Research Report

20515545, 2018, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/wps.20490 by INASP/HINARI - INDONESIA, Wiley Online Library on [16/10/2022].
See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RESEARCH REPORT
What causes psychosis? An umbrella review of risk and

protective factors
Joaquim Radua1-3, Valentina Ramella-Cravaro1,4, John P.A. Ioannidis5-8, Abraham Reichenberg9-12, Nacharin Phiphopthatsanee1, Taha
Amir1,
Hyi Yenn Thoo1, Dominic Oliver1, Cathy Davies1, Craig Morgan9,13, Philip McGuire9,13, Robin M. Murray9,13, Paolo Fusar-Poli1,13,14
1
Early Psychosis: Interventions & Clinical-detection (EPIC) Lab, Department of Psychosis Studies, Institute of Psychiatry, Psychology & Neuroscience, King’s
College London, London, UK; 2FIDMAG Germanes Hospitalaries, CIBERSAM, Sant Boi de Llobregat, Spain; 3Department of Clinical Neuroscience, Centre for
Psychiatry Research, Karolinska Institutet, Stockholm, Sweden; 4Department of Neurosciences, Psychology, Drug Research and Child Health, University of
Florence, Florence, Italy; 5Department of Medicine, Stanford Prevention Research Center, Stanford, CA, USA; 6Department of Health Research and Policy,
Stanford University School of Medicine, Stanford, CA, USA;
7 8
Meta-Research Innovation Center at Stanford, Stanford University, Stanford, CA, USA; Department of Statistics, Stanford University School of Humanities
and Sciences, Stanford, CA, USA; 9Institute of Psychiatry, Psychology & Neuroscience, King’s College London, London, UK; 10Department of Psychiatry,
Icahn School of Medicine at Mount Sinai, New York, NY, USA; 11Department of Preventive Medicine, Icahn School of Medicine at Mount Sinai, New York,
NY, USA; 12Frieman Brain Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA; 13National Institute for Health Research (NIHR) Maudsley
Biomedical Research Center, London, UK; 14OASIS Service, South London and Maudsley NHS Foundation Trust, London, UK
Psychosis is a heterogeneous psychiatric condition for which a multitude of risk and protective factors have been suggested. This umbrella review
aimed to classify the strength of evidence for the associations between each factor and psychotic disorders whilst controlling for several biases.
The Web of Knowledge database was searched to identify systematic reviews and meta-analyses of observational studies which examined
associations between socio-demographic, parental, perinatal, later factors or antecedents and psychotic disorders, and which included a
comparison group of healthy controls, published from 1965 to January 31, 2017. The literature search and data extraction followed PRISMA and
MOOSE guidelines. The association between each factor and ICD or DSM diagnoses of non-organic psychotic disorders was graded into
convincing, highly suggestive, suggestive, weak, or non-significant according to a standardized classification based on: number of psychotic
cases, random-effects p value, largest study 95% confidence interval, heterogeneity between studies, 95% prediction interval, small study effect,
and excess significance bias. In order to assess evidence for temporality of association, we also conducted sensitivity analyses restricted to data
from prospective studies. Fifty-five meta-analyses or systematic reviews were included in the umbrella review, corresponding to 683 individual
studies and 170 putative risk or protective factors for psychotic disorders. Only the ultra-high-risk state for psychosis (odds ratio, OR59.32, 95%
CI: 4.91-17.72) and Black-Caribbean ethnicity in England (OR54.87, 95% CI: 3.96-6.00) showed convincing evidence of association. Six factors
were highly suggestive (ethnic minority in low ethnic density area, second generation immigrants, trait anhedonia, premorbid IQ, minor physical
anomalies, and olfactory identification ability), and nine were suggestive (urbanicity, ethnic minority in high ethnic density area, first generation
immigrants, North-African immigrants in Europe, winter/spring season of birth in Northern hemisphere, childhood social withdrawal, childhood
trauma, Toxoplasma gondii IgG, and non-right handedness). When only prospective studies were considered, the evidence was convincing for
ultra-high-risk state and suggestive for urbanicity only. In summary, this umbrella review found several factors to be associated with psychotic
disorders with different levels of evidence. These risk or protective factors represent a starting point for further etiopathological research and for
the improvement of the prediction of psychosis.
Key words: Schizophrenia, psychosis, risk, environment, socio-demographic factors, parental factors, perinatal factors, antecedents, ultra-
high-risk state for psychosis, Black-Caribbean ethnicity, urbanicity risk factors that have been associated with psychotic disorders
– such as family history of mental illness – include both a
(World Psychiatry 2018;17:49–66)
genetic and an environmental component, and hence a
distinction between genetic and environmental risk factors may
be spurious.
Psychotic disorders like schizophrenia are among the With this in mind, in this study we adopted a pragmatic
world’s leading causes of disability1. They have a mean approach and used the term “non-purely genetic factors” to
incidence of 31.7 per 100,000 person-years in England2 and a define socio-demographic, parental, perinatal, later factors and
12-month prevalence of 1.1% among the US population 3. antecedents7-9 that may increase (risk factors) or decrease
Despite many decades of research, the etiology of these (protective factors) the likelihood of developing psychotic
disorders remains undetermined4. disorders. The clinical importance of investigating these factors
The model that has received most empirical support is threefold. First, they could potentially be used to advance the
suggests that the etiology of psychotic disorders, schizophrenia prediction of psychosis in populations at risk of developing the
for example, involves direct genetic and environmental risk disorder10,11. Second, some, albeit not all, of these factors may
factors along with their interaction 5,6. In reality, some of the be potentially modifiable by preventive interventions 4. Third,
World Psychiatry 17:1 - February 2018

20515545, 2018, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/wps.20490 by INASP/HINARI - INDONESIA, Wiley Online Library on [16/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
they could inform outreach campaigns targeting the general METHODS
public to raise awareness of risk factors for psychosis and to
promote mental health. The protocol of the study was registered on PROSPERO
Numerous studies investigating the association between 2016: CRD42016054101.
potential risk or protective factors and psychotic disorders have
been published. The body of literature in this area is
Search strategy and selection criteria
substantial, presumably due to the severe societal burden that is
associated with these disorders and thus the urgent need to An umbrella review (i.e., a systematic collection and
understand their causes. However, to date, for all of those assessment of multiple systematic reviews and meta-analyses
factors, there is no conclusive evidence with respect to both the published on a specific research topic) 35 was conducted. We
association itself and its direction (i.e., risk or protective), searched the Web of Knowledge database (incorporating Web
because published findings have often been conflicting. of Science and MEDLINE) to identify systematic reviews or
Furthermore, some of these results have been found to be meta-analyses of observational studies that examined the
affected by several types of biases12,13. These are particularly association between a number of factors and psychotic
relevant to this area of research because experimental support disorders, published from 1965 to January 31, 2017. The
for etiology, in the sense of randomized allocation to the search strategy used the keywords (“systematic review” OR
above-mentioned exposures13, is naturally lacking, and most “meta-analysis”) and (“psychosis” OR “schizophrenia”). We
evidence is based on observational studies. Finally, previously then conducted a manual search of the reference lists of the
published literature did not generate clear hierarchies of retrieved articles.
evidence across those factors, rendering the overall Articles were initially screened on the basis of title and
interpretation of the findings particularly complex. In fact, until abstract reading. The full texts of potentially eligible articles
recently there were no stringent evaluation criteria by which to were then independently scrutinized by two investigators (PFP,
hierarchically stratify the robustness of the evidence whilst at VRC), with no language restrictions. We selected systematic
the same time controlling for the presence of biases. reviews or meta-analyses of individual observational studies
Umbrella reviews can overcome these problems by (case-control, cohort, cross-sectional and ecological studies)
assessing the level of the evidence provided by systematic that examined the association between socio-demographic,
reviews and meta-analyses14 for each risk or protective factor, parental, perinatal, later factors or antecedents and any non-
through strict criteria that probe a standard list of potential organic psychotic disorder as defined by any edition of the ICD
biases. These criteria have been extensively validated in or the DSM, including a comparison group of non-psychotic
various areas of medicine, such as neurology, oncology, healthy controls, and reporting enough data to perform the
nutrition medicine, internal medicine, psychiatry, paediatrics, analyses.
dermatology and neurosurgery15-33. In the current study, we When data were incomplete, the corresponding author was
applied the umbrella review approach to the published contacted and invited to send additional information. When
evidence on risk or protective factors for psychotic disorders. two articles presented overlapping datasets on the same factor,
Our umbrella review advances knowledge in the field of only the article with the largest dataset was retained for the
psychosis etiology, providing the first state-of-the-art main analysis. However, if the overlap was minimal, the
classification based on the robustness of associations between articles were used conjointly, counting overlapping individual
putative risk or protective factors and psychotic disorders, studies once only36-43. Moreover, we also excluded articles that
controlling at the same time for several biases. The use of did not report quantitative data, and articles with an outcome
classification criteria for levels of evidence can help overcome other than the onset of an established psychotic disorder, such
some of the ambiguity experienced by clinicians and as those related to relapse, remission or treatment response.
researchers when confronted with conflicting meta-analyses 34 Articles that investigated pure genetic markers of psychotic
on complex topics and trying to base their decisions on them. disorders were excluded, because they have been examined
Furthermore, our analysis will hopefully promote further extensively elsewhere44,45. Articles that investigated the
etiological clinical research in psychosis, support the association between biomarkers and psychotic disorders were
refinement of risk prediction in at-risk populations, and inform not included, because this would have required specific
future preventive strategies. methodological approaches and separate analyses. However,
some putative biomarkers have been defined as antecedents
(e.g., premorbid IQ38,39, minor physical anomalies46, non-right
handedness47, dermatoglyphic abnormalities48 and neurological
soft signs49) or perinatal factors (vitamin D 50), and the relevant
articles were therefore included.

The same inclusion/exclusion criteria were checked for each the sample size and therefore biasing the hierarchical
individual study comprised in every eligible meta-analysis or classification of the evidence. The exception was for analyses
systematic review. The Preferred Reporting Items for not based on individual-level data, for which we specifically
Systematic Reviews and Meta-analyses (PRISMA) created new risk factors55,56, as detailed in the statistical
recommendations51 and the Meta-analysis of Observational analysis section.
Studies in Epidemiology (MOOSE) guidelines52 were followed. For descriptive purposes, risk/protective factors for
psychotic disorders were clustered as previously suggested:
socio-demographic and parental factors, perinatal factors, later
Data extraction
factors (i.e., factors intervening in the post-perinatal period)
Data extraction was performed independently by at least two and antecedents7-9. In line with previous definitions 7,8,
investigators. Any existing discrepancies were resolved in antecedents were conceptualized as premorbid deviations in
consensus meetings with two of the authors (VRC, PFP). functioning and developmental milestones that could indicate
Factors were extracted as defined in the corresponding meta- an early expression of the disorder or active risk-modifying
analysis or systematic review. We did not combine similar mechanisms and processes involved in psychosis onset. Risk
factors if they were considered and analyzed separately by factors, instead, would indicate a passive exposure to
meta-analyses/ systematic reviews53. Similarly, we did not split environmental agents that could play a role in the development
factors into subgroups if they were considered as a whole 54. of psychosis. One could argue that this distinction remains
When a metaanalysis or systematic review reported both arbitrary, since the exact timing and mechanisms involved in
pooled results and results divided according to subgroups, the etiology of psychotic disorders remain to be elucidated, but
pooled results were preferred, since they had a larger sample this issue is beyond the aim of this review.
size. Several variables were recorded: the type of factor studied,
Such a conservative approach was adopted to minimize the the first author of the paper, the year of publication, the type of
chance of introducing risk or protective factors that had not psychotic diagnosis, and the measure of association between
been defined by the corresponding articles, and that may have the factor and psychotic disorders (preferably unadjusted), with
been too heterogeneous to allow meaningful interpretation. the corresponding 95% confidence interval (CI) and the sample
This approach also minimized the risk of artificially inflating size (where available). If studies contained several types of
Figure 1 PRISMA flow chart control groups, data from healthy controls were used. When

data were reported only in graphic form, they were digitally sample, whilst allowing estimation of heterogeneity across the
measured and extracted using WebPlotDigitizer 57. The exposed samples62.
methodological quality of included studies was assessed using Some factors required special adjustments, such as: a) the
the validated AMSTAR (A Measurement Tool to Assess transformation of measures other than OR into OR in factors
Systematic Reviews) instrument58-60. where effect size was reported using different types of
measures (to have a single outcome measure), and b) the
combination of effect sizes for left and right nostrils in
Statistical analysis olfaction studies63, conservatively assuming a weak to
moderate correlation (r50.3)64. Ultimately, we used the
This umbrella review is composed of a number of
“metainc” (IRR), “metabin” (RR and OR), or “metacont”
metaanalyses of the included articles conducted separately with
(Hedges’ g) functions in the R “meta” package 65 to calculate
a series of scripts in R 61. The effect size measures of the
the meta-analytic effect size and its p value, the CI, and the
associa-
Table 1 Characteristics of meta-analyses and systematic reviews studying the association between psychotic disorders and sociodemographic and
parental factors
Study Factors examined k Diagnosis AMSTAR index
Bosqui et al71 Ethnic minority in high ethnic density area; ethnic 5, 5 FEP, SZ, NAP, AP 9/11
minority in low ethnic density area
Bourque et al53 First generation immigrants; second generation 12, 9 SZ, NAP, AP 10/11
immigrants
Torrey et al72 Paternal age >35 years; paternal age >45 years; paternal age 8, 7, 4 SZ, NAP, AP 3/11
>55 years
Kinney et al55 Disadvantaged group; latitude 2, 29 SZ 2/11
Kirkbride et al 2
Age/gender; African ethnicity; Asian ethnicity; 9, 4, 4, 2, 3 FEP, SZ, NAP, AP 11/11
mixed ethnicity; other white ethnicity (all
examined only in England)
Kwok73 Low paternal socio-economic status 9 FEP, SZ, NAP 6/11
O’Donoghue et al74 Neighbourhood level social deprivation 3 FEP 8/11
Rasic et al 75
Parental severe mental illness 9 FEP, SZ, NAP, AP 6/11
Saha et al56 Gross national income per capita 88 SZ 9/11
Tortelli et al 76
Black-Caribbean ethnicity in England 7 FEP, SZ, NAP, AP 10/11
van der Ven et al 77

