Asian
pearls
Validation
of
the
Chinese
version
of
Community
Assessment
of
Psychic
Experiences
(CAPE)
in
an
adolescent
general
population
Winifred
Mark
a,
Timothea
Toulopoulou
b,c,d,e,*
aKwaiChungHospital,HospitalAuthority,HongKongSpecialAdministrativeRegion b
DepartmentofPsychology,TheUniversityofHongKong,HongKongSpecialAdministrativeRegion
c
TheStateKeyLaboratoryofBrainandCognitiveSciences,TheUniversityofHongKong,HongKongSpecialAdministrativeRegion
d
DepartmentofPsychology,BilkentUniversity,Ankara,Turkey
e
DepartmentofBasicandClinicalNeuroscience,theInstituteofPsychiatry,PsychologyandNeuroscience,King’sCollegeLondon,UK
ARTICLE INFO
Articlehistory: Received11May2016
Receivedinrevisedform5January2017 Accepted16January2017
Keywords:
CommunityAssessmentofPsychic Experiences(CAPE) Psychosisproneness Schizophrenia Chinese Translation Psychometrics ABSTRACT
TheCommunityAssessmentofPsychicExperiences(CAPE)isapopular42-itemself-reportassessmentof psychosisproneness(PP)thathasbeenwidely-translated.However,thereisasyetnovalidationofCAPE innon-Westernlanguages.Here,wevalidatedaChinesetranslationofCAPE(“CAPE-C”)inayoung Chinese communitysample.Factoranalyses wereemployedinasampleof 660 individuals(mean age=18.63)toidentifyaculturally-sensitivefactorstructureforCAPE-C(Study1).Sinceconfirmatory factoranalysis(CFA)suggestedthatCAPE-Cdidnotfollowtheoriginalfactorstructure,exploratoryfactor analysisandfollow-upCFAwereemployedtoestablishanalternativestructure,resultingina15-item “CAPE-C15”whichretainedathree-factorstructuretappingpositive,negativeanddepressivesymptoms. TodemonstratethespecificityofCAPE-C15asameasureofPP,weconductedregressionanalysesto examineassociationsbetweenCAPE-C15dimensionsandothermeasuresofpsychoticanddepressive symptoms(Study2).ResultsconfirmedthatCAPE-C15dimensionsshowedspecificassociationswith relevant symptom dimensions of other measures, but not with irrelevant ones. Finally, to aid interpretationofCAPE-C15scores,ReceiverOperatingCharacteristicanalysiswasconductedtoestablish acut-offscorethatcouldindicatetest-takers’needforclinicalattention(Study3).Wefoundthatacut-off scoreof8.18onCAPE-C15positiveandnegativesymptomfrequencyanddistressscoresdistinguished individualswhosePPwaswithinnormalrangesfromthoseatpsychometrichigh-risk(sensitivity:78.6%; specificity:77.7%).CAPE-C15willlikelyproverelevanttoresearchersandhealthcareproviderswhoserve Chinese-speakingadolescentsandyoungadults.
©2017ElsevierB.V.Allrightsreserved.
1.Introduction
Psychosis proneness (PP) refers to the extent to which an individualexperiencespsychotic-like experiences(PLEs)that do notreachclinicalthresholdforpsychoticdisorders.Despitebeing relativelyprevalentinadolescenceandyoungadulthood(vanOs et al., 2009), such isolated symptoms are associated with an increasedriskofdebilitatingpsychiatricoutcomes(Kelleheretal., 2012a,2012b;Rössleretal.,2007;Werbeloffetal.,2012),and self-harmorsuicidebehaviors(Capraetal.,2015;Honingsetal.,2016). Assuch,thereisaneedforaccuratemeasurementofPPinresearch andclinicalpractice.OnepopularPPmeasureistheCommunity
Assessment of Psychic Experiences (CAPE-42) (Stefanis et al., 2002),a42-itemself-reportmeasureoffrequencyanddistressof psychotic-likefeelings,thoughtsormentalexperiences.
SeeingthatCAPE-42hasnotbeenvalidatedinanon-Western language,herewereportavalidationstudyofaChinesetranslation of CAPE-42 (“CAPE-C”) in an adolescent and young adult community sample.Validationis importantbecausetranslation canresultin difference initem meaning,anddifferentcultures may have different manifestations of PP. A late adolescent populationwaschosensinceyoungpeopleareatthehighestrisk todisplayPLEs(Laurensetal.,2008;Poultonetal.,2000),andto transitiontoapsychoticdisorder(HarropandTrower,2001).By validatingCAPE-C inadolescentsand youngadults,wehopeto recastthisapertureofvulnerabilityintoawindowofopportunity forearlydetectionandintervention.
Inaseriesofstudies,wefirststatisticallyestablishedafactor structure sensitive to young Chinese, resulting in a shortened *Corresponding author at: Department of Psychology, 6/F, the Centennial
Campus,PokfulamRoad,HongKongSpecialAdministrativeRegion. E-mailaddress:timothea@hku.hk(T.Toulopoulou).
http://dx.doi.org/10.1016/j.ajp.2017.01.012
1876-2018/©2017ElsevierB.V.Allrightsreserved.
ContentslistsavailableatScienceDirect
Asian
Journal
of
Psychiatry
questionnaire with 15 items (“CAPE-C15”) (Study 1). We then comparedCAPE-C15tootherestablishedmeasuresofpsychotic anddepressionsymptoms,inordertodemonstratethespecificity ofCAPE-C15subscalesinmeasuringpositivepsychoticsymptoms, negativepsychoticsymptomsanddepressivesymptoms(Study2). Finally,we established a cut-offscore that couldindicate test-takers’needforclinicalattention(Study3).Thefruitofthiseffortis a comprehensive psychometric profile of CAPE-C15, which can facilitateitsinformedusebyresearcherandclinicians.
2.Study1:translationandfactorstructureofCAPE-C
Study 1 concerned the psychometric robustness of CAPE-C, whichis importantbecauseit informswhetherscoresobtained couldbetrusted.Psychometricrobustnessofascalecanbegleaned byitsscore“internalreliability” and “factorstructure”.Internal reliability concerns consistencyof item scores (i.e., do CAPE-C items consistently measure the same construct?), while factor structure concerns scale composition (i.e., which items group togethertoformusefulsubscalesinCAPE-C?).
