Prediction of a single psychotic episode: A 7.5-year, prospective study in first-episode psychosis

Schizophrenia Research xxx (2010) xxx–xxx

SCHRES-04443; No of Pages 11

Contents lists available at ScienceDirect

Schizophrenia Research

j ourna l homepage: www.e lsev ie r.com/ locate /schres

Prediction of a single psychotic episode: A 7.5-year, prospective study infirst-episode psychosis

M. Alvarez-Jimenez a,b,⁎, J.F. Gleeson c,d, L.P. Henry a,b, S.M. Harrigan a,b, M.G. Harris e,G.P. Amminger a,b,g, E. Killackey a,b, A.R. Yung a,b,f, H. Herrman a,b, H.J. Jackson c, P.D. McGorry a,b

a Centre for Youth Mental Health, The University of Melbourne, Australiab Orygen Youth Health Research Centre, Australiac The Department of Psychology, The University of Melbourne, Australiad Northwestern Mental Health Program, Melbourne, Victoria, Australiae School of Population Health, The University of Queeensland, Brisbane, Australiaf School of Community Based Medicine, University of Manchester, United Kingdomg Department of Child and Adolescent Psychiatry, Medical University Vienna, Austria

a r t i c l e i n f o

⁎ Corresponding author. Centre for Youth Mental HeMelbourne, Orygen Youth Health Research Centre, 35,3054,Victoria,Melbourne,Australia. Tel.:+6193422805

E-mail address: [email protected] (M. Alv

0920-9964/$ – see front matter © 2010 Elsevier B.V.doi:10.1016/j.schres.2010.10.020

Please cite this article as: Alvarez-Jimenezfirst-episode psychosis, Schizophr. Res. (2

a b s t r a c t

Article history:Received 16 August 2010Received in revised form 28 September 2010Accepted 5 October 2010Available online xxxx

Background: Around 20% of patients who suffer from psychosis will experience a singlepsychotic episode (SPE), but relatively little is known about the characteristics and predictorsfor this group of patients. This study sought to: 1) characterise the subgroup of first-episodepsychosis (FEP) patients who experienced a SPE over a 7.5-year follow-up; and 2) to identifysignificant predictors for this subgroup independent of potential confounders.Methods: A representative sample of 413 FEP patients treated at a specialist early psychosis servicewere assessed at baseline and followed-up for 7.5 years. Binary logistic regression models wereemployed to investigate univariate and adjusted associations between baseline predictors andexperiencingaSPE.Resultswereadjusted for the influenceof knownprognostic factors forpsychosis.Results: Follow-up data was available for 274 participants. Forty-six (16.5%) achieved clinicalremission and experiencedno recurrence over the follow-upperiod. Durationof untreatedpsychosis(DUP) shorter than 60 days (OR=3.89, p=0.007), more rapid response to antipsychotic treatment(OR=0.33, p=0.019) and no parental loss (OR=5.25, p=0.045) significantly predicted a SPE. Theassociation remained significant after controlling for potential confounders.Conclusions: Early treatment (within two months of onset of psychotic symptoms) and socialsupport significantly reduce vulnerability to subsequent psychotic episodes. Future studies need toinvestigate the interplay between biological factors (i.e. sensitized dopaminergic system),environmental variables (i.e. exposure to trauma, stigma and discrimination), and psychologicalattributes (i.e. cognitive schemata) in order to elucidate the processes underlying the vulnerabilityto recurrent psychotic episodes.

© 2010 Elsevier B.V. All rights reserved.

Keywords:Early psychosisDuration of untreated psychosisSocial supportEarly intervention

1. Introduction

Despite previous longitudinal studies showing that around20% of patients who suffer from psychotic disorders will only

alth, The University ofPoplar Road, Parkville; fax:+61401772668.arez-Jimenez).

All rights reserved.

, M., et al., Prediction of010), doi:10.1016/j.sch

experience onepsychotic episode (Linszen et al., 2001; Shepherdet al., 1989; Wiersma et al., 1998), little research has beendedicated to identifying this subgroup of patients. There aremany potential benefits to both research and clinical practice inestablishing the factors associated with only experiencing asingle psychotic episode (SPE). Firstly, the identification of aseparate group of patients who achieve clinical remission with anon-relapsing course of illness may enable more benigntreatment approaches tailored to this subgroup's particularneeds, minimizing side effects associated with prolonged

a single psychotic episode: A 7.5-year, prospective study inres.2010.10.020

http://dx.doi.org/10.1016/j.schres.2010.10.020

mailto:[email protected]


http://www.sciencedirect.com/science/journal/09209964


2 M. Alvarez-Jimenez et al. / Schizophrenia Research xxx (2010) xxx–xxx

exposure to antipsychotic medication (Alvarez-Jimenez et al.,2008, 2009a,b; an der Heiden and Hafner, 2000; Bola andMosher, 2002; Carpenter and Heinrichs, 1981). Secondly,establishing treatment and environmental factors that have aneffect on the long-term course of the illness would shed light onmodifiable prognostic characteristics. Thiswould, in turn, informpublic health policies and the development of novel therapiestargeting these factors (van Os and Kapur, 2009).

A study that aims to discriminate factors associated withexperiencing a SPE from potential confounders needs to meetstrict methodological criteria. Firstly, a prospective first-episode psychosis (FEP) study is necessary in which partici-pants are recruited and assessed at first treatment contact, freefrom the confounding effects of previous treatment interven-tions or secondarymorbidity (Ram et al., 1992). Secondly, sucha study should include recruitment of an incident, multi-diagnostic FEP cohort, as opposed to convenience samples oflimited diagnostic range which are likely to create selectionbiases that impede generalizability of the findings to ‘realworld’ settings (Henry et al., 2007). Thirdly, the study shouldbe properly powered to investigate the influence of relevantpredictors independent of potential confounders. Finally, theduration of follow-up should extend beyond the critical period(i.e. the first 5 years after the onset of psychosis) after whichthe level of disability sustained, or recovery achieved, isthought to endure into the long-term (Birchwood et al.,1998; Crumlish et al., 2009).

Only a few longitudinal FEP studies have conductedprospective assessments beyond 5 years; however they haveexclusively focused on patients with schizophrenia (an derHeiden andHafner, 2000; Breier et al., 1991;Mason et al., 1995;Munk-Jorgensen and Mortensen, 1989; Wiersma et al., 1998),recruited convenience or small samples of newly admittedinpatients (Bottlender et al., 2003; Carpenter and Strauss, 1991;Linszen et al., 2001), or they have examined predictors ofrecovery without isolating the subset of patients who onlyexperience one psychotic episode (Crumlish et al., 2009;Robinson et al., 2004) or controlling for potential confounders(Robinson et al., 2004). To date, no single study has reliablycharacterized the group of patients who only experience onepsychotic episode or identified prognostic factors for thissubgroup independent of potential confounders.

In the present study we recruited a large, epidemiologicallyrepresentative cohort of FEPpatients presenting at a specializedFEP service followed themupover a 7.5-year period.We soughtto: 1) characterise the subgroup of FEP patients who experi-enced a SPE; 2) identify independent predictors of SPE aftercontrolling for potential confounders; and 3) evaluate anhypothesised model of the relationships between relevantpredictive factors and experiencing a SPE.

2. Methods

2.1. Design and setting

The studyparticipantswere recruited into the Early PsychosisPrevention and Intervention Centre (EPPIC) long-term follow-upstudy, a longitudinal 7.5-year follow-up study of consecutive FEPpatients admitted into EPPIC. The EPPIC programme is a front-line, youth-orientated, specialist mental health programmewhich provides comprehensive, community-based treatment

Please cite this article as: Alvarez-Jimenez, M., et al., Prediction offirst-episode psychosis, Schizophr. Res. (2010), doi:10.1016/j.sch

for FEP patients originating from a geographically definedcatchment area in metropolitan Melbourne, Australia with apopulation of approximately 800,000 (McGorry et al., 1996). Allother publicly funded mental health services in the area refertheir FEP patients to EPPIC. There are very fewprivate psychiatricservices in the area, including no private psychiatric inpatientsservices. Full details of the EPPIC long-term follow-up studydesign, sampling strategy, attrition and follow-up are availableelsewhere (Henry et al., 2007, 2010).

