Domains are designated by the neuropsychological test used. Interaction of age by group is depicted on the graphs. The Wechsler Test of Adult Reading assessed verbal knowledge; the Spatial Delayed Response Task, working memory; the Visual Object Learning Task, visual memory; and the digit symbol coding task, processing speed.
Data are depicted before and after adjusting for confounders. The Wechsler Test of Adult Reading (WTAR) assessed verbal knowledge; the Spatial Delayed Response Task (SDRT), working memory; the Visual Object Learning Task (VOLT), visual memory; and the digit symbol coding task (DSCT), processing speed. Error bars indicate 95% CIs.
Effect sizes are adjusted for all confounders. The Wechsler Test of Adult Reading (WTAR) assessed verbal knowledge; the Spatial Delayed Response Task (SDRT), working memory; the Visual Object Learning Task (VOLT), visual memory; and the digit symbol coding task (DSCT), processing speed. Error bars indicate 95% CIs.
eMethods. Summary of Meta-analysis
eTable 1. Psychosis Screening Questionnaire (PSQ) Items and Responses
eTable 2. Neuropsychological Battery
eTable 3. Correlation Matrix of All Variables
eTable 4. Characteristics of Participants by Psychotic Experiences Group
eTable 5. Group Means and Cohen d Effect Sizes by Psychotic Experience
eTable 6. Cohen d Effect Sizes After Controlling for Confounders Plus Educational Level
eTable 7. Sample Characteristics by Group
eFigure 1. PRISMA Diagram
eFigure 2. Effect Sizes for Each Neuropsychological Test Before (Left) and After (Right) Adjusting for Confounders
eFigure 3. Effect Sizes for Each Neuropsychological Test After Adjusting for Confounders Plus Educational Level
eFigure 4. Local Regression Curves of Neuropsychological Tests on Age-by Group-Interaction
eFigure 5. Hypothesized Model of the Etiology of Neuropsychological Impairments Associated With Psychotic Experiences Throughout Adulthood
Mollon J, David AS, Morgan C, Frissa S, Glahn D, Pilecka I, Hatch SL, Hotopf M, Reichenberg A. Psychotic Experiences and Neuropsychological Functioning in a Population-based Sample. JAMA Psychiatry. 2016;73(2):129-138. doi:10.1001/jamapsychiatry.2015.2551
Psychotic experiences in early life are associated with neuropsychological impairment and the risk for later psychiatric disorders. Psychotic experiences are also prevalent in adults, but neuropsychological investigations spanning adulthood are limited, and confounding factors have not been examined rigorously.
To characterize neuropsychological functioning in adults with psychotic experiences while adjusting for important sociodemographic characteristics and familial factors and investigating the effect of age.
Design, Setting, and Participants
The South East London Community Health (SELCoH) study is a population-based household survey of physical and mental health in individuals 16 years or older conducted from June 1, 2008, to December 31, 2010, in 2 London boroughs. The study included 1698 participants from 1075 households. Data were analyzed from May 6, 2014, to April 22, 2015.
Psychotic experiences measured using the Psychosis Screening Questionnaire.
Main Outcomes and Measures
Neuropsychological functioning measured using tests assessing verbal knowledge (Wechsler Test of Adult Reading), working memory (Spatial Delayed Response Task), memory (Visual Object Learning Task), and processing speed (digit symbol coding task). A composite IQ score of general cognitive ability was calculated.
