The Relationship Between Trauma Exposure and Nonclinical Psychosis: A Systematic Review and Meta-Analysis

Abstract

Background:

Trauma exposure is linked to increased psychosis risk; however, critical questions remain about how trauma exposure relates to psychotic-like experiences (PLEs), including the overall effect and whether trauma type, number of exposures, and methodological/sample characteristics influence the relationship.

Method:

A meta-analysis (following PRISMA guidelines) on the relationship between childhood trauma and PLEs was conducted. Effect size by trauma type (emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, bullying, discrimination) and number, assessment methods for trauma and PLEs, and the sex and race/ethnicity compositions of the sample were recorded.

Results:

92 studies (104 samples; 171,828 individuals) were included. The mean effect size was r=.23 and there was some evidence of publication bias, z=2.039, p=.04. A fail-safe n calculation indicated that 7,881 analyses with null results would need to be added for the effect to be non-significant (p≥.05). Effect sizes by trauma type ranged from r=.16 (sexual abuse) to r=.26 (emotional abuse). Bullying had an effect size of r=.28 and discrimination had an effect size of r=.23. Notably, effect size increased as number of trauma exposures increased.

Discussion:

This meta-analysis highlights the importance of expanding the assessment of trauma, including information about chronicity and about exposures such as bullying and discrimination. Research examining the relationship of these more nuanced measures of trauma exposure and later psychosis risk is critical to inform public health and policy decisions.

Introduction

With a prevalence of 1–2%, schizophrenia is a leading cause of disability worldwide and has a yearly economic burden of greater than US$343 billion in the United States alone (Kadakia et al., 2022). Considerable research has been conducted on the relationship between trauma exposure and psychotic symptoms, with evidence that trauma exposure is more common in individuals diagnosed with a psychotic disorder or meeting criteria for a clinical high risk syndrome compared to those at low risk for developing psychosis (Kline & Schiffman, 2014; Mayo et al., 2017; Peters et al., 2022). In addition, trauma exposure is also correlated with the severity of positive psychotic symptoms (Kline & Schiffman, 2014; Varese et al., 2021), and predicts conversion from clinical high risk to a psychotic disorder (Cannon, 2016; Kraan, 2015; Read, 2014). Research suggests that earlier exposure to trauma interacts with developmentally critical periods to produce distinct neurobiological changes associated with different types of psychopathology, with trauma exposure during developmentally critical periods such as those in childhood and adolescence more likely to result in long-term changes consistent with increased risk for or the presence of psychotic disorders (Gomes et al., 2019).

In addition to full-threshold and sub-threshold psychotic symptoms, the psychosis continuum also includes psychotic-like experiences (PLEs), which are qualitatively similar but not as distressing or impairing and occur in an ~8% of the general population (van Os et al., 2009). For example, someone may think they hear their name being called when no one is present and find this confusing or occasionally think that they may not exist without being preoccupied with the idea. These are similar to psychotic experiences of hallucinations or delusions but do not have the same impact clinically, as they do not lead to lasting distress or functional impairment, meaning they are considered PLEs rather than psychotic symptoms. The proneness-persistence-impairment model of psychosis suggests that genetic factors lead to psychosis-proneness (i.e., likelihood of experiencing transient PLEs) and interact with environmental risk factors (e.g., trauma exposure, cannabis use) to result in the persistence of PLEs and then the development of clinically-relevant distress and impairment (van Os et al., 2009). Notably, risk factors for developing a psychotic disorder play a similar role in risk at the persistence and impairment levels of the continuum (van Os et al., 2009), which strengthens the conceptual and mechanistic continuity across these levels and suggests that studying PLEs can expand upon research conducted in clinical populations to improve individual probabilistic prediction of psychosis risk (Fusar-Poli et al., 2020; Fusar-Poli et al., 2017; Kaymaz, 2012; Linscott, 2013; Zammit et al., 2013). In addition, longitudinal research has demonstrated that endorsement of PLEs in individuals without a psychiatric diagnosis predicts the onset of both psychotic disorders and other psychiatric disorders (Dominguez, 2011; Lindgren et al., 2022; Zammit et al., 2013). Thus, being able to identify and predict risk factors for PLEs not only provides benefit to research on psychosis spectrum disorders by providing a useful, more easily recruited analog, but also serves as a clinically meaningful predictor of subsequent mental health which could provide information about who may be best served by preventative interventions. However, although evidence from individual studies provides compelling evidence that trauma exposure influences the development of PLEs, many questions remain which can only be addressed in a meta-analytic summary, and there has not been a meta-analytic summary conducted with PLEs (i.e., non-clinical psychosis).