2.2. Sample and procedure

Patients from the EPPIC programmewere recruited betweenApril 1993 and July 1997. Inclusion criteria for entry to the studywere: 1) aged between 15 and 30 years; 2) a DSM-III-R (APA,1987) and from 1995, a DSM-IV (APA, 1994) diagnosis of apsychotic disorder (schizophrenia, schizophreniform disorder,schizoaffective disorder, delusional disorder, bipolar psychoticdisorder, major depressive disorder with psychotic features,brief reactive psychosis/brief psychosis and psychosis nototherwise specified); 3) informed consent for research partic-ipation; 4) living in the geographical catchment of the EPPICservice; 5) adequate English-language comprehension; and 6)experienced the first treated episode of psychosis with less than6 months of prior neuroleptic medication. Exclusion criteriawere: 1) primary organic mental syndrome; 2) intellectualdisability; 3) drug and/or alcohol induced psychosis; and 4)epilepsy. Assessments were conducted at multiple time pointsincluding: within the first few days following entry intotreatment (index presentation; T1); at the time of symptomremission or stabilization (median 8.9 weeks after indexpresentation; T2); approximately 6 months after stabilization(T3); 12 months after stabilization (T3); and amedian 7.5 yearsafter index presentation (T4). All assessments were conductedby interviewers with graduate psychology qualifications, whoreceived specific training in the administration of all instru-ments, and ongoing supervision to maintain reliability andconsistency. Very good to excellent inter-rater reliability wasfound for baseline measures including DSM diagnosis ofschizophrenia (kappa=0.92) and the onset and duration ofsymptoms (mean kappa=0.79). Inter-rater reliability for theprimary outcome variables at T4was assessed by calculating thepercentage of discrepancies between raters, with only 2% ofratings found to be discrepant (Henry et al., 2007). Raters wereblind todiagnostic informationandclinical ratings frompreviousassessments. The study was approved by relevant institutionalhuman research ethics committees.

2.3. Measures

2.3.1. Number of psychotic episodesThe main outcome variable was dichotomized into ‘experi-

enced a SPE’ vs. ‘experienced continuous or recurrent psychosis’.A psychotic episode was defined as a discrete period ofsymptomatology characterized by psychotic signs and symp-toms of hallucinations, delusions, cognitive disorganization,marked psychomotor disturbance, and/or grossly inappropriatebehaviour. Patients were classified as ‘experienced a SPE’ if they:1) experienced only one discrete psychotic episode with aduration not exceeding 12 months over the 7.5-year follow-upperiod (i.e. no recurrence); and 2) achieved complete remission



3M. Alvarez-Jimenez et al. / Schizophrenia Research xxx (2010) xxx–xxx

following the index episode. Remission was defined as a statefollowing a psychotic episode in which the participants werevirtually symptom-free and showed their usual premorbidpersonality over a minimum period of 4 weeks. In contrast,patients with two or more psychotic episodes, continuouspsychotic symptoms,multiplehospital admissions or incompleteclinical remission were categorized as ‘experienced continuousor recurrent psychosis’.

To determine the number of psychotic episodes throughoutthe7.5-year follow-upperiod theWHOLife Chart Schedule (LCS)(WHO, 1992) was employed. The LCS longitudinally assessesfour domains via a semi-structured interview. These includeresidency, employment, course of psychotic symptoms andservice utilization. LCS ratings were carried out on the basis ofinformation obtained from the patient, family members, mem-bers of the specialist treatment team or general practitioner andexamination of psychiatric/research medical records.

2.3.2. Predictor measuresWe considered predictors from four different domains,

based on previous research into prognostic factors for psycho-sis. These were sociodemographic and psychosocial variables,clinical variables, premorbid adjustment, andpsychopathology.

2.3.2.1. Clinical variables. At baseline, the Royal Park Multi-diagnostic Instrument for Psychosis (RPMIP) (McGorry et al.,1990a,b) was employed to evaluate illness duration compo-nents, psychiatric diagnosis, and other clinical and socio-demographic variables. The RPMIP is a comprehensive andvalid assessment procedure developed for the acute psychoticepisode using serial interviews and multiple informationsources (McGorry et al., 1990a,b). Onset of psychosis wasdefined as the emergence of the first sustained positivepsychotic symptom (i.e. the presence of delusions, hallucina-tions, clear-cut disorder of thinking/speech, or disorganised,bizarre or markedly inappropriate behaviour) at thresholdlevel, dated as precisely as possible to the nearest day, weekor month. Onset was additionally recorded as acute (symp-toms fully developed from a clearly non-symptomatic statewithin a week) or insidious (symptoms developed over amonth or longer, or the beginning of the episode was difficultto ascertain with accuracy). Age at onset was the patient's ageat the date of onset of psychosis. Duration of untreatedpsychosis (DUP) was defined as the number of days betweenthe onset of psychosis and the initiation of treatment, withthe latter defined as the first recorded date of admission oracceptance into the service. Previous research has demon-strated that a controlled reduction of DUP from a median of16 weeks to 5 weeks via an early detection programmeresulted in significantly lower symptom levels at treatmentinitiation and three months follow-up (Joa et al., 2008; Melleet al., 2004). Similarly, a recent 8-year follow-up report of aFEP cohort has established DUP cut-offs ranging from 1 to3 months for improved remission rates and symptom scores,respectively (Crumlish et al., 2009). Thus, to identify preciselya clinically meaningful cut-off point, DUP was further dividedinto four time points (30 days, 60 days, 90 days and 1 year).The onset of prodrome was defined as the earliest deviationfrom the patient's premorbid personality, behaviour or levelof functioning prior to the onset of psychotic symptoms. Theduration of prodrome was the period of time in days between


the onset of the prodrome and the onset of psychoticsymptoms. Duration of untreated illness (DUI) was the sumof the duration of prodrome and DUP. DUI was furtherdivided into three time points (6 months, 12 months and2 years) based on previously published longitudinal data(Crumlish et al., 2009). Time to treatment response wasdefined as the number of days between the onset ofpharmacological treatment and achievement of clinicalremission. Duration of hospital admission was defined asthe number of days of psychiatric hospitalization during theindex episode.

DSM-III-R diagnoses obtained from the RPMIP were subse-quently converted to DSM-IV (APA, 1994) and categorized intotwo primary groups: schizophrenia spectrum disorders (i.e.Schizophrenia disorder, Schizophreniform disorder, Schizoaffec-tive disorder, Delusional disorder, Psychotic disorder NOS, andBrief reactive psychosis), and affective psychosis (i.e. Bipolardisorder and Depressive psychosis). The RPMIP also assessedsymptomatology includingpresenceofmajordepressive episodeaccording to DSM-IV criteria, history of deliberate previous self-harm and prior violent history.

2.3.2.2. Sociodemographic and psychosocial variables. Baselinesociodemographic and psychosocial characteristics investi-gated included sex, age, employment status (employed orstudying vs. not-employed), marital status, living arrange-ments (family housing vs. living independently), level ofeducation (classified as tertiary education vs. primary/secondary education), country of birth (Australia vs. others),parents country of birth (classified as both parents born inAustralia vs. one or two immigrant parents), parental loss(categorized as having both parents alive vs. death of or lossof contact with one or both parents), presence of problemalcohol use and problem substance use (categorized as ‘no oroccasional use’; ‘moderate use’; ‘severe use’), psychosocialstressor in the year before onset, psychosocial stressorimmediately prior onset, family history of psychiatric illness,and family history of suicide.

2.3.2.3. Premorbid adjustment. Premorbid functioning wasassessed through the Premorbid Adjustment Scale (PAS)(Cannon-spoor et al., 1982). The PAS is designed to measurepremorbid functioning from a developmental perspective,conceptualizing good premorbid adjustment as the achieve-ment of age-appropriate developmental and social mile-stones. To avoid confounding factors a procedure wasdeveloped to estimate premorbid functioning which focusedupon the period preceding the prodromal phase (Harriganet al., 2003).

2.3.2.4. Psychopathology. Psychopathologywas assessed via the18-item Brief Psychiatric Rating Scale (BPRS) (Lukoff et al.,1986; Overall and Gorham, 1962), which provides severityratings across a broad range of psychotic and nonpsychoticsymptoms and, and the Schedule for the Assessment ofNegative Symptoms (SANS) (Andreasen, 1984), a scale specif-ically developed to measure negative symptoms. A positivesymptom subscale (BPRS-PS; range of possible scores 0–24)was derived from the BPRS, comprising items measuringconceptual disorganization, hallucinations, unusual thoughtcontent and suspiciousness.




