Content uploaded by Manisha Sawhney
Author content
All content in this area was uploaded by Manisha Sawhney on Dec 29, 2018
Content may be subject to copyright.
Received: 9 August 2016 Revised: 29 May 2017 Accepted: 1 June 2017
DOI: 10.1002/da.22666
RESEARCH ARTICLE
Atypical depression and double depression predict new-onset
cardiovascular disease in U.S. adults
Stephanie M. Case MS1Manisha Sawhney PhD2Jesse C. Stewart PhD3
1School of Psychological Sciences, University of
Indianapolis, Indianapolis, IN, USA
2Department of Psychology, Liffrig Family
School of Education and Behavioral Sciences,
University of Mary, Bismarck, ND, USA
3Department of Psychology,Indiana
University-Purdue University Indianapolis
(IUPUI), Indianapolis, IN, USA
Correspondence
Jesse C. Stewart, PhD, Department of
Psychology,Indiana University-
Purdue University Indianapolis, 402 North
Blackford Street, LD 100E, Indianapolis, IN
46202.
Email: jstew@iupui.edu
Funding Information
Grant sponsor: National Institute on Alcohol
Abuse and Alcoholism; Grant sponsor: The
National Institute on Drug Abuse; Grant spon-
sor: National Heart, Lung, and Blood Institute;
Contract grant number: R01HL122245.
BACKGROUND: Although depression is a risk factor for cardiovascular disease (CVD), it is
unknown whether this risk varies across depressive disorder subtypes. Thus, we investigated
atypical major depressive disorder (MDD) and double depression as predictors of new-onset CVD
in a nationally representative sample of U.S. adults.
METHODS: Prospective data from 28,726 adults initially free of CVD who participated in Wave
1 (2001–2002) and Wave 2 (2004–2005) of the National Epidemiologic Survey on Alcohol and
Related Conditions (NESARC) were examined. Lifetime depressive disorder subtypes (Wave 1)
and incident CVD (Wave 2) were determined bystructured interviews.
RESULTS:We identified 1,116 incident CVD cases. In demographics adjusted models, the atypical
MDD group had a higher odds of incident CVD than the no depression history (OR =2.19, 95%
CI: 1.71–2.81, P<.001), dysthymic disorder only (OR =1.61, 95% CI: 1.08–2.39, P=.019), and
nonatypical MDD (OR =1.46, 95% CI: 1.11–1.91, P=.006) groups. Likewise, the double depres-
sion group had a higher odds of incident CVD than the no depression history (OR =2.17, 95% CI:
1.92–2.45, P<.001), dysthymic disorder only (OR =1.59, 95% CI: 1.16–2.19, P=.004), and MDD
only (OR =1.46, 95% CI: 1.20–1.77, P<.001) groups. Relationships were similar but attenuated
after adjustment for CVD risk factors and anxiety disorders.
CONCLUSIONS: Adults with atypical MDD or double depression may be subgroups of the
depressed population at particularly high risk of new-onset CVD. Thus, these subgroups may (a) be
driving the overall depression–CVD relationship and (b) be in need of earlier and/or more intense
CVD primary prevention efforts to reduce their excess CVD burden.
KEYWORDS
angina pectoris, arteriosclerosis, depressive disorder, epidemiologic studies, myocardial infarc-
tion, prospective studies
1INTRODUCTION
Considerable evidence indicates that depression is an independent risk
factor for cardiovascular disease (CVD) (Van der Kooy et al., 2007).
However, it is unknown whether the CVD risk conferred by depres-
sion varies across disorder subtypes. Atypical depression accounts
for 15–40% of depression cases, and among its key features are the
reversed somatic-vegetative symptoms of hyperphagia and hypersom-
nia (APA, 2013; Grant et al., 2009; Quitkin, 2002). Adults with atypi-
cal depression, compared to those with nonatypical depression, have
higher rates or levels of several CVD risk factors, including dyslipi-
demia, hypertension, diabetes, obesity, metabolic syndrome, physi-
cal inactivity, and systemic inflammation (Chou & Yu, 2013; Cizza
et al., 2012; Glaus et al., 2013; Hickman, Khambaty, & Stewart, 2014;
Lamers et al., 2013; Lasserre et al., 2014; Levitan et al., 2012; Niran-
jan, Corujo, Ziegelstein, & Nwulia, 2012; Rudolf, Greggersen, Kahl,
Huppe, & Schweiger, 2014; Takeuchi, Nakao, Kachi, & Yano, 2013; van
Reedt Dortland et al., 2010). Because of this higher risk factor bur-
den, atypical depression may be a stronger predictor of CVD than
nonatypical depression. The few existing studies, however, do not
support this notion. Niranjan et al. (2012) found no difference in
prevalent CVD between depressed adults with versus without atypical
features, and atypical MDD symptoms were not associated with preva-
lent CVD in two other studies (Fraguas et al., 2007; Vogelzangs et al.,
2010). A key limitation of all three studies, however, is their cross-
sectional design, especially considering that reverse causality is plau-
sible (Spijkerman et al., 2005) and could mask depression subtype
differences.
Depress Anxiety.2018;35:10–17. c
2017 Wiley Periodicals, Inc. 10wileyonlinelibrary.com/journal/da
CASE ET AL.11
Double depression refers to major depressive disorder (MDD)
superimposed on dysthymia (Keller & Shapiro, 1982), a chronic, low-
grade depressive disorder. About three quarters of patients with dys-
thymia have or will experience a major depressive episode and, thus,
suffer from double depression (Hellerstein & Eipper, 2013). To date,
associations of double depression with CVD risk markers or outcomes
have not been examined. Nonetheless, double depression may be a
stronger predictor of CVD due to its longer duration and higher recur-
rence rate than MDD alone and its greater symptom severity than
dysthymia alone (Keller, Hirschfeld, & Hanks, 1997).