One critical question addresses the trauma type, as recent research has explored the distinct psychological and neurological mechanisms of change following trauma exposures high in the threat dimension (i.e., presence of atypical threats to an individual), those high in the deprivation dimension (i.e., absence of expected inputs), and those high in the discrepancy dimension (i.e., social exclusion, lack of belonging) (Sheridan & McLaughlin, 2014; T. Vargas et al., 2022). Exposures in the threat dimension are associated with changes in fear learning, elevated amygdala reactivity to threatening stimuli like angry faces, attentional biases facilitating threat identification, reduced ventromedial prefrontal cortex and hippocampal volumes, and poor hippocampal function related to learning and memory (Sheridan & McLaughlin, 2014). Exposures in the deprivation dimension are associated with reduced cortical thickness and deficits in cognitive and social functions such as language, executive function, spatial navigation, and social cognition (Sheridan & McLaughlin, 2014). Exposures in the discrepancy dimension are associated with reduced social support and increased psychological distress, likely through reductions in oxytocinergic signaling, alterations in amygdala, orbitofrontal cortex, and anterior cingulate cortex which are implicated in processing social threat, rejection, and lack of belonging, and activation of areas related in pain-related processing such as the anterior insula and periaqueductal gray (T. Vargas et al., 2020). Critically, many of these alterations in brain structure and function closely map onto those commonly observed in psychosis (Vargas & Mittal, 2022; T. G. Vargas & Mittal, 2022).

Another key question is the role of cumulative exposure to trauma, including exposure to multiple trauma types and repeated exposures to the same trauma type. Evidence has established that more types of trauma exposure leads to higher risk of developing psychotic symptoms (Croft, 2019; Stanton et al., 2020) and that chronic exposure to trauma, especially during the developmentally critical periods of childhood and adolescence, results in a different constellation of trauma-related symptoms (Ford et al., 2018; Panayi et al., 2022; Rodolfo Rossi et al., 2023). Although there is limited information about the mechanistic link between chronicity of trauma and psychosis proneness, it may be that oxidative stress (i.e., a biological mechanism of aging elevated by chronic stress) (Miller & Sadeh, 2014) results in elevated risk through increased inflammation (Delaney et al., 2019; Fraguas et al., 2019).

Finally, questions remain about whether these relationships are or are not consistent across various research methodologies (e.g., self-report/interview, participant-reported/parent-reported) and across different demographic groups (e.g., male/female, white/non-white). Addressing sample characteristics allows for nuance in considering the most relevant mechanisms and interventions for specific individuals or populations as precision medicine in mental health becomes closer to clinical practice. To be able to apply those conclusions, however, it is necessary to understand whether assessment methods may result in artifacts influencing research findings.

The current study used meta-analytic approaches to address cross-cutting questions about 1) the overall effect size between childhood trauma exposure and PLEs, 2) type of trauma exposure, 3) cumulative trauma exposures, 4) the impact of methodological variations (e.g., how trauma and PLEs were assessed), and 5) the influence of sample characteristics (e.g., sex, race/ethnicity).

Method

Eligibility Criteria.

