A large European, multicenter, multinational validation study of the Brief Negative Symptom Scale

(1)

Article

Reference

A large European, multicenter, multinational validation study of the Brief Negative Symptom Scale

MUCCI, Armida, et al.

Abstract

Negative symptoms represent an unmet need of treatment in schizophrenia. Although a consensus exists on negative symptom construct, and second generation assessment instruments reflecting the consensus are available, studies still rely upon old assessment instruments, that do not reflect recent conceptualizations and might limit progress in the search for effective treatments. This is often the case in the European context, where one of the challenges encountered in designing large studies is the availability of validated instruments in the many languages of the continent. To address this challenge and promote sound research on negative symptoms in Europe, the ECNP Schizophrenia Network coordinated a large multicenter, multinational validation study of the Brief Negative Symptom Scale (BNSS). Clinically-stable subjects with schizophrenia (SCZ, N = 249) were recruited from 10 European Countries. Apart from BNSS, subjects were administered the Positive and Negative Syndrome Scale (PANSS) and standardized instruments for depression, extrapyramidal symptoms and psychosocial functioning. Results showed an excellent [...]

MUCCI, Armida, et al . A large European, multicenter, multinational validation study of the Brief Negative Symptom Scale. European Neuropsychopharmacology , 2019, vol. 29, no. 8, p.

947-959

DOI : 10.1016/j.euroneuro.2019.05.006 PMID : 31255394

Available at:

http://archive-ouverte.unige.ch/unige:132683

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

www.elsevier.com/locate/euroneuro

A large European, multicenter,

multinational validation study of the Brief Negative Symptom Scale

Armida Mucci

^a^,^∗

, Annarita Vignapiano

^a

, István Bitter

^b

,

Stephen F. Austin

^c

, Camille Delouche

^d^,^e^,^f

, Sonia Dollfus

^d^,^e^,^f

, Andreas Erfurth

^g

, W. Wolfgang Fleischhacker

^h

,

Giulia M. Giordano

^a

, Igor Gladyshev

ⁱ

, Birte Glenthøj

^j^,^k

,

Karoline Gütter

^l

, Alex Hofer

^h

, Jan Hube ˇ nák

^m

, Stefan Kaiser

ⁿ

, Jan Libiger

^m

, Ingrid Melle

^o

, Mette Ø. Nielsen

^j^,^k

, Oleg Papsuev

ⁱ

, Janusz K. Rybakowski

^p

, Gabriele Sachs

^q

, Alp Üçok

^r

,

Pawel Wojciak

^p

, Silvana Galderisi

^a

aDepartmentofPsychiatry,UniversityofCampania“LuigiVanvitelli”,LargoMadonnadelleGrazie, 80138Naples,Italy

bDepartmentofPsychiatryandPsychotherapy,SemmelweisUniversity,Budapest,Hungary

cPsychiatricResearchUnit,PsychiatryRegionZealand,Hillerød,Denmark

dServicedePsychiatrie,CHUdeCaen,Caen,France

eUFRdeMédecine,UNICAEN,NormandieUniversité,Caen,France

fISTS,UNICAEN,NormandieUniversité,Caen,France

g6thPsychiatricDepartment,Otto-Wagner-Spital,Vienna,Austria

hDepartmentofPsychiatry,PsychotherapyandPsychosomatics,DivisionofPsychiatryI,Medical UniversityInnsbruck,Innsbruck,Austria

iDepartmentofSocio-clinicalandBiologicalResearchofPsychoticspectrumdisorders,MoscowResearch InstituteofPsychiatry,Moscow,Russia

jCenterforNeuropsychiatricSchizophreniaResearch(CNSR)andCenterforClinicalInterventionand NeuropsychiatricSchizophreniaResearch(CINS),MentalHealthCenterGlostrup,Glostrup,Denmark

kDepartmentofClinicalMedicine,FacultyofHealthandMedicalSciences,UniversityofCopenhagen, Copenhagen,Denmark

lDepartmentofPsychiatryandPsychotherapy,PsychiatricHospital,UniversityofZurich,Zurich, Switzerland

mPsychiatricDepartment,CharlesUniversityMedicalSchoolandFacultyHospitalHradecKrálové, HradecKrálové,Czechia

∗Correspondingauthor.

E-mailaddress:armida.mucci@unicampania.it(A.Mucci).

https://doi.org/10.1016/j.euroneuro.2019.05.006

(http://creativecommons.org/licenses/by-nc-nd/4.0/)

(3)

nAdultPsychiatryDivision,DepartmentofPsychiatry,UniversityofGenevaHospitals,Geneva, Switzerland

oNORMENTCentre,InstituteofClinicalPsychiatry,UniversityofOsloandOsloUniversityHospital,Oslo, Norway

pDepartmentofAdultPsychiatry,PoznanUniversityofMedicalSciences,Poznan,Poland

qDepartmentofPsychiatryandPsychotherapy,MedicalUniversityofVienna,Vienna,Austria

rIstanbulFacultyofMedicine,PsychoticDisordersResearchProgram,IstanbulUniversity,Istanbul, Turkey

Received 13February2019;receivedinrevisedform11April2019;accepted29May2019

KEYWORDS Negativesymptoms;

Avolition;

Functionaloutcome;

BNSS;

