Schizophrenia · Schizophrenia taken into account in most studies. The two areas of the United Kingdom which showed an increased incidence of schizophrenia are both areas with a high

64ournal ofNeurology, Neurosurgery, and Psychiatry 1996;61:604-613

NEUROEPIDEMIOLOGY

Schizophrenia

Mary Cannon, Peter Jones

Psychiatric epidemiology is the study of thedistribution and causes of mental disorder inthe population. Over the past 30 yearsprogress in psychiatric epidemiology has beenslower than in other areas-for example, car-diovascular disease or cancer epidemiology,because of certain methodological problemsthat are just now being overcome. In particularthe epidemiology of schizophrenia has sufferedfrom two major difficulties: uncertainty abouthow to define a case, and the relative rarity ofschizophrenia in the population.

Department ofPsychologicalMedicine, Institute ofPsychiatry, DenmarkHill, London SE5 8AF,UKM Cannon

Department ofPsychiatry, UniversityofNottingham,Duncan MacmillanHouse, PorchesterRoad, NottinghamNG3 6AA, UKP JonesCorrespondence to:Dr Peter Jones.

Special methodological problems inschizophrenia epidemiologyWHAT IS A CASE

The lack of physical signs or laboratory testsmeans that a diagnosis of schizophrenia isbased on evaluation of patients' self reported,subjective experiences. Until the late 1960s,there was a relatively loose definition of schizo-phrenia and a good deal of diagnostic latitudewas allowed to individual psychiatrists. How-ever, a report published in 1972,' showed thatthe diagnostic habits of psychiatrists in theUnited States and United Kingdom differed toan unacceptable degree, and led to the intro-duction of strict operational diagnostic criteriafor schizophrenia. The Diagnostic andStatistical Manual DSM-III,2 published in1980, reduced the reliance on symptoms aloneby incorporating an element of chronicity-sixmonths prior duration of symptoms and an

upper age limit of 45 years for a first diagnosisof schizophrenia. Since 1980 there have beentwo more restrictive revisions of this diagnosticsystem, DSM-III-R3 and DSM-IV,4 and theInternational Classification of Diseases,(ICD), the diagnostic system favoured byEuropean psychiatrists, has also undergone a

similar "narrowing" process with the changefrom ICD-9 to ICD-10.5

But has this process gone too far? Althoughthe restrictive diagnostic criteria have certainlyincreased the reliability of the diagnosis, theyhave not improved the validity of "DSM-IV"or "ICD-10 schizophrenia".6 Recent familystudies indicate that a broader concept ofschizophrenia actually fits genetic modelsmore readily than a restrictive definition. Inaddition, the frequent changes in diagnosticcriteria have affected the comparability of

studies of time trends and outcome in schizo-phrenia. Also, the commitment to a phenotypebased on symptoms in adult life may haveignored an important developmental or life-long dimension to the disorder.

SCHIZOPHRENIA AS A "RARE DISEASE"The low incidence (10-40 cases per 100 000per year) and relatively low lifetime prevalence(0 5%-1%) of schizophrenia in the populationhave led to a reliance on case-control studydesigns in research. Chronic patients recruitedfrom hospital wards are compared with volun-teer controls from the community and conse-quent problems of bias and confounding haveoften led to unreplicated and contradictoryfindings. It seemed for a time that schizophre-nia, already known as the "graveyard of neu-ropathologists",7 would also prove to be theundoing of epidemiologists. Over the past twodecades, however, the application of advancesin case-control methodology to psychiatricepidemiology has led to more robust results.8Also, cohort designs, for long thought toocostly and time consuming to use in schizo-phrenia research, have made a "comeback".

For the purposes of this review schizophre-nia refers to a recent, operational definitionsuch as CATEGO,9 ICD-10, DSM-III, DSM-III-R, or DSM-IV. Robust findings are thosefrom studies with an epidemiological design-that is, population based studies with welldefined samples and appropriate controls. Weaim to show how the findings from robust epi-demiological research can help to unravel thecomplex aetiology of schizophrenia.

GeographyIn 1978, a large multicentre study of schizo-phrenia was initiated byWHO the 10 countrystudy (also known as the Determinants ofOutcome of Severe Mental Disorders Study),to provide information about the incidence,course, and outcome of schizophrenia in dif-ferent cultures.10 Two case definitions of schiz-ophrenia were used: a broad, clinicaldefinition comprising ICD-9 schizophreniaand paranoid psychoses, and a narrow, restric-tive definition including only cases classified as"nuclear" schizophrenia using the CATEGOcomputer programme 9 Figures 1A and Bshow the incidence rates for both definitions.

604

on May 22, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. D

ownloaded from

Cannon, _rones

Figure 1 (A) Incidence ofCATEGO S narrow schiz-ophrenia with 95% confi-dence intervals from eightcentres in the WHODeterminants of OutcomeStudy. (B) Incidence ofCATEGO S, P, and 0broad schizophrenia with95% confidence limits fromeight centres in the WHOdeterminants ofoutcomestudy. Data from J7ablenskyet. al.'0

Cu

m

C)00

._00a)-0

.5C-

Cu

coCu

00000C-

C.)C

'a.5C

15 H

10i

5

0 LoB

B

60_

30 _

20

6

5

4

3

2

of non-affective psychosis was 0.7%.12 Thisdiscrepancy may be related to issues of sam-pling, interview methodology, or actualchange over time and remains to be clarifiedby future reports.A major prevalence study of psychiatric

morbidity carried out in the United Kingdombetween April and September 1993,13 found aprevalence rate of 0 4% for functional psy-chosis among persons aged 16-64 living in pri-vate households. This rate is even lower thanthat found in the national comorbidity surveybut as 37% of those who screened positive forpsychosis refused a second diagnostic inter-view, the true prevalence may be higher.'4 Atpresent, it seems that the lifetime prevalence ofschizophrenia in the western world lies some-where between 0 4% and 1-4%, and may have

TI decreased over the past decade.,o- + T i

,\v

ItNX00A

There is little variation between centres fornarrowly defined schizophrenia with ratesranging only between 7 and 14/100 000, (fig1A). Although the parsimonious conclusionwould be that the rates for narrowly definedschizophrenia are the same, the confidenceintervals for these estimates were wide andthere may not have been sufficient statisticalpower to detect differences.The incidence rates for broadly defined

schizophrenia do seem to vary between coun-

tries (range 16 to 42/100 000 per year), andthe rates for centres in the developing worldare about twice as high as those in the devel-oped world (fig 1B). Further studies of schizo-phrenia in the developing world are needed toreplicate and further investigate these interest-ing findings, and the problem of mental healthin the developing world can no longer remain a

low priority for funding agencies and govern-ments. On the whole, however, the variationin incidence rates for schizophrenia worldwide is very small compared with illnessessuch as ischaemic heart disease or cancer,which are known to have major environmentalrisk factors.

Time trendsPREVALENCE DATATwo major prevalence studies of psychiatricillness have been carried out in the UnitedStates, which indicate a decrease in prevalenceof schizophrenia over one decade. TheEpidemiological Catchment Area (ECA)Study surveyed 17 803 persons between 1980and 1984 and found a lifetime prevalence ofschizophrenia of 1-4%." The NationalComorbidity Survey (NCS) interviewed 8098people between 1990 and 1992 and found thatlifetime prevalence for the summary category

"INCIDENCE" DATAPrevalence of schizophrenia can be influencedby factors such as change in treatment, changein mortality rate, and changes in the age andsex structure of the population, and therefore,examination of incidence rates may be moreinformative. Ideally, such studies should bebased on community incidence samples but,as this is difficult to achieve, case registers andhospital admission data are commonly used.