Because no studies have examined atypical or double depression as
predictors of new-onset CVD, our primary aim was to address these
key gaps. Our secondary aim was to evaluate whether these rela-
tionships are independent of anxiety disorders. As anxiety disorders
are highly comorbid with depressive disorders (Kessler et al., 2003)
and also predict incident CVD (Roest, Martens, de Jonge, & Denol-
let, 2010), it is important to adjust for them to isolate depression
effects. We examined Wave 1 (2001–2002) and Wave 2 (2004–2005)
data from a large, nationally representative sample of U.S. adults from
the National Epidemiologic Survey on Alcohol and Related Conditions
(NESARC).
2MATERIALS AND METHODS
2.1 Study design and sample
NESARC is a prospective cohort study designed to determine the
prevalence of alcohol use disorders and associated disabilities in the
U.S. civilian noninstitutionalized population ≥18 years. Descriptions
of study methods are provided elsewhere (Grant et al., 2005, 2009;
Hasin & Grant, 2015). NESARC received ethical approval from the
U.S. Census Bureau and the U.S. Office of Management and Bud-
get. At Wave 1, 43,093 respondents (81.0% response rate) completed
computer-assisted home interviews assessing substance use disor-
ders, psychiatric disorders, and medical conditions. Three years later
(mean =36.6 months) at Wave 2, 34,653 of the eligible Wave 1 respon-
dents (86.7% response rate) completed a second home interview. A
total of 3,134 Wave 1 respondents were not eligible for Wave 2 due to
being deceased, deported, mentally or physically impaired, or on active
duty in the armed forces. Respondents who participated in Waves 1
and 2, versus Wave1 only, were younger (46.0 vs. 48.2 years) and more
likely to be female (58.0 vs. 53.2%), to be non-Hispanic White (58.2 vs.
51.3%), to have a high school education or more (83.4 vs. 75.1%), and
to have a lifetime depressive disorder (17.3 vs. 13.3%; all Ps<.001).
We applied three exclusioncriteria to the Wave 2 sample. Respondents
were excluded if: (1) CVD status at Wave 1 was positive (n=1,742) or
missing (n=1,719), (2) CVD status at Wave 2 was missing (n=1,065),
or (3) CVD risk factors at Wave 1 were missing (n=1,401). Character-
istics of our final sample of 28,726 adults are shown in Table 1.
2.2 Measures and procedures
2.2.1 Lifetime depressive disorder subtypes
Lifetime dysthymic disorder and MDD were determined by the Alco-
hol Use Disorder and Associated Disabilities Interview Schedule-IV
TAB L E 1 TABLE 1 Characteristics of respondents (N=28,726)
Demographic factors
Age, years (SD) 44.8 (17.0)
Female, % 57.5
Race/ethnicity
Non-Hispanic White, % 58.2
Non-Hispanic Black, % 18.6
Hispanic/Latino, % 18.7
Other, % 4.5
Educatio n level
Less than high school, % 15.2
High school or equivalent, % 28.5
Some college or Associate’s degree, % 30.6
Bachelor’s degree or higher, % 25.7
Cardiovascular risk factors
Hypertension, % 18.2
Hypercholesterolemia, % 19.3
Diabetes, % 8.0
Tobacco use, % 25.6
Body mass index, kg/m227.0 (5.6)
Anxiety disorders
Lifetime anxiety disorder, % 10.5
Note: Continuous variables are presented as mean (standard deviation), and
categorical variables are presented as percentage. Lifetime anxiety disor-
der consists of panic disorder, agoraphobia, generalized anxiety disorder,
and social phobia.
(AUDADIS-IV), a fully structured diagnostic interview administered
by lay interviewers assessing mental disorders using DSM-IV crite-
ria (Ruan et al., 2008). NESARC personnel coded diagnostic variables
for the past year and prior to the past year. We used the NESARC
variables that excluded illness-induced and substance-induced disor-
ders and ruled out bereavement (Grant et al., 2005). The AUDADIS-IV
has demonstrated good test-retest reliability for depressive disorders
(Grant et al., 2003) and generally good agreement with clinician evalu-
ations (Hasin & Grant, 2015).
From the NESARC variables, we computed two variables. Our first
variable, atypical depression, had four levels: no depressive disorder,
dysthymic disorder only, nonatypical MDD, and atypical MDD. First,
we classified respondents into no depressive disorder history (never
met criteria), lifetime dysthymic disorder only (past year or prior), and
lifetime MDD (past year or prior) groups. Those who met criteria for
both lifetime dysthymic disorder and MDD were placed into the MDD
group. Then, we further classified respondents with lifetime MDD. The
atypical MDD group consisted of respondents with both hyperpha-
gia and hypersomnia. We coded individuals as having hyperphagia if
they answered “yes” to either of the following AUDADIS-IV questions:
“During that time when your mood was at its lowest/you enjoyed or
cared the least about things, did you gain at least 2 pounds a week
for several weeks or at least 10 pounds altogether within a month
(other than when you were growing or pregnant)?” or “During that
time…, did you find that you wanted to eat a lot more than usual for no
12 CASE ET AL.
special reason, most days for at least 2 weeks?” We coded respon-
dents as having hypersomnia if they answered “yes” to the AUDADIS-
IV question, “During that time…, did you sleep more than usual nearly
every day for at least 2 weeks?” While other criteria for atypical MDD
exist (APA, 2013), using only the reversed somatic-vegetative symp-
toms is a valid approach (Benazzi, 2002) and has been utilized in
past studies (Blanco et al., 2012; Chou & Yu, 2013; Horwath, John-
son, Weissman, & Hornig, 1992; Matza, Revicki, Davidson, & Stewart,
2003).
Our second variable, double depression, also had four levels: no
depressive disorder, dysthymic disorder only, MDD only, and double
depression. The definitions for the no depressive disorder and dys-
thymic disorder groups were the same as above. Respondents with
lifetime MDD were further classified into two groups: MDD only (if
only lifetime MDD was present) and double depression (if both lifetime
dysthymic disorder and MDD were present).
