Journal of Affective Disorders 183 (2015) 195204

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

Research report

Validation of the rule of three, the red sign and temperament

as behavioral markers of bipolar spectrum disorders in a large sample
Diogo R. Lara a,n, Lusa W. Bisol b, Gustavo L. Ottoni c, Hudson W. de Carvalho d,
Dipavo Banerjee e, Shahrokh Golshan f, Kareen Akiskal e, Hagop S. Akiskal f
a

Faculdade de Biocincias, Pontifcia Universidade Catlica do Rio Grande do Sul, Porto Alegre, Brazil
Servio de Psiquiatria, Hospital So Lucas da Pontifcia Universidade Catlica do Rio Grande do Sul, Porto Alegre, Brazil
c
Servio de Psiquiatria, Hospital Presidente Vargas, Fundao Faculdade Federal de Cincias Mdicas de Porto Alegre, Porto Alegre, Brazil
d
Departamento de Psicologia, Universidade Federal de Pelotas, Pelotas, Brazil
e
International Mood Center, La Jolla, CA USA and Paris, France
f
Department of Psychiatry, University of California at San Diego, USA
b

art ic l e i nf o

a b s t r a c t

Article history:
Received 15 February 2013
Received in revised form
4 March 2015
Accepted 27 April 2015
Available online 19 May 2015

Background: Akiskal proposed the rule of three for behavioral indicators with high specicity for
bipolarity in patients with major depression episodes. We evaluated these distinctive behaviors in
controls and subjects with major depression or bipolar disorder.
Methods: data was collected in the BRAINSTEP project with questions on general behaviors, style and
talents. Univariate analysis was rst conducted in 36,742 subjects and conrmatory multivariate analysis
in further 34,505 subjects (22% with a mood disorder). Odds ratios were calculated adjusting for age.
Results: Univariate analysis showed that 29 behavioral markers differentiated bipolar subjects from
those with unipolar depression. The most robust differences in those with bipolarity (ORs 44) were Z 3
religion changes, Z 3 marriages, cheating the partner regularly, having Z60 lifetime sexual partners,
pathological love, heavy cursing, speaking Z3 foreign languages, having Z2 apparent tattoos, circadian
dysregulation and high debts. Most behaviors were expressed in a minority of patients (usually around
530%) and usually the rule of three was the best numerical marker to distinguish those with
bipolarity. However, multivariate analysis conrmed 11 of these markers for differentiating bipolar
disorder from unipolar depression (reversed circadian rhythm and high debts for both genders, Z 3
provoked car accidents and talent for poetry in men, and frequent book reading, Z3 religion changes,
Z60 sexual partners, pathological love Z2 times, heavy cursing and extravagant dressing style in
women).
Limitations: Self-report data collection only.
Conclusions: These behavioral markers should alert the clinician to perform a thorough investigation of
bipolarity in patients presenting with a depressive episode.
& 2015 Elsevier B.V. All rights reserved.

Keywords:
Behavior
Diagnostic markers
Bipolar disorder
Depression
Temperament
Extravagance

1. Introduction
The differential diagnosis between major depression and bipolar disorders currently relies exclusively on the presence of mania
for bipolar type I and hypomania for bipolar type II. This imposes a
major problem since hypomania is often unrecognized by the
patient, undetected by the clinician, may not reach the four-day
cut-off criteria dened in DSM-V, or has not yet emerged in the
course of the mood disorder. Thus, these guidelines suggest that
patients who present with a depressive episode should always be
n
Correspondence to: Faculdade de Biocincias PUCRS, Av. Ipiranga, 6681
Pd12A, Porto Alegre, RS 90619-900, Brazil.
Tel.: 55 51 81219187; fax: 55 51 33203612.
E-mail address: drlara@pucrs.br (D.R. Lara).

http://dx.doi.org/10.1016/j.jad.2015.04.046
0165-0327/& 2015 Elsevier B.V. All rights reserved.

treated for unipolar major depression unless a spontaneous (i.e.

not induced by antidepressants, stimulants or drugs of abuse)
(hypo)manic episode has developed. However, there is substantial
evidence that around half of the patients with a depressive
episode belongs to the bipolar spectrum or meet bipolarity
speciers (Akiskal and Benazzi, 2005; Angst et al., 2010, 2011).
In clinical practice, information on course (e.g. age at onset,
number of episodes, mood uctuations), family history, temperament, comorbidities (e.g. bulimia), intrusive hypomanic symptoms
within the depressive episode (i.e. mixed features) and response to
medication are useful for identifying bipolar spectrum disorders
(Akiskal et al., 1977; Akiskal et al., 1995; McElroy et al., 2005;
Benazzi, 2006; Angst et al., 2011), with or without documented
(hypo)mania. These features transcend ofcial diagnoses but
enable the clinician to provide adequate treatment to these

196

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

patients. Unfortunately, even the currently available diagnosis of

cyclothymia has been understudied and subdiagnosed, and when
detected in patients with an episode of major depression, should
direct treatment towards mood stabilization besides targeting the
acute episode.
Akiskal (2005) advanced the eld of differential diagnosis of
mood disorders by focusing on behavioral markers distinctively
associated with bipolarity. Based on extensive clinical observations
and case records, he listed several behaviors that putatively have
low sensitivity but high specicity for the diagnosis of bipolarity, i.
e. they are relatively rare, but when present, strongly indicate a
bipolar disorder. These biographic signs and behaviors followed
the pattern named as rule of three given their repeated and
cyclic nature (e.g. three or more weddings, religions, foreign
languages). Another facet of the rule of three is the presence
of a triad of specic characteristics, such mood lability, energy
activity and daydreaming (Akiskal et al., 1995). Also, the red
sign denoted a tendency of individuals with bipolarity for
ornamentation and amboyance, as in preference for items with
vibrant colors (Akiskal, 2005). To understand the rationale for this
kind of approach, it is important to consider that different aspects
of the mind, such as mood, personality, cognition and behavior
tend to operate in concert and share neurobiological substrates
(Lara and Akiskal, 2006; Anderson, 2010), meaning that certain
predispositions to particular states (e.g. hypomania), may also be
manifest at the trait behavioral level. Also, temperament and
personality of those with bipolar disorders differ from healthy
controls and those with major depression, typically showing more
cyclothymic and novelty-seeking traits (Nowakowska et al., 2005).
Thus, given the externalized and unstable nature of bipolar
disorders and their underlying temperaments, it is reasonable to
conceive that their biography and behaviors would show distinctive patterns and features.
We have developed a web-based system to anonymously
evaluate the general population in Brazil, aiming to have a large
sample with a thorough assessment of behavioral, psychological
and psychiatric measures. In this enterprise, called the Brazilian
Internet Study on Temperament and Psychopathology (BRAINSTEP), we included questions about the behavioral markers
suggestive of bipolarity proposed by Akiskal (2005) and other
features observed in our clinical practice. The current study sought
to investigate behavioral markers and temperament variables
distinctively associated with bipolar individuals in comparison to
controls and individuals with major depression.
Given that Akiskal's (2005) approach for using behavioral
indicators of bipolarity has received insufcient attention in the
literature (Fulford 2008; Manning, 2010; Culpepper 2010;
Strakowski et al. 2011), in the present research we seek to test
his views on a large sample of subjects in a different culture. As his
proposal includes many behavioral markers, we rst analyzed
each marker separately using an univariate model, which would
better represent a clinical perspective. Next, in an independent
sample, we performed a multivariate model to nd out the core
behavioral markers. We considered that abehavioral marker to be
validated when signicantly more expressed in a bipolar disorder
group compared to controls and unipolar major depression in the
multivariate analysis.

