Key PointsQuestion
What are the factors associated with psychotic experiences?
Findings
In this cohort study of 155 247 UK Biobank participants, exposome-wide association analysis yielded 148 correlates of psychotic experiences, with 36 independent associations further identified in the fully adjusted multivariable model. Mendelian randomization analyses of these 36 variables indicated a forward association with ever having experienced sexual assault and pleiotropy of risk-taking behavior and a reverse association with ever having experienced a physically violent crime, cannabis use, and worrying too long after embarrassment.
Meaning
The finding that both well-studied and unexplored multiple correlated variables were associated with psychotic experiences underlines the importance of systematic agnostic approaches and triangulation of evidence with genetically informed approaches to probe associations in the big-data era.
Importance
Although hypothesis-driven research has identified several factors associated with psychosis, this one-exposure-to-one-outcome approach fails to embrace the multiplicity of exposures. Systematic approaches, similar to agnostic genome-wide analyses, are needed to identify genuine signals.
Objective
To systematically investigate nongenetic correlates of psychotic experiences through data-driven agnostic analyses and genetically informed approaches to evaluate associations.
Design, Setting, Participants
This cohort study analyzed data from the UK Biobank Mental Health Survey from January 1 to June 1, 2021. An exposome-wide association study was performed in 2 equal-sized split discovery and replication data sets. Variables associated with psychotic experiences in the exposome-wide analysis were tested in a multivariable model. For the variables associated with psychotic experiences in the final multivariable model, the single-nucleotide variant–based heritability and genetic overlap with psychotic experiences using linkage disequilibrium score regression were estimated, and mendelian randomization (MR) approaches were applied to test potential causality. The significant associations observed in 1-sample MR analyses were further tested in multiple sensitivity tests, including collider-correction MR, 2-sample MR, and multivariable MR analyses.
Exposures
After quality control based on a priori criteria, 247 environmental, lifestyle, behavioral, and economic variables.
Main Outcomes and Measures
Psychotic experiences.
Results
The study included 155 247 participants (87 896 [57%] female; mean [SD] age, 55.94 [7.74] years). In the discovery data set, 162 variables (66%) were associated with psychotic experiences. Of these, 148 (91%) were replicated. The multivariable analysis identified 36 variables that were associated with psychotic experiences. Of these, 28 had significant genetic overlap with psychotic experiences. One-sample MR analyses revealed forward associations with 3 variables and reverse associations with 3. Forward associations with ever having experienced sexual assault and pleiotropy of risk-taking behavior and reverse associations without pleiotropy of experiencing a physically violent crime as well as cannabis use and the reverse association with pleiotropy of worrying too long after embarrassment were confirmed in sensitivity tests. Thus, associations with psychotic experiences were found with both well-studied and unexplored multiple correlated variables. For several variables, the direction of the association was reversed in the final multivariable and MR analyses.
Conclusions and Relevance
The findings of this study underscore the need for systematic approaches and triangulation of evidence to build a knowledge base from ever-growing observational data to guide population-level prevention strategies for psychosis.
Hypothesis-driven observational studies have identified various nongenetic factors associated with psychosis. These environmental factors include relatively well-studied exposures, such as childhood adversity, immigration, racial or ethnic minority status, urbanicity, cannabis use, and obstetric and pregnancy complications,1,2 as well as less studied exposures and lifestyle factors, such as physical activity;3 toxins, such as lead poisoning4 and nitrogen dioxide air pollution;5,6 and nutrients, such as caffeine and magnesium.7,8 Although hypothesis testing is essential and much knowledge on the environmental epidemiology of psychosis has been gained over the years, several limitations of this approach should be acknowledged. First, exposures form highly interconnected clusters.9 Therefore, single-exposure analyses are more prone to yield biased and often overestimated effect sizes and type I errors.9,10 The complexity of associations is also sometimes to the degree that it is difficult to differentiate an exposure from a behavioral outcome in the temporal sequence—for instance, exposure to cannabis vs cannabis use disorder. Second, preconceptions appear to introduce selective reporting and publication bias.11 Third, variation in analytical decisions and variable definitions across studies makes reliable comparison of findings extremely challenging.10,12 Therefore, systematic and agnostic approaches are needed to dissect strong and consistent signals from selective reporting.10
Large-scale systematic evaluation offers several advantages over studies on single-candidate exposures. First, the association of exposures that have previously been implicated in hypothesis-driven research (ie, the candidate-exposure approach) can be confirmed. An exposome-wide approach limits sources of bias and decreases the risk of false-positive findings.13 Second, large-scale systematic investigation may identify novel correlates that have not been considered thus far. Similar to genome-wide association studies (GWAS), researchers have conducted exposome-wide studies of several phenotypes, such as behavioral problems in children,14 HIV,15 and diabetes.16 Third, mendelian randomization (MR) may help triangulate findings and estimate associations with target variables.17 We conducted what is to our knowledge the first systematic and agnostic exposome-wide analysis to identify correlates of psychotic experiences and sequentially applied genetically informed approaches to probe potential associations.