2.2.2 Incident CVD
Using data from the NESARC Medical Conditions and Practices ques-
tionnaire administered at Wave 2, we computed an incident CVD vari-
able comprised of new-onset arteriosclerosis, angina, or myocardial
infarction (MI) based on self-reported physician diagnoses. In Part A,
respondents were asked, “In the last 12 months, did you have: (1) hard-
ening of the arteries or arteriosclerosis? (2) chest pain or angina pec-
toris? (3) a heart attack or MI?” If the answer to Part A was “yes,” in Part
B respondents were asked, “Did a doctor or other health professional
tell you that you had (name of condition)?” We coded respondents as
positive for incident CVD if they answered “yes” to Parts A and B for at
least one CVD question, and we coded respondents as negative if they
answered “no” to all three Part A questions. Those who were coded
as “unknown” for Part A or B for one or more questions and who did
not answer “yes” to Part A and B for at least one question were coded
as missing for incident CVD and were excluded. To compute a corre-
sponding baseline CVD variable, we applied the same coding scheme
to the identical Wave 1 CVD questions. Because our focus is predict-
ing new-onset CVD, we included only respondents coded negative for
baseline CVD.
2.2.3 Potential confounders
The following variables—which could operate as potential confounders
of depression-CVD associations (Luger, Suls, & Vander Weg, 2014;
Luppino et al., 2010; Nouwen et al., 2010)—were included as con-
trol variables in the models: age (years), sex (0 =male, 1 =female),
race/ethnicity, education level, hypertension, hypercholesterolemia,
diabetes, tobacco use, body mass index (BMI), and lifetime anxiety dis-
order. These variables were based on self-reported data from Wave
1 except hypercholesterolemia and diabetes, which were assessed
only at Wave 2. We recoded race/ethnicity into a four-level variable
(0 =non-Hispanic White, 1 =non-Hispanic Black, 2 =Hispanic or
Latino, 3 =Other). Next, we created three dummy-coded variables
using non-Hispanic White as the reference category. Education level
was assessed by the question, “Highest grade or year of school com-
pleted?” From these data, we computed a 4-level variable (0 =less than
high school, 1 =high school or equivalent, 2 =some college or Asso-
ciate’s degree, 3 =Bachelor’s degree or higher). Three dummy-coded
variables using less than high school as the reference category were
then created.
We coded respondents as positive for hypertension, hypercholes-
terolemia, and diabetes, respectively, if they answered “yes” to “In the
past 12 months, have you had: (1) high blood pressure or hyperten-
sion? (2) high cholesterol? (3) diabetes or sugar diabetes?" and “yes” to
“Did a doctor or other health professional tell you that you had (name
of condition)?” We coded respondents as negative for each condition
if they answered "no" to the first question. Those coded by NESARC
personnel as “unknown” for either question were coded as missing for
that condition and were excluded. We recoded NESARC’s tobacco use
variable (current user, former user, lifetime nonuser) into a dichoto-
mous variable (0 =current nonuser, 1 =current user). BMI (kg/m2)was
computed from self-reported height and weight.
We computed a lifetime anxiety disorder variable using AUDADIS-
IV data collected at Wave 1. Respondents who were coded by NESARC
personnel as meeting diagnostic criteria for panic disorder, agorapho-
bia, generalized anxiety disorder, or social phobia in the past year or
prior (illness- and substance-induced disorders excluded) were coded
as positive for lifetime anxiety disorder. Those not meeting criteria for
any of these disorders were coded as negative.
2.3 Data analysis
We ran three sets of logistic regression analyses—demographics
adjusted, CVD risk factor adjusted, and anxiety disorder adjusted—
examining the atypical depression variable as a predictor of incident
CVD. For each set, we ran three models, each of which used a different
reference category (1 =no depressive disorder, 2 =dysthymic disor-
der only, 3 =nonatypical MDD). Control variables in the demographics
adjusted models were age, sex, race/ethnicity, and education level. The
CVD risk factor adjusted models were further adjusted for hyperten-
sion, hypercholesterolemia, diabetes, tobacco use, and BMI. The anxi-
ety disorder adjusted models additionally included lifetime anxiety dis-
order. Finally, we ran an identical series of logistic regression analy-
ses for the double depression variable. All selected control variables
were significantly associated with the atypical depression and double
depression variables.
Analyses were conducted with SAS statistical software, version 9.3.
Models were weighted to account for oversampling, probabilities of
selection, and nonresponse. Weighted analyses provide estimates for
U.S. civilian noninstitutionalized population based on the 2000 Decen-
nial Census (Hasin & Grant, 2015).
3RESULTS
3.1 Depressive disorder subtypes and incident CVD
The lifetime prevalence was 3.4% (964 cases) for atypical MDD and
3.1% (833 cases) for double depression. The degree of overlap between
these two subtypes was modest, as the phi coefficient (r𝜑) was 0.22,
CASE ET AL.13
TAB L E 2 Logistic regression models examining atypical depression as a predictor of incident cardiovascular disease
Demographics adjusted
models
CVD risk factor adjusted
models
Anxiety disorder adjusted
models
Incident CVD cases
(% of group) OR 95% CI OR 95% CI OR 95% CI
No depressive disorder
(reference)
n=23,788 (82.8%)
903 (3.8%) 1.00 – 1.00 – 1.00 –
Dysthymic disorder only
n=227 (0.8%)
13 (5.7%) 1.37a,d 1.01–1.85 1.22 0.89–1.66 1.12 0.82–1.54
Nonatypical major
depressive disorder
n=3,747 (13.0%)
155 (4.1%) 1.51a,d 1.31–1.74 1.42a1.23–1.65 1.28a1.08–1.51
Atypical major
depressive disorder
n=964 (3.4%)
45 (4.7%) 2.19a,b,c 1.71–2.81 1.78a1.37–2.30 1.56a1.19–2.03
Note: N =28,726 (1,116 incident CVD cases). Control variables in the demographic-adjusted models were age, sex, race/ethnicity, and education level. The
CVD risk factor adjusted models were further adjusted for hypertension, hypercholesterolemia, diabetes, tobacco use, and body mass index. The anxiety
disorder adjusted models also included lifetime anxiety disorder. CVD, cardiovasculardisease; OR, odds ratio; CI, confidence interval.
aSignificantly different from no depressive disorder (P<.05).
bSignificantly different from dysthymic disorder only (P<.05).
cSignificantly different from nonatypical major depressive disorder (P<.05).
dSignificantly different from atypical major depressive disorder (P<.05).
