a meta-analysis of neuropsychological functioning in first-episode bipolar disorders

lable at ScienceDirect

Journal of Psychiatric Research xxx (2014) 1e11

Contents lists avai

Journal of Psychiatric Research

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

Review

A meta-analysis of neuropsychological functioning in first-episodebipolar disorders

Rico S.C. Lee a, b, *, Daniel F. Hermens a, Jan Scott c, d, M. Antoinette Redoblado-Hodge e,Sharon L. Naismith a, Jim Lagopoulos a, Kristi R. Griffiths a, Melanie A. Porter b,Ian B. Hickie a

a Clinical Research Unit, Brain and Mind Research Institute, University of Sydney, Sydney, Australiab Department of Psychology, Macquarie University, Sydney, Australiac Academic Psychiatry, Institute of Neuroscience, Newcastle University, UKd Centre for Affective Disorders, Institute of Psychiatry, London, UKe Child Development Unit, The Children's Hospital at Westmead, Sydney, Australia

a r t i c l e i n f o

Article history:Received 10 February 2014Received in revised form25 May 2014Accepted 23 June 2014

Keywords:NeuropsychologyCognitive deficitsBipolar disorderFirst-episodeMeta-analysisEarly intervention

* Corresponding author. 100 Mallett Street, BrainCamperdown, NSW, 2050, Australia. Tel.: þ61 2 9351

E-mail address: [email protected] (R.S.C. Lee)

http://dx.doi.org/10.1016/j.jpsychires.2014.06.0190022-3956/© 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Lee RSC, etPsychiatric Research (2014), http://dx.doi.or

a b s t r a c t

Broad neuropsychological deficits have been consistently demonstrated in well-established bipolardisorder. The aim of the current study was to systematically review neuropsychological studies in first-episode bipolar disorders to determine the breadth, extent and predictors of cognitive dysfunction at thisearly stage of illness through meta-analytic procedures. Electronic databases were searched for studiespublished between January 1980 and December 2013. Twelve studies met eligibility criteria (N ¼ 341,mean age ¼ 28.2 years), and pooled effect sizes (ES) were calculated across eight cognitive domains.Moderator analyses were conducted to identify predictors of between-study heterogeneity. Controllingfor known confounds, medium to large deficits (ES � 0.5) in psychomotor speed, attention and workingmemory, and cognitive flexibility were identified, whereas smaller deficits (ES 0.20e0.49) were found inthe domains of verbal learning and memory, attentional switching, and verbal fluency. A medium tolarge deficit in response inhibition was only detected in non-euthymic cases. Visual learning andmemory functioning was not significantly worse in cases compared with controls. Overall, first-episodebipolar disorders are associated with widespread cognitive dysfunction. Since euthymia was not asso-ciated with superior cognitive performance in most domains, these results indicate that even in theearliest stages of disease, cognitive deficits are not mood-state dependent. The current findings haveimportant implications for whether cognitive impairments represent neurodevelopmental or neurode-generative processes. Future studies need to more clearly characterise the presence of psychotic features,and the nature and number of previous mood episodes.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Numerous studies on bipolar disorders show that neuropsy-chological deficits are detectable in euthymia and contribute topoor outcome and impaired functioning independently of othersymptoms and factors (Bourne et al., 2013; Depp et al., 2012;Murphy and Sahakian, 2001; Olley et al., 2005; Robinson andFerrier, 2006). A spate of systematic reviews and meta-analysesin bipolar disorder (BD) indicate that these cognitive impairments

and Mind Research Institute,0763..

al., A meta-analysis of neuropg/10.1016/j.jpsychires.2014.06

are similar in range to those seen in schizophrenia, but that they areusually of lessermagnitude (Arts et al., 2008; Bora et al., 2009, 2010,2011; Bourne et al., 2013; Kurtz and Gerraty, 2009; Mann-Wrobelet al., 2011; Murphy and Sahakian, 2001; Robinson and Ferrier,2006; Torres et al., 2007). There are also suggestions that neuro-psychological impairments may represent a putative ‘endopheno-type’ for BD, or intermediate phenotype between genotype and theclinical syndrome (Daban et al., 2012; Glahn et al., 2010, 2006;Robbins et al., 2012). A recent review of cognitive dysfunction inBD and schizophrenia concluded that there were differences in thedeficits observed prior to illness onset (partly associated with dif-ferences in pre-morbid intellectual functioning; IQ), but that post-illness onset, BD and schizophrenia were associated with commondeficits across diagnostic boundaries (Lewandowski et al., 2011).

sychological functioning in first-episode bipolar disorders, Journal of.019

Delta:1_given name

Delta:1_surname

Delta:1_given name

Delta:1_surname

Delta:1_given name

Delta:1_surname

Delta:1_given name

Delta:1_surname

Delta:1_given name

mailto:[email protected]

www.sciencedirect.com/science/journal/00223956

http://www.elsevier.com/locate/psychires

http://dx.doi.org/10.1016/j.jpsychires.2014.06.019



R.S.C. Lee et al. / Journal of Psychiatric Research xxx (2014) 1e112

Currently, it remains unclear whether the neuropsychologicaldeficits in BD can be viewed as supportive of underlying neuro-developmental or neurodegenerative processes (Goodwin et al.,2008).

One approach to further clarify the time-course and aetiology ofcognitive dysfunction in BD is to examine neuropsychological and/or neuroimaging studies in first-episode cases. Despite a sizablebody of neuroimaging evidence demonstrating structural brainchanges in first-episode bipolar disorders (Vita et al., 2009), andsystematic reviews of neuropsychological functioning in first-episode major depressive disorder (Lee et al., 2012), andschizophrenia-spectrum psychoses (Mesholam-Gately et al., 2009),there are no meta-analyses to date that specifically examinecognitive functioning in those presenting with a first episode of BD.This represents a significant gap in the evidence-base that we seekto address in this systematic review and meta-analysis.

2. Methods

2.1. Data sources

Searches of peer-reviewed empirical studies between January1980 and December 2013 were conducted in PubMed and PsycInfodatabases (1980 was selected since this was the first year that theterm ‘bipolar disorder’ was codified in the DSM). Various combi-nations of search terms were used, including: first, single, early,episode, bipolar, manic, mania, neuropsycho*, neurocognit*, cog-niti*, impairment, deficit, and functioning. Citations were alsoreviewed for additional studies.

2.2. Study selection

Inclusion criteria were: 1/ an adult sample was recruited(age � 18 years) given evidence that paediatric BD may bephenomenologically different from adult BD (Pavuluri et al., 2005);2/ diagnoses of BD were made according to DSM (i.e. Bipolar IDisorder, Bipolar II Disorder) or ICD (i.e. Mania, or Bipolar AffectiveDisorder) criteria; 3/ findings for BD cases were compared withcontrols; 4/ the assessment of traditional neuropsychologicalfunctions using valid and reliable tests (Strauss et al., 2006) usedroutinely in clinical practice (i.e. excluding measures of cognitivebiases, or experimental paradigms with wide-ranging design pa-rameters and administration procedures, such as N-back tasks); 5/sufficient statistical data were reported for transformation intoeffect sizes (ES), or the relevant data were available from theoriginal researchers; and 6/ studies were published in the Englishlanguage. We did not impose a strict criteria regarding whether the‘first episode’ constituted amanic, hypomanic, mixed, depressive orpsychotic episode (according to the DSM or ICD), or first hospital-isation, since very few studies recruited a homogenous clinicalgroup based on current mood state.

Studies were excluded if they included cases who: i/ werediagnosed with a schizophrenia-spectrum psychosis (e.g. schizo-phrenia, schizoaffective disorder) according to DSM or ICD criteria;ii/ had electroconvulsive therapy in the preceding 12months; or iii/included cases potentially confounded by neurodegenerative dis-ease (i.e. age>64). Fulfilment of eligibility criteriawas confirmed bytwo authors (RSCL, KRG), who independently rated the studies; anydiscordance was then resolved by consensus. Only studies with thelargest sample were included in the instance of overlappingsamples.

As shown in Fig. 1, 309 titles and abstracts were initially iden-tified, with the final inclusion of 12 studies that met eligibilitycriteria in full.

Please cite this article in press as: Lee RSC, et al., A meta-analysis of neuropPsychiatric Research (2014), http://dx.doi.org/10.1016/j.jpsychires.2014.06

2.3. Meta-analytic procedure

Pooling of ES, tests of homogeneity, assessment of publicationbias, and moderator analyses, were all conducted using randomeffects modelling in Comprehensive Meta-Analysis Version 2.0software (Borenstein et al., 2005). Random effects modelling forpooling ES was used because the population varied across studies(Hedges and Olkin, 1985). Random effects modelling was imple-mented for meta-regression (unrestricted-maximum likelihood)and subgroup analyses (method of moments) since this carried theleast restrictive set of statistical assumptions.

