Schizophrenia, "just the facts" what we know in 2008. 2. Epidemiology and etiology.

Abstract

Although we have studied schizophrenia as a major disease entity over the past century, its causes and pathogenesis remain obscure. In this article, we critically review genetic and other epidemiological findings and discuss the insights they provide into the causes of schizophrenia. The annual incidence of schizophrenia averages 15 per 100,000, the point prevalence averages approximately 4.5 per population of 1000, and the risk of developing the illness over one's lifetime averages 0.7%. Schizophrenia runs in families and there are significant variations in the incidence of schizophrenia, with urbanicity, male gender, and a history of migration being associated with a higher risk for developing the illness. Genetic factors and gene-environment interactions together contribute over 80% of the liability for developing schizophrenia and a number of chromosomal regions and genes have been "linked" to the risk for developing the disease. Despite intensive research and spectacular advances in molecular biology, however, no single gene variation has been consistently associated with a greater likelihood of developing the illness and the precise nature of the genetic contribution remains obscure at this time. Environmental factors linked to a higher likelihood of developing schizophrenia include cannabis use, prenatal infection or malnutrition, perinatal complications, and a history of winter birth; the exact relevance or nature of these contributions is, however, unclear. How various genetic and environmental factors interact to cause schizophrenia and via which precise neurobiological mechanisms they mediate this effect is not understood. Etiological heterogeneity, complex patterns of gene-gene and gene-environment interaction, and inadequately elucidated schizophrenia pathophysiology are among the explanations invoked to explain our inadequate understanding of the etio-pathogenesis of schizophrenia. The ability to question some of our basic assumptions about the etiology and nature of schizophrenia and greater rigor in its study appear critical to improving our understanding about its causation.

Full text

Page 1

Schizophrenia, “Just the Facts” What we know in 2008.
2. Epidemiology and etiology
Rajiv Tandona, Matcheri S. Keshavanb, Henry A. Nasrallahc
aUniversity of Florida, 3706 Glin Circle, Tallahassee, FL 32309, United States
bWayne State University, Detroit, Michigan, United States
cUniversity of Cinncinnatti, Cinncinnatti, Ohio, United States
Received 15 February 2008; received in revised form 31 March 2008; accepted 2 April 2008
Abstract
Although wehave studied schizophreniaasa major disease entity overthe past century, its causesand pathogenesis remainobscure.
In this article, we critically review genetic and other epidemiological findings and discuss the insights they provide into the causes of
schizophrenia. The annual incidence of schizophrenia averages 15 per 100,000, the point prevalence averages approximately 4.5 per
populationof1000,andtheriskofdevelopingtheillnessoverone'slifetimeaverages0.7%.Schizophreniarunsinfamiliesandthereare
significant variations in the incidence of schizophrenia, with urbanicity, male gender, and a history of migration being associated with a
higherriskfordevelopingtheillness.Geneticfactorsandgene-environmentinteractionstogethercontributeover80%oftheliabilityfor
developing schizophrenia and a number of chromosomal regions and genes have been “linked” to the risk for developing the disease.
Despite intensive research and spectacular advances in molecular biology, however, no single gene variation has been consistently
associated with a greater likelihood of developing the illness and the precise nature of the genetic contribution remains obscure at this
time. Environmental factors linked to a higher likelihood of developing schizophrenia include cannabis use, prenatal infection or
malnutrition, perinatal complications, and a history of winter birth; the exact relevance or nature of these contributions is, however,
unclear. How various genetic and environmental factors interact to cause schizophrenia and via which precise neurobiological
mechanismstheymediatethiseffectisnotunderstood.Etiologicalheterogeneity,complexpatternsofgene–geneandgene–environment
interaction, and inadequately elucidated schizophrenia pathophysiology are among the explanations invoked to explain our inadequate
understanding of the etio-pathogenesis of schizophrenia. The ability to question some of our basic assumptions about the etiology and
nature of schizophrenia and greater rigor in its study appear critical to improving our understanding about its causation.
© 2008 Elsevier B.V. All rights reserved.
Keywords: Schizophrenia; Facts; Epidemiology; Genetics; Environmental risk factors; Incidence; Chromosomes; Genes; Prevalence; Linkage;
Association; Pregnancy; Etiology; Causes
1. Introduction
Epidemiology is the study of distribution and
determinants of disease (MacMahon and Pugh, 1970).
Distinguishing characteristics and experiences of persons
who develop a disease from those of individuals who do
not allows one to identify factors related to causation of
that disease. Determinants of disease constitute the
essenceofepidemiology;withreferencetoschizophrenia,
this includes both genetic and environmental risk factors
which need to be considered together since both are
important in the etiology of schizophrenia and neither
appears to operate in isolation (Tsuang et al., 2004).
In this paper, we summarize major epidemiological
findings in schizophrenia and discuss what they tell us
about genetic and environmental factors involved in its
Available online at www.sciencedirect.com
Schizophrenia Research 102 (2008) 1–18
www.elsevier.com/locate/schres
0920-9964/$ - see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.schres.2008.04.011