A total of 1677 participants with a mean (SD) age of 40 (17) years were included in the analysis. Compared with the group without psychotic experiences, the 171 (9.7%) adults with psychotic experiences did not show a statistically significant impairment on mean (SD) measures of IQ (95.25 [16.58] vs 100.45 [14.77]; Cohen d, −0.22; P = .06) or processing speed (40.63 [13.06] vs 42.17 [13.79]; Cohen d, −0.03; P = .73) but were impaired on measures of verbal knowledge (31.36 [15.78] vs 38.83 [12.64]; Cohen d, −0.37; P = .003), working memory (20.97 [4.12] vs 22.51 [3.26]; Cohen d, −0.34; P = .005), and memory (43.80 [8.45] vs 46.53 [7.06]; Cohen d, −0.28; P = .01). Only participants 50 years and older with psychotic experiences showed medium to large impairments in neuropsychological functioning (mean [SD]) on measures of IQ (81.22 [15.97] vs 91.28 [14.31]; Cohen d, −0.70), verbal knowledge (28.31 [13.83] vs 38.51 [11.50]; Cohen d, −0.88), working memory (19.11 [4.77] vs 21.99 [3.42]; Cohen d, −0.82), and memory (39.17 [8.23] vs 44.09 [6.51]; Cohen d, −0.45) after adjusting for socioeconomic status, cannabis use, and common mental disorders. Medium impairments (mean [SD]) on measures of working memory (21.27 [3.64] vs 22.62 [2.97]; Cohen d, −0.45) and memory (44.32 [5.84] vs 46.91 [5.74]; Cohen d, −0.45) were seen in those aged 35 to 49 years and on a measure of verbal knowledge (30.81 [14.17] vs 37.60 [10.48]; Cohen d, −0.62) in those aged 16 to 24 years. First-degree relatives of adults with psychotic experiences showed a small impairment on a measure of verbal knowledge (34.71 [12.10] vs 38.63 [10.97]; Cohen d, −0.36; P = .02), and unrelated cohabitants showed no neuropsychological impairment.
Conclusions and Relevance
The profile of cognitive impairment in adults with psychotic experiences differed from that seen in adults with psychotic disorders, suggesting important differences between subclinical and clinical psychosis.
Schizophrenia has a lifetime prevalence of approximately 1% and combined psychotic disorders of approximately 3%.1,2 A substantial minority of the general population also reports subclinical psychotic experiences, with the World Health Organization World Mental Health Surveys3 reporting a lifetime prevalence of 5.8% worldwide and of 6.8%, 7.2%, and 3.2% in high-, middle-, and low-income countries, respectively. Evidence suggests that subclinical psychotic experiences may lie on a continuum with clinically significant psychotic symptoms and therefore be informative for research into the cause of psychotic illness. First, psychotic experiences and psychotic disorders share risk factors, including low IQ, childhood maltreatment, and stressful life events.4,5 Second, imaging studies report pathophysiologic overlaps between subclinical and clinical psychosis, including hypofrontality, frontotemporal disconnection, and deficits in gray and white matter.6- 8 Finally, psychotic experiences in early life are associated with an increased risk for later psychotic illness9,10 and in adulthood with later hospitalization for a psychotic disorder.11 However, psychotic experiences are also associated with nonpsychotic psychiatric disorders, including anxiety, depression,12,13 and suicidal thoughts and behavior.14,15
Lending support to this hypothesized psychosis continuum are the small neuropsychological impairments seen in people with psychotic experiences. (Table 1 and eMethods and eFigure 1 in the Supplement report results of a meta-analysis of previous population studies of neuropsychological functioning and psychotic experiences).4,5,16- 40 Neuropsychological impairment is a core feature of schizophrenia41,42; it emerges early and remains relatively stable throughout the course of the illness.43 The most severe impairment is reported in processing speed,44,45 but deficits in episodic memory and working memory have been proposed as core features.46 A similar profile of impairment has been reported in people with psychotic experiences,19,20 but most of the studies have focused on child and adolescent samples, despite evidence that psychotic experiences are prevalent across the life course.3,5,47,48 Only 1 study in our meta-analysis investigated the neuropsychological correlates of psychotic experiences across adulthood5 and reported on a single cognitive domain. Moreover, previous studies have not adjusted for key sociodemographic confounders, whose importance is highlighted by the World Mental Health Surveys’ finding of higher prevalences of psychotic experiences in middle- and high-income countries compared with low-income countries.3 Finally, the association between psychotic experiences and neuropsychological functioning may be confounded by shared familial factors.49
In the present study we examined subclinical psychotic experiences and neuropsychological functioning in adults from an ethnically and sociodemographically diverse population. Our study was unique in examining the effects of (1) important confounders, including cannabis use and psychiatric morbidity, (2) age, and (3) confounding by familial factors on the association between psychotic experiences and neuropsychological functioning. We hypothesized that psychotic experiences would be associated with a profile of cognitive impairment similar, yet milder, to that of schizophrenia, characterized by specific deficits in processing speed and memory seen in the context of a generalized deficit.