All ES was represented by the standardised mean differences(SMD) in neuropsychological performance between BD cases andcontrols, and was calculated using Hedges' correction for bias insmall samples (Hedges' g; Hedges and Olkin, 1985). A positive SMDindicated poorer performance in BD cases compared with controls;the size of the ES was interpreted according to Cohen'sd (0.2 ¼ small; 0.5 ¼ medium; 0.8 ¼ large; Cohen, 1988; Lee et al.,2012). As described in Table 1, cognitive domains included werepsychomotor speed, attention and working memory, verballearning and memory, visual learning and memory, attentionalswitching, verbal fluency, cognitive flexibility, and response inhi-bition. These were grouped according to empirical models of neu-ropsychological functioning (Table 1; see Lezak et al., 2012, for adetailed overview). In longitudinal studies, only baseline neuro-psychological results were used to circumvent practice effects. If astudy used more than one test for the same domain, the averagedES was calculated.

Homogeneity was tested using the Q-statistic (c2). The I2 sta-tistic was used to quantify the proportion of between-study vari-ability that reflected differences in real effects rather than randomerror. In keeping with the literature, I2 values of 25%, 50% and 75%were interpreted as indicating small, moderate and high pro-portions of heterogeneity, respectively (Higgins et al., 2003). Tau2

was used to quantify the absolute degree of heterogeneity. Eggers'test (Egger et al., 1997) was conducted to determine the presence ofpublication bias, and Rosenthal's Fail-Safe N (Rosenthal, 1979) wassubsequently conducted to determine the number of studies thatwould be theoretically required to render the pooled ES non-significant. Duval and Tweedie's Trim and Fill method (Duval andTweedie, 2000a, 2000b) was applied to hypothetically incorpo-rate studies that may have been unpublished due to publicationbias, yielding an adjusted pooled ES.

Sensitivity analyses were conducted for each cognitive domainto determine the robustness of the current findings. That is, theresiliency of every pooled ES to individual studies was tested tocontrol for the potential peculiarity of each single study.

2.4. Moderator analyses

Variability in ES was examined using meta-regression (forcontinuous predictors) and subgroup analyses (Qbet, or betweensubgroup homogeneity, for dichotomous/categorical predictors).The predictors of heterogeneity were selected on the basis of pre-vious reviews and meta-analyses, and were included if a sufficientnumber of studies had reported these variables (i.e. no more thanone study missing per cognitive domain). These were grouped aseither:

1) Demographic predictors included ‘age’, ‘sex’, and ‘IQ’. IQ was notseparated into premorbid and current IQ since, contrary toschizophrenia-spectrum psychoses, BD is not associated withpremorbid intellectual decline, and as such, premorbid andcurrent IQ should be equivalent (Lewandowski et al., 2011;


Fig. 1. Study selection flowchart.

R.S.C. Lee et al. / Journal of Psychiatric Research xxx (2014) 1e11 3

Reichenberg et al., 2002; Torres et al., 2007; Toulopoulou et al.,2006).

2) Clinical predictors included clinical symptom status (‘euthymic’versus ‘not euthymic’). Symptom status was not treated as acontinuous variable (e.g. Hamilton Depression Rating Scalescore) since subsequent review of the studies revealed thatclinical rating scales were highly heterogeneous (i.e. not com-parable), and not every study reported this data. Exclusion ofcases diagnosed with substance use disorders (‘SUD’ versus ‘noSUD’) was also included as a clinical predictor to determinewhether substance misuse potentially influenced cognitivefindings.

3) Methodological predictors included whether controls werehealthy controls, or utilised non-psychotic or healthy relativesas controls (‘HC’ versus ‘not HC’). In addition, whether BD cases


were demographically-matched to controls (‘matched CTL’versus ‘not matched CTL’) was also included as an index ofmethodological rigour.

In the case that a categorical predictor was significantly asso-ciated with a cognitive outcome, we reported the ES for eachsubgroup. Multiple predictors were not simultaneously enteredinto a single meta-regression or subgroup analysis as a multivariatemodel, since this would detrimentally affect the validity of findingsthrough low case-to-variable ratios and potential multicollinearity.A more liberal critical value of 0.10 was used for tests of homoge-neity (overall Q and Qbet analyses) since the Q-statistic has beenshown to lack power to detect heterogeneity and, as such, wasmore susceptible to Type II errors (Borenstein et al., 2009). Indeed,the use of 0.10 as the critical alpha is now considered common


Table 1Cognitive domains and the corresponding neuropsychological tests included in eachanalysis.

Cognitive domain Neuropsychological test

Psychomotor Speed Trail Making Test-Part A; WAIS Digit Symbol-CodingAttention/Working

MemoryWAIS Digit Span Longest Digits Forward, LongestDigits Backward; WAIS Letter-Number Sequencing

Learning and MemoryVerbal WMS Logical Memory I and II, California Verbal

Learning Test Total, Short Delay Free Recall orLong Delay Free Recall; Rey Auditory VerbalLearning Test Total and Long Delay Recall

Visual WMS Visual Reproduction I and II; ReyeOsterriethComplex Figure Short Delay Recall; CANTAB PairedAssociates Learning

Executive FunctioningAttentional Switching Trail Making Test-Part BVerbal Fluency Letter Fluency (F-A-S, C-F-L); Category Fluency

(animals, body parts, fruits, animals, boys' names)Cognitive flexibility Wisconsin Card Sorting Test; CANTAB

Intra-/Extradimensional Set Shift; Brixton SpatialAnticipation Test

Response Inhibition Stroop Color-Word Interference; Haylings SentenceCompletion Task

CANTAB ¼ Cambridge Neuropsychological Test Automated Battery.WAIS ¼ Wechsler Adult Intelligence Scale. WMS ¼ Wechsler Memory Scale.


practice (Zeggini and Ioannidis, 2009). In addition, since the pre-dictors were selected a priori, and due to the low sensitivity of testsof homogeneity to heterogeneity, moderator analyses were con-ducted irrespective of whether the omnibus Q-statistic was sig-nificant at the domain-level.

3. Results

3.1. Included studies

As shown in Table 2, nine studies utilised either the DSM-III-R orDSM-IV for diagnosis, whereas three used the ICD-10. Three studiesincluded solely euthymic patients with first-episode BD cases(L�opez-Jaramillo et al., 2010; Nehra et al., 2006; Torres et al., 2010),whereas nine studies included cases that were symptomatic at thetime of neuropsychological testing (Ayres et al., 2007; Barrett et al.,2009; Fleck et al., 2008; Gruber et al., 2008; Hellvin et al., 2012; Hillet al., 2009; Hirayasu et al., 2000; Lebowitz et al., 2001; Zanelliet al., 2010). The countries of origin were diverse, with fivestudies coming from the United States, two from the UnitedKingdom, and one each from Brazil, Canada, Colombia, India, andNorway. Sample sizes were generally small, ranging from 16 to 45.There were a total of 341 BD cases, compared with a total of 1009controls. The proportion of male participants was a weightedaverage of 52.5% for BD cases and 53.6% for controls. The mean agewas aweighted average of 28.2 years for BD cases and 32.4 years forcontrols. The average IQ ranged from 84 to 115 (reported only in 11studies). Four studies explicitly excluded participants with sub-stance abuse or dependence. Eight studies used healthy controlgroups, whereas two used non-psychotic controls, one usedhealthy first-degree relatives and another used healthy second-degree relatives. Six studies reported discrepant demographiccharacteristics between BD cases and controls on at least one var-iable (e.g. age, gender, or educational level) and, thus, wereconsidered not matched on demographic factors.

3.2. Pooled effect sizes and heterogeneity

As shown in Table 3, initial analyses of first-episode BD casesdemonstrated poorer neuropsychological functioning compared


with controls across all domains (p < 0.05) except visual learningand memory (see Supplementary Figures for forest plots of eachcognitive domain).

Patients demonstrated poorer psychomotor speed (Hedges'g ¼ 0.56, p < 0.01), with a moderate level of heterogeneity acrossstudies (c2 ¼ 15.67, p < 0.05, I2 ¼ 68%, tau2 ¼ 0.14). Similarly,attention and working memory was poorer in patients than incontrols (Hedges g ¼ 0.37, p < 0.001), although there was no evi-dence that the ES varied across studies (c2 ¼ 6.28, p > 0.10).

With respect to learning and memory functioning, BD caseswere significantly worse than controls in the verbal domain(Hedges' g¼ 0.30, p < 0.001), although there was no evivdence thatthe ES varied across studies (c2 ¼ 4.38, p > 0.10). By contrast,learning andmemory for visual informationwas notmore impairedin patients compared to controls (Hedges' g ¼ 0.17, p > 0.10).