Studies were screened by the first and second authors and were considered eligible for inclusion if they 1) assessed both trauma exposure during childhood/adolescence and non-clinical PLEs, 2) reported on sample size, 3) reported an effect size of the relationship between trauma exposure and non-clinical PLEs as a correlation coefficient or provided sufficient information to calculate a correlation coefficient, and 4) were available in English. Any analysis which was available and met these criteria were included in the meta-analysis, including peer-reviewed studies and dissertations. The correlation coefficient was chosen as the primary effect size because both number of trauma exposures and experience of PLEs are conceptualized as being continuous variables in which the amount of exposures or PLEs reported is relevant for the current questions. Formulas used to convert effect sizes to correlation coefficients are available in the supplemental materials. Trauma for the purposes of this study was defined in alignment with previous literature on the relationship between trauma and psychosis as a highly distressing event in childhood which was beyond the coping capacity of the individual (e.g., abuse, neglect, having a mentally ill parent; Giannopoulou et al., 2023). Bullying and discrimination were included as potentially traumatic events due to the recent and expanding literature linking exposure to these events and increased risk of psychotic disorders (Giannopoulou et al., 2023) and because they often fit the identified definition of trauma (i.e., highly distressing and beyond the coping capacity of the individual). Non-clinical PLEs for the purposes of this study were defined as self-, parent-, or interviewer-reported experiences which were qualitatively similar to psychosis (e.g., hallucination-like or delusion-like) without producing clinically meaningful distress or impairment. Studies were excluded if the “PLE” measure was capturing degree of distress/impairment (e.g., only rating distressing experiences, summing frequency and distress scores), only included participants above a “clinically relevant” cut-off score, or constituted equivalency to a clinical-high risk syndrome. Included measures of PLEs were typically summing the total number of types of PLE endorsed, summing the frequency at which different PLEs were endorsed, or grouping participants in the sample by presence/absence or high/low numbers/frequency of PLEs. Studies were included if no specific age cut off was reported for the trauma exposure assessment (i.e., if trauma assessment included the childhood/adolescent period but may have also included adulthood exposures) but excluded if participants reported only on traumatic exposures experienced as an adult. Studies which included both clinical and non-clinical psychosis were included if they reported sample size and effect size information on the subsamples; only the non-clinical subsample was included in analysis. Additionally, if studies reported on multiple independent samples within one manuscript, each sample was coded and graphed separately. In the case of the small number of manuscripts which were conducted with the same individuals (e.g., two studies from the same birth cohort), if they were examining different types of trauma and/or PLEs, they were coded as though separate measurements within the same study. If they reported on the same relationships within the same sample, the earliest manuscript was included and others were excluded.

Search and Selection.

A literature search was conducted by the first author on January 31^st, 2025 by entering the terms “psychotic-like experiences” OR “PLEs” OR “non-clinical psychosis” OR “psychosis-like experiences” AND “trauma” OR “adversity” OR “childhood adversity” OR “adverse childhood experience” OR “early life stress” directly in the Psych Info, PubMed, Medline, and EMBASE databases. No criteria about publication date were included in the search (i.e., manuscripts published in any year were included). After this initial search was screened for inclusion criteria, the references of included studies were compiled and searched for any relevant studies which were not already identified.

Data Extraction and Calculation.

Identified studies were screened by title, abstract, and full text to determine eligibility. The first author then coded each study meeting eligibility criteria for sample size, overall effect size (converted into correlation coefficients), number of male and female participants, race/ethnicity composition of the sample, trauma types assessed, and assessment method for both trauma exposures and PLEs. Formulas used to convert from various effect sizes to correlation coefficients can be found in the Supplementary Material. If any of this information was not available, it was recorded as “not reported”. Separate effect sizes were also recorded for the relationship between PLEs and the following variables if available: total abuse (proxy for “threat” dimension), total neglect (proxy for “deprivation” dimension), emotional abuse, physical abuse, sexual abuse, emotional neglect, or physical neglect, and bullying or discrimination (proxy for “discrepancy” dimension). Finally, if studies reported separate effect sizes for different levels of trauma exposure (e.g., 1–2 traumas, 3–5 traumas), the effect size for each group was recorded and a difference score was calculated between the effect size of each group and the next-lowest group (e.g., low-trauma minus no trauma, medium-trauma minus low-trauma). This calculation method was consistent across studies regardless of how each group was defined (i.e., low-trauma defined as 1–2 exposures was subtracted from higher-trauma defined as 3–4 exposures in one study; in another, low-trauma defined as 0–5 exposures was subtracted from higher-trauma defined as 6+ exposures). Summary variables were created for predominant sex of the sample, with 50% male +/− 10% recorded as approximately 50/50, and predominant race/ethnicity of the sample (white/non-white/not reported). Characteristics of each included study are reported in Table 1.

Table 1.

Summary of characteristics of all studies included in the analysis.