TAB L E 3 Logistic regression models examining double depression as a predictor of incident cardiovascular disease
Demographics adjusted
models
CVD risk factor adjusted
models
Anxiety disorder adjusted
models
Incident CVD Cases
(% of group) OR 95% CI OR 95% CI OR 95% CI
No depressive disorder
(reference)
n=23,788 (82.8%)
903 (3.8%) 1.00 – 1.00 – 1.00 –
Dysthymic disorder only
n=227 (0.8%)
13 (5.7%) 1.36a,d 1.01–1.84 1.22d0.89–1.66 1.12d0.82–1.54
Major depressive
disorder
n=3,828 (13.3%)
150 (3.9%) 1.49a,d 1.27–1.76 1.38a,d 1.16–1.63 1.26a,d 1.04–1.51
Double depression
n=883 (3.1%)
50 (5.7%) 2.17a,b,c 1.92–2.45 1.95a,b,c 1.72–2.21 1.65a,b,c 1.46–1.87
Note: N =28,726 (1,116 incident CVD cases). Control variables in the demographic-adjusted models were age, sex, race/ethnicity, and education level. The
CVD risk factor adjusted models were further adjusted for hypertension, hypercholesterolemia, diabetes, tobacco use, and body mass index. The anxiety
disorder adjusted models also included lifetime anxiety disorder. CVD, cardiovasculardisease; OR, odds ratio; CI, confidence interval.
aSignificantly different from no depressive disorder (P<.05).
bSignificantly different from dysthymic disorder only (P<.05).
cSignificantly different from major depressive disorder (P<.05).
dSignificantly different from double depression (P<.05).
and 228 cases qualified for both atypical MDD and double depres-
sion. We identified 1,116 cases (3.9%) of incident CVD: 264 with arte-
riosclerosis only, 625 with angina only, 75 with MI only, 53 with arte-
riosclerosis and angina, 11 with arteriosclerosis and MI, 58 with angina
and MI, and 30 with all three outcomes. Tables 2 and 3 display the
number of CVD cases and the unadjusted case rate for each depres-
sive disorder subtype. Due to its smaller size, the dysthymic disorder
only group had a low number of CVD cases, which reduced power for
comparisons involving this group.
3.2 Atypical depression as a predictor of incident
CVD
Demographic adjusted logistic regression models (Table 2) revealed
that respondents with atypical MDD had over twice the odds of
incident CVD than those with no depressive disorder (P<.001).
Although adults with dysthymic disorder only (37% greater odds;
P=.043) or nonatypical MDD (51% greater odds; P<.001) were more
likely to develop CVD than nondepressed adults, the magnitude of
these associations was less than half of that for atypical MDD. These
models also indicated that the odds of CVD in the atypical MDD group
were greater than in the dysthymic disorder only (OR =1.61, 95%
CI: 1.08–2.39, P=.019) and nonatypical MDD (OR =1.46, 95% CI:
1.11–1.91, P=.006) groups.
CVD risk factor adjusted and anxiety disorder adjusted models
(Table 2) yielded a similar pattern of results, although associations
were attenuated and some fell short of significance. Respondents with
atypical MDD remained at the highest risk of incident CVD, with a
78% (P<.001) and 56% (P=.001) greater odds than nondepressed
14 CASE ET AL.
adults. However, comparisons of the atypical MDD group with the
dysthymic disorder only group (CVD risk factor adjusted OR =1.46,
95% CI: 0.97–2.18, P=.068; anxiety disorder adjusted OR =1.39,
95% CI: 0.92–2.09, P=.12) and the nonatypical MDD group (CVD
risk factor adjusted OR =1.25, 95% CI: 0.94–1.65, P=.13; anxiety
disorder adjusted OR =1.22, 95% CI: 0.92–1.61, P=.18) were no
longer significant. In the CVD risk factor adjusted and anxiety disor-
der adjusted models, nonatypical MDD also continued to predict inci-
dent CVD (P<.001 and .004), whereas dysthymic disorder did not
(P=.21 and .48).
3.3 Double depression as a predictor of incident
CVD
Demographics adjusted models (Table 3) indicated that respondents
those with double depression had more than twice the odds of inci-
dent CVD than nondepressed adults (P<.001). Dysthymic disorder
only (36% greater odds; P=.044) and MDD only (49% greater odds;
P<.001) also predicted incident CVD but these relationships were not
as strong as that for double depression. The double depression group
also had higher odds of incident CVD than the dysthymic disorder only
(OR =1.59, 95% CI: 1.16–2.19, P=.004) and MDD only (OR =1.46,
95% CI: 1.20–1.77, P<.001) groups.
Although associations were again attenuated, the pattern of results
was similar in the CVD risk factor adjusted and anxiety disorder
adjusted models (Table 3). Adults with double depression remained at
the highest odds of incident CVD. This group had a 95% (P<.001)
and 65% (P<.001) greater odds than those in the no depressive dis-
order group, and comparisons with the dysthymic disorder only group
(CVD risk factor adjusted OR =1.61, 95% CI: 1.16–2.21, P=.004; anx-
iety disorder adjusted OR =1.47, 95% CI: 1.07–2.02, P=.018) and
the MDD only group (CVD risk factor adjusted OR =1.42, 95% CI:
1.16–1.73, P=.001; anxiety disorder adjusted OR =1.31, 95% CI:
1.08–1.59, P=.006) continued to be significant. In the CVD risk fac-
tor adjusted and anxiety disorder adjusted models, MDD only (P<.001
and P=.016)—but not dysthymic disorder only (P=.22 and .48)—
remained a predictor of incident CVD.