Data were retrieved from the UK Biobank (UKB), a population-based cohort study that included approximately 500 000 participants from the United Kingdom.18 All participants provided written consent, and ethical approval was given by the National Research Ethics Service Committee North West Multi-Centre Haydock (committee reference: 11/NW/0382).19 The current study (UKB project number: 55392) analyzed participants with complete data on the mental health questionnaire19 that assessed psychotic experiences (N = 155 247). The study followed the Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization (STROBE-MR) reporting guideline. Guided by previous reports,20,21 a binary variable of psychotic experiences (n = 7803) was defined as an endorsement of any of the following 4 lifetime experiences: visual hallucination, auditory hallucination, reference delusion, and persecutory delusion (UKB field IDs f20471, f20463, f20474, f20468). After quality control and preprocessing steps (eMethods and eTables 1-3 in Supplements 1, 2, 3, and 4), the final data set included 247 variables (eTable 4 in Supplement 1). Figure 1 provides an overview of the analytical pipeline.
Analyses were performed from January 1 to June 1, 2021, using R version 4.0.4 (R Foundation). There were 3 sequential analytical steps (Figure 1). Guided by previous exposome-wide studies,15,22-24 we split the data into 2 equal-sized discovery and replication data sets by selecting random samples of participants matched in the frequency of psychotic experiences. To conduct the exposome-wide association study (XWAS), logistic regression analyses with psychotic experiences as the outcome were conducted in the discovery and replication data sets. Variables associated with psychotic experiences (threshold for significance, Bonferroni-corrected P < 2.02 × 10−4) in both discovery and replication data sets were tested in a mutually adjusted multivariable model using complete data (n = 57 702). All analyses were adjusted for age and sex. Variables associated with psychotic experiences in the final multivariable model (threshold for significance, P < .05) were further analyzed using genetically informed approaches to probe potential associations.25,26 A Bonferroni-corrected significance threshold (.0014) was subsequently applied for genetic analyses based on the multivariable results. Psychotic experiences GWAS summary statistics from the UKB were used to estimate single-nucleotide variant (SNV)–based heritability and genetic overlap with psychotic experiences. One-sample bidirectional MR analyses were conducted, as detailed in the eMethods in Supplement 1. The significant associations identified by the 1-sample MR analyses were further analyzed using sensitivity tests, including 1-sample MR analyses controlling for potential confounders (ie, variables that were significantly associated with allele scores of the variables that were significant in the initial 1-sample MR analyses) and collider-correction (CC) 2-sample MR using individual level data from the UKB to apply 3 models (inverse variance weighted [IVW], MR-Egger, and least absolute deviation [LAD] regression).27,28 Additionally, statistically significant variables identified in the initial 1-sample MR analyses were tested in 2-sample MR models (eMethods in Supplement 1). For the 2-sample MR, we used GWAS data from an independent adolescent cohort.29 To our knowledge, this adolescent cohort provides the only available GWAS data of psychotic experiences independent of UKB samples. However, the sample size was relatively small for GWAS (N = 8665; minimum P = 1.32 × 10−6), possibly inflating the risk of type II error. Therefore, as schizophrenia may be considered the severe end of the psychosis spectrum, we also applied 2-sample bidirectional MR using schizophrenia GWAS data30 with the IVW fixed-effect model. We then conducted sensitivity analyses using weighted median (testing associations when up to 50% of SNVs are invalid instruments), MR-Egger (testing associations when all genetic variants are invalid), generalized summary-databased MR (GSMR), and pleiotropy residual sum and outlier (PRESSO). We additionally applied a multivariable MR model to test statistically significant variables identified in the 1-sample forward MR analyses.
