You are on page 1of 11


ANP0010.1177/0004867416642021ANZJP ArticleEtain et al.


Australian & New Zealand Journal of Psychiatry

Association between childhood 1­–11

DOI: 10.1177/0004867416642021

dimensions of attention deficit © The Royal Australian and

New Zealand College of Psychiatrists 2016

hyperactivity disorder and adulthood Reprints and permissions:
clinical severity of bipolar disorders

Bruno Etain1,2,3,4, M Lajnef2, J Loftus4,5, C Henry1,2,3,4, A Raust3,

S Gard4,6, JP Kahn4,7, M Leboyer1,2,3,4, J Scott8 and F Bellivier4,9,10

Background: Clinical features of attention deficit hyperactivity disorder can be frequently observed in cases with
bipolar disorders and associated with greater severity of bipolar disorders. Although designed as a screening tool for
attention deficit hyperactivity disorder, the Wender Utah Rating Scale could, given its factorial structure, be useful in
investigating the early history of impulsive, inattentive or mood-related symptoms among patients with bipolar disorders.
Methods: We rated the Wender Utah Rating Scale in 276 adult bipolar disorder cases and 228 healthy controls and
tested its factorial structure and any associations with bipolar disorder phenomenology.
Results: We confirmed a three-factor structure for the Wender Utah Rating Scale (‘impulsivity/temper’, ‘inattentiveness’
and ‘mood/self-esteem’). Cases and controls differed significantly on Wender Utah Rating Scale total score and sub-scale
scores (p-values < 10−5). About 23% of bipolar disorder cases versus 5% of controls were classified as ‘WURS positive’
(odds ratio = 5.21 [2.73–9.95]). In bipolar disorders, higher Wender Utah Rating Scale score was associated with earlier
age at onset, severity of suicidal behaviors and polysubstance misuse; multivariate analyses, controlling for age and gen-
der, confirmed the associations with age at onset (p = 0.001) and alcohol and substance misuse (p = 0.001).
Conclusion: Adults with bipolar disorders who reported higher levels of childhood symptoms on the Wender Utah
Rating Scale presented a more severe expression of bipolar disorders in terms of age at onset and comorbidity. The
Wender Utah Rating Scale could be employed to screen for attention deficit hyperactivity disorder but also for ‘at-risk
behaviors’ in adult bipolar disorder cases and possibly for prodromal signs of early onset in high-risk subjects.

Bipolar disorder, attention deficit, hyperactivity, impulsivity, onset, misuse, suicide

1Faculté de Médecine, Université Paris-Est Créteil, Créteil, France

2Inserm U955, Equipe Psychiatrie Translationnelle, Créteil, France
3AP-HP, Hôpitaux Universitaires Henri Mondor, DHU Pepsy, Pôle de Psychiatrie et d’Addictologie, Créteil, France
4Fondation Fondamental, Créteil, France
5Centre Expert Trouble Bipolaire, Hospital Princesse Grace, Monaco
6Service de Psychiatrie Adulte, Hôpital Charles Perrens Bordeaux, Bordeaux, France
7CHU de Nancy—Hôpitaux de Brabois, Service de Psychiatrie et Psychologie Clinique, Vandoeuvre Les Nancy, France
8Institute of Neuroscience, Newcastle University, Newcastle upon Tyne, UK
9AP-HP, GH Saint-Louis–Lariboisière–Fernand-Widal, Pôle Neurosciences, Paris, France
10Université Paris Diderot, UMR-S 1144, Paris, France

Corresponding author:
Bruno Etain, Centre Expert Troubles Bipolaires, Pôle de Psychiatrie, Hôpital Albert Chenevier, 40, rue de Mesly, 94000 Créteil Cedex, France.

Australian & New Zealand Journal of Psychiatry

Downloaded from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
2 ANZJP Articles