PANSS-negative subscale

Abstract

Negativesymptomsrepresentanunmetneedoftreatmentinschizophrenia.Althoughacon- sensusexistsonnegativesymptomconstruct,andsecondgenerationassessmentinstruments reflectingtheconsensusareavailable,studiesstillrelyuponoldassessmentinstruments,that donotreflectrecentconceptualizationsandmight limitprogressinthesearchforeffective treatments.ThisisoftenthecaseintheEuropeancontext,whereoneofthechallengesen- counteredindesigninglargestudiesistheavailabilityofvalidatedinstruments inthemany languagesofthecontinent.Toaddressthischallengeandpromotesound researchonnega- tivesymptomsinEurope,theECNPSchizophreniaNetworkcoordinatedalargemulticenter, multinational validation studyoftheBrief NegativeSymptomScale(BNSS). Clinically-stable subjectswithschizophrenia(SCZ,N=249)wererecruitedfrom10EuropeanCountries.Apart fromBNSS,subjectswereadministeredthePositiveandNegativeSyndromeScale(PANSS)and standardizedinstrumentsfordepression,extrapyramidalsymptomsandpsychosocialfunction- ing.Resultsshowedanexcellentinternalconsistency,convergentanddiscriminantvalidityof BNSSandreplicateda5factor-model.Alargernumberofsubjectswithpredominantnegative symptoms,i.e.thetargetpopulationforclinicaltrials,wasidentifiedbyusingtheBNSScom- paredtothePANSS.RegressionanalysisshowedthatBNSS-avolition,akeynegativesymptom poorlyassessedbyPANSS,explained23.9%ofpsychosocialfunctioning,whilenocombinationof thePANSScorenegativesymptomsshowedthesameimpactonfunctioning.Thestudydemon- stratedthatBNSShassubstantialadvantageswithrespecttoPANSSfortheidentificationofthe avolitiondomainandsubjectswithpredominantnegativesymptoms.

This is an open access article under the CC BY-NC-ND license.

(http://creativecommons.org/licenses/by-nc-nd/4.0/)

1. Introduction

Despiteseveral improvementsin diagnosisand treatment, deﬁcitsinreallifefunctioning,suchasindependentliving, socialfunctioningandremunerativeemployment,persistin mostpeoplewithschizophrenia(Fleischhackeretal.,2014;

Velthorstetal.,2017).

Negative symptoms represent one the most important contributorsto disability(Erguland Ucok, 2015; Galderisi et al., 2014, 2017, 2018a; Marder and Galderisi, 2017;

Quinlanetal.,2014;Straussetal.,2012b;Venturaetal., 2015). Even low levels of negative symptoms were found topredictreal-lifefunctionalimpairment(Strassnigetal., 2018).

Primarynegativesymptomsseemtoberesistanttomost psychopharmacologicaltreatmentsandrehabilitationinter- ventions(Keshavanetal.,2017;Sarkaretal.,2015;Talpos, 2017;Veerman etal.,2017).Therefore,todate, negative symptomsinschizophreniaremainasigniﬁcantunmetneed (Galderisietal.,2018a;MarderandGalderisi,2017).

The heterogeneity of negative symptoms and the lim- itations in their deﬁnition and assessment might have

contributedtothelackofsigniﬁcantprogressindeveloping effectivetherapeuticinterventions(Galderisietal.,2018a).

In2005,theNationalInstituteof MentalHealth(NIMH)- MATRICS consensus statement on negative symptoms de- finedtheconstructstoberegardedaspartofthenegative symptom dimension (Kirkpatrick et al., 2006; Kirkpatrick andFischer,2006).Theconsensusstatementidentifiedfive maindomainsofthenegativesymptoms:anhedonia,avolition,bluntedaffect,alogia,and asociality(Blanchardand Cohen,2006; Horanetal., 2011;Kirkpatricketal., 2006, 2011;Kringetal.,2013;Millanetal.,2014).Subsequently, several factor analysis studies showed that the five con- structslistedaboveclusteredintotwoindependentdimen- sions: avolition/apathy (including avolition, asociality and anhedonia)andexpressivedeficit(includingbluntedaffect andalogia)(Bischofetal.,2016;Jangetal.,2016;Kimhy etal.,2006;Muccietal.,2015;NakayaandOhmori,2008;

Straussetal.,2013).

The two-factor solution hasbeen identiﬁed in subjects withprimarynegativesymptoms(deﬁcitschizophrenia)and in those with a combination of primary and secondary

(4)

negativesymptoms,especiallyafterexcludingitemsunre- lated tonegativesymptoms,suchasinattentivenessorin- appropriate affect (Blanchard and Cohen, 2006; Galderisi etal.,2013a,2018a;Kimhyetal.,2006;Kirkpatricketal., 2011; Kring et al., 2013; Mucci et al., 2015; Nakaya and Ohmori,2008;Straussetal.,2013).

The two dimensions are independent of medication (Kelleyetal.,1999; Tremeauetal.,2008)and havebeen cross-culturallyvalidated(Bischofetal.,2016;Maneetal., 2014;Muccietal.,2015;Straussetal.,2012a).

However,someﬁndingshavequestionedthetwofactors solution (Ahmed etal.,2018; Garcia-Portilla etal.,2015;

Strauss et al.,2018a, 2018b)in particular, alarge, multi- centricstudyinsubjectswithschizophrenia,usingtheBNSS toassessnegativesymptoms,suggestedthatafive-domain construct might have better fit statistics (Ahmed et al., 2018; Strauss et al., 2018a, 2018b). These latter findings awaitindependentreplications.