North African immigrants in Europe 5 NAP 9/11
Vassos et al78 Urbanicity 8 SZ, NAP 6/11
k – number of studies for each factor, FEP – first episode of psychosis, SZ – schizophrenia, NAP – non-affective psychosis other than schizophrenia, AP – affective
psychosis, AMSTAR – A Measurement Tool to Assess Systematic Reviews heterogeneity (summarized with the I 2 statistic and the p value
associated with the Q value). Resulting statistics were also used
to calculate the prediction interval66.
tion between each factor and psychotic disorders were: A few specific adjustments were also adopted for age and
incidence rate ratio (IRR), odds ratio (OR), risk ratio (RR), and gender, where IRRs were available for schizophrenia and
standardized mean difference (Hedges’ g) for continuous affective psychosis separately, and stratified by 5 or 10-year
measures. Primarily, the effect size measure and its CI were age ranges and gender2,67. We combined schizophrenia and
used. affective psychoses and then meta-analyzed the IRR of each
Since authors usually round off the measures, the first step 10-year age range (vs. other ages), and the IRR of males (vs.
was to “unround” them by estimating a more exact measure females, globally and within each 10-year age range). Since
and CI, in which the (logarithm of the) lower and upper bounds age and gender were considered as basic factors and excluded
were symmetrical around the (logarithm of the) measure. by previous reviews on psychosis8,9 and by umbrella reviews on
Subsequently, the variance was calculated from the standard other neuropsychiatric conditions23,25,27, these analyses were
formula for the CI. If two or more studies shared the non- considered exploratory.
exposed sample, the size of this sample was divided equally Alternative analyses were also conducted for latitude 55 and
between these studies. This approach minimized the gross national income per capita (GNI) 56, when the prevalence
dependence produced by the sharing of the non-exposed rates were provided in a series of locations. Specifically, the

Table 2 Characteristics of meta-analyses and systematic reviews studying the association between psychotic disorders and perinatal factors
AMSTAR
Study Factors examined k Diagnosis index
Cai et al40 Gestational influenza 6 SZ, NAP, AP 10/11
Cannon et al 37
Anaemia in pregnancy; antepartum haemorrhage; asphyxia; 2, 4, 3, 2, 2, 4, 3, 3, 5, 2, 3, 2, 3, SZ 6/11
baby detained in hospital; birth weight <2000 g; birth weight 6, 4, 3, 2, 2, 5, 2, 5, 2, 3, 5, 2, 1, 2, 2, 3
<2500 g; birth weight <2500 g1prematurity; breech delivery;
caesarean section; cephalopelvic disproportion; congenital
malformations; diabetes in pregnancy; emergency caesarean
section; forceps/vacuum delivery; gestational age <37
weeks; gestational age >42 weeks; induction of labour; low
Apgar score; nonvertex presentation; placental abruption;
preeclampsia; Rhesus incompatibility; small birth length;
being small for gestational age; small head circumference;
smoking in pregnancy; threatened premature delivery;
urinary infection in pregnancy; uterine atony
Christesen et al50 Neonatal vitamin D (<19.7 vs. 40.5-50.9 nmol/L); neonatal 1, 1, 1, 1 SZ 6/11
vitamin D (19.7-30.9 vs. 40.5-50.9 nmol/L); neonatal vitamin
D (30.9-40.4 vs. 40.5-50.9 nmol/L); neonatal vitamin D (>50.9
vs. 40.5-50.9 nmol/L)
Davies et al79 Winter/spring season of birth in Northern hemisphere 7 SZ 6/11
Geddes & Lawrie81 Obstetric complications 10 SZ 6/11
Geddes et al 36
Antepartum haemorrhage; birth weight <2500 g; caesarean 9, 9, 9, 7, 9, 9, 3, 8, 4, 9, 9, 7, 9, 9 SZ 4/11
section; congenital malformations; cord complications; forceps
delivery; gestational age <37 weeks; incubator or resuscitation;
labour >24 hours; non-vertex presentation; preeclampsia;
Rhesus incompatibility; rubella or syphilis; twin birth
McGrath & Welham80 Winter/spring season of birth in Southern hemisphere 7 SZ 9/11
Selten et al 106
Maternal stress during pregnancy 4 SZ 5/11
Selten & Termoshuizen41 Gestational influenza 7 SZ, AP 7/11
Van Lieshout et al 82
Pre-pregnancy and pregnancy maternal obesity 4 SZ, NAP 10/11
k – number of studies for each factor, FEP – first episode of psychosis, SZ – schizophrenia, NAP – non-affective psychosis other than schizophrenia, AP – affective
incidence in each location was (logistic) regressed by the psychosis, AMSTAR – A Measurement Tool to Assess Systematic Reviews
latitude or GNI, obtaining the OR of 108 increase in latitude or
10,000 USD increase in GNI. These results were also
considered exploratory because they are based on ecological simulated study using binomial or Poisson random cases was
analyses rather than individual-level data, and were considered “statistically significant”; the simulated studies had
traditionally excluded from previous umbrella reviews of risk the same mean incidence and person-times or sample sizes as
factors23,25,27. study A (using the full sample sizes in case of sharing a
Complementary analyses included: a) an Egger test to assess sample), and the same effect size as the largest study in the
small-study effects that lead to potential reporting or metaanalysis.
publication bias68; b) a test of excess significance bias 69 as Small-study effects and excess significance bias were
described below, and c) an OR equivalent. The test of excess claimed at one-sided p values <0.05, as in previous studies27. In
significance bias consisted of a binomial test to compare the order to easily compare meta-analyses using different outcome
observed vs. the expected number of studies yielding measures, OR equivalents were provided for the above
statistically significant results. This expected number was measures. Given the low incidence of psychotic disorders, RR
calculated as the sum of the statistical power of the studies, was assumed to be equivalent to OR, after checking that the
which was estimated using the standard t-test formulas for difference between an OR and a RR of the same data was
Hedges’ g, and random simulations for OR, RR and IRR. negligible. IRR was assumed to be equivalent to RR, and
Specifically, the statistical power of study A was estimated as Hedges’ g was converted to OR using a standard formula70.
the proportion of times in which a

IRR, OR and RR greater than 1 or Hedges’ g greater than 0 association, thus dealing with the problem of reverse causation
indicated that the factor was associated with an increased that may affect, for example, case-control studies 23,25,27.
likelihood of psychotic disorders. IRR, OR and RR lower than
1 or Hedges’ g lower than 0 indicated that the factor was
associated with a reduced likelihood of psychotic disorders, i.e.
it was protective. RESULTS
The levels of evidence of the associations between putative
risk (or protective) factors and psychotic disorders were Database
classified in accordance with previous umbrella reviews 23,25,27:
Overall, 4,023 records were searched, 302 were screened
convincing (class I) when number of cases >1000, p<1026, I2
and 55 articles were eligible 2,36-43,46-50,53-56,63,71-106 (see Figure 1).
<50%, 95% prediction interval excluding the null, no small-
The eligible articles were published between 1995 (when
study effects, and no excess significance bias; highly
metaanalyses in this field first became available) 107 and 2017.
suggestive (class II) when number of cases >1000, p<1026, All of the studies utilized a healthy control group except one,
largest study with a statistically significant effect, and class I investigating the ultra-high-risk state98. This latter study used as
criteria not met; suggestive (class III) when number of cases controls help-seeking individuals undergoing an ultra-high-risk
>1000, p<1023, and class I-II criteria not met; weak (class IV) assessment but not meeting the relevant criteria. The mental
when p<0.05 and class I-III criteria not met; non-significant health status of this control group was not well defined.
when p>0.05. Overall, the 55 eligible meta-analyses or systematic reviews,
Finally, a sensitivity analysis was conducted for the factors including 683 individual studies, reported on 170 putative risk/
classified as class I-III by using only prospective studies (as protective factors of psychotic disorders (Tables 1-4). For
defined in each meta-analysis/systematic review or, when this paternal socio-economic status73, neighbourhood-level social
was not provided, as defined by each individual study). deprivation74, pre-pregnancy and pregnancy maternal obesity 82,
Prospective studies allow one to address the temporality of the neonatal levels of vitamin D50, polluting agents (benzene,
Table 3 Characteristics of meta-analyses and systematic reviews studying the association between psychotic disorders and later factors
AMSTAR
Arias et al83 1, 8, 2, 2, 8, 3, 4, 1, 4, 5, 2, 1, 1, 4 SZ 8/11
BK virus; Borna disease virus; Chlamydia psittaci;

Chlamydia trachomatis; cytomegalovirus;
Epstein-Barr virus; human endogenous retrovirus; human
endogenous retrovirus type k115; human endogenous
retrovirus type W; human herpes virus 2; influenza; JC
virus; Toxocara spp; varicella zoster virus
Attademo et al84 Benzene; carbon monoxide; nitrogen dioxide; nitrogen oxides; 1, 1, 1, 1, 1, 1 SZ 2/11
tetrachloroethylene; traffic
Beards et al85 Adult life events 6 FEP, SZ, NAP, AP 8/11
Clancy et al 86
Epilepsy 1 SZ 6/11
Cunningham et al87 Bullying 1 SZ, NAP 7/11
De Sousa et al 88
Parental communication deviance 4 SZ 6/11
Gurillo et al89 Tobacco use 5 FEP, SZ, NAP, AP 11/11
Gutierrez-Fernandez et al90 Chlamydia pneumoniae; human herpes virus 1; human herpes virus 3, 11, 3 SZ, NAP 8/11
6
Khandaker et al91 Central nervous system infection during childhood 2 SZ, NAP 10/11
Linszen et al92 Hearing impairment 5 SZ 8/11
Marconi et al 93
Heavy cannabis use 2 FEP, SZ, NAP 7/11
Molloy et al94 Traumatic brain injury 8 SZ 7/11
Sutterland et al 95
Toxoplasma gondii IgG; Toxoplasma gondii IgM 40, 15 FEP, SZ 9/11
Varese et al 54
Childhood trauma 20 FEP, SZ, NAP, AP 10/11
k – number of studies for each factor, FEP – first episode of psychosis, SZ – schizophrenia, NAP – non-affective psychosis other than schizophrenia, AP – affective psychosis,
AMSTAR – A Measurement Tool to Assess Systematic Reviews, Ig – immunoglobulin

Table 4 Characteristics of meta-analyses and systematic reviews studying the association between psychotic disorders and antecedents
AMSTAR
Dickson et al96 4, 4, 3 FEP, SZ, NAP 7/11
Motor function pre-onset of psychosis; poor academic achievement pre-onset of

psychosis; poor mathematic academic achievement pre-onset of psychosis
Filatova et al97 Delay in grabbing object; delay in holding head up; delay in sitting unsupported; 3, 3, 4, 4, 5 SZ, NAP 9/11
delay in standing unsupported; delay in walking unsupported
Fusar-Poli et al98 Ultra-high-risk state for psychosis 9 FEP 9/11
Golembo-Smith et al48 ATD angle; fingertip pattern asymmetry; fluctuating asymmetry A-B ridge 5, 4, 3, 3, 13, 13 SZ 6/11
count; fluctuating asymmetry finger ridge count; total A-B ridge count; total
finger ridge count
Hirnstein & Hugdahl47 Non-right handedness 41 SZ, NAP 5/11
Khandaker et al39 Premorbid IQ 5 SZ, NAP 8/11
Kaymaz et al 99
Psychotic-like experiences 4 FEP, SZ, NAP, AP 10/11
Koning et al100 Dyskinesia in antipsychotic-na€ıve schizophrenic patients; parkinsonism in 5, 3 FEP, SZ 5/11

antipsychotic-na€ıve schizophrenic patients
Matheson et al43 Childhood social withdrawal 5 SZ, NAP 8/11
Moberg et al 63
Olfactory detection ability; olfactory identification ability; olfactory 18, 51, 8, 2, 9, 8, 7 SZ, NAP 9/11
discrimination ability; olfactory memory ability; olfactory hedonics ability
(pleasant odours); olfactory hedonics ability (unpleasant odours);
olfactory hedonics ability (unspecified odours)
Neelam et al49 Neurological soft signs 7 SZ, NAP 8/11
Ohi et al101 Cooperativeness; harm avoidance; novelty seeking; persistence; reward 7, 7, 7, 7, 7, 7, 7 SZ 4/11
dependence; self-directedness; self-transcendence
Ohi et al102 Agreeableness; conscientiousness; extraversion; neuroticism; openness 6, 7, 8, 8, 7 SZ 6/11
Potvin & Marchand103 Hypoalgesia 9 SZ 5/11
Tarbox & Pogue-Geile42 Childhood antisocial and externalizing behaviour; childhood social withdrawal and 2, 6 SZ, NAP 3/11
internalizing behaviour
Ward et al104 Extracranial size 7 SZ, NAP 3/11
Woodberry et al 38
Premorbid IQ 11 SZ 7/11
Xu et al46 Minor physical anomalies 14 SZ 4/11
Yan et al 105
Trait anhedonia 44 SZ, NAP 6/11
k – number of studies for each factor, FEP – first episode of psychosis, SZ – schizophrenia, NAP – non-affective psychosis other than schizophrenia, AP – affective psychosis,
AMSTAR – A Measurement Tool to Assess Systematic Reviews, ATD angle – dermatoglyphic feature that compares the length of the hand to the width
carbon monoxide, nitrogen dioxide, nitrogen oxides, prediction interval did not include the null. Additionally, the
tetrachloroethylene, traffic)84 and the ultra-high-risk state98, the evidence for small-study effects and excess significance bias
studies did not provide a quantitative synthesis of individual was noted for 16 (9.4%) and 17 (10.0%) factors, respectively.
findings, but reported adequate data to allow meta-analyses.
Classification of level of evidence of associations