Introduction (Duffy et al., 2010). Longitudinal studies are necessary to

help to identify clinical signs that predict transitions to
Bipolar disorder (BD) is highly heterogeneous in terms of nosographic clinical entities. However, such prospective
its clinical expression, course and comorbidities. For most studies are resource-intensive, often suffer from high drop-
patients, onset occurs during adolescence or young adult- out rates (Leopold et al., 2012) and very large samples of
hood (Bellivier et al., 2014; Etain et al., 2012), but some young people are required in order to capture the minority
(non-specific) symptoms are likely to occur very early dur- who will develop BD. Although methodologically weaker,
ing childhood (Duffy, 2010; Duffy and Carlson, 2013; Scott retrospective studies in adults with established BD can help
et al., 2013). Retrospective assessment of childhood symp- point toward some of the sub-syndromal states or precur-
toms and clinical dimensions among adult cases with BD sors that form part of a neurodevelopmental pathway to
may shed light on several issues such as (1) understanding BD. In this context, both prospective and retrospective
links between adult BD syndromes and various childhood- studies can foster the identification of risk constellations
onset psychiatric conditions (typically attention deficit prior to the development of BD that include multiple ele-
hyperactivity disorder [ADHD]), (2) the identification of ments such as genetic risk, substance misuse, ADHD fea-
(non-)specific precursors of BD according to staging mod- tures, sub-threshold affective symptoms, changes in sleep
els and (3) the investigation of associations between these and circadian rhythms and mood swings/affective lability
dimensions and the severity of clinical trajectories of BD in (Leopold et al., 2012). The use of the WURS-derived
adults. The Wender Utah Rating Scale (WURS) has been dimensions in adult BD cases may thus help to clarify
developed to retrospectively assess childhood symptoms of whether the inclusion of ADHD features in this pathway of
ADHD (Ward et al., 1993). Although typically employed as childhood symptoms is clinically relevant and worth inves-
a screening tool, the WURS is not specific for ADHD and tigating prospectively.
its factorial structure captures different childhood dimen- Third, it has been suggested that a history of ADHD fea-
sions (impulsivity/temper, inattentiveness and mood labil- tures in childhood may be associated with a more severe
ity) that transcend diagnostic categories. It is likely that clinical trajectory for BD and a worse prognosis. For exam-
trans-nosographic symptoms and dimensions assessed by ple, the presence of ADHD has been associated with earlier
self-report questionnaires such as the WURS can help onset of BD (Masi et al., 2006; Perlis et al., 2004), and a
detect features of a BD prodrome characterized by exter- recent systematic review has suggested that BD type I and
nalizing behaviors, inattention or emotion dysregulation. comorbid ADHD were significantly associated with suicide
It is noteworthy that BD and ADHD are frequently iden- attempts (SA) in children and adolescents with BD (Hauser
tified as co-occurring and ADHD is the second most com- et al., 2013). The latter finding has been reinforced by a
mon comorbid condition (after anxiety disorders) observed Taiwanese nationwide longitudinal study that reported that
in pediatric BD (Frias et al., 2015). Furthermore, a recent ADHD was an independent risk factor for attempted suicide
Swedish population-based study with the largest BD cohort later in life among adolescents and young adults with BD
ever reported that ADHD was the most common Diagnostic (Lan et al., 2015). Also, compared to patients without a life-
and Statistical Manual of Mental Disorders (DSM) axis I time history of ADHD, BD cases with comorbid ADHD
comorbid condition in adults (Song et al., 2015). It is pro- have higher comorbidity rates for axis I disorders, such as
posed that ADHD can be part of a typical developmental panic disorder and alcohol abuse/dependence (Perugi et al.,
illness trajectory of BD, although this notion is controver- 2013; Tamam et al., 2008; Torres et al., 2015). In a sample
sial (Duffy, 2012; Skirrow et al., 2012) and the association of 90 patients with BD, Karaahmet et al. (2013) demon-
may be an artifact arising from the overlap in clinical symp- strated that ‘WURS-positive’ cases (defined by a total WURS
toms across the syndromes (for review, see Youngstrom score >36, which is the cut-off for the Turkish language
et al., 2010). However, it is also possible that a shared version) had an earlier age at onset and a trend to higher
genetic vulnerability could explain overlapping symptoma- rates of alcohol dependence (Karaahmet et al., 2013). In a
tology and/or the existence of a clinically homogeneous study of 250 adult patients with mood disorders assessed
sub-group of BD with ADHD-like features (Faraone et al., with the WURS, Joo et al. (2012) showed that patients with
2012). The use of the WURS in adult patients with BD can BD II and recurrent depressive disorders scored higher on
provide further insights regarding the frequency of such the WURS total score as compared to controls, with BD I
symptoms during childhood and clarify whether sub- cases displaying intermediate scores (Joo et al., 2012).
dimensions are particularly encountered in children who However, with the exception of one study (Lan et al., 2015),
subsequently develop BD. these results have been obtained in small-scale studies,
Second, staging models of BD (Scott et al., 2013) mixed samples of BD and non-BD mood disorders or stud-
describe a continuum from unaffected, but an at-risk state ies using different assessments of ADHD (presence/absence
through to non-mood disorders or sub-threshold presenta- of ADHD, WURS scores, etc.). The available research indi-
tions (e.g. anxiety, ADHD symptoms) through to minor cates that the influence of ADHD symptoms or sub-syndro-
then major mood disturbances, depression and mania mal ADHD presentations on the clinical expression of adult

Australian & New Zealand Journal ofDownloaded

from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
Etain et al. 3

BD requires further clarification and the use of an estab- et al., 2010). A threshold of 46 is indicative of a possible
lished tool such as the WURS might enhance our under- ADHD in childhood and can be used to classify cases as
standing and allow greater cross-study comparisons. ‘WURS positive’ (Ward et al., 1993).
The aims of this study are (1) to explore the factorial
structure of the 25-item WURS in an independent sample,
Clinical indicators of the course of BD
(2) to compare total WURS and WURS sub-scale scores in
BD cases and healthy controls and (3) to investigate any We selected key clinical variables identified as important
association between ADHD features and markers of the characteristics of BD and indicators of a more severe or
clinical severity of BD (e.g. suicidal behavior) and illness complex course including subtype of BD, mode of onset
complexity (e.g. substance misuse). (age at onset and polarity at onset), lifetime history of SA,
rapid cycling, substance misuse (alcohol, cannabis and
other drugs misuse) or psychotic features (lifetime presence
Methods during at least one mood episode).
The research protocol (INSERM C0829) received approval Age at onset has been defined as the age at which the
by the Ethical Committee and institutional review board; patient fulfilled for the first time the DSM-IV criteria for a
written informed consent was obtained from all study par- depressive, (hypo)manic or mixed episode based on the
ticipants. The consent process involved a detailed descrip- DIGS.
tion of the study by a psychiatrist supplemented by written The DIGS allows the recording of several variables for
information summarizing the protocol and project for the suicidal behaviors (lifetime history of and number of SA
individual. and lifetime occurrence of at least one violent SA), so we
used this information to create a composite variable with a
0–3 rating, with higher scores indicative of more of the
Sample construct under study. The rating is the sum of the follow-
Individuals referred to the clinic services at four university- ing: lifetime presence of SA (rated 0 if absent or 1 if pre-
affiliated psychiatric departments in France (Paris/Créteil, sent), lifetime presence of violent SA (rated 0 if absent or 1
Bordeaux, Nancy) with a presumed diagnosis of BD were if present) and presence of multiple SA (0 if only one SA; 1
interviewed using the French version of the Diagnostic if two or more SA). We used the term violent suicidal
Interview for Genetic Studies (DIGS) (Nurnberger et al., attempt for any attempt characterized by attempted hang-
1994) to assess lifetime Diagnostic and Statistical Manual ing, the use of firearms, jumping from heights, severe deep
of Mental Disorders, 4th edition (DSM-IV) axis I diagno- cuts, car crash, burning, gas poisoning, drowning, electro-
ses (American Psychiatric Association, 1994). The study cution and jumping under a train (Asberg et al., 1976).
sample comprised 276 outpatients who met the following Drug overdoses were considered to be non-violent SA.
inclusion criteria: aged ⩾18 years old; met DSM-IV criteria Individuals with no SA received a score of 0, and those with
for BD I, II or NOS (not otherwise specified); currently lifetime presence of multiple SA including at least one vio-
euthymic (i.e. scoring ⩽5 on both the Montgomery–Asberg lent SA scored 3.
Depression Rating Scale and the Mania Rating Scale) We created a similar composite score for substance mis-
(Bech et al., 1978; Montgomery and Asberg, 1979) and use, categorizing the patients according to no misuse (rated
willing and able to give written informed consent. 0), presence of cannabis misuse (rated 1), presence of alco-
Healthy controls (N = 228) were recruited from adult hol misuse (rated 2) and presence of cannabis and alcohol
blood donors and the general population. Only those who misuse or other drug (cocaine, opiates, amphetamines) mis-
were free of any personal history of DSM-IV axis I psychi- use (rated 3). We used the term misuse for both abuse and/
atric disorders, as assessed with the DIGS (Nurnberger or dependence.
et al., 1994), and had no first-degree relatives with a history
of mood disorders, schizophrenia or SA (as assessed with Statistical analysis
the Family Interview for Genetic Studies [FIGS]) were
included in the study. Data analysis was conducted using R program. To take
multiple testing into account, we used a Bonferroni correc-
tion to determine the threshold for significance, which was
WURS set at p ⩽ 0.003.
The WURS is a 25-item self-administered instrument that
retrospectively measures ADHD symptoms experienced by Factor analysis. We began by conducting an exploratory
individuals before the age of 12 years (Ward et al., 1993). factor analysis using a Principal Component Analysis
Each item is rated on a 4-point scale. All participants com- (PCA) with promax rotation. To measure the appropriate-
pleted the French version of the WURS, the psychometric ness of PCA, we used the Kaiser–Meyer–Olkin index
properties of which have been previously described (Caci (KMO) and Bartlett’s test of sphericity. To define the

