Acta Ps.1Ichiatr Scand 2003: 107 (Suppl. 416): 16-23Printed in UK. All rights reser1led

Copyright @ B/ackweE-Munksgaard 2003

ACT A PSYCH[ATRICA

SCAND[NAV[CA[SSN 0065-[59[

Haro JM, Kamath SA, Ochoa S, Novick D, Rele K, Fargas A,Rodriguez MJ, Rele R, Orta J, Kharbeng A, Araya S, Gervin M,Alonso J, Mavreas V, Lavrentzou E, Liontos N, Gregor K, Jones PBon behalf of the SOHO Study Group. The Clinical Global Impression-Schizophrenia scale: a simple instrument to measure the diversity ofsymptoms present in schizophrenia,Acta Psychiatr Scand 2003: 107 (Suppl. 416): 16-23.@ B1ackwel1 Munksgaard 2003

J. M. Haro', S. A. Kamath2,S. Ochoa " D. Novick3, K. Rele2,

A. Fargas', M. J. Rodriguez',

R. Rele2, J. Orta', A. Kharbeng2,

S. Araya', M. Gervin2, J. Alonso,

V. Mavreas4, E. Lavrentzou4,

N. Liontos4, K. Gregor3,

P. B. Jones2 on behalf of the

SOHO Study Group*

1Research and Development Unit. Sant Joan deDeu-SSM. Sant Boi. Barcelona. Spain. zUniversity ofCambridge. Cambridge. UK. JEuropean Health OutcomesResearch. Eli Lilly and Company Limited. Windlesham.Surrey. UK and 4University of loannina. Greece

Objective: To describe the development and validation of the ClinicalGlobal Impression-Schizophrenia (CGI-SCH) scale, designed to assesspositive, negative, depressive and cognitive symptoms inschizophrenia.Method: The CGI-SCH scale was adapted from the CGI scale.Concurrent validity and sensitivity to change were assessed bycomparison with the Positive and Negative Symptom Severity(PANSS) and Global Assessment of Functioning (GAF) scales. Toevaluate inter-rater reliability, all patients were assessed by twoclinicians.Results: Symptoms were assessed in 114 patients. Correlationcoefficients between the CGI-SCH and the GAF and PANSS scoreswere high (most above 0.75), and were highest for positive and negativesymptoms. Reliability was substantial (intraclass correlationcoefficient, ICC > 0.70) in all but one dimension (depressivedimension, ICC = 0.64).Conclusion: The CGI-SCH scale is a valid, reliable instrument toevaluate severity and treatment response in schizophrenia. Given itssimplicity, brevity and clinical face validity, the scale is appropriate foruse in observational studies and routine clinical practice.

Key words' psychiatric status rating scales;

psychometrics; questionnaires; schizophrenia; signsand symptoms

Josep Maria HarD. Research and Development Unit.Dr Antoni Pujades 42. E-0883o-Sant Boi de L. Barcelona,

SpainTel: +3493 6002682; Fax. +34936520051;E-mail: 27652jha@comb.es

'florid' or 'productive', and 'defect' or 'deficit',which correspond roughly to the more up-to-dateterms of positive and negative symptoms. Depres-sion and cognitive symptoms also accompanypositive and negative symptoms as psychopatho-logical manifestations of schizophrenia. Depressedmood is common in individuals who suffer fromschizophrenia, often arising from the individual'sappraisal of psychosis and its implications for hisor her perceived social identity, position and 'groupfit'. Post-psychotic depression is associated with anincreased risk of suicide (1,2). Cognitive symptoms

Introduction

Schizophrenia is a serious mental disorder charac-terized by a number of symptoms. In the past, thesymptoms of schizophrenia were classified as

This paper is one of a suite of papers reporting aspects of theSchizophrenia Outpatient Health Outcomes (SOHO) Study.The study and this publication were funded by Eli Lilly andCompany Limited, Windlesham, Surrey, UK.

*The SOHO Study Group is listed in the Acknowledgementssection.

16

CGI-SCH validity in the SOHO study

were defined by Kraepelin in his first descriptionsof dementia praecox. Taken together, these symp-toms affect several areas of functioning, such asattention, executive functioning and memory .