It is widelyrecognizedthat primarynegativesymptoms are related to the core pathophysiology of schizophrenia andrepresentthetrueunmetneedofschizophreniatreat- ment. Secondary negativesymptoms,due toother identi- ﬁable causes such as positive symptoms, depression, extrapyramidal side effects of antipsychotic drugs or isola- tion, should thus be excluded in studies aiming to inves- tigate the pathophysiology of negative symptom domains or totest newtreatment effects (Carpenter etal.,1985;

Galderisietal.,2018a;Kirschneretal.,2017;Muccietal., 2017;Treenetal.,2019).

Forclinicaltrialstargetingnegativesymptomsregulatory agenciesrequiretheexclusionofmainconfoundingfactors, suchashighseverityofpositiveanddepressivesymptoms, aswell asparkinsonism (EMA,2012; Marderetal.,2011).

Two main concepts wereproposedtoaddress the issueof pseudospeciﬁceffectsonnegativesymptomsinclinicaltri- als(Buchanan,2007;Muccietal.,2017):the“predominant negativesymptoms” (withnegativesymptomsmoresevere than positive ones combinedwith a thresholdfor depres- sivesymptomsorparkinsonism),andthe“persistentnega- tivesymptoms” (inwhichamaximumthresholdforpositive symptoms,depressionandparkinsonismonvalidatedscales isdeﬁned).Thelatterconceptalsorequiresastabilityover 6monthsforthenegativesymptomsofmoderateseverity.

The advantages of newly developed scales over those largelyusedinthepast,suchasthePositiveandNegative SyndromeScale(PANSS)– Negativesubscale,havenotbeen systematicallyinvestigated.

Theavailabilityofvalidatedstateoftheartinstruments isofcrucialimportanceinlargemulticenter,multinational studies,whichrepresentthestandardcontextofclinicaltri- als. The validation shouldalso demonstrate that the new scaleidentifiesacomparableoralargernumberofsubjects with negative symptoms of moderate severity in samples withandwithoutconfoundingfactors(i.e.,severepositive symptoms,depressionorparkinsonism).Finally,alargerim- pactonfunctionaloutcomeofthenegativesymptomsiden- tifiedbythenewscale,incomparisonwiththoseidentified byestablishedscales,shouldbedemonstrated.

The purposes of our study were: (1) to validate the BNSS within a large European multicenter, multinational and multilingual study carried out in stabilized subjects with schizophrenia, demonstrating the convergent and

discriminant validity with respect to established rating scales to assess negative, positive and depressive symptoms,namelythePANSSandtheCalgaryDepressionRating Scale for schizophrenia(CDSS); (2) toinvestigate the frequency of negative symptoms in subjects with and with- outconfoundingfactors;(3) toevaluate whetherthenew scaleidentiﬁesmoresubjectswithcorenegativesymptoms in both groups and(4) toinvestigate the impactonfunc- tionaloutcomeofcorenegativesymptomsasidentiﬁedby the BNSS or thePANSS, controlling for majorconfounding factors.

2. Experimentalprocedures 2.1. Participants

The European College of Neuropsychopharmacology(ECNP) Net- workonSchizophreniaStudyontheAssessmentofNegativeSymp- toms(ENSANeS)wascarriedoutwithintheactivitiesoftheECNP SchizophreniaNetwork.

Studyparticipantswererecruitedamongsubjectsattendingthe outpatientandinpatientunitsofthePsychiatricDepartmentsof12 Europeancenters(seeappendixforthelistofthecenters).

RecruitmentwascarriedoutfromOctober31,2016toJuly15, 2017.

Inclusion criteriawerea diagnosis ofschizophrenia according toDSM-IV, conﬁrmedbytheMiniNeuropsychiatric Interview-Plus (MINI-Plus),andanagebetween18and65years.

Exclusioncriteriawere:(a)treatmentmodiﬁcationsand/orhos- pitalizationduetosymptomexacerbationinthelastthreemonths;

(b)ahistoryofmoderatetosevereintellectualdisabilityorofneu- rologicaldiseases;(c)ahistoryofalcoholand/orsubstanceabuse inthelastsixmonths;(d)currentpregnancy;and(e)inabilityto provideaninformedconsent.

ThelocalEthicsCommitteeoftheinvolvedinstitutionsapproved thestudy,andthestudywasperformedinaccordancewiththeeth- icalstandardslaiddowninthe1964DeclarationofHelsinki.

Allparticipantssignedawritteninformedconsenttoparticipate inthestudy,afterreceiving adetailed explanationofthestudy proceduresandgoals.

2.2. Assessment

2.2.1. Demographicandclinicalinformation

Anad-hocformwasﬁlledinwithdataonage,gender,race,education(numberofyearsatschool),paternalandmaternaleduca- tion(numberofyearsatschool), aswellasclinical information, suchasfamilyhistoryofschizophrenia-spectrumdisordersorma- jormooddisorders(bipolarandmajordepressivedisorderswithor withoutpsychoticfeatures),durationofillness,numberofhospital- ization,typeanddosageofantipsychoticmedicationsandconcomi- tantmedications,usingalltheavailablesourcesofinformation.

Thedailyantipsychoticdosewasconvertedtoolanzapineequiv- alentsdosebasedondeﬁneddailydoses(DDDs),accordingtothe WorldHealthOrganization’sCollaborativeCenterforDrugStatistics Methodology(Leuchtetal.,2016).