Toanswerthesequestions,wefirstemployedaconfirmatory factoranalysis(CFA)totestwhethertheoriginalCAPE-42factor structure modelfittedourdata.An acceptablemodelfit would mean that the scoring method of CAPE-42 could be directly adoptedtogeneratemeaningfulCAPE-Cscores.Anunacceptable model fit, however, would mean that scorescalculated by the originalscoringmethodwouldnotbeinterpretableforourdata. Sincemodelfitwasnotobtainedinourdata,exploratoryfactor analysis(EFA)wasusedtoexposehowbesttorestructureCAPE-C (e.g.,bydeletingquestions,orregroupingquestionsintodifferent subscales)inordertoobtainmeaningfulscores.Tomakesurethat scores generated from this alternative structure were reliable, Cronbach’salphacoefficientswerecalculated.Finally,tomakesure that the structure suggested by EFA represented a good fit of Chinesedata,follow-up CFA was conducted.Sucha meticulous approach ensured that the factor structure reported here for CAPE-Cwasculturally-sensitiveforaChineseyoungpopulation. 2.1.Materialsandmethods
2.1.1.Participants
669 Chinese adolescents and young adults, including 349 singletons(201females,148males)and 320twinsor similarly-aged siblings (199 females, 118 males, 3 did not declare their gender),wererecruitedfortheTwinsscanChinaproject,atwinand familystudyofPP intheChinesepopulation. Participantswere recruitedthroughtwinregistries,secondaryschoolsand universi-tiesinfivecitiesofThePeople’sRepublicofChina:HongKong, Beijing,Nanjing,QingdaoandGuangzhou.
Singleton and sibling/twin samples weredivided into three sub-samples for factor analyzes. In order to minimize familial
covariance between participants in each sub-sample, siblings/ twinswereassignedtoseparatesubsamplesinapseudo-random fashion.Wefirstgeneratedrandomizedintegersof“1”,“2”and“3” using a computerized “random integer generator” as a group number for each participant. We then manually inspected the dataset to ensure that twins/siblings were assigned a group numberthatwasdifferentfromthatassignedtohis/herco-twin/ sibling. The three sub-samples consisted of, respectively, 224 individuals(thereafter“Sample1”),250individuals(“Sample2”), and186individuals(“Sample3”).
2.1.2.Instruments
2.1.2.1. Community Assessment of Psychic Experiences (CAPE-42). CAPE-42 contained three dimensions: the positive dimension(CAPE-pos;20items)tappedbehaviorspertainingto reality distortion (e.g.,“hear voices when you are alone”); the negativedimension(CAPE-neg;14items)tappeddisruptionsto normal behavior (e.g., “experience few or no emotions at importantevents”);and thedepressivedimension(CAPE-dep;8 items) includedmainlycognitive symptoms ofdepression (e.g., “feellikeafailure”).
To ensure adequate translation of CAPE-42 from English to Chinese,translationsbythreeChinese-Englishbilingualspeakers were back-translated by three other Chinese-English bilingual speakersblindtothestudyhypothesesandtheEnglishCAPE-42. Discrepanciesintheforwardandbacktranslationswerereviewed bythepresentinvestigators.
Administration and scoring of CAPE-C followed that of the original version (Stefanis et al., 2002).Participants rated their frequency of PLEs on a four-point Likert scale: “never”=1, “sometimes”=2, “often”=3, “nearly always”=4. Apart from an endorsementof“never”,responderswereaskedtoratethedegree of subjective distress associated withthe psychotic experience from“notdistressed”=1,“abitdistressed”=2,“quitedistressed”= 3 to “very distressed”=4. To take into account partial non-response,aweightedscorewascalculatedbysumofscoresdivided bythenumberofcompletedquestions.Onlyfrequencyscoreswere usedforanalysis.
2.2.Results
2.2.1.Descriptivestatistics
Participants’ responses were screened for missing data, resultinginafinaldatasetof660participants(394females,263 males)withanaverageageof18.63(SD=1.99).Samples1,2and3 did not differ in proportion of participants from each data collection site, mean age of participants, gender ratio, and weightedtotalfrequencyscoresonCAPE-C(Table1).Hence,the randomizationwasconsideredsuccessful.
Table1
DemographiccharacteristicsofrandomizedsamplesandinternalreliabilityofCAPE-42scalescores.
Sample1 Sample2 Sample3
N 224 250 186
Mean(SD)ageinyears 18.80(1.90) 18.63(2.01) 18.42(2.08) RatioofparticipantscollectedfromBeijing:Guangzhou:HongKong:Qingdao:Nanjing 18:15:157:28:6 23:20:170:30:7 19:17:119:26:5
Ratiooffemale:male 132:92 144:105 118:66
Ratioofsingleton:sibling/twin 127:97 138:112 84:102 WeightedCAPE-Ctotalfrequencyscore 1.89(.45) 1.88(.37) 1.82(.37) CAPE-CtotalfrequencyscoreCronbach’salpha 0.94 0.92 0.91 CAPE-CpositivefrequencyscoreCronbach’salpha 0.87 0.85 0.82 CAPE-CnegativefrequencyscoreCronbach’salpha 0.88 0.85 0.85 CAPE-CdepressivefrequencyscoreCronbach’salpha 0.82 0.76 0.85 Notes:CAPE-C=Chinese-translatedCommunityAssessmentofPsychicExperiences;SD=Standarddeviation.
2.2.2.InternalreliabilityofCAPE-C
Internal reliability of CAPE-C scores was measured by Cronbach’salpha.Alphavaluesgreaterthan0.70wereconsidered tobe acceptable for further analysis (Nunnally, 1978). Overall, frequencyscoresonCAPE-C(Table1)showed“good”to“excellent” consistency across samples according to George and Mallery (2010).
2.2.3.ConfirmatoryfactoranalysisoforiginalCAPE-42structure The internal structure of Sample1 was fitted tothat ofthe originalCAPE-42modelwithCFAusingLISREL9.1(Joreskogand Sorbom,2006).Ofthe224participantsincluded,effectivesample size (ESS) for analysis (excluding data of participants that completed less than 70% of CAPE-42 items or those with acquiescenceresponsebias)was 217,yieldinga participant:item ratio of 5.17. Questionnaire items were assumed to contribute information to only one latent factor in the model. Maximum likelihoodestimationwasused.Themodelproduced goodness-of-fit indices as follows: RMSEA=0.08, CFI=0.87, and SRMR=0.09, whichdidnotmeetgeneralcriteriaofmodelfit(i.e.,RMSEA.07, CFI.95,SRMR<0.08)(HuandBentler,1999).