Australian & New Zealand Journal of Psychiatry

Downloaded from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
4 ANZJP Articles

number of factors to retain, we used Parallel Analysis (PA) of the variance, and the 5 items included ‘feeling sad or
(PA implies a Monte Carlo simulation process since blue, depressed, unhappy’ or ‘having low opinion of
‘expected’ Eigenvalues are obtained by simulating normal myself’. Four items had loadings <0.40 (items 3, 8, 21,
random samples that parallel the observed data in terms of 22) and were allocated as noted in the methods. Item 13
sample size and number of variables) combined with Cat- (‘moody, ups and downs’) had equal loadings on two fac-
tell’s Scree Test (which identifies the inflexion point on the tors, so it was classified with the ‘impulsivity/temper’
Eigenvalue graph). Factor loadings ⩾0.40 were considered factor as this was where it loaded in a previous study
as significant and thus were retained. As we consider inde- (Caci et al., 2010).
pendent factors, when an item has a factor loading ⩾0.40
for several factors, it was considered as unclassified. For Case–control comparisons.  As shown in Table 3, cases and
analyses that required the inclusion of all WURS items, we controls were significantly different for all WURS scores.
classified items with loadings <0.4 with the factor they Furthermore, 22.5% of BD cases versus 5.3% of controls
showed the highest loading for. were classified as WURS positive (OR  = 
5.21; 95%
CI = [2.73, 9.95]; p < 0.0001).
Case–control comparisons and analysis of WURS in the BD Linear regression showed that the differences between
group.  Between-group comparisons were performed using cases and controls remained significant when controlling
chi-squared tests for categorical variables and t-tests or for age and gender (see Table 4).
non-parametric tests for continuous variables as appropri-
ate. Odds ratios (OR) and 95% confidence intervals (95%
Associations between WURS scores and clinical characteristics
CI) are given for selected categorical comparisons.
of BD.  An earlier age at onset of BD was significantly cor-
We used linear regressions to take into account the
related with a higher WURS total score (r = −0.23; p < 10−4).
effects of age at interview and gender as appropriate, e.g.,
A ‘positive WURS’ was associated with early onset (onset
when exploring associations between suicidality ratings
before 22 years old) (32% vs 15% for later onset, OR = 2.70,
and WURS scores (as this behavior can vary according to
95% CI = [1.49, 4.89], p = 0.001).
age and gender). BoxCox power transformations were used
A ‘positive WURS’ was also associated with alcohol
for all scores to fulfill the normality assumption required
misuse (38% vs 20% for no alcohol misuse, OR = 2.46,
for parametric procedures. When an association was identi-
95% CI = [1.22, 4.98], p = 0.01) and cannabis misuse (44%
fied for the WURS total score, this was further investigated
vs 20% for no cannabis misuse, OR = 3.01, 95% CI = [1.29,
for each sub-scale (representing the factors derived from
7.05], p = 0.01).
the PCA).
An association was found between WURS total
score and the composite scores for increasing severity
Results of substance use (Figure 1) and for suicidal behaviors
(Figure 2).
Sample.  The BD group consisted of 276 patients with BD
No differences for WURS total score were observed for
(62% female) and 228 healthy controls (51% female).
cases classified according to BD subtype (p = 0.65), polar-
Cases and controls showed significantly different gender
ity at onset (p = 0.46), lifetime history of psychotic symp-
distributions (p = 0.02), but mean age at interview was not
toms during mood episodes (p = 0.77) or rapid cycling
significantly different (44.13  ± 12.2 vs 41.83  ± 12;
(p = 0.99).
p = 0.08). The clinical characteristics of BD cases are shown
in Table 1.