The Positive and Negative Symptom Severity(PANSS) scale (3) is the scale used most oftenwhen assessing treatment response or clinicalseverity in schizophrenia, and allows evaluationof the symptoms of this condition. Factor analysisstudies performed with the PANSS on largepopulations of patients with schizophrenia haveidentified five components in the symptomatology:positive, negative and cognitive/disorganizationsymptoms and two other affective dimensions(4-6). The findings using the PANSS scalehave been consistent in different populations ofpatients (4-6). The PANSS depressive scale hasbeen shown to be a valid measure of depres-sive symptoms in schizophrenia when comparedwith the Hamilton Rating Scale for Depression(HAM-D) and the Calgary Depression Scale for

Schizophrenia (7).However, the PANSS (like most scales that

assess clinical severity) has been developed for usein a research environment, and while suitable forassessing treatment response in clinical trials, istime-consuming to administer (typically taking 30-45 min). Shorter, simpler and easier-to-administerscales are badly needed, particularly for use instudies of treatment effectiveness, where evaluationof treatment occurs in a real practice environment.In this situation rapid assessment is mandatory, asa longer assessment would alter the normal courseof the care that is under evaluation. Quick, simpleinstruments could also be used in routine clinicalpractice. There is a need therefore for a simple,quick and easy-to-administer scale that is suitablefor use in observational studies and routine clinical

practice.The objective of this paper is to describe the

development and validation of the Clinical GlobalImpression-Schizophrenia (CGI-SCH) scale, abrief assessment instrument adapted from theClinical Global Impression (CGI) scale. TheCGI-SCH scale is designed to assess the mainsymptom dimensions in schizophrenia.

was to produce a simple, easy-to-administerinstrument that could be used in observationalstudies and routine clinical practice in schizophre-nia. It was decided that the instrument should:

.include evaluation of positive, negative, depres-sive and cognitive symptoms;

.be easy to understand;

.be quick and easy to administer;

.be valid and reliable; and

.be sensitive to change.

The CGI-SCH scale was adapted from the CGIscale (9) and the CGI-Bipolar Patients (CGI-BP)scale (10). The CGI scale is a simple instrumentthat evaluates the overall severity of mental disor-ders. The complete CGI scale consists of threedifferent global measures designed to rate theeffectiveness of a particular treatment:

(i) severity of the illness (assessment of thecurrent severity of symptoms);

(ii) global improvement (comparison of thepatient's baseline condition to his or hercurrent condition); and

(iii) efficacy index (evaluation of the patient'simprovement from baseline in relation totreatment side-effects).

The CGI has been used previously in efficacyand effectiveness studies in schizophrenia (11-13),and has been shown to be sensitive to change:changes recorded by the CGI correlate withchanges observed with more complex scales(14, 15). Nevertheless, the CGI has been criticizedfor being inconsistent and unreliable (10, 16, 17).Specific criticism includes the fact that the scalehas asymmetric scaling, lacks standard definitionsof illness severity and change, the change meas-ures are redundant and the assessment of side-effects mixed with the evaluation of treatmentchange can complicate evaluation and interpret-ation ( 10).

Spearing et al. modified the CGI scale toimprove its applicability in bipolar disorder (10).The CGI-BP overcomes the shortcomings of theCGI by eliminating the efficacy index, betterdefining the items, changing the anchor pointsand differentiating the rating of different types ofsymptoms (mania, depression and overall bipolarillness). The CGI-BP scale includes three categories(severity of illness, change from preceding phaseand change from the worst phase of illness) and theevaluation of significant side-effects. Each of thecategories has a different rating for manic, depres-sive and global symptoms. The CGI-BP has beenused in recently conducted treatment trials inbipolar disorder (18, 19).

Material and methods

Development of the CGI-SCH scale

The CGI-SCH scale was developed for use in theSchizophrenia Outpatient Heath Outcomes{SOHO) Study {8), an observational study of theoutcomes of antipsychotic treatment in schizophre-nia. In creating the CGI-SCH scale, the objective

17

Haro et al.

I. Severity of illnessConsidering your total clinical experience wit

Normal.not ill

latients with schizophrenia, how severely ill has the patient been during the last week?