2.2.4.Exploratoryfactoranalysisforalternativefactorstructure Sincegoodness-of-fitindexesoftheaboveCFAdidnotexceed establishedcut-offs,EFAwasperformedinSPSS20.0(IBMCorp., 2011)onSample2,toobtainafactormodelthatoptimallyaccounts for thedata in our Chinesepopulation. Of the250 individuals included,ESSwas 212, yieldinga participant:item ratioof 5.05. DimensionsofPLEswereidentifiedusingEFAwithprincipalaxis factor extraction with Promax rotation for conservativeness (OsborneandCostello,2009).Thenumberoffactorstoberetained wasdeterminedbyeigenvalues>1andthescreeplot(Fig.1).EFA suggestedamultiple-factorsolutionwith12eigenvalues>1and thescreeplotsuggestedathree-factorsolution.
The three-factor model accounted for 36.42% of the total variance. The first factor explained 23.69% of the variance,
corresponding to items tapping depressive experiences. The secondfactorexplained7.32%ofthevariance,andcomprisedof itemsrelated tonegativesymptoms.The thirdfactor explained 5.42% of thevariance, and containeditemsconcerning positive symptoms.Theresultantmodelconsistedof15items,hereafter knownas“CAPE-C15”(Table2).
2.2.5.InternalreliabilityofCAPE-C15
Cronbach’scoefficientalphawascalculatedforCAPE-C15scores in Sample 2. Alpha value for CAPE-C15 total=.82 (15 items), positivescale=.74(4items),negative=.69(5items)and depres-sivescale=.75(6items).
2.2.6.ConfirmatoryfactoranalysisofCAPE-C15
TheinternalstructureofSample3wasfittedtotheCAPE-C15 factorstructure,withaCFAusingLISREL9.1.Ofthe186participants included in Sample 3, ESS for analysis was 180, yielding a participant:itemratioof12.Themodel producedgoodness-of-fit indices as follows: RMSEA=.04, CFI=.96,and SRMR=.061, which represented acceptable model fit (i.e., RMSEA.07, SRMR<.08, CFI.95) (Hu and Bentler, 1999). Hence, the CAPE-C15 factor structurewasstatisticallyrobust.
2.3.Discussion
Recognizingtheresearchand clinicalvaluesof CAPE-42,the present study was conducted to rectify the lack of validated Chinesetranslationsinthecurrentliterature.Wefoundthatthe originalstructureofCAPE-42didnotachieveacceptablemodelfit inouryoungChinesecommunitysample.Instead,asmallersetof 15itemswasretainedinCAPE-C15,clusteredintothreesubscales correspondingtoapositive,anegative,andadepressivedimension (i.e.,a tri-dimensionalmodel that mirrorstheoriginalCAPE-42 structure as proposed by Stefanis and colleagues). Our results corroborateourpreviousfindingthatitemnumbersandloadings tendedtobedifferentintranslatedversionsofCAPE-42(Markand
Toulopoulou, 2016), further reinforcing the need for proper validationoftranslations.
Deleteditems tendedtohave lowcommunalities(i.e.,these itemsexplainedtoosmallanamountofvarianceintheretained factorstobepreserved)orwereassessedbysimilaritemsthat had been retained. In spite of the greatly reduced number of items, theCAPE-possubscale structure previouslyfoundin our meta-analysis of CAPE-pos factor analytic studies (Mark and Toulopoulou,2016)wasretainedinCAPE-C15.Specifically,items pertaining to“Bizarre experiences”, “Delusional ideations” and “Perceptualanomalies”couldbefoundinthepositivesubscaleof CAPE-C15whileitemspertainingto“Socialwithdrawal”, “Affec-tive flattening” and “Avolition” could befound in the negative subscaleofCAPE-C15.Itis,however,worthyofnotethattwoitems originally proposed to be in CAPE-neg were loaded onto the depressivefactorinCAPE-C15.Thisiscongruentwiththeviewthat negative symptoms show phenomenological similarities to depressive symptoms (Murali and Kumar, 2008; Newcomer et al.,1990; Saxet al.,1996).Longitudinal studiessuggest that thetwocouldbedifferentiatedbytheirtemporalcharacteristics: depressive symptoms varied more over time while negative symptomstendedtobemorestableandtrait-like(Häfneretal., 2005).However,since CAPEwas designedtobea self-reportof
lifetimepsychoticexperience,itmightbedifficulttodifferentiate the lack of motivation to establish regular activities such as hobbies (negative symptom) and lack of motivationto pursue establishedhobbies(depression)withoutfurtherinformationon thetimecourseofthesymptom.
Onelimitationofourstudyistherelativelysmallsamplesize relativetothepopulationofChina.However,ourfactoranalytic studiesallyieldedparticipant:itemratiosthatliedinorexceeded thesuggestedrangeof3to6(Cattell,1978).Hence,ourresultsare statisticallyvalid.ArelatedcaveatisthatCAPEproducesskewed dataandlowratesofendorsementoncertainitemsinthegeneral population, asisthecase formostmeasurementsofsubclinical psychopathology. Thisraises theissue of low variation in item responsesduetosmallsamplesize.However,wedonotthinkthis couldexplainourresultssincefurtheritemresponseanalysisof thedatasetindicatedthatmostitemsachievedatleastmoderate itemdiscrimination.Anothercaveatisthatthescorereliabilityin CAPE-C15 was lower than in the initialversion. However, the possibility that this relatestothelower number ofscale items couldnotbeexcluded(Streineretal.,2014).Toconclude,Study1 presentedpreliminarydatathatCAPE-C15issuitableforindexing PPinayoungChinesepopulation.Furtherreplicationstudieswill servetoconfirmthis.
Table2
ExploratoryFactorAnalysisoftheChineseCAPE-42itemsinSample2.