Factor analysis.  We established that PCA was an appropri-

Multivariate analysis
ate procedure (KMO equal to 0.93 and Bartlett’s test As shown in Table 5, linear regression analysis (controlling
p < 10−6). Using both the PA and Cattell’s Scree Test, three for potential confounders) confirmed significant associa-
factors were extracted, accounting for 52% of the variance. tions between WURS total score, earlier age at onset of BD
Items and PCA factor loadings in the whole sample are pre- and higher severity of alcohol and substance misuse, but
sented in Table 2. did not find a significant association with severity of sui-
Factor 1 was labeled ‘impulsivity/temper’ and cidal behaviors.
explained 37% of the variance. It included 11 items such Scores for the three factors (derived from the PCA)
as ‘having temper outburst, tantrums’ or ‘being hot tem- identified significant associations between age at onset,
pered, low boiling point’. Factor 2 was labeled ‘inatten- severity of misuse and the Impulsivity/Temper,
tiveness’ and explained 8% of the variance. It contained Inattentiveness and Mood/Self-esteem sub-scales (all
9 items such as ‘having concentration problems, easily p-values ⩽ 0.003), but again there was no association
distracted’ or ‘did not achieve up to potential’. Factor 3 between any of these factors and the composite measure of
was labeled ‘mood/self-esteem’ and explained about 7% suicidal behaviors (data not shown).

Australian & New Zealand Journal ofDownloaded

from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
Etain et al. 5

Table 1.  Clinical characteristics of cases with bipolar disorder (BD).

Variables N included in analysis Cases with BD

BD I 276 71%

BD II 23.2%

BD NOS 5.8%

Mean age at onset (SD) 265 25.41 (9.6)

Mean duration of the illness in years (SD) 261 18.83 (12.2)

Depressive polarity at onset 260 62%

Mean number of major mood episodes (SD) 224 6.07 (4.4)

⩾1 psychotic mood episode 263 56%

Rapid cycling 243 23%

⩾1 mixed episode 234 24%

⩾1 suicide attempt 262 42%

Mean number of suicide attempts (SD) 110 2.35 (2.0)

Mean age at first suicide attempt (SD) 108 30.04 (11.7)

⩾1 violent suicide attempt 110 26%

Tobacco smoking status 263  

 Never-smoker 37%
  Former smoker 22%
  Current smoker 41%

Alcohol misuse 262 16%

Cannabis misuse 262 10%

Other substances misuse 260 4%

NOS: not otherwise specified; SD: standard deviation.

Percentages are reported as the nearest whole number.

Discussion it is worthwhile highlighting some of the issues that com-

plicate the measurement of ADHD symptoms in BD (and
We confirmed the factorial structure of the French version vice versa).
of the 25-item WURS in a sample of euthymic BD cases First, it is notable that the design of our study does not
and healthy controls. The three factors identified allow us to determine the sensitivity or specificity of the
(Impulsivity/Temper, Inattentiveness and Mood/Self- WURS as a tool to detect ADHD caseness in our BD sam-
esteem) replicated the structure previously described in a ple or the healthy controls. We did not include a structured
French validation study (Caci et al., 2010) and that reported interview to ascertain a formal diagnosis of ADHD. This
in studies of patients with other mood disorders (Greenwood limitation arose mainly because the version of the DIGS
et al., 2013; Joo et al., 2012). Adult BD cases differed sig- that is validated for use in French does not include a section
nificantly from healthy controls for the WURS total score of the diagnosis of ADHD (unlike later versions of the
and the three factor scores. A greater magnitude of child- DIGS). We note that about 20% of WURS scores exceeded
hood ADHD features was associated with more evidence of the arbitrary cut-off score (total >46) used to identify prob-
clinical severity and complexity of BD including an earlier able caseness, so some or all of these individuals may have
age at onset and comorbid alcohol and substance misuse true ADHD comorbidity. Having said that, screening tools
(mainly cannabis), but only a trend for a greater array of such as the WURS can generate a substantial rate of false
suicidal behaviors. In addition to discussing these findings, positives in clinical samples (as high as

Australian & New Zealand Journal of Psychiatry

Downloaded from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
6 ANZJP Articles

Table 2.  Principal Component Analysis in the sample of cases and controls.

Item number Factor 1 Factor 2 Factor 3

WURS 61 WURS 25 Impulsivity/temper Inattentiveness Mood/self-esteem
3 1 0.62  
4 2 −0.66
5 3 0.36 0.37  
6 4 0.49  
7 5 0.82  
9 6 0.88  
10 7 0.48  
11 8 0.39  
12 9 −0.79
15 10 0.52  
16 11 −0.73
17 12 0.71  
20 13 0.41 −0.41
21 14 0.84  
24 15 0.43  
25 16 0.49  
26 17 −0.63
27 18 0.44  
28 19 0.46  
29 20 −0.41
40 21 0.23 0.28  
41 22 0.34 0.29  
51 23 0.47  
56 24 0.41  
59 25 0.54  

WURS: Wender Utah Rating Scale.

Darker shades of gray indicate loadings >0.40.
Slighter shades of gray indicate loadings <0.40.
Items in italic are those with no clear factor loading.

Table 3. WURS scores in cases with bipolar disorder (BD) and healthy controls.

WURS scores BD cases (N = 276) Healthy controls (N = 228) p-value

Total score 31.32 (19.0) 16.12 (14.2) <10−6

Impulsivity/temper 12.64 (10.1)   6.22 (6.8) <10−6

Inattentiveness 11.31 (7.3)   7.01 (6.5) <10−6

Mood/self-esteem   7.37 (5.3)   2.89 (3.4) <10−6

WURS: Wender Utah Rating Scale.

Australian & New Zealand Journal ofDownloaded

from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
Etain et al. 7

Table 4. WURS scores in cases with bipolar disorder (BD) and healthy controls adjusted for gender and age at interview.