Study	Sample Size	Majority Race/Ethnicity	Majority Sex	PLEs –method	PLEs – measure	Trauma - method	Trauma – measure	Trauma types assessed
Adjorlolo et al., 2023	1886	Not reported	Female	Self-report questionnaire	CAPE	Self-report questionnaire	Custom questions	Substance use, head trauma, violence victimization, school-level stress, disorganized neighborhood, family history of mental illness
Anglin et al., 2018	644	Balanced Asian/Black/Hispanic	Female	Self-report questionnaire	PQ (45 positive items)	Self-report questionnaire	EOD	Racial discrimination
Anglin et al., 2015	549	Black	Female	Self-report questionnaire	PQ	Self-report questionnaire	LEC	Natural disaster, fire or explosion, transportation accident, serious accident, exposure to toxic substance, physical assault, assault with a weapon, sexual assault, other unwanted or uncomfortable sexual experience, combat or exposure to a war-zone, captivity, life-threatening illness or injury, severe human suffering, sudden violent death, sudden accidental death, serious injury/harm/death you caused to someone else, other very stressful event or experience
Ayazi et al., 2016	1200	Not reported	50/50	Interview	MINI	Self-report questionnaire	HTQ	Lack of food/clean water, lacked shelter, property looted/confiscated/destroyed, witnessed someone being physically harmed, suffered ill health without access to medical care or medicine, murder or violent death of family member, search, exposed to combat situation or landmine, forced to leave country/hometown, family member kidnapped or taken as a hostage, physically harmed
Baryshnikov et al., 2018	282	Not reported	Female	Self-report questionnaire	CAPE-42	Self-report questionnaire	TADS	Emotional, physical, sexual abuse and emotional and physical neglect
Berry et al., 2018	123	White	Female	Self-report questionnaire	LSHS	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Binbay et al., 2012	4011	Turkish	50/50	Interview	CIDI, SCID-IV	Interview	Interview questions	Death of parents, divorce of parents, separation from parents for at least 3 months
Bolhuis et al., 2018	3984	White	50/50	Self-report questionnaire	YSR (3 items)	Mother interview	Custom questions	Parental divorce/separation, transferring schools, physical or sexual maltreatment, and others
Boyda et al., 2018	302	Not reported	Female	Self-report questionnaire	CAPE	Self-report questionnaire	Questions derived from ACE-IQ	Emotional, physical, sexual abuse, neglect, witnessing interpersonal violence, family environment, relationship with parents
Campbell & Morrison, 2007	373	Not reported	50/50	self-report questionnaire	PS, LSHS-R auditory, BAPS, DES, ambiguous stimuli task	self-report questionnaire	OBVQ (29 victimization questions)	Bullying
Carlyle et al., 2021	2630	Not reported	50/50	Self-report questionnaire	CAPE-15	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Choi, 2017	169	Not reported	50/50	Self-report questionnaire	RC 6 and RC 8 on MMPI-2	Self-report questionnaire	Korean CTQ, IES-R-K	Emotional, physical, sexual abuse and emotional and physical neglect
Cole et al., 2016	200	White	Female	Self-report questionnaire	LSHS-R, PDI	Self-report questionnaire	CATS	Sexual abuse, punishment and physical abuse, neglect and degree of negativity, emotional abuse
Croft et al., 2019	3758	Not reported	50/50	Semi-structured Interview	PLIKSi	Self-report questionnaire, parent report	121 questions from assessment battery	Physical abuse, emotional abuse, bullying, sexual abuse, domestic violence, emotional neglect
Cullen et al., 2014	90	White	50/50	Self-report questionnaire	Antecedent screening questionnaire	Self-report questionnaire	Negative life events questions	Death of someone close, parental separation/divorce, exposure to fire or natural disaster, involved in serious car accident, serious illness requiring hospitalization, parent experienced serious accident or illness, victim of burglary, victim of other crime
de Castro-Catala et al., 2017	808	White	Female	Self-report questionnaire	WSS, CAPE	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
DeRosse et al., 2014	447	White	Female	self-report questionnaire	CAPE	self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
DeVylder et al., 2013	2,434	Asian/Latino	50/50	Semi-structured Interview	WHO-CIDI 3.0	Self-report questionnaire	Nine dichotomous acculturative stress items	Acculturative stress