2.2.2. Psychopathologicalassessment

TheBriefNegativeSymptomScale(BNSS;Kirkpatricketal.,2011), developed after the MATRICS consensus initiative (Kirkpatrick etal.,2006),wasusedtoassessnegativesymptomdomains.The BNSSisasemi-structuredinterviewwith13items,organizedinto 5 negative symptom subscales (anhedonia, asociality, avolition, bluntedaffectandalogia),plusandadditionalsubscaleincluding

(5)

oneitemassessingthelackofthenormalexperienceofdistressing, unpleasant emotions,which wasincludedtohelp thedistinction ofprimaryfromsecondarynegativesymptoms(Kirkpatricketal., 2011;Muccietal.,2015).

Forallitemshigherscoresare associatedwithgreaterimpair- ment/presenceofsymptoms.Ascaletotalscoreiscalculatedby summingupthe13individualitems;subscalescoresarecalculated bysumminguptheindividualitemswithineachsubscale.Thedis- tresssubscale,thatquantiﬁes theabsenceofdistress,istreated in the same manner as the other subscales. It is not included inthecorenegativesymptom domains(Kirkpatricket al.,2006) and does notconsistentlyloadon thenegativesymptom factors (Kirkpatricketal.,2006,2011;Muccietal.,2015;Ahmedetal., 2018).

Allnon-EnglishversionsoftheBNSSweredevelopedusingthe translation–backtranslationmethod.Thetranslatedversionswere back-translatedintoEnglishbyanativespeakerofthesamelan- guageinwhichthescalewasdeveloped.Theback-translatedver- sions were reviewed and approved byBrian Kirkpatrick or Gre- gory Strauss who participated in the development of the scale (Kirkpatricketal.,2011).

ThePositive andNegative SyndromeScale(PANSS; Kayet al., 1987) is one ofthemost widelyusedinstruments toinvestigate psychopathology inschizophrenia. Thescale includes a subscale forpositivesymptoms;asubscalefornegativesymptoms,thatin- cludes 2symptoms recently excluded,i.e.,difficulty inabstract thinkingandstereotypedthinking,asbothofthemarepartofthe cognitivedimension;andageneralpsychopathologyscalewith16 items.According tothe five-factormodel proposedbyWallwork et al. (2012), we computed the positivedimension bysumming upthescoresontheitems“Delusions” (P1),“Hallucinatorybehav- ior” (P3),“Grandiosity” (P5)and“Unusualthoughtcontent” (G9), andthedisorganizationdimensionbysummingupthethreePANSS items: “Conceptualdisorganization” (P2), “Difficulty in abstract thinking” (N5),and“Poorattention” (G11).

To compare the ability of BNSS and PANSS-negative subscale (PANSS-NEG)toidentifycorenegativesymptomsandassesstheim- pactof identiﬁed negativesymptoms on functional outcome we usedthe 5 subscalesof BNSS(alogia,anhedonia, avolition,aso- cialityand bluntedaffect)and the5corenegativesymptomsof the PANSS(bluntedaffect, emotional withdrawal, poor rapport, passive-apatheticsocial withdrawal,and lackofspontaneityand ﬂowofconversation).

TheCDSS(Addingtonetal.,1993)wasusedtoassessdepressive symptoms.Ratings>6onthetotalscoreindicateclinicallysignif- icantdepression.Non-EnglishversionsoftheCDSSareavailableon theofﬁcialwebsite(http://www.ucalgary.ca/cdss/).

TheSHSR(Gerlachetal.,1993)wasusedtoquantifythepres- enceofextrapyramidalsymptoms.Clinicallysigniﬁcantparkinson- ism,whichmightconfoundtheassessmentofnegativesymptoms, isdeﬁnedbya“mild” (2)ratingonatleastthreeitems,ora“mild”

ratingfortremororrigidity plusa“mild” ratingon atleastan- otheritem,ora“mild-moderate” (3ormore)ratingonatleastone item.

Bradykinesiaand reductionoffacialexpressionhavenotbeen evaluatedasconfoundingfactorssincetheycouldbeanaspectof negativesymptoms(WaltherandStrik,2012).

Raterswereresearch staff membersofthe participatingcen- ters.Allratershadextensivepriorexperienceinconductingclin- icalinterviews,buttheinter-raterreliabilityacrossthesiteswas notevaluatedforthepresentstudy.

2.2.3. Functionaloutcome

PersonalandSocialPerformance(PSP)scale(Morosinietal.,2000) wasusedtoevaluatethefunctionaloutcome.ThePSPisa0–100 single-item rating scale, in which higher scores indicate better functioning.

2.3. Assessmentoffrequencyofatleastmoderate severitynegativesymptoms

AstheEuropeanMedicinesAgency(EMA)guidelinesonclinicaltrial designfornegativesymptomsrequiretheinclusionofsubjectswith atleastmoderateseverity negativesymptoms,deﬁnedonanac- ceptedandvalidatedratingscale,weinvestigated thefrequency ofthese symptoms using both theBNSS and thePANSS. For the BNSSascore≥3(moderate)wasnecessaryoneachofthe5BNSS subscalesforcorenegativesymptoms;forPANSSascore≥4(moderate)forthefollowingcorenegativesymptoms:Blunted Affect (N1),EmotionalWithdrawal(N2),PoorRapport(N3),Passivesocial withdrawal(N4)andLackofSpontaneity(N6).

2.4. Statisticalanalysis

2.4.1. Internalconsistency,convergentanddiscriminant validity

Datadistributionswereexaminedfornormalityand homogeneity ofvariance and presenceof outliers(subjects whose scores ex- ceededthe75thorthe25thpercentileby1.5timestheinterquar- tilerange).

Thedemographic and clinical characteristicsof studypartici- pantsweresummarized as mean ± StandardDeviation (SD)and percentageswhereappropriate.