Eagles and Whalley'5 first reported a declinein the diagnosis of schizophrenia among firstadmissions in Scotland between 1969 and1978. There have since been 14 papers exam-ining this issue in England,'6-'0 Scotland,"'3Denmark,'425 New Zealand,'6 Canada,'7Ireland,'8 the United States,'9 and theNetherlands.30 Those based on national statis-tics have found a large (40%-50%) decline infirst admission rates for schizophrenia duringthe 1970s and the 1980s. 17 18 25 26 Findingsbased on case register data have been less con-sistent and indeed, two such studies, fromCamberwell"7 and Salford'8 in the UnitedKingdom, actually found a slight increase inthe incidence of schizophrenia during thesame period. Others have reported a decreasein first admission rates for schizophreniaamong female patients only." 2830The major question is whether the decrease

noted in first admission rates corresponds toan actual decrease in the incidence of the con-dition. Many factors influence this apparent"administrative decline" in schizophrenia: (1)The introduction of more restrictive diagnos-tic criteria for schizophrenia may cause a shiftto diagnoses such as "borderline states"24 oraffective psychosis.'930 (2) The move to com-munity care over the past two decades couldhave affected hospital admission rates.'7 (3)Clinicians have become more reluctant tomake a diagnosis of schizophrenia on the firsthospital admission,'4 and this could lead to aspurious fall in incidence rates over the lastfew years of the period under observation. Thepolicies of private health insurance companiesregarding schizophrenia may be partly respon-sible for this effect.'9 (4) Changes in the age,sex, and ethnic structure of the populationover the period under study have not been

605

20 FA

on May 22, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. D

ownloaded from

Schizophrenia

taken into account in most studies. The twoareas of the United Kingdom which showedan increased incidence of schizophrenia areboth areas with a high proportion of immi-grants.'7 18 Unfortunately, schizophrenia hassuch a low incidence that it may be impossibleto disentangle all these effects and reach anyfirm conclusions regarding changes in inci-dence in the developed world in recentdecades.31

Characteristics of patientsAGE AND SEX DISTRIBUTIONTotal lifetime risk for schizophrenia seems tobe equal in both sexes.1032 Schizophrenia canoccur at any age,3335 but onset is commonly inearly adulthood-over 70% of incident casesin the WHO 10 country study were between15 and 35 years of age.'0 The mean age ofonset for male patients is three to four yearsearlier than for female patients, irrespective ofwhether onset of schizophrenia is defined asthe first sign of mental disorder, the first psy-chotic symptom, or the first hospital admis-sion.36-38 The peak age of onset for males isbetween 15 and 30 whereas females have aslower and more even rate of onset with a peakbetween the ages of 20 and 35 and a secondsmaller peak after the age of 45, (fig 2). Nosatisfactory explanation yet exists for this sexdifference. A protective effect for female sexhormones has been suggested,39 but has notbeen proved empirically. The same patternoccurs across countries indicating that it is notan effect of cultural factors.10

SOCIAL CLASSThe long recognised association betweenschizophrenia and low social class was con-firmed by the ECA study, which showed thatschizophrenic patients in the United Stateswere 10 times more likely to be in the lowestsocioeconomic group than in the highest.35Evidence from birth cohorts in Britain,40 41 andFinland42 show that this relation is notcausal schizophrenic patients at birth havethe same socioeconomic distribution as thegeneral population. What is remarkable is thesteep decline in social state which accompa-nies the illness and is evident even before theclinical onset.42 43 This decline is far greaterthan that experienced by patients with severeaffective disorder.4243

Figure 2 Distribution ofage at first admission forschizophrenia in males andfemales. Source: Htfner etal.38

COC

U)a-41

30---Female

20 -t= Male

10 i' /

12-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59

Age group (y)

MARITAL STATESchizophrenic patients have low rates of mar-riage and low fertility32-a challenge to geneti-cists. Married patients have a better clinicaland social outcome than single patients,1'0although marriage may, of course, reflect betterpremorbid social adjustment and later age atonset, which are both independent predictorsof good outcome. However, the prevalence ofschizophrenia among separated or divorcedpeople in the United States (2 9%) is similarto that among single people (2-1%), suggest-ing that marriage may confer an independentprotective effect.32 An interaction betweenmarital state and gender was found in the oneyear follow up of the ECA study.44 Singlewomen were 14 times more likely to developschizophrenia than married women but singlemen were almost 50 times more likely todevelop schizophrenia than their marriedcounterparts. The role of marital state as a riskindicator or modifier for schizophreniadeserves further study.

ETHNICITY AND MIGRANT STATETo date no single indigeneous ethnic groupseems to have a significantly higher rate ofschizophrenia than any other,10 although somepockets of high prevalence may exist in iso-lated areas like north Sweden.45 The west ofIreland46 47 and the Istrian peninsula inCroatia48 had previously been identified ashigh risk cultures but the higher rates origi-nally found among these peoples are nowthought to be due to methodological factors.49Populations with increased prevalence, if care-fully identified, could be very useful for geneti-cists, but they seem to be local exceptionsrather than the general rule in the epidemiol-ogy of schizophrenia.50 The ECA study founda higher prevalence for schizophrenia amongblack people than among white people in theUnited States (2-1% v 1 -4%).32 However, whencontrolled for age, gender, marital state, andmost importantly, socioeconomic group, thesignificant difference disappeared. Of course in across sectional study such as the ECA study, it isnot possible to state definitively that currentmarital state and socioeconomic state com-pletely explain the association found betweenrace and disorder, because schizophrenia canlead to low social state and marital breakup.

Schizophrenia in immigrantsIn 1988, the psychiatric community was sur-prised by a report that the incidence of schizo-phrenia among the Afro-Caribbean populationin Nottingham was more than 10 times higherthan in the general population.5' This findinghas been convincingly replicated in other cen-tres in the United Kingdom5>255 and in TheNetherlands,56 although the incidence ratio isnow thought to be rather lower (3-5) when thedenominator is adjusted for possible underre-porting in census data.5455 An increased inci-dence ratio for schizophrenia has also beenfound among African5455 and Asian54 immi-grants in the United Kingdom, indicating thatthe effect is not confined to a single ethnicminority.

606

on May 22, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. D

ownloaded from

Cannon, _Jones

The hospital admission rate for schizophre-nia among migrants is higher in their hostcountry than in their country of origin,57-9which implicates factors occurring principallyafter migration. The risk of schizophrenia isgreater among second generation migrantsthan first generation migrants,51-54 60 arguingagainst selective migration of preschizophrenicpersons. The fact that immigrants from poorcountries tend to show higher rates of schizo-phrenia than immigrants from affluent coun-tries, implies that factors associated withimproved living conditions, industrialisation,or urbanisation are involved. Obstetric com-plications and exposure to novel viruses duringpregnancy have been proposed as biologicalexplanations,61 62 but empirical evidence islacking. One study has found an increased riskfor schizophrenia among the siblings of UnitedKingdom born Afro-Caribbean schizophrenicpatients compared with the risk among sib-lings of white patients,63 which could indicatea gene-environment interaction effect.

Despite these intriguing clues, the evidencefor a unique epidemic of schizophrenia amongcertain immigrant groups, although sugges-tive, is not conclusive. No studies have con-trolled adequately for possible confounderssuch as low socioeconomic class and maritalstate, which may reduce the incidence ratioeven further. It is likely that ethnicity repre-sents a "proxy" variable for various social andperhaps biological factors. Once these are con-trolled, there may be little or no residual effectof ethnicity itself but we would gain informa-tion about other, perhaps preventable, risk fac-tors involved. The most parsiminiousconclusion would be that schizophrenia inimmigrants is caused by the same factors thatcause schizophrenia in other groups but thatthese factors are more common, (and there-fore more conspicious), after migration. Thealternatives-that there are specific causes inimmigrants which do not occur in other popu-lations or that there is true interaction withubiquitious factors causing schizophrenia inonly some groups-both require furtherresearch.