In terms of executive functioning, attentional switching (Hed-ges' g ¼ 0.52, p < 0.001), verbal fluency (Hedges' g ¼ 0.32,p < 0.001), cognitive flexibility (Hedges' g ¼ 0.41, p < 0.01), andresponse inhibition (Hedges' g ¼ 0.44, p < 0.001) were all worse inpatients than in controls, with a moderate level of ES variability forcognitive flexibility (c2 ¼ 14.69, p < 0.05, I2 ¼ 59%, tau2 ¼ 0.09). Bycomparison, therewas no evidence that the ES varied across studiesfor attentional switching (c2 ¼ 6.70, p > 0.10), verbal fluency(c2 ¼ 9.44, p > 0.10) and response inhibition (c2 ¼ 3.82, p > 0.10).

3.3. Moderator analyses

In terms of demographic factors, there was no evidence that anyneuropsychological function differed according to age, sex distri-bution (% male) or IQ (p > 0.05; see Table 4).

Compared with controls, symptomatic BD cases demonstratedmore severe response inhibition deficits than euthymic cases(Qbet ¼ 3.13, p < 0.10). When analysed in subgroups (see Table 5),only symptomatic BD cases demonstrated significant response in-hibition deficits (Hedges' g ¼ 0.59, p < 0.001). Explicit exclusion ofcases with SUDs was not systematically associated with ES in anycognitive domain (p > 0.10).

Methodologically, studies using HC found more severe deficitsin attention and working memory (Qbet ¼ 3.52, p < 0.10), andcognitive flexibility (Qbet ¼ 3.15, p < 0.10), than studies using non-psychotic controls or healthy relatives. As such, studies using HCdemonstrated medium sized deficits in attention and workingmemory (Hedges' g ¼ 0.51, p < 0.001), whereas those using othertypes of controls detected a small ES (Hedges' g ¼ 0.22, p < 0.05).Only studies using an HC group detected a pooled cognitive flexi-bility deficit (Hedges' g ¼ 0.55, p < 0.001). Studies that properlymatched BD cases to controls on demographic factors showed alarger pooled ES than those that did not in the domain of verballearning and memory (Qbet ¼ 3.44, p < 0.10), with only well-matched studies demonstrating a significant deficit (Hedges'g ¼ 0.44, p < 0.001). No other cognitive domain was systematicallyaffected by methodological factors (p > 0.10).

3.4. Publication bias

Of the neuropsychological domains that differentiated casesfrom controls, psychomotor speed (t ¼ 3.27, df ¼ 4, p < 0.05),attentional switching (t ¼ 6.30, df ¼ 4, p < 0.05), and verbal fluency(t¼ 2.42, df¼ 8, p < 0.05) were influenced by publication bias, withsmaller and less reliable (i.e. larger standard errors) studiesreporting larger ES. Using Rosenthal's Fail-Safe N, 35, 37 and 51studies would be required to nullify the significant pooled ES ineach of these cognitive domains, respectively. Using Duval andTweedie's Trim & Fill method, the ES (Hedges' g) for attentionalswitching deficit was re-adjusted to 0.34 (p < 0.05), whereas the


Table 2Summary of key characteristics of all studies that were eligible for inclusion in the meta-analysis of neuropsychological profile in cases of first episode bipolar disorders.

Study (year, country) CTL BD Cognitive domains assessed

N (% male) Mean age(years) ± SD

Type N (% male) Mean age(years) ± SD

Diagnosticsystem

Diagnosticinstrument

Diagnosis(verbatim)

Clinicalratinginstrument

Euthymic at testing

DSM-III-R or DSM-IV first episodea bipolar disorderL�opez-Jaramillo et al.

(2010, Colombia)60 (66.6) 37.4 ± 8.6 H-FD-REL 24 (66.7) 32.0 ± 10.2 DSM-IV DIGS “BD-I” HDRS,

YMRSYes Attention and working

memory, attentional switching,cognitive flexibility, cognitiveinhibition, psychomotor speed,verbal fluency, verbal learningand memory, visual learningand memory

Torres et al.(2010, Canada)

25 (48.0) 22.5 ± 4.8 HC 45 (51.1) 22.2 ± 3.9 DSM-IV MINI “Bipolar I disorder” BPRS,HDRS,PANSS,YMRS

Yes Attention and workingmemory, attentional switching,cognitive flexibility, cognitiveinhibition, psychomotor speed,verbal fluency, verbal learningand memory, visual learningand memory

Hirayasu et al.(2000, United States)

22 (90.9) 24.5 ± 4.7 HC 24 (75.0) 23.6 ± 5.0 DSM-III-R SCID “Bipolar disorder maniawith psychotic features”

BPRS No Attention and working memory

Lebowitz et al.(2001, United States)

30 (43.3) 31.2 ± 8.1 HC 19 (52.6) 27.4 ± 7.2 DSM-IV SCID “BPD, manic or mixedepisode”

HDRS,SAPS,YMRS

No Verbal fluency

Ayres et al. (2007, Brazil) 383 (47.3) 32.1 ± 11.3 NP-CTL 41 (36.6) 33.5 ± 12.5 DSM-IV SCID “Bipolar disorder” PANSS No Attention and workingmemory, verbal fluency

Fleck et al.(2008, United States)

48 (41.7) 28.2 ± 7.9 HC 21 (52.4) 25.7 ± 9.2 DSM-IV SCID “Bipolar I disorder” HDRS,YMRS

No Cognitive flexibility

Gruber et al.(2008, United States)

20 (75.0) 25.3 ± 4.7 HC 26 (73.1) 24.4 ± 5.5 DSM-IV SCID “Bipolar disorder” “Clinicalratingscales”

No Attentional switching, cognitiveflexibility, cognitive inhibition,psychomotor speed, verbalfluency

Hill et al.(2009, United States)

41 (58.5) 24.9 ± 8.8 HC 22 (59.0) 22.7 ± 6.4 DSM-IV SCID “Bipolar with psychosis (BP)” HDRS,PANSS

No Attention and workingmemory, attentional switching,psychomotor speed, verbalfluency, verbal learning andmemory

Hellvin et al.(2012, Norway)

110 (44.5) 31.1 ± 9.8 HC 34 (44.1) 31.2 ± 9.6 DSM-IV SCID “BD-I” IDS-C,PANSS,YMRS

No Attention and workingmemory, cognitive flexibility,cognitive inhibition, verbalfluency, verbal learning andmemory, visual learning andmemory

ICD-10 first episodea bipolar affective disorder and/or maniaNehra et al. (2006, India) 20 (60.0) 38.7 ± 12.2 H-SD-REL 16 (50.0) 28.4 ± 9.3 ICD-10 ICD-10-DCR “BPAD/hypomanic/manic

episode”HDRS,YMRS

Yes Attentional switching, cognitiveflexibility, psychomotor speed,verbal fluency

Barrett et al.(2009, United Kingdom)

67 (58.2) 33.2 ± 11.0 HC 32 (50.0) 36.7 ± 9.3 ICD-10 SCAN “Bipolar disorder/manic” BDI, PANSS No Attention and workingmemory, cognitive flexibility,cognitive inhibition, verbalfluency, verbal learning andmemory, visual learning andmemory

(continued on next page)

R.S.C.Leeet

al./Journal

ofPsychiatric

Researchxxx

(2014)1e11

5

Pleasecite

thisarticle

inpress

as:LeeRSC,etal.,A

meta-analysis

ofneuropsychologicalfunctioningin

first-episodebipolar

disorders,JournalofPsychiatric

Research(2014),http://dx.doi.org/10.1016/j.jpsychires.2014.06.019

Table

2(con

tinu

ed)

Study(yea

r,co

untry)

CTL

BD

Cog

nitivedom

ainsassessed

N(%

male)

Mea

nag

e(yea

rs)±SD

Type

N(%

male)

Mea

nag

e(yea

rs)±SD

Diagn

ostic

system

Diagn

ostic

instrumen

tDiagn

osis

(verba

tim)

Clin

ical

rating

instrumen

t

Euthym

icat

testing

Zanelliet

al.

(201

0,United

Kingd

om)

177(63.3)

37.2

±12

.9NP-CTL

37(40.5)

28.1

±8.1

ICD-10

SCAN

“Bipolar

disorder

orman

ia(F30

.2,F

31.2,F

31.5)”

SCAN-IGC

No

Atten

tion

andworking

mem

ory,

attention

alsw

itch

ing,

psych

omotor

spee

d,v

erba

lfluen

cy,v

erba

llea

rningan

dmem

ory,

visu

allearningan

dmem

ory

BD

¼Bipolar

Disorder.B

DI¼

BeckDep

ressionInve

ntory.B

PAD

Bipolar

Affective

Disorder.B

PD¼

Bipolar

Disorder.B

PRS¼

Brief

Psyc

hiatric

RatingScale.