TheconstructvalidityoftheBNSSwasanalyzedbyassessingits internalconsistencyand convergentand discriminant validity,as reportedinpreviouspapers(Kirkpatricketal.,2011;Muccietal., 2015;Straussetal.,2012a).Internalconsistencyacrossallitemsof theBNSSwasassessedusingtheCronbach’salphavalue.

Convergentvaliditywasevaluatedbyexaminingtheassociation oftheBNSStotalscore withthePANSSnegative subscalescore, whichmeasuresthesameconstruct,andPANSStotalscore,asthe latterincludesthenegativesubscalescoreandthusshouldbeat leastmoderatelycorrelatedwiththeBNSStotal.Discriminantva- liditywasassessedbyexaminingcorrelationsoftheBNSStotalscore withPANSSpositivesubscalescoreandCDSStotalscore.TheSpear- manrhowasusedforcorrelationsasscoredistributionsdeviated fromnormalityforthescales.Withlargesamplesize(N>200),p valuesaregenerallyhighlysignificantevenforverylowcorrelation coefficients(r=0.10,p<0.01);forthisreason,theabsolutevalue ofthecorrelationcoefficientisamoreconservativeestimationof association.Correlationcoefficients(inabsolutevalue)<0.35are generallyconsideredtorepresentloworweakcorrelations,those from0.36to0.67modesttomoderatecorrelations,andthosefrom 0.68to1.0strongcorrelations(Evans,1996).

2.4.2. FactorstructureoftheBNSS

ThefactorstructureoftheBNSSwasexaminedusingaconfirma- toryfactoranalysis(CFA),inordertoevaluatethefitof4modelsof thelatentstructureofnegativesymptoms(Figs.S1–S4),according toStraussetal.(2018a)andAhmedetal.(2018).Thefirstmodel (unidimensionalmodel)evaluatedwhetherall fivedomains(avolition,anhedonia,asociality,blunted affectand alogia)reflected asinglenegativesymptomstructure(Fig. S1);thesecondmodel (Fig.S2)consideredthetwonegativesymptomdimensions:moti- vationandpleasure(MAP)and expressivedeficit(EXP);thethird modeltestedthe5-factormodelofnegativesymptoms(Fig.S3);

finallythefourthmodel(hierarchicalmodel)(Fig.S4)wasdesigned with 5 first-order factors (five negative symptom domains) and 2second-orderfactors(MAPandEXP).InalltheCFAmodelsthe item“Lackofnormal distress” wasnotincludedbecause it was notpartoftheNIMHconsensusconferencedomains,andpriorex- ploratoryfactor analysisstudiesreportedlow communalitiesfor thisitem(Straussetal.2012a;Muccietal.,2015).

(6)

Toassesstheglobalfit,thefollowingindiceswereapplied:χ2 value,thecomparativefitindex(CFI),theTuckerLewisindex(TLI), therootmeansquareerrorofapproximation(RMSEA),theAkaike informationcriterion(AIC)andthesamplesize-adjustedBayesian informationcriterion(BIC).Theinformationcriteria,AICandBIC, arerelativefitindicesofmodelparsimonythattakeintoaccount modelcomplexitybasedondegreesoffreedom.Lowervaluesindi- catebettermodelfit.Agoodfitincludedaχ2valuenotstatistically significant,CFIandTLIvaluesofatleast0.95,RMSEAnogreater than0.08andlowerBICandAICvalues.

2.4.3. Frequencyofnegativesymptomsofatleastmoderate severity

Frequenciesofatleastonenegativesymptomofmoderatesever- ity, assessed usingthe PANSSand theBNSS, werecalculatedfor eachBNSSsubscaleassessingcorenegativesymptoms(i.e.,exclud- ing“Distress”)andPANSSﬁvecorenegativesymptomitemsinthe wholeSCZsampleandinmaleandfemalesubgroups.Furthermore, subjectsweredividedinthosewithandwithoutconfoundingfac- torsbasedonthefollowingthresholds:positivesymptoms(P1,P3, P5,P6,G9)anddisorganization(P2)score>4onthePANSS,CDSS Totalscore>6(Addingtonetal.,1993)andpresenceofparkinson- ism evaluatedwithSHRS asoutlinedabove.Inthesegroups,the frequencyofatleast1,2and3negativesymptomsofmoderate severitywasassessed.

2.4.4. Associationofcorenegativesymptomswithfunctional outcome

Theassociationofcorenegativesymptomswithfunctionaloutcome wasstudiedusingmultiplestepwiselinearregression.

Separatestepwisemultipleregressionswereconductedforneg- ativesymptomsassessed usingthe5 BNSSdomainsorthe5core PANSSitemsaspreviouslydeﬁned.

Foreachregressionanalysis,thePSPtotalscorewasthedepen- dentvariable,whiletheindependentpredictorswereage,gender, education, durationofillness,olanzapineequivalentdose,PANSS disorganizationandpositivedimensions,depression,SHRSParkin- sonism, andscores onBNSSsubscales(anhedonia,avolition,asociality,bluntedaffectandalogia)orscoresonPANSScorenegative items(BluntedAffect,EmotionalWithdrawal,PoorRapport,Passive withdrawalandLackofSpontaneity).

Statisticalsigniﬁcancelevelwassetatp≤0.05foralltests.

Conﬁrmatory factor analyses(maximum likelihood)were esti- matedusingAMOS21.0.Allotheranalyseswerecarriedoutusing SPSS22(IBMCorp,Armonk,NY).