The South East London Community Health (SELCoH) study is a population-based household survey completed in 2010 in the London boroughs of Lambeth and Southwark. The aim was to provide prevalence estimates of mental and physical health symptoms in an ethnically and socioeconomically diverse, geographically defined, inner-city community. A random sample of households was identified using the Small User Postcode Address File (http://www.poweredbypaf.com/), which has nearly complete coverage of private households in the United Kingdom. Introductory letters were sent to households, which were visited up to 4 times at different times of the day and the week. When contact was made, written informed consent was sought from as many eligible (aged ≥16 years) members of the household as possible; all study participants provided written informed consent. Contact was established with 2070 private households, of which 1075 had at least 1 member interviewed, representing a 51.9% household participation rate. Of 2359 individuals eligible within the participating households, 1698 (72.0%) participated (a detailed description can be found in Hatch et al50). This study was approved by the research ethics committee of King’s College London.
Data were collected using a computer-assisted interview schedule, which was piloted to establish reliability, validity, and feasibility. Data were collected from June 1, 2008, to December 31, 2010. Participants received £15 for participation.
The Psychosis Screening Questionnaire (PSQ)51 was used to measure psychotic experiences. The PSQ is an interviewer-administered questionnaire that assesses psychotic experiences in the preceding year and consists of 5 sections covering hypomania, thought disorder, paranoia, strange experiences, and hallucinations. Items on hypomania were discarded because the focus was on psychosis. Each section has an initial probe, followed by secondary question(s), which establish the quality of psychotic experiences. The PSQ has been validated in 2 national surveys in the United Kingdom.52,53 In this study, those participants who endorsed 1 or more secondary questions at the highest level on the PSQ were compared with those who did not5 (eTable 1 in the Supplement).
Verbal knowledge was assessed using the Wechsler Test of Adult Reading (WTAR),54 working memory with the Spatial Delayed Response Task (SDRT),55 visual memory with the Visual Object Learning Task (VOLT),56 and processing speed with a digit symbol coding task (DSCT)57,58 (eTable 2 in the Supplement). Neuropsychological tests were administered using a computer as in previous schizophrenia and bipolar disorder studies in which impairments of expected effect sizes were shown.59- 63 We calculated a general cognitive ability composite score as the first principal component of a factor analysis using all the neuropsychological tests administered.64 Scores were transformed to an IQ-like score with mean of 100 and SD of 15.
The interview established ethnicity, age, and occupation classified according to the Registrar General65 as professional (I), managerial/technical (II), skilled nonmanual (III-NM), skilled manual (III-M), semiskilled (IV), unskilled (V), and unclassified. Common mental disorders were assessed using the Revised Clinical Interview Schedule (CIS-R),66 which asks about the following 14 symptom domains during the previous week: fatigue, sleep problems, irritability, worry, depression, depressive ideas, anxiety, obsessions, subjective memory and concentration, somatic symptoms, compulsions, phobias, physical health worries, and panic. A score of 18 or more indicates the presence of a common mental disorder. Good reliability and validity of the Revised Clinical Interview Schedule have been reported.66,67 Cannabis use in the past year was also reported. Ethnicity, occupation, cannabis use, and common mental disorders were correlated significantly with psychotic experiences and neuropsychological functioning, but minimally correlated with one another (eTable 3 in the Supplement). Interactions of group (participants with psychotic experiences or control participants) by confounder (ethnicity, occupation, cannabis use, and common mental disorders) with centered variables68 on all tests were not statistically significant except for the group by ethnicity interaction on IQ (P = .03).
We investigated familial factors by dividing the cohabitants of participants with psychotic experiences into 2 groups. The first group included first-degree relatives (eg, biological child, biological sibling); the second, nongenetically related cohabitants (eg, spouse, nonbiological child).