COURSE AND OUTCOMEOutcome studies in schizophrenia are usuallybased on hospital treatment samples and maynot be representative of the population ofschizophrenic patients64; indeed, the true nat-ural course of schizophrenia is almost alwaysmasked by treatment these days. The results ofoutcome studies are difficult to summarisebecause the definitions of outcome and meth-ods of assessment used are so varied.However, all studies attest to the striking vari-ability of course and outcome of schizophre-nia.64 At the extremes of outcome, 20% ofpatients seem to recover completely after oneepisode of psychosis,65-68 whereas 14%-19% ofpatients develop a chronic unremitting psy-chosis and never fully recover.'0 65 69 In general,clinical outcome at five years seems to followthe rule of thirds: with about 35% of patientsin the poor outcome category65 69; 36% in thegood outcome category,65707' and the remain-

der with intermediate outcome. Prognosis inschizophrenia does not seem to worsen afterfive years.507'The effect of changes in diagnostic criteria

has been highlighted in a recent meta-analysisof outcome studies over 100 years,70 whichfound that prognosis for schizophrenia hasworsened over the past decade. The morerestrictive definitions of schizophrenia post-DSM-III, predict worse prognosis, partly as a"self fulfilling prophecy", as chronicity over sixmonths is already built into the definition ofthe disorder.6

Predictors of course and outcomeImmutable factors-Patients from developingcountries have a better outcome than patientsfrom developed countries,' 065 but it is notknown which aspects of non-Western cultureare responsible for this effect. Possible socialfactors are lower levels of expressed emotion,72stronger social support networks, or lack ofstigma. Ethnicity may be related to outcomebut conclusions are preliminary.73 Otherfavourable prognostic factors are good pre-morbid social adjustment, female sex, beingmarried, and later age at onset,'065 but thesefactors are unlikely to act independently.74Acute onset of illness and the experience ofnegative life events before illness are alsorelated to better outcome.6575

Mutable factors-High levels of "expressedemotion" among close relatives-namely, crit-icism, hostility and, overinvolvement76 77 -pre-dict early relapse. Substance misuse'078 alsopredicts poor prognosis. The important find-ing that delay in receiving treatment results inpoorer outcome,79-8 might be due to bias, andconfounding with certain chemical character-istics resulting in delayed treatment, but it ispossible that a more biological process isinvolved, with persistence of untreated posi-tive symptoms altering the "hard wiring" inthe brain.

MORTALITYStandardised mortality ratios (SMRs) forschizophrenic patients are estimated to be twoto four times higher than the general popula-tion,82 and their life expectancy overall is 20%shorter than for the general population.8' Themost common cause of death (in 10% ofpatients), is suicide-the risk of suicide is 20times higher than for the general population.828'Young men with a chronic illness are at partic-ular risk.84 More work needs to be carried outto identify patients at risk of suicide, and todetermine the effect of secular trends such asdeinstitutionalisation on suicide rates. Deathsfrom heart disease and from diseases ofthe respiratory and digestive system are alsoincreased among schizophrenic patients.828'Certainly the lower socioeconomic statusassociated with schizophrenia carries with itmany risk factors for physical diseases, such assmoking and poor diet, all of which are pre-ventable.

Risk factorsUntil recently, research into the aetiology of

607

on May 22, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. D

ownloaded from

Schizophrenia

schizophrenia focused principally on familyrelationships and social stressors. A remark-able "paradigm shift" has taken place over thepast decade; schizophrenia is now consideredto be a brain disease and emphasis is placed onbiological determinants.89 The impetus for thischange came from neuroimaging and neu-ropathological studies showing evidence ofbrain abnormalities in schizophrenicpatients.86 The timing of these pathologicalchanges is unclear but is likely to be a defect inearly brain development.86 Profound changeshave also occurred in hypotheses concerningneurotransmitter abnormalities in schizophre-nia. The dopamine hypothesis has been exten-sively revised and is no longer a considered aprimary causative model-current researchfocuses on the interaction between differentneurotransmitter systems.87

GENETIC FACTORSFamily studiesSchizophrenia runs in families. First degreerelatives of schizophrenic patients have a mor-bid risk of developing schizophrenia which is10 times higher than in the general popula-tion.88 89 The excess risk of developing schizo-phrenia occurs principally among the childrenand siblings of patients. Parents are at lowerrisk of developing schizophrenia, probablybecause the low fertility associated with havingschizophrenia means that parents are, ofnecessity, "selected for health", and havealready survived much of the period of risk.90However, "familiality" of an illness does notnecessarily indicate a genetic effect and canalso be due to environmental factors.Adoption studies and twin studies allow theseeffects to be examined separately.

Adoption studiesResults from the influential Danish adoptionstudy91 show that the risk for schizophreniaamong the biological first degree relatives ofthe schizophrenic adoptees is almost eighttimes higher than the risk among biologicalrelatives of control adoptees.92 Preliminaryresults from another large adoption studybeing carried out in Finland concur with thesefindings.93 An increased risk of schizophrenia,in the absence of any contact with biologicalrelatives, indicates that the familial aggre-gation of schizophrenia is due largely togenetic factors and not to some other aspect ofthe familial environment. Of course theadopted-away child has still spent the prenatalperiod with the biological mother and there-fore prenatal environmental risk factorscannot be ruled out. However, the risk ofschizophrenia spectrum disorders, (seebelow), was also increased in the paternal halfsiblings of the schizophrenic adoptees,who shared neither the prenatal nor familialenvironment.9

Twin studiesTwelve major twin studies of schizophreniahave been carried out, all of which show thatthe risk for schizophrenia in the co-twin of aschizophrenic proband (the probandwise con-

cordance) is substantially higher for MZ thanDZ twins.94 Although the average probandwiseconcordance in MZ twins is 46% (comparedwith 14% in DZ twins),95 the offspring of dis-cordant MZ twins have all been found to sharethe same risk of developing the disorder,96 rais-ing the possibility that the unaffected MZ twinmay represent an unexpressed genotype.Genetic factors are undoubtedly predominantbut there is evidence (discussed below) thatnon-genetic factors may be involved in theaetiology of schizophrenia. The average esti-mate of the proportion of liability attributableto non-genetic factors in schizophrenia is0.18.94

What is transmitted?The "schizophrenia spectrum" Certain psychi-atric illnesses and personality disorders,known as the schizophrenia spectrum, arethought to be genetically linked to schizophre-nia. The most important of these disorders isschizotypal personality disorder, and the rela-tive risk for schizotypal personality disorder inthe first degree relatives of schizophrenicprobands compared with controls is aboutfive.97 98 Parents of schizophrenic patients have ahigher risk of schizotypal personality disorderthan siblings, suggesting that individuals whoinherit this milder genetic vulnerability areresponsible for the maintenance of schizophre-nia in the population.89 Within the cluster ofsymptoms and signs which comprise thediagnostic entity of schizotypal personality dis-order aloofness, poor rapport, social dys-function, odd speech patterns, and avoidantsymptoms in particular predict genetic vul-nerability to schizophrenia.99

Other disorders which form part of theschizophrenia spectrum are schizoaffective dis-order,'00 paranoid personality disorder,98 andschizoid personality disorder,91 97 althoughthere is some debate about the second.98 Noexcess of anxiety disorder or alcoholism hasbeen found in the relatives of schizophrenicpatients compared with the relatives of con-trols, indicating that the genetic transmissionof schizophrenia and schizophrenia spectrumdisorders is relatively specific and does notinclude a generalised liability to all psychiatricillness.9 97 10' The debate about whether affec-tive disorders occur to excess in the relatives ofschizophrenic patients has not yet beenresolved,8810102 although relatives of schizo-phrenic patients seem to have an increasedpredisposition to develop psychotic symptomsas part of an affective illness.'o'

DEVELOPMENTAL ABNORMALITIESEvidence for developmental abnormalities inschizophrenia has come from two sources ofprospective data: (a) high risk studies and (b)general population birth cohorts.