2.4. Statistical analysis

Data screening was conducted to determine the presenceof outliers, non-normality, heterogeneity of variance, andheteroscedasticity. When data deviated from the normalcurve, logarithmic transformations were conducted. Descrip-tive statistics are presented for untransformed data. Samplerepresentativeness was assessed using logistic regression tocompare the study sample with the non-completer group ona range of baseline demographic and clinical characteristics.

The primary analysis focused on the identification ofpredictors of ‘experienced a SPE’. A two-stage analysisprocedure was implemented. Firstly, a series of univariatelogistic regression models were fitted to assess the unadjustedassociations between each candidate predictor and ‘experi-enced a SPE’. The relevance of each variable was assessedfollowing the Hosmer–Lemeshow method (Hosmer andLesmeshow, 2000). This method recommends excludingvariables with little predictive value (i.e. pN0.15) from furtheranalysis in order to obtain stable and clinically relevantmodels.Subsequently, all possible multilevel logistic regressionmodelsfor multiple predictor variables were fitted and ranked byMallows' Cp statistic (a measure of goodness of fit of themodel). The performance of the models was additionallyexamined via the area under the receiver operation character-istic curve (Hanley and McNeil, 1982) and the Nagelkerke's R2

statistic (a measure of the proportion of explained variation inthe logistic model) (Nagelkerle, 1991). Next, the significantvariables resulting from the analysis were force-entered into amultilevel logistic regression model before the inclusion of arange of potential confounders. These were: age of onset,gender, premorbid adjustment, severity of symptoms at entry,suddenness of onset, and baseline diagnosis. Age of onset andgender were specified as male gender and earlier onset havebeen found to predict poor outcome (Davidson andMcGlashan,1997). Premorbid adjustment was specified as it has beenshown to be a prognostic factor in early onset psychosis(Verdoux et al., 2001). Severity of symptoms at admission wasincluded in light of previous findings on the associationbetween baseline symptoms and long-term outcomes (Gift etal., 1980). Suddenness of onset was included as insidious onsetpredicts poor long-term outcomes (Wiersma et al., 1998).Finally, baselinediagnosiswas specified as affective psychosis isthought to have a more favourable prognosis compared withnon-affective psychosis (Langfeldt, 1969). Premorbid adjust-ment and severity of symptoms at entry contained 21.6 and6.6% missing data, respectively. We employed multiple impu-tation to account for missing values on these variables(Mackinnon, 2010). In contrast to listwise deletion, which hasthe disadvantages of reduced statistical power and potentialbias of estimates, multiple imputation permits analysis ofcomplete data by calculating estimates ofmissing datawith theuse of other variables in the model as predictors (Little andRubin, 1987; Schafer and Graham, 2002). Details of the dataimputation procedure are provided in the online supplement.

All regression analyseswere repeated substituting DUP forDUI to examine whether including DUI improved theexplanatory power of the models. Given the high correlationbetween DUP and DUI (Spearman's rho=0.64, pb0.001),they were not entered simultaneously in the same model. Allmodels were validated graphically and analytically.


Next, results from previous analyses regarding therelationship between potential predictors of experiencing aSPE were examined using structural equation modelling(SEM). SEM encompasses the use of path models thatmathematically represent the causal influences on thevariables of interest. Constructed models are tested for fitagainst the data. To quantify the overall fit of the hypothe-sised models to the empirical data the maximum likelihoodmethod which generates a χ2 goodness-of-fit statistic isemployed. For any proposed model a lower, non-significant(p≥0.05) χ2 value indicates minimum significant differencesbetween the hypothesizedmodel data and the empirical data.Model fit was further examined by using two additionalindices of goodness-of-fit, the Comparative Fit Index (CFI)(Bentler and Bonett, 1980) and the Root Mean Square Error ofApproximation (RMSEA) (Browne and Cudeck, 1993). Good-to-excellent model fit is indicated if the following criteria aremet: χ2 difference test ≥0.05, CFIN0.95 and RMSEAb0.05(Browne and Cudeck, 1993).

Descriptive and regression analyses were conducted usingthe Statistical Package for Social Sciences (SPSS) for Win-dows, version 18.0 (SPSS, Chicago, IL). For SEM analyses theAMOS 8.0 SEM software was employed.

3. Results

3.1. Follow-up and characteristics of the sample

Four hundred and thirteen patients agreed to take part inthe study, and reliable information on number of psychoticepisodeswere available for 274 participants from the 7.5-yearfollow-up interviews. Loss to follow-up (n=139) was due torefusal to be interviewed (n=79; 19.1%), previous refusal ofall further follow-up (n=3; 0.7%), deceased (n=26; 2.3%),failure to locate the participant for interview (n=27; 6.5%),or lack of reliable data on number of psychotic episodesdespite being interviewed (n=4; 0.9%). Results from theanalyses examining the representativeness of the sampleshowed no significant differences on baseline demographicand clinical characteristics between completers and non-completers (Table 1).

Of the 274 included participants, 46 (16.5%) onlyexperienced a SPE over the 7.5-year period of enrolment inthe study, and 228 (83.5%) experienced continuous or severalpsychotic episodes.

3.2. Univariate predictors of ‘experienced a SPE’

Potential predictors of ‘experienced a SPE’ were firstlyinvestigated using univariate logistic regression. As depictedin Table 2, statistically significant associations were foundbetween ‘experienced a SPE’ and ten of the candidatepredictors. Specifically, subjects with post-secondary educa-tion were twice as likely to only experience a SPE as thosewith primary/secondary education. Likewise, subjects withacute onset of psychosis were three times more likely to onlyexperience a SPE compared with those with insidious onset.Conversely, the loss of one or both parents increased the riskof having more than one psychotic episode fourfold. Theoptimal cut-off points for DUI (i.e. 2 years) and DUP (i.e.60 days) increased the risk of experiencing two or more


http://dx.doi.org/10.1111/j.13652010.02274.x


Table 1Comparison of study sample and non-completer groups (total n=413) on baseline demographic and clinical variables with means (SDs) or percentages (n).

Baseline characteristics Non-completers (n=139) Study sample (n=274) OR p-value a

Sociodemographic and psychosocial variablesAge at service entry 21.8 (3.6) 21.8 (3.4) 0.99 0.922Gender

% male 78.4 (109) 70.1 (192) 1.55 0.073Marital status

% never married 83.5 (116) 85.4 (234) 1.16 0.603Education

% post-secondary 20.1 (28) 25.3 (69) 0.74 0.247Work status

% unemployed 13.7 (19) 15.7 (43) 0.85 0.587Living arrangements b

Family housing 68.1 (92) 70.7 (191) 0.88 0.592Country of birth

% Immigrant 23.0 (32) 16.4 (45) 0.65 0.105Patient' country of birth

% Immigrant parents 69.3 (95) 63.9 (175) 0.78 0.271Parental loss 15.1 (21) 15.0 (41) 0.98 0.969Psychosocial stressor immediately prior onset b 20.0 (26) 22.1 (58) 1.13 0.641Psychosocial stressor in the year before onset b 46.3 (62) 48.9 (132) 1.11 0.620Illicit drug use

No or occasional use c 36.7 (51) 39.4 (108)Moderate use 22.3 (31) 21.2 (58) 0.88 0.658Severe use 41.0 (57) 39.4 (108) 0.89 0.639

Alcohol useNo or occasional use c 56.1 (78) 56.2 (154)Moderate use 26.6 (37) 28.8 (79) 1.08 0.747Severe use 17.3 (24) 15.0 (41) 0.86 0.620

Clinical variablesAge at onset of psychosis symptoms 21.3 (3.6) 21.3 (3.3) 0.99 0.914DSM-IV diagnosis

% Affective psychosis d 28.1 (39) 24.5 (67) 0.83 0.428Duration of untreated psychosis (DUP) in days e

Median 61.0 60.0 1.00 0.966% Longer than 60 days 50.4 (70) 49.6 (136) 0.89 0.889

Duration of prodromal symptoms in days e

Median 157.0 171.5 0.99 0.942Duration of untreated illness (DUI) in days e

Median 293 342 1.03 0.816% Longer than 2 years 25.9 (36) 28.1 (77) 1.11 0.635

Time to treatment response in days e

Median 46 45 0.95 0.842Duration of hospital admission b,e

Median 14.5 17.0 1.05 0.716Suddenness of onset b

% Acute 48.9 (68) 52.4 (143) 0.87 0.871Presence of major depressive episode 33.8 (47) 32.5 (89) 0.94 0.786Family history of psychiatric illness b 59.1 (81) 60.5 (164) 1.06 0.786Family history of suicide b 5.1 (7) 4.9 (13) 0.93 0.897Previous self-harm b 22.1 (30) 23.5 (64) 1.08 0.740Previous violent history b 23.2 (32) 17.1 (46) 0.68 0.141

