PSYCHIATRIC SERVICES http://ps.psychiatryonline.org September 2005 Vol. 56 No. 911008844

Diagnostic stability is the de-gree to which a diagnosis re-mains constant at subse-quent patient assessments (1). Thediagnosis of a psychotic disorder isbased on the presence or absence ofcharacteristic symptoms. However,the presence of such symptoms variesduring the course and treatment ofthese illnesses, which raises the ques-tion of how stable a diagnosis of psy-chosis remains over time.

Diagnostic instability is an impor-

tant issue, because it has potential im-plications in terms of service planningand resource allocation. For example,the services required for treating pa-tients with drug-induced psychosesare likely to differ from those re-quired for treating patients withnonsubstance-related psychotic dis-orders. Similarly, patients with schiz-ophrenia are likely to require differ-ent medical and psychosocial treat-ments compared with patients whohave affective psychoses, in which

negative symptoms are less likely topersist after the acute phase of illness.

Several readmission studies haveexamined the stability of clinical diag-noses over the course of illness (29).However, such studies are limited byinadequate reliability of clinical diag-noses and the bias that is inherent insampling rehospitalized patients (10).Consequently, other authors exam-ined diagnostic stability prospectively(1,1118) with varying durations offollow-up. These studies concludedthat schizophrenia is the most stableinitial diagnosis (about 90 percent ofinitial diagnoses are retained at fol-low-up), followed by affective psy-chosis (around 80 percent). Schizo-phreniform disorder, schizoaffectivedisorder, and psychosis not otherwisespecified are the least stable initialdiagnoses.

Even though these studies repre-sented an improvement in methodol-ogy compared with earlier studies,they were limited by the fact that theyinvolved first-admission patients only,which raises the question of whetherthese findings are applicable to all pa-tients with psychosis. The most con-clusive way to achieve diagnostic sta-bility is to conduct a prospective lon-gitudinal study of incident cases.However, such studies are difficult toconduct because of the time requiredto accrue a sample that is largeenough for investigating factors relat-ed to diagnostic change. Two previousstudies (19,20) examined inpatientsand outpatients who experienced a

Diagnostic Stability Four Years After a First Episode of PsychosisPPeetteerr WWhhiittttyy,, MM..DD..,, MM..RR..CC..PPssyycchh..MMaarryy CCllaarrkkee,, MM..DD..,, MM..RR..CC..PPssyycchh..OOrrffhhllaaiitthh MMccTTiigguuee,, MM..BB..,, MM..RR..CC..PPssyycchh..SStteepphheenn BBrroowwnnee,, MM..DD..,, MM..RR..CC..PPssyycchh..MMooyyaaaadd KKaammaallii,, MM..BB..,, MM..RR..CC..PPssyycchh..CCoonnaallll LLaarrkkiinn,, MM..BB..,, FF..RR..CC..PPssyycchh..EEaaddbbhhaarrdd OOCCaallllaagghhaann,, MM..DD..,, FF..RR..CC..PPssyycchh..

Dr. Whitty, Dr. McTigue, Dr. Kamali, and Professor OCallaghan are affiliated with Clu-ain Mhuire Family Centre in Blackrock, County Dublin, Ireland. Professor OCallaghanis also with St. John of God Hospital in Stillorgan, County Dublin, with which Dr. Clarkeand Dr. Larkin are affiliated. Dr. Browne is with Waterford Regional Hospital in Coun-ty Waterford, Ireland. Send correspondence to Professor OCallaghan at Stanley Re-search Unit, Cluain Mhuire Family Centre, Newtownpark Avenue, Blackrock, CountyDublin, Ireland (e-mail, eadbhard@gmail.com).