Dhondt et al., 2022	6039	Not reported	50/50	Self-report questionnaire	7 items of APSS	Parent-report questionnaire	Custom questions	Death of an immediate family member, death of a close friends, parent in prison, drug taking or alcoholism in immediate family, mental disorder in the immediate family, serious illness or injury, stay in foster/residential care, or three or more of the following: divorce, separation of parents, death of a close family member, serious illness or injury of a family member, conflict between parents, moving to a new house, moving to a new country, other unspecified event
Ered & Ellman, 2019	116	White	Female	Semi-structured interview	SIPS	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Ered et al., 2017	454	White	Female	Self-report questionnaire	PQ (45 positive items)	Self-report questionnaire	LEC	Natural disaster, fire or explosion, transportation accident, serious accident, exposure to toxic substance, physical assault, assault with a weapon, sexual assault, other unwanted or uncomfortable sexual experience, combat or exposure to a war-zone, captivity, life-threatening illness or injury, severe human suffering, sudden violent death, sudden accidental death, serious injury/harm/death you caused to someone else, other very stressful event or experience
Fekih-Romdhane et al., 2020	135	Not reported	50/50	Self-report questionnaire	CAPE	Self-report questionnaire	CTQ (28 item version)	Emotional, physical, sexual abuse and emotional and physical neglect
Fekih-Romdhane et al., 2019	482	Not reported	Female	Self-report questionnaire	CAPE	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Freeman & Fowler, 2009	200	White	50/50	Self-report questionnaire	G-PTS Part B, Cardiff anomalous perceptions scale	Self-report questionnaire	Life stressor checklist (only criterion A events)	Non-victimization, victimization, and childhood physical or sexual abuse
Gawęda et al., 2019	649	Not reported	50/50	Self-report questionnaire	PQ-16	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Gawęda, Pionke et al., 2018	690	Not reported	Female	Self-report questionnaire	PQ-16	Self-report questionnaire	TEC	Emotional neglect, emotional abuse, physical abuse, threat to life, sexual harassment, sexual abuse
Gawęda, Prochwicz et al., 2018	653	Not reported	Female	self-report questionnaire	CAPE	Self-report questionnaire	TEC	Emotional neglect, emotional abuse, physical abuse, threat to life, sexual harassment, sexual abuse
Gelner et al., 2024	188	Not reported	Female	Experience sampling method	Questions from previous studies and adapted from the PQ-16	Self-report questionnaire	TEC	Emotional neglect, emotional abuse, physical abuse, threat to life, sexual harassment, sexual abuse, bizarre punishment, intense pain
Gibson, 2018	1673	White	Female	self-report questionnaire	PQ	Self-report questionnaire	CTQ-SF	Emotional, physical, sexual abuse and emotional and physical neglect
Giesbrecht et al., 2007	185	Not reported	Female	self-report questionnaire	DES, STA, CEQ, CFQ	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Goldstone et al., 2011	133	Not reported	50/50	Self-report questionnaire	PDI	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Goldstone et al., 2012	133	Not reported	50/50	Self-report questionnaire	LSHS-R	Self-report questionnaire	ETI-SR	Physical and emotional trauma, sexual abuse
Goodall et al., 2015	283	Not reported	Female	Self-report questionnaire	SQB-B	Self-report questionnaire	ITEM	Life-threatening illness, someone close died, life threatened, physical assault, sexual assault, sexual harassment, war/combat, captive/tortured, caused suffering or death, witnessed suffering or death, natural disaster, human-made disaster, stalked, bullied, emotional abuse, neglect
Gracie et al., 2007	228	White	Female	Self-report and interview	PS, LSHS, SIAPA	Self-report questionnaire	TLEQ, SRS-PTSD	Physical, emotional, sexual abuse, physical and emotional neglect, witnessing family violence, bullying, sexual harassment, being beaten up by a stranger
Grattan et al., 2023	466	White	50/50	Self-report questionnaire	PQ-B	Self-report questionnaire	ACE	Psychological, physical, or sexual abuse, violence against mother, imprisoned family member, living with people who abuse substances, is mentally ill, or suicidal
Green & Webster, 2022	528	White	Female	Self-report questionnaire	CAPE-42	Self-report questionnaire	C-BSA, Chinese CTQ	Bullying, emotional, physical, sexual abuse and emotional and physical neglect