Of the 155 247 individuals included in this study, 87 896 (57%) were female, and the mean (SD) age was 55.94 (7.74) years. Of 162 variables that were associated with psychotic experiences in the discovery data set, 148 (91%) were replicated (eTable 5 and eFigures 1 and 2 in Supplement 1). Figure 2 shows the odds ratios (ORs) and 95% CIs of 148 variables under 13 categories in the whole data set. The multivariable analysis of the 148 replicated variables revealed that 36 (24%) were associated with psychotic experiences (eTable 6 in Supplement 1). The correlation matrix of these 36 variables is provided in eFigure 3 in Supplement 1. The Table reports the ORs and 95% CIs of the 36 variables derived from the discovery, replication, and mutually adjusted multivariable analysis. Compared with the XWAS, the associations of the 5 following variables with psychotic experiences were in the opposite direction (ie, the so-called Janus effect) in the multivariable analysis: frequency of unenthusiasm or disinterest in last 2 weeks; nitrogen dioxide air pollution, annual average 2007; number of operations, self-reported; recent feelings of inadequacy; and worrying too long after embarrassment.
Estimating Heritability and Genetic Overlap With Psychotic Experiences
Figure 3 shows the SNV-based heritability and genetic overlap of the 36 variables with psychotic experiences (UKB and adolescent cohort), as detailed in eTable 7 in Supplement 1. The SNV-based heritability of these 36 variables ranged from 0.016 to 0.141 (Figure 3A). Twenty-eight variables were genetically correlated with psychotic experiences in the UKB (Figure 3B). The top hit was chest pain or discomfort (rg, 0.808; 95% CI, 0.615-1.001; P = 2.5 × 10−16). The following 3 variables showed Janus effects, with a genetic correlation in the opposite direction of the XWAS: receipt of attendance, disability, or mobility allowance; major dietary changes in the last 5 years; and regular vitamin and mineral supplement intake. For the analysis using the adolescent cohort, we only reported the genetic covariance (Figure 3C) and not the genetic correlation, as the SNV-based heritability was out of bounds (ie, negative SNV-based heritability). The top hit was feeling hated by family member as a child (genetic covariance, 0.026; 95% CI, 0.011-0.041). Six variables showed Janus effects compared with the XWAS: major dietary changes in the last 5 years; worrying too long after embarrassment; sexual interference by partner or former partner without consent as an adult; nitrogen dioxide air pollution, annual average 2007; receipt of attendance, disability, or mobility allowance; and regular vitamin and mineral supplement intake.
The 1-Sample Bidirectional MR Analyses
Figure 4 shows the 1-sample bidirectional MR analyses results (eTables 8 and 9 in Supplement 1). Among the 130 363 unrelated participants of European ancestry in the UKB, the allele scores explained fractional variance of the 36 variables ranging from 0.04% to 7.85%. The concordance between the XWAS and the 1-sample MR is shown in eFigure 4 in Supplement 1. The 1-sample forward MR analyses confirmed associations with ever having experienced sexual assault (OR, 1.32; 95% CI, 1.14-1.52; P = 2.67 × 10−4), ever having experienced a physically violent crime (OR, 1.25; 95% CI, 1.11-1.41; P = 3.28 × 10−4), and risk-taking behavior (OR, 1.21; 95% CI, 1.08-1.35; P = 1.34 × 10−3). The allele scores for these 3 variables explained 0.03% to 0.23% variance of the corresponding variable. F statistics ranged from 21.53 to 181.84, indicating that the results did not suffer from a weak-instrument bias.