Premorbid adjustment f 0.32 (0.19) 0.29 (0.18) 0.45 0.225

PsychopathologyBPRS total score g 29.9 (9.4) 29.1 (9.3) 0.99 0.668BPRS psychotic subscale g 10.8 (3.5) 10.9 (3.7) 1.01 0.713SANS g 24.9 (15.5) 23.8 (14.6) 0.99 0.526

a Based on univariate logistic regression.b Subject numbers vary between n=389 and n=412 for these variables.c Reference category against which other categories are compared.d Non affective psychosis' diagnostic category includes Schizophrenia, Schizophreniform, Schizoaffective, Delusional, Psychotic disorder NOS (not otherwise

specified) and Brief psychotic disorder. Affective psychosis' diagnostic category includes Bipolar disorder and Depressive psychosis.e Log-transformed due to positive skewness but untransformed data are presented here.f Estimation premorbid functioning prior to the onset of the prodromal period (Harrigan et al., 2003). Included subjects n=314.g Subject numbers vary between n=335 and n=387 for these variables.


Please cite this article as: Alvarez-Jimenez, M., et al., Prediction of a single psychotic episode: A 7.5-year, prospective study infirst-episode psychosis, Schizophr. Res. (2010), doi:10.1016/j.schres.2010.10.020


Table 2Baseline psychosocial and clinical predictors of ‘experiencing a single psychotic episode’ over 8 years follow-up (n=274), with means (SDs) or percentages (n) a.

Baseline characteristics Two or more psychotic episodes (n=228) Single psychotic episode (n=46) OR p-value a

Sociodemographic and psychosocial variablesAge at service entry 21.9 (3.4) 21.2 (3.1) 0.94 0.224Gender

% female 28.9 (66) 34.8 (16) 1.30 0.431Marital status

% married 10.1 (23) 13.3 (6) 1.33 0.553Education

% post-secondary 22.5 (51) 39.1 (18) 2.21 0.020Work status

% employed 82.9 (189) 91.3 (42) 2.16 0.161Living arrangements b

% Living with family 71.9 (161) 65.2 (30) 0.73 0.367Patient country of birth

% No immigrant 83.3 (190) 84.8 (39) 1.11 0.809Parents' country of birth

% No immigrant 34.2 (78) 45.7 (21) 1.61 0.143No parental loss 82.9 (189) 95.7 (44) 4.54 0.042No psychosocial stressor immediately prior onset b 78.5 (172) 75.0 (33) 0.82 0.606No psychosocial stressor in the year before onset b 51.6 (116) 48.9 (22) 0.89 0.744Illicit drug use

No or occasional use c,d 39.5 (90) 39.1 (18)Moderate use 21.5 (49) 19.6 (9) 0.91 0.848Severe use 39.0 (89) 41.3 (19) 1.06 0.857

Alcohol useNo or occasional use c 57.0 (130) 52.2 (24)Moderate use 27.6 (63) 34.8 (16) 1.37 0.372Severe use 15.4 (35) 13.0 (6) 0.92 0.881

Clinical variablesAge at onset of psychosis symptoms 21.3 (3.4) 21.0 (3.2) 0.96 0.512DSM-IV diagnosis

% Affective psychosis d 19.7 (45) 47.8 (22) 3.72 b0.001Duration of untreated psychosis (DUP) in days e

Median 62.5 21.0 0.43 0.001% Shorter than 30 days 35.5 (81) 65.2 (30) 3.40 b0.001% Shorter than 60 days 44.3 (101) 80.4 (37) 5.16 b0.001% Shorter than 90 days 56.1 (128) 82.6 (38) 3.71 0.001% Shorter than 1 year 86.0 (196) 93.5 (43) 2.34 0.175

Duration of prodromal symptoms in days e

Median 182.5 118.0 0.70 0.046Duration of untreated illness (DUI) in days e

Median 365.0 142.5 0.44 b0.001% Shorter than 6 month 32.0 (73) 54.3 (25) 2.52 0.005% Shorter than 1 year 51.3 (117) 76.1 (35) 3.01 0.003% Shorter than 2 years 68.4 (156) 89.1 (41) 3.78 0.007

Time to treatment response in days e

Median 52.0 24.0 0.29 0.001Duration of hospital admission b,e

Median 18.0 16.0 1.14 0.563Suddenness of onset b

% Acute 48.5 (110) 71.7 (33) 2.70 0.005Absence of major depressive episode 67.1 (153) 69.6 (32) 1.12 0.745No family history of psychiatric illness b 38.2 (86) 45.7 (21) 1.35 0.349No family history of suicide b 94.6 (210) 97.8 (45) 2.57 0.370No previous self-harm b 74.4 (169) 86.7 (39) 2.23 0.084No previous violent history b 82.5 (184) 84.8 (39) 1.18 0.710

Premorbid adjustment f 0.31 (0.17) 0.24 (0.19) 0.12 0.050

PsychopathologyBPRS total score g 29.0 (8.9) 29.7 (11.0) 1.00 0.672BPRS psychotic subscale g 10.9 (3.7) 11.0 (4.0) 1.00 0.871SANS g 24.7 (14.6) 19.5 (13.6) 0.97 0.041

a Based on univariate logistic regression.b Subject numbers vary between n=257 and n=273 for these variables.c Reference category against which other categories are compared.d Non affective psychosis' diagnostic category includes Schizophrenia, Schizophreniform, Schizoaffective, Delusional, Psychotic disorder NOS (not otherwise

specified) and Brief psychotic disorder. Affective psychosis' diagnostic category includes Bipolar disorder and Depressive psychosis.e Log-transformed due to positive skewness but untransformed data are presented here.f Estimation premorbid functioning prior to the onset of the prodromal period (Harrigan et al., 2003). Included subjects n=218.g Subject numbers vary between n=223 and n=256 for these variables.


Please cite this article as: Alvarez-Jimenez, M., et al., Prediction of a single psychotic episode: A 7.5-year, prospective study infirst-episode psychosis, Schizophr. Res. (2010), doi:10.1016/j.schres.2010.10.020



episodes threefold and fivefold, respectively. Those subjectswith a diagnosis of schizophrenia spectrum disorders were3.7 times as likely to experience more than one psychoticepisode compared with those with a diagnosis of affectivepsychosis; similarly, participants with poorer premorbidadjustment were more likely to experience several psychoticepisodes. In addition, longer time to treatment responseincreased the risk of experiencing more than one psychoticepisode threefold. Finally, longer duration of prodromalsymptoms and higher levels of negative symptoms atbaseline significantly increased the risk of suffering contin-uous or several psychotic episodes.

3.3. Multivariate predictors of ‘experienced a SPE’

Secondly, variables with predictive power (pb0.15) wereretained for the multivariate analysis. Parents' country of birth,history of previous self-harm and poor premorbid socialadjustment or work history were associated with experienceda SPE below the a priori specified probability level of 0.15 andwere additionally included in subsequent multivariate analy-ses. Next, all possible multilevel logistic regression models forrelevant candidates and first-order interactionswere fitted andtheir performance assessed by Mallow's Cp statistic, the areaunder the receiver operation characteristic curve, and theNagelkerke's R2 statistic. Multivariate models identified fourfactors which were significant predictors of experienced a SPE(OR for time to treatment response was reversed for ease ofinterpretation): DUPb60 days (OR=3.49; 95% CI 1.54–7.94;p=0.003), more rapid time to treatment response (OR=0.4095% CI 0.17–0.91; p=0.029; reversed OR=2.49; 95% CI 1.09–5.64; p=0.029), no parental loss (OR=5.03; 95% CI 1.05–23.97; p=0.042) and a diagnosis of affective psychosis(OR=2.14; 95% CI 1.04–4.41; p=0.039). The model whichincluded thesevariableshadgoodpowers of discrimination andexplained a significantproportionof the variance of onlyhavinga SPE (Nagelkerke R2=0.21) (pb0.001) (Table 3). Each of thesignificant predictors were then force-entered into amultilevellogistic regression model before the introduction of potentialconfounders including suddenness of onset, premorbid adjust-ment, baseline psychopathology, age of onset, and gender. Inthe adjusted model, DUPb60 days, time to treatment responseand no parental loss remained unchanged and significantlyassociated with SPE, whereas a diagnosis of affective psychosiswas no longer significant (Table 3). The adjusted modelperformance (Nagelkerke R2=0.21) was similar to that of the

Table 3Multivariate model of predictive factors of ‘experiencing a single psychotic episode

Model variable a OR (95% CI) p-value

DUPb60 days 3.49 (1.54–7.94) 0.003Time to treatment response 0.40 (0.17–0.91) 0.029No parental loss 5.03 (1.05–23.97) 0.042Affective psychosis 2.14 (1.04–4.41) 0.039

a Best fit logistic regression model for ‘experiencing a single psychotic episodemultilevel logistic regression models for multiple predictor variables were fittedcharacteristic curve. For overall model including DUP, Time to treatment response, PCp=5.00 and area under the curve=0.76.

b Adjusted for sex (p=0.894), age of onset (p=0.586), premorbid adjustment (Subject number for the adjusted model n=274.


best model obtained in the multivariate analysis (NagelkerkeR2=0.22) which indicated little additional predictive value forthe control variables.