3. Results

3.1. Sociodemographicandclinicalvariables

Two hundred forty-nine subjects with a DSM-IV diagnosis of schizophrenia were enrolled. Data on demographic and illness related variables are providedin Table 1. En- rolled subjects were predominantly males, with a mean age of37years,withameaneducationlevelof 13years, all in a chronic phase of the illness. Almost all subjects withschizophreniaweretreatedwithantipsychotics(94%), mostlywithsecond-generationantipsychotics(Table1).

3.2. BNSSinternalconsistency,constructvalidity andfactormodels

Two hundred thirty-six subjects withschizophrenia had a completedatasetwithrespecttotheconsideredmeasures

Table1 Demographicandillness-relatedvariablesofthe studysample(SCZ=249).

N %

Gender§(%males) 158 63.5

Caucasian§(%) 202 81.1

Antipsychotictreatment(%)

Firstgeneration 14 5.6

Secondgeneration 193 77.5

Both 27 10.8

None 13 5.2

N/A 4 0.8

N Mean±SD

Age(years) 249 37.3±11.3

Education(completedyears) 249 12.9±3.2 Durationofillness(years) 249 12.4±9.1 Olanzapineequivalentdose 232 14.75±9.29

PANSStotal 240 61.21±17.20

PANSSpositive 244 13.1±5.3

PANSSnegative 247 17.54±16.62

PANSSdisorganization 246 6.70±2.83

BNSStotal 246 26.5±15.5

Anhedoniasubscale 248 6.31±4.67 Asocialitysubscale 249 4.20±2.58

Avolitionsubscale 249 4.58±3

Bluntedaffectsubscale 249 6.71±4.43

Alogiasubscale 249 3.20±3.09

CDSStotal 236 3.5±3.6

SHRSParkinsonismtotal 248 3.78±5.0

PSPtotal 234 57.38±15.32

SCZ=subjectswithschizophrenia;PANSS=PositiveandNeg- ative Syndrome Scale; CDSS = Calgary Depression Scale for Schizophrenia;SHRS=St.HansRatingScale;BNSS=BriefNeg- ativeSymptomScale; PSP=Personaland SocialPerformance scale;§ – Datanotavailableforoneoftheparticipantcenters.

usedtoassessBNSSinternalconsistency,constructvalidity anditsfactorstructure.

Cronbach’s alpha, calculated to examine internal consistency, was0.94, indicating high internal consistency of BNSS.

BNSS total score was significantly correlated with the PANSS negativesubscale score (ρ=0.77) and was moder- ately associated with PANSS total score (ρ=0.57) in the whole sample of subjectswith schizophrenia. Convergent validity was assessed also in the subsample without clinically significantparkinsonism and/or depressionincluding 164subjects;wefoundastrongcorrelationwiththePANSS negativesubscalescore(ρ=0.76)andamoderateassocia- tion withPANSS total score (ρ=0.61). These correlations confirm that both scales assess a similar underlying con- structofnegativesymptoms(Table2).

As to the discriminant validity, the BNSS total score showed weak correlations with the PANSS positive subscale(ρ=0.21)andwithCDSStotalscore(ρ=0.27)inthe wholesample. In thesubsamplewithout confoundingfactors,thecorrelationwasweakwiththePANSSpositivesub- scale(ρ=0.23)andextremelyweakwithCDSStotalscore (ρ=0.13).

(7)

Table 2 BNSSconvergent anddiscriminant validity (rho-values) inthemain sample (n =236) andinthesubsample without confoundingfactors§ (n=136).

BNSStotalscore (mainsample)

pvalue BNSStotalscore(subsamplewithout depressionandextrapyramidalsymptoms)

pvalue

Convergentvalidity

PANSSnegativesubscale 0.77^∗∗∗ <0.0001 0.76^∗∗∗ <0.0001

PANSStotalscore 0.57^∗∗ <0.0001 0.61^∗∗ <0.0001

Discriminantvalidity

PANSSpositivesubscale 0.21^∗ <0.001 0.23^∗ 0.003

CDSStotalscore 0.27^∗ <0.0001 0.13 0.08

BNSS: BriefNegativeSymptom Scale;PANSS:Positive andNegativeSyndrome Scale;CDSS: CalgaryDepressionScaleforSchizophrenia.

§ Confoundingfactors=clinicallysigniﬁcantlevelsofdepression(CDSStotal>6),extrapyramidalsymptoms(parkinsonismontheSt.Hans scale)andpositiveordisorganizationsymptoms(>4onthePANSS).

Withlargesamplesize(N>200),pvaluesarehighlysignificantevenforverylowcorrelationcoefficients(r=0.10,p<0.01);theabsolute valueofthecorrelationcoefficientisamoreconservativeestimationofassociation:^∗Weakcorrelation0.20–0.35;^∗∗Moderatecorrelation 0.36–0.67;^∗∗∗Strongcorrelation:0.68–1.0.

Table3 ModelﬁtresultsfromCFAonnegativesymptomsasassessedbytheBriefNegativeSymptomScale(BNSS).

Model Numberofdistinct

parameterstobeestimated

AIC BIC X²value(df) TLI CFI RMSEA

1Factor 36 962.464 966.447 890.5(54) .531 .676 .250

2Factor 37 490.028 494.122 416.0(53) .793 .859 .166

5Factor 46 200.100 205.190 108.1(44) .956 .975 .077

Hierarchical 42 200.083 204.730 116.1(48) .957 .974 .076

CFA=confirmatoryfactoranalysis;AIC=Akaikeinformationcriterion;BIC=Bayesianinformationcriterion;CFI=confirmatoryfitindex;

RMSEA=rootmeansquareerrorofapproximation;TLI=TuckerLewisindex.