High risk studiesThe high risk study design involves followingup offspring of schizophrenic mothers untilthey have passed through the period of risk forschizophrenia. Only the Copenhagen high risk

608

on May 22, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. D

ownloaded from

Cannon, 3rones

study'03 has completed follow up of itsprobands, but because probands were recruitedat age 15 there are no data on early develop-ment. The second generation of high risk studiesin schizophrenia, which were started during themiddle and late 1970s, began their observationson high risk children from the time of birth.The most influential of these studies are theNew York High Risk Project,'04 the JerusalemInfant Development Study,'05 and the high riskstudy of Barbara Fish.'06 As the subjects arenow still in their teens and 20s, full informationon outcome in the cohorts will not be availablefor another 10 years or so, but interesting dataon infant and early childhood developmentamong the high risk children have already beenpublished and show a high degree of consis-tency between studies.The developmental abnormalities found in

25%-56% of high risk children during differentstages of childhood can be summarised as fol-lows:

* Neonatal period: hypoactivity; extreme varia-tion in alertness; hypotonia; poor "cuddli-ness"

* Infancy: a delay in acquisition of develop-mental milestones and a disordered patternof acquisition of milestones-termed"pandysmaturation" by Fish'06 107

* Early childhood: soft neurological signs, inparticular poor motor coordination

* Late childhood: deficits in information pro-cessing and deficits in attention.

The question which remains to be answered iswhether the children who have displayed theseneurodevelopmental abnormalities throughoutchildhood will develop schizophrenia or a schiz-ophrenia spectrum disorder in adulthood.Results from general population birth cohortstudies (discussed below) and from a studybased on examination of childhood "homemovies" of schizophrenic patients'08 have repli-cated some of these developmental findings.

General population studiesPopulation based cohort studies can overcomethe problem of generalisability associated withhigh risk studies. Unfortunately the informationcollected during childhood, although safe fromselection and information bias, is often notideal. In the United Kingdom, two such studieshave now reported findings concerning schizo-phrenia.40 41The National Survey of Health and

Development, also referred to as the British1946 birth cohort, comprises a sample of 5362people born in the week 3-9 March 1946, whohave been regularly followed up over fourdecades. Jones and colleagues40 identified allcases of schizophrenia (30 in total) from thiscohort and using the detailed, unbiased assess-ments made during childhood, investigatedchildhood developmental risk factors for adultschizophrenia in these patients using theremaining sample as a comparison group.Children who went on to develop schizophreniaas adults could be distinguished from theirpeers in the following ways:

* Motor development milestones, in particu-lar walking, were delayed by an average of1-2 months

* More speech problems (odds ratio/OR 2-8)* Lower educational test scores at ages 8, 1 1,

and 15* Preference for solitary play at age 4 (OR

2-1)* When the children were aged 4, health visi-

tors rated the mother as having below aver-age mothering skills and understanding ofher child (OR 5 8): a finding which mayindicate deviance in the child, the mother,or both.

The National Child Development Survey, alsoknown as the British perinatal mortalitysurvey, comprises all those born in the week3-9 March 1958. Forty patients with schizo-phrenia were identified from this cohort. Aninvestigation of childhood behavioural charac-teristics found that at the age of 7 patientswere rated by their teachers as more sociallymaladjusted than controls.41 Findings regard-ing low IQ were similar to the 1946 cohort.

These developmental findings are presumedto show differences in the way the brain isdeveloping. Whether or not the developmentaldifferences themselves modify risk, they haveto be explained by any aetiological theory ofschizophrenia, genetic or otherwise, and add alongitudinal aspect to the phenotype.

NON-GENETIC ENVIRONMENTAL RISK FACTORSBirth complicationsCase-control studies have found that schizo-phrenic patients as a group expernence morebirth complications than controls,'09 but studiesmay have been subject to many types of biasincluding recall, selection, and publicationbias."10 Three studies have used a cohort fol-low up design and have found modestincreases in relative risk (about 2) for schizo-phrenia associated with certain birth complica-tions: chronic fetal hypoxia,"' low maternalweight,"'2 and rhesus incompatibility.'3

Place and time of birthThere is a small increase in risk for schizophre-nia (OR 1.15), among people born in winteror early spring.' '4 The reason for this season-of-birth effect is unknown, although it may bedue to infectious diseases."5 Being born in acity seems to be associated with a slightincrease in risk for schizophrenia (OR 1 38),after adjusting for socioeconomic factors,"6although not all studies have found such aneffect.40 Urbanisation, like social class, is aproxy measure for many other risk-increasingagents which are likely to operate synergisti-cally.

Prenatal risk factorsEcological studies have shown a modestincrease in risk of schizophrenia (OR 2 0),associated with exposure to influenza in thesecond trimester of fetal life, particularly forthe 1957 influenza epidemic."7-"9 Cohort fol-low up studies have had low power to examinethis association,'20 121 and the case remains

609 on M

ay 22, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. Dow

nloaded from

Schizophrenia

unproved. Nutritional deficiency during thefirst trimester may also increase risk (OR2 0).122 123 One intriguing, although unrepli-cated, study from Finland found that the riskof schizophrenia was higher (OR 6 2) amongoffspring of women who lost their spouse dur-ing pregnancy than for those whose spousedied during the child's first year, possibly indi-cating that severe maternal stress can affectfetal development.'24

Heavy cannabis intakeHeavy cannabis consumption at the age of 18was associated with an increased risk, adjustedfor confounders, of later psychosis (OR 2 3) ina large cohort of military conscripts inSweden.125 A dose-response relation was con-vincing but the direction of causality remainsin question.

Cerebral ventricular enlargementSmall case-control studies have been themainstay of neuroimaging research in schizo-phrenia and findings are remarkably contra-dictory. 126 It is likely that the previouslymentioned problems of bias, confounding,and inadequate power are at least partlyresponsible. The most consistent finding isthat schizophrenic patients have larger lateralcerebral ventricles than controls. The level ofrisk associated with ventricular enlargement,adjusted for intracranial volume, age, sex, eth-nicity, and social class, is about 2-0, althoughthe whole notion of categorical definitions ofpathology in terms of enlargement or shrink-age has been challenged.'27

RISK FACTORS AS CLUES TO AETIOLOGYExamination of the effect sizes of various riskfactors can give clues to causation. The tableshows that only genetic risk factors come any-where near the effect size expected for a strongcausal agent in schizophrenia. However, theexistence of so many environmental risk factorswith small effect sizes cannot be ignored. Thesemay be proxy measures for an as yet unrecog-nised major environmental causal agent, or mayact additively with each other or with chanceevents. They could also indicate the existenceof gene-environment interactions.