Summary of associations between socio-demographic and parental, perinatal,
later factors or antecedents and psychotic disorders
The number of cases was greater than 1,000 for 48 factors
(28.2%). One hundred three of the 170 analyzed factors Among the 170 factors, one socio-demographic factor
(60.6%) presented a statistically significant effect (p<0.05) (BlackCaribbean ethnicity in England: OR54.87, 95% CI:
under the random-effects model, but only 39 (22.9%) reached 3.96-6.00) and
p<1026. Fifty-three factors (31.2%) presented a large one antecedent (ultra-high-risk state: OR59.32, 95% CI:
heterogeneity (I2>50%), while for 28 (16.5%) the 95% 4.9117.72) presented a convincing level of association (class I:

Table 5 Level of evidence for the association of socio-demographic and parental factors and psychotic disorders
Features used for classification of level of evidence
Random-effects p randomeffects
Factor k measure, ES (95% CI) N I2 (p) PI (95% CI) SSE/ESB LS eOR CE
Black-Caribbean ethnicity in England76 9 IRR, 4.87 (3.96-6.00) 3,446 2.8 3 10250 38% (0.12) 2.95-8.03 No/No Yes 4.87 I
Ethnic minority in low ethnic 5 IRR, 3.71 (2.47-5.58) 1,328 3.1 3 10210 70% (0.09) 0.95-14.43 Yes/No Yes 3.71 II
density area71
Second generation immigrants 53 26 IRR, 1.68 (1.42-1.92) 28,753 7.6 3 10210 77% (<0.001) 0.92-3.06 No/No Yes 1.68 II
North African immigrants in Europe 77
12 IRR, 2.22 (1.58-3.12) 2,577 4.2 3 1026
65% (0.001) 0.77-6.41 No/NA Yes 2.22 III
Urbanicity78 8 OR, 2.19 (1.55-3.09) 45,791 8.9 x 1025 99% (<0.001) 0.62-7.77 No/No Yes 2.19 III
Ethnic minority in high ethnic 5 IRR, 2.11 (1.39-3.20) 1,328 4.3 3 1024
58% (0.04) 0.57-7.81 No/No Yes 2.11 III
density area71
First generation immigrants 53 42 IRR, 2.10 (1.72-2.56) 25,063 1.9 3 10213 89% (<0.001) 0.75-5.89 No/Yes No 2.10 III
Parental severe mental illness 75

9 RR, 5.94 (2.99-11.79) 90 3.5 3 1027 0% (0.85) 2.60-13.59 No/No Yes 5.94 IV
Black African ethnicity in England 2

4 IRR, 4.72 (3.30-6.77) 452 2.3 3 10217
49% (0.12) 1.25-17.82 No/NA Yes 4.72 IV
Asian ethnicity in England2 6 IRR, 2.83 (1.59-5.05) 613 4.2 3 1024 55% (0.05) 0.53-15.00 No/Yes No 2.83 IV
Other white ethnicity in England 2

3 IRR, 2.62 (1.35-5.10) 274 0.004 87% (<0.001) 0.93-21.88 No/NA Yes 2.62 IV
Paternal age >45 years72 4 OR, 2.36 (1.35-4.11) 392 0.003 0% (0.66) 0.69-8.01 No/Yes No 2.36 IV
Disadvantaged vs. advantaged groups 55

3 RR, 2.27 (1.21-4.27) 532 0.010 69% (0.04) 0-2016.72 No/No Yes 2.27 IV
Mixed ethnicity in England2 3 IRR, 2.19 (1.08-4.44) 330 0.030 0% (0.41) 0.02-14.53 No/NA No 2.19 IV
Low paternal socio-economic status73 9 OR, 1.30 (1.02-1.65) 15,922 0.032 94% (<0.001) 0.58-2.90 No/No Yes 1.30 IV
Paternal age >35 years 72

9 OR, 1.22 (1.06-1.41) 2,181 0.007 30% (0.18) 0.89-1.67 No/Yes No 1.22 IV
Neighbourhood level social 3 OR, 1.64 (0.83-3.23) 5,560 0.156 88% (<0.001) 0-5961.52 No/No No 1.64 ns
deprivation74
Paternal age >55 years72 7 OR, 1.21 (0.82-1.78) 57 0.341 47% (0.07) 0.45-3.22 No/No No 1.21 ns
k – number of samples for each factor, ES – effect size, N – number of cases, PI – prediction interval, CI – confidence interval, SSE – small-study effect, ESB – excess significance
bias, LS – largest study with significant effect, eOR – equivalent odds ratio, CE – class of evidence, IRR – incidence rate ratio, OR – odds ratio,
>1000 cases, p<1026, no evidence of small-study effects or nonright handedness: OR51.58). There was either weak (class
excess significance bias, 95% prediction interval not including IV)
the null, and no large heterogeneity). RR – relative risk, NA – not assessable, ns – not significant
For six factors there was highly suggestive evidence for

association (class II: >1000 cases, p<1026, largest study with a
statistically significant effect, and class I criteria not met).
or no evidence of association with psychotic disorders for all
These were two socio-demographic and parental factors (ethnic
other factors (see Tables 5-8).
minority in low ethnic density area: OR53.71; and second
generation immigrants: OR51.68); none of the perinatal and
later factors; and four antecedents (minor physical anomalies: Exploratory analyses
OR55.30; trait anhedonia: OR54.41; olfactory identification
ability: OR50.19; and premorbid IQ: OR50.47). Results of the exploratory analyses on the association
There was suggestive evidence for association (class III) for between age and gender strata (total of 23 strata) showed a
nine further factors: four socio-demographic and parental main effect for male gender (males vs. females: IRR51.34,
factors (North-African immigrants in Europe: OR52.22; 95% CI: 1.05-1.71, class IV). There was also a main effect for
urbanicity: OR52.19; ethnic minority in high ethnic density 15-35 year-old age (2534 year-old vs. other: IRR51.45, 95%
area: OR52.11; and first generation immigrants: OR52.10); one CI: 1.29-1.63, class II; 20-29 year-old vs. other: IRR52.43,
perinatal factor (winter/spring season of birth in Northern 95% CI: 1.58-3.74, class IV; 15-24 year-old vs. other:
hemisphere: OR51.04); two later factors (childhood trauma: IRR51.46, 95% CI: 1.14-1.87, class IV). Age older than 35 was
OR52.87; and Toxoplasma gondii IgG: OR51.82); and two found to be a protective factor (60-69 year-old vs. other:
antecedents (childhood social withdrawal: OR52.91; and IRR50.26, 95% CI: 0.14-0.51, class IV; 55-64 year-old vs.
other: IRR50.30, 95% CI: 0.17-0.51, class IV; 50-59 year-old

vs. other: IRR50.50, 95% CI: 0.27-0.93, class IV; 40-49 year- associations, with ORs of 1.22 and 0.80, respectively.
old vs. other: IRR50.54, 95% CI: 0.35-0.83, class IV; 35-44 Although these factors include >1000 patients and have a
year-old vs. other: IRR50.80, 95% CI: 0.70-0.93, class IV). p<0.001, it was not possible to apply the classification of the
There was also weak (class IV) association between evidence.
psychotic disorders and male gender for 15-40 year-old age
(male vs. female within 20-29 year-old: IRR52.19, 95% CI:
Classification of level of evidence of associations
1.69-2.84; male vs. female within 15-24 year-old: IRR51.98,
between socio-demographic and parental, perinatal,
95% CI: 1.62-2.41; male vs. female within 30-39 year-old:
later factors or antecedents and psychotic disorders
IRR51.72, 95% CI: 1.22-
after sensitivity analysis
2.41; male vs. female within 25-34 year-old: IRR51.60, 95%
CI: 1.26-2.03). The other ten strata were all not associated with A sensitivity analysis was not possible for four of the
psychotic disorders. associations categorized as class I-III in the overall analysis
Additional exploratory analyses on latitude (per 108) 55and (winter/ spring season of birth in Northern hemisphere,
GNI per capita (per 10,000 USD) 56 found significant olfactory iden-
Table 6 Level of evidence for the association of perinatal factors and psychotic disorders
Random-effects measure, p randomeffects

Factor k ES (95% CI) N I2 (p) PI (95% CI) SSE/ESB LS eOR CE
Winter/spring season of birth in 27 OR, 1.04 (1.02-1.06) 115,010 2.1 3 1026 0% (0.99) 1.02-1.06 No/No Yes 1.04 III
Northern hemisphere79
Diabetes in pregnancy37 2 OR, 10.12 (1.84-55.72) 243 0.008 0% (0.69) NA NA/NA No 10.12 IV
Emergency caesarean section37 3 OR, 3.36 (1.48-7.63) 825 0.004 0% (0.92) 0.02-685.69 No/No No 3.36 IV
Birth weight <2000 g 37

2 OR, 2.46 (1.11-5.46) 507 0.027 0% (0.85) NA NA/NA Yes 2.46 IV
Congenital malformations36,37 10 OR, 2.31 (1.29-4.13) 1,080 0.005 0% (0.99) 1.16-4.57 No/Yes Yes 2.31 IV
Incubator or resuscitation 36
8 OR, 2.12 (1.29-3.47) 438 0.003 0% (0.85) 1.14-3.92 No/No No 2.12 IV
Neonatal vitamin D (<19.7 vs. 1 RR, 2.11 (1.28-3.49) 424 0.004 NA (1.00) NA NA/NA Yes 2.11 IV
40.5-50.9 nmol/L)50
Neonatal vitamin D (30.9-40.4 vs. 1 RR, 2.10 (1.30-3.40) 424 0.003 NA (1.00) NA NA/NA Yes 2.10 IV
40.5-50.9 nmol/L)50
Threatened premature delivery37 2 OR, 2.05 (1.02-4.10) 314 0.043 0% (0.56) NA NA/NA Yes 2.05 IV
Neonatal vitamin D (19.7-30.9 vs. 1 RR, 2.02 (1.27-3.19) 424 0.003 NA (1.00) NA NA/NA Yes 2.02 IV
40.5-50.9 nmol/L)50
Pre-pregnancy and pregnancy maternal 4 OR, 1.99 (1.26-3.14) 305 0.003 27% (0.24) 0.47-8.50 No/No No 1.99 IV
obesity82
Uterine atony37 3 OR, 1.93 (1.35-2.76) 836 3.3 3 1024 0% (0.37) 0.19-19.78 No/No Yes 1.93 IV
Obstetric complications 81
10 OR, 1.84 (1.25-2.70) 373 0.002 25% (0.21) 0.80-4.22 Yes/Yes No 1.84 IV
Neonatal vitamin D (>50.9 vs. 1 RR, 1.71 (1.04-2.80) 424 0.033 NA (1.00) NA NA/NA Yes 1.71 IV
40.5-50.9 nmol/L)50
Antepartum haemorrhage36,37 14 OR, 1.63 (1.12-2.38) 1,489 0.011 6% (0.38) 0.92-2.89 No/No No 1.63 IV
Birth weight <2500 g 36,37

13 OR, 1.57 (1.20-2.07) 1,815 0.001 0% (0.45) 1.16-2.14 No/Yes Yes 1.57 IV
Small head circumference37 2 OR, 1.41 (1.00-1.97) 762 0.048 0% (0.58) NA NA/NA No 1.41 IV
Placental abruption 37
2 OR, 4.54 (0.32-64.63) 314 0.264 72% (0.05) NA NA/NA No 4.54 ns
Rhesus incompatibility 36,37 9 OR, 1.96 (0.88-4.33) 1,097 0.098 0% (0.98) 0.75-5.11 No/NA No 1.96 ns
Asphyxia 37
3 OR, 1.95 (0.77-4.97) 1,122 0.160 76% (0.01) 0-108727 No/No Yes 1.95 ns
Forceps delivery 36 9 OR, 1.67 (0.90-3.08) 554 0.103 42% (0.08) 0.34-8.15 Yes/No Yes 1.67 ns
Rubella or syphilis 36
9 OR, 1.64 (0.47-5.71) 567 0.435 0% (0.099) 0.37-7.39 No/No No 1.64 ns
Twin birth36 9 OR, 1.53 (0.79-2.97) 558 0.208 0% (0.45) 0.69-3.40 Yes/No No 1.53 ns
Gestational age <37 weeks 36,37

7 OR, 1.35 (0.99-1.84) 1,502 0.057 0% (0.66) 0.90-2.03 Yes/No No 1.35 ns
Being small for gestational age37 5 OR, 1.34 (0.82-2.19) 1,436 0.240 58% (0.04) 0.28-6.41 No/No Yes 1.34 ns