2. Methods
2.1. Ethics
The Ethics Committee of Hospital So Lucas (PUCRS) approved
the protocol of this study. All participants gave their electronic
informed consent before entering the system. This form was

created to fulll the requirements of the National Health Council

of Brazil (Resolution 196/1996) and the Code of Ethics of the World
Medical Association (Declaration of Helsinki). Probands could
cancel their voluntary and anonymous participation at any
moment without justication. Personal information was sent via
a secure and encrypted connection and was stored behind a
rewall. Our system guaranteed anonymity by coding the e-mail
addresses when data was stored, so that no one (even research
staff) could have access to the e-mails of specic participants.
2.2. Sampling and procedures
The data presented are part of the BRAINSTEP web-based
survey (Lara et al., 2012a): a noncommercial, advertisement-free
website in Brazilian Portuguese (www.temperamento.com.br),
which was broadcast on national TV news programs in Brazil
and in major city newspapers. Volunteers reported demographic
variables, which psychiatric diagnoses they had received from
mental health professionals, and answered several self-report
instruments, including the Affective and Emotional Composite
Temperament Scale (AFECTS, Lara et al., 2012b), the Adult SelfReport Inventory (ASRI; Gadow et al., 2004), the Hypomania/
Mania Symptom Checklist (HCL-32; Angst et al., 2005), and the
questions on behavior and style described here.
To ensure data validity and to promote participation and
adherence, participants were promptly informed about the scientic purposes of the study, that participation was anonymous, free
of charge, and that they would receive a 67 page report on their
temperament prole and the probability of having a positive
screening for 19 psychiatric disorders currently and in the past.
They were also informed that the whole system could take 23 h
to complete; however, they could interrupt the process and come
back to where they had stopped for a period of one week after
they had started. After accepting participation, a password to log
in was sent to their email address with a copy of instructions.
To ensure reliability of the data, questions checking for attention were inserted within the instruments. Also, at the end of the
system there were two specic questions on the degree of
attention, sincerity and seriousness of the volunteer while
responding to the instruments. Only those who stated being
attentive, sincere and serious throughout the study and had
correct answers in the attention validity items were included.
Finally, to evaluate the validity status of the diagnosis of bipolarity
received by the mental health professional, the mean scores of the
HCL-32 and the past manic symptoms from the ASRI were used.
All subjects that completed the instruments used in this study
with age ranging from 18 to 65 were included, resulting in a total
of 86,135 volunteers (26.9% males). After the exclusion of those
who failed in the validity checks, the nal sample was composed
of 71,247 subjects, with 19,412 males (27.3%, 31.9 710.6 years old)
and 51,835 females (77.7%, 31.6710.4 years old). This sample was
further divided into two independent sets: the rst set, with
36,742 participants, was used to explore and identify individual
behavioral markers associated with bipolarity via univariate statistics. The second set, composed of 34,505 participants, was used
to conrm and test the pertinence of selected behavioral markers
for bipolarity via multivariate statistics. Fig. 1 depicts the methodological approach regarding the sample.
The prevalence of mood disorders and mean age for each
diagnostic group are shown in Table 1. Diagnostic groups were
dened according to diagnoses ever received by participants from
mental health professionals. Therefore, there were those with
major depression diagnosis only, bipolar disorder diagnosis only
or both diagnoses. Those with major depression only were further
divided according to their affective temperament (internalized/
stable or externalized/unstable see below). Individuals without

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

psychiatric diagnoses (control group) were younger than those

with mood disorders (F3, 36739 391.74, p o0.05) and, therefore,
age was included as a covariate in all analyses.

197

2.3.3. Hypomania/Mania Symptom Checklist (HCL-32)

The HCL-32 is a self-report screening for hypomania and mania
that evaluates 32 symptoms assessed via dichotomous yes or no
scale. A total score of 14 or more is considered to be an indicative
of bipolarity.

2.3. Instruments
2.4. Assessment of behavioral markers
2.3.1. The Affective and Emotional Composite Temperament Scale
(AFECTS)
The affective section of the AFECT scale (Lara et al., 2012b)
contains short descriptions of the twelve affective temperaments,
divided in 4 groups: internalized (depressive, anxious, apathetic),
unstable (cyclothymic, dysphoric, volatile), stable (obsessive,
euthymic, hyperthymic), and externalized (irritable, disinhibited,
and euphoric), which are presented and the subject has to select
which prole is the most suitable to represent his/her temperament. This allows for a categorical evaluation of affective temperament. Participants who had received a diagnosis of major
depression were further categorized in depression with an internalized or stable affective temperament (Depression Int-Stab) or
with an externalized or unstable affective temperament (Depression Ext-Unstab) in order to better explore the concept of bipolar
spectrum.

Since there is no established instrument to assess such a

diversity of behaviors, we developed a specic questionnaire on
general behaviors, social and sexual behavior, appearance style
and artistic talents. Whenever possible, the questions were asked
for quantication of specic behaviors or features starting the
question with How many, such as How many university
courses have you started? followed by radio button options for
the number of times. Exceptions to this format were in the
following questions:

2.3.2. Adult Self-Report Inventory-4 (ASRI)

The ASRI is a self-report psychiatric rating scale that evaluates
various DSM-IV psychiatric disorders. It is composed of 136 items
that describe psychiatric symptoms related to mood, anxiety,
impulsive, compulsive, anti-social, and psychotic symptoms
assessed by a 4-point Likert scale that varies from never, sometimes, often and very often. In the current study, only the items that
estimate past mania episode were used.

 Have you ever been through times when you slept during the






BRAINSTEP
N = 86,135
Validity checks
Participants Excluded
N = 14,889

Selected Sample
N = 71,247

Exploratory Phase
N = 36,742

Confirmatory Phase
N = 34,505

Univariate analyses
Groups based on
mood spectrum and
controls

Multivariate analyses

Test of selected
behavioral markers
for bipolarity

Selection of
behavioral markers
for bipolarity (OR2)

Fig. 1. Diagram of the methodological approach.

day and were awake at night (not due to work)? (a) no; (b) yes,
for less than a month; (c) yes, for more than a month.
Do you usually have the feeling that your mind speeds up or
turns on only at night? (a) no; (b) sometimes; (c) yes.
How do you manage your nances? (a) I spend less than I earn
or have; (b) I spend what I earn or have; (c) I spend more than I
earn or have; (d) I spend much more than I earn or have and I
have high debts.
Do you bite your nails? (a) no; (b) occasionally; (c) regularly;
(d) regularly and it deforms my ngers.
When you are in a stable relationship you: (a) have never
cheated; (b) have kissed someone else but have not had sex;
(c) have had sex with someone else; (d) usually have sex with
someone else (cheats regularly); (e) have not been in a stable
relationship; (f) prefer not to answer.
For pathological love: Have you ever been so in love with or
obsessed about someone that nothing else mattered to you, you
felt you could not live without this person, felt bad when away
from this person and tried to monitor his/her activities? (a) no;
(b) maybe; (c) yes, only once; (d) yes, twice or more.
For swearing: How often do you say bad words? (a) never,
(b) rarely, (c) sometimes, (d) frequently, (e) frequently and with
heavy bad words.
How many tattoos do you have? (a) none, (b) 1 small and easy
to hide, (c) 2 or more easy to hide, (d) 1 apparent (regardless of
having others easy to hide), (e) 2 or more apparent (regardless
of having others easy to hide).
How is your dressing style? (a) conservative or classic (I do not
like to stand out), (b) casual and informal, (c) casual but with
something to stand out, (d) extravagant.
For colors: How do you like the color (color name)? (a) not at
all, (b) like a little, (c) like, (d) like very much.