Best estimate sizes of various genetic and environmental risk factors for schizophrenia(expressed as odds ratios or relative risks)

Best estimate ofCategory of nrsk factor Specific risk factors effect size

Genetic* MZ twin of schizophrenic patients 46Child of two parents with schizophrenia 40DZ twin of schizophrenic patient 14Child of one schizophrenic patient 10Sibling of schizophrenic patient 10Parent of schizophrenic patient 5

Developmentalt Delayed milestones 3Speech problems 3Cerebral ventricular enlargement 2

Postnatal Immigrant/ethnic minority status 5environmentt Chronic cannabis consumption 2

Solitariness as child 2Prenatal and perinatal Birth complications 2environmentt Severe undemutrition (1st trimester) 2

Maternal influenza (2nd trimester) 2Born in city 1-4Born in winter/early spring 1.1

*Relative risks; todds ratios

Gene-environment interactionTraditional aetiological models have consid-ered vulnerability to schizophrenia as the sumof the impact of genetic and environmentalrisk factors-an additive model. But the con-cept of gene-environment interaction that is,genes controlling sensitivity to the disease-predisposing effects of the environment, canno longer be neglected,128 with many examplesbeing found in the fields of medicine'29 andpsychiatry. 130

Both psychosocial and biological environ-mental factors may interact with genetic vul-nerability to increase the risk forschizophrenia. Initial results from the Finnishadoption study of schizophrenia show agreater risk of psychosis among the biologicaloffspring of schizophrenic mothers who wereplaced in poorly functioning adoptive homesthan among those in well functioning adoptivehomes.93 High risk children who were rearedin a kibbutz had increased rates of schizophre-nia in adulthood compared with those whowere reared in family homes, whereas control(low risk) children reared in a kibbutz had noincreased risk of psychopathology.'3' A seriesof analyses from the Copenhagen high riskstudy has shown that birth complications mayinteract with genetic factors to increase therisk of schizophrenia.92 132133

In other words, the pathogenic effects ofbirth complications, an adverse adoptivehome, or some aspect of life in a kibbutz seemto increase risk of schizophrenia only in chil-dren who already have some degree of geneticvulnerability. Although the evidence is prelim-inary, it indicates that the environment shouldno longer be considered a "nuisance variable"by psychiatric geneticists. Common environ-mental factors may work in combination withrare high risk genotypes to maintain schizo-phrenia in the population.

NEW STRATEGIES IN MOLECULAR GENETICSThe search for the "psychosis gene"'134 has metwith substantial difficulties, including lack of avalid phenotype; the possibility of genetic het-erogeneity; the possibility of environmentaleffects acting alone or in combination withgenetic factors; and lack of knowledge aboutthe exact mode of transmission.94 Traditionallinkage techniques are only able to exclude theexistence of single major genes. As the geneticaetiology of schizophrenia is likely to be com-plex, involving genes of small effect, new tech-niques, and strategies are necessary.9094 134

Linkage studiesLinkage strategies for detecting genes of smalleffect include non-parametric techniques,such as affected sib pair and affected relativepair approaches, which make no assumptionsabout mode of transmission. A recent varia-tion on the sib pair design is the quantitativetrait locus approach which involves measuringliability to schizophrenia as a continuous vari-able in siblings.135 136 The application of thetechniques of behavioural genetics to the studyof schizophrenia is likely to prove a fruitfularea for the future.'35

610 on M

ay 22, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. Dow

nloaded from

Cannon,3'ones

After a false positive report of linkage onchromosome 5,137 current interest centres onchromosome 22138 and particularly on chro-mosome 6.139 A large study of 265 Irish multi-plex families has found evidence for linkage onchromosome 6p24-22 using a model assuminga broad phenotype and genetic heterogene-ity.'39 This locus is thought to confer suscepti-bility to schizophrenia in 1 5%-30% offamilies studied.

Association studiesAllellic association studies, which compare thefrequency of marker alleles in a sample ofpatients compared with ethnically matchedcontrols, are another method useful for exam-ining genes of small effect."37 An associationbetween HLA A9 and schizophrenia has beenfound, but this effect is weak (OR 1 6), and itssignificance is uncertain.90 Another promisingarea for the future is the use of newly devel-oped DNA polymorphisms to examine candi-date genes (genes coding for proteins whichare likely a priori to be involved in the patho-genesis of schizophrenia)-for example, genesinvolved in neurodevelopment. 140 Thisapproach offers the possibility of elucidatingthe cause of schizophrenia at a cellular level,bridging the gap between epidemiology andthe basic sciences.

ConclusionSchizophrenia occurs in all countries and in allraces. On the whole, there is little variation inincidence or prevalence between countries,but there seems to be an increase in the inci-dence of schizophrenia when populationsmigrate. The reason for this immigration effectis unknown and is currently the subject ofintensive study in the United Kingdom.Schizophrenia is associated with considerablemorbidity and high mortality, and more infor-mation is needed on predictors of suicideamong patients. However, on the positive side,the clinical outcome is not as hopeless as previ-ously thought-almost 20% of patients make afull recovery after the first episode of psychosisand there is suggestive, although not conclu-sive, evidence for a decrease in incidence inthe developing world.The aggregation of schizophrenia in fami-

lies, the evidence from twin and adoptionstudies, and the lack of variation in incidenceworld wide, indicate that schizophrenia is pri-marily a genetic condition, although environ-mental risk factors are also involved at somelevel as necessary, sufficient, or interactivecauses. The persistence of schizophrenia in thepopulation despite low fertility and high mor-tality, suggests that genetic transmissionoccurs principally through persons who do nothave the illness. Further refinement of thediagnostic concept of the schizophrenia spec-trum in both cross sectional and longitudinalterms will help elucidate the mode of trans-mission, and reliable instruments must bedeveloped to detect these disorders in thecommunity.

Areas of particular interest for the future are

studies of risk factors and developmentalabnormalities in those with known genetic vul-nerability for schizophrenia-for example, sib-lings; the use of new molecular geneticstrategies-including behavioural genetics-and the study of gene-environment interac-tions. If current progress in epidemiologicaland genetic research in schizophrenia contin-ues, it is likely that the aetiology of this com-plex disorder will at least begin to beunravelled before the "decade of the brain" isover.

MC is supported by a Wellcome Trust Research TrainingFellowship in clinical epidemiology.

1 Cooper JE, Kendell RE, Gurland BJ, et al. Psychiatricdiagnosis in New York and London. Oxford: OxfordUniversity Press, 1972.

2 American Psychiatric Association. Diagnostic and statisticalmanual. 3rd ed. Washington DC: APA, 1980.

3 American Psychiatric Association. Diagnostic and statisticalmanual. 3rd ed, revised. Washington DC: APA, 1987.

4 American Psychiatric Association. Diagnostic and statisticalmanual. 4th ed. Washington DC: APA, 1994.

5 World Health Organisation. TheICD-10 classification ofmental and behavioural disorders: diagnostic criteria forresearch. Geneva, Switzerland: WHO, 1994.

6 Andreasen N. Changing concepts of schizophrenia andthe ahistorical fallacy. Am Jf Psychiatry 1994;151:1405-7.

7 Plum F. Prospects for research on schizophrenia. 3.Neuropsychology. Neuropathological findings. Neuro-sciences ResearchProgram Bulletin 1972;10:384-8.

8 Lewis G, Pelosi A.The case-control study in psychiatry.Brj Psychiatry 1990;157: 197-207.

9 WingJK, Cooper JE, Sartorius N. The measurement andclassification of psychiatric symptoms. Cambridge:Cambridge University Press. 1974.

10 Jablensky A, Sartorius N, Emberg G, et al. Schizophrenia:manifestations, incidence and course in different cul-tures. A World Health Organisation ten country study.Psychol Med 1992; monograph supplement 20.

11 Keith SJ, Regier DA, Rae DS. Schizophrenic disorders.In: Robins LN, Regier DA, eds. Psychiatric disorders inAmerica: the Epidemiologic Catchment Area Study. NewYork: Free Press, 1991;33-52.

12 Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and12-month prevalence of DSM-III-R psychiatric disor-ders in the United States: results from the NationalComorbidity Survey. Arch Gen Psychiatry 1994;52:8-19.

13 Office of Population Censuses and Surveys. OPCS surveysof psychiatric morbidity in Great Britain: bulletin No 1.The prevalence of psychiatric morbidity among adults aged16-64 living in private households in Great Britain.London: OPCS.

14 Mason P, Wilkinson G. The prevalence of psychiatricmorbidity. OPCS survey of psychiatric morbidity inGreat Britain. BrJ Psychiatry 1996;168:1-3.