4DISCUSSION
Our examination of the NESARC data indicates that atypical MDD and
double depression are two depressive disorder subtypes that may be
particularly strong predictors of new-onset CVD. With respect to atyp-
ical MDD, U.S. adults with a lifetime history of this subtype had a higher
odds of incident CVD than those with no depression history, dysthymic
disorder only, or nonatypical MDD in demographics adjusted models.
Although a similar pattern of results was observed in CVD risk fac-
tor adjusted models, some comparisons fell short of significance. Con-
cerning double depression, U.S. adults with a lifetime history of sub-
type had a higher odds of incident CVD than those with no depression
history, dysthymic disorder only, or MDD only in both demographics
and CVD risk factor adjusted models. Further adjustment for lifetime
anxiety disorder attenuated associations but did not alter the pattern
of results for either subtype. The modest degree of overlap between
atypical MDD and double depression suggests that their associations
with incident CVD likely reflect separate relationships; however, the
228 cases with both subtypes did contribute to both relationships.
Collectively, our findings indicate that adults with atypical MDD or
double depression may be subgroups of the depressed population at
greatest risk of developing CVD and, thus, may be driving the overall
depression-CVD relationship.
Our study addresses a key gap in the literature—that is, the absence
of prospective studies examining atypical depression or double depres-
sion as predictors of incident CVD. Our findings do conflict with three
prior studies that observed no difference in prevalent CVD between
depressed adults with versus without atypical features (Niranjan et al.,
2012) and no associations between atypical MDD symptoms and
prevalent CVD (Fraguas et al., 2007; Vogelzangs et al., 2010). How-
ever, due to their cross-sectional design, reverse causality may have
obscured depression subtype differences (Spijkerman et al., 2005). To
our knowledge, there are no previous studies that have examinedasso-
ciations between double depression and CVD risk markers or out-
comes. Although one other prospective analysis of the NESARC data
reported the association of lifetime MDD with CVD, the focus of
that analysis was the bipolar disorder-incident CVD relationship, and
depressive disorder subtypes were not examined (Goldstein, Schaffer,
Wang, & Blanco, 2015).
There are multiple candidate mechanisms that could explain why
atypical MDD may be a stronger predictor of incident CVD. Evidence
suggests that conventional CVD risk factors are elevated in adults
with atypical versus nonatypical depression (Chou & Yu, 2013; Cizza
et al., 2012; Glaus et al., 2013; Lamers et al., 2013; Levitan et al., 2012;
Niranjan et al., 2012; Takeuchi et al., 2013; van Reedt Dortland et al.,
2010). Moreover, atypical MDD has been found to predict incident
obesity and increases in BMI, waist circumference, and fat mass over
time (Lasserre et al., 2014). In our models adjusting for Wave 1 CVD
risk factors, associations were attenuated. However, because depres-
sive disorders and CVD risk factors were assessed at the same point,
the NESARC data cannot be used to determine whether the CVD
risk factors were operating as confounders or mediators. Given that
our objective was to provide unbiased estimates of the associations
between depressive disorder subtypes and incident CVD, we chose to
treat the CVD risk factors as potential confounders to be conserva-
tive. In addition to conventional CVD risk factors, other mechanisms
may also be at work. Another candidate mechanism, which was not
assessed in NESARC, is greater systemic inflammation. Adults with
atypical depression, versus those with nonatypical depression, have
been found to have higher circulating levels of inflammatory markers
predictive of CVD, such as C-reactive protein and interleukin-6 (Hick-
man et al., 2014; Lamers et al., 2013; Rudolf, et al., 2014). Two other
candidate mechanisms are poor diet quality (Rahe et al., 2015) and low
physical activity (Glaus et al., 2013; Matza et al., 2003). In addition to
these biological and behavioral pathways, adults with atypical MDD
may have greater lifetime exposure to depression, given that it is char-
acterized by earlier age of onset, more severe symptoms, and a greater
number of episodes (Blanco et al., 2012; Matza et al., 2003; Novick
et al., 2005).
CASE ET AL.15
Less is known about the correlates of double depression that could
be operating as underlying mechanisms. Like atypical depression, asso-
ciations were attenuated after adjustment for conventional CVD risk
factors, suggesting that they may partially explain the elevated CVD
risk of this group. It is also plausible that double depression is a
stronger predictor of incident CVD on account of its longer duration
and higher recurrence rate than MDD alone and its greater symptom
severity than dysthymia alone (Keller et al., 1997), resulting in greater
lifetime exposure to depression and the associated atherogenic biolog-
ical and behavioral changes (Grippo & Johnson, 2002; Joynt, Whellan,
& O’Connor,2003). In addition, a smaller study found that patients with
double depression reported greater hopelessness than patients with
MDD or dysthymia (Joiner, Cook, Hersen, & Gordon, 2007). Hopeless-
ness has been linked with greater subclinical atherosclerosis (Whipple
et al., 2009) and an increased risk of CVD events (Everson et al., 1996),
independent of depressive symptoms. Clearly, there is a need for stud-
ies examining associations of double depression with CVD outcomes
and candidate mechanisms.
The present study has key strengths, including the longitudinal
design, large nationally representative sample, and structured inter-
view assessments of psychiatric disorders. NESARC is the largest and
most comprehensive psychiatric epidemiologic survey conducted in
the United States (Hasin & Grant, 2015), and its prospective data
allowed us to draw strong inferences regarding directionality. Our
study also has limitations that should be considered. First, epidemio-
logic surveys often assess CVD by self-report of physician diagnoses.