Similar predictors were obtained when analyses wererepeated replacing DUP with DUI, although DUI did not reachstatistical significance and the predictive power of the modelwas reduced from 21 to 18% of the variance of ‘experienced aSPE’ (DUIb2 years (OR=2.62; 95% CI 0.96 to 7.17; p=0.060),shorter time to treatment response (OR=0.37 95% CI 0.16–0.83; p=0.016; reversed OR=2.69; 95% CI 1.20–6.02;p=0.016), no parental loss (OR=4.94; 95% CI 1.05–23.16;p=0.043), diagnosis of affective psychosis (OR=2.67; 95% CI1.32–5.36; p=0.006)). In the adjusted model controlling forthe influence of potential confounders time to treatmentresponse (p=0.025), parental loss (p=0.044) and a diagnosisof affective psychosis (p=0.023) remained significant, whileDUI was not significantly associated with SPE (p=0.123).

3.4. Multivariate relationship between relevant predictors andexperiencing a SPE

Finally, given that DUP has been associated with treatmentresponse and a range of psychosocial variables (Norman andMalla, 2001; Verdoux et al., 1998) we examined the relation-ships between DUP and relevant predictors including time totreatment response, educational level, parental loss andparents' immigration status. The latter was specified becauseprior multivariate analysis showed it was somewhat related toSPE and immigration has been suggested to influence theexpression of psychotic disorders (van Os and Kapur, 2009).DUP shorter than 60 days was significantly associated withhigher educational level (χ2=4.44, p=0.03) and shorter timeto treatment response (t=4.13, pb0.001), whereas DUPshorter than 60 days, no parental loss (χ2=1.64, p=0.19)and no parental immigration (χ2=0.03, p=0.95) wereunrelated. Subsequently, we applied SEM to test this modelmathematically. The hypothesized model is presented in Fig. 1.Baseline educational level and time to treatment responsewerepredicted to mediate the effects of DUP on experiencing a SPE.Furthermore both, no parental loss and parental immigrationwere postulated to independently predict SPE over a 7.5-yearfollow-up period. The hypothesized model provided anexcellent fit for the data as suggested by a non-significant χ2

(χ2=4.50, df=8, p=0.809) and the CFI and RMSEA indices(1.00 and 0.00 respectively) (Fig. 1).

’ in FEP patients over 8 years follow-up.

AOR b (95% CI) p-value b SE Wald

3.89 (1.46–10.39) 0.007 0.41 8.950.33 (0.13–0.83) 0.019 0.41 4.775.25 (1.03–26.68) 0.045 0.79 4.121.97 (0.93–4.19) 0.076 0.36 4.28

’ vs. ‘experiencing two or more psychotic episodes’ (n=273). All possibleand ranked the models by Mallow's Cp statistic and the receiver operatingarental loss and Diagnosis, χ2=36.78, df=4, pb0.001, Nagelkerke R2=0.21,

p=0.520), baseline BPRS (p=0.261) and suddenness of onset (p=0.670).



Fig. 1. Hypothesized model of the relationship between ‘Experienced a single psychotic episode (SPE)’ and predictor variables.


4. Discussion

This is the first study, to our knowledge, to characterise thesubgroup of FEP patients who experience a single psychoticepisode (SPE) and remain in clinical remission during a long-term follow-up extending beyond the critical period forpsychosis. Both environmental and treatment-related variableswere shown to be relevant predictors of experiencing SPE.Shorter DUP,more rapid time to response to treatment, and ‘noparental loss’ emerged as the strongest independent predictorsof ‘experienced a SPE’, the association remaining strong andsignificant even after controlling for potential confounders.Results from multivariate and SEM analysis provided furthersupport to the hypothesis that treatment delay affects bothpsychosocial and treatment response variables while environ-mental variablesmay exert an independent effect on the courseof the illness.

4.1. DUP, DUI, time to treatment response and experiencing aSPE

Most studies of FEP have found an association betweenlong DUP and short-term outcomes (Marshall et al., 2005;Norman and Malla, 2001). Recent research has also foundDUP to predict positive symptoms and social functioning innon-affective psychosis at 7.5-year follow-up (Crumlish et al.,2009). However, the nature of this relationship remains to befully elucidated. While some have postulated that prolongedDUP has a toxic effect through an unknown neurologic orpsychosocial mechanism (McGlashan, 1999), others arguethat DUP may only be an epiphenomenon of pre-existingprognostic factors that contribute to poor outcome (Owenset al., 2010). In the present study, longer DUP emerged as astrong predictor of experiencing continuous or severalpsychotic episodes even when accounting for potentialconfounders including sex, age of onset, diagnosis, acuity ofonset, premorbid adjustment and baseline psychopathology.In addition, the finding that the association between DUI andexperiencing a SPE was not significant after controlling forpotential confounders further supports the contention thatDUP does not merely reflect pre-existing prognostic char-acteristics. Thus, the predictive power of DUP was unlikely tobe an epiphenomenon.

Results from SEM suggested that both psychosocial andbiological variables may mediate treatment delay andexperiencing a SPE. Firstly, it has been argued that if DUP


has a toxic effect upon the brain treatment delay would affectspeed of initial response as well as long-term outcomes(Keshavan and Amirsadri, 2007; Keshavan et al., 2003;Norman and Malla, 2001). Our finding that DUP waspositively associated with time to treatment responsewhich, in turn, predicted experiencing prolonged or severalpsychotic episodes provides support to this hypothesis. Onthe other hand, DUP has been suggested to generatedestructive psychological processes which lead to poorlong-term prognosis (Melle et al., 2004). We found evidencefor the association between longer DUP and lower level ofeducation which was further related to experiencing morethan one psychotic episode. Thus, the untreated psychosismay disrupt the developmental process by increasing socialproblems and stigma (Alvarez-Jimenez et al., 2009a,b, 2010),leading to a loss of potentially protective factors, such aseducation and social support. DUP may also foster socialdisadvantage and isolation which have been shown toincrease risk for psychosis (Morgan et al., 2008; Selten andCantor-Graae, 2005; Veling et al., 2007) and increase theexposure to social stress and experiences of trauma anddiscrimination. Experiences of social defeat and powerless-ness have been linked to a paranoid way of thinking (Janssenet al., 2003; Kuipers et al., 2006) and the expectation thatdistressing interpersonal interactions will continuously occur(Bentall and Fernyhough, 2008). When this is paired with theunderlying sensitisation of the dopaminergic system (Bentalland Fernyhough, 2008; Moutoussis et al., 2007; Selten andCantor-Graae, 2005), it could generate an ongoing vulnera-bility to experiencing psychotic episodes.

This is the first study to report that FEP patients with longerDUP have a higher risk of experiencing multiple psychoticepisodes during the first 7.5 years of the course of psychosis.Interestingly, the optimal cut-off point for DUP was 60 days,with the largest effect size in terms of risk (≥4). This finding isimportant as themedianDUPobserved in this study is amongstthe shorter DUPs previously reported (Marshall et al., 2005;Norman andMalla, 2001). When taken together these findingsindicate that the long-termharmful effects, either psychosocial,biological or both caused by DUP occur principally in the first60 days after onset. These results parallel those of Drake et al.(2000), who found that reducing DUP from 6 months to1 month produced clinical gains similar to reducing treatmentdelay from 6 years to 1 year. In addition, this would explainwhy previous reports, with considerably longer DUPs andchosen cut-off points further from the onset of psychosis failed




to find an association between DUP and relapse (Loebel et al.,1992; Marshall et al., 2005; Robinson et al., 1999).