Objective: The objective of this study was to determine the stability of a di-agnosis of psychosis four years after the first-episode diagnosis. Methods:The study was a prospective four-year follow-up study (1995 to 1999) of147 patients with schizophrenia, affective disorder, and other psychoseswho presented with a first episode of psychosis in an epidemiologiccatchment area in Ireland. All diagnoses were made on the basis of theStructured Clinical Interview for DSM-IV. Results: One quarter of thepatients evidenced a change in diagnosis at follow-up. The most com-mon change was to a diagnosis of schizophrenia. The positive predictivevalues of schizophrenia and bipolar affective disorder were 97 percentand 80 percent, respectively. Fewer years spent in education, lower lev-els of initial psychopathology, and presence of comorbid alcohol or sub-stance abuse were associated with change in diagnosis at follow-up.Conclusions: Among the diagnoses studied, schizophrenia was the moststable diagnosis after four years. The greatest instability occurred in thecategories of drug-induced psychosis and psychosis not otherwise spec-ified. (Psychiatric Services 56:10841088, 2005)

oca.qxd 8/22/2005 9:47 AM Page 1084

first episode of psychosis and report-ed findings similar to those above.However, those studies did notspecifically determine the factors as-sociated with or predictive of diag-nostic instability.

Thus we sought to determine diag-nostic stability over the first fouryears of illness in a sample of patientswho presented with a first episode ofpsychosis to a catchment area servicein Dublin, Ireland. We examined thestability of the diagnosis at a numberof levels. First, we sought to establishhow many patients who received a di-agnosis at first presentation retainedthis diagnosis at follow-up. Next, weestablished how many patients with agiven diagnosis at follow-up had alsoreceived that diagnosis at first presen-tation. Finally, we sought to deter-mine what factors predicted a changein diagnosis at four-year follow-up.

MethodsThe study was conducted between1995 and 1999 at Cluain MhuireFamily Centre, a catchment areaservice that provides community-based psychiatric care for an urbanpopulation (approximately 165,000)in County Dublin, and at St. John ofGod Hospital in County Dublin (21).The ethics committee of the hospital,which serves both institutions, ap-proved the study. First-episode psy-chosis was defined as a first-everpresentation to any psychiatric serv-ice with a psychotic episode. Patientswho commenced antipsychotic med-ication before referral were includedin the study, provided that they werebeing treated for a first psychoticepisode and their treatment had notcommenced more than 30 days be-fore assessment.

Patients were followed up fouryears after the initial presentation,and all patients gave informed con-sent to participate in the initial andfollow-up studies. Patients were re-assessed across the same clinicalmeasures, which included a diagnos-tic interview by an investigator whowas blinded to the original diagnosisand the initial assessments. Interraterreliability was established by the con-comitant examination of ten patientswith psychosis or suspected psychosisby study evaluators. The rater who

conducted follow-up assessments wasblinded to the initial diagnosis, whichreduced the chance of test-retest reli-abilitythat is, patients who weregiven a diagnosis at one time point bya rater are more likely to test thesame on subsequent assessments bythe same rater.

The follow-up was conducted bymeans of face-to-face interviews andby contacting treating teams, pa-tients relatives and carers, and gener-al practitioners as well as reviewingcase notes for additional informationrelevant to diagnosis.

AssessmentsAll diagnoses were made on the basisof the Structured Clinical Interviewfor DSM-IV (SCID) (22). In addi-tion, we assessed patients across anumber of variables to determine thefactors associated with a change in di-agnosis. These variables were dividedinto four groups: background charac-teristics (age at first presentation,gender, years spent in education,marital status, social class at first pres-entation, presence or absence ofmental retardation, and presence orabsence of a personality disorder),clinical history (Global Assessment ofFunctioning [GAF] score at first

presentation, previous treatment withan antipsychotic medication, lifetimehistory of substance use disorder, andduration of untreated initial psy-chosis), initial hospitalization charac-teristics (duration of first hospitaliza-tion, scores on the Positive and Nega-tive Symptoms Scale [PANSS] [23] atfirst presentation, current substanceuse disorder, admission status [tem-porary or voluntary], and presence orabsence of private health insurance),and outcome characteristics (GAFscore at four years, total number ofadmissions at follow-up, social andoccupational functioning at fouryears, psychopathology at four years,medication status, and insight).