Minimally Mildly Moderately Markedly Severelyill ill ill ill ill

Among the mostseverely ill

3

3 4

2 4 6i

1. Positive symptoms 1

(e.g. hallucinations, delusions or bizarre behaviourl

2. Negative symptoms 1

(e.g. affective flattening, avolition or anhedonia)

3. Depressive symptoms 1

(e.g. sadness, depressed mood or hopelessness)

4. Cognitive symptoms 1

(e.g. impaired attention, concentration or memory)

5. Overall severity 1

3

3 4

II. Degree of change

Compared to the previous evaluation*, how much has the patient changed? Rate improvement whether or not, in your judgement, is due entirely to treatment?

Very much Much Minimally No Minimally Much Very much

improved improved improved change worse worse worse N/A

93

2 9

4 9

3

, .Positive symptoms ,(e.g. hallucinations. delusions or bizarre behaviour)

2. Negative symptoms ,le.g. affective flattening. avolition or anhedonia)

3. Depressive symptoms ,(e.g. sadness. depressed mood or hopelessness)

4. Cognitive symptoms ,(e.g. impaired attention, concentration or memory)

5. Overall severity 1 3 4 1 9

*In treatment trials with several evaluation points, use 'Compared to the phase immediately preceding this treatment trial' instead of 'Compared to the previous evaluation'

Based on the CGI and CGI-BP, the CGI-SCHwas developed for use with patients with schizo-phrenia. The CGI-SCH is simpler than the CGIand the CGI-BP scales as it consists of only twocategories; severity of illness and degree of change(Table I). The severity of illness category evalu-ates the situation during the week previous to theassessment, while the degree of change categoryevaluates the change from the previous evaluation(or from the phase preceding the treatment trial).Each category contains five different ratings(positive, negative, depressive, cognitive andglobal) that are evaluated using a seven-pointordinal scale. To help understanding, a shortdefinition of each symptom is included in theinstrument, and the instruction manual contains amore detailed definition of each dimension. Com-pared with the CGI instrument, several importantchanges have been introduced. The scaling ofratings has been modified to achieve more con-sistent intervals and time' domains have beenclarified. For example, the CGI instrument asksfor the state of the patient 'at this time', while theCGI-SCH asks for the state of the patient 'duringthe last week'. The CGI efficacy index rates theimprovement due to pharmacological treatmentand relates this to the presence of side effects. As

this index combines two diverse constructs, it isdifficult to rate and probably not particularlyreliable. The efficacy index has been deleted fromthe CGI-SCH and it is suggested that the evalu-ation of side-etfects should be undertaken withspecific scales.

The meaning of each of the ratings of the CGI-SCH is similar to the P ANSS dimensions (positive,negative, depressive and cognitive/disorganiza-tion). The term 'CGI-SCH cognitive symptoms' isused instead of 'CGI-SCH cognitive/disorganiza-tion', as cognitive symptoms is a term with whichpsychiatrists are more likely to be familiar and thescale was designed for use by psychiatrists workingin clinical practice rather than a research environ-ment.

A brief user manual was developed to accom-pany the CGI-SCH (available from the authors).Following development of the CGI-SCH scale andthe user manual, a process of cognitive debriefingwas undertaken to test if the instrument wasunderstood as it was intended.

The CGI-SCH was developed in English. Theoriginal English version was converted to Spanishusing standard translation-backtranslation proce-dures, including expert panels, cognitive debriefingand pilot testing.

18

CGI-SCH validity in the SOHO study

Patients

The study was conducted in three centres: SantJoan de Deu-Serveis de Salut Mental in Barce-lona, Spain, the University of Cambridge in theUK and the University of Ioannina in Greece.The study sample was designed to include abroad representation of patients with schizophre-nia, including in-patients and outpatients, as wellas patients experiencing an acute episode andthose in a stable condition. Patients were recruit-ed from three acute in-patient units and threeoutpatient services. The heterogeneity of thispatient sample reflects the expectation that theCGI-SCH will be used in both in-patient andoutpatient settings. Patients with a diagnosis ofschizophrenia (according to IDC-I0 or DSM-IVcriteria), receiving psychiatric treatment, aged18 years or older, and who gave informed consentfor participation were included. No exclusioncriteria were applied.