(NP-)CTL

¼(N

onpsych

otic-)Con

trols.DIGS¼

Diagn

osticInterview

for

Gen

eticStudies.DSM

(-III-Ror

eIV)¼

Diagn

ostican

dStatisticalM

anual

ofMen

talD

isorders(-Th

irdEd

itionRev

ised

orFo

urthEd

ition).H-FD-R

EL¼

Hea

lthy-FirstDeg

ree-Relatives.H

-SD-R

EL¼

Hea

lthy-Se

condDeg

ree-Relatives.

HC¼Hea

lthyCon

trols.HDRS¼Ham

ilton

Dep

ressionRatingScale.ICD-10(-D

CR)¼

Intern

ational

Classification

ofDisea

seve

rsion10

(-Diagn

osticCriteriaforResea

rch).ID

S-C¼Inve

ntory

ofDep

ressiveSy

mptoms-Clin

icianrated.

K-SADS-PL

¼Kiddie-Sch

edule

forAffective

Disordersan

dSchizop

hrenia-Presentan

dLifetime.MIN

I¼MiniIntern

ational

Neu

ropsych

iatricInterview.P

ANSS

¼Po

sitive

andNeg

ativeSy

ndromeScale.SC

AN(-IGC)¼

Sched

ulesfor

Clin

ical

Assessm

entin

Neu

ropsych

iatry(-Item

GroupChecklist).S

CID

¼StructuredClin

ical

Interview

forDSM

-IIIor

IV.Y

MRS¼

You

ngMan

iaRatingScale.

aW

edid

not

imposeastrict

criteria

rega

rdingwhether

the‘firstep

isod

e’co

nstitutedaman

ic,h

ypom

anic,m

ixed

,dep

ressiveor

psych

otic

episod

e(accordingto

theDSM

orICD),or

firsthospitalisation,since

very

few

studies

recruited

ahom

ogen

ousclinical

grou

pon

theba

sisof

curren

tclinical

state.



pooled ES for psychomotor speed and verbal fluency remainedsignificant (p < 0.05) and unchanged in magnitude.

3.5. Sensitivity analyses

Sensitivity analyses were conducted for each significant pooledES. The statistical significance remained unchanged for eachcognitive domain no matter which study was removed from theanalysis, underscoring the robustness of the current findings.Indeed, by inspecting the range of pooled ES within each cognitivedomain, individual removal of each study from the pooled effectproduced only a set of relatively narrow pooled ES estimates.

4. Discussion

This is the first meta-analysis to systematically investigate theneuropsychological profile of first-episode bipolar disorders, andto delineate potential predictors of between-study heterogeneity.Consistent with previous meta-analyses of later stage and morechronic patients (Arts et al., 2008; Bora et al., 2009; Bourne et al.,2013; Kurtz and Gerraty, 2009; Mann-Wrobel et al., 2011; Torreset al., 2007), our findings indicate that widespread neuropsycho-logical dysfunction is already evident in the first episode of illness.The severity of impairment was generally comparable to the me-dium effects from studies including cases with recurrent episodes(Bourne et al., 2013), although only a limited number of cognitivedomains were explored in that meta-analysis. By contrast,compared with other meta-analyses where the majority of ESestimates have typically been large (Arts et al., 2008; Bora et al.,2009, 2011; Kurtz and Gerraty, 2009; Mann-Wrobel et al., 2011;Torres et al., 2007), our findings appear to show a more modestdegree of dysfunction. Taking solely into account studies usinghealthy or well-matched controls, medium to large ES decrementswere found for attention and working memory, and cognitiveflexibility, and small to medium ES decrements were found forverbal learning and memory. Small to medium ES reductions werealso detected for the domains of attentional switching and verbalfluency, whereas psychomotor slowing was evident at a mediumto large degree of severity. Dysfunction in response inhibition wasonly observed in those who were symptomatic at the time oftesting, at a similarly medium to large degree of severity. Takentogether, the present findings have key implications for neuro-developmental and neurodegenerative models of BD. Each ofthese findings and their implications will be addressed in turn.

4.1. Age, IQ and sex

Contrary to previous research (Bora et al., 2009), a younger ageof first-episode BD was not associated with more pronouncedneuropsychological dysfunction. This diverges from the notionthat a younger age of onset may lead to more severe cognitiveimpairment through a greater cumulative impact of illness in thiscritical growth period, particularly in light of evidence that dura-tions between illness onset and first treatment is typically greaterin younger BD cases (Suominen et al., 2007). The view thatyounger age of onset is associated with poorer cognitive outcomeis also supported by findings of poorer prognosis in those whodevelop BD at earlier ages, and with evidence that younger first-episode mania cases are associated with poorer syndromal andfunctional recovery (Tohen et al., 2003). It is likely that the modestnumber of studies currently reviewed, and the restriction of theage range due to the inclusion of only adult first-episode cases,contributed to obscuring any potential relationship between ageand cognitive functioning, particularly if the association is small.


Table 3Number of studies (k), pooled sample size (N), pooled ES (Hedges' g), homogeneity (Q, I, tau2), publication bias, and sensitivity analyses (pooled ES range; each study removed),by cognitive domain.

Cognitive domain k BD N CTL N Hedges' g 95% CI Q I tau2 Egger's test (t) Fail-safe N Trim & Fill Sensitivity analyses

Psychomotor speed 6 170 349 0.56** 0.19e0.92 15.67* 68% 0.14 3.27* 35 0.56* 0.42e0.67Attention/working memory 8 259 891 0.37*** 0.22e0.51 6.28 0% 0.00 1.77 e e 0.34e0.44Learning and MemoryVerbal 6 194 486 0.30*** 0.13e0.48 4.38 0% 0.00 1.46 e e 0.25e0.36Visual 5 172 445 0.17 �0.02e0.35 3.61 0% 0.00 e e e e

Executive FunctioningAttentional switching 6 170 349 0.52*** 0.28e0.78 6.70 25% 0.02 6.30** 37 0.34* 0.44e0.61Verbal fluency 10 296 939 0.32*** 0.18e0.47 9.43 5% 0.00 2.42* 51 0.32* 0.28e0.37Cognitive flexibility 7 198 356 0.41** 0.12e0.70 14.69* 59% 0.09 0.38 e e 0.33e0.49Response inhibition 5 161 288 0.44*** 0.24e0.65 3.82 0% 0.00 0.25 e e 0.36e0.51Total k/N 12 341 1009

BD ¼ Bipolar Disorder. CTL ¼ Control. ES ¼ effect size.*p � 0.05. **p � 0.01. ***p � 0.001.


Intellectual functioning, at least in those within the normalrange, was not associated with deficits in any of the investigatedcognitive domains. This was not surprising given evidence showingthat cognitive deficits remain significant despite controlling for IQat the individual-level among BD cases canvassed across all stagesof illness (Bourne et al., 2013). Taken together, low intelligence isunlikely to be a prominent risk factor for neuropsychologicaldysfunction in BD. Similarly, sex was not related to the severity ofcognitive dysfunction, a finding that is in keeping with previousmeta-analytic evidence (Mann-Wrobel et al., 2011).

4.2. Euthymia, psychosis and substance misuse

Response inhibition was the only neuropsychological functionassociated with current mood state. Studies including patientsexperiencing clinically significant (hypo-)manic, depressive ormixed states, showed response inhibition deficits that were notapparent in the studies of euthymic patients. This finding contrastswith a previous meta-analysis that included relatively older adultsand cases with primarily long-established BD, and their first andsecond-degree relatives, which found that response inhibitionmight be a robust marker of BD, even during remission (Bora et al.,2009). A possible explanation for this discrepancy may be thatresponse inhibition deficits in the earlier stages of BD are lessentrenched and, as such, are only expressed when the symptomburden is higher and cognitive resources are ‘taxed’. This assertionis consistent with functional neuroimaging evidence showing

Table 4Moderator analyses to determine predictors of heterogeneity.

Demographic factors

Age Sex (% male) IQ

Cognitive domain % k Z % k Z % k Z

Psychomotor speed 100 �1.47 100 0.14 100 �0.5Attention/working memory 100 �0.02 100 1.28 88 0.9Learning and memoryVerbal 100 �1.19 100 0.78 100 0.6Visual 100 �1.12 100 �0.15 100 1.5Executive functioningAttentional switching 100 �1.20 100 1.09 100 �0.4Verbal fluency 100 �0.61 100 0.84 90 �1.2Cognitive flexibility 100 �0.75 100 1.39 100 1.7Response inhibition 100 0.15 100 �0.63 100 1.4

CTL ¼ Controls. SUD ¼ Substance Use Disorder.y�0.10.


response inhibition deficits and reduced activation in associatedprefrontal cortical regions in highly symptomatic first-episodemania patients (Strakowski et al., 2008). In keeping with thisproposition, albeit to the contrary, response inhibition deficits seemto persist despite symptomatic remission in the later stages of BD(Townsend et al., 2012). All other areas of cognitive dysfunction infirst-episode BD, however, currently appear to be independent ofclinical state, consistent with findings in older cohorts comprisedlargely of later stage BD cases (Martinez-Aran et al., 2004).