CAPE-42Items Pos Neg Dep
01. Feelsad .908
02. Peopleseemtodrophintsaboutyouorsaythingswithadoublemeaning
03. Notaveryanimatedperson .799
04. Notmuchofatalkerwhenyouareconversingwithotherpeople .527 05. ThingsinmagazinesoronTVwerewrittenespeciallyforyou
06. Somepeoplearenotwhattheyseemtobe 07. Beingpersecutedinsomeway
08. Experiencefewornoemotionsatimportantevents .483 09. Pessimisticabouteverything
10. Conspiracyagainstyou .369
11. Destinedtobesomeoneveryimportant 12. Nofutureforyou
13. Youareaveryspecialorunusualperson
14. Donotwanttoliveanymore .533
15. Communicatetelepathically 16. Nointeresttobewithotherpeople
17. Electricaldevicessuchascomputerscaninfluencethewayyouthink
18. Lackinginmotivationtodothings .407
19. Cryaboutnothing .539
20. Believeinthepowerofwitchcraft,voodooortheoccult
21. Lackinginenergy .712
22. Peoplelookatyouoddlybecauseofyourappearance 23. Mindisempty
24. Thoughtsinyourheadarebeingtakenawayfromyou 25. Spendingallyourdaysdoingnothing
26. Thoughtsinyourheadarenotyourown 27. Feelingsarelackinginintensity
28. Thoughtssovividthatyouwereworriedotherpeoplewouldhearthem 29. Lackinginspontaneity
30. Hearyourownthoughtsbeingechoedbacktoyou 31. Underthecontrolofsomeforceorpowerotherthanyourself 32. Youremotionsareblunted
33. Hearvoiceswhenyouarealone
34. Hearvoicestalkingtoeachotherwhenyouarealone .515 35. Youareneglectingyourappearanceorpersonalhygiene
36. Youcannevergetthingsdone .506
37. Youhaveonlyfewhobbiesorinterests .440
38. Feelguilty
39. Feellikeafailure .385
40. Feeltense
41. Adoublehastakentheplaceofafamilymember,friendoracquaintance .640 42. Seeobjects,peopleoranimalsthatotherpeoplecannotsee .905
Notes:Onlyitemswithfactorloadings0.35areshown;CAPE–42=CommunityAssessmentofPsychicExperiences;Dep=Depressivedomain;EFA=Exploratoryfactor analysis;Neg=Negativedomain;Pos=Positivedomain.
3.Study2:convergent-divergentvalidityofCAPE-C
Tobeausefulassessmenttoolinresearchandclinicalsettings, CAPE-Cshouldonlymeasurewhatitsupposedlymeasures,andnot measure irrelevant constructs. This is known as “convergent-divergentvalidity”.Toexaminetheconvergent-divergentvalidity of CAPE-C15, CAPE-pos and CAPE-neg symptom dimensions of CAPE-C15needtobetestedagainstanotherPPinstrument.The questions of interest are: does CAPE-pos, but not CAPE-neg, correlate with assessments of positive symptoms? Conversely, doesCAPE-neg,butnotCAPE-pos,correlatewithassessmentsof negativesymptoms?Thismethodisknownas“cross-validation”. PreviousresearchhassuggestedthatCAPE-posandCAPE-neg had convergent-discriminant validity (Konings et al., 2006). However,cross-validationof CAPE-neg andCAPE-depwas often neglectedin theliterature.Given that CAPE-depwas especially addedtoCAPE-42inordertodiscriminatebetweennegativeand depressivesymptoms (Stefanis et al., 2002), it is important to establish whether CAPE-neg and CAPE-dep indeed specifically measured negative PP symptoms and depressive symptoms respectively.
Inthisstudy,cross-validationwascarriedoutwithCAPE-C15on the one hand; and an interview-based schizotypy measure “StructuredInterview forSchizotypy-Revised(SIS-R)”as wellas aself-reportdepressionmeasurein“SymptomChecklist90 (SCL-90)”ontheother.SIS-RwasselectedsothatsubtlePLEsthatmight be missed by self-report could be captured. The depression subscaleof SCL-90was selectedfor comparisonwith CAPE-dep sinceSCL-90hasbeenrepeatedlyvalidatedinChinesepopulations (Chen and Li, 2003;Feng and Zhang, 2001; Hu,2006; Liu and Zhang,2004),andhasseparatenormsforChineseadolescents(Liu andZhang,2004).ItwashypothesizedthatCAPE-posand CAPE-negwouldshowindependentandspecificassociationswiththe relevant symptom dimension in SIS-R or SCL-90 (convergent validity),andshownon-significantassociationswiththeirrelevant symptom dimension (divergent validity). In order to test the assumptionthat CAPE-C15 leads tomore a valid estimation of positivesymptoms,negativesymptoms,anddepressionsymptoms thanthe42-itemCAPE-CinaChinesepopulationsample,adirect comparisonofconvergent-divergentvalidityoftheCAPE-C15and CAPE-Cwasconducted.
3.1.Materialsandmethods 3.1.1.Participants
The sample included 212 participants recruited for TwinsscanChina (71 male, 139 female, 2 did not declare their gender).Onlysiblings/twins collectedinHongKongandBeijing wereincludedintheconvergent-divergentanalysisforCAPE-C15, sinceonlythesegroupscompletedCAPE,SIS-RandSCL-90. 3.1.2.Instruments
3.1.2.1. Chinese-translated Community Assessment of Psychic Experiences (CAPE-C)/15-item Community Assessment of Psychic Experiences(CAPE-C15). AllparticipantscompletedtheCAPE-Cas described in Study1. Separate analyses were donetocompare CAPE-CitemsandCAPE-C15items.CAPE-C15containedfouritems describingpositivePLEsofhallucinationandpersecutorybeliefs; fiveitemsdescribing negativesymptomsof affectiveflattening, socialwithdrawalandapathy;andsixitemsdescribingdepressive experiences including behavioral and cognitive aspects. Administrationandscoringprocedureswereidenticaltothatin Study1.ScorereliabilitiesofCAPE-Cfrequencysubscalewere0.93 fortotalscore,0.86forCAPE-pos,0.85forCAPE-neg,and0.78for CAPE-dep.ScorereliabilitiesofCAPE-C15frequencysubscalewere
0.86fortotalscore,0.74forCAPE-pos,0.67forCAPE-neg,and0.74 forCAPE-dep.