Statistical analysis

Concurrent validity (a type of construct validity) isthe capacity of an instrument to agree with otherconstructs that coexist with the one assessed bythe test. Concurrent validity of the CGI-SCH scalewas assessed by analysing the agreement betweenthe CGI-SCH ratings with the PANSS (positive,

negative, depressive, cognitive/disorganizationand global scores), and the GAP. The CGI-SCHseverity of illness (CGI-SCH SI) score for positivesymptoms was compared with the P ANSS positivescore, for example, and the CGI-SCH SI score fordepressive symptoms was compared with theP ANSS depressive score. Pearson correlationcoefficients were used to analyse the association.The P ANSS scores were calculated using thefollowing items (4, 23):

.positive (delusions, hallucinatory behaviour,grandiosity, suspiciousness, unusual thoughtcontent, lack of judgment and insight);

.negative (blunted affect, emotional withdrawal,poor rapport, passive/apathetic social with-drawal, lack of spontaneity and flow of conver-sation, motor retardation, active socialavoidance, disturbance of volition);

.depressive (anxiety, guilt feelings, depression);and

.cognitive/disorganization (poor attention, con-ceptual disorganization, difficulty in abstractthinking, disorientation).

Sensitivity to change was analysed by calculat-ing the effect size of the change of the CGI-SCH,the PANSS and the GAP ratings from admissionto discharge. The effect size was calculated bydividing the mean change in the scale by thestandard deviation. Sensitivity to change of theCGI-SCH scale was also evaluated by comparingthe change in the CGI-SCH scale with the changein the P ANSS and the GAP scales. As with thevalidity assessment, each of the CGI-SCH ratingswas compared to the rating in the other scalesthat measured the same construct. Pearson corre-lation coefficients were used to analyse thisassociation.

CGI-SCH degree of change (CGI-SCH DC)ratings measure the change of the severity of thedisorder between two time-points ( e.g. from theinitiation of treatment to the assessment of itseffectiveness). The CGI-SCH DC ratings at dis-charge were compared to the change in the CGI-SCH SI ratings from admission to discharge.Pearson correlation coefficients were used to ana-lyse the agreement.

Methods

The objectives of the evaluation were to determinethe concurrent validity, inter-rater reliability andsensitivity to change of the CGI-SCH scale. Abattery of instruments were administered to thepatients, including a sociodemographic and clinicalquestionnaire, the CGI-SCH scales, the PANSS(3, 20) and the Global Assessment of FunctioningScale (GAP) (21,22).

Out-patients included in the study were rated bytwo clinicians (one of whom was usually thetreating psychiatrist) using the battery of instru-ments. One of the clinicians conducted the inter-view and both clinicians completed the fourquestionnaires independently. Only the severity ofillness (and not the degree of change) part of theCGI-SCH was completed for outpatients, as therewas no follow-up assessment. In-patients wereevaluated twice. The first evaluation took placeduring the first days after admission (this evalua-tion was equivalent to the evaluation of out-patients), and the second evaluation was conductedat discharge by one of the clinicians. The secondevaluation included the same instruments, exceptthat both categories of the CGI-SCH scales(severity of illness and degree of change) wererated. The order of administration of the ques-tionnaires was the same in all cases and wassociodemographic and clinical questionnaire, CGI-SCH, PANSS and GAF.

The study protocol was approved by the ethicscommittees of the participating institutions.

19

Haro et at.

Inter-rater reliability was assessed by comparingthe ratings of each of the CGI-SCH dimensionsmade by the two clinicians for the same patient,analysed using intraclass correlation coefficients(ICC) (24). ICC values range from ° to 1; values of0.7 and over are considered to indicate 'substantialagreement' and values of 0.5-0.7 are considered toindicate 'moderate agreement' (25). As the GAFscale can have up to 100 possible scores, inter-raterreliability was calculated by grouping the scoresinto 5-point intervals.

Results

A total of 114 patients were included in thestudy; 50 patients from Spain (24 in-patients and26 out-patients), 34 from the United Kingdom (19in-patients and 15 out-patients) and 30 fromGreece (eight in-patients and 22 out-patients).The proportion of men was 66.7%, 82.8% and69.7% for Spain, United Kingdom and Greece,respectively. Mean age was 38.7 years (SD 10.2),37.0 years (SD 11.6) and 33.9 years (SD 10.6) forSpain, United Kingdom and Greece, respectively.Patient sociodemographic and clinical characteris-tics are outlined in Table 2.