Data were analyzed from May 6, 2014, to April 22, 2015. Analyses were completed in STATA software (version 13; StataCorp). Appropriate survey commands (svy) were used to generate robust SEs. All analyses of SELCoH data accounted for clustering by household inherent in the study design and were weighted for within-household nonresponse.50
We used linear regression (adjusting for ethnicity, occupational class, cannabis use in the past year, and common mental disorders) to test the hypothesis that psychotic experiences would be associated with impairment in general cognitive ability (IQ). Because the profile of cognitive impairment in schizophrenia is characterized by specific deficits in processing speed and memory in the context of a generalized deficit and because different neural systems underlie performance on different neuropsychological tests, secondary analyses examined the association between psychotic experiences and individual neuropsychological test results. We applied a Bonferroni correction to adjust for multiple comparisons in the secondary analyses, yielding a corrected significance level of P < .007 (0.05/7). We computed Cohen d effect sizes using postestimation margins (effect sizes of 0.2, 0.5 and 0.8 represent small, medium, and large effects, respectively69).
To explore the effect of age on the association between psychotic experiences and neuropsychological functioning, local regression curves for age on the interaction of cognition by group were plotted for each test. To test the effect of age formally, we entered an interaction of group (psychotic experiences vs control) by age (continuous) with centered variables68 into the regression model, adjusting for confounders as above. Linear regression analyses were subsequently stratified by age group (16-24, 25-34, 35-49, and ≥50 years), adjusting for confounders as above. Age categories were selected based on local regression curves and to ensure sufficient power. Owing to the exploratory nature of the age-stratified analyses, Cohen d effect sizes with associated 95% CIs only are presented herein.
Confounding by familial factors was first examined by plotting local regression curves for age on the interaction of cognition by group (controls, participants with psychotic experiences, first-degree relatives, and unrelated cohabitants) for each test and subsequently using linear regression, as described above, across the whole sample to ensure sufficient power. We conducted sensitivity analysis using a different cutoff for the presence of psychotic experiences (ie, yes to ≥1 secondary question at any level70,71) using linear regression, as described above.
Ten participants with missing PSQ data, 8 participants reporting a current or a past diagnosis of psychosis, and 3 participants currently taking antipsychotics were excluded from the analyses. The remaining 1677 participants were ethnically diverse (633 [37.7%] not white British), with a mean (SD) age of 40 (17) years (range, 16-90 years). Seven hundred thirty-three participants were male (43.7%).
Based on self-report and medication use, 11 participants (prevalence, 0.7%) had psychotic illness, which is consistent with previous reports of prevalence.1,2 The 1-year weighted prevalence of psychotic experiences, defined as positive responses to all secondary questions in 1 or more categories on the PSQ,5 included 171 participants (9.7%), which is consistent with findings from the World Health Survey.72
The group with psychotic experiences was more likely to be of a minority ethnic background, have a lower occupational status, have used cannabis in the past year, and have a common mental disorder (eTable 4 in the Supplement). Age differences in prevalence were not statistically significant.
Table 2 shows the relationship between psychotic experiences and neuropsychological performance before and after adjusting for confounders. The IQ impairment did not reach statistical significance when adjusting for confounders (β = −3.78; t1261 = −1.91; P = .06; Cohen d = −0.22). However, significant impairments were seen in the WTAR (β = −4.21; t1347 = −3.02; P = .003; Cohen d = −0.37), SDRT (β = −1.15; t1385 = −2.84; P = .005; Cohen d = −0.34), VOLT (β = −1.94; t1455 = −2.59; P = .01; Cohen d = −0.28), and VOLT delay (β = −0.66; t1411 = −2.37; P = .02; Cohen d = −0.24) scores. Impairments in the WTAR and SDRT scores remained statistically significant after correcting for multiple comparisons. No statistically significant impairment in the DSCT score was seen before or after adjusting for confounders.
We performed an exploratory analysis of the association between each psychotic experience (thought insertion, paranoia, strange experiences, and hallucinations) and neuropsychological performance. All experiences were associated with cognitive impairment, but effect sizes varied (eTable 5 in the Supplement). The frequency of individual symptoms did not allow stratification by age.