Table 1
Sample description.
Sample Subset 1
36,742 subjects

Controls
MD
MD/Bip
Bipolar
Total

Sample Subset 2
34,505 subjects

Male (%)

Female (%)

Age (SD)

Male (%)

Female (%)

Age (SD)

8573
1085
182
137
9977

20,790
4883
703
389
26,765

29.2(10.5)*
34.2(10.7)
33.2(10.7)
32.1(10)
31.3(10.3)

7848
1168
257
162
9435

18,392
5467
819
392
25,070

31.2 (10.6)*
35.4 (10.8)
34.7 (10.1)
33.4 (10.3)
32.3(10.7)

(85.9)
(10.9)
(1.8)
(1.4)
(27.2)

(77.6)
(18.2)
(2.6)
(1.5)
(72.8)

MD Major Depression diagnosis; MD/Bip both Major Depression and Bipolar Disorder diagnoses.
n

po 0.001.

(83.1)
(12.4)
(2.7)
(1.7)
(27.3)

(73.3)
(21.8)
(3.3)
(1.6)
(72.7)

198

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

 For artistic talents: Are you recognized by other people as

having talent in any of these areas? yes/no options for music,
poetry, writing, painting, photography, dance, drama.

2.5. Statistical analysis

Mean age between groups was compared with one-way
ANOVA. The proportion of the AFECTS affective temperament
categories in groups was analyzed with chi-square tests. HCL-32
score and past manic symptoms from the ASRI were compared
between groups with age as covariate using ANCOVA. The odds
ratios for the presence of each specic behavioral marker in the
exploratory phase of this study were calculated with Multinomial
Logistic Regression separately for males and females, with age as
covariate and controls as reference, the behavioral marker as the
dependent variable and diagnostic group as the independent
variable. The separate analysis for genders was conducted as
Akiskal's descriptions involved some gender-specic issues. In
the case of continuous variables (e.g. number of religion changes),
the rst response that differentiated groups signicantly was
chosen (e.g. 3) when further responses (e.g. 4 and 5) were also
indicative, then a new variable was created to summarize the
higher numbers (e.g. 3 and higher) and the odds ratio was
recalculated for this new variable. From this exploratory phase,
only variables with odds ratio greater than 2.0 for any bipolar
group (bipolar disorder alone or both bipolar and major depression diagnoses) are shown in the tables and were selected for the
conrmatory phase in order to increase clinical relevance. In the
conrmatory phase, the selected behavioral markers (OR Z2.0)
were entered simultaneously in the model using Multinomial
Logistic Regression. Results were considered signicant if
p o0.05 and shown as odds ratios (OR) and 95% condence
intervals (CI).

3. Results
3.1. Characterization of the sample according to affective
temperament and manic symptoms
The distribution of affective temperaments in sample 1 using
the AFECTS showed that individuals with bipolar disorder (with or
without a diagnosis of depression) more often ascribed to
cyclothymic, dysphoric and euphoric temperament and less often
too anxious, obsessive, euthymic or hyperthymic types
(x2 805,88; DF 11, p o0.001) (Fig. 2). Those with depression
were more often depressive, anxious, cyclothymic or obsessive.
The proportion of depressive and volatile temperaments was
higher in individuals with both depression and bipolar disorder
diagnoses. Considering this pattern of distribution, we divided the
group of individuals with depression in two types: those with
internalized or stable temperament types (Depression Int-Stab)
and those with externalized or unstable types (Depression ExtUnstab).
To examine if past manic symptoms were present in the bipolar
groups and not in the depression groups, we compared HCL-32
score and past manic symptoms from the ASRI between groups
with age as covariate using ANCOVA. HCL-32 (F(5, 20,912) 158,25;
p o0.001) scores (mean, 95%CI) were: controls (17.53, 17.4517.61),
Depression Int-Stab (17.69, 17.4817.91), Depression Ext-Unstab
(19.59, 19.3519.84), Depression-Bipolar (22.22, 21.7022.76) and
Bipolar (21.88, 21.4022.22). ASRS (F(5, 21,950) 400,03; p o0.001)
scores (mean, 95%CI) were: controls (6.66, 6.606.71), Depression
Int-Stab (7.10, 6.947.26), Depression Ext-Unstab (9.11, 8.929.29),
Depression-Bipolar (12.18, 11.7812.57) and bipolar (11.50,

Fig. 2. Distribution of affective temperaments in controls (n 27,972), patients

with depression diagnosis only (n5968), both depression and bipolar disorder
diagnoses (n 885), and only bipolar disorder diagnosis (n 526). n signicantly
higher frequency, # signicantly lower frequency, chi-square test (p o 0.05).

11.2011.81). These results show that bipolar groups differ from

controls and from depression groups for past manic symptoms and
suggest that the Depression Ext-Unstab is an intermediate group
in the continuum between pure unipolar depression and bipolar
disorder.

3.2. Exploratory phase for behavioral markers using univariate

analysis.
Table 2 shows the general behaviors that differentiated those
with bipolar disorder from controls and, in most cases, from
individuals in the Depression Int-Stab group in both males and
females., As indicated by higher ORs and lack of overlap between
condence intervals, the most robust differences were that individuals with bipolar disorder had more frequent alterations in
circadian cycle, speeding up of the mind at night, high spending
and debts and more religion changes than controls.
The social behaviors shown in Table 3 also discriminated
individuals with bipolar disorders, with more robust differences
between both groups with bipolar disorder from groups with
depression in females. Ors and condence intervals showed that
Bipolarity was associated with more romantic relationships and
with higher age difference from partners (except for males having
younger partners), and with more diverse sexual behavior, including number of partners, extramarital relationships, group sex and
sadomasoquist sex. Pathological love (twice or more) was more
common in all mood disorder groups, especially in women with
bipolar disorder. Bipolarity was also associated with more frequent
and heavy swearing and speaking more foreign languages.
Individuals with bipolar disorders more often had apparent
tattoos, piercings, change hairstyle frequently and adopted an
extravagant dressing style (Table 4). They also reported liking the
color pink, especially males, but the results for the color red were
not remarkable. Most parameters also differentiated the bipolar
groups from the Depression Int-Stab group, as evidenced by the
lack of overlap in condence intervals. In arts, the most pronounced difference in bipolar groups (OR 42) was for being more
talented in poetry. They were also signicantly more talented in
music, painting and drama, but with OR o2 (data not shown). As a
result, bipolarity was associated with having 4 or more talents
more often.

