged when we restricted the analyses to in- 6. Krueger RF. The structure of common mental disorders. Arch
dividuals 21 and older, reverse causality is unlikely to Gen Psychiatry 1999;56:921–926.
explain our results. Second, although our model is com- 7. Slade T, Watson D. The structure of common DSM-IV and ICD-
prehensive, it is not exhaustive. In order to be parsimo- 10 mental disorders in the Australian general population. Psychol
nious, in accord with established recommendations for Med 2006;36:1593–1600.
model[65] development, we limited the number of vari- 8. Tambs K, Czajkowsky N, Roysamb E, et al. Structure of genetic
ables based on a priori conceptual model.[26, 28] Future and environmental risk factors for dimensional representations of
investigations could examine the common and specific DSM-IV anxiety disorders. Brit J Psychiat 2009;195:301–307.
effects of other potential risk factors. Third, our sample 9. Vollebergh WAM, Iedema J, Bijl RV, et al. The structure and
stability of common mental disorders: the NEMESIS study. Arch
was composed of individuals 18 and older. The model Gen Psychiatry 2001;58:597–603.
may not be applicable to younger individuals or indi- 10. Kendler KS, Heath AC, Martin NG, Eaves LJ. Symptoms of anx-
viduals of other cohorts. Fourth, some variables, such iety and symptoms of depression: same genes, different environ-
as childhood sexual abuse, were assessed retrospectively ments? Arch Gen Psychiatry 1987;44:451–457.
and may be subject to recall bias, whereas others, such as 11. Andrews G, Stewart G, Allen R, Henderson A. The genetics of six
low self-esteem, were assessed with a single item. This neurotic disorders: a twin study. J Affect Disord 1990;19:23–29.
may have increased error variance and underestimated 12. Scherrer JF, True WR, Xian H, et al. Evidence for genetic influ-
the strength of association of the risk factors with the ences common and specific to symptoms of generalized anxiety
outcomes. Therefore, our estimates are likely to be con- and panic. J Affect Disord 2000;57:25–35.
13. Hettema JM, Prescott CA, Myers JM, et al. The structure of ge-
servative. Fifth, some variables of potential interest, such
netic and environmental risk factors for anxiety disorders in men
as family history of anxiety disorders, were not assessed and women. Arch Gen Psychiatry 2005;62:182–189.
in the NESARC and could not be included in the model. 14. Hettema JM, Neale MC, Kendler KS. A review and meta-analysis
In conclusion, risk for anxiety disorders and MDD ap- of the genetic epidemiology of anxiety disorders. Am J Psychiatry
pears to be mediated partially by a latent variable under- 2001;158:1568–1578.
lying these disorders, and partially by disorder-specific 15. Morcillo C, Duarte CS, Sala R, et al. Conduct disorder and adult
effects. The shared risk may contribute to the large rates psychiatric diagnoses: associations and gender differences in the
of comorbidity among disorders, help to identify com- US adult population. J Psychiatr Res 2012;46:323–330.
monalities in the etiologies of these disorders, and pro- 16. Kessler RC, McLaughlin KA, Green JG, et al. Childhood ad-
vide clues for the development of unified treatment pro- versities and adult psychopathology in the WHO World Mental
Health Surveys. Brit J Psychiat 2010;197:378–385.
tocols or integrated preventive interventions.[66] The
17. Magee WJ, Eaton WW, Wittchen HU, et al. Agoraphobia, simple
disorder-specific effects may contribute toward the dis- phobia, and social phobia in the National Comorbidity Survey.
tinct aspects of the etiology of each disorder and suggest Arch Gen Psychiatry 1996;53:159–168.
the need for intervention flexibility. These findings may 18. Marmorstein NR. Anxiety disorders and substance use disor-
help inform dimensional models of classification of psy- ders: different associations by anxiety disorder. J Anxiety Disord
chiatric disorders. 2012;26:88–94.
19. McEvoy PM, Grove R, Slade T. Epidemiology of anxiety dis-
orders in the Australian general population: findings of the 2007
Acknowledgments. This study was supported by
Australian National Survey of Mental Health and Wellbeing. Aust
NIH grants DA019606, DA023200, MH076051, and NZ J Psychiat 2011;45:957–967.
MH082773 (CB) and the New York State Psychiatric 20. Merikangas KR, Zhang H, Avenevoli S, et al. Longitudinal trajec-
Institute (CB and MW). tories of depression and anxiety in a prospective community study:
the Zurich Cohort Study. Arch Gen Psychiatry 2003;60:993–
1000.
REFERENCES 21. Mulhare ES, Ghesquiere A, Shear AK. Attachment, separation and
anxiety disorders. In: Simpson HB, Neria Y, Lewis-Fernandez
1. Kessler RC, Berglund P, Demler O, et al. Lifetime prevalence R, Schneier F, editors. Anxiety Disorders: Theory, Research
and age-of-onset distributions of DSM-IV disorders in the Na- and Clinical Perspectives. New York, NY: Cambridge University
tional Comorbidity Survey Replication. Arch Gen Psychiatry Press; 2010:227–238.
2005;62:593–602. 22. Pietrzak RH, Goldstein RB, Southwick SM, Grant BF. Prevalence
2. Wittchen HU. Generalized anxiety disorder: prevalence, burden, and axis I comorbidity of full and partial posttraumatic stress dis-
and cost to society. Depress Anxiety 2002;16:162–171. order in the United States: results from Wave2 of the National
3. World Health Organization, editor. World Health Report Epidemiologic Survey on Alcohol and Related Conditions. J Anx-
2001: Mental Health: New Understanding, New Hope. Geneva, iety Disord 2010;25:456–465.
Switzerland: WHO; 2001. 23. Richards M, Hardy R, Wadsworth M. The effects of divorce and
4. Grant BF, Stinson FS, Dawson DA, et al. Prevalence and co- separation on mental health in a national UK birth cohort. Psychol
occurrence of substance use disorders and independent mood and Med 1997;27:1121–1128.