The 1-sample reverse MR analyses were conducted using 2 instruments: 1 SNV significantly associated with psychotic experiences in our GWAS in the UKB (rs11792873) and 4 SNVs from a previous study (eTable 10 in Supplement 1). The rs11792873 explained 0.03% variance of psychotic experiences, with an F statistic of 27.34. The 1-sample reverse MR analyses revealed an association with ever having experienced a physically violent crime (OR, 1.17; 95% CI, 1.11-1.24; P = 2.72 × 10−9) and cannabis use (OR, 1.16; 95% CI, 1.10-1.22; P = 3.96 × 10−9) (eTable 11 in Supplement 1). We also calculated an instrument based on increasing psychotic experiences risk allele scores using 4 SNVs from a previous study.20 The increasing psychotic experience risk allele scores explained 0.14% variance of psychotic experiences, with an F statistic of 19.26. We validated the abovementioned association with cannabis use (OR, 1.11; 95% CI, 1.06-1.15; P = 2.64 × 10−6) and ever having experienced a physically violent crime (OR, 1.08; 95% CI, 1.04-1.13; P = 3.92 × 10−4) (eTable 12 in Supplement 1). Additionally, we detected an association with worrying too long after embarrassment (OR, 1.06; 95% CI, 1.03-1.10; P = 3.96 × 10−4).
Sensitivity Analyses for 1-Sample MR Analyses
The allele scores of ever having experienced sexual assault, ever having experienced physically violent crime, and risk-taking behavior were correlated with 5, 1, and 14 confounders, respectively (eTable 13 in Supplement 1). The 1-sample forward MR analyses adjusted for these potential confounders confirmed the association with ever having experienced a physically violent crime and ever having experienced sexual assault but not with risk-taking behavior (eTable 14 in Supplement 1). We also validated the forward association with risk-taking behavior in CC-IVW and CC-LAD. However, taking the horizontal pleiotropy effects into account using CC-MR-Egger, the association between risk-taking behavior and psychotic experiences was no longer statistically significant. The I2 statistic of CC-MR-Egger was 99.3%, which validated the suitability of the instruments in MR-Egger and confirmed the absence of substantial bias in the association estimates due to uncertainty in the genetic associations. The associations with experiencing physically violent crime and ever having experienced sexual assault could not be tested, as there were not enough independent SNVs (n = 1 at P < 10−6) to calculate instruments. The reverse associations with having experienced physically violent crime, cannabis use, and worrying too long after embarrassment were confirmed with the CC-MR-Egger and CC-LAD regression models. Of the associations identified in the 1-sample MR analyses, the 2-sample forward MR analyses using schizophrenia GWAS data30 confirmed the associations with having experienced sexual assault and the pleiotropy of risk-taking behavior, while the 2-sample reverse MR analyses using schizophrenia GWAS data30 confirmed the reverse associations with having experienced physically violent crime and cannabis use without pleiotropy, as well as the reverse association with worrying too long after embarrassment with pleiotropy (eMethods, eTables 15 to 20 and eFigures 5 to 8 in Supplements 1 and 5). In the multivariable IVW model, risk-taking behavior and having experienced sexual assault showed significant associations, while neither pleiotropy nor other associations were detected in the multivariable MR-Egger model (eTable 21 in Supplement 1). The consistency of the findings across different MR methods is demonstrated in eFigures 9 and 10 in Supplement 1.
The synopsis of the results from each main analytical step are provided in eTable 22 in Supplement 1. The contingency of the 5 variables identified in the MR analyses is provided in eTable 23 in Supplement 1. The presence of all 5 correlates was associated with increased odds of psychotic experiences (OR, 10.63; 95% CI, 8.27-13.65; P = 1.2 × 10−114).
This cohort study, to our knowledge constituting the largest systematic investigation of the nongenetic correlates of psychotic experiences, consisted of several sequential analytical steps. Exposome-wide analyses yielded 148 correlates. In line with the literature, environmental exposures, such as traumatic experiences (sexual assault, physical violence, partner abuse, and serious life-threatening event);2,31,32 hearing difficulties;2,32 neighborhood, social, and economic deprivation;33 cannabis use;32,34 multidimensional psychopathology domains;35,36 proxies of poor mental health outcome (disability allowance, self-harm, and suicidal ideation);37-39 and physical complaints (chest pain or discomfort or fall during the last year)40 were among the top correlates. Psychotic experience was also associated with relatively unexplored factors, including major dietary changes in the last 5 years, driving faster than the speed limit, hot drink temperature, playing computer games,41 regular vitamin and mineral supplement intake, alkaline phosphatase,42 and nitrogen dioxide air pollution.5,6 Of 36 variables that were significantly associated with psychotic experiences in the multivariable analysis, 28 had significant genetic overlap with psychotic experiences. MR analyses revealed the potential forward association with having experienced sexual assault and pleiotropy of risk-taking behavior and reverse associations with having experienced physically violent crime, cannabis use, and worrying too long after embarrassment.