Page 2

causation. We first summarize current knowledge about
variations in the occurrence of schizophrenia across
populations, socio-demographic characteristics and
time. We then briefly review key genetic findings and
environmental risk factors linked to the development of
schizophrenia and critically discuss our state of under-
standing about the etiology of schizophrenia. We
highlight key conceptual issues, outline approaches to
dissecting genetic and environmental contributions to its
causation, and consider major challenges towards
elucidating the etio-pathogenesis of this disease.
2. Incidence and prevalence
The distribution of a disease is generally expressed in
terms of incidence (new cases), and prevalence (total
number of cases: existing+new). Incidence rate refers to
the number of new cases of the disease that develop over a
specific period of time among people who are at risk for
developing the disease. Instead of a rate (new cases per
populationatriskpertime),incidencecanalsobeexpressed
astheprobabilitythatanindividualwilldevelopaparticular
disease (e.g., lifetime risk). Differences between the
characteristics of people who develop versus do not
develop the disease help to define the risk and protective
factors for and against development of the disease,
respectively. Thus knowledge about the distribution of
newdisease development (incidence) over time,place,and
person points to etiological factors (determinants) for
developing the disease. Prevalence refers to the proportion
ofpeopleinacommunitywhohaveaparticulardiseaseata
given time (point prevalence) or over a given time-span
(periodprevalence)includingbothpeoplewithpre-existing
disease and those who newly develop the disease over this
specified time period.
2.1. Incidence and lifetime risk
Since incidence is a measure of the number of new
cases of a disease occurring among people who are at risk
for developing the disease over a specified period of time
(generally one year), it requires knowing who is a case at
the beginning (and excluding them from both the
numerator and denominator), the number of people at
risk for newly developing the disease (denominator), and
the number of people who newly develop the disease
(numerator)overthisspecifiedperiodoftime.Thenumber
ofnew cases can be determined by communitysurveys or
by identifying them as they seek services (first contact
with health worker, hospitalization, etc.). Correctly
determining the distribution of a disease depends upon
havingareliableandvaliddiagnosisofthediseaseandthe
ability to precisely demarcate those with disease from
those without it. In the absence of any pathognomonic
feature and several impediments to constructing a fully
reliable patient clinical profile, it becomes a greater art to
diagnoseschizophreniathanmostotherdiseases(Eatonet
al., 2007). Additionally, consistent case definition (when
does a possible case become a case — e.g., persons with
prodromal symptoms) and identification (e.g., a person
may not seek treatment/s and is thereby missed as a case)
can be problematic. Furthermore, although we have made
substantial advances over the past two decades in being
abletodiagnoseschizophreniawithgreaterreliability,itis
unclear if we have made any progress with regard to
validity (McCormick and Flaum, 2005).
These challenges have contributed, in significant part,
to varying estimates of incidence obtained across the
multitude of studies. In the only global study that directly
generatedincidencedata(WHO10-nationstudy,Sartorius
et al., 1986; Jablensky et al., 1992), the annual incidence
wasfoundtorangefrom16–40/100,000/yearusingbroad
criteria(ICD-9;WorldHealthOrganization,1978)and7–
14/100,000 using narrow criteria (CATEGO class S+
identifying nuclear schizophrenia; Wing et al., 1974) for
diagnosingschizophrenia.A10-foldhigherincidencerate
was detected in the U.S.A.-based five community site
National Institute of Mental Health Epidemiological
Catchment Area (ECA) program (Tien and Eaton,
1992), but these results were derived from community
surveys conducted by individuals with minimal or no
clinical experience rather than on the basis of service
provision and clinician diagnosis, which is how most
other studies derived their data; this likely resulted in
several false positives (Anthony et al.,1985; Regier et al.,
1998). Studies based on service provision, on the other
hand, might underestimate rates of schizophrenia because
some affected individuals may not seek treatment. A
recent meta-analysis of all published studies between
1965 through 2001 obtained a median incidence rate of
15.2/100,000/year with 80% confidence interval rates
(10th–90th decile) ranging from 8–43 per 100,000 per
year (McGrath et al., 2004). Data were derived from 55
studies conducted across 33 countries. Rates were not
found to vary by broad world regions or economic status
of the country (Saha et al., 2006). Contrary to prior
assumptions of uniform rates of schizophrenia across the
world, however, this meta-analysis revealed robust
variations in incidence of schizophrenia, with urbanicity,
migration, and male gender found to be associated with
a higher risk for developing schizophrenia.
Recent findings confirm and clarify longstanding
suggestions of a link between urbanicity and schizo-
phrenia (Faris and Dunham, 1939). For many decades,
2 R. Tandon et al. / Schizophrenia Research 102 (2008) 1–18