Data analysisWe divided the sample into three di-agnostic subgroups: schizophrenia, af-fective disorders, and other psychoses(drug-induced psychosis, delusionaldisorder, psychosis not otherwisespecified, and schizoaffective disor-der). We evaluated diagnostic stabilityby using two measures in each diag-nostic subgroup. The first measure,prospective consistency, refers to theproportion of patients who were in adiagnostic category at presentationwho had the same diagnosis at follow-up; this value corresponds to the posi-tive predictive value, assuming follow-up diagnosis as the standard. The sec-ond measure, retrospective consisten-cy, refers to the proportion of patientswith a given diagnosis at follow-upwho had received that diagnosis atfirst presentation; this value corre-sponds to sensitivity.

We determined the reasons behinda change in diagnosis by grouping pa-tients into two groups for analysis:those whose diagnosis changed be-tween first presentation and follow-up, and those whose diagnosis waschanged to schizophrenia. We com-pared the patients in each group byusing t tests for continuous variablesand chi square tests for categoricalvariables. Only variables that weresignificantly associated with a changein diagnosis were entered into a di-rect logistic regression analysis to de-termine which factors were signifi-cantly associated with a diagnosticchange at follow-up. All data were an-alyzed with use of SPSS (24).

PSYCHIATRIC SERVICES http://ps.psychiatryonline.org September 2005 Vol. 56 No. 9 11008855

The

presence

of characteristic

symptoms of a psychotic

disorder varies throughout

the illness, which raises the

question of how stable a

diagnosis of psychosis

remains over

time.

oca.qxd 8/22/2005 9:47 AM Page 1085

ResultsA total of 171 patients were assessedat baseline and received diagnosesbased on the SCID I interview. Atfour-year follow-up, five patients haddied (four patients by suicide and onefrom natural causes); these patientsdiagnoses were organic psychosis(three patients), bipolar affective dis-order (one patient), and drug-in-duced psychosis (one patient). We ex-cluded cases of organic psychosis(one patient) from further analysisand made new diagnoses for 147 ofthe initial 165 patients at four years,which resulted in a follow-up rate of89 percent (Table 1). No statisticallysignificant difference was found be-tween the patients who received newdiagnoses and those who did not interms of age at first presentation, gen-der, duration of untreated initial psy-chosis, GAF score at first presenta-tion, psychopathology at first presen-tation, presence of substance use dis-order at first presentation, or initialtreatment setting (inpatient or outpa-tient). Of the 147 patients who werefollowed up, 131 (89 percent) re-ceived their follow-up diagnosis bymeans of face-to-face interviews.

SchizophreniaNinety patients received a diagnosisof schizophrenia or schizophreni-form disorder at first presentation.

Of these, 87 retained this diagnosis atfollow-up. This finding equates to aprospective consistency for a diagno-sis of schizophrenia or schizophreni-form disorder of 97 percent. We nextseparated out the subgroup of pa-tients with schizophreniform disor-der (N=15). Of the 75 patients whowere given a diagnosis of schizophre-nia at first presentation, 72 (96 per-cent) retained this diagnosis at fol-low-up. Of the 15 patients with schiz-ophreniform disorder, five retainedthis diagnosis at follow-up, for aprospective consistency rate of 33percent.

Of the 147 patients we followed up,106 received a diagnosis of schizo-phrenia or schizophreniform disor-der; of these, 87 received this diagno-sis at both first presentation and fol-low-up, for a retrospective consisten-cy of 82 percent. We repeated theanalysis after removing cases of schiz-ophreniform disorder (five patients).Among the 101 patients with schizo-phrenia, 71 (70 percent) who weregiven a diagnosis of schizophrenia atfollow-up had also received that diag-nosis at first presentation. All five pa-tients with schizophreniform disorderreceived that diagnosis at first presen-tation as well as at follow-up. Thusthe retrospective consistency for a di-agnosis of schizophreniform disorderwas 100 percent (five of five).