The forward MR analyses showed an association with having experienced sexual assault, which is in accordance with converging evidence suggesting that psychosis is associated with traumatic events and stress-related mechanisms.31,43 Sexual assault was 1 of the top associations with the largest odds for psychotic experiences in the World Mental Health Survey.44 Although the 1-sample MR analysis suggested an association between experiencing physically violent crime and psychotic experiences, this association could not be confirmed in the 2-sample MR analyses. Our analyses further indicated pleiotropy of risk-taking behavior. Risk-taking behavior is associated with various personality traits and mental disorders, such as schizophrenia, posttraumatic stress disorder, ADHD, and bipolar disorder.41,45-47 Genetic overlap of risk-taking behavior with psychiatric diagnoses, behavioral patterns (smoking, alcohol consumption, and cannabis use), body mass index, and IQ has also been found.48,49 Recent evidence suggests that the path from genetic predisposition for risk-taking behavior to schizophrenia might be through environmental factors, such as immigration, urbanicity, or drug use.41 In accordance, we detected 14 possible confounders and, controlling for these, uncovered the pleiotropy of risk-taking behavior.
The reverse MR analyses showed associations between psychotic experiences and having experienced physically violent crime, worrying too long after embarrassment, and cannabis use. The findings support research showing that individuals with mental health problems, particularly psychosis, more frequently experience crimes and that this experience may impact patient trajectories.50 These findings highlight the need for population-wide interventions that decrease violence against vulnerable individuals with mental health problems. The finding on worrying too long after embarrassment might be explained by the association of paranoia with rumination and affective regulation.51 Furthermore, our analyses detected a reverse association between psychotic experiences and cannabis use. These results are in agreement with previous MR studies showing a reverse association between schizophrenia risk and cannabis use.52,53 There is also evidence that genetic liability to schizophrenia is associated with cannabis use.54 However, these results contrast with findings showing that cannabis use is associated with an increase in risk of psychosis in a forward manner.55-59 There is an active debate on whether a bidirectional association between cannabis use and risk of psychosis may exist.52,53,59,60 Longitudinal cohort studies (particularly within-individual designs),56 genetically informative approaches,61 and experimental models62 are crucial to understanding the association between psychosis and cannabis use.
Our findings provide support to previous UKB reports showing that polygenic risk score for schizophrenia was associated with several parameters, including risk-taking behavior and psychiatric phenotypes.41 In accordance with a previous UKB finding20 that showed positive genetic correlations between psychotic experiences and mental disorders, our findings suggest a shared genetic etiology between psychotic experiences and behavioral phenotypes (eg, ever contemplated self-harm; ever had prolonged feelings of sadness or depression; and ever saw a psychiatrist for nerves, anxiety, tension, or depression). Furthermore, we replicated recent UKB findings showing no statistically significant genetic correlation between cannabis use and individual psychotic experience items.63 This is in contrast to several studies suggesting genetic correlation between substance use (eg, smoking, drinking, and cannabis use52,54,61,64-66) and psychiatric disorders, including schizophrenia. Other exposures that were previously found to be genetically correlated with schizophrenia in the UKB, such as population density67,68 and dietary intake,69 either failed the quality-control steps or did not reach significance in the XWAS. We also detected Janus effects for several variables across different analytical steps. This finding illustrates how variable selections and analytical modalities may impact study results.10,12 In accordance with previous studies in the UKB21,70,71 and other large cohorts,32,72,73 investigating gene-environment and environment-environment interactions may additionally help explain the variance in psychotic experiences in the UKB.