Smoking in pregnancy37 1 OR, 1.29 (0.72-2.31) 76 0.393 NA (1.00) NA NA/NA No 1.29 ns
Birth weight <2500 g and prematurity 37

4 OR, 1.25 (0.52-3.00) 959 0.610 65% (0.03) 0.03-46.31 No/No Yes 1.25 ns
Anaemia in pregnancy37 3 OR, 1.22 (0.46-3.28) 528 0.688 56% (0.10) 0- 41770 No/No No 1.22 ns
Maternal stress during pregnancy 106

5 RR, 1.16 (0.94-1.43) 4,412 0.166 71% (0.01) 0.60-2.25 No/No No 1.16 ns
Low Apgar score37 2 OR, 1.13 (0.69-1.84) 405 0.622 0% (0.67) NA NA/NA No 1.13 ns
Preeclampsia 36,37
15 OR, 1.07 (0.78-1.46) 2,277 0.690 22% (0.20) 0.53-2.15 No/No Yes 1.07 ns
Forceps/vacuum delivery37 7 OR, 1.07 (0.81-1.42) 1,888 0.643 34% (0.16) 0.55-2.09 No/No Yes 1.07 ns
Cord complications 36
9 OR, 1.06 (0.47-2.39) 549 0.894 0% (0.54) 0.40-2.83 No/No No 1.06 ns
Small birth length37 3 OR, 1.05 (0.86-1.30) 929 0.619 0% (0.91) 0.28-4.03 No/No No 1.05 ns
Baby detained in hospital 37

3 OR, 1.04 (0.59-1.86) 976 0.883 76% (0.01) 0-903.90 No/No Yes 1.04 ns
Winter/spring season of birth in 7 OR, 1.03 (0.88-1.19) 15,023 0.738 16% (0.30) 0.77-1.37 No/NA No 1.03 ns
Southern hemisphere80
Influenza during pregnancy40,41 14 OR, 0.99 (0.91-1.08) 7,620 0.867 46% (0.03) 0.79-1.24 No/No No 0.99 ns
Table 6 Level of evidence for the association of perinatal factors and psychotic disorders (continued)

Factor k ES (95% CI) N I2 (p) PI (95% CI) SSE/ESB LS eOR CE
Non-vertex presentation 36,37 15 OR, 0.99 (0.75-1.31) 2,272 0.953 6% (0.38) 0.65-1.51 No/No No 0.99 ns
Gestational age >42 weeks37 3 OR, 0.97 (0.48-1.95) 1,193 0.933 42% (0.18) 0-1000 No/No No 0.97 ns
Caesarean section 36,37 15 OR, 0.95 (0.71-1.28) 1,920 0.734 0% (0.46) 0.68-1.32 No/No No 0.95 ns
Breech delivery 37
3 OR, 0.95 (0.49-1.84) 470 0.879 0% (0.78) 0.01-68.26 No/No No 0.95 ns
Urinary infection in pregnancy37 3 OR, 0.90 (0.44-1.84) 690 0.776 29% (0.24) 0-498.73 No/No No 0.90 ns
Induction of labor 37
3 OR, 0.82 (0.53-1.28) 528 0.387 24% (0.26) 0.02-35.30 No/No No 0.82 ns
Cephalopelvic disproportion37 2 OR, 0.60 (0.18-1.99) 243 0.407 0% (0.48) NA NA/NA No 0.60 ns
Labour >24 hours 36

4 OR, 0.84 (0.39-1.78) 266 0.643 20% (0.28) 0.09-8.11 No/No No 0.84 ns
k – number of samples for each factor, ES – effect size, N – number of cases, PI – prediction interval, CI – confidence interval, SSE – small-study effect, ESB – excess
significance bias, LS – largest study with significant effect, eOR – equivalent odds ratio, CE – class of evidence, IRR – incidence rate ratio, OR – odds ratio,
RR – relative risk, NA – not assessable, ns – not significant DISCUSSION
To our knowledge, this is the first umbrella review of risk

and protective factors for psychotic disorders that includes a
tification ability, trait anhedonia and minor physical
robust hierarchical classification of the published evidence.
anomalies), because they did not include any prospective
Overall, 55 meta-analyses or systematic reviews, with a total of
studies.
683 individual studies and 170 socio-demographic and
Within class I factors, only ultra-high-risk state maintained
parental, perinatal, later factors or antecedents of psychotic
the same level of evidence, whereas Black-Caribbean ethnicity
disorders, were included. There was convincing evidence (class
in England downgraded to a weak (class IV) level of evidence.
I) for only two factors, which were the ultra-high-risk state for
Equally, all other available class II and III factors were
psychosis and BlackCaribbean ethnicity in England. However,
downgraded either to a weak (ethnic minority in low ethnic
six other factors were characterized by highly suggestive
density area, North-African immigrants in Europe, childhood
evidence (class II), and another nine by suggestive evidence
trauma, ethnic minority in high ethnic density area, childhood
(class III). Sensitivity analyses that limited data to prospective
social withdrawal, first and second generation immigrants,
studies indicated that ultra-high-risk state and urbanicity
Toxoplasma gondii IgG, and premorbid IQ) or a non-
showed the largest evidence of association (class I and class
significant (non-right handedness) level of evidence, except
III, respectively) with psychotic disorders.
urbanicity, that remained a class III risk factor (Table 9).
Overall, our umbrella review indicates that, although a large
number of risk factors for psychotic disorders have been
evaluated in multiple studies, reviews and meta-analyses, the
number of those that have suggestive or stronger support is far

more limited. This is consistent with previous findings about characterized by the accumulation of a number of risk factors 117
the etiology of other neuropsychiatric conditions where which enrich the risk in an uncontrolled manner118-122.
umbrella reviews have been performed, such as dementia, Ethnic minority status and urbanicity may better represent
Parkinson’s disease, amyotrophic lateral sclerosis, multiple true risk factors, contributing to the development of psychotic
sclerosis and bipolar disorder16,23,25-27. disorders through increased socio-environmental adversities 123.
Although the past two decades have clearly shown that the In fact, the effect of both factors on the risk of developing
ultra-high-risk state is substantially associated with an psychotic disorders may be explained (mediated) by
increased risk of psychosis11,108,109, this result should be environmental exposures at an individual level, such as
interpreted with caution. Firstly, this state is the closest substance use, social isolation, social defeat, social
antecedent of psychosis by definition, with onset of the fragmentation, and discrimination124. Interestingly, many of
disorder occurring from within a few months of ultra-high-risk these exposures appear to share a common factor of social
diagnosis110. Indeed, some ultra-highrisk individuals already stress and defeat125,126, and have been – mostly indirectly –
present with severe symptoms, including short-lived psychotic associated with various neurobiological sequelae of potential
episodes111,112, affective symptoms113 and impaired relevance to psychotic disorders127, such as alterations in the
functioning114. Secondly, the ultra-high-risk state is intrinsically hypothalamic-pituitary-adrenal axis128,129, inflammation130,
heterogeneous10,115, including different subgroups115 and varying altered brain functioning131,132, reduced brain volumes133, and
diagnostic operationalizations116. Furthermore, from an neurochemical dysfunctions126,134,135. However, studies to
epidemiological perspective, it is a spurious condition, directly assess the correlations between these factors
Table 7 Level of evidence for the association of later factors and psychotic disorders
Random-effects
Factor k measure, ES (95% CI) N p random-effects I2 (p) PI (95% CI) SSE/ESB LS eOR CE
Childhood trauma 54 20 OR, 2.87 (2.07-3.98) 2,363 2.5 3 10214 77% (<0.001) 0.75-11.01 No/Yes No 2.87 III
Toxoplasma gondii IgG 95
42 OR, 1.82 (1.51-2.18) 8,796 2.1 3 10210 78% (<0.001) 0.68-4.88 Yes/Yes No 1.82 III
Toxocara spp 83
1 OR, 41.61 (9.71-178.32) 98 5.1 3 10 27
NA (1.00) NA NA/NA Yes 41.61 IV
Chlamydia psittaci83 2 OR, 29.05 (8.91-94.69) 82 2.2 3 1028 0% (0.71) NA NA/NA Yes 29.05 IV
Human endogenous retrovirus 5 OR, 19.78 (6.50-60.22) 256 1.4 3 10 27

33% (0.20) 1.05-372.34 No/No Yes 19.78 IV
type W83
Parental communication deviance88 4 g, 1.35 (0.97-1.73) 74 2.3 3 10212 0% (0.41) 0.51-2.19 No/No No 11.55 IV
Chlamydia pneumoniae90 3 OR, 6.02 (2.86-12.66) 116 2.1 3 1026 0% (0.57) 0.05-745.30 No/No Yes 6.02 IV
Traffic 84
1 RR, 5.55 (1.63-18.87) 29 0.006 NA (<0.001) NA NA/NA Yes 5.55 IV
Adult life events85 6 OR, 5.34 (3.84-7.43) 317 2.1 3 10223 3% (0.39) 3.22-8.87 No/No Yes 5.34 IV
Heavy cannabis use 93

2 OR, 5.17 (3.64-7.36) 748 6.3 3 10220
42% (0.18) NA NA/NA Yes 5.17 IV
Benzene84 1 RR, 3.20 (1.01-10.12) 29 0.048 NA (1.00) NA NA/NA Yes 3.20 IV
Tobacco use 89
6 RR, 2.19 (1.36-3.53) 8,488 0.001 99% (<0.001) 0.38-12.50 No/No Yes 2.19 IV
Borna disease virus83 21 OR, 1.94 (1.30-2.91) 1,919 0.001 36% (0.05) 0.65-5.81 No/No Yes 1.94 IV
Traumatic brain injury 94

8 OR, 1.49 (1.09-2.05) 9,653 0.013 78% (<0.001) 0.57-3.89 Yes/No No 1.49 IV
Human herpes virus 283 5 OR, 1.44 (1.14-1.81) 901 0.002 0% (0.97) 0.99-2.09 No/No Yes 1.44 IV
Chlamydia trachomatis 83
2 OR, 4.39 (0.03-587.92) 82 0.554 85% (0.01) NA NA/NA No 4.39 ns
Human endogenous retrovirus83 4 OR, 3.64 (0.72-18.37) 128 0.117 36% (0.19) 0.01-1019 No/No Yes 3.64 ns
Tetrachloroethylene 84
1 RR, 3.41 (0.48-24.24) 4 0.219 NA (1.00) NA NA/NA No 3.41 ns
Carbon monoxide84 1 RR, 3.07 (0.96-9.82) 29 0.059 NA (1.00) NA NA/NA No 3.07 ns
Epilepsy 86
1 OR, 3.06 (0.31-29.95) 4 0.337 NA (1.00) NA NA/NA No 3.06 ns
Nitrogen oxides84 1 RR, 2.02 (0.74-5.53) 29 0.171 NA (1.00) NA NA/NA No 2.02 ns
Central nervous system infection during 2 RR, 1.99 (0.31-12.78) 2,369 0.466 80% (0.02) NA NA/NA No 1.99 ns
childhood91
Epstein-Barr virus83 3 OR, 1.98 (0.23-16.85) 55 0.532 0% (0.81) 0-2121495 No/No No 1.98 ns
Nitrogen dioxide84 1 RR, 1.91 (0.70-5.19) 29 0.205 NA (1.00) NA NA/NA No 1.91 ns

Hearing impairment92 5 OR, 1.64 (0.85-3.15) 597 0.141 76% (0.002) 0.18-15.17 No/No Yes 1.64 ns
Toxoplasma gondii IgM 95

15 OR, 1.24 (0.97-1.59) 2,867 0.083 2% (0.43) 0.91-1.70 No/No No 1.24 ns
Human herpes virus 190 11 OR, 1.24 (0.98-1.58) 1,117 0.074 5% (0.39) 0.87-1.78 No/No No 1.24 ns
Cytomegalovirus 83
8 OR, 1.20 (0.65-2.20) 171 0.558 0% (1.00) 0.56-2.56 No/No No 1.20 ns
Varicella zoster virus83 4 OR, 1.17 (0.16-8.58) 69 0.878 0% (0.99) 0.01-92.93 No/No No 1.17 ns
BK virus 83
1 OR, 1.05 (0.02-55.41) 20 0.979 NA (1.00) NA NA/NA No 1.05 ns
JC virus83 1 OR, 1.05 (0.02-55.41) 20 0.979 NA (1.00) NA NA/NA No 1.05 ns
Human endogenous retrovirus type 1 OR, 0.89 (0.43-1.84) 178 0.753 NA (1.00) NA NA/NA No 0.89 ns
k11583
Influenza83 2 OR, 0.87 (0.05-15.48) 33 0.925 0% (0.92) NA NA/NA No 0.87 ns
Human T-lymphotropic virus 183 2 OR, 0.57 (0.20-1.62) 209 0.294 0% (0.87) NA NA/NA No 0.57 ns
Bullying 87
1 OR, 0.38 (0.13-1.10) 30 0.075 NA (1.00) NA NA/NA No 0.38 ns
Human herpes virus 690 3 OR, 0.34 (0.05-2.42) 55 0.284 0% (0.71) 0-106440 No/No No 0.34 ns
k – number of samples for each factor, ES – effect size, N – number of cases, PI – prediction interval, CI – confidence interval, SSE – small-study effect, ESB – excess
significance bias, LS – largest study with significant effect, eOR – equivalent odds ratio, CE – class of evidence, IRR – incidence rate ratio, OR – odds ratio, RR –
relative risk, Ig – immunoglobulin, NA – not assessable, ns – not significant
Table 8 Level of evidence for the association of antecedents and psychotic disorders
Random-effects
Factor k measure, ES (95%CI) N p random-effects I2 (p) PI (95% CI) SSE/ESB LS eOR CE
Ultra-high-risk state for psychosis98 9 RR, 9.32 (4.91 to 17.72) 1,226 9.5 3 10212 0% (0.91) 4.30 to 20.24 No/No No 9.32 I
Minor physical anomalies 46
14 g, 0.92 (0.61 to 1.23) 1,212 5.8 3 1029 91% (<0.001) 20.34 to 2.18 No/Yes Yes 5.30 II
Trait anhedonia105 44 g, 0.82 (0.72 to 0.92) 1,601 9.2 3 10257 43% (0.002) 0.37 to 1.27 No/Yes Yes 4.41 II
Olfactory identification ability 63