Bipolar affective disorder and affective psychosisThe prospective consistency of a diag-nosis of bipolar affective disorder was80 percent; 16 of 20 patients whowere given this diagnosis at first pres-entation retained it at follow-up. Atfollow-up, 21 patients received a di-agnosis of bipolar disorder, of whom16 had received that diagnosis at bothtime points. Thus the retrospectiveconsistency for a diagnosis of bipolaraffective disorder was 76 percent (16of 21). We next added the subgroupof patients who had major depressionwith psychotic features (N=11). Ofthe 31 patients with affective psy-chosis at first presentation, 25 re-tained that diagnosis at follow-up.Thus the prospective consistency of adiagnosis of affective psychosis was 81percent (25 of 31). At follow-up 29patients were given a diagnosis of anaffective psychosis. Of these, 24 re-ceived that diagnosis at both firstpresentation and follow-up, for a ret-rospective consistency of 83 percent.

Other diagnostic subgroupsOf the initial 26 patients with otherforms of psychosis, nine received thatdiagnosis at follow-up, for a prospec-tive consistency of 35 percent.Among patients who were assigned adiagnosis of drug-induced psychosisat first presentation (N=11), two (27

PSYCHIATRIC SERVICES http://ps.psychiatryonline.org September 2005 Vol. 56 No. 911008866

TTaabbllee 11

Diagnoses at first presentation and at four-year follow-up in a sample of 147 patients with a first episode of psychosis

Diagnosis at follow-up

PsychosisSchizo- Bipolar Major Drug- not oth- Schizo-phreniform affective depressive induced Delusional erwise affective

Diagnosis at first presentation Schizophrenia disorder disorder disorder psychosis disorder specified disorder

Schizophrenia (N=75) 72 1 1 1Schizophreniform disorder

(N=15) 10 5 Bipolar affective disorder

(N=20) 3 16 1Major depressive disorder

(N=11) 2 1 8 Drug-induced psychosis

(N=11) 6 2 2 1 Delusional disorder (N=12) 5 1 5 1Psychosis not otherwise

specified (N=3) 3 Schizoaffective disorder (N=0) Total (N=147) 101 5 21 8 2 6 1 3

oca.qxd 8/22/2005 9:47 AM Page 1086

percent) retained that diagnosis atfollow-up. Among patients who weregiven a diagnosis of delusional disor-der at first presentation (N=12), fiveretained that diagnosis at follow-up,for a prospective consistency of 38percent (five of 13).

The retrospective consistency of adiagnosis of other psychoses was 78percent (seven of nine). The retro-spective consistency for a diagnosis ofdrug-induced psychosis was 100 per-cent; the two patients who receivedthat diagnosis at follow-up also re-ceived it at first presentation. Of thesix patients with delusional disorderat follow-up, five received that diag-nosis at first presentation, for a retro-spective consistency of 83 percent.

Factors related to diagnostic instabilityTotal group. Of the 147 patients whowere followed up, 37 (25 percent)were given a different diagnosis at fol-low-up than at first presentation; thediagnoses of 110 patients (75 percent)remained stable. We compared thesetwo patient groups to identify the fac-tors that distinguished one groupfrom the other. The groups did notdiffer in terms of background charac-teristics. In terms of clinical history,patients whose diagnosis changed ev-idenced a statistically significantshorter duration of untreated initialpsychosis compared with those whosediagnosis remained stable (t=2.18,df=145, p=.03) and were more likelyto present with a lifetime history of al-cohol or substance abuse or depend-ence (2=4.58, df=1, p=.03). In addi-tion, a statistically significant differ-ence in initial hospitalization charac-teristics was observed between thetwo groups; patients whose diagnosischanged were more likely to evidencelower total symptom scores (t=2.40,df=144, p=.02) and to have a comor-bid diagnosis of alcohol or substanceabuse or dependence (2=4.21, df=1,p=.04) at first presentation. No signif-icant differences in outcome charac-teristics were noted between the twogroups.