3.1.2.2.StructuredInterviewforSchizotypy-Revised(SIS-R). Ashort versionoftheSIS-Rconsistingof11symptomsand4signswas used(Collinetal.,2011;Latasteretal.,2014;vandenBergetal., 2013).Theshortened SIS-Rcouldbesubdivided intoa positive symptomdimensiontappingreferentialthinking,suspiciousness, magical ideation, illusions, psychotic symptoms, and derealization/depersonalization; and a negative symptom dimension tapping social isolation, introversion, hypersensitivity, restricted affect, thought disturbances (e.g., tangentiality),loosening of associations, poverty of speechand odd/eccentric behavior. To reduce interviewers’ idiosyncratic judgments, Vollema and Ormel (2000) explicitly defined the criteriaforeach symptomand sign,which researchersassessed with standardized questions tapping frequency, duration and degreeofconvictionorobservation.Basedonsuchinformation, researchersgaveanoverallratingofseverityonafour-pointLikert scale(“absent”=0,“mild”=1,“moderate”=2and“severe”=3). 3.1.2.3. Symptom Checklist 90 (SCL-90). The SCL-90 (Derogatis, 1983) is a self-report questionnaire assessing psychopathology, including somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism symptoms. Participants scored the 90 items on a five-point Likert scale based on distress (“none”=1, “a little bit”=2, “moderate”=3, “quite a bit”=4,“extreme”=5).Aweightedscorewascomputedforeach subscalebyaddingallscoresdividedbythenumberofitemsin each subscale. Only the depression subscale was used in this analysis.Thedepressionsubscalecontainedsymptomsconsistent withtheclinicaldiagnosisofamajordepressiveepisodeinDSM-5 (AmericanPsychologicalAssociation,2013).Theinternalreliability oftheSCL-90depressionsubscale(13items)inthissamplewas acceptable:Cronbach’salpha=0.90.
3.2.Results
3.2.1.Descriptivestatistics
Participants’ responses werescreened for missing data. The finalsampleconsistedof196participants(64male,130female; twoparticipantsdidnotdeclaredtheirgender)withameanageof 17.76(SD=1.97).Themean,standarddeviationandstandarderrors ofeachvariableofinterestarepresentedinTable3.
3.2.2.Convergent-divergentvalidity
Since thedatawashierarchicalin natureduetosiblingship, multilevelregressionswereperformedwithmaximumlikelihood estimationtoexamineassociationsbetweendimensionsof CAPE-C15 on the one hand, and the SIS-R dimensions or SCL-90 depressionscaleontheother.
Table3
MeanandstandarddeviationofCAPE-C15,SIS-RandSCL-90scores.
Mean SD
CAPE-C15positivedimensionfrequency 1.59 0.62 CAPE-C15negativedimensionfrequency 2.09 0.58 CAPE-C15depressivedimensionfrequency 1.95 0.52 SIS-Rpositivedimension 3.55 2.75 SIS-Rnegativedimension 3.09 2.14 SCL-90depressionsubscale 1.66 0.62 Notes:CAPE-C15=15-itemChineseCommunityAssessmentofPsychicExperiences; SCL–90=Symptoms Checklist 90; SIS-R=Structured Interview for Schizotypy, Revised.
Multilevel regression with CAPE-C15 positive and negative subscalescoresaspredictorsandSIS-Rpositivedomainscorethe outcome showed that CAPE-C15 positive domain significantly predicted SIS-Rpositivedimensionscores(ß=1.70, 95%CI=1.07, 2.33,p<.001).TheCAPE-C15negativedomaindidnotsignificantly predict SIS-R positive symptom domain scores (ß=0.47, 95% CI=0.20,1.13,p=.17).Hence,onlyCAPE-pos,butnotCAPE-neg,of CAPE-C15wasassociatedwithSIS-Rpositivesymptomdimension. Multilevel regression with CAPE-C15 positive and negative subscalescoresaspredictorsandSIS-Rnegativedomainscoreas theoutcomeshowedthatCAPE-C15negativedomainsignificantly predictedSIS-Rnegativedimensionscores(ß=1.03,95%CI=0.50, 1.57,p<.001).TheCAPE-C15positivedomaindidnotsignificantly predict SIS-R negative symptom domain scores (ß=.34, 95% CI=0.17,0.85,p=.19).Hence,onlyCAPE-neg,butnotCAPE-pos, of CAPE-C15 was associated with SIS-R negative symptom dimension.
MultilevelregressionwithCAPE-C15negativeanddepressive subscale scores as predictors and SCL-90 depression subscale scoresastheoutcomeshowedthatCAPE-C15depressivedomain significantlypredictedSCL-90depressionsubscalescores(ß=0.48, 95%CI=0.29,0.67,p<.001).TheCAPE-C15negativedomaindidnot significantly predict SIS-R depression symptom domain scores (ß=0.13,95%CI=0.05,0.30, p=.15). Hence,onlyCAPE-dep,but notCAPE-neg,ofCAPE-C15wasassociatedwithSCL-90depression dimension.
Furthermore,totest theassumptionthat CAPE-C15 leadsto moreavalidestimationofpositivesymptoms,negativesymptoms, and depression symptomsthan the42-itemCAPE-C, multilevel regressionswereperformedwithmaximumlikelihoodestimation toexamineassociationsbetweendimensionsofthe42-item CAPE-Contheonehand,andtheSIS-RdimensionsorSCL-90depression scaleontheother.MultilevelregressionwithCAPE-Cpositiveand negativesubscalescoresaspredictorsandSIS-Rpositivedomain score as the outcome showed that CAPE-C positive domain significantlypredicted SIS-Rpositivedimensionscores(ß=1.98, 95%CI=0.89,3.08,p<.001).TheCAPE-Cnegativedomaindidnot significantly predict SIS-R positive symptom domain scores (ß=.83,95%CI=.16,1.83,p=.10).Hence,onlyCAPE-pos,butnot CAPE-neg, of the 42-item CAPE-C was associated with SIS-R positivesymptomdimension.
Multilevel regression with CAPE-C positive and negative subscale scores as predictors and SIS-R negativedomain score astheoutcomeshowedthatCAPE-Cnegativedomainsignificantly predictedSIS-Rnegativedimensionscores(ß=1.52,95%CI=0.71, 2.33,p<.001).TheCAPE-Cpositivedomaindidnotsignificantly predict SIS-R negative symptom domain scores (ß=.10, 95% CI=.81,1.01, p=.83).Hence,onlyCAPE-neg, but notCAPE-pos, of the 42-item CAPE-C was associated with SIS-R negative symptomdimension.
Multilevel regression with CAPE-C negative and depressive subscale scores as predictors and SCL-90 depression subscale scoresastheoutcomeshowedthatCAPE-C15depressivedomain significantlypredictedSCL-90depressionsubscalescores(ß=0.33, 95%CI=0.11, 0.55, p=.003). The CAPE-C negative domain also significantlypredictedSIS-Rdepressionsymptomdomainscores (ß=0.39, 95%CI=0.18,0.61,p<.001).Hence,bothCAPE-depand CAPE-neg of the 42-item CAPE-C wereassociated with SCL-90 depressiondimension.