55 g, 20.91 (–1.05 to 20.78) 1,703 4.0 3 10 241
67% (<0.001) 21.72 to 20.10 Yes/Yes Yes 0.19 II
Premorbid IQ38,39 16 g, 20.42 (20.52 to 20.33) 4,459 1.1 3 10218 73% (<0.001) 20.70 to 20.14 No/No Yes 0.47 II
Childhood social withdrawal 42,43

15 g, 0.59 (0.33 to 0.85) 1,810 6.4 3 10 26
93% (<0.001) 20.44 to 1.62 No/No Yes 2.91 III
Non-right handedness47 41 OR, 1.58 (1.35 to 1.86) 2,652 2.0 3 1028 21% (0.12) 0.99 to 2.54 No/No No 1.58 III
Neurological soft signs 49

8 g, 1.83 (1.28 to 2.38) 564 7.7 3 10 211
93% (<0.001) 20.15 to 3.81 Yes/No Yes 27.59 IV
Neuroticism102 8 g, 1.20 (0.88 to 1.52) 430 2.7 3 10213 73% (<0.001) 0.18 to.21 No/No Yes 8.76 IV
Harm avoidance 101

7 g, 0.98 (0.78 to 1.18) 384 4.5 3 10 221
48% (0.07) 0.43 to 1.53 No/No Yes 5.92 IV
Parkinsonism in antipsychotic-na€ıve 3 OR, 5.33 (1.75 to 16.23) 84 0.003 0% (0.81) 0 to 7310 No/No Yes 5.33 IV
schizophrenic patients100
Psychotic like experiences 99 4 RR, 3.84 (2.55 to 5.79) 118 1.2 3 10210 0% (0.65) 1.56 to 9.45 No/No No 3.84 IV
Dyskinesia in antipsychotic-na€ıve 5 OR, 3.59 (1.53 to 8.42) 189 0.003 0% (0.75) 0.90 to 14.32 No/No Yes 3.59 IV
schizophrenic patients100
Self-transcendence101 7 g, 0.61 (0.48 to 0.75) 384 7.8 3 10219 0% (0.67) 0.43 to 0.79 No/No Yes 3.03 IV
Antisocial and externalizing behaviour 42

3 g, 0.48 (0.22 to 0.74) 68 3.1 3 10 24
36% (0.20) 21.97 to 2.93 No/No Yes 2.39 IV
Delay in walking unsupported97 5 g, 0.48 (0.27 to 0.68) 368 4.3 3 1026 81% (<0.001) 20.27 to 1.22 Yes/NA Yes 2.37 IV
Hypoalgesia103 9 g, 0.46 (0.13 to 0.79) 204 0.006 64% (0.005) 20.57 to 1.49 No/No No 2.31 IV
Extracranial size 104

7 g, 0.27 (0.05 to 0.50) 192 0.018 15% (0.31) 20.15 to 0.70 No/No Yes 1.64 IV
Delay in standing unsupported97 4 g, 0.25 (0.12 to 0.39) 307 2.6 3 1024 48% (0.12) 20.26 to 0.76 Yes/NA No 1.58 IV
Delay in sitting unsupported 97

4 g, 0.19 (0.05 to 0.33) 386 0.006 48% (0.12) 20.33 to 0.70 Yes/NA No 1.41 IV
Delay in holding head up97 3 g, 0.13 (0.01 to 0.24) 352 0.029 0% (0.91) 20.61 to 0.86 Yes/NA No 1.26 IV
Olfactory memory ability 63

2 g, 21.62 (–2.24 to 21.01) 67 2.0 3 10 27
56% (0.13) NA NA/NA Yes 0.05 IV
Self-directedness101 7 g, 20.96 (–1.10 to 20.82) 384 7.7 3 10242 0% (0.75) 21.14 to 20.78 No/No Yes 0.17 IV
Extraversion 102
8 g, 20.90 (–1.05 to 20.75) 430 3.6 3 10 232
5% (0.38) 21.13 to 20.67 No/No Yes 0.20 IV
Olfactory discrimination ability 63 8 g, 20.88 (–1.16 to 20.60) 226 4.1 3 10210 45% (0.07) 21.61 to 20.15 No/No Yes 0.20 IV

Olfactory hedonics ability 10 g, 20.76 (20.99 to 20.54) 298 2.5 3 10211 38% (0.10) 21.34 to 20.19 No/No Yes 0.25 IV
(pleasant odours)63
Conscientiousness102 7 g, 20.68 (20.92 to 20.44) 399 2.2 3 1028 51% (0.05) 21.33 to 20.04 No/No Yes 0.29 IV
Olfactory detection ability 63

18 g, 20.63 (20.94 to 20.32) 498 5.9 3 10 25
80% (<0.001) 21.92 to 0.66 Yes/Yes No 0.32 IV
Motor function pre-onset of 4 g, 20.56 (20.73 to 20.38) 152 4.1 3 10210 0% (0.60) 20.94 to 20.17 No/No Yes 0.36 IV
psychosis96
Olfactory hedonics ability 7 g, 20.51 (20.78 to 20.24) 142 2.1 3 1024 21% (0.26) 21.06 to 0.05 No/No No 0.40 IV
(unspecified odours)63
Agreeableness102 6 g, 20.47 (20.88 to 20.07) 375 0.022 81% (<0.001) 21.82 to 0.88 No/No Yes 0.42 IV
Cooperativeness 101
7 g, 20.47 (20.60 to 20.33) 384 7.9 3 10 212
0% (0.88) 20.64 to 20.29 Yes/Yes Yes 0.43 IV
Reward dependence101 7 g, 20.43 (20.56 to 20.30) 384 2.7 3 10210 0% (0.43) 20.61 to 20.26 No/No Yes 0.46 IV
Openness 102
7 g, 20.40 (20.67 to 20.13) 399 0.003 62% (0.01) 21.18 to 0.38 No/Yes No 0.49 IV
Olfactory hedonics ability 9 g, 20.35 (20.53 to 20.17) 244 1.3 3 1024 0% (0.79) 20.57 to 20.13 No/No No 0.53 IV
(unpleasant odours)63
Persistence101 7 g, 20.24 (20.39 to 20.08) 384 0.003 22% (0.26) 20.56 to 0.09 No/No No 0.65 IV
Total A-B ridge count48 13 g, 20.15 (20.28 to 20.02) 979 0.027 46% (0.35) 20.53 to 0.23 No/No No 0.76 IV
Table 8 Level of evidence for the association of antecedents and psychotic disorders (continued)
Random-effects
Factor k measure, ES (95%CI) N p random-effects I2 (p) PI (95% CI) SSE/ESB LS eOR CE
Fluctuating asymmetry A-B ridge 4 g, 0.74 (20.65 to 2.13) 241 0.295 98% (<0.001) 26.00 to 7.49 No/Yes Yes 3.84 ns
count48
Fluctuating asymmetry finger 4 g, 0.31 (20.50 to 1.12) 233 0.448 94% (<0.001) 23.54 to 4.17 No/No Yes 1.76 ns
ridge count48
Fingertip pattern asymmetry48 5 g, 0.25 (20.08 to 0.59) 249 0.138 66% (0.02) 20.85 to 1.35 No/No Yes 1.58 ns
Poor general academic 4 g, 0.20 (20.12 to 0.51) 1,007 0.219 93% (<0.001) 21.25 to 1.65 No/No Yes 1.43 ns
achievement pre-onset of
psychosis96
ATD angle48 5 g, 0.16 (20.02 to 0.34) 261 0.083 0% (0.54) 20.13 to 0.46 No/No No 1.34 ns
Poor mathematic academic 3 g, 0.11 (20.24 to 0.47) 136 0.527 63% (0.06) 23.77 to 3.99 No/No Yes 1.23 ns
achievement pre-onset of
psychosis96
Delay in grabbing object97 3 g, 0.05 (20.07 to 0.17) 351 0.440 14% (0.31) 20.90 to 1.00 Yes/NA No 1.09 ns
Novelty seeking101 7 g, 20.31 (20.68 to 0.05) 384 0.092 85% (<0.001) 21.56 to 0.93 No/No No 0.57 ns
Total finger ridge count 48

13 g, 20.12 (20.29 to 0.04) 935 0.149 65% (0.001) 20.69 to 0.44 No/Yes No 0.80 ns
k – number of samples for each factor, ES – effect size, N – number of cases, PI – prediction interval, CI – confidence interval, SSE – small study effect, ESB – excess
significance bias, LS – largest study with significant effect, eOR – equivalent odds ratio, CE – class of evidence, IRR – incidence rate ratio, OR – odds ratio,
RR – relative risk, ATD angle – dermatoglyphic feature that compares the length of the hand to the width, NA – not assessable, ns – not significant
(e.g., urbanicity) and neurobiological alterations in psychotic (winter/spring season of birth in Northern hemisphere) or later
disorders have only just started to emerge 133,136. Until the exact factors/antecedents (childhood trauma and childhood social
mechanisms that lead to an increased risk of psychosis are withdrawal, Toxoplasma gondii IgG, minor physical
determined, the requirement for biological or psychological anomalies, trait anhedonia, low olfactory identification ability,
plausibility for these factors cannot be fully met. Importantly, low premorbid IQ, and non-right handedness) might have in
future research is required to clarify the contextual specifics of psychosis onset. At the same time, a number of the explored
ethnic minority status and urbanicity, because their effects may factors showed only weak evidence of association with
also be modulated by geographical location or predominant psychotic disorders. Some of these factors, such as heavy
population factors, rather than having universal value. cannabis use and obstetric complications, were expected to
Several other factors beyond the ultra-high-risk state, ethnic have stronger evidence. However, weak findings in these areas
minority status, and urbanicity provided a highly suggestive or may simply indicate that there are not yet enough data. Our
a suggestive level of evidence of association with psychotic umbrella review also identified only a few putative protective
disorders, mostly confirming the role that perinatal factors factors, indicating that the vast majority of available studies

Table 9 Sensitivity analysis for the associations of socio-demographic and parental, perinatal, later factors, antecedents and psychotic disorders within
individual prospective studies of class I-III factors
Factor CE k ES (95% CI) N>1000 I2 (p) PI (95% CI) SSE/ESB LS eOR CES
Ultra-high-risk state for I 9 RR, 9.32 (4.91 to 17.72) Yes 9.5 3 10212 0% (0.91) 4.30 to 20.24 No/No No 9.32 I
psychosis98
Urbanicity78 III 8 OR, 2.19 (1.55 to 3.09) Yes 8.9 3 1026 99% (<0.001) 0.62 to 7.77 No/No Yes 2.19 III
Black-Caribbean ethnicity in I 7 IRR, 5.54 (4.50 to 6.82) No 4.9 3 10259

0% (0.48) 4.22 to 7.27 No/No Yes 5.54 IV
England76
Ethnic minority in low ethnic II 3 IRR, 4.27 (1.89 to 9.68) No 4.9 3 1024 82% (0.004) 0 to 75335 Yes/No Yes 4.27 IV
density area71
North African immigrants in III 8 IRR, 3.20 (2.36 to 4.35) No 1.0 3 10213 21% (0.27) 1.73 to 5.94 No/NA Yes 3.20 IV
Europe77
Childhood trauma 54 III 4 OR, 2.52 (1.27 to 5.02) Yes 0.009 71% (0.016) 0.14 to 46.01 No/Yes No 2.52 IV
Ethnic minority in high ethnic III 3 IRR, 2.51 (1.10 to 5.71) No 0.028 70% (0.037) 0 to 28153 No/No Yes 2.51 IV
density area71
Childhood social withdrawal42,43 III 11 g, 0.43 (0.14 to 0.71) Yes 0.003 94% (<0.001) 20.63 to 1.48 No/No Yes 2.16 IV
First generation immigrants 53 III 12 IRR, 1.83 (1.40 to 2.38) No 9.6 3 1026 0% (0.82) 1.35 to 2.47 No/Yes No 1.83 IV
Second generation immigrants 53

II 10 IRR, 1.45 (1.05 to 2.01) Yes 0.023 76% (<0.001) 0.54 to 3.95 Yes/No No 1.45 IV
Toxoplasma gondii IgG95 III 7 OR, 1.28 (1.06 to 1.55) Yes 0.012 22% (0.26) 0.86 to 1.91 Yes/No No 1.28 IV
Premorbid IQ 38,39
III 9 g, 20.43 (–0.64 to 20.22) No 5.2 3 10 25
62% (0.007) 21.04 to 0.18 No/No Yes 0.46 IV
Non-right handedness47 III 1 OR, 1.83 (0.62 to 5.39) No 0.273 NA NA NA/NA No 1.83 ns
Minor physical anomalies 46