Previous studies have used samples consisting mostly of children and adolescents or of older adults, but inspection of Table 1 suggests that the association between psychotic experiences and neuropsychological functioning may differ by age. Figure 1 shows an overall age-associated cognitive decline in both groups but also a difference in the severity of neuropsychological impairment associated with psychotic experiences at different ages. We found significant group-by-age interactions for all neuropsychological measures, including IQ (P = .006), WTAR (P = .01), SDRT (P = .03), VOLT (P = .001), VOLT delay (P = .02), and DSCT (P = .01) scores when adjusting for confounders.
After stratifying by age group and adjusting for confounders, group differences in mean (SD) IQ (81.22 [15.97] vs 91.28 [14.31]; Cohen d = −0.70), WTAR (28.31 [13.83] vs 38.51 [11.50]; Cohen d = −0.88), SDRT (19.11 [4.77] vs 21.99 [3.42]; Cohen d = −0.82), VOLT (39.17 [8.23] vs 44.09 [6.51]; Cohen d = −0.45), and VOLT delay (13.09 [2.74] vs 14.34 [2.40]; Cohen d = −0.52) scores were medium to large in participants with psychotic experiences 50 years and older (Figure 2). Medium impairments in SDRT (21.27 [3.64] vs 22.62 [2.97]; Cohen d = −0.45) and VOLT (44.32 [5.84] vs 46.91 [5.74]; Cohen d = −0.45) scores were also found in participants aged 35 to 49 years with psychotic experiences and a medium impairment (30.81 [14.17] vs 37.60 [10.48]; Cohen d = −0.62) and a medium advantage (44.80 [6.61] vs 38.63 [8.43]; Cohen d = 0.74) in WTAR scores in those aged 16 to 24 and 25 to 34 years, respectively. All other impairments in those younger than 50 years were small.
Sensitivity analysis using a different cutoff for the presence of psychotic experiences70,71 produced similar results (eFigure 2 in the Supplement). Adjusting additionally for educational level (eTable 6 and eFigure 3 in the Supplement) also produced similar results.
Characteristics of each group are shown in eTable 7 in the Supplement. Exploratory analyses (eFigure 4 in the Supplement) suggested that WTAR performance was the most and VOLT performance was the least familial. First-degree relatives of probands were impaired on the WTAR score (β = −3.93; t1347 = −2.33; P = .02; Cohen d = −0.36) after adjusting for confounders (Figure 3). Nongenetically related cohabitants did not show statistically significant neuropsychological deficits.
This study provides evidence that subclinical psychotic experiences are associated with mild neuropsychological impairment in adults. The magnitude of impairment in specific domains suggests impaired verbal and memory functions but spared processing speed. Only older adults with psychotic experiences showed medium to large impairments in working memory and memory when adjusting for sociodemographic factors, psychiatric morbidity, and cannabis use. First-degree relatives of probands also had a significant verbal impairment, but not a significant memory impairment. Our findings introduce new knowledge and propose new hypotheses regarding the neuropsychology of psychotic experiences in adults.
Our results are in line with those of the National Survey of Psychiatric Morbidity in Great Britain,5 which reported a small verbal IQ deficit in adults reporting psychotic experiences. Although not directly comparable to those from other previous studies, which have mostly focused on younger or older samples,19,20,23,24 the adjusted effect sizes we report closely approximate the pooled effect sizes in our meta-analytic summary (Table 1). Adjustment for confounders attenuated the processing speed deficit, suggesting that the significant processing speed deficits reported in children and adolescents with psychotic experiences19,20 may be partly confounded. Alternatively, the cause of processing speed deficits associated with psychotic experiences may differ between adolescence and adulthood. Indeed, a recent study that used a dimensional categorization of psychotic experiences also did not find differences in processing speed between adults with low and high levels of psychotic experiences.73
Our findings highlight similarities, but also differences, between neuropsychological dysfunction associated with psychotic experiences vs disorders. Meta-analyses have shown the most severe cognitive impairment in schizophrenia to be in processing speed.44,45 We found only a weak association between psychotic experiences and DSCT performance. The discrepancy between a substantial processing speed impairment in psychotic disorder and a negligible impairment in psychotic experiences suggests that clinical psychosis is associated with increasing abnormality in processing speed.74 Our findings add to a growing body of literature that challenges psychotic experiences as a subclinical phenotype of psychosis. Identification of factors associated with psychotic experiences that predict transition to psychotic disorder is needed if they are to be useful in the etiologic investigation of psychosis. Neuropsychological impairment, in processing speed specifically, may be one factor. Individuals with psychotic experiences and processing speed deficits may be most at risk for psychosis, and this combination of risk factors warrants further longitudinal study.