Table4 Frequencyofnegativesymptomsofmoderateseverityinthewholestudysample(N=249).

A)BNSS B)PANSS

Negativesymptoms Frequency % Negativesymptoms Frequency %

Atleast1 147 59 Atleast1 128 51.4

Atleast2 107 43 Atleast2 78 31.3

Atleast3 73 29 Atleast3 68 27.3

Anhedonia 79 31.7 N/A

Bluntedaffect 90 36.1 Bluntedaffect(N1) 96 38.6

Avolition 100 40.2 Emotionalwithdrawal(N2) 62 24.9

Asociality 70 28.1 Passivesocialwithdrawal(N4) 64 25.7

Alogia 60 24.1 Lackofspontaneity(N6) 52 20.9

N/A Poorrapport(N3) 31 12.4

BNSS=BriefNegativeSymptomScale;PANSS=PositiveandNegativeSyndromeScale.

ResultsoftheCFAanalysesarereportedinTable3.The 1-factorand2-factormodelsprovidedpoorfit,whilethe5- factormodelandthehierarchicalmodelprovidedthebest fit,withasmalladvantageofthe5-factorfit.

3.3. Frequencyofnegativesymptomsofatleast moderateseverity

Frequency of at least moderate severity (≥3) negative symptomsfortheﬁveBNSSsubscales inthewholesample are presented in Table 4A. One hundred and forty-seven subjects (59%) had a score ≥3 for at least one negative symptom on the BNSS scale. Frequency of the negative

symptomsofmoderateseverity(score≥4),evaluatedusing PANSS ﬁve core items is reported in the Table 4B: 128 subjects (51.4%) showed at least one negative symptom of moderate severity. The frequency of each negative symptom wassystematically higher usingthe BNSS versus thePANSS(Table4AandB).

The frequency of at least moderate severity negative symptomswashigherinmalesthaninfemalesusingeither theBNSSorthePANSS(Tables5Aand5B).

Thefrequencyofnegativesymptomsofmoderatesever- ityin subjectswithandwithoutconfoundingfactorsisre- portedinTable6AandB,respectively.Inbothsubgroups, BNSS identiﬁed more subjects with at least 1 or 2 nega- tivesymptomsthanthePANSS;thedifferencebetweenthe

(8)

Table5A FrequencyofatleastmoderateseveritynegativesymptomsusingtheBriefNegativeSymptomScale(BNSS)inthewhole studysampleandinmaleandfemalesubsamples.

BNSSsubscale Frequencyinthewholesample:N(%) Frequencyinmales(M)andfemales(F):N(%)

Anhedonia 79/249(31.7) M:64/158(40.5)

F:15/91(16.5)

Asociality 70/249(28.1) M:54/158(34.2)

F:16/91(17.6)

Avolition 100/249(40.2) M:70/158(44.3)

F:30/91(33)

Bluntedaffect 90/249(36.1) M:69/158(43.7)

F:21/91(23.1)

Alogia 60/249(24.1) M:45/158(28.5)

F:15/91(16.5)

Table5B FrequencyofthedifferentnegativesymptomswithatleastmoderateseverityusingthePositiveandNegativeSyndrome Scale(PANSS)inthewholestudysampleandmaleandfemalesubsamples.

PANSSitem Frequencyinthewholesample:N(%) Frequencyinmales(M)andfemales(F):N(%)

Bluntedaffect(N1) 96/249(38.6) M:73/158(46.2)

F:23/91(25.3)

Emotionalwithdrawal(N2) 62/249(24.9) M:46/158(29.1)

F:16/91(17.6)

Poorrapport(N3) 31/249(12.4) M:23/158(14.6)

F:8/91(8.8) Passivesocialwithdrawal(N4) 64/249(25.7) M:49/158(31)

F:15/91(16.5)

Lackofspontaneity(N6) 52/249(20.9) M:39/158(24.7)

F:13/91(14.3)

Table6 Frequencyofnegativesymptomsofmoderateseverityinthesubgroupswithandwithoutconfoundingfactorssuch as positivesymptoms,disorganization,depressionandextrapyramidalsymptoms.

A)Subgroupwithconfoundingfactors (N=113)

B)Subgroupwithoutconfounding factors(N=136)

BNSS PANSS BNSS PANSS

N(%) N(%)

Atleast1negativesymptom 79(69.9) 74(65.5) 68(50) 54(39.7)

Atleast2negativesymptoms 58(51.3) 47(41.6) 49(36) 31(22.8)

Atleast3negativesymptoms 44(38.9) 40(35.4) 29(21.3) 28(20.6)

BNSS=BriefNegativeSymptomScale;PANSS=PositiveandNegativeSyndromeScale.

scaleswasnegligiblefortheidentiﬁcationofsubjectswith at least 3negative symptoms(i.e., the most severely af- fectedsubjects).

3.4. Impactofnegativesymptomsonfunctional outcome

Table 7illustrates theresultsof themultiple stepwisere- gressionanalysesonPSPtotalscore.