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

199

Table 2
General behaviors in controls and individuals with mood disorders (univariate analysis).
Males

Day-night
switch4 1month
vs never

OR
CI
Mind speeds up at night %
vs not
OR
CI
Z3 university courses
%
started vs none
OR
CI
Z3 professions
%
vs none
OR
CI
Z3 driving tickets last 12 %
months vs none
OR
CI
Z3 car accidents lifetime %
provoked vs none
OR
CI
High spending and debts %
vs low spending
OR
CI
46 scars 4 1cm
%
vs none
OR
CI
Reads4 1 book per
%
month
vs does not read books
OR
CI
Z3 religions
%
vs no religion change
OR
CI
Nail-biting with
%
deformed
OR
ngers vs does not bite
nails
CI

Females

Controls

Depression
Int-Stab

Depression
Ext-Unstab

Depression
Bipolar

Bipolar

Controls

Depression
Int-stab

Depression
Ext-Unstab

Depression
Bipolar

Bipolar

12.0

19.4

22.2

30.1

32.9

7.3

12.3

16.8

21.9

24.8

2.01
1.722.34
30.2
2.03
1.762.35
11.5
1.70
1.382.10
14.5
0.89
0.731.01
5.8
1.13
0.881.44
4.5
0.95
0.721.25
3.9
1.38
1.031.87
2.6
0.63
0.440.89
3.4

2.89
2.423.45
34.8
3.09
2.593.70
11.5
1.60
1.232.06
15.6
1.17
0.921.48
8.0
1.71
1.312.21
8.1
2.06
1.582.68
7.4
3.34
2.514.40
5.3
1.52
1.102.11
3.9

3.23
2.434.31
32.7
2.90
2.183.85
17.0
2.78
1.894.07
19.6
1.62
1.112.37
8.3
1.93
1.272.95
11.9
3.24
2.244.70
8.7
4.00
2.586.18
6.4
1.54
0.942.52
5.1

5.27
4.176.67
41.3
4.47
3.485.75
17.6
3.23
2.324.49
18.0
1.42
1.031.97
8.0
1.63
1.142.33
8.0
1.89
1.312.72
11.8
5.84
4.208.16
6.9
2.15
1.423.24
5.1

2.36
2.172.57
23.2
1.82
1.691.95
7.6
1.40
1.241.58
10.3
1.25
1.121.38
1.7
0.95
0.771.17
1.2
1.22
0.941.58
4.4
1.50
1.301.72
1.8
1.09
0.891.33
3.3

3.09
2.813.39
28.6
2.55
2.342.78
7.7
1.27
1.111.46
10.6
1.40
1.231.58
3.3
2.07
1.712.51
1.6
1.91
1.462.50
7.4
3.62
3.144.20
3.1
1.99
1.642.42
2.8

5.48
4.956.54
32.0
3.55
2.964.25
10.8
2.89
2.193.83
11.6
1.79
1.382.32
4.2
2.80
1.973.97
3.2
4.02
2.676.06
9.6
5.07
3.886.63
3.3
1.85
1.232.76
5.6

6.98
6.147.92
42.9
4.83
4.215.5
12.1
2.34
1.942.83
13.4
1.82
1.532.18
3.1
1.99
1.482.68
2.0
2.37
1.633.43
14.0
8.20
6.839.80
4.3
2.82
2.163.66
5.1

1.36
1.091.70
5.3
1.51
1.161.96
5.4

1.54
1.162.03
6.8
1.99
1.502.65
8.5

2.66
1.744.05
11.2
3.39
2.305.00
8.7

2.11
1.492.99
12.2
3.98
2.895.49
8.9

1.14
1.011.29
5.1
1.60
1.401.83
3.3

0.91
0.791.05
6.8
2.23
1.942.57
5.1

2.08
1.542.80
10.5
3.68
2.874.71
5.3

1.55
1.251.93
11.7
4.30
3.605.13
6.6

1.11
0.861.42

1.80
1.402.32

1.70
1.122.58

2.02
1.432.85

1.23
1.061.44

1.85
1.592.16

1.85
1.352.53

2.65
2.143.27

19.7
1
6.6
1
12.4
1
4.3
1
3.6
1
2.9
1
4.1
1
2.5
1
3.7
1
5.8
1

3.3. Conrmatory phase using multivariate analysis

Based on results from the univariate analyses, 29 behavioral
markers with OR 42 comparing controls to depression/bipolar
and bipolar disorder groups were selected and tested. As shown in
Table 5, using multivariate analysis, most ORs decreased and 19
markers remained statistically signicant comparing controls to
any bipolar disorder group. Of these, 15 showed OR4 2. However,
only 11 behavioral markers were signicantly more frequent in a
bipolar disorder groups compared to the Depression Int-Stab
group in at least one gender (shown in bold in Table 5). Two
behavioral markers were signicant for both genders, while three
separate markers for males and seven separate markers for
females were signicant (Table 6).

4. Discussion
In the univariate analysis, 29 behavioral markers were able to
differentiate at least one bipolar group in at least one gender from
those with depression with internalized or stable temperament
(Depression Int-Stab). Moreover, many behavioral markers differentiated both bipolar groups from controls and the Depression IntStab group in both genders. As originally proposed by Akiskal
(2005) on the basis of clinical observation, in most cases the rule
of three was the best numerical marker to distinguish those with
bipolarity. However, in the conrmatory phase using an

18.5
1
4.9
1
7.7
1
1.5
1
0.8
1
3.0
1
2.0
1
2.8
1
3.4
1
3.2
1

independent sample and multivariate analysis, only 2 markers

were signicantly different in bipolar groups compared to Depression Int-Stab group in both genders (day-night switch and high
debts) and another 9 markers were gender-specic.
The most robust and consistent differences in those with
bipolarity were seen in day-night switch longer than one month
and having high debts in both genders. A lifetime history of Z3
provoked car accidents and talent for poetry were signicant
markers for men only. Women with a bipolar diagnosis stood
out for having Z3 religion changes, Z60 or more lifetime sexual
partners, pathological love Z2 times, heavy cursing, mind speeding up at night, frequent book reading and extravagant dressing
style. Previous ndings of large odds ratios for marital divorce (3.3
in men and 4.8 in women) (Kessler and Walters, 1998) and
polyglotism (Rihmer, 1982) were conrmed in the univariate
analysis, but were not signicant in the multivariate analysis.
The presence of the observed behavioral markers in a patient
should alert and motivate the clinician to perform a thorough
investigation of bipolarity in patients presenting with a depressive
episode. However, lack of these traits should not be regarded as
protective against a bipolar diagnosis. These results also identify
some readily observable features and behaviors predictive of bipolarity during psychiatric evaluation, such as extravagant dressing and
use of bad words in the clinical setting, especially in women. The
other markers are easily accessible only during interview, with the
exception of intimate issues, such as sexual behavior, which requires
skillful questioning. Of note, these features are also found in the

200

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

Table 3
Social behavior in controls and individuals with mood disorders (univariate analysis).
Males

Z3 marriages
vs none
Z7 relationships lasting
42
months vs none
Had partner 410 years
younger vs never had
younger partner
Had partner 410 years
older
vs never had older
partner
Z60 sexual partners
vs no sexual partner

Regularly cheats partner

vs never cheats
Sadomasoquist sex ever
vs never
Group sex or swing ever
vs never
Pathological love Z 2
times
vs never