WURS scores Disease status (case vs control) Gender (female vs male) Age at interview

  Beta (SD) p-value Beta (SD) p-value Beta (SD) p-value

Total score 2.36 (0.3) <10−5 −0.34 (0.27) 0.21 −0.01 (0.01) 0.21

Impulsivity/temper 1.47 (0.23) <10−5 −0.32 (0.23) 0.16 −0.02 (0.01) 0.07

Inattentiveness 1.91 (0.22) <10−5 0.17 (0.21) 0.42 −0.01 (0.01) 0.19

Mood/self-esteem 1.84 (0.22) <10−5 0.19 (0.12) 0.38 −0.01 (0.01) 0.24

WURS: Wender Utah Rating Scale; SD: standard deviation.

Positive values of beta indicated association between the scores and BD. p- values in bold remained significant after correction for multiple testing.

Figure 1. WURS total score as a function of substance misuses.

WURS: Wender Utah Rating Scale.

p(Kruskall Wallis test) = 0.005.
After multiple corrections test, no misuse versus cannabis, alcohol and others remained significant (p = 0.0003).

75% in some studies) (Suhr et al., 2009), and the positive actually measuring clinical manifestations of externalizing
predictive values of the WURS are only around 45% among behaviors or other symptoms that are far from specific to
alcohol dependent patients (Daigre et al., 2015), around ADHD (Suhr et al., 2009).
35% in individuals seeking treatment for substance use dis- An alternative view of our approach is that we have used
orders (Dakwar et al., 2012) and about the same in patients the WURS as a multi-dimensional measure to capture
with BD (Oncu et al., 2005). Screening for adult ADHD in aspects of the complex clinical phenomenology and over-
BD samples gives rise to similar issues, and about 40% of lapping features of BD and ADHD. This overlap typically
the individuals who score positively on the adult self-rating includes impulsivity and inattention and also emotional
scale (ASRS-v1.1) do not meet ADHD caseness criteria lability, and all these dimensions are represented in the fac-
(Edebol et al., 2012; McCann and Roy-Byrne, 2004; torial structure of the WURS scale. It has been highlighted
Perroud et al., 2014; Torres et al., 2015). However, we were that ADHD manifestations in childhood also include an
primarily interested in considering dimensions of ADHD emotional component (mood swings and emotional labil-
and potential symptom clusters, not simply studying pres- ity), and Shaw et al. (2014) report that emotion dysregula-
entations that exceeded diagnostic thresholds. To this end, tion is found in up to 50% of children with ADHD. The
several authors have thus proposed that the WURS rather is clinical overlap between ADHD and bipolar mania includes

Australian & New Zealand Journal of Psychiatry

Downloaded from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
8 ANZJP Articles

Figure 2. WURS total score as a function of severity of suicidal behaviors.

WURS: Wender Utah Rating Scale.

p(Kruskall Wallis test) = 0.02.
After correction for multiple test, 1 versus 2 remained significant (p = 0.0003).

Table 5.  Multivariate analysis using WURS total score as the irritable mood, poor concentration, increased motor activ-
dependent variables and age at onset, severity of misuse and of ity and features that can also be observed in certain sub-
suicide behavior as independent variables. types of bipolar depression. Indeed, the clinical
manifestations assessed in the WURS belong to different
Variables Beta (SD) p-value
psychological domains such as impulsivity, temper, inat-
Intercept 11.3 (0.8) 0.001 tention, unstable mood, low self-esteem and anxiety. As
such, the conclusion in favor of any one position over
Severity Misuse 1a 0.61 (1.33) 0.65
another (i.e. the WURS demonstrates the co-occurrence of
Severity Misuse 2a 1.63 (0.83) 0.06 BD and ADHD phenomena and represents true comorbid-
ity or it highlights early manifestations of BD or indicates
Severity Misuse 3a 3.93 (1.09) 0.001
non-specific psychopathological manifestations of several
Severity Suicide 1b −0.53 (0.63) 0.41 putative disease trajectories) is far from easy. Our own
review of the literature indicated that more than half of
Severity Suicide 2b 1.03 (0.73) 0.16
individuals who developed BD presented a putative pro-
Severity Suicide 3b 1.37 (1.13) 0.23 drome prior to 14 years of age and that this frequently
included a mixture of precursors such as mood lability,
Age At Onset (AAO) −0.09 (0.03) 0.001 depressive episodes, anxiety, sleep disorders, conduct prob-
WURS: Wender Utah Rating Scale; SD: standard deviation. lems, attention and concentration impairments, altered
Severity Suicide 0: no suicide attempt; Severity Suicide 1: unique non- energy patterns and/or a family history of mania or depres-
violent suicide attempt; Severity Suicide 2: multiple or violent suicide sion (Geoffroy et al., 2015; Leopold et al., 2012).
attempts; Severity Suicide 3: multiple suicide and violent attempts, The validation study of the WURS (Ward et al., 1993)
– Severity Misuse 0: no lifetime substance misuse (excluding tobacco);
Severity Misuse 1: alcohol misuse alone; Severity Misuse 2: cannabis mis-
selected the 25 items that showed the greatest mean differ-
use alone; Severity Misuse 3: alcohol and cannabis misuse or any other ence between subjects with ADHD and healthy controls.
substances misuse (cocaine, opiates, amphetamines). However, in addition to externalizing behaviors, the items
aScore for the construct compared to the reference level (absence of
included (anxiety, worry, inattention, daydreaming, sleep
lifetime misuse).
bScore for the construct compared to the reference level (absence of
problems, stubborn behavior, poor concentration) bear a
suicidal behavior). striking resemblance to the prodromal signs identified in
longitudinal studies of high-risk offspring of BD parents