II NC NC NC NC NC NC NC NC NC NC
Olfactory identification ability63 II NC NC NC NC NC NC NC NC NC NC

Trait anhedonia 105
II NC NC NC NC NC NC NC NC NC NC
Winter/spring season of birth in III NC NC NC NC NC NC NC NC NC NC

Northern hemisphere79
CE – class of evidence, k – number of samples for each factor within prospective studies, ES – effect size, CI – confidence interval, N – number of cases, PI – prediction interval, SSE
– small study effect, ESB – excess significance bias, LS – largest study with significant effect, eOR – equivalent odds ratio, CES – class of evidence after sensitivity analysis, RR –
relative risk, OR – odds ratio, IRR – incidence rate ratio, NA – not assessable, Ig – immunoglobulin, ns – non-significant, NC
have focused on the adverse or negative end of several factors. psychosis in populations at risk, paralleling the recent
Future research is required to actively seek unstudied advancements observed in genetics.
protective factors that are not reciprocal to risk factors, such as In this latter area, the availability of robust meta-analytical
specific characteristics of the individual, family or wider evidence of associations between genetic loci and psychotic
environment that improve the likelihood of positive disorders – provided by the genome-wide association study
outcomes137. (GWAS) meta-analysis conducted by the Schizophrenia
This study has several conceptual implications. On an Working Group of the Psychiatric Genomics Consortium 44 –
etiopathological level, our findings corroborate the notion that has ultimately led to the development of polygenic risk scores
psychotic disorders can be related to adversities in an to assess the en masse genetic effect of several loci 140.
individual’s social milieu, whereby environmental exposures Polygenic risk scores have been used to predict case-control
during critical developmental periods impact brain, status at the time of a first-episode psychosis, explaining
neurocognition, affect, and social cognition 13,138. It is also approximately 9% of the variance 140. The small proportion of
apparent that most of these factors are likely not specific to variance explained indicates that the use of polygenic risk
psychosis, but also associated with other mental disorders 139. scores in clinical routine would be unwarranted 44 without first
From a transdiagnostic perspective, the current study can boosting them with other non-purely genetic factors.
provide a benchmark for comparing the magnitude of – not calculable (no prospective studies to be analyzed)
association of these factors with other nonpsychotic mental
disorders. On a risk prediction level, these results may
substantially advance our ability to prognosticate the onset of

To date, the integration of multiple non-purely genetic factors that are difficult to study (or uncommonly studied), the
factors into “polyrisk” scores has been limited by the lack of current grade of evidence is unlikely to be remarkable, given
established and robust a priori knowledge on their association the limited data.
with psychotic disorders139. The current umbrella review A third limitation is that the definition of healthy control
attempted to fill this knowledge gap by providing the most groups employed by each meta-analysis/systematic review or,
robust estimates of association (ORs) between several non- when this was not provided, by each individual study, may not
purely genetic risk (or protective) factors and psychotic be entirely accurate. Moreover, some of the factors included in
disorders. Assessing these predictive factors may offer some this umbrella review may be better conceptualized as risk
logistic advantages over more complex measurements that are markers, because they may be the result of different interacting
based on cognitive, imaging, central or peripheral measures. risk factors. The ultra-high-risk state 98, ethnicity76 and
Simple demographic factors have already been used to develop immigration status53,77 are prototypical examples of risk
an individualized risk estimation tool to predict psychosis markers, and their limitations have already been addressed
onset in at-risk individuals in clinical practice141. above.
Recently, geneticists have advocated the development of Another caution is that the categories of socio-demographic
polyrisk scores encompassing socio-demographic, parental, and parental, perinatal, later factors, and antecedents 7-9 were
perinatal, later factors, antecedents, and genetic risk used only for descriptive purposes. As noted in the Methods
profiling139,142. Such an approach may ultimately reveal new, section, these categories may actually overlap to some extent.
clinically useful predictors, because even the risk factors that Finally, the relevance of epigenetic risk factors, and the
we found to be weakly associated with psychotic disorders may interaction between environmental and genetic factors in
eventually contribute to the predictive accuracy of the model, psychotic disorders, remains to be elucidated157.
as previously observed for genetic associations44. The current In conclusion, we found several factors to be associated with
umbrella review lays the groundwork for testing the predictive psychotic disorders at different levels of evidence. These
accuracy of integrated polyrisk scores in independent factors represent a starting point of knowledge that can be used
samples139. to advance etiological research and improve the prediction of
Finally, on a pragmatic level, the current stratification of psychosis.
evidence can be used by clinicians, policy makers and
regulatory bodies to inform and strategically target outreach
campaigns, to promote the prevention of mental disorders in ACKNOWLEDGEMENTS
the youth population, and to raise awareness of risk factors for This study was supported in part by the National Institute for Health Research
Biomedical Research Centre at South London and Maudsley National Health
psychotic disorders. Service Foundation Trust and King’s College London, by a 2017 Medical
This study also has some limitations. First, association is not Research Council Confidence in Concept grant to P. Fusar-Poli, and by Instituto
de Salud Carlos III and FEDER grants (CP14/00041 and PI14/00292) to J.
necessarily causation. Reverse causation is a particular
Radua. The funders had no influence on the design and conduct of the study;
concern13, and thus establishing the temporality of the collection, management, analysis, and interpretation of the data; preparation,
association is critical. It is possible that some of the later review, or approval of the manuscript; and decision to submit the manuscript for
publication. The views expressed are those of the authors and not necessarily
factors and antecedents are actually characteristics of psychotic those of the funders. J. Radua and V. Ramella-Cravaro contributed equally to this
disorders themselves or secondary to their appearance. To work.
specifically address these problems and the effect of
temporality, we conducted a sensitivity analysis restricted to
individual studies with prospective designs.
REFERENCES
A second limitation is that the umbrella review approach
1. Global Burden of Disease Study Collaborators. Global, regional, and
may favour the selection of more commonly and readily national incidence, prevalence, and years lived with disability for 301
studied factors, since they are more likely to be included in a acute and chronic diseases and injuries in 188 countries, 1990-2013: a
metaanalysis. We cannot exclude the possibility that some systematic analysis for the Global Burden of Disease Study 2013. Lancet
2015;386:743-800.
promising factors, despite having sufficient data, do not have a 2. Kirkbride JB, Errazuriz A, Croudace TJ et al. Incidence of schizophrenia
corresponding eligible meta-analysis, such as mood and and other psychoses in England, 1950–2009: a systematic review and
meta-analyses. PLoS One 2012;7:e31660.
anxiety disorders143-145, personality disorders146, attachment147,
3. Cloutier M, Aigbogun MS, Guerin A et al. The economic burden of
alcohol and psychoactive substances 148-151, sleep dysfunction152, schizophrenia in the United States in 2013. J Clin Psychiatry 2016;77:764-
homelessness153 or pervasive developmental disorders154. 71.
4. Fusar-Poli P, McGorry PD, Kane J. Improving outcomes of first-episode
However, this possibility is becoming less likely in the current psychosis: an overview. World Psychiatry 2017;16:251-65.
era, with meta-analyses being performed massively, to the 5. van Os J, Rutten BPF, Poulton R. Gene-environment interactions in
point that for several topics multiple meta-analyses are schizophrenia: review of epidemiological findings and future directions.
Schizophr Bull 2008;34:1066-82.
available155,156. In any case, for most putative risk or protective

6. Howes OD, Murray RM. Schizophrenia: an integrated randomized trials and 29 meta-analyses. Intensive Care Med 2014;40:
sociodevelopmentalcognitive model. Lancet 2014;383:1677-87. 769-87.
7. Matheson SL, Shepherd AM, Laurens KR et al. A systematic meta-review 31. Ioannidis JP, Zhou Y, Chang CQ et al. Potential increased risk of cancer
grading the evidence for non-genetic risk factors and putative antecedents from commonly used medications: an umbrella review of meta-analyses.
of schizophrenia. Schizophr Res 2011;133:133-42. Ann Oncol 2014;25:16-23.
8. Laurens KR, Luo L, Matheson SL et al. Common or distinct pathways to 32. Doufas AG, Panagiotou OA, Ioannidis JP. Concordance of sleep and pain
psychosis? A systematic review of evidence from prospective studies for outcomes of diverse interventions: an umbrella review. PLoS One 2012;7:
developmental risk factors and antecedents of the schizophrenia spectrum e40891.
disorders and affective psychoses. BMC Psychiatry 2015;15:205. 33. Contopoulos-Ioannidis DG, Ioannidis JP. Claims for improved survival
9. Dean K MR. Environmental risk factors for psychosis. Dialogues Clin from systemic corticosteroids in diverse conditions: an umbrella review.
Neurosci 2005;7:69-80. Eur J Clin Invest 2012;42:233-44.
10. Fusar-Poli P. The Clinical High-Risk State for Psychosis (CHR-P), 34. Fleischhacker WW. A meta view on meta-analyses. JAMA Psychiatry
Version II. Schizophr Bull 2017;43:44-7. 2017;
11. Fusar-Poli P, Schultze-Lutter F. Predicting the onset of psychosis in 74:684-5.
patients at clinical high risk: practical guide to probabilistic prognostic 35. Ioannidis JPA. Integration of evidence from multiple meta-analyses: a
reasoning. Evid Based Ment Health 2016;19:10-5. primer on umbrella reviews, treatment networks and multiple treatments
12. Ioannidis JP, Munafo MR, Fusar-Poli P et al. Publication and other meta-analyses. CMAJ 2009;181:488-93.
reporting biases in cognitive sciences: detection, prevalence, and 36. Geddes JR, Verdoux H, Takei N et al. Schizophrenia and complications of
prevention. Trends Cogn Sci 2014;18:235-41. pregnancy and labor: an individual patient data meta-analysis. Schizophr
13. van Os J, Kenis G, Rutten BPF. The environment and schizophrenia. Bull 1999;25:413-23.
Nature 2010;468:203-12. 37. Cannon M, Jones PB, Murray RM. Obstetric complications and
14. Aromataris E, Fernandez R, Godfrey CM et al. Summarizing systematic schizophrenia: historical and meta-analytic review. Am J Psychiatry
reviews: methodological development, conduct and reporting of an 2002;159: 1080-92.
umbrella review approach. Int J Evid Based Health 2015;13:132-40. 38. Woodberry KA, Giuliano AJ, Seidman LJ. Premorbid IQ in schizophrenia:
15. Dinu M, Pagliai G, Casini A et al. Mediterranean diet and multiple health a meta-analytic review. Am J Psychiatry 2008;165:579-87.
outcomes: an umbrella review of meta-analyses of observational studies 39. Khandaker GM, Barnett JH, White IR et al. A quantitative meta-analysis
and randomised trials. Eur J Clin Nutr (in press). of population-based studies of premorbid intelligence and schizophrenia.
16. Bortolato B, Kohler CA, Evangelou E et al. Systematic assessment of Schizophr Res 2011;132:220-7.
environmental risk factors for bipolar disorder: an umbrella review of 40. Cai L, Wan CL, He L et al. Gestational influenza increases the risk of
systematic reviews and meta-analyses. Bipolar Disord 2017;19:84-96. psychosis in adults. Med Chem 2015;11:676-82.
17. Kyrgiou M, Kalliala I, Markozannes G et al. Adiposity and cancer at 41. Selten J-P, Termorshuizen F. The serological evidence for maternal
major anatomical sites: umbrella review of the literature. BMJ influenza as risk factor for psychosis in offspring is insufficient: critical
2017;356:j477. review and meta-analysis. Schizophr Res 2017;183:2-9.
18. Dragioti E, Karathanos V, Gerdle B et al. Does psychotherapy work? An 42. Tarbox SI, Pogue-Geile MF. Development of social functioning in
umbrella review of meta-analyses of randomized controlled trials. Acta preschizophrenia children and adolescents: a systematic review. Psychol
Psychiatr Scand 2017;136:236-46. Bull 2008;134:561-83.
19. Markozannes G, Tzoulaki I, Karli D et al. Diet, body size, physical 43. Matheson SL, Vijayan H, Dickson H et al. Systematic meta-analysis of
activity and risk of prostate cancer: an umbrella review of the evidence. childhood social withdrawal in schizophrenia, and comparison with data
Eur J Cancer 2016;69:61-9. from at-risk children aged 9-14 years. J Psychiatr Res 2013;47:1061-8.
20. Belbasis L, Savvidou MD, Kanu C et al. Birth weight in relation to health 44. Schizophrenia Working Group of the Psychiatric Genomics Consortium.
and disease in later life: an umbrella review of systematic reviews and Biological insights from 108 schizophrenia-associated genetic loci. Nature
meta-analyses. BMC Med 2016;14:147. 2014;511:421-7.
21. Belbasis L, Stefanaki I, Stratigos AJ et al. Non-genetic risk factors for 45. O’Dushlaine C, Rossin L, Lee PH et al. Psychiatric genome-wide
cutaneous melanoma and keratinocyte skin cancers: an umbrella review of association study analyses implicate neuronal, immune and histone
meta-analyses. J Dermatol Sci 2016;84:330-9. pathways. Nat Neurosci 2015;18:199-209.
22. Papageorgiou PN, Deschner J, Papageorgiou SN. Effectiveness and 46. Xu T, Chan RCK, Compton MT. Minor physical anomalies in patients
adverse effects of deep brain stimulation: umbrella review of meta- with schizophrenia, unaffected first-degree relatives, and healthy controls:
analyses. J Neurol Surg A Cent Eur Neurosurg 2017;78:180-90. a meta-analysis. PLoS One 2011;6:e24129.
23. Bellou V, Belbasis L, Tzoulaki I et al. Systematic evaluation of the 47. Hirnstein M, Hugdahl K. Excess of non-right-handedness in
associations between environmental risk factors and dementia: an umbrella schizophrenia: meta-analysis of gender effects and potential biases in
review of systematic reviews and meta-analyses. Alzheimers Dement handedness assessment. Br J Psychiatry 2014;205:260-7.
2017; 13:406-18. 48. Golembo-Smith S, Walder DJ, Daly MP et al. The presentation of
24. Carvalho AF, Kohler CA, Fernandes BS et al. Bias in emerging dermatoglyphic abnormalities in schizophrenia: a meta-analytic review.
biomarkers for bipolar disorder. Psychol Med 2016;46:2287-97. Schizophr Res 2012;142:1-11.
25. Bellou V, Belbasis L, Tzoulaki I et al. Environmental risk factors and 49. Neelam K, Garg D, Marshall M. A systematic review and meta-analysis of
Parkinson’s disease: an umbrella review of meta-analyses. Parkinsonism neurological soft signs in relatives of people with schizophrenia. BMC
Relat Disord 2016;23:1-9. Psychiatry 2011;11:139.
26. Belbasis L, Bellou V, Evangelou E. Environmental risk factors and 50. Christesen HT, Elvander C, Lamont RF et al. The impact of vitamin D in
amyotrophic lateral sclerosis: an umbrella review and critical assessment pregnancy on extraskeletal health in children: a systematic review. Acta
of current evidence from systematic reviews and meta-analyses of Obstet Gynecol Scand 2012;91:1368-80.
observational studies. Neuroepidemiology 2016;46:96-105. 51. Moher D, Liberati A, Tetzlaff J et al. Preferred reporting items for
27. Belbasis L, Bellou V, Evangelou E et al. Environmental risk factors and systematic reviews and meta-analyses: the PRISMA statement. J Clin
multiple sclerosis: an umbrella review of systematic reviews and Epidemiol 2009;62:1006-12.
metaanalyses. Lancet Neurol 2015;14:263-73. 52. Stroup DF, Berlin JA, Morton SC et al. Meta-analysis of observational
28. Tsilidis KK, Kasimis JC, Lopez DS et al. Type 2 diabetes and cancer: studies in epidemiology - a proposal for reporting. JAMA
umbrella review of meta-analyses of observational studies. BMJ 2015; 2000;283:200812.
350:g7607. 53. Bourque F, van der Ven E, Malla A. A meta-analysis of the risk for
29. Theodoratou E, Tzoulaki I, Zgaga L et al. Vitamin D and multiple health psychotic disorders among first- and second-generation immigrants.
outcomes: umbrella review of systematic reviews and meta-analyses of Psychol Med 2011;41:897-910.
observational studies and randomised trials. BMJ 2014;348:g2035. 54. Varese F, Smeets F, Drukker M et al. Childhood adversities increase the
30. Tonelli AR, Zein J, Adams J et al. Effects of interventions on survival in risk of psychosis: a meta-analysis of patient-control, prospective- and
acute respiratory distress syndrome: an umbrella review of 159 published cross-sectional cohort studies. Schizophr Bull 2012;38:661-71.