%
OR
IC
%
OR
IC
%
OR
IC
%

Controls

Depression
Int-Stab

Depression
Ext-Unstab

Depression
Bipolar

Bipolar

Controls

Depression
Int-Stab

Depression
Ext-Unstab

Depression
Bipolar

Bipolar

3.3
1

4.6
0.83
0.621.12
10.1

6.8
1.54
1.122.12
13.4

7.4
2.35
1.413.93
13.8

9.1
1.68
1.122.51
17.1

2
1

2.0
1.36
1.161.60
9.6

3.6
2.65
2.223.16
11.2

4.6
3.04
2.114.38
14.2

8.1
4.60
3.705.75
14.9

0.68
0.530.86
14.0
0.97
0.791.20
9.2

1.35
1.021.80
12.7
1.32
1.001.73
10.5

1.52
0.952.43
14.4
1.24
0.811.90
12.5

1.57
1.092.29
13.1
0.92
0.641.32
13.8

1.16
1.031.32
5.2
1.35
1.181.55
21.7

2.05
1.772.38
5.1
1.80
1.532.11
23.6

3.24
2.364.45
6.1
2.30
1.683.15
27.4

2.96
2.363.71
6.9
2.56
2.063.19
28.1

1.18
0.961.45
8.8
1.21
0.871.69

1.49
1.181.89
12.2
2.57
1.683.94

1.82
1.262.63
18.3
4.63
2.269.55

2.14
1.582.90
14.3
4.11
2.088.12

1.25
1.131.38
2.2
1.36
1.091.70

1.66
1.481.87
2.8
2.93
2.273.79

2.15
1.672.77
4.6
8.80
5.0315.42

6.4
0.80
0.621.01
6.2
1.29
1.021.63
16.5
1.00
0.861.17
19.0

11.5
1.87
1.482.37
8.1
1.70
1.322.18
24.6
1.67
1.421.95
21.9

10.6
1.72
1.152.57
9.3
1.98
1.342.92
27.6
1.94
1.512.50
22.4

10.0
1.53
1.092.16
9.6
2.06
1.492.85
25.2
1.71
1.372.13
24.3

1.2
1

1.3
1.10
0.871.41
5.0
1.21
1.061.37
4.2
1.10
0.961.27
15.9

2.4
2.41
1.943.01
7.3
1.69
1.481.91
6.7
1.81
1.582.07
19.6

3.9
3.94
2.715.71
10.3
2.43
1.933.06
10.2
2.83
2.253.57
23.4

2.00
1.672.40
6.3
7.81
5.52
11.04
3.4
3.63
2.714.86
10.5
2.56
2.163.02
10.1
2.81
2.363.34
28.3

2.09
1.772.48
5.3
1.40
0.991.98
3.5
1.45
1.061.98

3.22
2.623.97
8.4
3.11
2.034.78
2.9
1.22
0.811.84

2.93
2.114.10
6.4
1.62
0.833.16
7.4
3.18
2.025.01

3.45
2.614.57
8.5
2.07
1.233.48
5.6
2.47
1.603.81

2.00
1.842.19
2.6
1.24
1.031.50
2.4
1.32
1.091.58

3.08
2.783.40
4.5
3.76
3.064.62
2.0
1.04
0.821.31

3.93
3.194.84
7.2
4.65
3.226.73
4.2
2.82
1.984.02

5.25
4.516.10
7.8
6.37
4.858.41
2.1
1.34
0.941.91

9.7
1
8.5
1
7.9

OR
IC
%
OR
IC

%
OR
IC
%
OR
IC
%
OR
IC
%

7.0
1

OR
IC
Frequent use of heavy bad %
words vs never
OR
IC
Z3 foreign languages
%
vs none
OR
IC

Females

6.6
1

5.0
1
16.0
1
11.4
1
4.7
1
2.6
1

general population and do not necessarily imply a psychiatric

diagnosis, but in the clinical setting, especially a combination of
them, should raise the suspicion for a bipolar spectrum disorder.
Our ndings also showed that cyclothymic, dysphoric and euphoric
affective temperaments were more frequent among subjects with a
bipolar disorder diagnosis, which was compensated by their less
frequent identication with anxious, obsessive, euthymic or hyperthymic types. The concept of affective temperament from the AFECTS was
based on the TEMPS-A (Akiskal et al., 2005), but with some distinctions and inclusion of other types. The hyperthymic temperament in
the AFECTS is dened as a high energy and stable prole, whereas the
new euphoric type includes more externalizing features of excessive
behaviors, impulsivity, emotional intensity, distractibility and rulebreaking traits. Our data shows that it is the euphoric temperament
that is overrepresented in bipolar disorders, whereas this adaptive
denition of hyperthymic temperament is underrepresented. However, our analysis did not include the affective temperament prole
previous to the development of the mood disorder, which may reveal a
different proportion of hyperthymics. Nevertheless, cyclothymic was
the most prevalent temperament in the bipolar groups (32%), as
repeatedly shown to be associated with the bipolar spectrum (Akiskal
et al., 1983; Akiskal, 1994; Hantouche et al., 2003; Kesebir et al., 2005;
Akiskal et al., 2006; Perugi et al., 2012). Despite four decades of
research on the bipolar spectrum and the importance of early
diagnostic markers for an illness as devastating as bipolar disorder,
this approach continues to be criticized (Kuiper et al., 2013). The

6.3
1
2.6
1
17.5

1.5
1

4.7
1
3.9
1
10.1
1
2.5
1
1.8
1

position espoused in the present paper within the time-honored

theoretical and clinical position that what we call bipolar disorder
today and which was called manic-depressive psychosis in its historical origins, arises from temperamental substrates which represent
accentuation of normal emotionality and temperaments (Kraepelin,
1921; Akiskal et al. 1977; Marneros 2001; Akiskal, 2006). Indeed, some
of the most accomplished men and women since Aristotle's historical
observation have been characterized by temperamental excesses
ranging from normality to full-blown bipolar disorder (Akiskal and
Akiskal, 2007). Current data, two millennia later, support the hypothesis that the genetics of bipolar disorder may reside in such temperamental features characteristic of the hyperthymic/euphoric and
choleric types identied within families with mania (Greenwood
et al., 2012).
In this Brazilian internet sample, we found 3.24.1% prevalence
of bipolar disorders and observed that 40% of individuals with a
diagnosis of depression, but not bipolar disorder, had externalized
or unstable temperaments commonly associated with bipolar
disorder. This group also showed intermediate values of manic
symptoms and odds ratios for most behavioral markers, which is
in line with the bipolar spectrum concept. Interestingly, the
National Comorbidity Survey replication in the United States
reported 2.1% as the lifetime prevalence of bipolar disorder and
2.4% subthreshold bipolarity dened as recurrent hypomania
without a major depressive episode or with fewer symptoms than
required for threshold hypomania (Merikangas et al., 2007). A

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

201

Table 4
Appearance, style and talents in controls and individuals with mood disorders (univariate analysis).
Males

Z2 apparent tattoos
vs none
Z3 piercings
vs none
Z3 hair style changes/
year
vs no change
Extravagant dressing
style vs discreet
style
Likes pink very much
vs likes pink
Talent for poetry
vs no talent for poetry
Z4 talents
vs none