Depression and Anxiety

 Research Article: Risk Factors for Anxiety Disorders 763

24. Zoccolillo M. Co-occurrence of conduct disorder and its adult 41. Blanco C, Vesga-Lopez O, Stewart JW, et al. Epidemiology of
outcomes with depressive and anxiety disorders: a review. J Am major depression with atypical features: results from the National
Acad Child Adolesc Psychiatry 1992;31:547–556. Epidemiologic Survey on Alcohol and Related Conditions (NE-
25. Brandes M, Bienvenu OJ. Personality and anxiety disorders. Curr SARC). J Clin Psychiatry 2012;73:224–232.
Psychiatry Rep 2006;8:263–269. 42. Lopez-Quintero C, Hasin DS, De Los Cobos JP, et al. Probabil-
26. Kendler KS, Gardner CO, Prescott CA. Toward a comprehen- ity and predictors of remission from life-time nicotine, alcohol,
sive developmental model for major depression in women. Am J cannabis or cocaine dependence: results from the National Epi-
Psychiatry 2002;159:1133–1145. demiologic Survey on Alcohol and Related Conditions. Addiction
27. Kendler KS, Prescott CA. Genes, Environment and Psy- 2011;106:657–669.
chopathology: Understanding the Causes of Psychiatric and Sub- 43. Lopez-Quintero C, Cobos JP, Hasin DS, et al. Probability and
stance Use Disorders. New York: The Guilford Press; 2006. predictors of transition from first use to dependence on nicotine,
28. Alegria AA, Hasin DS, Nunes EV, et al. Comorbidity of gener- alcohol, cannabis, and cocaine: results of the National Epidemi-
alized anxiety disorder and substance use disorders: results from ologic Survey on Alcohol and Related Conditions (NESARC).
the National Epidemiologic Survey on Alcohol and Related Con- Drug Alcohol Depend. 2011;115:120–130.
ditions. J Clin Psychiatry 2010;71:1187–1195. 44. Hu LT, Bentler PM, Kano Y. Can test statistics in covariance
29. Noyes R, Holt CS, Woodman CL. Natural course of anxiety dis- structure analysis be trusted? Psychol Bull 1992;112:351–362.
orders. In: Mavissakalian M, Prien R, editors. Long-Term Treat- 45. Kim ES, Yoon M, Lee T. Testing measurement invariance using
ments of Anxiety Disorders. Washington, DC: American Psychi- MIMIC: likelihood ratio test with a critical value adjustment. Educ
atric Press; 1996: 1–48. Psychol Meas 2011;72:469–492.
30. Blanco C, Xu Y, Schneier FR, et al. Predictors of persistence 46. Muthen LK, Muthen BO. Mplus User’s Guide. Los Angeles, CA:
of social anxiety disorder: a national study. J Psychiatr Res Muthen & Muthen; 1998–2006.
2011;45:1557–1563. 47. Stark S, Chernyshenko OS, Drasgow F. Detecting differen-
31. Yonkers KA, Bruce SE, Dyck IR, Keller MB. Chronicity, relapse, tial item functioning with confirmatory factor analysis and item
and illness—course of panic disorder, social phobia, and general- response theory: toward a unified strategy. J Appl Psychol
ized anxiety disorder: findings in men and women from 8 years of 2006;91:1292–1306.
follow-up. Depress Anxiety 2003;17:173–179. 48. Blanco C, Rubio JM, Wall MM, et al. The latent structure and
32. Grant B, Moore, TC, Shepard J, Kaplan K. Source and Accuracy comorbidity patterns of generalized anxiety disorder and major
Statement: Wave 1. National Epidemiologic Survey on Alcohol depressive disorder: a national study. Depress Anxiety 2014;31:
and Related Conditions (NESARC). Bethesda, MD: National In- 214–222.
stitute on Alcohol Abuse and Alcoholism; 2003. 49. Pérez-Fuentes GOM, Villegas L, Morcillo C, Wang S, Blanco C.
33. Grant BF, Dawson DA, Stinson FS, et al. The Alcohol Use Prevalence and correlates of child sexual abuse: a national study.