Our systematic approach aimed to overcome biases (eg, selective reporting and data dredging), but it was not without limitations. The sequential replication procedure and stringent multiple-testing correction might have led to type II errors. Contrarily, statistically significant but trivial effects are also likely to emerge in large data analyses. The universally applied data preprocessing steps aim to eliminate confirmation bias and a posteriori decision-making. However, some relevant correlates might have been omitted because of missingness or collinearity. Also, these preprocessing steps might have introduced uninformed categorizations. Although we identified several potential associations in MR analyses, the lack of comparable GWAS data (only available data: adolescent psychotic experiences or schizophrenia), lack of power in the adolescent cohort, and violation of assumptions (eg, weak instruments or pleiotropy effects) posed a challenge for the 2-sample MR analyses. Especially, the associations of psychotic experiences with having experienced sexual assault and having experienced physically violent crime need further validation, as the instruments for these analyses were each based on a single SNV, thereby decreasing statistical power. Furthermore, genetic findings may be biased by a winner’s curse for instrument selection, given that most instruments were calculated based on the discovery UKB results rather than an independent data set.74
The findings in this exposome-wide study revealed associations of psychotic experiences with both well-studied and unexplored parameters, some of which were correlated and showed Janus effects. MR analyses revealed an association with having experienced sexual assault and pleiotropy of risk-taking behavior and a reverse association with having experienced physically violent crime, cannabis use, and worrying too long after embarrassment. The findings underline the need for systematic exposome-wide analyses and triangulation of evidence with genetically informed approaches to probe potential causality in the era of big data. To guide public health policies and implementation, future studies aiming for mechanistic understanding are needed.
Accepted for Publication: May 10, 2022.
Published Online: July 20, 2022. doi:10.1001/jamapsychiatry.2022.1655
Corresponding Author: Sinan Guloksuz, MD, PhD, Department of Psychiatry and Neuropsychology, School for Mental Health and Neuroscience, Maastricht University Medical Center, Vijverdalseweg 1, PO Box 616, 6200 MD, Maastricht, the Netherlands (sinan.guloksuz@maastrichtuniversity.nl).
Author Contributions: Drs Lin and Guloksuz had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Lin and Pries shared contribution.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: Lin, Pries, Sarac, Rutten, Luykx, Guloksuz.
Drafting of the manuscript: Lin, Pries, Sarac, Luykx, Guloksuz.
Critical revision of the manuscript for important intellectual content: Lin, Pries, van Os, Rutten, Luykx, Guloksuz.
Statistical analysis: Lin, Pries, Sarac, Guloksuz.
Obtained funding: Rutten, Luykx, Guloksuz.
Administrative, technical, or material support: Luykx, Guloksuz.
Supervision: van Os, Luykx, Guloksuz.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Pries is supported by the Kootstra Talent Fellowship of Maastricht University. Drs van Os and Guloksuz are supported by the Ophelia research project, ZonMw grant 636340001. Dr Rutten was funded by a Vidi award (91718336) from the Netherlands Scientific Organisation.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
1.Belbasis
L, Köhler
CA, Stefanis
N,
et al. Risk factors and peripheral biomarkers for schizophrenia spectrum disorders: an umbrella review of meta-analyses.
Acta Psychiatr Scand. 2018;137(2):88-97. doi:
10.1111/acps.12847PubMedGoogle ScholarCrossref 3.Firth
J, Solmi
M, Wootton
RE,
et al. A meta-review of “lifestyle psychiatry”: the role of exercise, smoking, diet and sleep in the prevention and treatment of mental disorders.
World Psychiatry. 2020;19(3):360-380. doi:
10.1002/wps.20773PubMedGoogle ScholarCrossref 8.Botturi
A, Ciappolino
V, Delvecchio
G, Boscutti
A, Viscardi
B, Brambilla
P. The role and the effect of magnesium in mental disorders: a systematic review.
Nutrients. 2020;12(6):1661. doi:
10.3390/nu12061661PubMedGoogle ScholarCrossref 10.Guloksuz
S, Rutten
BPF, Pries
L-K,
et al; European Network of National Schizophrenia Networks Studying Gene-Environment Interactions Work Package 6 (EU-GEI WP6) Group. The complexities of evaluating the exposome in psychiatry: a data-driven illustration of challenges and some propositions for amendments.