Given the range of diagnostic systems and varying sets ofdiagnostic criteria, there was insufficient certainty regardingwhether studies included cases with psychotic features and, assuch, we were unable to code studies accordingly. Indeed, somestudies (Ayres et al., 2007; Gruber et al., 2008) only reported datafor diagnosed cases of ‘bipolar disorder’ without further diagnosticinformation. Others reported BD cases as part of their ‘first-episodepsychosis’ cohort without specifying whether this amounted to theinclusion of solely psychotic patients or whether they simplydefined BD as a form of psychotic illness. As a result of these lim-itations, we were constrained in our ability to systematicallycorroborate prior findings (Martinez-Aran et al., 2008; Simonsenet al., 2011) of more pronounced verbal memory and executivefunctioning deficits in BD cases with psychosis for this early episodecohort. We were similarly unable to examine whether psychosisseverity, viewed dimensionally and measured by clinical ratingscales, was related tomore severe cognitive deficits, since a numberof studies did not report this explicitly, and the studies that did

Clinical factors Methodological factors

Euthymia SUD Healthy CTL Matched CTL

% k Qbet % k Qbet % k Qbet % k Qbet

5 100 0.12 100 1.35 100 0.52 100 0.710 100 0.01 100 0.24 100 3.52y 100 1.95

7 100 2.20 100 0.02 100 0.30 100 3.44y2 100 0.40 100 0.00 100 1.57 100 1.68

5 100 0.14 100 1.36 100 1.34 100 0.143 100 0.52 100 0.72 100 0.00 100 0.452 100 1.22 100 0.74 100 3.15y 100 0.256 100 3.13y 100 0.00 100 1.81 100 0.35


Table 5Subgroup analyses of cognitive domains controlling for known clinical or method-ological confounds.

Cognitive domain Moderator k Studies Hedges' g

Attention/Working Memory HC 5 0.51***Not HC 3 0.22*

Verbal Learning and Memory Matched CTL 4 0.44***Not Matched CTL 2 0.11

Cognitive Flexibility HC 5 0.55***Not HC 2 0.00

Response Inhibition Euthymic 2 0.20Not Euthymic 3 0.59***

Note: all between group differences (e.g. HC vs. Not HC) are significant (p < 0.10; seeTable 3).HC ¼ healthy controls. CTL ¼ controls.*p � 0.05. ***p � 0.001.


used varying rating scales. There is good evidence in the literatureto suggest that non-psychotic BD, at least in the later stages, sharesa similar profile of neuropsychological deficits with psychotic BD(Lewandowski et al., 2011). Thus, it remains to be clarified whetherindividuals experiencing BD with psychosis follow a separate, moremaligned, cognitive trajectory early in the course of illness. Datafrom our group suggest that this appears to be largely undetectableat this early stage (Hermens et al., 2011).

Surprisingly, the exclusion of cases with comorbid substanceabuse and dependence did not affect the magnitude of cognitivedeficits identified. This finding corroborates the within-studyfindings of Torres et al. (2010), who showed that there were nodifferences in cognitive functioning between those with andwithout SUDs in first-episode mania cases. Given data to the con-trary from our group, suggesting that problematic drinking maycompound neuropsychological dysfunction in affective disorders(Hermens et al., 2013a), the effects of substance misuse in BDclearly necessitates further study, particularly given evidence thatSUDs are especially common in young adults presenting with se-vere mental illnesses (Hermens et al., 2013b), and that more than50% of young mania patients meet criteria for a lifetime alcohol usedisorder (Wittchen et al., 2003). Indeed, the failure to find morepronounced cognitive deficits in cohorts not explicitly excludingSUDs may be related to the young age of these samples, when thecumulative effects of substancemisuse is hypothetically less than inthose with long-established BD.

4.3. Neuroprogression in BD

The failure to find a deficit in visual learning and memory isnoteworthy given the emerging view that visual memory, as con-trasted with verbal memory, deteriorates by the time BD is fullyestablished in cases largely comprised of chronic patients (Maziadeet al., 2011). Taken together with the current results, it is likely thatvisual memory deficits do not significantly deteriorate by the first-episode of illness in BD, or are undetectable by traditional neuro-psychological instruments. Thus, perhaps visual mnemonicdysfunction as measured by commonly used and standardisedneuropsychological tests only becomes detectable through recur-rence and chronicity (Lewandowski et al., 2011). This contrasts withthe psychosis literature (comprised largely of schizophrenia-spectrum cases), where cognitive deficits can decline almost adecade prior to the onset of full-threshold positive symptoms, withfurther deterioration on transition to first-episode of illness (Kahnand Keefe, 2013). In these psychosis cohorts, visual memory im-pairments are also predictive of those who transition from psy-chosis at-risk states to frank psychosis (Brewer et al., 2005; Linet al., 2013). Interestingly, the magnitude of cognitive deficits in


BD from the current meta-analysis was generally smaller thanthose identified in first-episode schizophrenia and schizoaffectivedisorder (Mesholam-Gately et al., 2009), and may indicate thepresence of distinct pathways to the development of thesedisorders.

Overall, the current findings have key implications for theoriesof neuroprogression in BD. Various studies in children and ado-lescents have suggested that those who go on to develop BD do notexperience early premorbid neurodevelopmental decline in neu-ropsychological (Reichenberg et al., 2002), intellectual(Toulopoulou et al., 2006), or academic (Kutcher et al., 1998;Quackenbush et al., 1996) domains of functioning, although thishas not always been replicated (Meyer et al., 2004; Tiihonen et al.,2005). On balance, the evidence shows that premorbid neuro-developmental decline akin to that observed in schizophrenia doesnot occur (Murray et al., 2004). Subtle neuropsychological deficitshave been observed in adolescents who later develop BD (Meyeret al., 2004), which may worsen during the ‘emerging’ (Hermenset al., 2010) or ‘prodrome’ phase immediately preceding illnessonset (Ratheesh et al., 2013), with additional decline occurring withrepeated manic episodes (L�opez-Jaramillo et al., 2010). Thesefindings parallel neuroimaging evidence of relatively intact grossbrain structure in the early phases of BD (Lisy et al., 2011), withprogressive grey matter (Lyoo et al., 2006) and ventricular(Strakowski et al., 2002) changes evolving as illness progresses.Thus, BD may be seen as a neuroprogressive disorder with a subtleneurodevelopmental component immediately preceding illnessonset (Lewandowski et al., 2011). This may be followed by subse-quent neurodegeneration following first episode (as currentlydemonstrated in most cognitive domains), and progressive cogni-tive decline through prolonged recurrence (at least in the case ofvisual memory). It remains to be clarified in longitudinal studies ordirect head-to-head comparisons between solely first-episode andsolely multiple-episode cases whether neuropsychologicaldysfunction in other cognitive domains undergoes further declinethrough illness progression. Further, the long-term impact of psy-chotropic medication-use on the neuropsychological trajectory ofBD warrants greater attention. Nonetheless, the current finding ofsignificant widespread cognitive deficits supports the neuro-progressive model of BD and further underscores the importance ofearly intervention in BD (Berk et al., 2013), particularly given evi-dence that earlier intervention in this cohort is associated withmore favourable outcomes (Scott et al., 2006).

4.4. Nature of control groups

Our data show that studies using healthy comparison cohortsdemonstrated more severe cognitive dysfunction compared tostudies using non-psychotic controls or ‘healthy’ familial cohorts.This is a critical finding given the burgeoning debate regarding theimportance of choosing appropriate control samples in psychiatricresearch (Lee et al., 2007; Schwartz and Susser, 2011). Studies usinghealthy control samples identified deficits in the domains ofattention and working memory, and cognitive flexibility, whereasstudies using non-psychotic controls and those putatively at ge-netic risk of BD (first and second-degree relatives) found smallerdeficits (in the former cognitive domain) or failed to find a deficitaltogether. Thus, non-psychotic or familial controls demonstratedmore similar levels of cognitive functioning to probands, thanhealthy controls to probands, and are in keeping with the conceptof cognitive endophenotypes in those with a genetic predispositionto BD (Arts et al., 2008; Bora et al., 2009).

Studies using control samples that were well matched to BDcases on demographic factors were associated with worse verballearning and memory functioning. Specifically, only studies that



properly matched controls elicited a small to medium verballearning and memory deficit, whereas those that did not failed tofind a significant deficit. This finding extends previous evidence innormative populations to bipolar illness, which consistentlydemonstrate a role for age, sex, and education attainment inverbal memory functions (Carstairs et al., 2012; Strauss et al.,2006).