55. Kinney DK, Teixeira P, Hsu D et al. Relation of schizophrenia prevalence 81. Geddes JR, Lawrie SM. Obstetric complications and schizophrenia: a
to latitude, climate, fish consumption, infant mortality, and skin color: a meta-analysis. Br J Psychiatry 1995;167:786-93.
role for prenatal vitamin D deficiency and infections? Schizophr Bull 82. Van Lieshout RJ, Taylor VH, Boyle MH. Pre-pregnancy and pregnancy
2009;35:582-95. obesity and neurodevelopmental outcomes in offspring: a systematic
56. Saha S, Chant D, Welham J et al. A systematic review of the prevalence of review. Obes Rev 2011;12:e548-59.
schizophrenia. PLoS Med 2005;2:413-33. 83. Arias I, Sorlozano A, Villegas E et al. Infectious agents associated with
57. Rohatgi A. WebPlotDigitizer 3.10. arohatgi.info/WebPlotDigitizer/. schizophrenia: a meta-analysis. Schizophr Res 2012;136:128-36.
84. Attademo L, Bernardini F, Garinella R et al. Environmental pollution and
risk of psychotic disorders: a review of the science to date. Schizophr Res
58. Shea BJ, Grimshaw JM, Wells GA et al. Development of AMSTAR: a 2017;181:55-9.
measurement tool to assess the methodological quality of systematic 85. Beards S, Gayer-Anderson C, Borges S et al. Life events and psychosis: a
reviews. BMC Med Res Methodol 2007;7:10. review and meta-analysis. Schizophr Bull 2013;39:740-7.
59. Shea BJ, Bouter LM, Peterson J et al. External validation of a 86. Clancy MJ, Clarke MC, Connor DJ et al. The prevalence of psychosis in
measurement tool to assess systematic reviews (AMSTAR). PLoS One epilepsy; a systematic review and meta-analysis. BMC Psychiatry 2014;
2007;2: e1350. 14:75.
60. Pieper D, Mathes T, Eikermann M. Can AMSTAR also be applied to 87. Cunningham T, Hoy K, Shannon C. Does childhood bullying lead to the
systematic reviews of non-randomized studies? BMC Res Notes development of psychotic symptoms? A meta-analysis and review of
2014;7:609. prospective studies. Psychosis 2015;8:48-59.
61. R Development Core Team. R: a language and environment for statistical 88. de Sousa P, Varese F, Sellwood W et al. Parental communication and
computing. Vienna: R Foundation for Statistical Computing, 2011. psychosis: a meta-analysis. Schizophr Bull 2014;40:756-68.
62. Higgins JPT, Green S (eds). Cochrane handbook for systematic reviews of 89. Gurillo P, Jauhar S, Murray RM et al. Does tobacco use cause psychosis?
interventions version 5.1.0. The Cochrane Collaboration, 2011. Systematic review and meta-analysis. Lancet Psychiatry 2015;2:718-25.
63. Moberg P, Kamath V, Marchetto D et al. Meta-analysis of olfactory 90. Gutierrez-Fernandez J, de Dios Luna del Castillo J, Mananes-Gonzalez S
function in schizophrenia, first-degree family members, and youths at-risk et al. Different presence of Chlamydia pneumoniae, herpes simplex virus
for psychosis. Schizophr Bull 2014;40:50-9. type 1, human herpes virus 6, and Toxoplasma gondii in schizophrenia:
64. Rubia K, Alegria AA, Cubillo AI et al. Effects of stimulants on brain meta-analysis and analytical study. Neuropsychiatr Dis Treat 2015;11:
function in attention-deficit/hyperactivity disorder: a systematic review 843-52.
and meta-analysis. Biol Psychiatry 2014;76:616-28. 91. Khandaker GM, Zimbron J, Dalman C et al. Childhood infection and adult
65. Schwarzer G. Meta: an R package for meta-analysis. R News 2007;7:40-5. schizophrenia: a meta-analysis of population-based studies. Schizophr Res
66. Higgins JPT, Thompson SG, Spiegelhalter DJ. A re-evaluation of 2012;139:161-8.
randomeffects meta-analysis. J R Stat Soc A Stat 2009;172:137-59. 92. Linszen MMJ, Brouwer RM, Heringa SM et al. Increased risk of psychosis
67. Jackson D, Kirkbride J, Croudace T et al. Meta-analytic approaches to in patients with hearing impairment: review and meta-analyses. Neurosci
determine gender differences in the age-incidence characteristics of Biobehav Rev 2016;62:1-20.
schizophrenia and related psychoses. Int J Methods Psychiatr Res 2013; 93. Marconi A, Di Forti M, Lewis CM et al. Meta-analysis of the association
22:36-45. between the level of cannabis use and risk of psychosis. Schizophr Bull
68. Egger M, Smith GD, Schneider M et al. Bias in meta-analysis detected by 2016;42:1262-9.
a simple, graphical test. BMJ 1997;315:629-34. 94. Molloy C, Conroy RM, Cotter DR et al. Is traumatic brain injury a risk
69. Ioannidis JPA, Trikalinos TA. An exploratory test for an excess of factor for schizophrenia? A meta-analysis of case-controlled
significant findings. Clin Trials 2007;4:245-53. populationbased studies. Schizophr Bull 2011;37:1104-10.
70. Chinn S. A simple method for converting an odds ratio to effect size for 95. Sutterland AL, Fond G, Kuin A et al. Beyond the association. Toxoplasma
use in meta-analysis. Stat Med 2000;19:3127-31. gondii in schizophrenia, bipolar disorder, and addiction: systematic review
71. Bosqui TJ, Hoy K, Shannon C. A systematic review and meta-analysis of and meta-analysis. Acta Psychiatr Scand 2015;132:161-79.
the ethnic density effect in psychotic disorders. Soc Psychiatry Psychiatr 96. Dickson H, Laurens KR, Cullen AE et al. Meta-analyses of cognitive and
Epidemiol 2014;49:519-29. motor function in youth aged 16 years and younger who subsequently
72. Torrey EF, Buka S, Cannon TD et al. Paternal age as a risk factor for develop schizophrenia. Psychol Med 2012;42:743-55.
schizophrenia: how important is it? Schizophr Res 2009;114:1-5. 97. Filatova S, Koivumaa-Honkanen H, Hirvonen N et al. Early motor
73. Kwok W. Is there evidence that social class at birth increases risk of developmental milestones and schizophrenia: a systematic review and
psychosis? A systematic review. Int J Soc Psychiatry 2014;60:801-8. metaanalysis. Schizophr Res 2017;188:13-20.
74. O’Donoghue B, Roche E, Lane A. Neighbourhood level social deprivation 98. Fusar-Poli P, Cappucciati M, Rutigliano G et al. At risk or not at risk? A
and the risk of psychotic disorders: a systematic review. Soc Psychiatry meta-analysis of the prognostic accuracy of psychometric interviews for
Psychiatr Epidemiol 2016;51:941-50. psychosis prediction. World Psychiatry 2015;14:322-32.
75. Rasic D, Hajek T, Alda M et al. Risk of mental illness in offspring of 99. Kaymaz N, Drukker M, Lieb R et al. Do subthreshold psychotic
parents with schizophrenia, bipolar disorder, and major depressive experiences predict clinical outcomes in unselected non-help-seeking
disorder: a meta-analysis of family high-risk studies. Schizophr Bull population-based samples? A systematic review and meta-analysis,
2014;40:28-38. enriched with new results. Psychol Med 2012;42:2239-53.
76. Tortelli A, Errazuriz A, Croudace T et al. Schizophrenia and other 100. Koning JPF, Tenback DE, van Os J et al. Dyskinesia and Parkinsonism
psychotic disorders in Caribbean-born migrants and their descendants in inantipsychotic-naive patients with schizophrenia, first-degree relatives
England: systematic review and meta-analysis of incidence rates, 1950– and healthy controls: a meta-analysis. Schizophr Bull 2010;36:723-31.
2013. Soc Psychiatry Psychiatr Epidemiol 2015;50:1039-55. 101. Ohi K, Hashimoto R, Yasuda Y et al. Personality traits and schizophrenia:
77. van der Ven E, Veling W, Tortelli A et al. Evidence of an excessive evidence from a case-control study and meta-analysis. Psychiatry Res
gender gap in the risk of psychotic disorder among North African 2012;198:7-11.
immigrants in Europe: a systematic review and meta-analysis. Soc 102. Ohi K, Shimada T, Nitta Y et al. The Five-Factor Model personality
Psychiatry Psychiatr Epidemiol 2016;51:1603-13. traitsin schizophrenia: a meta-analysis. Psychiatry Res 2016;240:34-41.
78. Vassos E, Pedersen CB, Murray RM et al. Meta-analysis of the association 103. Potvin S, Marchand S. Hypoalgesia in schizophrenia is independent
of urbanicity with schizophrenia. Schizophr Bull 2012;38:1118-23. ofantipsychotic drugs: a systematic quantitative review of experimental
79. Davies G, Welham J, Chant D et al. A systematic review and metaanalysis studies. Pain 2008;138:70-8.
of Northern Hemisphere season of birth studies in schizophrenia. 104. Ward KE, Friedman L, Wise A et al. Meta-analysis of brain and
Schizophr Bull 2003;29:587-93. cranialsize in schizophrenia. Schizophr Res 1996;22:197-213.
80. McGrath JJ, Welham JL. Season of birth and schizophrenia: a systematic 105. Yan C, Cao Y, Zhang Y et al. Trait and state positive emotional
review and meta-analysis of data from the Southern Hemisphere. experiencein schizophrenia: a meta-analysis. PLoS One 2012;7:e40672.
Schizophr Res 1999;35:237-42.