The present findings provide support to the recentlyformulated clinical stagingmodel where treatment effects arethought to be greatest when delivered as early as possible(McGorry, 2007). Two primary assumptions underlie thismodel. First, patients in the earliest stages of psychosis have abetter response to treatment and a better prognosis thanthose in later stages. Second, treatments offered in the earlystages should be more benign as well as more effective(McGorry et al., 2006). Our results show that reducing DUP tominimum levels enhances treatment effectiveness and mayhave an effect on the long-term course of psychosis byreducing vulnerability to suffer multi-episode psychosis.Importantly, these results further suggest that there is asmall time interval where interventions can effect substantiallong-term changes in the course of the psychosis.

4.2. Parental loss, immigration and experiencing a SPE

Previous cohort studies have found that being anunwanted child increased the risk for developing psychosislater in life fourfold (Myhrman et al., 1996); likewise, earlyseparation from or adverse relationship with parents havebeen found to significantly increase the risk of psychosis ingenetically vulnerable children (Agid et al., 1999). This is thefirst study to link the loss of one or both parents to the long-term course of psychosis by showing a strong associationbetween parental loss and risk for experiencing multi-episode psychosis independently of potential confounders.Likewise, results from SEM indicated that parents' immigra-tion status may be associated with ‘experiencing a SPE’.

The mechanisms underlying this association may be bothpsychological and biological. Firstly, insecure attachment as aresult of parental separation, victimization, or separation inmigrant populations (Morgan et al., 2007) has been associ-ated with paranoid thinking (Bentall and Fernyhough, 2008;MacBeth et al., 2008). Bentall and Fernyhough (2008) arguedthat psychological mechanisms such as low self-esteem anddifficulty in trusting others may mediate the relationshipbetween insecure attachment and paranoia. Furthermore, theloss of one or both parents may increase the exposure tostressful life events in the absence of the potential buffer ofsocial support. On the other hand, maternal deprivation hasbeen linked to an enduring sensitization of the dopaminesystem (Hall et al., 1999), which paired with a cognitiveschemata of interpersonal distrust, may induce a continuingvulnerability to experience several psychotic episodes.

4.3. Strengths and limitations

Themain strengths of this study are its design and sampling.The study sample comprised a multidiagnositic, epidemiolog-ically representative ‘realworld’ FEP cohort recruited soon afterillness onset (Henry et al., 2007). We employed standardizedand valid instruments to assess all outcomes variables.Furthermore, we examined the effect of potential predictorsof ‘experiencing a SPE’ independent of a broad range ofpotential confounders reportedbyprevious literature includingsex, age of onset, premorbid adjustment, diagnosis, baselinepsychopathology and acuity of onset.


There are a number of limitations to this work that need tobe acknowledged. Firstly, of the 413 participants initiallyrecruited into the study 139 could not be assessed at 7.5-yearfollow-up. That said, our attrition rate is relatively lowconsidering the cohort size and length of follow-up andsomewhat lower than that of the OPUS sample at 5 years(Bertelsen et al., 2008) and a recent 7.5-year FEP study(Crumlish et al., 2009). In addition, analyses found noevidence of selection bias due to attrition over the follow-up period. Secondly, the 7.5 year assessment point relied onthe recall of participants of previous psychotic episodes. Toaddress this issue, considerable efforts were made to obtaininformation from object and reliable sources such aspsychiatric records, psychiatric treating teams, generalpracticitioners, friends and family members. Thirdly, itcould be argued that specific diagnostic categories such asdrug-induced psychosis and brief psychotic disorders maynot provide a representative picture of FEP. Of note, the studysample included no subjects with drug-induced psychosisand only 3 (1.1%) with brief psychotic disorder. Additionally,analyses found no evidence of association between drug useprior to first onset of psychosis and experiencing a SPE andcontrolled for the effects of affective vs. non-affectivepsychosis. Thus, our findings are likely to be generalisableto other incident FEP cohorts. Conversely, the initial exclusionof participants with poor English-language comprehensionmay have artificially reduced the number of immigrant FEPpatients with low social support. Fourthly, although not alimitation as such, the current study set out to examine theearly predictors of experiencing a SPE and therefore cannotcomment upon the influence of either social or psychiatricfactors in the interim period. In addition, we examined aplausible model of the hypothesized relationships betweenbaseline predictors and SPE. Further studies should investi-gate the goodness of fit of the postulated model againstcompeting explanations which include additional or con-trasting variables. Fifthly, baseline predictors explained arelatively small proportion of variance of SPE (around 20%),which suggest that other factors such as personality char-acteristics, stigma, resilience variables, cognitive schemata,interim life events and their interaction may play animportant role in determining the course of psychosis. Finally,while the follow-up extended well beyond the consideredcritical period of psychosis (Birchwood et al., 1998; Carpenterand Strauss, 1991), the possibility of experiencing furtherpsychotic episodes beyond the first 7.5 years after onset ofpsychosis cannot be completely disregarded.

4.4. Clinical implications and future research

This study demonstrated that prompt delivery of treat-ment (i.e. within 60 days of onset) and less social disadvan-tage independently predicted experiencing a SPE eight yearsafter illness presentation. This emphasizes the importance ofearly intervention efforts which are likely to improve thelong-term course of psychosis—beyond the level that would beaccomplished by similar interventions initiated later—by meansof reducing the biological and psychosocial damages of theuntreated illness. Likewise, treatment strategies which addressthe needs of those socially disadvantaged and tackle theexperiences of victimization, stigma and low-self-esteem




suffered by FEP patients could significantly reduce the risk forfurther psychotic episodes. Ultimately, the present findingssuggest that more benign and cost-effective treatmentapproaches which minimize the side effects of prolongedexposure to antipsychotic medication should be evaluated forFEP patients with short illness duration, rapid response toantipsychotic treatment and adequate social and family support.

Our study shows that environmental, biological variablesand their interaction may play a pivotal role in shaping thecourse of psychosis. Further research should investigate theeffects of the first weeks of untreated psychosis on both brainabnormalities (e.g. sensitization of the basal ganglia dopaminesystem) and cognitive schemata relevant to experiences oftrauma, stigma and discrimination as well as their complexinterplay. This would shed light on the processes underlyingthe vulnerability to experience several or continuous psychoticepisodes. Moreover, future research should focus on theidentification of biological and psychological resilience factorsin those who, being exposed to disadvantaged social environ-ments, do not suffer subsequent psychotic episodes. The basisfor this resilience could help to devise interventions whichtarget psychological and biological processes known to berelevant for the course of psychosis.

Role of funding sourceThis work was supported by the Victorian Health Promotion Foundation

(grant number 91-0084C) Victoria, Australia; the Australian National Healthand Medical Research Council (grant number 350241) Canberra, AustralianCapital Territory, Australia; and the Colonial Foundation, Victoria, Australia.The funding bodies had no further role in study design; in the collection,analysis and interpretation of data; in the writing of the report; and in thedecision to submit the paper for publication.

ContributorsMA-J and JF-G designed the study and wrote the first draft of the

manuscript. E-K and AR-Y contributed to the design of the study and criticallyrevised the manuscript. PD-M, HJ-J, H-H, LP-H, MG-H, GP-A and SM-Hsupervised the research project and critically revised the manuscript. Allauthors contributed to and have approved the final manuscript.

Conflict of interestThe authors report no additional financial or other affiliation relevant to

the subject of this article.

AcknowledgmentsThis work was supported by the Victorian Health Promotion Foundation

(grant number 91-0084C) Victoria, Australia; the Australian National Healthand Medical Research Council (grant number 350241) Canberra, AustralianCapital Territory, Australia; and the Colonial Foundation, Victoria, Australia.The funding bodies did not contribute to the design and conduct of the study;collection, management, analysis, and interpretation of the data; andpreparation, review, or approval of the manuscript. The authors wish tothank all of the young people and family members who participated in thestudy and the dedicated research team.

References

Agid,O., Shapira, B., Zislin, J., Ritsner,M.,Hanin,B.,Murad,H., Troudart, T., Bloch,M.,Heresco-Levy, U., Lerer, B., 1999. Environment and vulnerability to majorpsychiatric illness: a case control study of early parental loss in majordepression, bipolar disorder and schizophrenia. Mol. Psychiatry 4 (2),163–172.