This study was the first, to our knowledge, to investigate the effect of age on neuropsychological impairment associated with psychotic experiences in adults. Some studies suggest that these experiences are most prevalent in adolescence75 and old age,76 whereas meta-analyses have not found significant age differences.3,47,48 In our sample, prevalence of psychotic experiences was greatest in the youngest group but remained sizable in the other age groups (eTable 4 in the Supplement). Only older adults showed medium to large deficits in IQ, verbal knowledge, working memory, and memory after adjustment for multiple demographic and psychosocial factors. Medium impairments in working memory and memory were seen in the group aged 35 to 49 years and in verbal knowledge in the group aged 16 to 24 years. These findings highlight the heterogeneity of extended phenotypic expression associated with psychotic experiences throughout adulthood. Etiopathologic pathways to subclinical psychotic experiences and neuropsychological dysfunction may be age dependent (eFigure 5 in the Supplement). In young adults with psychotic experiences, neuropsychological impairment may signal a risk for general psychiatric disorders, possibly owing to psychosocial stress and/or substance use; however, in older adults, neuropsychological impairment may indicate vulnerability to accelerated cognitive aging. A faster trajectory of cognitive decline for patients with Alzheimer disease with concurrent delusions77 and hallucinations78 has been reported. Accelerated shrinkage of prefrontal and hippocampal regions seen in normal aging79,80 may lead to disrupted dopamine release and psychotic phenomena81 but also to more rapid cognitive decline.
By including cohabitants of participants with psychotic experiences, we explored potential mechanisms behind the association between psychotic experiences and cognition. First-degree relatives were significantly impaired on verbal knowledge, whereas unrelated cohabitants showed no impairment. Our findings suggest that a complex interplay of genetic, biological, and psychosocial factors lies behind the association between psychotic experiences and neuropsychological impairment. This pattern of verbal knowledge impairment suggests common genetic and/or family environmental factors. On the other hand, unimpaired memory functions in both groups of cohabitants support biological and/or psychosocial effects of psychotic experiences on memory consistent with a causal effect.
Although we hypothesize that sociodemographic factors, cannabis use, and common mental disorders are confounders, they could also be mediators or moderators. Mediation and confounding are identical statistically but can be distinguished conceptually.82 Changes associated with the hypothalamic-pituitary-adrenal axis may be one area of investigation because the hypothalamic-pituitary-adrenal axis has been associated with all of these factors and with psychosis and cognition.83 Shared genetic factors could be another area of investigation.84,85 Examining interactions between multiple risk factors is important because complex multifactorial traits are likely to result from such interactions, and future studies with large samples are required for such investigations.86
The present study has a number of strengths. It is the first, to our knowledge, to use a large, heterogeneous, representative sample drawn from an urban community to investigate the effect of age and familial factors on the association between psychotic experiences and neuropsychological functioning while adjusting for important confounders. Nevertheless, several methodologic limitations require consideration. First, the 51.9% household participation rate was high, but we were not able to characterize nonresponders on demographic variables and rule out possible bias owing to nonparticipation. However, the sample was representative of the local population on most sociodemographic characteristics.50,71 Second, psychotic experiences are fairly common in old age, presumably owing to neurodegenerative processes,87 and we could not exclude participants with dementia. However, most of the participants in the oldest category were aged 50 to 65 years, when dementia is rare,88 and medium to large impairments in IQ, WTAR, SDRT, and VOLT scores (Cohen d > 0.50 for all) remained when we excluded participants older than 65 years. Third, interpretation of effect sizes generally depends on the assumption of normality, which held true for all tasks except the WTAR, meaning that effect sizes in verbal knowledge may have been underestimated.