4.5. Limitations

Due to a number of studies aggregating cases across differentclinical states as well as using varying rating scales, wewere unableto compare cognitive functioning between different mood states inBD (e.g. depressed versus manic). Moreover, studies seldom pro-vided a clear definition of whether a diagnosis of first-episode wasdefined as; the first episode of mania, hypomania, depression ormixed states, or first episode of psychosis; whether first-episodewas operationalised as first hospitalisation; and how many pre-ceding episodes of depression, hypomania and mixed episodeswere present. Nevertheless, the mean age in the current meta-analysis (approximately 28 years of age) supports the view thatBD cases in the present investigation were early in their course ofBD (Baldessarini et al., 2010), and more proximal to the mean onsetage of BD as compared with previous meta-analyses, which have allincluded cases across the spectrum of chronicity. The current meanage is also in keeping with the mean age of BD onset of approxi-mately 18 years (Merikangas et al., 2007), with a delay in diagnosistypically 5e10 years post-onset (Baldessarini et al., 2007).

Poorly defined BD groups in terms of the course of prior moodepisodes precluded us from disentangling its contributing effectson the identified cognitive deficits. This is a particular concerngiven evidence that the first episode in BD is most often adepressive episode (Angst and Sellaro, 2000), and first-episodemajor depressive disorder is associated with small to medium de-gree of similarly widespread cognitive dysfunction (Lee et al.,2012). Nonetheless, the included studies in Lee et al. (2012) pri-marily consisted of middle-aged cases (approximately 39 years ofage) and, as such, are not directly comparable to the current find-ings. Similarly, it is difficult to determine the extent that other (non-substance related) comorbid Axis I psychiatric diagnoses contrib-uted to the current findings. Most studies did not list comorbidpsychiatric diagnoses in the event that they were not explicitlyexcluded.

All studies currently reviewed included patients who weremedicated, on a variety of psychotropic regimens, and thus, sys-tematically delineating its effects was not possible. Mood stabilisers(Bourne et al., 2013; Pachet andWisniewski, 2003), antidepressants(Bora et al., 2009; Bourne et al., 2013), and antipsychotics (Boraet al., 2009) have been shown to affect psychomotor speed andverbal learning and memory abilities. However, this is not alwaysconsistently found, with some studies finding effects in othercognitive domains (Torres et al., 2010), or not at all (Kurtz andGerraty, 2009). A feasible way to disentangle the effects of medi-cation may be to investigate medication-naïve first-episode sam-ples, or the effects of medically managed withdrawal frommedication in those who have been clinically stable for sufficientlylong periods of time and who are deemed to be in completeremission.

Furthermore, the number of included studies in the currentstudy is still modest. Therefore, the present findings stands to bereplicated in future, larger quantitative reviews. Nevertheless, thecurrent results provide the first systematic evidence of widespreadneuropsychological dysfunction in first-episode bipolar disorders,with limited evidence these deficits are linked to acutemood states.In addition, due to the limited number of studies using the same


cognitive tests, we were unable to meaningfully pool ES estimatesaccording to every individual neuropsychological measure. Instead,we chose to pool the overall ES from the average performance acrossvarious measures in each study, which provided a more statisticallyreliable estimate of neuropsychological functioning. It stands to beclarified in future meta-analyses whether the current findings holdwhen the covariance matrices among non-independent test scoresare accounted forwithin each included study. In termsof the relativeseverity of neuropsychological dysfunction, since the cognitivemeasures were not psychometrically equivalent, the differences inES may also be a result of the varying sensitivities of the includedmeasures to the underlying cognitive constructs, as opposed todifferences in deficit severity per se.

There is also mounting evidence indicating that social cognitioncan mediate the impact of neuropsychological abilities on func-tional outcome in severe mental illnesses (Addington et al., 2010).In BD, however, there remains conflicting evidence of whethersocial cognition is indeed impaired (Lee et al., 2013a; Samam�e,2013). Unfortunately, none of the studies included in the currentmeta-analysis examined social cognitive abilities and this clearlyrepresents a gap in the literature that stands to be explicitlyexamined in future investigations.

4.6. Conclusions and implications

The current meta-analysis is the first to systematically delin-eate and characterise neuropsychological functioning in cases offirst-episode BD. The findings suggest that a broad range ofcognitive deficits is already present at this early stage, and sug-gestive of dysfunction in fronto-subcortical and fronto-temporalneural circuitry. It remains to be clarified whether the magnitudeof decrements is comparable to those seen in later stages of BD,although for visual learning andmemory it appears that a deficit atthis early stage is undetectable. The severity of deficits does notappear to be linked to mood state, with the exception of responseinhibition, which was found in symptomatic first-episode casesonly. Taken together, these findings indicate that the neuropsy-chological profile of BD may potentially represent a mix oflate neurodevelopmental and neurodegenerative processes withwidespread reductions in cognitive skills by the time of transitionto the first episode of illness. However, this model stands to beempirically tested in longitudinal studies of individuals at clinicalor genetic risk of developing BD, through prospectively trackingthese cases from at risk, to prodromal and first episode stages of BD(alongside well-matched healthy comparison groups, while con-trolling for the effects that medications may have on early neu-ropsychological trajectories). To this end, clinical staging modelsmay have particular utility (Hickie et al., 2013). Whether inter-individual differences in age, sex, IQ, number of previous depres-sive episodes, and experience of psychotic symptoms or episodescan explain the neuropsychological deficits that are found whencases present with a first episode of BD is debatable, but warrantsgreater attention. In addition, future studies of early episode BDshould consider separating cases according to subtypes (e.g. Bi-polar I versus II Disorder) to delineate whether these groups showdistinct neuropsychological profiles, as previously investigated inolder and more recurrent cohorts (Bora et al., 2011). Clinically, thepresence of neuropsychological deficits early in the illness, whichtheoretically may be more amenable to change, indicates that therecent introduction of cognitive remediation for depression (Leeet al., 2013b; Naismith et al., 2010), and functional remediationfor BD adults with established impairments (Torrent et al., 2013),might also need to be adapted to individuals with first-episodebipolar illness.



Role of funding source

IBH, DFH and JL were supported by a National Health andMedical Research Council Australia Fellowship awarded to IBH(464914). SLN is funded by an National Health and MedicalResearch Council Clinical Research Fellowship (402864). The fun-ders had no role in study design, data collection and analysis, de-cision to publish, or preparation of the manuscript.

Contributors

RSCL designed the study, conducted the literature searches andstatistical analyses, and wrote the first draft of the manuscript. Allauthors contributed to and have approved the final manuscript.

Conflicts of interest

None.

Acknowledgements

None.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jpsychires.2014.06.019.

References

Addington J, Girard TA, Christensen BK, Addington D. Social cognition mediatesillness-related and cognitive influences on social function in patients withschizophrenia-spectrum disorders. J Psychiatry Neurosci 2010;35:49e54.

Angst J, Sellaro R. Historical perspectives and natural history of bipolar disorder.Biol Psychiatry 2000;48:445e57.

Arts B, Jabben N, Krabbendam L, van Os J. Meta-analyses of cognitive functioning ineuthymic bipolar patients and their first-degree relatives. Psychol Med2008;38:771e85.

Ayres AM, Busatto GF, Menezes PR, Schaufelberger MS, Coutinho L, Murray RM,et al. Cognitive deficits in first-episode psychosis: a population-based study inSao Paulo. Schizophr Res 2007;90:338e43.

Baldessarini RJ, Bolzani L, Cruz N, Jones PB, Lai M, Lepri B, et al. Onset-age of bipolardisorders at six international sites. J Affect Disord 2010;121:143e6.

Baldessarini RJ, Tondo L, Baethge CJ, Lepri B, Bratti IM. Effects of treatment latencyon response to maintenance treatment in manic-depressive disorders. BipolarDisord 2007;9:386e93.

Barrett SL, Mulholland CC, Cooper SJ, Rushe TM. Patterns of neurocognitiveimpairment in first-episode bipolar disorder and schizophrenia. Br J Psychiatry2009;195:67e72.

Berk M, Berk L, Dodd S, Cotton S, Macneil C, Daglas R, et al. Stage managing bipolardisorder. Bipolar Disord 2013. http://dx.doi.org/10.1111/bdi.12099.

Bora E, Yucel M, Pantelis C. Cognitive endophenotypes of bipolar disorder: a meta-analysis of neuropsychological deficits in euthymic patients and their first-degree relatives. J Affect Disord 2009;113:1e20.

Bora E, Yucel M, Pantelis C. Neurocognitive markers of psychosis in bipolar disorder:a meta-analytic study. J Affect Disord 2010;127:1e9.