%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI

Females

Controls

Depression
Int-Stab

Depression
Ext-Unstab

Depression
Bipolar

Bipolar

Controls

Depression
Int-Stab

Depression
Ext-Unstab

Depression
Bipolar

Bipolar

2.0
1

2.1
1.14
0.771.67
1.8
1.53
0.992.36
2.7

2.9
1.70
1.132.56
1.8
1.50
0.902.52
5.6

5.8
3.31
2.025.41
3.2
2.82
1.475.40
5.4

4.2
2.58
1.604.17
2.7
2.48
1.374.49
7.6

2.4
1

3.1
1.51
1.281.77
1.1
1.25
0.971.61
12.5

5.6
2.92
2.513.39
1.9
1.87
1.462.38
16.6

6.3
3.34
2.574.66
2.8
2.62
1.714.03
19.2

7.9
4.38
3.595.34
2.5
2.61
1.873.63
18.7

1.02
0.721.42
9.4

2.27
1.6730.07
14.0

2.20
1.323.66
16.7

3.33
2.304.82
22.5

1.01
0.931.10
24.7

1.60
1.451.76
32.9

1.67
1.382.02
38.1

1.73
1.502.00
34.7

0.81
0.661.00
6.6
1.33
0.991.78
12.8
1.10
0.851.42
9.2
1.18
0.981.43

1.48
1.171.87
7.8
1.80
1.282.59
16.0
1.36
10.21.82
10.5
1.35
1.081.69

1.62
1.132.33
12.6
3.34
2.055.48
25.3
2.63
1.724.02
14.4
1.93
1.402.67

2.69
2.013.60
8.6
1.95
1.263.06
21.2
2.15
1.463.16
14.9
2.05
1.572.68

0.74
0.680.80
27.5
1.01
0.941.08
10.6
1.25
1.091.43
8.2
1.01
0.921.11

1.46
1.321.62
30.5
1.14
1.051.23
12.1
1.23
1.111.51
10.7
1.31
1.181.46

1.77
1.412.21
34.9
1.13
0.921.39
16.4
1.92
1.422.59
16.2
2.10
1.742.54

1.35
1.151.57
33.8
1.36
1.181.67
16.3
1.85
1.472.32
11.7
1.46
1.251.71

1.5
1
3.2
1
11.5
1
5.1
1
12.3
1
8.4
1

reanalysis of these data also demonstrated that 40% of those with

depression had a history of subthreshold hypomania dened as a
positive screening question for hypomania but not meeting its full
diagnostic criteria (Angst et al., 2010). More recently, the BRIDGE
global survey also showed that 16% of patients with a depressive
episode have bipolar disorder and another 47% met the criteria for
the bipolarity specier (Angst et al., 2011).
These behavioral markers are also revealing about the underlying nature of bipolarity. Akiskal and Akiskal (2007) have proposed that the cyclothymic temperament is involved in romance
and creativity and hyperthymic traits favor leadership and territoriality. The behavior overexpressed in those with bipolar disorders indicate in general they push the borders of convention
and differentiate these individuals by extending activity
throughout the night, seeking novelty and excitement in several
areas, engaging more times and more intensely in romantic
relationships, communicating profusely, and adorning themselves.
Many of these behaviors have a symbolic nature, such as body
ornaments, arts and developed language abilities, whereas others
seem to be manifestations of unrestrained appetitive impulses,
such as in sex and drug use. These symbolic attributes are
distinctively developed in humans, suggesting that their genetic
background has substantial evolutionary relevance, whereas
increased sexual behavior obviously favors transmission of these
traits (Akiskal and Akiskal, 2005). The observations of Akiskal in
the United States were mostly conrmed in this large Brazilian
sample, suggesting some transcultural validity of his ndings.
Indeed these two studies have been carried out in rather distinct
cultures; the original Akiskal study was done in Memphis, which
does not include a signicant Latino population. Therefore, our
general contention for transcultural validity is upheld, not on the
basis of countries of origin, but on the basis dominant subcultures
in each of the countries.
This study has limitations to be considered. First, the data was
collected by internet in Brazilian Portuguese, relying solely on the
participants' information. Also, the diagnosis of cases was made by
their mental health clinicians and informed to them, without
standardized measures. However, bipolar groups had higher

1.3
1
11.6
1
27.8
1
30.6
1
9.0
1
8.7
1

scores of past manic symptoms, showing at least a distinction in

this regard. There may be undiagnosed cases of mood disorders in
the control group, but the large number of individuals without a
major psychiatric disorder probably diluted their contribution.
Furthermore, the higher proportion of cyclothymic, dysphoric
and euphoric temperaments in the bipolar groups further indicates the validity of these reported diagnoses. Mood state was not
evaluated, but most questions are on trait behaviors, not on states,
and marked depressive or manic states are incompatible with
responding a long internet system (at least 450 questions to be
included in this analysis). Finally, to reduce the number of
undetected bipolarity in the depression group, the depression
Ext-Unstab group was created, which indeed had intermediate
scores in past manic symptoms. Another limitation was the lack of
standardized instruments to assess most of these behavioral
markers. On the other hand, the strengths of this study were the
large sample size and the use of anonymous internet-based
assessment, which facilitates the acquisition of reliable information on sensitive issues such as sexual behavior and drug abuse
(Turner et al., 1998; Pealer et al., 2001).
In conclusion, several behavioral markers and temperamental
features were distinctively expressed in bipolar disorder patients
compared to controls and patients with depression using univariate analysis. Eleven of these traits remained differently expressed
in subjects with bipolar disorders compared to those with unipolar
depression. These traits can be used to enhance clinical wisdom,
especially when evaluating patients during a depressive episode.
Future studies should evaluate the predictive validity of these
traits for the diagnosis of bipolarity. Finally, transcultural applicability must be evaluated in other countries.

Role of funding source

The funding agency had no inuence on design of the study or interpretation of
results.

Conict of interest
The authors declare no nancial disclosures relevant to this paper.

202

Table 5
Behavioral markers in controls and individuals with mood disorders (multivariate analysis).
MALES

Day-night switch 41month

vs never
Mind speeds up at night
vs not

Z3 driving tickets last 12

months vs none
Z3 car accidents lifetime
provoked vs none
High spending and debts
vs low spending
Reads 41 book per month
vs does not read books
Z3 religions
vs no religion change
Nail-biting with deformed
ngers vs does not bite nails
3 Z marriages
vs none
Z 60 sexual partners
vs no sexual partner
Regularly cheats partner
vs never cheats
Pathological love Z2 times
vs never
Frequent use of heavy bad
words vs never
Z3 foreign languages
vs none
Z2 apparent tattoos
vs none

Controls

Depression Int-Stab

Depression Ext-Unstab

Depression Bipolar

Bipolar

Controls

Depression Int-Stab

Depression Ext-Unstab

Depression Bipolar

Bipolar

13.0
1

19.6
1.59
1.271.99
30.2
1.59
1.281.98
11.7
1.60
1.182.16
5.2
1.22
0.841.76
4.1
0.93
0.611.14
3.9
0.77
0.471.26
19.1
1.11
0.851.45
4.7
1.70
0.721.47
5.4
1.03
0.721.47
4.4
0.75
0.481.17
7.9
1.17
0.811.70
16.0
1.12
0.871.44
20.9
0.58
0.350.94
4.8
1.23
0.742.05
3.7
1.46
0.932.28
2.8
0.98
0.591.62

23.8
1.78
1.352.34
34.8
1.64
1.242.17
11.3
1.17
0.801.70
8.1
1.28
0.851.91
9.7
2.12
1.443.11
9.5
2.41
1.384.21
20.1
1.29
0.921.82
6.3
1.19
0.761.86
9.7
1.60
1.112.30
6.8
0.88
0.541.44
9.5
1.19
0.721.96
20.1
1.52
1.112.08
22.6
0.97
0.571.62
8.1
1.58
0.852.95
2.7
1.05
0.551.99
3.2
0.98
0.541.78