Correlation coefficients for the ratings in theCGI-SCH SI scales and the PANSS, and GAPscores are shown in Table 3. Values in bold arecorrelations that compare the CGI-SCH SI scaleswith the corresponding assessment in the otherinstruments. CGI-SCH SI ratings for positive,negative, cognitive symptoms and overall severityshowed substantial agreement with the P ANSSpositive, negative, cognitive/disorganization and

Table 2. Patient sociodemographic and clinical characteristics

laracteristic

44.7

55.3

69.7

36.9 110.8)

24.1 16.6)

total scores, respectively (Pearson correlationcoefficients ranging from 0.75 to 0.86). Moderateagreement was found between the CGI-SCHdepressive score and the P ANSS depressive dimen-sion and the CGI-SCH global score and the GAPscale (Pearson correlation coefficients of 0.60 and0.67, respectively). As symptom dimensions are nottotally independent, correlations between the CGI-SCH SI scales and the other scales that assessedsymptoms not directly related to the symptomsbeing evaluated in that CGI-SCH dimension werealso present. However, as expected, the values ofthe Pearson correlation coefficients were low(values ranging from 0.02 to 0.37), except for thecognitive and negative dimensions, where correla-tion coefficients were around 0.5. Scales that assessglobal symptomatology or functioning (CGI-SCHglobal score, P ANSS total and GAP) were corre-lated to symptom dimensions, as global symptomsinclude the individual dimensions.

Sensitivity to change wasanalysed by assessing theeffect size of the change in ratings during admission(Table 4). The effect sizes for CGI-SCH SI positiveand global scores were higher than for negative,depressive and cognitive symptoms, and similar tothose for the P ANSS positive, total and GAP scores.Hospital admission to in-patient units is usuallycaused by an increase in positive symptoms and itsimprovement is the main objective of treatment. Theeffect sizes of the P ANSS depressive and cogni-tive/disorganization scores were higher than theCGI-SCH SI depressive and cognitive ratings.

The Pearson correlation coefficients of thechange in CGI-SCH SI scores with the corres-ponding PANSS dimension and GAP score (CGI-SCH SI positive with P ANSS positive score,CGI-SCH SI negative with PANSS negativescore, etc.) ranged from 0.62 (P < 0.001) fordepressive symptoms to 0.70 (P < 0.001) forpositive symptoms (data not shown). The correla-tion coefficients between the change in CGI-SCHSI scores and the CGI-SCH degree of changescores ranged from 0.63 (depressive symptoms) to0.75 (cognitive symptoms).

Inter-rater reliability was substantial for theCGI-SCH SI positive, negative, cognitive andglobal scores (ICC ranged from 0.73 to 0.82) andmoderate for the depressive scores (ICC = 0.64)(Table 5). Inter-rater reliability was slightly higherfor the PANSS and GAP instruments than for theCGI-SCH scores.

31.3

43.7

20.5

3.6

0.9

SettingIn-patient (%)

Outpatient (%)Gender (% male}

Mean (SD) age (years)

Mean (SD) age at first treatment contact (years}

HousingIndependent residence 1%)

Residence as dependent family member (%)

Supervised residence (%)

Homeless (%)

Other (%)

Mean (SD} PANSS score

Global

Positive

Negative

Depressive

CognitiveMean (SD) GAF score

Number of patients

70.5

18.2

20.0

6.3

8.8

44.2

114

Discussion

The CGI-SCH scale is a brief assessment instru-

ment designed to evaluate positive, negative,

20

(22.4)

(7.7)

(9.1)

(2.51

(4.1)

(19.9)

CGI-SCH validity in the SOHO study

Table Concurrent validity: correlation coefficients for the CGI-SCH severity of illness scales and PANSS and GAF scores

CGI-SCIpositive

CGI-SCH

negative

CGI-SCH

depressivE

CGI-SCHcognitive

CGI-SCHglobal

PANSSpositive

PANSS

negativePANSS

depressivePANSS

:ognitivePANSS

totalScale

Table 4. Analysis of sensitivity to change. effect size of the change in CGI-SCH.PANSS and GAF scores from admission to discharge in in-patients