Insummary,thediscriminantvalidityofCAPE-C15negativeand depressivescalesaresuperiortothatofthe42-itemCAPE-C. 3.3.Discussion
Study2wasconductedtoexaminetheconvergent-divergent validity of CAPE-C15. Consistent with our hypotheses, results
confirmed independent significant associations with relevant subscalesin SIS-RandSCL-90, andnon-significantassociations with irrelevant subscales. This showed that subscales of CAPE-C15 had convergent-divergent validity, supporting the individual subscales as specific measures of positive PP, negative PP and depressivesymptoms. In addition, this study also found that CAPE-neg of CAPE-C15 could discriminate between negative and depressive symptoms better than that of the 42-item CAPE-C. This is important because negative symptoms show phenomenological similarities to depressive symptoms(MuraliandKumar,2008;Newcomeretal.,1990;Sax et al., 1996), and must be accurately classified to avoid mis-assessment. Taken together, CAPE-C15 could measure positive, negativeanddepressivesymptomsrobustly,showingpromiseas a screening tool for PP.
4.Study3:CAPE-C15asascreeningtool:ROCanalysis HavingestablishedCAPE-C15’spsychometricproperties,the nextstep is toconsiderits applicationin research andclinical settings. To do so, researchers and clinicians must know at which point a score on CAPE-C15 indicated elevated PP. One way to approach this is to conduct a Receiver Operating Characteristic (ROC) analysis. ROC analysis works by using an “indextest”(i.e.,thetestof interest;inourcase,CAPE-C15)to predictclassificationof individualswhodidordidnotwarrant clinical attention as determinedbya “golden standard” (i.e.,a validatedmeasureorcliniciandiagnosis).Thisresultsina CAPE-C15 cut-off score that classifies individuals into those who experience PLEs but are not distressed by them (prevalence 8%; van Os et al., 2009), and psychometric high-risk individualswhoreportPLEsthatdonotcrossclinicalthreshold for a diagnosis of psychotic disorder, but who nonetheless experience distress and may deserve clinical attention (prevalence=4%; van Oset al., 2009).
CAPEcut-offscoresfromROCanalyseshavebeenpreviously reported in help-seeking populations. Research showed that using theDutchCAPE-42asanadjuncttoclinicalinterviewing improved detection of first episode psychosis (FEP) in new referralsatamentalhealthserviceclinic(Boonstraetal.,2009). That data yielded a cut-off of 50 (out of 80) on either the frequencyordistress dimensionofCAPE-pos,whichprovideda sensitivity of 77.5%and a specificity of 70.5%.This meansthat, usingarawscorecut-offof50oneitherCAPE-posfrequencyor CAPE-posdistressruledin77.5%ofindividualswhohaveFEP,while ruling out70.5%ofindividuals whodidnothaveFEP.Similarly, thereisevidencethattheGermanCAPE-42wasusefulindetecting individuals at ultra-high risk for psychosis in a help-seeking clinical population. Two cut-off points were suggested for a weightedCAPE-posfrequencyanddistressscore:3.20(sensitivity: 67%;specificity:73%)and2.80(sensitivity:83%;specificity:49%) (Mossahebetal.,2012).
Study 3 aimedtoestablish a cut-offscorefor CAPE-C15.In contrasttoaforementionedstudiesthatemployedhelp-seeking populations, we were interested in using CAPE to screen for psychometric high-risk populations in a general population. Given the association of PP with clinical outcomes (Kelleher et al., 2012a; Kelleher et al., 2012b; Rössler et al., 2007; Werbeloff et al., 2012), such a cut-off score would facilitate earlydetectioninnon-clinicalsettings(e.g.,schools)forreferral toprimaryhealthcare.A“CAPE-C15PLEfrequencyanddistress score” composed of the frequency and distress scores of CAPE-pos and CAPE-neg were used as the index scores, and were compared to a gold standard as defined by the SCL-90 psychoticism subscale score.
4.1.Materialsandmethods 4.1.1.Participants
401 singletonparticipants (161male, 238female; 2 didnot declarehis/hergender)withameanageof19.68(SD=1.54)were included.TheywererecruitedthroughuniversitiesinHongKong. 4.1.2.Instruments
4.1.2.1. Chinese-translated Community Assessment of Psychic Experiences (CAPE-C)/15-item Chinese-translated Community Assessment of Psychic Experiences (CAPE-C15). All participants completedCAPE-CasdescribedinStudy1.CAPE-CandCAPE-C15 weretheindex tests ofseparate analyses. AweightedCAPE-C/ CAPE-C15 PLE frequency and distress score was calculated by summing frequency and distress scores for each item of the positiveandnegativescales,dividedbythenumberofquestions completed.Scorereliabilitywasacceptable:Cronbach’salphawas .86fortheCAPE-Cpositivefrequencyanddistressscoreand.84for the CAPE-C negative frequency and distress score. Cronbach’s alphawas .88for theCAPE-C15positive frequencyanddistress scoreand.72 fortheCAPE-C15negativefrequencyanddistress score.
4.1.2.2. Symptom Checklist 90 (SCL-90). The SCL-90 (Derogatis, 1983)waschosenasthegoldenstandardduetotheavailabilityof well-established norms for Chinese adolescents (Chen and Li, 2003;FengandZhang,2001;Hu,2006;LiuandZhang,2004).Only the psychoticism subscale was used in this analysis. The psychoticism subscale taps experiences associated with the psychosis continuum, from social withdrawal to hallucinations and delusions. In accordance with guidelines from the Beijing SuicideandResearchPreventionCenter,acut-offof3wasusedto indicate an elevated level of symptomatology that warranted clinical attention in Chinese populations. SCL-90 psychoticism subscale scores in our sample were acceptable: Cronbach’s alpha=.83.
4.2.Results
4.2.1.Descriptivestatistics
Participants’ responses were screened for missing data on eitherCAPE-CweightedPLEfrequencyanddistressscoresor SCL-90 psychoticism subscale. Five participants were excluded for incompleteSCL-90psychoticismsubscalescores.Thefinalsample included396participants(160maleand236female)withmean ageof19.68(SD=1.54).Fourteen(3.54%)participantswere “SCL-90-positive”(i.e.,abovecut-off)and382(96.46%)were “SCL-90-negative” (i.e., below cut-off). Mean weighted CAPE-C15 PLE frequencyanddistressscorewas6.67(SD=2.15),whilethatof SCL-90psychoticismsubscalewas1.56(SD=.58).