32.7
3.03
2.154.27
41.2
1.89
1.312.72
18.7
2.73
1.724.33
8.2
1.30
0.772.19
8.6
1.73
1.032.92
12.8
3.85
1.887.90
26.1
0.99
0.641.52
11.3
1.85
1.142.99
7.4
1.17
0.691.99
9.7
1.31
0.742.31
7.4
2.57
1.205.46
17.1
0.97
0.641.48
23.0
1.13
0.641.99
9.3
1.18
0.562.46
6.2
2.31
1.254.26
3.5
0.80
0.381.69

22.8
1.47
0.932.32
40.1
1.92
1.232.99
16.7
2.05
1.153.62
10.5
2.31
1.284.14
14.2
3.19
1.845.51
8.6
1.72
0.694.25
27.8
2.18
1.213.93
10.5
1.82
0.983.37
4.3
0.75
0.331.69
8.0
1.34
0.632.85
16.7
2.08
0.864.98
18.5
1.07
0.641.78
24.1
0.85
0.461.56
5.6
2.37
0.609.29
8.0
2.14
1.084.23
6.8
1.49
0.723.09

8.8
1

14.3
1.92
1.702.17
28.3
1.42
1.281.58
7.6
1.22
1.021.45
1.7
0.85
0.621.16
1.2
1.17
0.791.72
4.4
1.00
0.791.27
21.7
1.07
0.931.24
4.7
1.19
0.981.44
3.7
0.97
0.791.19
3.6
1.09
0.861.37
2.4
1.22
1.011.48
5.9
0.90
0.761.07
12.2
0.66
0.470.91
2.6
0.92
0.701.20
2.1
1.21
0.911.60
4.0
1.17
0.951.43

16.7
1.81
1.562.09
32.2
1.56
1.361.78
7.2
1.06
0.861.31
2.5
1.23
0.891.70
1.6
1.68
1.122.52
7.5
1.63
1.242.15
18.7
0.84
0.701.00
6.7
1.42
1.151.75
5.7
1.30
1.051.60
4.4
1.43
1.101.87
2.7
1.36
10.71.73
6.2
0.92
0.751.14
16.7
0.94
0.671.32
4.0
1.52
1.122.07
1.6
0.88
0.611.29
6.9
1.37
1.121.69

29.5
3.09
2.533.77
43.5
2.01
1.612.44
11.0
1.50
1.202.01
2.2
0.75
0.441.26
1.5
1.20
0.632.26
13.6
2.22
1.523.22
24.4
1.02
0.771.35
10.6
2.08
1.572.74
7.0
1.55
1.142.11
7.0
1.48
1.042.10
5.9
1.93
1.312.85
7.0
0.93
0.681.28
21.2
1.09
0.771.55
7.7
1.92
1.272.90
2.1
1.01
0.591.72
9.3
1.56
1.172.08

26.3
2.56
1.923.42
41.3
1.57
1.172.09
8.9
1.76
1.112.77
4.6
1.53
0.892.62
2.6
1.84
0.903.76
9.4
1.30
0.762.21
31.9
2.04
1.313.18
12.5
2.50
1.743.60
5.9
1.15
0.731.82
5.4
1.52
0.892.59
3.3
2.81
1.545.12
7.9
1.16
0.761.76
30.8
1.55
1.052.28
7.4
1.82
1.023.23
3.6
1.37
0.752.50
6.9
0.94
0.601.48

23.4
1
6.3
1
3.8
1
3.5
1
3.3
1
16.7
1
3.9
1
5.8
1
3.7
1
5.3
1
16.0
1
13.8
1
4.4
1
2.3
1
2.7
1

22.0
1
4.5
1
1.3
1
0.7
1
3.3
1
19.3
1
3.5
1
4.0
1
2.2
1
2.1
1
1
7.3
1
10.3
1
2.4
1
1.6
1
3.7
1

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

Z3 university courses
started vs none

%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI
%
OR
CI

FEMALES

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

203

4.1
2.34
1.214.51
34.9
0.75
0.521.08
17.1
1.30
0.921.84

Table 6
Signicant behavioral markers for bipolarity compared to Depression Int-Stab
group (multivariate analysis).

1.2
0.87
0.551.40
30.6
0.94
0.781.13
11.2
1.16
0.971.39

3.2
1.26
0.752.10
34.2
1.13
0.851.51
15.6
1.38
071.78

Both males and females

Daynight switch for Z 1 month
High spending and high debts
Males
Z 3 car accidents provoked lifetime (males)
Talent for poetry
Females
Speeding up of the mind at night
Reading Z 1 book per month
Z 3 religion changes
Having Z 60 lifetime sexual partners
Pathological love Z 2 times
Heavy cursing
Extravagant dressing style

References

11.6
1
Talent for poetry
vs none

The behavioral markers in bold were signicantly different between a bipolar disorder group and Depression Int-Stab group.

8.9
1

30.6
1
5.2
1
Likes pink very much
vs likes pink

Extravagant dressing
style vs discrete

%
OR
CI
%
OR
CI
%
OR
CI

0.6
1

0.7
1.07
0.402.85
6.5
1.30
0.921.85
12.1
0.92
0.681.22

1.6
1.56
0.623.95
7.7
1.22
0.801.84
15.2
1.13
0.791.61

1.6
2.18
0.667.20
7.8
0.96
0.551.66
18.7
1.32
0.852.04

1.9
1.34
0.335.44
12.3
2.01
1.133.58
24.1
2.51
1.474.30

0.9
1

0.8
0.69
0.441.06
27.4
1.06
0.911.24
10.1
1.11
0.951.29

Acknowledgment
This research was funded by PRONEX (10/0055-0) and DRL is a CNPq research
fellow.

Akiskal, H.S., 2005. Searching for behavioral indicators of bipolar II in patients