Scale Effect size

CGI-SCI SI positiveCGI-SCH SI negativeCGI-SCH SI depressiveCGI-SCH SI cognitiveCGI-SCH globalPANSS positivePANSS negativePANSS depressivePANSS cognitivePANSS totalGAF

0.81

0.14

0.31

0.25

0.79

0.93

0.24

0.49

0.50

0.80

0.90

Table 5. Inter-rater reliability analysis: ICC of the evaluations otthe two cliniciansand the battery of instruments

Scale ICC

CGI-SCI SI positiveCGI-SCH SI negativeCGI-SCH SI depressiveCGI-SCH SI cognitiveCGI-SCH globalPANSS positivePANSS negativePANSS depressivePANSS cognitivePANSS totalGAF

0.820.730.640.770.750.880.770.800.850.870.87

depressive, cognitive symptoms and overall sever-ity in schizophrenia. The scale aims to translateclinical judgement into ratings that reflect thediversity of symptoms present in schizophrenia.The ratings are based on clinical judgement and theassessment is not time consuming to administer.

Overall, the psychometric properties of the CGI-SCH scale were good. CGI-SCH inter-rater reli-ability measured with the ICC was similar to thePANSS dimension score ICC in most ratings,

except for the depressive score (CGI-SCH depres-sive rating ICC was 0.64 compared with 0.80 in theP ANSS depressive dimension). The association ofthe CGI-SCH scales with the PANSS dimensionratings measured with the Pearson correlationcoefficient was high for all of the ratings, except,again, for the depressive symptoms (pearson cor-relation coefficient of 0.6, indicating a moderate

relationship).The CGI-SCH global rating correlation with the

P ANSS total and GAF scores was apparentlylower than that for the positive, negative andcognitive scores (Pearson correlation coefficients of0.75 and 0.67 for the correlation with the PANSStotal and GAF scores, respectively, compared with0.86, 0,80 and 0.78, for the positive, negative andcognitive scores, respectively). However, it shouldbe remembered that the PANSS global, GAF andCGI-SCH global rating do not measure the sameconstructs. The CGI-SCH global score assessesglobal severity of the disorder, including bothsymptoms and interference with functioning. TheP ANSS total score only evaluates symptoms andnot interference, and the GAF scale is made up oftwo independent scales (severity of symptomsand interference) and the final rating is the lowestof both. The correlation between the GAF andPANSS total scores was 0.66.

Some correlation exists between the intensity ofthe symptoms in different dimensions. For exam-ple, the CGI-SCH negative symptoms score isrelated to the CGI-SCH cognitive score. Clinicalsense dictates that patients with more negativesymptoms are also likely to score higher in termsof cognitive symptoms. The correlation betweendepressive, positive and negative symptoms hasalso been found by other authors (26-28).

Sensitivity to change for the CGI-SCH scale wassimilar to sensitivity to change for the PANSS andGAF, except for the depressive dimension, where

21

Haro et at.

UK; (UK) Martin Knapp, London School of Economics,Centre for the Economics of Mental Health, Institute ofPsychiatry, London, UK.

sensitivity to change was lower for the CGI-SCHdepressive score.

When considering these findings, it should beremembered that the design of the study tried tomimic routine clinical practice. The rating of theCGI-SCH scale was recorded after an interviewthat lasted approximately the same time as aclinical visit. After the rating of the CGI-SCHscale, the interview was extended to administer therest of the battery of tests and further questioningwas done. This series of events and timing wasdesigned to compare the information obtainedfrom the CGI-SCH scale administered at the endof a short visit, with the information obtainedduring a typical evaluation with the p ANSS andGAF scales. The training of psychiatrists in the useof the CGI-SCH scale was based only on the CGI-SCH instruction manual. Previous research hasshown that training duration is associated withincreased inter-rater reliability (29, 30), and it isprobable that with additional training, correlationof the CGI-SCH with the pANSS and inter-raterreliability would be higher.

Conclusion

From these results, it can be concluded that theCGI-SCH scale is a valid and reliable instrumentwith which to evaluate severity and treatmentresponse in schizophrenia. Support was strongerfor the positive, negative, cognitive and globalratings than for the depressive ratings. The simpli-city of the instrument and the fact that it is quick toadminister make it appropriate for use in observa-tional studies and routine clinical practice.