Logistic regression analysis usingthe presence or absence of a changein diagnosis as the dependent variablewith psychopathology, current andlifetime history of substance abuse or

dependence, and duration of untreat-ed psychosis as the predictor vari-ables indicated that a change in diag-nosis at follow-up was significantly as-sociated with low levels of psy-chopathology at first presentation(=.03, Wald=4.29, p=.04); there wasa trend toward significance for ashorter duration of untreated initialpsychosis.

Schizophrenia. Of the 101 patientswho were given a diagnosis of schizo-phrenia at follow-up, 29 (20 percent)had not been given that diagnosis ini-tially, whereas for 72 (80 percent) thefollow-up diagnosis represented a sta-ble diagnosis. Ten patients evidenceda shift in diagnosis from schizophreni-form disorder to schizophrenia andwere excluded from further analysisbecause their cases reflected the nat-ural history of the disorder and not di-agnostic change per se.

The 19 patients whose diagnosischanged to schizophrenia evidenceda significantly higher initial GAFscore (t=3.24, df=89, p=.003) andcompleted fewer years of education(t=.48, df=89, p=.02) compared withthe patients with a stable diagnosis ofschizophrenia (N=72). In terms ofinitial hospitalization characteristics,a change to a diagnosis of schizophre-nia was associated with significantlyfewer positive symptoms (t=2.93,df=89, p=.01) and negative symptoms(t=3.93, df=89, p=.001) at first pres-entation. A change to a diagnosis ofschizophrenia was also significantlyassociated with a current (2=9.92,df=1, p=.01) and lifetime (2=6.10,df=1, p=.47) diagnosis of alcohol orsubstance abuse or dependence.

We next performed a logistic re-gression analysis to determinewhether these factors were significantpredictors of change to a diagnosis ofschizophrenia. This analysis revealedthat a shift to a diagnosis of schizo-phrenia was associated with feweryears of education (=.39, Wald=5.19, p=.02) and fewer negativesymptoms (=.18, Wald=5.11, p=.02)at first presentation.

DiscussionMain findingsThe main finding of this study was thatone-quarter of the patients who pre-sented with a first episode of psychosis

were given a different diagnosis atfour-year follow-up. The most com-mon change was to a diagnosis ofschizophrenia. Patients with an initialdiagnosis of drug-induced psychosis,psychosis not otherwise specified, orschizophreniform disorder were themost likely to evidence this change.Compared with patients who had astable diagnosis of schizophrenia,those who were given a diagnosis ofschizophrenia as a new diagnosis at fol-low-up completed fewer years of edu-cation and had lower levels of psy-chopathology at first presentation.These patients were also more likely tohave a comorbid diagnosis of alcoholor substance abuse or dependence.

Diagnostic subgroupsSchizophrenia was the most stable di-agnosis, followed by affective psy-chosis. Overall, both disorders dis-played high levels of stability, whichsupports the view that they are dis-tinct clinical entities (25). The mostunstable diagnoses were psychosisnot otherwise specified, schizo-phreniform disorder, and drug-in-duced psychosis. The low prospectiveconsistency of a diagnosis of drug-in-duced psychosis is of particular inter-est, because this finding suggests thatmost people who presented in thisfashion had an underlying major psy-chotic illness. Indeed, almost three-quarters of such patients evidenced ashift in diagnosis into a different cate-gory of psychotic illness at follow-up.However, it was beyond the scope ofthis study to determine the nature ofthe relationship between psychosisand drug abusethat is, whetherdrug abuse causes the psychotic ill-ness or, rather, leads to psychoticsymptoms in predisposed or vulnera-ble patients.