Bora E, Yucel M, Pantelis C, Berk M. Meta-analytic review of neurocognition in bi-polar II disorder. Acta Psychiatr Scand 2011;123:165e74.

Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive meta-analysisversion 2.0. Biostat 2005.

Borenstein M, Hedges L, Higgins J, Rothstein H. Introduction to meta-analysis. UK:John Wiley & Sons Ltd; 2009.

Bourne C, Aydemir O, Balanza-Martinez V, Bora E, Brissos S, Cavanagh JT, et al.Neuropsychological testing of cognitive impairment in euthymic bipolar dis-order: an individual patient data meta-analysis. Acta Psychiatr Scand 2013;128:149e62.

Brewer WJ, Francey SM, Wood SJ, Jackson HJ, Pantelis C, Phillips LJ, et al. Memoryimpairments identified in people at ultra-high risk for psychosis who laterdevelop first-episode psychosis. Am J Psychiatry 2005;162:71e8.

Carstairs JR, Shores EA, Myors B. Australian norms and retest data for the ReyAuditory and verbal learning test. Aust Psychol 2012;47:191e7.

Cohen J. Set correlation and contingency tables. Appl Psychol Meas 1988;12:425e34.


Daban C, Mathieu F, Raust A, Cochet B, Scott J, Etain B, et al. Is processing speed avalid cognitive endophenotype for bipolar disorder? J Affect Disord 2012;139:98e101.

Depp CA, Mausbach BT, Harmell AL, Savla GN, Bowie CR, Harvey PD, et al. Meta-analysis of the association between cognitive abilities and everyday functioningin bipolar disorder. Bipolar Disord 2012;14:217e26.

Duval S, Tweedie R. A Nonparametric “Trim and Fill” method of accounting forpublication bias in meta-analysis. J Am Stat Assoc 2000a;95:89e98.

Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing andadjusting for publication bias in meta-analysis. Biometrics 2000b;56:455e63.

Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by asimple, graphical test. BMJ 1997;315:629e34.

Fleck DE, Shear PK, Madore M, Strakowski SM. Wisconsin card sorting test per-formance in bipolar disorder: effects of mood state and early course. BipolarDisord 2008;10:539e45.

Glahn DC, Almasy L, Barguil M, Hare E, Peralta JM, Kent Jr JW, et al. Neurocognitiveendophenotypes for bipolar disorder identified in multiplex multigenerationalfamilies. Arch Gen Psychiatry 2010;67:168e77.

Glahn DC, Bearden CE, Bowden CL, Soares JC. Reduced educational attainment inbipolar disorder. J Affect Disord 2006;92:309e12.

Goodwin GM, Martinez-Aran A, Glahn DC, Vieta E. Cognitive impairment in bipolardisorder: neurodevelopment or neurodegeneration? an ECNP expert meetingreport. Eur Neuropsychopharmacol 2008;18:787e93.

Gruber SA, Rosso IM, Yurgelun-Todd D. Neuropsychological performance predictsclinical recovery in bipolar patients. J Affect Disord 2008;105:253e60.

Hedges LV, Olkin I. Statistical methods for meta-analysis. Academic Press; 1985.Hellvin T, Sundet K, Simonsen C, Aminoff SR, Lagerberg TV, Andreassen OA, et al.

Neurocognitive functioning in patients recently diagnosed with bipolar disor-der. Bipolar Disord 2012;14:227e38.

Hermens DF, Lee RSC, De Regt T, Lagopoulos J, Naismith SL, Scott EM, et al. Neu-ropsychological functioning is compromised in binge drinking youngadults with depression. Psychiatry Res 2013b;210(1):256e62. http://dx.doi.org/10.1016/j.psychres.2013.05.001.

Hermens DF, Naismith SL, Redoblado Hodge MA, Scott EM, Hickie IB. Impairedverbal memory in young adults with unipolar and bipolar depression. EarlyInterv Psychiatry 2010;4:227e33.

Hermens DF, Redoblado Hodge MA, Naismith SL, Kaur M, Scott E, Hickie IB. Neu-ropsychological clustering highlights cognitive differences in young peoplepresenting with depressive symptoms. J Int Neuropsychol Soc 2011;17:267e76.

Hermens DF, Scott EM, White D, Lynch M, Lagopoulos J, Whitwell BG, et al. Frequentalcohol, nicotine or cannabis use is common in young persons presenting formental healthcare: a cross-sectional study. BMJ Open 2013a;3:e002229.

Hickie IB, Scott EM, Hermens DF, Naismith SL, Guastella AJ, Kaur M, et al. Applyingclinical staging to young people who present for mental health care. EarlyInterv Psychiatry 2013;7:31e43.

Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557e60.

Hill SK, Reilly JL, Harris MSH, Rosen C, Marvin RW, DeLeon O, et al. A comparison ofneuropsychological dysfunction in first-episode psychosis patients with uni-polar depression, bipolar disorder, and schizophrenia. Schizophr Res 2009;113:167e75.

Hirayasu Y, McCarley RW, Salisbury DF, Tanaka S, Kwon JS, Frumin M, et al. Planumtemporale and heschl gyrus volume reduction in schizophrenia: a magneticresonance imaging study of first-episode patients. Arch Gen Psychiatry2000;57:692e9.

Kahn RS, Keefe RS. Schizophrenia is a cognitive illness: time for a change in focus.JAMA Psychiatry 2013. http://dx.doi.org/10.1001/jamapsychiatry.2013.155.

Kurtz MM, Gerraty RT. A meta-analytic investigation of neurocognitive deficits inbipolar illness: profile and effects of clinical state. Neuropsychology 2009;23:551e62.

Kutcher S, Robertson HA, Bird D. Premorbid functioning in adolescent onset bipolarI disorder: a preliminary report from an ongoing study. J Affect Disord 1998;51:137e44.

Lebowitz BK, Shear PK, Steed MA, Strakowski SM. Verbal fluency in mania: rela-tionship to number of manic episodes. Neuropsychiatry Neuropsychol BehavNeurol 2001;14:177e82.

Lee J, Altshuler L, Glahn DC, Miklowitz DJ, Ochsner K, Green MF. Social andnonsocial cognition in bipolar disorder and schizophrenia: relative levels ofimpairment. Am J Psychiatry 2013a;170:334e41.

Lee RSC, Hermens DF, Porter MA, Redoblado-Hodge MA. A meta-analysis ofcognitive deficits in first-episode Major Depressive Disorder. J Affect Disord2012;140:113e24.

Lee RSC, Redoblado Hodge MA, Naismith SL, Hermens DF, Porter MA, Hickie IB.Cognitive remediation improves memory and psychosocial functioning in first-episode psychiatric out-patients. Psychol Med 2013b;43:1161e73.

Lee W, Bindman J, Ford T, Glozier N, Moran P, Stewart R, et al. Bias in psychiatriccase-control studies: literature survey. Br J Psychiatry 2007;190:204e9.

Lewandowski KE, Cohen BM, Ongur D. Evolution of neuropsychological dysfunctionduring the course of schizophrenia and bipolar disorder. Psychol Med 2011;41:225e41.

Lezak MD, Howieson DB, Bigler ED, Tranel D. Neuropsychological assessment. 5thed. New York: Oxford University Press; 2012.

Lin A, Yung AR, Nelson B, Brewer WJ, Riley R, Simmons M, et al. Neurocognitivepredictors of transition to psychosis: medium- to long-term findings from asample at ultra-high risk for psychosis. Psychol Med 2013:1e12.




http://refhub.elsevier.com/S0022-3956(14)00194-0/sref1


























http://dx.doi.org/10.1111/bdi.12099







































































http://dx.doi.org/10.1016/j.psychres.2013.05.001

http://dx.doi.org/10.1016/j.psychres.2013.05.001





























http://dx.doi.org/10.1001/jamapsychiatry.2013.155







































Lisy ME, Jarvis KB, DelBello MP, Mills NP, Weber WA, Fleck D, et al. Progressiveneurostructural changes in adolescent and adult patients with bipolar disorder.Bipolar Disord 2011;13:396e405.

L�opez-Jaramillo C, Lopera-V�asquez J, Gallo A, Ospina-Duque J, Bell V, Torrent C, et al.Effects of recurrence on the cognitive performance of patients with bipolar Idisorder: implications for relapse prevention and treatment adherence. BipolarDisord 2010;12:557e67.

Lyoo IK, Sung YH, Dager SR, Friedman SD, Lee JY, Kim SJ, et al. Regional cerebralcortical thinning in bipolar disorder. Bipolar Disord 2006;8:65e74.

Mann-Wrobel MC, Carreno JT, Dickinson D. Meta-analysis of neuropsychologicalfunctioning in euthymic bipolar disorder: an update and investigation ofmoderator variables. Bipolar Disord 2011;13:334e42.