When the negative symptoms assessed using the ﬁve BNSS subscaleswereentered amongthe independentpre- dictors (Table 7A), the model was signiﬁcant (adjusted R²=0.387,p<0.00001) andaccounted forapproximately 39%ofthevarianceofPSPtotalscore.Lowerfunctionalout- comewaspredictedbyhigheravolitionscores(β=−0.404,

p < 0.00001),which accounted for mostof theexplained PSP variance (29.3/39%), higher disorganization and posi- tivedimensionscores(β=−0.193,p=0.009,β=−0.154, p = 0.026, respectively) and male gender (β=0.134, p =0.015). Theother independentvariables(age,education, duration of illness, olanzapine equivalent dose, the other negative symptoms,depression, SHRSParkinsonism) didnotentertheregressionequation.

Theregressionmodelwasalsosigniﬁcantwhentheneg- ativesymptomswereevaluatedusingPANSSﬁvecoreitems (adjusted R²=0.378, p < 0.00001)and accounted for ap- proximately 38% of the variance of functional outcome (Table7B).Intheregressionanalysis,lowerfunctionalout- come was predicted by higher scores on disorganization dimension (β=−0.238, p = 0.001), which accounted for most of the explained PSP variance (21.4/37.8%), Passive

(9)

Table7 ResultsofmultipleregressionanalysesonPersonalandSocialPerformance(PSP)totalscoreinsubjectswithschizophrenia toinvestigatetheimpactofnegativesymptomsonfunctioning.

A)NegativesymptomdomainsfromtheBNSS

Variablesenteringtheequation Rsquarechange Beta T P

BNSSavolition 0.297 −0.404 −6.378 <0.00001

PANSSdisorganization 0.071 −0.193 −2.628 0.009

Gender 0.017 0.134 2.449 0.015

PANSSpositive 0.015 −0.154 −2.239 0.026

Variablesnotenteringtheequation Rsquarechange Beta T P

BNSSanhedonia NA 0.050 0.639 0.524

BNSSasociality NA −0.008 −0.100 0.920

BNSSbluntedaffect NA −0.115 −1.630 0.105

BNSSalogia NA 0.022 0.325 0.746

Age NA 0.005 0.089 0.930

Education NA 0.087 1.515 0.131

Durationofillness NA 0.016 0.285 0.776

Olanzapineequivalentdose NA −0.027 −0.480 0.632

Depression NA 0.035 0.609 0.543

SHRSParkinsonism NA −0.060 −1.049 0.295

Theregressionmodelwassigniﬁcant(adjustedR²=0.387,p<0.00001)andaccountedforapproximately39%ofthePSPtotal variance.

B)NegativesymptomdomainsfromthePANSS

Variablesenteringtheequation Rsquarechange Beta T P

PANSSdisorganization 0.218 −0.238 −3.245 0.001

Passivesocialwithdrawal(N4) 0.105 −0.232 −3.529 0.001

Poorrapport(N3) 0.023 −0.179 −2.722 0.007

Gender 0.017 0.134 2.398 0.017

PANSSpositive 0.015 −0.156 −2.221 0.027

Variablesnotenteringtheequation Rsquarechange Beta T P

PANSSbluntedaffect(N1) NA −0.024 −0.310 0.756

PANSSemotionalwithdrawal(N2) NA −0.099 −1.138 0.257

PANSSlackofspontaneity(N6) NA 0.094 1.197 0.233

Age NA −0.052 −0.908 0.365

Education NA 0.105 1.788 0.075

Durationofillness NA −0.041 −0.691 0.491

Olanzapineequivalentdose NA −0.057 −1.014 0.312

Depression NA 0.021 0.351 0.726

SHRSParkinsonism NA −0.075 −1.298 0.196

Theregressionmodelwassigniﬁcant(adjustedR²=0.378,p<0.00001)andaccountedforapproximately38%ofthePSPtotal variance.

InA)the5subscalesoftheBriefNegativeSymptomScale(BNSS)wereincludedasindependentpredictors,whileinB)the5corenegative symptomitems ofthePositiveand NegativeSyndrome Scale(PANSS)wereincluded.Thenegativestandardizedbetaforallcontinuous independentpredictorsindicatesthathigherseverityofpsychopathologyisassociatedwithlowerscoresonPSP(i.e.,poorerfunctioning).

MalegenderwasassociatedwithlowerPSPscoresthanfemalegender.

Social Withdrawal (β=−0.232, p=0.001), Poor Rapport (β=−0.179,p=0.007),malegender(β=0.134,p=0.017) andhigherpositivescores(β=−0.156,p=0.027).Thetwo negativesymptomsaccountedforabout12%ofthePSPvari- ance.Theotherindependentvariables(age,education,durationofillness,olanzapineequivalentdose,theotherneg- ativesymptoms,depressionandSHRSParkinsonism)didnot entertheregressionequation.

3.5. Controlanalyses

As control analyses, to evaluate the comparability of BNSS across countries/languages, we assessed the presenceof acenter effecton:(1) thefrequency of negative

symptoms, as evaluated by the BNSS and PANSS; (2) the correlations of the BNSS and PANSS core negative symp- tomtotals;(3)thecorrelationsofthePSPtotalscorewith the BNSS and PANSS core negative symptom total scores.

As reported in Table S1 of the supplemental materials, in all centers the frequency of negative symptoms of at least moderate severity was higher using the BNSS than thePANSS, except in twocentersin which it wascompa- rable. All centers showed a positive correlation between BNSSandPANSScorenegativesymptomtotals>0.70(Table S2).AstothecorrelationsofthePSP totalwithbothBNSS andPANSS core negativesymptoms(Table S3), allcenters showed a negative association, although in some centers theassociationwasnotsigniﬁcant.However,thedifference between the highest and lowest correlation coefﬁcients