15 Eagles JM, Whalley U. Decline in the diagnosis of schizo-phrenia among first admissions to the Scottish mentalhospitals from 1969-78. Br J Psychiatry 1985;146:151-4.

16 Der G, Gupta S, Murray RM. Is schizophrenia disappear-ing? Lancet 1990;335:513-6.

17 Castle D, Wessley S, Der G, Murray RM. The incidence ofoperationally-defined schizophrenia in Camberwell1965-84. BrJPsychiatry 1991;159:790-4.

18 Bamrah JS, Freeman HL, Goldberg DP. Epidemiology ofschizophrenia in Salford 1974-1984. Changes in anurban community over ten years. Br Jf Psychiatry1991;159:802-10.

19 Harrison G, Cooper JE, Gancarczyk R. Changes in theadministrative incidence of schizophrenia. Br JPsychiatry 1991;159:811-16.

20 De Alarcon J, Seagroatt V, Goldacre M. Trends in schizo-phrenia. Lancet 1990;335:852-3.

21 Eagles JM, Hunter D, McCance C. Decline in the diagno-sis of schizophrenia among first contacts with psychiatricservices in North East Scotland, 1969-84. Br JPsychiatry 1988;152:793-8.

22 Kendell RE, Malcolm DE, Adams W. The problem ofdetecting changes in the incidence of schizophrenia. BrJ_Psychiatry 1993;162:212-8.

23 Geddes JR, Black RI, Whalley Uj, et al. Persistence of thedecline in the diagnosis of schizophrenia among firstadmissions to Scottish hospitals from 1969-1988. Br3JPsychiatry 1993;163:620-6.

24 Munk-Jorgensen P. Decreasing first-admission rates ofschizophrenia among males in Denmark from 1970 to1984. Acta Psychiatr Scand 1986;73:645-50.

25 Munk-Jorgensen P, Mortensen PB. Incidence and otheraspects of the epidemiology of schizophrenia inDenmark, 1971-87. Br J Psychiatry 1992;161:489-95.

26 Joyce PR. Changing trends in first admissions and read-

611 on M

ay 22, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. Dow

nloaded from

Schizophrenia

missions for mania and schizophrenia in New Zealand.Aus NZJ Psychiatry 1987;21:82-6.

27 Nicole L, Lesage A, Lalonde P. Lower incidence andincreased male:female ratio in schizophrenia. Br JPsychiatry 1992;161:557-6.

28 Waddington JIL, Youssef HA. Evidence for gender-spe-cific decline in the rate of schizophrenia in a rural popu-lation over a fifty-year period. Br J7 Psychiatry 1994;164:171-6.

29 Stoll AL, Tohen M, Baldessarini RJ, et al. Shifts in diag-nostic frequencies of schizophrenia and major affectivedisorders at six North American psychiatric hospitals,1972-88. AmJ Psychiatry 1993;150: 1668-73.

30 Oldehinkel AJ, Giel R. Time trends in the care-based inci-dence of schizophrenia. BrJ Psychiatry 1995;167:777-8.

31 Jablensky A. Schizophrenia: recent epidemiologic issues.Epidemiol Rev 1995;17:10-20.

32 Keith SJ, Regier DA, Rae DS. Schizophrenic disorders.In: Robins L, Regier DA, eds. Psychiatric Disorders inAmerica. New York: Free Press, 1991:33-52.

33 Asarnow RF, Asarnow JR. Childhood onset schizophre-nia. Schizophr Bull 1994;20:591-7.

34 Castle D, Murray RM. The epidemiology of late-onsetschizophrenia. Schizophr Bull 1993;19:691-700.

35 Van Os J, Howard R, Takei N, Murray R. Increasing age isa risk factor for psychosis in the elderly. Soc PsychiatryPsychiatr Epidemiol 1995;30:161-4.

36 Loranger AW. Sex differences in age at onset of schizo-phrenia. Arch Gen Psychiatry 1984;41:157-61.

37 Lewine RRJ. Sex differences in age of symptom onsetand first hospitalisation in schizophrenia. Am JOrthopsychiatry 1980;50:316-22.

38 Hafner H, Maurer K, Loeffler W, Reicher-Rossler A. Theinfluence of age and sex on the onset and early course ofschizophrenia. BrJ Psychiatry 1993;162:80-6.

39 Hafner H, Reicher-Rossler A, An der Heiden H, MaurerK, Fatkenheuer B, Loffler W. Generating and testing acausal explanation of the gender difference in age at firstonset of schizophrenia. Psychol Med 1993;23:924-40.

40 Jones P, Rodgers B, Murray R, Marmot M. Childhooddevelopmental risk factors for schizophrenia in the 1946national birth cohort. Lancet 1994;344:1398-402.

41 Done DJ, Crow TJ, Johnstone EC, Sacker A. Childhoodantecedents of schizophrenia and affective illness: socialadjustment at ages 7 and 11. BMJ 1994;309:699-703.

42 Aro S, Aro H, Kesmaki I. Socioeconomic mobility amongpatients with schizophrenia or major affective disorder: a17-year retrospective follow-up. Br J Psychiatry 1995;166:759-67.

43 Jones PB, Bebbington P, Foerster A, et al. Premorbidsocial underachievement in schizophrenia: results fromthe Camberwell collaborative psychosis study. Br JPsychiatry 1993;162:65-71.

44 Tien AY, Eaton WW. Psychopathological precursors andsociodemographic risk factors for the schizophreniaspectrum. Arch Gen Psychiatry 1992;49:37-46.

45 Book JA, Wetterberg L, Modrzewska K. Schizophrenia ina North Swedish geographical isolate 1900-1977: epi-demiology, genetics and biochemistry. Clinical Genetics1978;14:373-94.

46 Torrey EF. Schizophrenia and civilisation. New York: JasonAronson, 1980.

47 Torrey EF. Prevalence studies in schizophrenia. Br J7Psychiatry 1987;150:598-608.

48 Crocetti GJ, Lemkau PV, Kulcar Z, Kesic B. Selectedaspects of the epidemiology of psychoses in Croatia,Yugoslavia. II. The cluster sample and the results of thepilot survey. Am J Epidemiology 197 1;94: 126-34.

49 Eaton WW. Update on the epidemiology of schizophre-nia. Epidemiol Revs 1991;13:320-8.

50 Jablensky A. Schizophrenia: the epidemiological horizon.In: Hirsch S, Weinberger D, eds. Schizophrenia. Oxford:Blackwell Science, 1995:206-53.

51 Harrison G, Owens D, Holton A, Neilson D, Boot D. Aprospective study of severe mental disorder in Afro-Caribbean patients. Psychol Med 1988;18:643-57.

52 Wessley S, Castle D, Der G, Murray R. Schizophreniaand Afro-Caribbeans. A case-control study. Br JPsychiatry 1991;159:795-801.

53 Thomas CS, Stone K, Osborn M, Thomas PF, Fisher M.Psychiatric morbidity and compulsory admission amongUK-born Europeans, Afro-Caribbeans and Asians incentral Manchester. Bry Psychiatry 1993;163:91-99.

54 King M, Coker E, Leavey G, Hoare A, Johnson-Sabine E.Incidence of psychotic illness in London: a comparisonof ethnic groups. BMJ 1994;309:1115-9.

55 Van Os J, Castle DJ, Takei N, Der G, Murray RM.Psychotic illness in ethnic minorities: clarification fromthe 1991 census. Psychol Med 1996;26:203-8.

56 Selten JP, Sijben N. First admission rate for schizophreniain immigrants to The Netherlands: the Dutch NationalRegister. Soc Psychiatry Psychiatr Epidemiol 1994;29:71-2.