Schizophr Bull. 2018;44(6):1175-1179. doi:
10.1093/schbul/sby118PubMedGoogle ScholarCrossref 16.He
Y, Lakhani
CM, Rasooly
D, Manrai
AK, Tzoulaki
I, Patel
CJ. Comparisons of polyexposure, polygenic, and clinical risk scores in risk prediction of type 2 diabetes.
Diabetes Care. 2021;44(4):935-943. doi:
10.2337/dc20-2049PubMedGoogle ScholarCrossref 17.Choi
KW, Stein
MB, Nishimi
KM,
et al; 23andMe Research Team; Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium. An exposure-wide and mendelian randomization approach to identifying modifiable factors for the prevention of depression.
Am J Psychiatry. 2020;177(10):944-954. doi:
10.1176/appi.ajp.2020.19111158PubMedGoogle ScholarCrossref 19.Davis
KAS, Coleman
JRI, Adams
M,
et al. Mental health in UK Biobank—development, implementation and results from an online questionnaire completed by 157 366 participants: a reanalysis.
BJPsych Open. 2020;6(2):e18. doi:
10.1192/bjo.2019.100PubMedGoogle ScholarCrossref 21.García-González
J, Ramírez
J, Howard
DM, Brennan
CH, Munroe
PB, Keers
R. The effects of polygenic risk for psychiatric disorders and smoking behaviour on psychotic experiences in UK Biobank.
Transl Psychiatry. 2020;10(1):330. doi:
10.1038/s41398-020-01009-8PubMedGoogle ScholarCrossref 23.Patel
CJ, Cullen
MR, Ioannidis
JP, Butte
AJ. Systematic evaluation of environmental factors: persistent pollutants and nutrients correlated with serum lipid levels.
Int J Epidemiol. 2012;41(3):828-843. doi:
10.1093/ije/dys003PubMedGoogle ScholarCrossref 25.Lin
BD, Alkema
A, Peters
T,
et al. Assessing causal links between metabolic traits, inflammation and schizophrenia: a univariable and multivariable, bidirectional mendelian-randomization study.
Int J Epidemiol. 2019;48(5):1505-1514. doi:
10.1093/ije/dyz176PubMedGoogle ScholarCrossref 26.Bulik-Sullivan
BK, Loh
PR, Finucane
HK,
et al; Schizophrenia Working Group of the Psychiatric Genomics Consortium. LD score regression distinguishes confounding from polygenicity in genome-wide association studies.
Nat Genet. 2015;47(3):291-295. doi:
10.1038/ng.3211PubMedGoogle ScholarCrossref 27.Liu
J, Richmond
RC, Bowden
J,
et al. Assessing the causal role of sleep traits on glycated haemoglobin: a mendelian randomization study.
medRxiv. Posted December 20, 2020. doi:
10.1101/2020.12.18.20224733Google Scholar 29.Pain
O, Dudbridge
F, Cardno
AG,
et al. Genome-wide analysis of adolescent psychotic-like experiences shows genetic overlap with psychiatric disorders.
Am J Med Genet B Neuropsychiatr Genet. 2018;177(4):416-425. doi:
10.1002/ajmg.b.32630PubMedGoogle ScholarCrossref 30.Pardiñas
AF, Holmans
P, Pocklington
AJ,
et al; GERAD1 Consortium; CRESTAR Consortium. Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under strong background selection.
Nat Genet. 2018;50(3):381-389. doi:
10.1038/s41588-018-0059-2PubMedGoogle ScholarCrossref 31.Varese
F, Smeets
F, Drukker
M,
et al. Childhood adversities increase the risk of psychosis: a meta-analysis of patient-control, prospective- and cross-sectional cohort studies.
Schizophr Bull. 2012;38(4):661-671. doi:
10.1093/schbul/sbs050PubMedGoogle ScholarCrossref 32.Guloksuz
S, Pries
LK, Delespaul
P,
et al; Genetic Risk and Outcome of Psychosis (GROUP) investigators. Examining the independent and joint effects of molecular genetic liability and environmental exposures in schizophrenia: results from the EUGEI study.