The prospective consistency (posi-tive predictive value) of schizophre-nia was higher than the retrospectiveconsistency (sensitivity), which wasan encouraging finding. At first pres-entation it is especially important tomake a diagnosis with a high degreeof certainty. Clinically, the major con-cern would be a low specificity (thatis, many false-positives), because pa-tients would be incorrectly labeled ashaving schizophrenia, which wouldhave major implications for patients

PSYCHIATRIC SERVICES http://ps.psychiatryonline.org September 2005 Vol. 56 No. 9 11008877

oca.qxd 8/22/2005 9:47 AM Page 1087

and their families. The reverse situa-tionlower sensitivity with manyfalse negativesis less important, be-cause clinical treatment is initiallybased on symptoms, not diagnosticcategories (19). An alternative methodof examining this outcome is to calcu-late the probability of falsely assigninga diagnosis of schizophrenia, as op-posed to another psychotic category,at onset (1 minus positive predictivevalue). Thus, for patients with schizo-phrenia, a clinician would make afalse-positive diagnosis for one ofevery 25 patients when using theSCID I interview. Among patientswith bipolar affective disorder, thefalse-positive rate is one in five.Among patients who present with afirst episode of psychosis, a clinicianwill make the wrong diagnosis in oneof every four assessments.

Factors related to diagnostic changeA change in diagnosis at follow-upwas associated with low levels of psy-chopathology, poor educational func-tioning, and presence of substanceabuse or dependence at first presen-tation. This finding suggests that pa-tients whose diagnosis changed toschizophrenia evidenced early warn-ing signs in terms of poor educationalfunctioning, which may reflect a pro-dromal phase of illness. However, thelow levels of psychopathology dis-played make it more difficult to cor-rectly differentiate diagnostic sub-groups at first presentation, thus lead-ing to uncertainty of diagnosis. An ad-ditional complicating factor is thepresence of substance abuse or de-pendence at first presentation, whichserves to cause further diagnostic un-certainty. This finding suggests thatpatients presenting with an initial di-agnosis of a drug-induced psychosisneed to be followed up carefully, be-cause a significant proportion of thesepatients will evidence a shift into adifferent category of psychosis, whichmay require long-term treatmentsand interventions.

The main limitation of this studywas the small number of patients inthe subgroup without schizophrenia,which makes interpretation of diag-nostic shift for these patients difficultbecause of reduced power. However,

this was a study that examined out-comes for all psychoses. Future stud-ies of diagnostic stability may focus onspecific diagnostic subgroups to in-crease the power of their conclusions.A further potential limitation of thisstudy is that these findings are appli-cable only to the relatively early phaseof illness, and reassessment at a latertime may reveal further patterns ofdiagnostic instability.

ConclusionsApproximately one-quarter of pa-tients who presented with a firstepisode of psychosis received a differ-ent diagnosis at four-year follow-up.Furthermore, changes in diagnosisover time were associated with sub-stance abuse and low levels of psy-chopathology at first presentation.The greatest instability occurred inthe less frequent diagnostic cate-gories of drug-induced psychosis andpsychosis not otherwise specified.

References1. Fennig S, Kovasznay B, Rich C, et al: Six-

month stability of psychiatric diagnoses infirst-admission patients with psychosis.American Journal of Psychiatry 151:12001208, 1994

2. Amara IB: Consistency in the diagnosis ofthe functional psychoses. Canadian Psychi-atric Association Journal 23:329336, 1978

3. Babigian HM, Gardner EA, Miles HC, etal: Diagnostic consistency and change in afollow-up study of 1215 patients. AmericanJournal of Psychiatry 121:895901, 1965

4. Chen YR, Swann AC, Burt DB: Stability ofdiagnosis in schizophrenia. American Jour-nal of Psychiatry 153:682686, 1996

5. Jorgensen P, Mortensen PB: Admissionpattern and diagnostic stability of patientswith functional psychosis in Denmark dur-ing a two-year observation period. ActaPsychiatrica Scandinavica 78:361365,1988

6. Kendell RE: The stability of psychiatric di-agnoses. British Journal of Psychiatry124:352356, 1974

7. Rabinowitz J, Slyuzberg M, Ritsner M, etal: Changes in diagnosis in a 9-year nation-al longitudinal sample. ComprehensivePsychiatry 35:361365, 1994

8. Stanton MW, Joyce PR: Stability of psychi-atric diagnoses in New Zealand psychiatrichospitals. Australia and New Zealand Jour-nal of Psychiatry 27:28, 1993