Page 3

there was controversy about whether this association
meantthaturbandwellingcausedschizophrenia(“breeder
hypothesis”) or whether persons with schizophrenia
migrated to urban settings (“selection hypothesis”). For
the past half-century, this argument was seemingly
resolvedinfavor ofthe later perspective andthe observed
association was ascribed to the “social drift” of persons
with schizophrenia to inner city areas with cheaper ac-
commodationandrelativeanonymity(Dohrenwendetal.,
1992). Even as the social drift phenomenon has been
corroborated, well designed recent studies have also
confirmed an association between urban birth and
upbringing (upto the age of 15) and an increased risk of
developing schizophrenia (Lewis et al., 1992; Mortensen
etal.,1999;Kirkbrideetal.,2006).Thefindingofadose–
response relationship between degree of urbanicity and
risk of schizophrenia strongly supports the proposition
thatsomefactorassociatedwithurbanicitymaybecausally
related to schizophrenia (Pedersen and Mortensen, 2001).
What that specific risk-modifying factor linked to
urbanicity might be, however, is unclear. Several
candidates have been proposed — these include urban–
rural differences in rates of cannabis and other substance
use, prenatal and perinatal health, degree of social stress
and social connectedness, poverty, rates and nature of
migration, environmental toxins, various infectious dis-
eases, or vitamin D deficiency. Whereas some of these
factors have independently been linked to schizophrenia
(e.g., migration) and others to urbanicity (e.g., vitamin D
deficiency), none has been convincinglylinked toboth or
established as the schizophrenia risk-modifying factor
that satisfactorily explains the urbanicity–schizophrenia
connection.
Ever since Odegaard (1932) documented a higher
occurrence of “schizophrenic breakdown” among Nor-
wegians who had migrated to Minnesota than among
those who remained in Norway, several studies have
confirmed an association between migration and an
increased risk of developing schizophrenia (Malzberg,
1964; Bhugra, 2004). A meta-analysis of 18 studies
published between 1977 through 2003 identified a
personal or family history of migration as a significant
risk factor for schizophrenia (Cantor-Graae and Selten,
2005); the relative risk for developing schizophrenia was
foundtobe2.7forfirst-generationimmigrantsand4.5for
second-generation immigrants. Both selective migration
(Odegaard, 1932) and diagnostic bias (Sashidharan,
1993) have been cited as explanations; neither of them
appears to adequately account for the association. The
migration–schizophrenia link has been found to be more
robustamongindividualsmigratingfromacountrywhere
the population is predominantly black to a country where
the population is predominantly white; similarly, migrat-
ing to areas with a lower density of people with a similar
ethnic background has been found to be associated with a
higher liability for psychotic illness (Kirkbride et al.,
2007; Veling et al., 2008). What factor/s then might
mediate the link between migration and schizophrenia?
Social adversity associated with being a migrant (social
isolation, discrimination and “racism”, experience of
“social defeat”, etc.) has been cited as the major factor
(Boydell et al., 2001; Cooper et al., 2008), although
“biological” explanationssuchasvitaminD insufficiency
and epigenetic mechanisms have also been suggested
(Dealberto,2007).Theassociationbetweenmigrationand
increased risk of developing schizophrenia provides
compelling evidence supporting a role for social factors
in its etiology (Cantor-Graae, 2007); the specific risk-
mediating factor (social or biological), however, remains
to be elucidated.
Estimates of the risk of developing schizophrenia over
one's lifetime range from 0.3–2.0% with an average of
approximately 0.7% (Saha et al., 2005). Although gender
differences in the clinical expression and outcome of
schizophrenia have long been recognized (Seeman,
1982), it has generally been believed that the risk of
developing schizophrenia over one's lifetime is similar
among males and females (e.g., Wyatt et al., 1988). More
recent studies have undermined this assumption and two
recent meta-analyses revealed that males have a higher
lifetime risk of developing schizophrenia with a male–
female relative risk of about 1.4 (Aleman et al., 2003;
McGrath et al., 2004). The male–female ratio in
morbid risk for schizophrenia is found to increase as
more stringent current diagnostic criteria are utilized
(Beauchamp and Gagnon, 2004). Studies from develop-
ing countries have not found such a difference and study
samples obtained prior to 1980 were much less likely
to reveal a gender difference in the risk of schizo-
phrenia (Aleman et al., 2003). This observed discrepancy
between findings of studies conducted in the past two
decades and those conducted earlier is of obvious
importance but poorly understood, although proposed
explanations include differences between the two time
periods with regard to diagnostic criteria, case-ascertain-
ment methods, or differential changes in a variety of
physicaland/orsocialenvironmentriskfactorsamongthe
two genders (Hambrecht et al., 1992; Aleman et al.,
2003).
2.1.1. Incidence of schizophrenia over time: is schizo-
phrenia a new disease?
Schizophrenia has been more consistently and
frequently described over the past two centuries
3R. Tandon et al. / Schizophrenia Research 102 (2008) 1–18