Moreover, those participants with poorer neuropsychological functioning may have been more likely to endorse psychotic experiences, and yet the association was not present at all ages or in all domains. We cannot infer from the current cross-sectional data that psychotic experiences lead to deficits in neuropsychological functioning. Another plausible explanation is that psychotic experiences and neuropsychological functioning are manifestations of common underlying processes.89,90 Finally, the timing and history of psychotic experiences cannot be established from the PSQ, meaning that some psychotic experiences may be long-standing. The present data provide valuable insight into potential pathways to adult psychopathologic disorders, but future longitudinal studies that are able to disentangle their temporal sequence and to determine whether these findings also apply to lifetime psychotic experiences are needed.
The profile of cognitive impairment in adults with psychotic experiences differed from that found in psychotic disorders. Our findings highlight the importance of considering age, familial factors, and the psychosocial context in neuropsychological studies of psychotic experiences.
Corresponding Author: Josephine Mollon, MSc, Department of Psychosis Studies, Institute of Psychiatry, Psychology, and Neuroscience, King’s College London, London SE5 8AF, England (email@example.com).
Submitted for Publication: June 25, 2015; final revision received October 14, 2015; accepted October 20, 2015.
Published Online: December 30, 2015. doi:10.1001/jamapsychiatry.2015.2551.
Author Contributions: Ms Mollon had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: David, Morgan, Glahn, Hatch, Hotopf, Reichenberg.
Acquisition, analysis, or interpretation of data: Mollon, David, Morgan, Frissa, Pilecka, Hatch, Hotopf, Reichenberg.
Drafting of the manuscript: Mollon, David, Reichenberg.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Mollon, Glahn, Pilecka, Reichenberg.
Obtained funding: David, Hatch, Hotopf.
Administrative, technical, or material support: David, Morgan, Frissa, Glahn, Hatch, Hotopf.
Study supervision: David, Morgan, Reichenberg.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the Biomedical Research Nucleus data management and informatics facility at South London and the Maudsley NHS (National Health Service) Foundation Trust, which is funded by the National Institute for Health Research (NIHR) Mental Health Biomedical Research Centreat South London, Maudsley NHS Foundation Trust, King’s College London, and a joint infrastructure grant from Guy’s and St Thomas’ Charity and the Maudsley Charity. Ms Mollon is supported by the Medical Research Council. Drs Frissa, Hatch, and Hotopf receive salary support from the NIHR Mental Health Biomedical Research Centre at South London, Maudsley NHS Foundation Trust, and King’s College London.
Role of the Funder/Sponsor: The funding sources 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.
Disclaimer: The views expressed herein are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health.
Additional Contributions: We thank all the study participants. The following additional members of the SELCoH (South East London Community Health) study team assisted: project coordinator Maria Verdecchia, MSc, and project manager Aysha Begum, DClinPsych (both of Department of Psychological Medicine, Institute of Psychiatry, Psychology, and Neuroscience, King’s College London, London, England); coinvestigators Robert Stewart, PhD (Department of Psychological Medicine, Institute of Psychiatry, Psychology, and Neuroscience, King’s College London), and Nicola T. Fear, PhD (King’s Centre for Military Health Research, King’s College London); and research assistants Bwalya Kankulu, MSc, Jennifer Clark, PhD, Billy Gazard, MSc, Robert Medcalf, DClinPsych, Alistair Baile, BSc, Elizabeth Doherty, BSc, Deborah Bekele, MSc, Joshua Buckman, DClinPsych, Grant Henderson, BSc (all of Department of Psychological Medicine, Institute of Psychiatry, Psychology, and Neuroscience, King’s College London), Helena Brundy, BSc (Department of Psychosis Studies, Institute of Psychiatry, Psychology, and Neuroscience, King’s College London), and Benedict Weobong, PhD (Kintampo Health Research Centre, Ghana Health Service, Kintampo, Brong Ahafo Region, Ghana). The research assistants were paid to collect data. No other compensation was given.