Hayes, 2022	212	White	Female	Self-report questionnaire	CAPE-42	Self-report questionnaire	CATS	Sexual abuse, physical abuse/punishment, negative home environment/neglect, emotional abuse
Heins et al., 2011	476	Not reported	Female	Interview	SIS-R	Self-report questionnaire	CTQ-SF	Emotional abuse, physical abuse, “general” childhood abuse, emotional neglect, physical neglect
Hu et al., 2024	4302	Han	50/50	Self-report questionnaire	CAPE-P8 (Chinese version)	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Huque et al., 2024	698	Asian/Black	Female	Self-report questionnaire	PQ	Self-report questionnaire	EOD	Discrimination
Isaac, 2018	29	Black	Female	Self-report questionnaire	PQ	Self-report questionnaire	LEC	Natural disaster, fire or explosion, transportation accident, serious accident, exposure to toxic substance, physical assault, assault with a weapon, sexual assault, other unwanted or uncomfortable sexual experience, combat or exposure to a war-zone, captivity, life-threatening illness or injury, severe human suffering, sudden violent death, sudden accidental death, serious injury/harm/death you caused to someone else, other very stressful event or experience
Jankowski, 2024	115	White	50/50	Self-report questionnaire	CAPE-42	Self-report questionnaire	LEC	Natural disaster, fire or explosion, transportation accident, serious accident, exposure to toxic substance, physical assault, assault with a weapon, sexual assault, other unwanted or uncomfortable sexual experience, combat or exposure to a war-zone, captivity, life-threatening illness or injury, severe human suffering, sudden violent death, sudden accidental death, serious injury/harm/death you caused to someone else, other very stressful event or experience
Kisely et al., 2022	2,427	White	Female	Self-report questionnaire	PDI	Self-report questionnaire, agency reports	CTQ, agency-reported maltreatment	Emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect
Lataster et al., 2006	1290	Not reported	50/50	Self-report questionnaire	Derived from DISC-C	Self-report questionnaire	Custom questions	Bullying, sexual trauma
Lataster et al., 2011	1,722	Not reported	50/50	Interview, Self-report questionnaire	M-CIDI, SLC-90-R	Interview	MEL	Early life parental loss and separation, terrible experiences in a war, serious physical threats or attacks, rape, sexual abuse, natural catastrophes, serious accidents, imprisonment, hostage experiences, kidnap experiences
Li et al., 2022	3346	Note reported	50/50	Self-report questionnaire	SCL-90-R (Chinese version)	Self-report questionnaire	Questions adopted from the OBVQ	Bullying
Loore, 2007	1129	Not reported	50/50	Self-report questionnaire	Derived from DISC-C	Self-report questionnaire	Custom questions	Bullying, sexual trauma
Lopez, 2020	261	Hispanic	Female	Self-report questionnaire	SPQ, PQ-B, O-LIFE	Self-report questionnaire	EDS	9 discriminatory events
Lu et al., 2020	4360	Not reported	50/50	self-report questionnaire	CAPE-42	self-report questionnaire	CTQ-SF	Emotional, physical, sexual abuse and emotional and physical neglect
Marlowe et al., 2020	298	Not reported	Female	Self-report questionnaire	CAPE-42	Self-report questionnaire	CTQ-SF	Emotional, physical, sexual abuse and emotional and physical neglect
McMahon et al., 2021	973	Not reported	50/50	Self-report questionnaire	APSS	Self-report questionnaire	Custom questions	Trouble with bullies, theft of personal belongings, lower grades than expected, change of school, serious argument with a friend, minor violation of the law, alcohol or drug use by a family member (past 3 months); been physically attacked, having trouble with parents, death in the family (past 12 months)
Mętel et al., 2020	2614	Not reported	Female	Self-report questionnaire	PQ-16	Self-report questions	TEC (3 items), CECA.Q (3 items)	Emotional neglect, emotional abuse, bullying, sexual trauma
Misiak et al., 2023	3000	Not reported	50/50	Self-report questionnaire	Items from Revised Hallucination Scale, R-GPTS, PQ	Self-report questionnaire	Items from TEC and CECA.Q	Emotional, physical, sexual abuse and emotional neglect
Mongan et al., 2019	748	White	50/50	Self-report questionnaire	PQ-16	Self-report questionnaire	ACE-Q	Emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, a parent with an alcohol or drug problem, a parent with a mental illness, witnessing domestic violence, having an incarcerated relative, and parental separation (divorce, death, abandonment)