Alvarez-Jimenez, M., Gonzalez-Blanch, C., Crespo-Facorro, B., Hetrick, S.,Rodriguez-Sanchez, J.M., Perez-Iglesias, R., Vazquez-Barquero, J.L., 2008.Antipsychotic-induced weight gain in chronic and first-episode psy-chotic disorders: a systematic critical reappraisal. CNS Drugs 22 (7),547–562.


Alvarez-Jimenez, M., Gleeson, J.F., Cotton, S., Wade, D., Gee, D., Pearce, T.,Crisp, K., Spiliotacopoulos, D., Newman, B., McGorry, P.D., 2009a.Predictors of adherence to cognitive-behavioural therapy in first-episodepsychosis. Can. J. Psychiatry 54 (10), 710–718.

Alvarez-Jimenez, M., Parker, A.G., Hetrick, S.E., McGorry, P.D., Gleeson, J.F.,2009b. Preventing the second episode: a systematic review and meta-analysis of psychosocial and pharmacological trials in first-episodepsychosis. Schizophr. Bull. (Nov 9, Electronic publication ahead of print).

Alvarez-Jimenez, M., Gleeson, J.F., Cotton, S.M., Wade, D., Crisp, K., Yap, M.B.,McGorry, P.D., 2010. Differential predictors of critical commentsand emotional over-involvement in first-episode psychosis. Psychol.Med. 40 (1), 63–72.

an der Heiden, W., Hafner, H., 2000. The epidemiology of onset and course ofschizophrenia. Eur. Arch. Psychiatry Clin. Neurosci. 250 (6), 292–303.

Andreasen, N.C., 1984. Scale for Assessment of Negative Symptoms (SANS).University of Iowa, Iowa City.

APA, 1987. Diagnostic and Statistical Manual of Mental Disorders, 3rd edn.American Psychiatric Association, Washington DC.

APA, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4 edn.American Psychiatric Association, Washington, DC.

Bentall, R.P., Fernyhough, C., 2008. Social predictors of psychotic experiences:specificity and psychological mechanisms. Schizophr. Bull. 34 (6),1012–1020.

Bentler, P.M., Bonett, D.G., 1980. Significance tests and goodness of fit in theanalysis of covariance-structures. Psychol. Bull. 88 (3), 588–606.

Bertelsen, M., Jeppesen, P., Petersen, L., Thorup, A., Ohlenschlaeger, J., Quach, P.L.,Christensen, T.O., Krarup, G., Jorgensen, P., Nordentoft, M., 2008. Course ofillness in a sample of 265 patients with first-episode psychosis-five-yearfollow-up of the Danish OPUS trial. Schizophr. Res. 107 (2–3), 173–178.

Birchwood, M., Todd, P., Jackson, C., 1998. Early intervention in psychosis.The critical period hypothesis. Br. J. Psychiatry Suppl. 172 (33), 53–59.

Bola, J.R., Mosher, L.R., 2002. At issue: predicting drug-free treatment responsein acute psychosis from the Soteria project. Schizophr. Bull. 28 (4),559–575.

Bottlender, R., Sato, T., Jager, M., Wegener, U., Wittmann, J., Strauss, A.,Moller, H.J., 2003. The impact of the duration of untreated psychosisprior to first psychiatric admission on the 15-year outcome inschizophrenia. Schizophr. Res. 62 (1–2), 37–44.

Breier, A., Schreiber, J.L., Dyer, J., Pickar, D., 1991. National Institute of MentalHealth longitudinal study of chronic schizophrenia. Prognosis andpredictors of outcome. Arch. Gen. Psychiatry 48 (3), 239–246.

Browne, M.W., Cudeck, R., 1993. Alternative ways of assesing model fit. In:Bollen, K.A., Long, J.S. (Eds.), Testing Structural Equation Models. SAGE,Thousand Oaks, CA, pp. 136–162.

Cannon-spoor, H.E., Potkin, S.G., Wyatt, R.J., 1982. Measurement ofpremorbid adjustment in chronic schizophrenia. Schizophr. Bull. 8 (3),470–484.

Carpenter Jr., W.T., Heinrichs, D.W., 1981. Treatment-relevant subtypes ofschizophrenia. J. Nerv. Ment. Dis. 169 (2), 113–119.

Carpenter Jr., W.T., Strauss, J.S., 1991. The prediction of outcome inschizophrenia. IV: Eleven-year follow-up of the Washington IPSS cohort.J. Nerv. Ment. Dis. 179 (9), 517–525.

Crumlish, N., Whitty, P., Clarke, M., Browne, S., Kamali, M., Gervin, M.,McTigue, O., Kinsella, A., Waddington, J.L., Larkin, C., O'Callaghan, E.,2009. Beyond the critical period: longitudinal study of 8-year outcome infirst-episode non-affective psychosis. Br. J. Psychiatry 194 (1), 18–24.

Davidson, L., McGlashan, T.H., 1997. The varied outcomes of schizophrenia.Can. J. Psychiatry 42 (1), 34–43.

Drake, R.J., Haley, C.J., Akhtar, S., Lewis, S.W., 2000. Causes and consequences ofduration of untreated psychosis in schizophrenia. Br. J. Psychiatry 177,511–515.

Gift, T.E., Strauss, J.S., Kokes, R.F., Harder, D.W., Ritzler, B.A., 1980.Schizophrenia: affect and outcome. Am. J. Psychiatry 137 (5), 580–585.

Hall, F.S., Wilkinson, L.S., Humby, T., Robbins, T.W., 1999. Maternaldeprivation of neonatal rats produces enduring changes in dopaminefunction. Synapse 32 (1), 37–43.

Hanley, J.A., McNeil, B.J., 1982. The meaning and use of the area under thereceiver operating characteristic (ROC). Radiology 143, 29–36.

Harrigan, S.M., McGorry, P.D., Krstev, H., 2003. Does treatment delay in first-episode psychosis really matter? Psychol. Med. 33 (1), 97–110.

Henry, L., Harris, M.G., Amminger, G.P., Yuen, H.P., Harrigan, S.M., Lambert,M., Conus, P., Schwartz, O., Prosser, A., Farrelly, S., Purcell, R., Herrman, H.,Jackson, H.J., McGorry, P.D., 2007. The EPPIC long term follow-up study offirst episode psychosis: methodology and baseline characteristics. Early.Interv. Psychiatry 1, 49–60.

Henry, L.P., Amminger, G.P., Harris, M.G., Yuen, H.P., Harrigan, S.M., Prosser, A.L.,Schwartz, O.S., Farrelly, S.E., Herrman, H., Jackson, H.J., McGorry, P.D., 2010.The EPPIC follow-up study of first-episode psychosis: longer-termclinical and functional outcome 7 years after index admission. J. Clin.Psychiatry 71 (6), 716–728.




Hosmer, D.W., Lesmeshow, S., 2000. Applied Logistic Regression, 2nd edn.Wiley, New York.

Janssen, I., Hanssen, M., Bak, M., Bijl, R.V., de Graaf, R., Vollebergh, W.,McKenzie, K., van Os, J., 2003. Discrimination and delusional ideation. Br.J. Psychiatry 182, 71–76.

Joa, I., Johannessen, J.O., Auestad, B., Friis, S., McGlashan, T., Melle, I.,Opjordsmoen, S., Simonsen, E., Vaglum, P., Larsen, T.K., 2008. The key toreducing duration of untreated first psychosis: information campaigns.Schizophr. Bull. 34 (3), 466–472.

Keshavan, M.S., Amirsadri, A., 2007. Early intervention in schizophrenia:current and future perspectives. Curr. Psychiatry Rep. 9 (4), 325–328.

Keshavan, M.S., Haas, G., Miewald, J., Montrose, D.M., Reddy, R., Schooler, N.R.,Sweeney, J.A., 2003. Prolonged untreated illness duration from prodromalonset predicts outcome in first episode psychoses. Schizophr. Bull. 29 (4),757–769.

Kuipers, E., Garety, P., Fowler, D., Freeman, D., Dunn, G., Bebbington, P., 2006.Cognitive, emotional, and social processes in psychosis: refiningcognitive behavioral therapy for persistent positive symptoms. Schi-zophr. Bull. 32 (Suppl 1), S24–S31.

Langfeldt, G., 1969. Schizophrenia: diagnosis and prognosis. Behav. Sci. 14 (3),173–182.

Linszen, D., Dingemans, P., Lenior, M., 2001. Early intervention and a five yearfollow up in young adults with a short duration of untreated psychosis:ethical implications. Schizophr. Res. 51 (1), 55–61.