Australian & New Zealand Journal ofDownloaded

from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
Etain et al. 9

(Duffy, 2010; Duffy et al., 2010; Egeland et al., 2012) and 2013) and to dimensions of hostility/impulsivity to suicidal
other risk syndromes (Scott et al., 2013). A review of pro- behaviors (Parmentier et al., 2012). Two meta-analyses have
spective studies exploring clinical risk factors for BD found demonstrated that children with ADHD were more likely to
evidence for at least two heterotypic trajectories, one char- develop disorders of abuse/dependence for nicotine, alco-
acterized by prodromal anxiety symptoms and the other by hol, marijuana, cocaine and other substances (Charach et al.,
high impulsivity, aggression, externalizing behaviors and 2011; Lee et al., 2011). The use of the WURS might shed
mixed dysphoric features (Faedda et al., 2014). As such, light on how trans-diagnostic symptoms may accelerate or
high WURS scores may simply reflect heterotypia in the exaggerate disorder-related phenomena such as age at onset
developmental trajectory of BD and such findings should or patterns of risky or self-harming behaviors. The clinical
encourage clinicians to explore not only problems related implication is to systematically screen patients with BD for
to ‘early mood deregulation’ but also a larger range of this early component to identify those who might be at risk
(non-)specific precursors that may be markers of future for such high-risk behaviors.
morbidity independent of the specific late-stage syndrome This study has several methodological limitations. As
(Hickie et al., 2013). noted, using the WURS to retrospectively measure child-
The utility of the research assessment we have under- hood ADHD features is the main weakness, and recall biases
taken is that it demonstrates an association between high and cognitive deficits may further confound the assessment.
scores on the WURS and specific aspects of a more severe, Responses could also be biased by residual mood symptoms
complex, BD phenotype characterized by an earlier age at although our sample has been assessed during a period of
onset and an increased lifetime severity of suicidal attempts euthymia. We used no external validation of the self-assess-
and substance misuse. The association between higher ment made by individuals. We used the WURS as an indica-
WURS and earlier onset of BD could represent higher levels tor of ADHD during childhood but provided no formal
of true ADHD comorbidity in the early-onset BD sub-group diagnosis for childhood or adulthood persistent ADHD.
(since 32% of those with early onset vs 15% of those with Therefore, our study was not an attempt to assess a retro-
later onset have a positive WURS), and there is corroborat- spective diagnosis of ADHD. Finally, the cut-off of 46 that
ing evidence for such a finding in the literature. For exam- we used to define WURS-positive cases is not universally
ple, Greenwood et al. (2013) reported a negative correlation accepted as the optimal cut-off score (Suhr et al., 2009). We
between age of onset and all WURS scores except for inat- also did not provide any adult ADHD assessment to study
tention. The bidirectional and robust comorbidity between the continuity between ADHD symptoms from childhood to
pediatric BD type I and ADHD has been well documented adulthood. Although this is one of the larger studies on this
in pediatric and adult studies in both clinical and epidemio- topic, false negative findings could have emerged such as
logical samples (Biederman et al., 2013). In the Systematic the reduced magnitude in association with suicidal behav-
Treatment Enhancement Program for Bipolar Disorder iors (after controlling for other variables).
(STEP-BD) cohort of adults, the overall lifetime prevalence
of ADHD was about 10% and patients with BD and comor-
bid ADHD reported the onset of their mood disorder about
5 years earlier than other cases (Nierenberg et al., 2005). We found that ADHD features were common in adults with
Also, the bidirectional association between pediatric BD BD, with 22% of cases versus 5% of controls having a posi-
and ADHD has been documented in clinical and epidemio- tive WURS score. The presence of ADHD features in child-
logical samples (Biederman et al., 2013) and ADHD is hood was associated with early onset, suicidal behavior and
reported to be a significant risk factor for switching from substance misuse. Both the total WURS score and scores
unipolar to BD in youth (Biederman et al., 2009; Chen et al., on the three factor sub-scales (Impulsivity/Temper,
2015). However, the evidence does not suggest the associa- Inattentiveness and Mood/Self-esteem) were linked with
tion between the WURS ratings, and an earlier age is due to early onset and substance misuse. Further clinical studies
misdiagnosis of ADHD and juvenile mania in our sample. It of BD cases and individuals at high risk of developing BD
is notable that the earliest age at onset in our sample is will enable better understanding of how ADHD features fit
14 years old, corresponding to the often-noted lack of pre- into a neurodevelopmental model of BD and whether the
pubertal mania in European clinical populations (Douglas features predict a specific illness trajectory or represent an
and Scott, 2014; James et al., 2014). Furthermore, it is epiphenomena (such as overlapping clinical features).
unlikely that mood swings or emotion lability dramatically Additional research would also be useful to delineate how
accounts for the differences observed between groups in our the WURS can be integrated into assessment packages of
study (as this factor accounts for only 7% variance in the ‘risk constellations’ for adults with complex BD presenta-
WURS total score). The largest proportion of the variance tions (Leopold et al., 2012). The systematic investigation of
was explained by the impulsivity/temper factor. ADHD features (and a possible diagnosis of childhood
We have previously reported evidence linking impulsiv- ADHD) especially in those with an early-onset BD and evi-
ity to alcohol and substance misuse in BD (Etain et al., dence of substance misuse or suicidal behaviors could help

Australian & New Zealand Journal of Psychiatry

Downloaded from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
10 ANZJP Articles