Page 4

(Bleuler, 1950; Kraepelin, 1971). Although loose
descriptions resembling schizophrenia are obtained in
texts dating back several thousand years (Jeste et al.,
1985; Ellard, 1987), easily recognizable descriptions of
schizophrenia are much less common than those of other
psychiatric or neurological disorders (Evans et al.,
2003). This has led some to suggest that schizophrenia is
a disease that has afflicted humans only over the past
two centuries and that some factor related to indus-
trialization, urbanization, or increasing population
density may have contributed to the emergence of this
disease (Torrey, 1980; Hare, 1988). The broadly even
distribution of schizophrenia across the world in the
context of its strong genetic underpinnings (Crow,
1995), however, argues against the proposition that it is
a recent disease and most experts believe that schizo-
phrenia, like many other diseases, had been present for a
long time before its first lucid description in the early
19th century. Resolution of this controversy is unlikely
until we have a better understanding of the etio-
pathophysiology of the illness.
What is less controversial is the fact that descriptions
of schizophrenia have been fairly consistent over the
past two centuries and that its occurrence has been
relatively stable over this period even though specific
diagnostic criteria have changed. Its presentation has,
however, evolved over the past century with a modest
improvement in the overall prognosis (Bleuler, 1972;
Hegarty et al., 1994) and reduced occurrence of more
severe forms of the disease such as hebephrenia and
catatonia (Morrison, 1974; Stompe et al., 2002). Some
studies suggest a decline in the incidence of schizo-
phrenia (Eagles et al., 1988; Woogh, 2001) whereas
others suggest an increase (Tsuchiya and Munk-
Jorgensen, 2002; Bray et al., 2006); changing diagnostic
criteria and case detection methods make such compar-
isons difficult (Stromgren, 1987; Kendell et al., 1993).
2.2. Prevalence
Prevalence is a measure of the proportion of
individuals in a defined population who either are
manifesting a given disease at a particular time (point or
period prevalence) or have manifested the disease at any
time during their life (lifetime prevalence). Point
prevalence thus refers to the proportion of the popula-
tion with the disease on a given day, period prevalence
to the proportion with the disease during a particular
time-frame (generally 6 months or one year), and
lifetime prevalence to the proportion that ever had the
disease at any time during their life regardless of
whether they currently do or do not have the disease.
Logically, estimates of period prevalence should equal
or exceed those of point prevalence and estimates of
lifetime prevalence should equal or exceed those of
period prevalence. The prevalence of any condition in a
community is influenced by several factors including
rates of new case development (incidence), duration of
the condition, and differential mortality or migration
patterns associated with the condition (Fig. 1). Although
variable degrees of “recovery” do occur (Harding et al.,
1987), complete cures are uncommon and the average
duration that an affected person lives with schizophrenia
is approximately 30 years. Based on a median incidence
rate of 15.2/100,000/year, one would roughly predict a
median point prevalence of about 456/100,000 or 4.56/
1000 population.
Althoughfindingsfromthe approximately200studies
are found tovaryseveral-fold,the average isveryclose to
this predicted estimate. Saha et al. (2005) conducted a
systematic review of 188 studies from 46 countries and
derived a range of different prevalence estimates in
schizophrenia. Based on a meta-analysis of 21 studies,
they obtained a median point prevalence estimate of 4.6
per 1000 persons with 80% confidence interval estimates
(10th–90thdecile)rangingfrom1.9–10/1000.Basedona
meta-analysis of 34 studies, they obtained a median
period (upto 1 year) prevalence estimate of 3.3 per 1000
persons with 80% confidence interval estimates (10th–
90thdecile)rangingfrom1.3–8.2/1000.Basedonameta-
analysis of 24 studies, they obtained a median lifetime
prevalence estimate of 4.0 per 1000 persons with 80%
confidence interval estimates (10th–90th decile) ranging
from 1.6–12.1/1000. Sixty-seven distinct studies pro-
vided data towards these three estimates: two studies for
all three meta-analyses,8for twoofthe estimates,andthe
remainder for just one of these analyses.
Although these estimates of prevalence were some-
what lower than previously believed (e.g., Kessler et al.,
1994) and more recently reported (e.g., Perala et al.,
2007), this systematic review confirmed our notions of
worldwide prevalence with pockets of low and high
prevalence. Similar to the observed higher incidence of
schizophrenia among migrants, this comprehensive
review also found the prevalence of schizophrenia to
be higher among this group. In contrast to observed
differences in incidence across gender and urban–rural
settings, however, no such differences in prevalence
were observed. Prevalence was found to be similar
among males and females as also among urban and rural
dwellers. In comparison to the similar incidence across
less developed and more developed countries, on the
other hand, a significantly higher prevalence of schizo-
phrenia was observed in more developed versus less
4R. Tandon et al. / Schizophrenia Research 102 (2008) 1–18