4.2.2.Receiveroperatingcharacteristicanalysis
ROCanalysiswasconductedinSPSS20.0(IBMCorp.,2011).The areaundertheROCcurveisthemeasureoftheeffectivenessand validityof theindex test (Hajian-Tilaki, 2013), which was .817 (SE=.065)inouranalysis,confirmingCAPE-C15PLEfrequencyand distress scores as a “good” predictor (Swets et al., 2000). The optimal cut-off score was determined by the Youden criteria (Youden,1950),whichrepresentedtheoptimalstatisticalcut-off correspondingtoacombinationofhighsensitivityandspecificity. Acut-offof8.18resultedinasensitivityof78.6%andaspecificityof 77.7%.
ROCanalysisofthe42-itemCAPE-Cresultedinanareaunder curveof .828(SE=.055)inouranalysis,confirmingCAPE-C PLE frequencyanddistressscoresasa“good”predictor(Swetsetal.,
2000).UsingtheYoudencriteriarevealedthatanoptimalcut-offof 9.14resultedinasensitivityof92.9%andaspecificityof74.1%. 4.3.Discussion
Our studyis thefirsttodatetoestablisha cut-offscorefor differentiating individuals with normal-range PP from psycho-metrichigh-riskindividualswhomightwarrantclinicalattention, inthegeneralpopulation.ROCanalysissuggestedacut-offscoreof 8.18fortheweightedCAPE-C15PLEfrequencyanddistressscores. Comparingittoacut-offof9.14forCAPE-C(sensitivity=92.9%; specificity=74.1%),onecouldseethatincludingall42 questions onlyincreasedtheoptimalcut-offby1pointandresultedina dropinspecificity,whichsupportsitemreduction.CAPE-C15may be used in routine student health checkups for screening individualswho areexperiencingpsychoticsymptomsandwho maybedistressedbythem.
5.Conclusion
When conducting cross-cultural replication research, the significance of employing properly translated and validated questionnaires cannotbeover-emphasized.Given thatany data collectedisonlyasrobustastheinstrumentthatcollectedthem,a poorly-translatedandvalidatedquestionnaireleadsinevitablyto corruptdata,whichinturnresultsinspuriousandunsubstantiated conclusions. OurChinese CAPE-C15enriches the predominately WesternCAPEliterature,andcouldfacilitatePPresearchaswellas earlydetectionand interventionintheChinesepopulation.Our resultsarerelevanttotheyoungpopulationtargetedinthestudy, butfurtherstudiesareneededtoconfirmtheextenttowhichthey generalizetoolderpopulations.
Thestrengthofthisstudyliesinthemeticuloustranslationand validationprocedures.Thisisthefirsttranslationvalidationreport ofCAPEwithcomprehensivedocumentationofscorereliability, factorstructure,convergent-divergentvalidity,aswellasacut-off score that indicated clinical needs in a community sample. However,thepresentstudywasnotimmunetoproblemsinherent toself-reportmeasures,includingsocialdesirabilityandover-or under-reporting of symptoms. Possible cultural influences on socialdesirabilityfurthercomplicatethetaskoftestdevelopment (Green,2009).Thatbeingsaid,CAPE-C150spsychometric robust-nessanduser-friendlinessrendersitanaccessibletoolforquick self-evaluationandself-referral,orforriskassessmentforreferral toprimaryhealthcareinnon-clinicalsettings(e.g.,schools).Allin all,ourthreestudiessupportedtheuseofCAPE-C15inresearch andcommunitysamplesofChineseadolescentand youngadult populations.
References
AmericanPsychologicalAssociation,2013.DiagnosticandStatisticalManualof MentalDisorders:DSM-5.AmericanPsychologicalAssociation.
Boonstra,N.,Wunderink,L.,Systema,S.,Wiersma,D.,2009.Improvingdetectionof firstepisodepsychosisbymentalhealthcareservicesusingself-report questionnaire.EarlyInterv.Psychiatry3(4),289–295.
Capra,C.,Kavanagh,D.J.,Hides,L.,Scott,J.G.,2015.Subtypesofpsychotic-like experiencesaredifferentiallyassociatedwithsuicidalideation,plansand attemptsinyoungadults.PsychiatryRes.228(3),894–898.
Cattell,R.B.,1978.TheScientificUseofFactorAnalysisinBehavioralandLife Sciences.PlenumPress,NewYork.
Chen,S.,Li,L.,2003.Re-testingreliability,validityandnormapplicatilityofSCL-90. Chin.J.Nerv.MentalDis.29(5),323–327.
Collin,G.,HulshoffPol,H.E.,Haijma,S.V.,Cahn,W.,Kahn,R.S.,vandenHeuvel,M.P., 2011.Impairedcerebellarfunctionalconnectivityinschizophreniapatientsand theirhealthysiblings.Front.Psychiatry2,73.
Derogatis,L.R.,1983.SCL–90–R:Administration,Scoring,andProceduresManualII. ClinicalPsychometricResearchBaltimore.
Feng,Z.Z.,Zhang,D.J.,2001.Studyonthevalidityofthesymptomcheck-list-90of Chineseversion.ActaAcad.Med.MilitarisTertiae23,481–483.
George,D.,Mallery,M.,2010.SPSSforWindowsStepbyStep:ASimpleGuideand Reference,17.0Update,10thed.Pearson,Boston.
Green,B.A.,2009.CultureandMentalHealth:SocioculturalInfluences,Theory,and Practice.Wiley-Blackwell,Oxford.
Häfner,H.,Maurer,K.,Trendler,G.,anderHeiden,W.,Schmidt,M.,Könnecke,R., 2005.Schizophreniaanddepression:challengingtheparadigmoftwoseparate diseases—acontrolledstudyofschizophrenia,depressionandhealthycontrols. Schiz.Res.77(1),11–24.doi:http://dx.doi.org/10.1016/j.schres.2005.01.004.
Hajian-Tilaki,K.,2013.Receiveroperatingcharacteristic(ROC)curveanalysisfor medicaldiagnostictestevaluation.Casp.J.Intern.Med.4(2),627–635.
Harrop,C.,Trower,P.,2001.Whydoesschizophreniadevelopatlateadolescence? Clin.Psychol.Rev.21(2),241–265.
Honings,S.,Drukker,M.,Groen,R.,vanOs,J.,2016.Psychoticexperiencesandriskof self-injuriousbehaviourinthegeneralpopulation:asystematicreviewand meta-analysis.Psychol.Med.46(02),237–251.
Hu,L.T.,Bentler,P.M.,1999.Cutoffcriteriaforfitindexesincovariancestructure analysis:conventionalcriteriaversusnewalternatives.Struct.Equ.Model.6(1), 1–55.