Acknowledgements

The SOHO Study Group: (Denmark) Karsten Haderup

Kristensen, Aalborg Psykiatrisk Hospital, Aalborg, Denmark;(France) Jean-Pierre Lepine, Hopital Femand Widal, Paris,France; (France) Isabelle Gasquet, Hopital Paul Brousse,Villejuif, France; (Germany) Dieter Naber, Universitatskran-kenhaus-Eppendorf, Klinik fur Psychiatrie und Psychothera-pie, Hamburg, Germany; (Greece) Venetsanos G. Mavreas,Department of Psychiatry, University of Ioannina, Greece;(Ireland) Declan Murray, St Ita's Hospital, Dublin, Ireland;(Italy) Paolo Pancheri, Fondazione Italiana per 10 studio dellaSchizofrenia (FIS), Rome, Italy; (the Netherlands) CJ Slootf,Psychosencluster GGX N-Drenthe, Kenniscentrum Scizofre-nie, RA Assen, the Netherlands; (portugal) Joao MarquesTeixeira, University do Porto, Porto, Portugal; (Spain) JordiAlonso, Health Services Research Unit, Institut Municipald'Investigacio Medica, University of Barcelona, Barcelona,Spain; (Spain) Josep Maria Haro, Research and DevelopmentUnit, Sant Joan de Deu-ssM, Sant Boi, Barcelona, Spain;(UK) Tim Croudace, Department of Psychiatry, Adden-brooke's Hospital, Cambridge, UK; (UK) Peter B. Jones, uni-versity of Cambridge, Addenbrooke's Hospital, Cambridge,

References

I. ADDINGTON DD, AzORIN IM, FALLOON IR, GERLACH I, HIRSCHSR, SIRIS Sa. Clinical issues related to depression in schi-zophrenia: an international survey of psychiatrists. ActaPsychiatr Scand 2002;105:189-195.

2. BIRCHWOOD M, IQBAL Z, CHADWICK P, TRoWER P. Cognitiveapproach to depression and suicidal thinking in psychosis.I. Ontogeny of post-psychotic. Br I Psychiatry 2000;177:516-521.

3. KAY SR, OPLER LA, FISZBEIN A. The positive and negativesymptom scale (PANSS). Rating manual. Soc Behav SciDoc 1986;17:28-29.

4. KAY SR, SEVY S. pyraniidical model of schizophrenia.Schizophr Bull 1990;16:537-545.

5. LANCON C, AuQUIER P, NAYT a, REINE a. Stability of thefive-factor structure of the Positive and Negative Syn-drome Scale (PANSS). Schizophr Res 2000;42:231-239.

6. BELL MD, LYSAKER PH, BEAM-GOULET IL, MILSTEIN RM,LINDENMAYER JP. Five-component model of schizophrenia:assessing the factorial invariance of the positive and neg-ative syndrome scale. Psychiatry Res 1994;52:259-303.

7. EL yAZAJI M, BATTAS 0, AGOUB M et al. Validity of thedepressive dimension extracted from principal componentanalysis of the PANSS in drug-free patients with schizo-phrenia. Schizophr Res 2002;56:121-127.

8. HARo IM, EooELL ET, IONES PB et al. on behalf ofthe SOHO STUDY GROUP. The European SchizophreniaOutpatient Health Outcome (SOHO) Study: rationale,methods and recruitment. Acta Psychiatr Scand 2003;107:222-232.

9. GUY W, ed. Clinical global impression. In: ECDEUAssessment Manual for Psychopharmacology (revised).Rockville, MD: National Institute of Mental Health, 1976:217-221.

10. SPEARING MK, POST RM, LEVERICH as, BRANDT D, NoLEN W.Modification of the Clinical Global Impression (CGI) scalefor use in bipolar illness: the CGI-BP. Psychiatry Res

1997;73:159-171.II. KINON BI, BASSON BR, GILMORE IA, MALCOLM S, STAUFFER

VL. Strategies for switching from conventional antipsy-chotic drugs or risperidone to olanzapine. I Clin Psychiatry

2000;61:833-840.12. PERRY PI, LUND BC, SANGER T, BEASLEY C. Olanzapine

plasma concentrations and clinical response: acute phaseresults of the North American Olanzapine Trial. I Clin

PsychopharmacoI2001;21:14-20.13. SACRISTAN IA, GOMEZ IC, MoNTEJO AL, VIETA E, GREGOR KJ.