presenting with major depressive episodes: the red sign, the rule of three
and other biographic signs of temperamental extravagance, activation and
hypomania. J. Affect. Disord. 84 (23), 279290.
Akiskal, H.S., 2006. Special issue on circular insanity and beyond: historic
contributions of French psychiatry to contemporary concepts and research on
bipolar disorder. J. Affect. Disord. 96 (3), 141143.
Akiskal, H.S., Akiskal, K.K., 2007. In search of Aristotle: temperament, human
nature, melancholia, creativity and eminence. J. Affect. Disord. 100 (13), 16.
Akiskal, H.S., 1994. The temperamental borders of affective disorders. Acta
Psychiatr. Scand. Suppl. 379, 3237.
Akiskal, H.S., Djenderedjian, A.M., Rosenthal, R.H., Khani, M.K., 1977. Cyclothymic
disorder: validating criteria for inclusion in the bipolar affective group. Am. J.
Psychiatry 134 (11), 12271233.
Akiskal, H.S., Hirschfeld, R.M., Yerevanian, B.I., 1983. The relationship of personality
to affective disorders. Arch. Gen. Psychiatry 40 (7), 801810.
Akiskal, H.S., Kilzieh, N., Maser, J.D., Clayton, P.J., Schettler, P.J., Traci Shea, M.,
Endicott, J., Scheftner, W., Hirschfeld, R.M., Keller, M.B., 2006. The distinct
temperament proles of bipolar I, bipolar II and unipolar patients. J. Affect.
Disord. 92 (1), 1933.
Akiskal, K.K., Akiskal, H.S., 2005. The theoretical underpinnings of affective
temperaments: implications for evolutionary foundations of bipolar disorder
and human nature. J. Affect. Disord. 85, 231239.
Akiskal, H.S., Maser, J.D., Zeller, P.J., Endicott, J., Coryell, W., Keller, M., Warshaw, M.,
Clayton, P., Goodwin, F., 1995. Switching from unipolar to bipolar II. An 11-year
prospective study of clinical and temperamental predictors in 559 patients.
Arch. Gen. Psychiatry 52 (2), 114123.
Akiskal, H.S., Mendlowicz, M.V., Jean-Louis, G., Rapaport, M.H., Kelsoe, J.R., Gillin, J.
C., Smith, T.L., 2005. TEMPS-A: validation of a short version of a self-rated
instrument designed to measure variations in temperament. J. Affect. Disord 85
(12), 4552.
Akiskal, H.S., Benazzi, F., 2005. Optimizing the detection of bipolar II disorder in
outpatient clinical practice: toward a systematization of clinical diagnostic
wisdom. J. Clin. Psychiatry 66 (7), 914921.
Anderson, M.L., 2010. Neural reuse: a fundamental organization principle of the
brain. Behav. Brain Sci. 33 (4), 245266.
Angst, J., Adolfsson, R., Benazzi, F., Gamma, A., Hantouche, E., Meyer, T.D., Skeppar,
P., Vieta, E., Scott, J., 2005. The HCL-32: towards a self-assessment tool for
hypomanic symptoms in outpatients. J. Affect. Disord. 88 (2), 217233.
Angst, J., Azorin, J.M., Bowden, C.L., Perugi, G., Vieta, E., Gamma, A., Young, A.H.,
BRIDGE Study Group, 2011. Prevalence and characteristics of undiagnosed
bipolar disorders in patients with a major depressive episode: the BRIDGE
study. Arch. Gen. Psychiatry 68 (8), 791798.
Angst, J., Cui, L., Swendsen, J., Rothen, S., Cravchik, A., Kessler, R.C., Merikangas, K.R.,
2010. Major depressive disorder with subthreshold bipolarity in the National
Comorbidity Survey Replication. Am. J. Psychiatry 167 (10), 11941201.
Benazzi, F., 2006. Mood patterns and classication in bipolar disorder. Curr. Opin.
Psychiatry 19 (1), 18.
Culpepper, L., 2010. The role of primary care clinicians in diagnosing and treating
bipolar disorder. Prim. Care Companion J. Clin. Psychiatry 12 (Suppl 1),
S4S9.
Gadow, K.D., Sprafkin, J., Weiss, M.D., 2004. AdultSelf-Report Inventory-4 and Adult
Inventory Manual. Checkmate Plus, Stony Brook, NY.
Greenwood, T.A., Akiskal, H.S., Akiskal, K.K., Bipolar Genome StudyKelsoe, J.R., 2012.
Genome-wide association study of temperament in bipolar disorder

204

D.R. Lara et al. / Journal of Affective Disorders 183 (2015) 195204

reveals signicant associations with three novel Loci. Biol. Psychiatry 72 (4),
303310.
Hantouche, E.G., Angst, J., Akiskal, H.S., 2003. Factor structure of hypomania:
interrelationships with cyclothymia and the soft bipolar spectrum. J. Affect.
Disord. 73 (12), 3947.
Kesebir, S., Vahip, S., Akdeniz, F., Ync, Z., Alkan, M., Akiskal, H., 2005. Affective
temperaments as measured by TEMPS-A in patients with bipolar I disorder and
their rst-degree relatives: a controlled study. J. Affect. Disord. 85 (12),
127133.
Kessler, R.C., Walters, E.E., Forthofer, M.S., 1998. The social consequences of
psychiatric disorders, III: probability of marital stability. Am. J. Psychiatry 155,
10921096.
Kuiper, S., McLean, L., Malhi, G.S., 2013. To BD or not to BD: function neuroimaging
and the boundaries of bipolarity. Expert Rev. Neurother. 13 (1), 7587.
Kraepelin, 1921. Manic-Depressive Insanity and Paranoia. Edinburgh: E.&S. Livingstone. Originally published as Psychiatrie.EinLehrbuch fur Studierende und
Arzte, ed. 2. KlinischePsychiatrie. II. Johann Ambrosius Barth, Leipzig (1899).
Lara, D.R., Akiskal, H.S., 2006. Toward an integrative model of the spectrum of
mood, behavioral and personality disorders based on fear and anger traits: II.
Implications for neurobiology, genetics and psychopharmacological treatment.
J. Affect. Disord. 94, 89103.
Lara, D.R., Ottoni, G.L., Brunstein, M.G., Frozi, J., de Carvalho, H.W., Bisol, L.W., 2012a.
Development and validity data of the Brazilian Internet Study on Temperament
and Psychopathology (BRAINSTEP). J. Affect Disord. 141 (2-3), 390398.
Lara, D.R, Bisol, L.W., Brunstein, M.G., Reppold, C.T., de Carvalho, H.W., Ottoni, G.L.,
2012b. The Affective and Emotional Composite Temperament (AFECT) model
and scale: a system-based integrative approach. J Affect Disord 140 (1), 1437.

Manning, J.S., 2010. Tools to improve differential diagnosis of bipolar disorder in

primary care. Prim. Care Companion J. Clin. Psychiatry 12 (Suppl 1), S17S22.
Marneros, A., 2001. Expanding the group of bipolar disorders. J. Affect. Disord. 62
(12), 3944.
McElroy, S.L., Kotwal, R., Keck Jr, P.E., Akiskal, H.S., 2005. Comorbidity of bipolar
andeating disorders: distinct or related disorders with shared dysregulations?
J. Affect. Disord. 86 (23), 107127.
Merikangas, K.R., Akiskal, H.S., Angst, J., Greenberg, P.E., Hirschfeld, R.M., Petukhova,
M., Kessler, R.C., 2007. Lifetime and 12-month prevalence of bipolar spectrum
disorder in the National Comorbidity Survey replication. Arch. Gen. Psychiatry
64 (5), 543552.
Nowakowska, C., Strong, C.M., Santos, C.M., Wang, P.W., Ketter, T.A., 2005.
Temperamental commonalities and differences in euthymic mood disorder
patients, creativecontrols, and healthycontrols. J. Affect. Disord. 85, 207215.
Pealer, L.N., Weiler, R.M., Pigg Jr., R.M., et al., 2001. The feasibility of a web-based
surveillance system to collect health risk behavior data from college students.
Health Educ. Behav. 28 (5), 547559.
Perugi, G., Toni, C., Maremmani, I., Tusini, G., Ramacciotti, S., Madia, A., Fornaro, M.,
Akiskal, H.S., 2012. The inuence of affective temperaments and psychopathological traits on the denition of bipolar disorder subtypes: a study on bipolar I
Italian national sample. J. Affect. Disord. 136 (12), e41e49.
Rihmer, Z., 1982. Polyglottism and depression. Br. J. Psychiatry 140, 550.
Strakowski, S.M., Fleck, D.E., Maj, M., 2011. Broadening the diagnosis of bipolar
disorder: benets vs. risks. World Psychiatry 10 (3), 181186.
Turner, C.F., Ku, L., Rogers, S.M., Lindberg, L.D., Pleck, J.H., Sonenstein, F.L., 1998.
Adolescent sexual behavior, drug use, and violence: increased reporting with
computer survey technology. Science 280 (5365), 867873.