treatment planning and modify outcomes. Indeed, some Dakwar E, Mahony A, Pavlicova M, et al. (2012) The utility of attention-
recommendations have been proposed for the management deficit/hyperactivity disorder screening instruments in individuals
seeking treatment for substance use disorders. Journal of Clinical
of patients with mood disorders and comorbid ADHD in Psychiatry 73: e1372–e1378.
national clinical guidelines (Bond et al., 2012). Douglas J and Scott J (2014) A systematic review of gender-specific rates
of unipolar and bipolar disorders in community studies of pre-puber-
tal children. Bipolar Disorders 16: 5–15.
Duffy A (2010) The early natural history of bipolar disorder: What we
We thank the patients and controls for participating in this study. have learned from longitudinal high-risk research. Canadian Journal
We thank E Abadie, and JR Richard for their assistance. of Psychiatry 55: 477–485.
Duffy A (2012) The nature of the association between childhood ADHD
and the development of bipolar disorder: A review of prospec-
Declaration of Conflicting Interests tive high-risk studies. The American Journal of Psychiatry 169:
The author(s) declared no potential conflicts of interest with respect 1247–1255.
to the research, authorship and/or publication of this article. Duffy A, Alda M, Hajek T, et al. (2010) Early stages in the development of
bipolar disorder. Journal of Affective Disorders 121: 127–135.
Duffy A and Carlson GA (2013) How does a developmental perspec-
Funding tive inform us about the early Natural History of Bipolar Disorder?
Journal of Canadian Academy of Child and Adolescent Psychiatry
The author(s) disclosed receipt of the following financial support
22: 6–12.
for the research, authorship, and/or publication of this article: This
Edebol H, Helldin L and Norlander T (2012) Objective measures of
work was supported by Institut National de la Santé et de la behavior manifestations in adult ADHD and differentiation from par-
Recherche Médicale (INSERM), Assistance Publique des ticipants with bipolar II disorder, borderline personality disorder, par-
Hôpitaux de Paris and Fondation FondaMental. ticipants with disconfirmed ADHD as well as normative participants.
Clinical Practice and Epidemiology in Mental Health 8: 134–143.
Egeland JA, Endicott J, Hostetter AM, et al. (2012) A 16-year prospective
study of prodromal features prior to BPI onset in well Amish children.
American Psychiatric Association (APA) (1994) Diagnostic and Statistical Journal of Affective Disorders 142: 186–192.
Manual of Mental Disorders, 4th edn. Washington, DC: APA. Etain B, Lajnef M, Bellivier F, et al. (2012) Clinical expression of bipolar
Asberg M, Traskman L and Thoren P (1976) 5-HIAA in the cerebro- disorder type I as a function of age and polarity at onset: Convergent
spinal fluid. A biochemical suicide predictor? Archives of General findings in samples from France and the United States. Journal of
Psychiatry 33: 1193–1197. Clinical Psychiatry 73: e561–e566.
Bech P, Rafaelsen OJ, Kramp P, et al. (1978) The mania rating scale: Scale Etain B, Mathieu F, Liquet S, et al. (2013) Clinical features associated with
construction and inter-observer agreement. Neuropharmacology 17: trait-impulsiveness in euthymic bipolar disorder patients. Journal of
430–431. Affective Disorders 144: 240–247.
Bellivier F, Etain B, Malafosse A, et al. (2014) Age at onset in bipolar Faedda GL, Serra G, Marangoni C, et al. (2014) Clinical risk factors for
I affective disorder in the USA and Europe. The World Journal of bipolar disorders: A systematic review of prospective studies. Journal
Biological Psychiatry 15: 369–376. of Affective Disorders 168: 314–321.
Biederman J, Faraone SV, Petty C, et al. (2013) Further evidence that Faraone SV, Biederman J and Wozniak J (2012) Examining the comor-
pediatric-onset bipolar disorder comorbid with ADHD represents a bidity between attention deficit hyperactivity disorder and bipolar I
distinct subtype: Results from a large controlled family study. Journal disorder: A meta-analysis of family genetic studies. The American
of Psychiatric Research 47: 15–22. Journal of Psychiatry 169: 1256–1266.
Biederman J, Petty CR, Byrne D, et al. (2009) Risk for switch from unipo- Frias A, Palma C and Farriols N (2015) Comorbidity in pediatric bipolar
lar to bipolar disorder in youth with ADHD: A long term prospective disorder: Prevalence, clinical impact, etiology and treatment. Journal
controlled study. Journal of Affective Disorders 119: 16–21. of Affective Disorders 174: 378–389.
Bond DJ, Hadjipavlou G, Lam RW, et al. (2012) The Canadian Network Geoffroy PA, Leboyer M and Scott J (2015) Predicting bipolar disorder:
for Mood and Anxiety Treatments (CANMAT) task force recom- What can we learn from prospective cohort studies? L’Encéphale 41:
mendations for the management of patients with mood disorders and 10–16.
comorbid attention-deficit/hyperactivity disorder. Annals of Clinical Greenwood TA, Joo EJ, Shekhtman T, et al. (2013) Association of dopa-
Psychiatry 24: 23–37. mine transporter gene variants with childhood ADHD features in
Caci HM, Bouchez J and Bayle FJ (2010) An aid for diagnosing attention- bipolar disorder. American Journal of Medical Genetics. Part B,
deficit/hyperactivity disorder at adulthood: Psychometric properties Neuropsychiatric Genetics 162B: 137–145.
of the French versions of two Wender Utah Rating Scales (WURS-25 Hauser M, Galling B and Correll CU (2013) Suicidal ideation and suicide
and WURS-K). Comprehensive Psychiatry 51: 325–331. attempts in children and adolescents with bipolar disorder: A system-
Charach A, Yeung E, Climans T, et al. (2011) Childhood attention- atic review of prevalence and incidence rates, correlates, and targeted
deficit/hyperactivity disorder and future substance use disorders: interventions. Bipolar Disorders 15: 507–523.
Comparative meta-analyses. Journal of the American Academy of Hickie IB, Scott J, Hermens DF, et al. (2013) Clinical classification
Child and Adolescent Psychiatry 50: 9–21. in mental health at the cross-roads: Which direction next? BMC
Chen MH, Chen YS, Hsu JW, et al. (2015) Comorbidity of ADHD and Medicine 11: 125.
subsequent bipolar disorder among adolescents and young adults with James A, Hoang U, Seagroatt V, et al. (2014) A comparison of American
major depression: A nationwide longitudinal study. Bipolar Disorders and English hospital discharge rates for pediatric bipolar disor-
17: 315–322. der, 2000 to 2010. Journal of the American Academy of Child and
Daigre C, Roncero C, Rodriguez-Cintas L, et al. (2015) Adult ADHD Adolescent Psychiatry 53: 614–624.
screening in alcohol-dependent patients using the Wender-Utah Joo EJ, Lee KY, Choi KS, et al. (2012) Childhood attention deficit hyper-
Rating Scale and the adult ADHD Self-Report Scale. Journal of activity disorder features in adult mood disorders. Comprehensive
Attention Disorders 19: 328–334. Psychiatry 53: 217–223.