Supporting this approach, agreement between self-reported and med-
ical record-ascertained CVD has been found to be acceptable to good
(Barr, Tonkin, Welborn, & Shaw, 2009; Bergmann, Byers, Freedman, &
Mokdad, 1998; Heckbert et al., 2004; Lampe, Walker, Lennon, Whin-
cup, & Ebrahim, 1999; Machon et al., 2013; Okura, Urban, Mahoney,
Jacobsen, & Rodeheffer, 2004). A recent study (Yasaitis, Berkman, &
Chandra, 2015) comparing self-reported and Medicare claims iden-
tified MIs did observe lower agreement than past studies; however,
the authors speculated that this may have been due to their sample’s
older age and their narrower MI definition. Nonetheless, because some
degree of misclassification occurs with self-reports of physician diag-
noses, there is a need for future studies examining depressive disor-
der subtypes as predictors of incident CVD adjudicated by a review
of medical records. Second, incident fatal CVD events were not cap-
tured. Respondents who died between Waves1 and 2 of NESARC were
excluded from the Wave 2 cohort, and information regarding cause of
death is not available. While this could have compromised power, that
does not appear to be the case, as we observed 1,116 cases of inci-
dent CVD. Third, some incident nonfatal MIs may have not have been
detected because the NESARC Wave 2 questions inquired about CVD
diagnoses in the past 12 months only. This is less of a concern for arte-
riosclerosis and angina, as these are chronic conditions and not dis-
crete events. Our composite incident CVD outcome also reduces the
potential for misclassification, given that respondents who suffered
nonfatal MIs between Waves 1 and 2, but prior to the past 12 months,
may have also been diagnosed with one of the other CVD conditions
during follow-up. Fourth, due to the limited temporal resolution of the
diagnostic variables, our double depression definition did not take into
account the order of onset or co-occurrence of dysthymic disorder and
MDD. Future studies with diagnostic variables possessing greater tem-
poral resolution are needed to examine the importance of these char-
acteristics in predicting incident CVD. Fifth, although age ranged from
18 to 97 years in our sample, the mean age was only 45 years, and
the follow-up period was only three years. Both of these factors likely
contributed to the lower rate (3.9%) of incident CVD.
5CONCLUSION
We report prospective evidence from a nationally representative sam-
ple indicating that U.S. adults with atypical MDD or double depres-
sion may be subgroups of the depressed population at particularly
high risk of new-onset CVD who may be driving the depression–CVD
relationship. With respect to research implications, our results sug-
gest that clinical trials evaluating whether successful depression treat-
ment reduces CVD risk should consider specifically recruiting patients
with atypical MDD or double depression, as it is in these subgroups
where most of the excess CVD risk seems to reside. Furthermore, our
findings underscore the need to continue to test existing treatments
(Fournier et al., 2013) or to develop new treatments to address resid-
ual depressive symptoms and syndromes, such as reversed somatic-
vegetative symptoms and dysthymia. Concerning clinical practice, our
results highlight the potential importance of depression screening that
allows for depressive disorder subtyping. Finally, our findings raise the
possibility that CVD primary prevention efforts should be initiated
earlier and/or intensified among adults with atypical MDD or double
depression to prevent or delay clinical CVD onset, thereby reducing
the excess CVD burden of the depressed population.
ACKNOWLEDGEMENTS
NESARC is funded by the National Institute on Alcohol Abuse and
Alcoholism with supplemental support from the National Institute
on Drug Abuse. A portion of Dr. Stewart’s time was supported by
the National Heart, Lung, and Blood Institute under Award Number
R01HL122245. The content is solely the responsibility of the authors
and does not necessarily represent the official views of the National
Institutes of Health.
CONFLICT OF INTEREST
The authors have no possible conflicts of interest to declare.
REFERENCES
American Psychiatric Association. (2013). Diagnostic and statistical man-
ual of mental disorders, fifth edition (DSM-5), Washington, DC: American
Psychiatric Association.
Barr, E. L., Tonkin, A. M., Welborn, T. A., & Shaw, J. E. (2009). Validity of self-
reported cardiovascular disease events in comparison to medical record
adjudication and a statewide hospital morbidity database: The AusDiab
study. Internal Medicine Journal,39, 49–53.
Benazzi, F. (2002). Can only reversed vegetative symptoms define atypical
depression? Europen Archives of Psychiatry and Clinical Neuroscience,252,
288–293.
16 CASE ET AL.
Bergmann, M. M., Byers, T., Freedman, D. S., & Mokdad, A. (1998). Valid-
ity of self-reported diagnoses leading to hospitalization: A compar-
ison of self-reports with hospital records in a prospective study
of American adults. American Journal of Epidemiology,147, 969–
977.
Blanco, C., Vesga-Lopez, O., Stewart, J. W., Liu, S. M., Grant, B. F., &
Hasin, D. S. (2012). Epidemiology of major depression with atypical fea-
tures: Results from the National Epidemiologic Survey on Alcohol and
Related Conditions (NESARC). JournalofClinicalPsychiatry,73, 224–
232.
Chou, K. L., & Yu, K. M. (2013). Atypical depressive symptoms and obesity in
a national sample of older adults with major depressive disorder.Depres-
sion and Anxiety,30, 574–579.
Cizza, G., Ronsaville, D. S., Kleitz, H., Eskandari, F., Mistry, S., Torvik, S., …
Martinez, P. E. (2012). Clinical subtypes of depression are associated
with specific metabolic parameters and circadian endocrine profiles in
women: The power study. PLoS One,7, e28912.
Everson, S. A., Goldberg, D. E., Kaplan, G. A., Cohen, R. D., Pukkala, E.,
Tuomilehto, J., & Salonen, J. T. (1996). Hopelessness and risk of mor-
tality and incidence of myocardial infarction and cancer. Psychosomatic
Medicine,58, 113–121.
Fournier, J. C., DeRubeis, R. J.,Hollon, S. D., Gallop, R., Shelton, R. C., & Ams-
terdam, J. D. (2013). Differential change in specific depressive symp-
toms during antidepressant medication or cognitive therapy. Behaviour
Research and Therapy,51, 392–398.