57 Odegaard 0. Emigration and insanity. Acta PsychiatrScand 1932; suppl 4.

58 Burke AW. First admission rates and planning in Jamaica.Soc Psychiatry 1974;15:17-19.

59 Hickling FW. Psychiatric hospital admission rates inJamaica, 1971 and 1988. Br J Psychiatry 1991;159:817-21.

60 McGovern D, Cope RV. First psychiatric admission rateof first and second generation Afro-Caribbeans. SocPsychiatry 1987;22:139-49.

61 Warner R. Time trends in schizophrenia: changes in

obstetric risk factors with industrialization. SchizophrBull 1995;21:483-500.

62 Eagles JM. The relationship between schizophrenia andimmigration. Are there alternative hypotheses? Br JPsychiatry 1991;159:783-9.

63 Sugarman PA, Craufurd D. Schizophrenia in the Afro-Caribbean community. Br Jf Psychiatry 1994;164:474-80.

64 Ram R, Bromet EJ, Eaton WW, Pato C, Schwartz JE. Thenatural course of schizophrenia: a review of first admis-sion studies. Schizophr BuUl 1992;18:185-207.

65 Leff J, Sartorius N, Jablensky A, Korten A, Ernberg G.The International Pilot Study of Schizophrenia: five-year follow-up findings. PsycholMed 1992;22: 131-45.

66 The Scottish Schizophrenia Research Group. TheScottish first episode schizophrenia study. VIII. Fiveyear follow-up: clinical and psychosocial findings. Br JfPsychiatry 1992;161:496-500.

67 Tohen M, Stoll AL, Strakowski SM, et al. The McLeanfirst-episode psychosis project: six-month recovery andrecurrence outcome. Schizophr Bull 1992;18:273-82.

68 Geddes J, Mercer G, Frith CD et al. Prediction of out-come following a first episode of schizophrenia: a follow-up study of Northwick Park first episode study subjects.BrJ Psychiatry 1994;165:664-8.

69 Shepherd M, Watt D, Falloon I, Smeeton N. The naturalhistory of schizophrenia: a five-year follow-up study ofoutcome and prediction in a representative sample ofschizophrenics. Psychol Med 1989; monograph supple-ment 15.

70 Hegarty JD, Baldessarini RJ, Tohen M, Waterman C,Oepen G. One hundred years of schizophrenia: a meta-analysis of the outcome literature. Am J Psychiatry1994;151:1409-16.

71 Bleuler M. The Schizophrenic disorders: long-term patientand family studies. New Haven, CT: Yale UniversityPress, 1978.

72 Leff J, Wig NN, Bedi H, et al. Relatives expressed emo-tion and the course of schizophrenia in Chandigargh: atwo-year follow-up study of a first contact sample. Br JfPsychiatry 1990;156:351-6.

73 McKenzie K, Van Os J, Fahy T, Jones P, Harvey I, TooneB, Murray R. Psychosis with good prognosis in Afro-Caribbean people now living in the United Kingdom.BMJ 1995;311:1325-8.

74 Eaton WW. Update on the epidemiology of schizophre-nia. Epidemiol Rev 1991;13:320-8.

75 Van Os J, Fahy T, Bebbington P, et al. The influence oflife events on the subsequent course and outcome ofpsychotic illness. A prospective study of the CamberwellCollaborative Psychosis Study. Psychol Med 1994;24:503-13.

76 Brown GW, Birley JLT, Wing JK. Influence of family lifeon the course of schizophrenic disorders. BrJ Psychiatry1972;121:241-58.

77 LeffJ, Kuipers L, Berkowitz R, Vaughan C, Sturgeon D.Life events, relative's expressed emotion and mainte-nance neuroleptics in schizophrenic relapse. Psychol Med1983;13:799-800.

78 Linzen DH, Dingemans PM, Lenior ME. Cannabis abuseand the course of recent-onset schizophrenic disorders.Arch Gen Psychiatry 1994;51:273-9.

79 Johnstone EC, Crow TJ, Johnson AL, MacMillan JF. TheNorthwick Park study of first episode schizophrenia. 1.Presentation of the illness and problems relating toadmission. BrJ7 Psychiatry 1986;148: 115-20.

80 Wyatt RJ. Neuroleptics and the natural course of schizo-phrenia. Schizophr BuU 1991;17:325-51.

81 Loebel AD, Lieberman JA, Alvir JMJ, et al. Duration ofpsychosis and outcome in first-episode schizophrenia.AmJ Psychiatry 1992;149:1183-8.

82 Mortensen PB, Juel K. Mortality and causes of death infirst-admitted schizophrenic patients. Br J Psychiatry1993;163:183-9.

83 Newman SC, Bland RC. Mortality in a cohort of patientswith schizophrenia: a record linkage study. Can J7Psychiatry 1991;36:239-45.

84 Roy A. Suicide in chronic schizophrenia. Br J Psychiatry1982;141:171-7.

85 Carpenter WT, Buchanan RW. Schizophrenia. N Engl JMed 1994;330:681-90.

86 Weinberger DR. Schizophrenia: from neuropathology toneurodevelopment. Lancet 1995;346:552-7.

87 Davis KL, Kahn RS, Ko G, Davidson M. Dopamine inschizophrenia: a review and reconceptualisation. Am JPsychiatry 1991;148: 1474-86.

88 Maier W, Lichtermann D, Minges J, et al. Continuity anddiscontinuity of affective disorders and schizophrenia:results of a controlled family study. Arch Gen Psychiatry1993;50:871-83.

89 Kendler KS, McGuire M, Gruenberg AM, O'Hare A,Spellman M, Walsh D. The Roscommon Family Study.I: Methods, diagnosis of probands and risk of schizophre-nia in relatives. Arch Gen Psychiatry 1993;50:527-40.

90 McGuffin P, Owen M, Farmer AE. Genetic basis of schiz-ophrenia. Lancet 1995;346:678-82.

91 Kety SS, Wender PH, Jacobsen B, et al. Mental illness inthe biological and adoptive relatives of schizophrenicadoptees. Arch Gen Psychiatry 1994;51:442-55.

92 Kendler KS, Gruenberg AM, Kinney DK. Independentdiagnoses of adoptees and relatives as defined by DSM-III criteria, in the provincial and national samples of theDanish adoption study of schizophrenia. Arch GenPsychiatry 1994;51:456-68.

612 on M

ay 22, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. Dow

nloaded from

Cannon, Jones

93 Tienari P. Interaction between genetic vulnerability andfamily environment: theFinnish adoptive study of schiz-ophrenia. Acta PsychiatrScand 199 1;84:460-5.

94 Kendler KS, Diehl SR. Schizophrenia: genetics. In:Kaplan HI, Sadock BJ, eds. Comprehensive textbook ofpsychiatry VI. Vol 1. Baltimore: Williams and Wilkins,1995:942-57.

95 Gottesman II, Shields J. Schizophrenia, the epigenetic puz-zle. Cambridge: Cambridge University Press, 1982.

96 GottesmanII, Bertelsen A. Confirming unexpressedgenotypes for schizophrenia. Risks in the offspring ofFischer's Danish identical and fraternal twins. Arch GenPsychiatry 1989;46:867-72.

97 Parnas J, Cannon TD, Jacobsen B, et al. Lifetime DSM-III-R diagnostic outcomes in the offspring of schizo-phrenic mothers: results from the Copenhagen high-riskstudy. Arch Gen Psychiatry 1993;50:707-14.

98 Kendler KS, McGuire M, Gruemberg AM, et al. TheRoscommon family study. III. Schizophrenia-relatedpersonality disorders in relatives. Arch Gen Psychiatry1993;50:781-8.

99 Kendler KS, McGuire M, Gruenberg AM, Walsh D.Schizotypal symptoms and signs in the Roscommonfamily study: their factor structure and familial relation-ship with psychotic and affective disorder. Arch GenPsychiatry 1995;52:296-303.