9. Weeke A: Admission pattern and diagnosticstability among unipolar and bipolar manic-depressive patients. Acta Psychiatrica Scan-dinavica 70:603613, 1984

10. Cohen P, Cohen J: The clinicians illusion.Archives of General Psychiatry 41:1178

1182, 1984

11. Schwartz JE, Fennig S, Tanenberg-KarantM, et al: Congruence of diagnoses 2 yearsafter a first-admission of psychosis. Ar-chives of General Psychiatry 57:593600,2000

12. Beiser M, Iacono WG, Erickson D: Tem-poral stability in the major mental disor-ders, in The Validity of Psychiatric Diagno-sis. Edited by Robins LN, Barrett JE. NewYork, Raven, 1989

13. Lenz AS, Simhandl C, Thau K, et al: Tem-poral stability of diagnostic criteria forfunctional psychoses: results from the Vien-na follow-up study. Psychopathology 24:328335, 1991

14. Tohen M, Tsaung MT, Goodwin TC: Pre-diction of outcome in mania by mood-con-gruent or mood-incongruent psychotic fea-tures. American Journal of Psychiatry 149:15801584, 1992

15. Mason P, Harrison G, Croudace T, et al:The predictive validity of a diagnosis ofschizophrenia: a report from the Interna-tional Study of Schizophrenia (IsoS) coor-dinated by the World Health Organisationand Department of Psychiatry, Universityof Nottingham. British Journal of Psychia-try 170:321327, 1997

16. Lee AS, Murray RM: The long-term out-come of Maudsley depressives. BritishJournal of Psychiatry 153:741751, 1988

17. Marneros A, Deister A, Rohde A: Stabilityof diagnoses in affective, schizoaffective,and schizophrenic disorders: cross-section-al versus longitudinal diagnoses. EuropeanArchives of Psychiatry and Clinical Neuro-science 241:187192, 1991

18. Tsaung MT, Woolson RF, Winokur G, et al:Stability of psychiatric diagnosis: schizo-phrenia and affective disorders followed upover a 30- to 40-year period. Archives ofGeneral Psychiatry 38:535539, 1981

19. Amin S, Singh SP, Brewin J, et al: Diagnos-tic stability of first-episode psychosis.British Journal of Psychiatry 175:527543,1999

20. Veen ND, Selten JP, Schols D, et al: Diag-nostic stability in a Dutch psychosis inci-dence cohort. British Journal of Psychiatry185:460464, 2004

21. Browne S, Clarke M, Gervin M, et al: De-terminants of neurological dysfunction infirst episode schizophrenia. PsychologicalMedicine 30:14331441, 2000

22. Diagnostic and Statistical Manual of Men-tal Disorders, 4th ed. Washington, DC,American Psychiatric Association, 1994.

23. Kay SR, Fiszbein A, Opler LA: The positiveand negative syndrome scale (PANSS) forschizophrenia. Schizophrenia Bulletin 13:261276, 1987

24. Statistical Package for the Social Sciences,version 11. Chicago, SPSS, 2002

25. Winokur G, Monahan P, Coryell W, et al:Schizophrenia and affective disorder: dis-tinct entities or continuum? An analysisbased on a prospective 6-year follow-up.Comprehensive Psychiatry 37:7787, 1996

PSYCHIATRIC SERVICES http://ps.psychiatryonline.org September 2005 Vol. 56 No. 911008888

oca.qxd 8/22/2005 9:47 AM Page 1088

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /DownsampleGrayImages false /GrayImageDownsampleType /Average /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /DownsampleMonoImages false /MonoImageDownsampleType /Average /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError false /PDFXTrimBoxToMediaBoxOffset [ 0.50000 0.50000 0.50000 0.50000 ] /PDFXSetBleedBoxToMediaBox false /PDFXBleedBoxToTrimBoxOffset [ 0.12500 0.12500 0.12500 0.12500 ] /PDFXOutputIntentProfile (None) /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org) /PDFXTrapped /Unknown

/Description >>> setdistillerparams> setpagedevice