Fraguas, R., Jr., Iosifescu, D. V., Alpert, J., Wisniewski, S. R., Barkin, J. L.,
Trivedi,M. H., …Fava, M. (2007). Major depressive disorder and comor-
bid cardiac disease: Is there a depressive subtype with greater cardio-
vascular morbidity? Results from the STAR*D study. Psychosomatics,48,
418–425.
Glaus, J., Vandeleur, C., Gholam-Rezaee, M., Castelao, E., Perrin, M., Rothen,
S., …Preisig, M. (2013). Atypical depression and alcohol misuse are
related to the cardiovascular risk in the general population. Acta Psychi-
atrica Scandinavica,128, 282–293.
Goldstein, B. I., Schaffer, A., Wang, S., & Blanco, C. (2015). Excessive and
premature new-onset cardiovascular disease among adults with bipo-
lar disorder in the US NESARC cohort. Journal of Clinical Psychiatry,76,
163–169.
Grant, B. F., Dawson, D. A., Stinson, F. S., Chou, P. S., Kay, W., & Pickering, R.
(2003). The Alcohol Use Disorder and Associated Disabilities Interview
Schedule-IV (AUDADIS-IV): Reliability of alcohol consumption, tobacco
use, family history of depression and psychiatric diagnostic modules
in a general population sample. Drug and Alcohol Dependence,71,7–
16.
Grant, B. F., Goldstein, R. B., Chou, S. P., Huang, B., Stinson, F. S., Dawson,
D. A., …Compton, W. M. (2009). Sociodemographic and psychopatho-
logic predictors of first incidence of DSM-IV substance use, mood and
anxiety disorders: Results from the Wave2 National Epidemiologic Sur-
vey on Alcohol and Related Conditions. Molecular Psychiatry,14, 1051–
1066.
Grant, B. F., Stinson, F. S., Hasin, D. S., Dawson, D. A., Chou, P., Ruan, W. J.,
& Huang, B. (2005). Prevalence, correlates, and comorbidity of bipolar
I disorder and axis I and II disorders: Results from the National Epi-
demiologic Survey on Alcohol and Related Conditions. Journal of Clinical
Psychiatry,66, 1205–1215.
Grippo, A. J., & Johnson, A. K. (2002). Biological mechanisms in the relation-
ship between depression and heart disease. Neuroscience and Biobehav-
ioral Reviews,26, 941–962.
Hasin, D. S., & Grant, B. F. (2015). The National Epidemiologic Survey on
Alcohol and Related Conditions (NESARC) Waves 1 and 2: Review and
summary of findings. Social Psychiatry and Psychiatric Epidemiology,50,
1609–1640.
Heckbert, S. R., Kooperberg, C., Safford, M. M., Psaty, B. M., Hsia, J., McTier-
nan, A., …Curb, J. D. (2004). Comparison of self-report, hospital dis-
charge codes, and adjudication of cardiovascular events in the Women’s
Health Initiative. American Journal of Epidemiology,160, 1152–1158.
Hellerstein, D. J., & Eipper, J. W. (2013). Dysthymia and chronic depression.
In J. J. Mann , P. J. McGrath & S. P. Roose (Eds.), Clinical handbook for the
management of mood disorders (pp. 20–36). New York, NY: Cambridge
University Press.
Hickman, R. J., Khambaty, T., & Stewart, J. C. (2014). C-reactive protein
is elevated in atypical but not nonatypical depression: Data from the
National Health and Nutrition Examination survey (NHANES) 1999–
2004. Journal of Behavioral Medicine,37, 621–629.
Horwath, E., Johnson, J., Weissman, M. M., & Hornig, C. D.(1992). The valid-
ity of major depression with atypical features based on a community
study. Journal of Affective Disorders,26, 117–125.
Joiner, T. E., Jr., Cook, J. M., Hersen, M., & Gordon, K. H. (2007). Double
depression in older adult psychiatric outpatients: Hopelessness as a
defining feature. Journal of Affective Disorders,101, 235–238.
Joynt, K. E., Whellan, D. J., & O’Connor, C. M. (2003). Depression and car-
diovascular disease: Mechanisms of interaction. Biological Psychiatry,54,
248–261.
Keller, M. B., Hirschfeld, R. M., & Hanks, D. (1997). Double depression: A
distinctive subtype of unipolar depression. Journal of Affective Disorders,
45, 65–73.
Keller, M. B., & Shapiro, R. W.(1982). "Double depression": Superimposition
of acute depressive episodes on chronic depressive disorders. American
Journal of Psychiatry,139, 438–442.
Kessler, R. C., Berglund, P., Demler, O., Jin, R., Koretz, D., Merikangas, K.
R., …Wang, P. S. (2003). The epidemiology of major depressive disor-
der: Results from the National Comorbidity Survey Replication (NCS-R).
JAMA,289, 3095–3105.
Lamers, F., Vogelzangs, N., Merikangas, K. R., de Jonge, P., Beekman, A. T., &
Penninx, B. W. (2013). Evidence for a differential role of HPA-axis func-
tion, inflammation and metabolic syndrome in melancholic versus atyp-
ical depression. Molecular Psychiatry,18, 692–699.
Lampe, F. C., Walker, M., Lennon, L. T., Whincup, P. H., & Ebrahim, S. (1999).
Validity of a self-reported history of doctor-diagnosed angina. Journal of
Clinical Epidemiology,52, 73–81.
Lasserre,A.M.,Glaus,J.,Vandeleur,C.L.,Marques-Vidal,P.,Vaucher,J.,Bas-
tardot, F., …Preisig, M. (2014). Depression with atypical features and
increase in obesity, body mass index,waist circumference, and fat mass:
A prospective, population-based study. JAMA Psychiatry,71, 880–888.
Levitan, R. D., Davis, C., Kaplan, A. S., Arenovich, T., Phillips, D. I., & Ravin-
dran, A. V. (2012). Obesity comorbidity in unipolar major depressive
disorder: Refining the core phenotype. Journal of Clinical Psychiatry,
73(8)1119–1124.