Doses of olanzapine, risperidone, and haloperidol used inclinical practice: results of a prospective pharmacoepi-demiologic study. EFESO Study Group. Clin Ther

2000;22:583-599.14. KHAN A, KHAN SR, SHANKLES EB, POUSSAR NL. Relative

sensitivity of the Montgomery-Asberg Depression RatingScale, the Hamilton Depression rating scale and theClinical Global Impressions rating scale in antidepressantclinical trials. Int Clin Psychopharmacol 2002;17:281-285.

15. ANGRIST B, ROTRosEN I, GERSHON S. Responses to apomor-

phine, amphetamine, and neuroleptics in schizophrenic

subjects. Psychopharmacology (Berl) 1980;67:31-38.

22

CGI-SCH validity in the SOHO study

and negative syndrome scale (PANSS). Acta Psychiatr

Scand 1999;100:237-243.

24. PRIETO L, LAMARCA R, CASADO A, ALONSO J. The evaluation

of agreement on continuous variables by the intraclass

correlation coefficient. J Epidemiol Community Health

1997;51:579-580.25. SCHENE AH, KOETER M, WUNGAARDEN BY et al. Methodo-

logy of a multi-site reliability study. Br J Psychiatry

2000;177(Suppl.39):SI~S20.26. ADDINGTOND, ADDINGTON J, MATICKA-TYNDALE E. Specificity

of the Calgary Depression Scale for schizophrenics.

Schizophr Res 1994;11:239-244.

27. ADDINGTON D, ADDINGTON J, ATKINSON M. A psychometric

comparison of the Calgary Depression Scale for Schizo-

phrenia and the Hamilton Depression Rating Scale.

Schizophr Res 1996;19:20~212.

28. MULLER MJ, SZEGEDI A, WETZEL H, BENKERT 0. Depressive

factors and their relationships with other symptom

domains in schizophrenia, schizoaffective disorder, and

psychotic depression. Schizophr Bull 2001;27:19-28.29. MULLER MJ, WETZEL H. Improvement of inter-rater reli-

ability ofPANSS items and subscales by standardized rater

training. Acta Psychiatr Scand 1998; 98:13~139.

30. MULLER MJ, ROSSBACH W, DANNIGKErr P, MULLER-SIECHENEDER

F, SZEGEDI A, WETZEL H. Evaluation of standardized rater

training for the Positive and Negative Syndrome Scale

(PANSS). Schizophr Re& 1998; 32:151-160.

16. BENEKE M, RASMus W. Clinical global impressions(ECDEU): some critical comments. Pharmacopsychiatry

1992;25:171-176.17. DAHLKE F, LoHAUS A, GUIZMANN H. Reliability and clinical

concepts underlying global judgements in dementia:implications for clinical research. Psychopharmacol Bull

1992;28:425-432.18. ToHEN M, SANGER TM, McELROY SL et al. Olanzapine

versus placebo in the treatment of acute mania. OlanzapineHGEH Study Group. Am I Psychiatry 1999;156:702-709.

19. ToHEN M, CHENPAGGA KN, SUPPES T et al. Efficacy ofolanzapine in combination with valproate or lithium in thetreatment of mania in patients partially non responsive tovalproate or lithium monotherapy. Arch Gen Psychiatry

2002;59:62-69.20. PERALTA MARTIN V, CUESTA ZORITA MI. [Validation of

positive and negative symptom scale (PANSS) in a sampleof Spanish schizophrenic patients.] Actas Luso Esp NeurolPsiquiatr Cienc Afines 1994;22:171-177.

21. AMERICAN PSYCHIATRIC AsSOCIATION. Diagnostic and statisticalmanual of mental disorders. DSM-IV. Washington, DC:American Psychiatric Association, 1994.

22. AMERICAN PSYCHIATRIC AsSOCIATION. Manual diagnostico yestadistico de los trastornos mentales. DSM-IV. Barcelona:Masson, 1995.

23. LANCON C, REINE G, LLORCA PM, AuQUIER P. Validity andreliability of the French-Ianguage version of the positive

23