106. Selten JP, Cantor-Graae E, Nahon D et al. No relationship between risk 130. Powell ND, Sloan EK, Bailey MT et al. Social stress up-regulates
ofschizophrenia and prenatal exposure to stress during the Six-Day War or inflammatory gene expression in the leukocyte transcriptome via beta-
Yom Kippur War in Israel. Schizophr Res 2003;63:131-5. adrenergic induction of myelopoiesis. Proc Natl Acad Sci USA
107. O’Rourke K. An historical perspective on meta-analysis: dealing 2013;110:16574-
quantitatively with varying study results. J Roy Soc Med 2007;100:579- 9.
82. 131. Akdeniz C, Tost H, Streit F et al. Neuroimaging evidence for a role of
108. Fusar-Poli P, Borgwardt S, Bechdolf A et al. The psychosis high-risk state: neural social stress processing in ethnic minority-associated environmental
acomprehensive state-of-the-art review. JAMA Psychiatry 2013;70:107- risk. JAMA Psychiatry 2014;71:672-80.
20. 132. Lederbogen F, Kirsch P, Haddad L et al. City living and urban
109. Fusar-Poli P, Rutigliano G, Stahl D et al. Long-term validity of the At upbringingaffect neural social stress processing in humans. Nature
RiskMental State (ARMS) for predicting psychotic and non-psychotic 2011;474:498501.
mental disorders. Eur Psychiatry 2017;42:49-54. 133. Haddad L, Schafer A, Streit F et al. Brain structure correlates of
110. Kempton MJ, Bonoldi I, Valmaggia L et al. Speed of psychosis urbanupbringing, an environmental risk factor for schizophrenia.
progressionin people at ultra-high clinical risk: a complementary meta- Schizophr Bull 2015;41:115-22.
analysis. JAMA Psychiatry 2015;72:622-3. 134. Selten JP, Booij J, Buwalda B et al. Biological mechanisms whereby
111. Fusar-Poli P, Cappucciati M, De Micheli A et al. Diagnostic and socialexclusion may contribute to the etiology of psychosis: a narrative
prognostic significance of brief limited intermittent psychotic symptoms review. Schizophr Bull (in press).
(BLIPS) in individuals at ultra high risk. Schizophr Bull 2017;43:48-56. 135. Egerton A, Howes OD, Houle S et al. Elevated striatal dopamine
112. Fusar-Poli P, Cappucciati M, Bonoldi I et al. Prognosis of brief functionin immigrants and their children: a risk mechanism for psychosis.
psychoticepisodes: a meta-analysis. JAMA Psychiatry 2016;73:211-20. Schizophr Bull 2017;43:293-301.
113. Falkenberg I, Valmaggia L, Byrnes M et al. Why are help-seeking 136. Frissen A, van Os J, Habets P et al. No evidence of association
subjectsat ultra-high risk for psychosis help-seeking? Psychiatry Res betweenchildhood urban environment and cortical thinning in psychotic
2015;228: 808-15. disorder. PLoS One 2017;12:e0166651.
114. Fusar-Poli P, Rocchetti M, Sardella A et al. Disorder, not just state of risk: 137. Keskinen E, Marttila R, Koivumaa-Honkanen H et al. Search for
meta-analysis of functioning and quality of life in people at high risk of protective factors for psychosis - a population-based sample with special
psychosis. Br J Psychiatry 2015;207:198-206. interest in unaffected individuals with parental psychosis. Early Interv
115. Fusar-Poli P, Cappucciati M, Borgwardt S et al. Heterogeneity of Psychiatry (in press).
psychosis risk within individuals at clinical high risk: a meta-analytical 138. Millan MJ, Andrieux A, Bartzokis G et al. Altering the course of
stratification. JAMA Psychiatry 2016;73:113-20. schizophrenia: progress and perspectives. Nat Rev Drug Discov
116. Fusar-Poli P, Cappucciati M, Rutigliano G et al. Towards a standard 2016;15:485-515.
psychometric diagnostic interview for subjects at ultra high risk of 139. Uher R, Zwicker A. Etiology in psychiatry: embracing the reality of
psychosis: CAARMS versus SIPS. Psychiatry J 2016;2016:7146341. polygene-environmental causation of mental illness. World Psychiatry
117. Fusar-Poli P, Tantardini M, De Simone S et al. Deconstructing 2017; 16:121-9.
vulnerability for psychosis: meta-analysis of environmental risk factors for 140. Vassos E, Di Forti M, Coleman J et al. An examination of polygenic
psychosis in subjects at ultra high-risk. Eur Psychiatry 2017;40:65-75. scorerisk prediction in individuals with first episode psychosis. Biol
118. Fusar-Poli P, Rutigliano G, Stahl D et al. Deconstructing pretest Psychiatry 2016;81:470-7.
riskenrichment to optimize prediction of psychosis in individuals at 141. Fusar-Poli P, Rutigliano G, Stahl D et al. Development and validation of
clinical high risk. JAMA Psychiatry 2016;73:1260-7. aclinically based risk calculator for the transdiagnostic prediction of
119. Fusar-Poli P, Schultze-Lutter F, Cappucciati M et al. The dark side of psychosis. JAMA Psychiatry 2017;74:493-500.
themoon: meta-analytical impact of recruitment strategies on risk 142. Iyegbe C, Campbell D, Butler A et al. The emerging molecular
enrichment in the clinical high risk state for psychosis. Schizophr Bull architecture of schizophrenia, polygenic risk scores and the clinical
2016;42: 732-43. implications for GxE research. Soc Psychiatry Psychiatr Epidemiol
120. Fusar-Poli P, Schultze-Lutter F, Addington J. Intensive community 2014;49:169-82.
outreach for those at ultra high risk of psychosis: dilution, not solution. 143. Hartley S, Barrowclough C, Haddock G. Anxiety and depression in
Lancet Psychiatry 2016;3:18. psychosis: a systematic review of associations with positive psychotic
121. Fusar-Poli P, Palombini E, Davies C et al. Why transition risk to symptoms. Acta Psychiatr Scand 2013;128:327-46.
psychosisis not declining at the OASIS ultra high risk service: the hidden 144. Mishara AL, Fusar-Poli P. The phenomenology and neurobiology of
role of stable pretest risk enrichment. Schizophr Res (in press). delusion formation during psychosis onset: Jaspers, Truman symptoms,
122. Fusar-Poli P. Why ultra high risk criteria for psychosis prediction do and aberrant salience. Schizophr Bull 2013;39:278-86.
notwork well outside clinical samples and what to do about it. World 145. Achim AM, Maziade M, Raymond E et al. How prevalent are anxiety
Psychiatry 2017;16:212-3. disorders in schizophrenia? A meta-analysis and critical review on a
123. Kirkbride J, Hameed Y, Ankireddypalli G et al. The epidemiology of significant association. Schizophr Bull 2011;37:811-21.
firstepisode psychosis in early intervention in psychosis services: findings 146. Newton-Howes G, Tyrer P, North B et al. The prevalence of
from the Social Epidemiology of Psychoses in East Anglia [SEPEA] personalitydisorder in schizophrenia and psychotic disorders: systematic
Study. Am J Psychiatry 2017;174:143-53. review of rates and explanatory modelling. Psychol Med 2008;38:1075-82.
124. Heinz A, Deserno L, Reininghaus U. Urbanicity, social adversity and 147. Gumley AI, Taylor HEF, Schwannauer M et al. A systematic review
psychosis. World Psychiatry 2013;12:187-97. ofattachment and psychosis: measurement, construct validity and
125. Selten J-P, van Os J, Cantor-Graae E. The social defeat hypothesis outcomes. Acta Psychiatr Scand 2014;129:257-74.
ofschizophrenia: issues of measurement and reverse causality. World 148. Large M, Sharma S, Compton MT et al. Cannabis use and earlier onset
Psychiatry 2016;15:294-5. ofpsychosis. Arch Gen Psychiatry 2011;68:555-61.
126. Mizrahi R. Social stress and psychosis risk: common neurochemical 149. Murray RM, Quigley H, Quattrone D et al. Traditional marijuana,
substrates? Neuropsychopharmacology 2016;41:666-74. highpotency cannabis and synthetic cannabinoids: increasing risk for
127. Holtzman CW, Trotman HD, Goulding SM et al. Stress and psychosis. World Psychiatry 2016;15:195-204.
neurodevelopmental processes in the emergence of psychosis. 150. Bramness JG, Rognli EB. Psychosis induced by amphetamines. Curr
Neuroscience 2013;249: 172-91. OpinPsychiatry 2016;29:236-41.
128. Steinheuser V, Ackermann K, Schonfeld P et al. Stress and the city: 151. Roncero C, Ros-Cucurull E, Daigre C et al. Prevalence and risk factors
impactof urban upbringing on the (re)activity of the hypothalamus- ofpsychotic symptoms in cocaine-dependent patients. Actas Esp Psiquiatr
pituitaryadrenal axis. Psychosom Med 2014;76:678-85. 2012;40:187-97.
129. Mewes R, Reich H, Skoluda N et al. Elevated hair cortisol 152. Reeve S, Sheaves B, Freeman D. The role of sleep dysfunction in
concentrationsin recently fled asylum seekers in comparison to theoccurrence of delusions and hallucinations: a systematic review. Clin
permanently settled immigrants and non-immigrants. Transl Psychiatry Psychol Rev 2015;42:96-115.
2017;7:e1051.

153. Folsom D, Jeste DV. Schizophrenia in homeless persons: a
systematicreview of the literature. Acta Psychiatr Scand 2002;105:404-13.
154. Padgett FE, Miltsiou E, Tiffin PA. The co-occurrence of nonaffective
psychosis and the pervasive developmental disorders: a systematic review.
J Intellect Dev Disabil 2010;35:187-98.
155. Ioannidis JP. The mass production of redundant, misleading, and
conflicted systematic reviews and meta-analyses. Milbank Q
2016;94:485514.
156. Siontis KC, Hernandez-Boussard T, Ioannidis JP. Overlapping
metaanalyses on the same topic: survey of published studies. BMJ
2013;347: f4501.
157. Millan MJ. An epigenetic framework for neurodevelopmental disorders:
from pathogenesis to potential therapy. Neuropharmacology 2013;68:282.
DOI:10.1002/wps.20490

Research Report

Uploaded by

Copyright:

Available Formats

Research Report

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Research Report

Uploaded by

Copyright:

Available Formats

20515545, 2018, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/wps.20490 by INASP/HINARI - INDONESIA, Wiley Online Library on [16/10/2022].

What causes psychosis? An umbrella review of risk and

World Psychiatry 17:1 - February 2018

World Psychiatry 17:1 - February 2018

World Psychiatry 17:1 - February 2018

Kinney et al55 Disadvantaged group; latitude 2, 29 SZ 2/11

Kwok73 Low paternal socio-economic status 9 FEP, SZ, NAP 6/11

O’Donoghue et al74 Neighbourhood level social deprivation 3 FEP 8/11

Saha et al56 Gross national income per capita 88 SZ 9/11

van der Ven et al 77

Vassos et al78 Urbanicity 8 SZ, NAP 6/11

World Psychiatry 17:1 - February 2018

Cai et al40 Gestational influenza 6 SZ, NAP, AP 10/11

Davies et al79 Winter/spring season of birth in Northern hemisphere 7 SZ 6/11

Geddes & Lawrie81 Obstetric complications 10 SZ 6/11

McGrath & Welham80 Winter/spring season of birth in Southern hemisphere 7 SZ 9/11

Selten & Termoshuizen41 Gestational influenza 7 SZ, AP 7/11

World Psychiatry 17:1 - February 2018

Arias et al83 1, 8, 2, 2, 8, 3, 4, 1, 4, 5, 2, 1, 1, 4 SZ 8/11

BK virus; Borna disease virus; Chlamydia psittaci;

Beards et al85 Adult life events 6 FEP, SZ, NAP, AP 8/11

Cunningham et al87 Bullying 1 SZ, NAP 7/11

Gurillo et al89 Tobacco use 5 FEP, SZ, NAP, AP 11/11

Linszen et al92 Hearing impairment 5 SZ 8/11

Molloy et al94 Traumatic brain injury 8 SZ 7/11

World Psychiatry 17:1 - February 2018

Dickson et al96 4, 4, 3 FEP, SZ, NAP 7/11

Motor function pre-onset of psychosis; poor academic achievement pre-onset of

Fusar-Poli et al98 Ultra-high-risk state for psychosis 9 FEP 9/11

Hirnstein & Hugdahl47 Non-right handedness 41 SZ, NAP 5/11

Khandaker et al39 Premorbid IQ 5 SZ, NAP 8/11

Koning et al100 Dyskinesia in antipsychotic-na€ıve schizophrenic patients; parkinsonism in 5, 3 FEP, SZ 5/11

Matheson et al43 Childhood social withdrawal 5 SZ, NAP 8/11

Neelam et al49 Neurological soft signs 7 SZ, NAP 8/11

Ohi et al102 Agreeableness; conscientiousness; extraversion; neuroticism; openness 6, 7, 8, 8, 7 SZ 6/11

Potvin & Marchand103 Hypoalgesia 9 SZ 5/11

Ward et al104 Extracranial size 7 SZ, NAP 3/11

Xu et al46 Minor physical anomalies 14 SZ 4/11

Classification of level of evidence of associations

World Psychiatry 17:1 - February 2018

Parental severe mental illness 75

Black African ethnicity in England 2

Other white ethnicity in England 2

Disadvantaged vs. advantaged groups 55

Paternal age >35 years 72

For six factors there was highly suggestive evidence for

World Psychiatry 17:1 - February 2018

Random-effects measure, p randomeffects

Birth weight <2000 g 37

Birth weight <2500 g 36,37

Gestational age <37 weeks 36,37

World Psychiatry 17:1 - February 2018

Birth weight <2500 g and prematurity 37

Maternal stress during pregnancy 106

Baby detained in hospital 37

Random-effects measure, p randomeffects

Labour >24 hours 36

To our knowledge, this is the first umbrella review of risk

World Psychiatry 17:1 - February 2018