World Psychiatry. 2019;18(2):173-182. doi:
10.1002/wps.20629PubMedGoogle ScholarCrossref 35.Pries
LK, Guloksuz
S, Ten Have
M,
et al. Evidence that environmental and familial risks for psychosis additively impact a multidimensional subthreshold psychosis syndrome.
Schizophr Bull. 2018;44(4):710-719. doi:
10.1093/schbul/sby051PubMedGoogle ScholarCrossref 37.Topor
A, Stefansson
CG, Denhov
A, Bülow
P, Andersson
G. Recovery and economy; salary and allowances: a 10-year follow-up of income for persons diagnosed with first-time psychosis.
Soc Psychiatry Psychiatr Epidemiol. 2019;54(8):919-926. doi:
10.1007/s00127-019-01655-4PubMedGoogle ScholarCrossref 39.Bromet
EJ, Nock
MK, Saha
S,
et al; World Health Organization World Mental Health Survey Collaborators. Association between psychotic experiences and subsequent suicidal thoughts and behaviors: a cross-national analysis from the World Health Organization World Mental Health Surveys.
JAMA Psychiatry. 2017;74(11):1136-1144. doi:
10.1001/jamapsychiatry.2017.2647PubMedGoogle ScholarCrossref 46.Ramírez-Martín
A, Ramos-Martín
J, Mayoral-Cleries
F, Moreno-Küstner
B, Guzman-Parra
J. Impulsivity, decision-making and risk-taking behaviour in bipolar disorder: a systematic review and meta-analysis.
Psychol Med. 2020;50(13):2141-2153. doi:
10.1017/S0033291720003086PubMedGoogle ScholarCrossref 53.Pasman
JA, Verweij
KJH, Gerring
Z,
et al; 23andMe Research Team; Substance Use Disorders Working Group of the Psychiatric Genomics Consortium; International Cannabis Consortium. GWAS of lifetime cannabis use reveals new risk loci, genetic overlap with psychiatric traits, and a causal influence of schizophrenia.
Nat Neurosci. 2018;21(9):1161-1170. doi:
10.1038/s41593-018-0206-1PubMedGoogle ScholarCrossref 56.van Os
J, Pries
L-K, Ten Have
M,
et al. Schizophrenia and the environment: within-person analyses may be required to yield evidence of unconfounded and causal association-the example of cannabis and psychosis.
Schizophr Bull. 2021;47(3):594-603. doi:
10.1093/schbul/sbab019PubMedGoogle ScholarCrossref 57.Kuepper
R, van Os
J, Lieb
R, Wittchen
H-U, Höfler
M, Henquet
C. Continued cannabis use and risk of incidence and persistence of psychotic symptoms: 10 year follow-up cohort study.
BMJ. 2011;342:d738. doi:
10.1136/bmj.d738PubMedGoogle ScholarCrossref 64.Johnson
EC, Demontis
D, Thorgeirsson
TE,
et al; Psychiatric Genomics Consortium Substance Use Disorders Workgroup. A large-scale genome-wide association study meta-analysis of cannabis use disorder.
Lancet Psychiatry. 2020;7(12):1032-1045. doi:
10.1016/S2215-0366(20)30339-4PubMedGoogle ScholarCrossref 71.Coleman
JRI, Peyrot
WJ, Purves
KL,
et al; Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium. Genome-wide gene-environment analyses of major depressive disorder and reported lifetime traumatic experiences in UK Biobank.
Mol Psychiatry. 2020;25(7):1430-1446. doi:
10.1038/s41380-019-0546-6PubMedGoogle ScholarCrossref 73.Pries
LK, Dal Ferro
GA, van Os
J,
et al; Genetic Risk and Outcome of Psychosis (GROUP) investigators. Examining the independent and joint effects of genomic and exposomic liabilities for schizophrenia across the psychosis spectrum.
Epidemiol Psychiatr Sci. 2020;29:e182. doi:
10.1017/S2045796020000943PubMedGoogle ScholarCrossref 74.Sadreev
II, Elsworth
BL, Mitchell
RE,
et al. Navigating sample overlap, winner’s curse and weak instrument bias in mendelian randomization studies using the UK Biobank.
medRxiv. Posted July 1, 2021. doi:
10.1101/2021.06.28.21259622Google Scholar