Little, R.J.A., Rubin, D.B., 1987. Statistical Analysis With Missing Data. JohnWiley & Sons Inc., New York, NY.

Loebel, A.D., Lieberman, J.A., Alvir, J.M., Mayerhoff, D.I., Geisler, S.H.,Szymanski, S.R., 1992. Duration of psychosis and outcome in first-episode schizophrenia. Am. J. Psychiatry 149 (9), 1183–1188.

Lukoff, D., Liberman, R.P., Nuechterlein, K.H., 1986. Symptom monitoring inthe rehabilitation of schizophrenic patients. Schizophr. Bull. 12 (4),578–593.

MacBeth, A., Schwannauer, M., Gumley, A., 2008. The association betweenattachment style, social mentalities, and paranoid ideation: an analoguestudy. Psychol. Psychother. 81 (Pt 1), 79–93.

Mackinnon, A., 2010. The use and reporting of multiple imputation in medicalresearch - a review. J. Intern. Med. doi:10.1111/j.1365-2796.2010.02274.x.(Aug 16, Electronic publication ahead of print).

Marshall, M., Lewis, S., Lockwood, A., Drake, R., Jones, P., Croudace, T., 2005.Association between duration of untreated psychosis and outcome incohorts of first-episode patients: a systematic review. Arch. Gen.Psychiatry 62 (9), 975–983.

Mason, P., Harrison, G., Glazebrook, C., Medley, I., Dalkin, T., Croudace, T.,1995. Characteristics of outcome in schizophrenia at 13 years. Br. J.Psychiatry 167 (5), 596–603.

McGlashan, T.H., 1999. Duration of untreated psychosis in first-episodeschizophrenia: marker or determinant of course? Biol. Psychiatry 46 (7),899–907.

McGorry, P.D., 2007. Issues for DSM-V: clinical staging: a heuristic pathwayto valid nosology and safer, more effective treatment in psychiatry. Am. J.Psychiatry 164 (6), 859–860.

McGorry, P.D., Copolov, D.L., Singh, B.S., 1990a. Royal Park MultidiagnosticInstrument for Psychosis: Part I. Rationale and review. Schizophr. Bull.16 (3), 501–515.

McGorry, P.D., Singh, B.S., Copolov, D.L., Kaplan, I., Dossetor, C.R., van Riel, R.J.,1990b. Royal Park Multidiagnostic Instrument for Psychosis: Part II.Development, reliability, and validity. Schizophr. Bull. 16 (3), 517–536.

McGorry, P.D., Edwards, J., Mihalopoulos, C., Harrigan, S.M., Jackson, H.J.,1996. EPPIC: an evolving system of early detection and optimalmanagement. Schizophr. Bull. 22 (2), 305–326.

McGorry, P.D., Hickie, I.B., Yung, A.R., Pantelis, C., Jackson, H.J., 2006. Clinicalstaging of psychiatric disorders: a heuristic framework for choosingearlier, safer and more effective interventions. Aust. N. Z. J. Psychiatry40 (8), 616–622.

Melle, I., Larsen, T.K., Haahr, U., Friis, S., Johannessen, J.O., Opjordsmoen, S.,Simonsen, E., Rund, B.R., Vaglum, P., McGlashan, T., 2004. Reducing the


duration of untreated first-episode psychosis: effects on clinicalpresentation. Arch. Gen. Psychiatry 61 (2), 143–150.

Morgan, C., Kirkbride, J., Leff, J., Craig, T., Hutchinson, G., McKenzie, K.,Morgan, K., Dazzan, P., Doody, G.A., Jones, P., Murray, R., Fearon, P., 2007.Parental separation, loss and psychosis in different ethnic groups: a case-control study. Psychol. Med. 37 (4), 495–503.

Morgan, C., Kirkbride, J., Hutchinson, G., Craig, T., Morgan, K., Dazzan, P.,Boydell, J., Doody, G.A., Jones, P.B., Murray, R.M., Leff, J., Fearon, P., 2008.Cumulative social disadvantage, ethnicity and first-episode psychosis: acase-control study. Psychol. Med. 38 (12), 1701–1715.

Moutoussis, M., Williams, J., Dayan, P., Bentall, R.P., 2007. Persecutorydelusions and the conditioned avoidance paradigm: towards anintegration of the psychology and biology of paranoia. Cogn. Neuropsy-chiatry 12 (6), 495–510.

Munk-Jorgensen, P., Mortensen, P.B., 1989. Schizophrenia: a 13-yearfollow-up. Diagnostic and psychopathological aspects. Acta Psychiatr.Scand. 79 (4), 391–399.

Myhrman, A., Rantakallio, P., Isohanni, M., Jones, P., Partanen, U., 1996.Unwantedness of a pregnancy and schizophrenia in the child. Br. J.Psychiatry 169 (5), 637–640.

Nagelkerle, N., 1991. A note on general definition of the coeficient ofdetermiantion. Biometrika 78, 791–792.

Norman, R.M., Malla, A.K., 2001. Duration of untreated psychosis: a criticalexamination of the concept and its importance. Psychol. Med. 31 (3),381–400.

Overall, J.E., Gorham, D.R., 1962. The brief psychiatric rating-scale. Psychol.Rep. 10 (3), 799–812.

Owens, D.C., Johnstone, E.C., Miller, P., Macmillan, J.F., Crow, T.J., 2010.Duration of untreated illness and outcome in schizophrenia: test ofpredictions in relation to relapse risk. Br. J. Psychiatry 196, 296–301.

Ram, R., Bromet, E.J., Eaton, W.W., Pato, C., Schwartz, J.E., 1992. The naturalcourse of schizophrenia: a review of first-admission studies. Schizophr.Bull. 18 (2), 185–207.

Robinson, D., Woerner, M.G., Alvir, J.M., Bilder, R., Goldman, R., Geisler, S.,Koreen, A., Sheitman, B., Chakos, M., Mayerhoff, D., Lieberman, J.A., 1999.Predictors of relapse following response from a first episode ofschizophrenia or schizoaffective disorder. Arch. Gen. Psychiatry 56 (3),241–247.

Robinson, D.G., Woerner, M.G., McMeniman, M., Mendelowitz, A., Bilder, R.M.,2004. Symptomatic and functional recovery from a first episode ofschizophrenia or schizoaffective disorder. Am. J. Psychiatry 161 (3),473–479.

Schafer, J.L., Graham, J.W., 2002. Missing data: our view of the state of the art.Psychol. Meth. 7 (2), 147–177.

Selten, J.P., Cantor-Graae, E., 2005. Social defeat: risk factor for schizophre-nia? Br. J. Psychiatry 187, 101–102.

Shepherd, M., Watt, D., Falloon, I., Smeeton, N., 1989. The natural history ofschizophrenia: a five-year follow-up study of outcome and prediction ina representative sample of schizophrenics. Psychol. Med. Monogr. Suppl.15, 1–46.

van Os, J., Kapur, S., 2009. Schizophrenia. Lancet 374 (9690), 635–645.Veling, W., Selten, J.P., Susser, E., Laan, W., Mackenbach, J.P., Hoek, H.W.,

2007. Discrimination and the incidence of psychotic disorders amongethnic minorities in The Netherlands. Int. J. Epidemiol. 36 (4), 761–768.

Verdoux, H., Bergey, C., Assens, F., Abalan, F., Gonzales, B., Pauillac, P.,Fournet, O., Liraud, F., Beaussier, J., Gaussares, C., Etchegaray, B.,Bourgeois, M., van Os, J., 1998. Prediction of duration of psychosis beforefirst admission. Eur. Psychiatry 13 (7), 346–352.

Verdoux, H., Liraud, F., Bergey, C., Assens, F., Abalan, F., van Os, J., 2001. Is theassociation between duration of untreated psychosis and outcomeconfounded? A two year follow-up study of first-admitted patients.Schizophr. Res. 49 (3), 231–241.

WHO, 1992. Life Chart Schedule. World Health Organization, Geneva.Wiersma, D., Nienhuis, F.J., Slooff, C.J., Giel, R., 1998. Natural course of

schizophrenic disorders: a 15-year followup of a Dutch incidence cohort.Schizophr. Bull. 24 (1), 75–85.



Prediction of a single psychotic episode: A 7.5-year, prospective study in first-episode psychosis

Documents

Transcript of Prediction of a single psychotic episode: A 7.5-year, prospective study in first-episode psychosis