Disorder and Associated Disabilities Interview Schedule-IV Compr Psychiat 2013;54:16–27.
(AUDADIS-IV): reliability of alcohol consumption, tobacco use, 50. Keyes KM, Eaton NR, Krueger RF, et al. Childhood maltreatment
family history of depression and psychiatric diagnostic modules in and the structure of common psychiatric disorders. Br J Psychiatry
a general population sample. Drug Alcohol Depend 2003;71:7–16. 2012;200:107–115.
34. Grant B, Moore TC., Shepard J, Kaplan K. Source and Accuracy 51. Heim C, Nemeroff CB. The impact of early adverse experiences
Statement: Wave 2. National Epidemiologic Survey on Alcohol on brain systems involved in the pathophysiology of anxiety and
and Related Conditions (NESARC). Bethesda, MD: National In- affective disorders. Biol Psychiatry 1999;46:1509–1522.
stitute on Alcohol Abuse and Alcoholism; 2005. 52. Heim C, Nemeroff CB. The role of childhood trauma in the neu-
35. American Psychiatric Association. Diagnostic and Statistical Man- robiology of mood and anxiety disorders: preclinical and clinical
ual of Mental Disorders. Washington, DC: American Psychiatric studies. Biol Psychiatry 2001;49:1023–1039.
Association; 2000. 53. Blanco C, Hasin DS, Liu SM, et al. Mapping common psychi-
36. Grant BF, Dawson DA, Hasin DS, editors. The Wave 2 National atric disorders: structure and predictive validity in the National
Epidemiologic Survey on Alcohol and Related Conditions Alcohol Epidemiologic Survey on Alcohol and Related Conditions. JAMA
Use Disorder and Associated Disabilities Interview Schedule— Psychiatry 2013;70:199–208.
DSM-IV Version. Bethesda, MD: National Institute on Alcohol 54. Kendler KS, Prescott CA, Myers J, Neale MC. The structure of
Abuse and Alcoholism; 2004. genetic and environmental risk factors for common psychiatric and
37. Grant B, Dawson D, Hasin D. The Alcohol Use Disorder substance use disorders in men and women. Arch Gen Psychiatry
and Associated Disabilities Interview Schedule-DSM-IV Version. 2003;60:929–937.
Bethesda, MD: National Institute on Alcohol Abuse and Alco- 55. Acarturk C, Smit F, De Graaf R, et al. Incidence of social phobia
holism; 2001. and identification of its risk indicators: a model for prevention.
38. Ruan W, Goldstein RB, Chou SP, et al. The Alcohol Use Disorder Acta Psychiat Scand 2009;119:62–70.
and Associated Disabilities Interview Schedule-IV (AUDADIS- 56. Silverstone PH. Low self-esteem in different psychiatric condi-
IV): reliability of new psychiatric diagnostic modules and risk tions. Brit J Clin Psychol 1991;30:185–188.
factors in a general population sample. Drug Alcohol Depend 57. Silverstone PH, Salsali M. Low self-esteem and psychiatric pa-
2008;92:27–36. tients: part I—the relationship between low self-esteem and psy-
39. Bernstein DP, Ahluvalia T, Pogge D, Handelsman L. Validity of chiatric diagnosis. Ann Gen Psychiatry 2003;2:2.
the Childhood Trauma Questionnaire in an adolescent psychiatric 58. Breslau N, Kessler RC, Chilcoat HD, et al. Trauma and post-
population. J Am Acad Child Adolesc Psychiatry 1997;36:340– traumatic stress disorder in the community: the 1996 De-
348. troit Area Survey of Trauma. Arch Gen Psychiatry 1998;55:
40. Vesga-Lopez O, Schneier FR, Wang S, et al. Gender differences in 626–632.
generalized anxiety disorder: results from the National Epidemi- 59. Cosci F, Schruers KRJ, Abrams K, Griez EJL. Alcohol use dis-
ologic Survey on Alcohol and Related Conditions (NESARC). J orders and panic disorder: a review of the evidence of a direct
Clin Psychiatry 2008;69:1606–1616. relationship. J Clin Psychiatry 2007;68:874–880.