100 Kendler KS, McGuire M, Gruenberg AM, et al. TheRoscommon family study. II.The risk of nonschizo-phrenic, nonaffective psychosis in relatives. Arch GenPsychiatry 1993;50:645-52.

101 Kendler KS, McGuire M, Gruenberg AM, et al. TheRoscommon family study. IV. Affective illness, anxietydisorders and alcoholism in relatives. Arch GenPsychiatry 1993;50:952-60.

102 Crow TJ. The continuum of psychosis and its genetic ori-gins. The sixty-fifth Maudsley lecture. BrJf Psychiatry1990;156:788-97.

103 Cannon TD, Mednick SA.The schizophrenia high-riskproject in Copenhagen: three decades of progress. ActaPsychiatr Scand 1993;370:33-47.

104 Erlenmeyer-Kimling L, Cornblatt B. The New York highrisk project: a follow-up report. Schizophr Bull 1987;13:451-62.

105 Marcus J, Hans SL, Auerbach JG, Auerbach AG.Children at risk for schizophrenia: the Jerusalem infantdevelopment study. Arch Gen Psychiatry 1993;50:797-809.

106 Fish B. Neurobiologic antecedents of schizophrenia inchildren. Arch Gen Psychiatry 1977;34:1297-313.

107 Fish B. Infants at risk for schizophrenia: sequelae of agenetic neurointegrative defect. Arch Gen Psychiatry1992;49:221-35.

108 Walker E, Lewine RJ. Prediction of adult-onset schizo-phrenia from childhood home movies of the patients.Am J Psychiatry 1994;147: 1052-6.

109 Lewis SW, Owen MJ, Murray RM. Obstetric complica-tions and schizophrenia: methodology and mechanisms.In: Schultz SC, Tamminga CA, eds. Schizophrenia: a sci-entific focus. New York; Oxford University Press,1989:56-9.

110 Geddes JR, Lawrie SM. Obstetric complications andschizophrenia: a meta-analysis. Br Jf Psychiatry 1995;167:786-93.

111 Buka S, Tsuang MT, Lipsitt LP. Pregnancy-delivery com-plications and psychiatric diagnosis: a prospective study.Arch Gen Psychiatry 1993;50:151-6.

112 Done DJ, Johnstone EC, Frith CD, et al. Complicationsof pregnancy and delivery in relation to psychosis inadult life: data from the British perinatal mortality sur-vey sample. BMJ 1991;302:1576-80.

113 Hollister M, Laing P, Mednick SA. Rhesus incompatibilityas a risk factor for schizophrenia in male adults. ArchGen Psychiatry 1996;53:19-24.

114 Bradbury TN, Miller GA. Season of birth in schizophre-nia: a review of the evidence, methodology and etiology.Psychol Bull 1985;98:569-94.

115 Cotter D, Larkin C, Waddington JL, O'Callaghan E.Season of birth in schizophrenia: clue or cul-de-sac? In:Waddington JL, Buckley PB, eds. The Neuro-developmental Basis of Schizophrenia. Georgetown: RGLandes Co, 1996:17-30.

116 Lewis G, David A, Andreasson S, Allebeck P.Schizophrenia and city life. Lancet 1992;340:137-40.

117 Mednick SA, Machon RA, Huttunen MO, Bonett D.Adult schizophrenia following prenatal exposure to aninfluenza epidemic. Arch Gen Psychiatry 1988;45: 171-6.

118 O'Callaghan E, Sham P, Takei N, Glover G, Murray RM.Schizophrenia after prenatal exposure to the 1957 A2influenza epidemic. Lancet 1991;337:1248-50.

119 Adams W, Kendell RE, Hare EH, Munk-Jorgensen P.Epidemiological evidence that maternal influenza con-tributes to the aetiology of schizophrenia: an analysis ofScottish, English and Danish data. BrJf Psychiatry1993;163:522-34.

120 Crow TJ, Done DJ. Prenatal exposure to influenza doesnot cause schizophrenia. Br J Psychiatry 1992;161:390-3.

121 Cannon M, Cotter D, Coffey VP, et al. Schizophreniaafter prenatal exposure to the 1957 influenza epidemic: afollow-up study. BrJ Psychiatry 1996; 168:368-71.

122 Susser E,Lin P. Schizophrenia after prenatal exposure tothe Dutch hunger winter of 1944-1945. Arch GenPsychiatry 1992;49:983-8.

123 Susser E, Neugebauer R, Hoek HW, et al. Schizophreniaafter prenatal famine: further evidence. Arch GenPsychiatry 1996;53:25-31.

124 Huttunen MO, Niskanen P. Prenatal loss of father andpsychiatric disorders. Arch Gen Psychiatry 1978;35:429-31.

125 Andreasson S, Allebeck P, Engstrom A, Rydberg U.Cannabis and schizophrenia. Lancet 1987;2:1483-6.

126 Chua SE, McKenna PJ. Schizophrenia-a brain disease?A critical review of structural and functional cerebralabnormality in the disorder. Br J Psychiatry 1995;166:563-82.

127 Jones PB, Harvey I, Lewis SW, et al. Cerebral ventricledimensions as risk factors for schizophrenia and affectivepsychosis: an epidemiological approach to analysis.Psychol Med 1994;24:995-101 1.

128 Kendler KS. Genetic epidemiology in psychiatry: takingboth genes and environment seriously. Arch GenPsychiatry 1995;52:895-9.

129 Khoury MJ, Beaty TH, Newill CA, BryantS, Cohen BH.Genetic-environmental interactions in chronic airwaysobstruction. AmJEpidemiol 1986;15:65-72.

130 Kendler KS, Kessler RC, Walters EE, et al. Stressful lifeevents, genetic liability and onset of an episode of majordepression in women. Am J Psychiatry 1995;152:833-42.

131 Mirsky AF, Silberman EK, Latz A, Nagler S. Adult out-comes of high-risk children: differential effects of townand kibbutz rearing. Schizophr Bull 1985;11:150-6.

132 Cannon TD, Mednick SA, Parnas J, et al. Developmentalbrain abnormalities in the offspring of schizophrenicmothers. I: Contributions of genetic and environmentalfactors. Arch Gen Psychiatry 1993;50:551-64.

133 Cannon TD, Mednick SA, Parnas J, et al. Developmentalbrain abnormalities in the offspring of schizophrenicmothers. II: Structural brain characteristics of schizo-phrenia and schizotypal personality disorder. Arch GenPsychiatry 1994;51:955-62.

134 Crow TJ. The search for the psychosis gene. Br JPsychiatry 1991;158:611-4.

135 Plomin R, Owen MJ, McGuffin P. The genetic basis ofcomplex human behaviours. Science 1994;264:1733-9.

136 Chapman JP, Chapman LJ, Kwapil TR. Scales for theassessment of schizotypy. In: Raine A, Lencz T,Mednick SA, eds. Schizotypal personality. New York:Cambridge University Press, 1995: Ch 5.

137 Sherrington R, Brynjolfsson J, Petursson H, et al.Localisation of a susceptibility locus for schizophreniaon chromosome 5. Nature 1988;336:164-7.

138 Vallada HP, Gill M, Sham P, et al. Linkage studies onchromosome 22 in familial schizophrenia. Am J MedGenet 1995;60: 139-46.

139 Straub RE, Maclean CI, O'Neill A, et al. A potential vul-nerability locus for schizophrenia on chromsomes 6p24-22: evidence for genetic interogeneity. NatureGenetics 1995;ii:287-93.

140 Jones P, Murray R. The genetics of schizophrenia is thegenetics of neurodevelopment. Br J Psychiatry 1991;158:615-23.

613

on May 22, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.60.6.604 on 1 June 1996. D

ownloaded from