Australian & New Zealand Journal ofDownloaded

from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016
Etain et al. 11

Karaahmet E, Konuk N, Dalkilic A, et al. (2013) The comorbidity of adult participants in the systematic treatment enhancement program for
attention-deficit/hyperactivity disorder in bipolar disorder patients. bipolar disorder (STEP-BD). Biological Psychiatry 55: 875–881.
Comprehensive Psychiatry 54: 549–555. Perroud N, Cordera P, Zimmermann J, et al. (2014) Comorbidity between
Lan WH, Bai YM, Hsu JW, et al. (2015) Comorbidity of ADHD and sui- attention deficit hyperactivity disorder (ADHD) and bipolar disorder
cide attempts among adolescents and young adults with bipolar disor- in a specialized mood disorders outpatient clinic. Journal of Affective
der: A nationwide longitudinal study. Journal of Affective Disorders Disorders 168: 161–166.
176, 171–175. Perugi G, Ceraudo G, Vannucchi G, et al. (2013) Attention deficit/hyper-
Lee SS, Humphreys KL, Flory K, et al. (2011) Prospective association of activity disorder symptoms in Italian bipolar adult patients: A prelimi-
childhood attention-deficit/hyperactivity disorder (ADHD) and sub- nary report. Journal of Affective Disorders 149, 430–434.
stance use and abuse/dependence: A meta-analytic review. Clinical Scott J, Leboyer M, Hickie I, et al. (2013) Clinical staging in psychiatry:
Psychology Review 31, 328–341. A cross-cutting model of diagnosis with heuristic and practical value.
Leopold K, Ritter P, Correll CU, et al. (2012) Risk constellations prior to British Journal of Psychiatry 202, 243–245.
the development of bipolar disorders: Rationale of a new risk assess- Shaw P, Stringaris A, Nigg J, et al. (2014) Emotion dysregulation in
ment tool. Journal of Affective Disorders 136, 1000–1010. attention deficit hyperactivity disorder. The American Journal of
McCann BS and Roy-Byrne P (2004) Screening and diagnostic utility of Psychiatry 171: 276–293.
self-report attention deficit hyperactivity disorder scales in adults. Skirrow C, Hosang GM, Farmer AE, et al. (2012) An update on the
Comprehensive Psychiatry 45: 175–183. debated association between ADHD and bipolar disorder across the
Masi G, Perugi G, Millepiedi S, et al. (2006) Developmental differences lifespan. Journal of Affective Disorders 141: 143–159.
according to age at onset in juvenile bipolar disorder. Journal of Child Song J, Bergen SE, Kuja-Halkola R, et al. (2015) Bipolar disorder and its
and Adolescent Psychopharmacology 16: 679–685. relation to major psychiatric disorders: A family-based study in the
Montgomery SA and Asberg M (1979) A new depression scale designed Swedish population. Bipolar Disorders 17: 184–193.
to be sensitive to change. British Journal of Psychiatry 134: 382–389. Suhr J, Zimak E, Buelow M, et al. (2009) Self-reported childhood atten-
Nierenberg AA, Miyahara S, Spencer T, et al.; STEP-BD Investigators tion-deficit/hyperactivity disorder symptoms are not specific to the
(2005) Clinical and diagnostic implications of lifetime attention-defi- disorder. Comprehensive Psychiatry 50: 269–275.
cit/hyperactivity disorder comorbidity in adults with bipolar disorder: Tamam L, Karakus G and Ozpoyraz N (2008) Comorbidity of adult atten-
Data from the first 1000 STEP-BD participants. Biological Psychiatry tion-deficit hyperactivity disorder and bipolar disorder: Prevalence
57: 1467–1473. and clinical correlates. European Archives of Psychiatry and Clinical
Nurnberger JI, Jr, Blehar MC, Kaufmann CA, et al. (1994) Diagnostic Neuroscience 258: 385–393.
interview for genetic studies. Rationale, unique features, and train- Torres I, Gomez N, Colom F, et al. (2015) Bipolar disorder with comorbid
ing. NIMH Genetics Initiative. Archives of General Psychiatry 51: attention-deficit and hyperactivity disorder. Main clinical features and
849–859; discussion 863–864. clues for an accurate diagnosis. Acta Psychiatrica Scandinavica 132:
Oncu B, Olmez S and Senturk V (2005) Validity and reliability of the 389–399.
Turkish version of the Wender Utah Rating Scale for attention-deficit/ Ward MF, Wender PH and Reimherr FW (1993) The Wender Utah Rating
hyperactivity disorder in adults. Türk Psikiyatri Dergisi 16: 252–259. Scale: An aid in the retrospective diagnosis of childhood attention
Parmentier C, Etain B, Yon L, et al. (2012) Clinical and dimensional deficit hyperactivity disorder. The American Journal of Psychiatry
characteristics of euthymic bipolar patients with or without suicidal 150: 885–890.
behavior. European Psychiatry 27: 570–576. Youngstrom EA, Arnold LE and Frazier TW (2010) Bipolar and ADHD
Perlis RH, Miyahara S, Marangell LB, et al. (2004) Long-term impli- comorbidity: Both artifact and outgrowth of shared mechanisms.
cations of early onset in bipolar disorder: Data from the first 1000 Clinical Psychology 17: 350–359.

Australian & New Zealand Journal of Psychiatry

Downloaded from at HOWARD UNIV UNDERGRAD LIBRARY on April 18, 2016