Page 5

developed countries. A higher prevalence of schizo-
phrenia among lower versus higher socio-economic
classes within communities has also been consistently
documented over the past century.
The combinations of incidence-prevalence findings
observed in schizophrenia are not easily explained.
Variables linked to both higher incidence and prevalence
are likely relevant to the etiology and possibly to
persistence of the disease. Factors associated with
higher incidence but equal prevalence of schizophrenia
may be related in complex ways to both etiology and
outcome differences (e.g., hypothetically, a greater risk
of developing the illness along with greater illness-
related mortality in males versus females might explain
the higher incidence but equal prevalence of schizo-
phrenia among males versus females; alternatively, this
set of gender findings could hypothetically also be
explained by a greater risk of development but higher
rate of “cure” among males versus females, Fig. 1).
Variables associated with equal incidence but higher
prevalence are probably unrelated to etiology but
relevant to outcome. Although the basis of these
incidence-lifetime risk-prevalence findings is not well
understood, they do provide a rich substrate for
developing and testing hypotheses about what causes
schizophrenia (McGrath, 2007): we need to better
understand what these patterns of distribution of
schizophrenia are telling us about the specific genetic
and environmental risk factors related to its causation
and the neurobiological mechanisms that might mediate
these effects?
3. The genetic basis for schizophrenia
It is well known that schizophrenia aggregates in
families. Although over two-thirds of the cases occur
sporadically, having an affected family member sub-
stantially increases the risk of developing schizophrenia.
This risk increases as the degree of genetic affinity with
the affected family member increases (Kendler et al.,
1993). Although a genetic basis for schizophrenia has
long been suggested (Kallman, 1946), family dynamic
and interactional explanations were commonly invoked
to explain this familiality until the 1960s (Bateson et al.,
1956; Lidz et al., 1965). To differentiate between these
explanations, a series of seminal studies (Heston, 1966;
Kety et al., 1968) examined the risk of schizophrenia in
the adopted-away offspring of parents with schizophre-
nia raised by parents without the illness and adopted-
away offspring of parents without schizophrenia raised
by parents with the illness; they found that the risk of
schizophrenia was related to the presence of the illness
in biological parents but not in the adoptive parents.
In keeping with the genetic basis for schizophrenia
implied by this finding, twin studies have consistently
found more than a three-fold greater concordance for the
Fig. 1. Relation between incidence and prevalence in a given population.
5 R. Tandon et al. / Schizophrenia Research 102 (2008) 1–18