Luger, T. M., Suls, J., & Weg, Vander, & M., W. (2014). How robust is the
association between smoking and depression in adults? A meta-analysis
using linear mixed-effects models. Addictive Behaviors,39, 1418–1429.
Luppino,F.S.,deWit,L.M.,Bouvy,P.F.,Stijnen,T.,Cuijpers,P.,Penninx,B.W.,
& Zitman, F. G. (2010). Overweight, obesity, and depression: A system-
atic review and meta-analysis of longitudinal studies. Archives of General
Psychiatry,67, 220–229.
Machon, M., Arriola, L., Larranaga, N., Amiano, P., Moreno-Iribas, C., Agudo,
A., …Dorronsoro, M. (2013). Validity of self-reported prevalent cases
of stroke and acute myocardial infarction in the Spanish cohort of the
EPIC study. Journal of Epidemiology and Community Health,67, 71–75.
Matza, L. S., Revicki, D. A., Davidson, J. R., & Stewart, J. W. (2003). Depres-
sion with atypical features in the National Comorbidity Survey: Classifi-
cation, description, and consequences. Archives of General Psychiatry,60,
817–826.
CASE ET AL.17
Niranjan, A., Corujo, A., Ziegelstein, R. C., & Nwulia, E. (2012). Depression
and heart disease in US adults. General Hospital Psychiatry,34, 254–261.
Nouwen, A., Winkley, K., Twisk, J., Lloyd, C. E., Peyrot, M., Ismail, K., …Con-
soritum, E. R. (2010). Type 2 diabetes mellitus as a risk factor for the
onset of depression: A systematic review and meta-analysis. Diabetolo-
gia,53, 2480–2486.
Novick, J. S., Stewart, J. W., Wisniewski, S. R., Cook, I. A., Manev, R., Nieren-
berg,A.A.,…Rush, A. J. (2005). Clinical and demographic features of
atypical depression in outpatients with major depressive disorder: Pre-
liminary findings from STAR*D. JournalofClinicalPsychiatry,66, 1002–
1011.
Okura, Y., Urban, L. H., Mahoney, D. W., Jacobsen, S. J., & Rodeheffer, R.
J. (2004). Agreement between self-report questionnaires and medi-
cal record data was substantial for diabetes, hypertension, myocardial
infarction and stroke but not for heart failure. Journal of Clinical Epidemi-
ology,57, 1096–1103.
Quitkin, F. M. (2002). Depression with atypical features: Diagnostic valid-
ity, prevalence, and treatment. Primary Care Companion to the Journal of
Clinical Psychiatry,4, 94–99.
Rahe, C., Baune, B. T., Unrath, M., Arolt, V., Wellmann, J., Wersching,
H., & Berger, K. (2015). Associations between depression subtypes,
depression severity and diet quality: Cross-sectional findings from
the BiDirect Study. BMC Psychiatry,15, 38. https://doi.org/10.1186/
s12888-015-0426-9
Roest, A. M., Martens, E. J., de Jonge, P., & Denollet, J. (2010). Anxiety and
risk of incident coronary heart disease: A meta-analysis. Journalofthe
American College of Cardiology,56, 38–46.
Ruan, W. J., Goldstein, R. B., Chou, S. P., Smith, S. M., Saha, T. D., Pickering, R.
P. , …Grant, B. F. (2008). The alcohol use disorder and associated disabil-
ities interview schedule-IV (AUDADIS-IV): Reliability of new psychiatric
diagnostic modules and risk factors in a general population sample. Drug
and Alcohol Dependence,92, 27–36.
Rudolf, S., Greggersen, W., Kahl, K. G., Huppe, M., & Schweiger, U. (2014).
Elevated IL-6 levels in patients with atypical depression but not in
patients with typical depression. Psychiatry Research,217, 34–38.
Spijkerman, T., de Jonge, P., van den Brink, R. H., Jansen, J. H., May, J. F.,
Crijns, H. J., & Ormel, J. (2005). Depression following myocardial infarc-
tion: First-ever versus ongoing and recurrent episodes. General Hospital
Psychiatry,27, 411–417.
Takeuchi, T., Nakao, M., Kachi, Y.,& Yano, E. (2013). Association of metabolic
syndrome with atypical features of depression in Japanese people.
Psychiatry and Clinical Neurosciences,67, 532–539.
Van der Kooy, K., van Hout, H., Marwijk, H., Marten, H., Stehouwer, C., &
Beekman, A. (2007). Depression and the risk for cardiovascular dis-
eases: Systematic review and meta analysis. International Journal of Geri-
atric Psychiatry,22, 613–626.
van Reedt Dortland, A., Giltay, E., van Veen, T., van Pelt, J., Zitman, F., & Pen-
ninx, B. (2010). Associations between serum lipids and major depressive
disorder: Results from the Netherlands Study of Depression and Anxi-
ety (NESDA). Journal of Clinical Psychiatry,71, 729–736.
Vogelzangs, N., Seldenrijk, A., Beekman, A. T., van Hout, H. P., de Jonge, P., &
Penninx, B. W. (2010). Cardiovascular disease in persons with depres-
sive and anxiety disorders. Journal of Affective Disorders,125, 241–
248.
Whipple, M. O., Lewis, T. T., Sutton-Tyrrell, K., Matthews, K. A., Barinas-
Mitchell, E., Powell, L. H., & Everson-Rose, S. A. (2009). Hopelessness,
depressive symptoms, and carotid atherosclerosis in women: The Study
of Women’s Health Across the Nation (SWAN) heart study. Stroke,40,
3166–3172.
Yasaitis, L. C., Berkman, L. F., & Chandra, A. (2015). Comparison of self-
reported and Medicare claims-identified acute myocardial infarction.
Circulation,131, 1477–1485.
How to cite this article: Case SM, Sawhney M, Stewart
JC. Atypical depression and double depression predict New-
Onset cardiovascular disease in U.S. adults. Depress Anxiety.
2018;35:10–17. https://doi.org/10.1002/da.22666