Martinez-Aran A, Torrent C, Tabares-Seisdedos R, Salamero M, Daban C, Balanza-Martinez V, et al. Neurocognitive impairment in bipolar patients with andwithout history of psychosis. J Clin Psychiatry 2008;69:233e9.

Martinez-Aran A, Vieta E, Reinares M, Colom F, Torrent C, Sanchez-Moreno J, et al.Cognitive function across manic or hypomanic, depressed, and euthymic statesin bipolar disorder. Am J Psychiatry 2004;161:262e70.

Maziade M, Rouleau N, M�erette C, Cellard C, Battaglia M, Marino C, et al. Verbal andvisual memory impairments among young offspring and healthy adult relativesof patients with schizophrenia and bipolar disorder: selective generationalpatterns indicate different developmental trajectories. Schizophr Bull 2011;37:1218e28.

Merikangas KR, Akiskal HS, Angst J, Greenberg PE, Hirschfeld RM, Petukhova M,et al. Lifetime and 12-month prevalence of bipolar spectrum disorder in theNational Comorbidity Survey replication. Arch Gen Psychiatry 2007;64:543e52.

Mesholam-Gately RI, Giuliano AJ, Goff KP, Faraone SV, Seidman LJ. Neurocognitionin first-episode schizophrenia: a meta-analytic review. Neuropsychology2009;23:315e36.

Meyer SE, Carlson GA, Wiggs EA, Martinez PE, Ronsaville DS, Klimes-Dougan B, et al.A prospective study of the association among impaired executive functioning,childhood attentional problems, and the development of bipolar disorder. DevPsychopathol 2004;16:461e76.

Murphy FC, Sahakian BJ. Neuropsychology of bipolar disorder. Br J Psychiatry2001;178:S120e7.

Murray RM, Sham P, Van Os J, Zanelli J, Cannon M, McDonald C. A developmentalmodel for similarities and dissimilarities between schizophrenia and bipolardisorder. Schizophr Res 2004;71:405e16.

Naismith SL, Redoblado-Hodge MA, Lewis SJG, Scott EM, Hickie IB. Cognitivetraining in affective disorders improves memory: a preliminary study using theNEAR approach. J Affect Disord 2010;121:258e62.

Nehra R, Chakrabarti S, Pradhan BK, Khehra N. Comparison of cognitive functionsbetween first- and multi-episode bipolar affective disorders. J Affect Disord2006;93:185e92.

Olley A, Malhi GS, Mitchell PB, Batchelor J, Lagopoulos J, Austin MP. When euthymiais just not good enough: the neuropsychology of bipolar disorder. J Nerv MentDis 2005;193:323e30.

Pachet AK, Wisniewski AM. The effects of lithium on cognition: an updated review.Psychopharmacology 2003;170:225e34.

Pavuluri MN, Birmaher B, Naylor MW. Pediatric bipolar disorder: a review of thePast 10 Years. J Am Acad Child Adolesc Psychiatry 2005;44:846e71.

Quackenbush D, Kutcher S, Robertson HA, Boulos C, Chaban P. Premorbid andpostmorbid school functioning in bipolar adolescents: description and sug-gested academic interventions. Can J Psychiatry 1996;41:16e22.

Ratheesh A, Lin A, Nelson B, Wood SJ, Brewer W, Betts J, et al. Neurocognitivefunctioning in the prodrome of mania-an exploratory study. J Affect Disord2013;147:441e5.

Reichenberg A, Weiser M, Rabinowitz J, Caspi A, Schmeidler J, Mark M, et al.A population-based cohort study of premorbid intellectual, language, andbehavioral functioning in patients with schizophrenia, schizoaffective disorder,and nonpsychotic bipolar disorder. Am J Psychiatry 2002;159:2027e35.


Robbins TW, Gillan CM, Smith DG, de Wit S, Ersche KD. Neurocognitive endophe-notypes of impulsivity and compulsivity: towards dimensional psychiatry. TICS2012;16:81e91.

Robinson LJ, Ferrier IN. Evolution of cognitive impairment in bipolar disorder: asystematic review of cross-sectional evidence. Bipolar Disord 2006;8:103e16.

Rosenthal R. The file drawer problem and tolerance for null results. Psychol Bull1979;86:638e41.

Samam�e C. Social cognition throughout the three phases of bipolar disorder: astate-of-the-art overview. Psychiatry Res 2013;210:1275e86.

Schwartz S, Susser E. The use of well controls: an unhealthy practice in psychiatricresearch. Psychol Med 2011;41:1127e31.

Scott J, Paykel E, Morriss R, Bentall R, Kinderman P, Johnson T, et al. Cognitive-behavioural therapy for severe and recurrent bipolar disorders: randomisedcontrolled trial. Br J Psychiatry 2006;188:313e20.

Simonsen C, Sundet K, Vaskinn A, Birkenaes AB, Engh JA, Faerden A, et al. Neuro-cognitive dysfunction in bipolar and schizophrenia spectrum disorders dependson history of psychosis rather than diagnostic group. Schizophr Bull 2011;37:73e83.

Strakowski SM, Adler CM, Cerullo M, Eliassen JC, Lamy M, Fleck DE, et al. Magneticresonance imaging brain activation in first-episode bipolar mania during aresponse inhibition task. Early Interv Psychiatry 2008;2:225e33.

Strakowski SM, DelBello MP, Zimmerman ME, Getz GE, Mills NP, Ret J, et al. Ven-tricular and periventricular structural volumes in first- versus multiple-episodebipolar disorder. Am J Psychiatry 2002;159:1841e7.

Strauss E, Sherman EMS, Spreen O. A Compendium of neuropsychological tests:administration, norms, and commentary. 3rd ed. New York: Oxford UniversityPress; 2006.

Suominen K, Mantere O, Valtonen H, Arvilommi P, Leppamaki S, Paunio T, et al.Early age at onset of bipolar disorder is associated with more severe clinicalfeatures but delayed treatment seeking. Bipolar Disord 2007;9:698e705.

Tiihonen J, Haukka J, Henriksson M, Cannon M, Kieseppa T, Laaksonen I, et al. Pre-morbid intellectual functioning in bipolar disorder and schizophrenia: resultsfrom a cohort study of male conscripts. Am J Psychiatry 2005;162:1904e10.

Tohen M, Zarate Jr CA, Hennen J, Khalsa HM, Strakowski SM, Gebre-Medhin P, et al.The McLean-Harvard first-episode mania study: prediction of recovery and firstrecurrence. Am J Psychiatry 2003;160:2099e107.

Torrent C, Del Mar Bonnin C, Martinez-Aran A, Valle J, Amann BL, Gonzalez-Pinto A,et al. Efficacy of functional remediation in bipolar disorder: a MulticenterRandomized controlled study. Am J Psychiatry 2013;170:852e9.

Torres IJ, Boudreau VG, Yatham LN. Neuropsychological functioning in euthymicbipolar disorder: a meta-analysis. Acta Psychiatr Scand Suppl 2007;434:17e26.

Torres IJ, DeFreitas VG, DeFreitas CM, Kauer-Sant'Anna M, Bond DJ, Honer WG, et al.Neurocognitive functioning in patients with bipolar I disorder recently recov-ered from a first manic episode. J Clin Psychiatry 2010;71:1234e42.

Toulopoulou T, Quraishi S, McDonald C, Murray RM. The Maudsley Family Study:premorbid and current general intellectual function levels in familial bipolar Idisorder and schizophrenia. J Clin Exp Neuropsychol 2006;28:243e59.

Townsend JD, Bookheimer SY, Foland-Ross LC, Moody TD, Eisenberger NI, Fischer JS,et al. Deficits in inferior frontal cortex activation in euthymic bipolar disorderpatients during a response inhibition task. Bipolar Disord 2012;14:442e50.

Vita A, De Peri L, Sacchetti E. Gray matter, white matter, brain, and intracranialvolumes in first-episode bipolar disorder: a meta-analysis of magnetic reso-nance imaging studies. Bipolar Disord 2009;11:807e14.

Wittchen H-U, Mühlig S, Pezawas L. Natural course and burden of bipolar disorders.Int J Neuropsychopharmacol 2003;6:145e54.

Zanelli J, Reichenberg A, Morgan K, Fearon P, Kravariti E, Dazzan P, et al. Specific andgeneralized neuropsychological deficits: a comparison of patients with variousfirst-episode psychosis presentations. Am J Psychiatry 2010;167:78e85.

Zeggini E, Ioannidis JP. Meta-analysis in genome-wide association studies. Phar-macogenomics 2009;10(2):191e201.








































































































































































a meta-analysis of neuropsychological functioning in first-episode bipolar disorders

Documents