Depression and Anxiety

764 Blanco et al.

60. Kendler KS, Neale MC, Kessler RC, et al. Major depression and 63. Neale MC, Kendler KS. Models of comorbidity for multifactorial
generalized anxiety disorder: same genes,(partly) different envi- disorders. Am J Hum Genet 1995;57:935–953.
ronments? Arch Gen Psychiatry 1992;49:716–722. 64. Kessler RC, Chiu WT, Demler O, Walters EE. Prevalence,
61. Insel TR, Cuthbert BN, Garvey MA, et al. Research domain cri- severity, and comorbidity of 12-month DSM-IV disorders in the
teria (RDoC): toward a new classification framework for research National Comorbidity Survey Replication. Arch Gen Psychiatry
on mental disorders. Am J Psychiatry 2010;167:748–751. 2005;62:617–627.
62. National Institute of Mental Health. NIMH Research Do- 65. Kline RB. Principles and Practice of Structural Equation Model-
main Criteria (RDoC) 2011. Available at: http://www.nimh. ing. 2nd ed. New York: Guilford Press; 2005.
nih.gov/research-funding/rdoc/nimh-research-domain-criteria- 66. Barlow D, Allen LB, Choate ML. Toward a unified treatment for
rdoc.shtml. emotional disorders. Behav Ther 2004;35:205–230.

Depression and Anxiety

Copyright of Depression & Anxiety (1091-4269) is the property of John Wiley & Sons, Inc.
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.