Hu,S.L.,2006.ConfirmatoryfactoranalysisandnormscomparisonofSCL-90in middleschoolstudents.Stud.Psychol.Behav.4(2),114–119.
IBMCorp,2011.IBMSPSSStatisticsforWindows,Version20.0.IBMCorp.,Armonk, NY.
Joreskog,K.G.,Sorbom,D.,2006.LISRELforWindows,9.1ed.ScientificSoftware International,Inc,Skokie,IL.
Kelleher,I.,Keeley,H.,Corcoran,P.,Lynch,F.,Fitzpatrick,C.,Devlin,N.,Molloy,C., Roddy,S.,Clarke,M.C.,Harley,M.,Arseneault,L.,Wasserman,C.,Carli,V., Sarchiapone,M.,Hoven,C.,Wasserman,D.,Cannon,M.,2012a.
Clinicopathologicalsignificanceofpsychoticexperiencesinnon-psychotic youngpeople:evidencefromfourpopulation-basedstudies.Br.J.Psychiatry 201(1),26–32.
Kelleher,I.,Lynch,F.,Harley,M.,etal.,2012b.Psychoticsymptomsinadolescence indexriskforsuicidalbehavior:findingsfrom2population-basedcase-control clinicalinterviewstudies.Arch.Gen.Psychiatry69(12),1277–1283.
Konings,M.,Bak,M.,Hanssen,M.,vanOs,J.,Krabbendam,L.,2006.Validityand reliabilityoftheCAPE:aself-reportinstrumentforthemeasurementof psychoticexperiencesinthegeneralpopulation.ActaPsychiatr.Scand.114(1), 55–61.
Lataster,T.,Verweij,K.,Viechtbauer,W.,2014.Effectofillnessexpressionand liabilityonfamilialassociationsofclinicalandsubclinicalpsychosis phenotypes.ActaPsychiatr.Scand.129(1),44–53.
Laurens,K.R.,West,S.A.,Murray,R.M.,Hodgins,S.,2008.Psychotic-likeexperiences andotherantecedentsofschizophreniainchildrenaged9–12years:a comparisonofethnicandmigrantgroupsintheUnitedKingdom.Psychol.Med. 38(8),1103–1111.
Liu,H.,Zhang,J.,2004.Normofsymptomchecklist(SCL-90)inchinesemiddle schoolstudents.Chin.MentalHealthJournal18(2),88–90.
Mark,W.,Toulopoulou,T.,2016.Psychometricpropertiesofcommunityassessment ofpsychicexperiences:reviewandmeta-analyses.Schizophr.Bull.42(1),34– 44.
Mossaheb,N.,Becker,J.,Schaefer,M.R.,Klier,C.M.,Schloegelhofer,M.,Papageorgiou, K.,Amminger,G.P.,2012.TheCommunityAssessmentofPsychicExperience (CAPE)questionnaireasascreening-instrumentinthedetectionofindividuals atultra-highriskforpsychosis.Schizophr.Res.141,210–214.
Murali,T.,Kumar,R.,2008.Symptomsofdepressioninschizophrenia.IndianJ.Med. Res.127,516–518.
Newcomer,J.W.,Faustman,W.O.,Yeh,W.,Csernansky,J.G.,1990.Distinguishing depressionandnegativesymptomsinunmedicatedpatientswith schizophrenia.PsychiatryRes.31(3),243–250.
Nunnally,J.C.,1978.PsychometricTheory,2nded.McGraw-Hill,NewYork.
Osborne,J.W.,Costello,A.,2009.Bestpracticesinexploratoryfactoranalysis:four recommendationsforgettingthemostfromyouranalysis.Pan-Pac.Manage. Rev.12(2),131–146.
Poulton,R.,Caspi,A.,Moffitt,T.E.,Cannon,M.,Murray,R.,Harrington,H.,2000. Children'sself-reportedpsychoticsymptomsandadultschizophreniform disorder:a15-yearlongitudinalstudy.Arch.Gen.Psychiatry57(11), 1053–1058.
Rössler,W.,Riecher-Rossler,A.,Angst,J.,Murray,R.,Gamma,A.,Eich,D.,vanOs,J., Gross,V.A.,2007.Psychoticexperiencesinthegeneralpopulation:a twenty-yearprospectivecommunitystudy.Schizophr.Res.92(1–3),1–14.
Sax,K.W.,Strakowski,S.M.,KeckJr.,P.E.,Upadhyaya,V.H.,West,S.A.,McElroy,S.L., 1996.Relationshipsamongnegative,positive,anddepressivesymptomsin schizophreniaandpsychoticdepression.Br.J.Psychiatry:J.MentalSci.168(1), 68–71.
Stefanis,N.C.,Hanssen,M.,Smirnis,N.K.,Avramopoulos,D.A.,Evdokimidis,I.K., Stefanis,C.N.,Verdoux,H.,vanOs,J.,2002.Evidencethatthreedimensionsof psychosishaveadistributioninthegeneralpopulation.Psyhcol.Med.32,347– 358.
Streiner,D.L.,Norman,G.R.,Cairney,J.,2014.HealthMeasurementScales:aPractical GuidetoTheirDevelopmentandUse.Oxforduniversitypress.
Swets,J.A.,Dawes,R.M.,Monahan,J.,2000.Psychologicalsciencecanimprove diagnosticdecisions.Psychol.Sci.PublicInterest1,1–26.
Vollema,M.G.,Ormel,J.,2000.Thereliabilityofthestructuredinterviewfor schizotypy-revised.Schizophr.Bull.26(3),619–629.
Werbeloff,N.,Drukker,M.,Dohrenwend,B.P.,Levav,I.,Yoffe,R.,vanOs,J.,Davidson, M.,Weiser,M.,2012.Self-reportedattenuatedpsychoticsymptomsas forerunnersofseverementaldisorderslaterinlife.Arch.Gen.Psychiatry69(5), 467–475.
vanOs,J.,Linscott,R.J.,Myin-Germeys,I.,Delespaul,P.,Krabbendam,L.,2009.A systematicreviewandmeta-analysisofthepsychosiscontinuum:evidencefor apsychosisproneness-persistence-impairmentmodelofpsychoticdisorder. Psychol.Med.39(2),179–195.
vandenBerg,S.M.,Paap,M.C.S.,Derks,E.M.,2013.GeneticRiskandOutcomeof Psychosis(GROUP)investigators,Usingmultidimensionalmodelingtocombine self-reportsymptomswithclinicaljudgmentofschizotypy.PsychiatryRes.206, 75–80.