Morgan et al., 2014	1680	White	50/50	Interviewer-administered questionnaire	PQ-Likert	Self-report questionnaire	Single questions	Separated from partner, loved one died, no money for rent, serious accident, witnessed violence, victim of crime, attacked, injured with weapon, been in war zone
Morrison & Petersen, 2003	64	Not reported	Female	self-report questionnaire	RHS	self-report measure	Custom questions	Bereavement, emotional abuse, bullying, physical assault, sexual assault
Nonweiler et al., 2023	1156	Spanish	Female	Self-report questionnaire	CAPE (20 positive items)	Self-report questionnaire	CTQ-B, CECA.Q	Emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, maternal anitpathy, paternal antipathy, maternal psychological abuse, paternal psychological abuse, parental loss, role reversal and support
O’Hare et al., 2023	26960	Not reported	50/50	Self-report questionnaire	PLEQ-C	Level of CPS involvement	Level of child protection response	Placed in care outside home, substantiated report, unsubstantiated report, non-threshold report
Oh et al., 2014	8990	Hispanic/Black	50/50	Interview	WHO-CIDI 3.0 psychosis screen	Self-report questionnaire	EDS	9 discriminatory events
Perona-Garcelán et al., 2014	83	Not reported	Female	Self-report questionnaire	LSHS-R	Self-report questionnaire	Trauma Questionnaire	Unexpected death of relative/friend, physical abuse, near-drowning experience, serious threats to the life of a family member or close person, car/train/plane accidence, sexual abuse, fire, flood
Pignon et al., 2021	1762	White	Female	Self-report questionnaire	CAPE	Self-report questionnaire	CTQ, discrimination measure, LTE	Emotional, physical, and sexual abuse, emotional and physical neglect, discrimination, major stressors like serious injury or illness (self or close relative), death of parent/child/partner, death of a family member, death of a friend, separation from a partner, loss of job or financial difficulties
Pionke-Ubych et al., 2022	139	Not reported	50/50	Self-report questionnaire, structured clinical interview	PQ-16, CAARMS, IPASE	Self-report questionnaire	CECA.Q	Psychological, physical, sexual trauma
Richardson et al., 2023	166	White	Female	Self-report questionnaire	PQ-B, Paranoia scale	Self-report questionnaire	CSTQ	Childhood sexual trauma
Rossi, Collazzoni et al., 2020	500	Not reported	Female	Self-report questionnaire	iPQ-16	Self-report questionnaire	RFQ	Exposure to harsh parenting, derived from the ACE-q
Rossi, Socci et al., 2020	494	Not reported	Female	Self-report questionnaire	Italian PQ-16	Self-report questions	Custom questions	Earthquake exposure
Rössler et al., 2016	820	Not reported	50/50	Self-report questionnaire	SIAPA, SPQ-B, PARA, 2 subscales from SCL90-R, CEQ	Self-report questionnaire	CTQ (short form)	Emotional, physical, sexual abuse and emotional and physical neglect
Sangimino et al., 2024	107	White	Female	Self-report questionnaire	PQ-16	Self-report questionnaire	BTQ	Physical, emotional, and sexual trauma
Schreier et al., 2009	6437	White	50/50	Semi-structured interview	PLIKSi	Structured interview, parent and teacher report	Bullying and Friendship Interview Schedule, Strength and Difficulties Questionnaire	Peer victimization and bullying
Sengutta et al., 2019	200	Not reported	Female	self-report questionnaire	PQ-16	self-report questionnaire	ACE	Emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, separation from a parent, violence against mother, substance abuse, substance dependence, mental illness of family member, imprisonment of family member
Sheinbaum et al., 2014	546	Not reported	Female	Self-report questionnaire	CAPE, SCL-90-R, SCID-I	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Sheinbaum et al., 2020	715	Not reported	Female	Self-report questionnaire	SPQ, WSS-SF	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Shevlin et al., 2013	4111	Not reported	Female	Self-report questionnaire	PSQ	Self-report questions	Custom questions	Domestic violence, sexual abuse before and after 16
Shevlin et al., 2007	5893	Not reported	50/50	Interview	psychosis screening section of CIDI	Interview	PTSD module of CIDI	Childhood neglect, childhood physical abuse, victim of physical attack/assault, rape, sexual molestation
Smigielski et al., 2021	24	White	Male	self-report questionnaire	PAGE-R, SPQ	self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Soosay et al., 2012	1245	Not reported	50/50	self-report questionnaire	PSQ	self-report questionnaire	HTQ	Flood/earthquake, been without food and water, house burnt down, family member disappeared, combat, been without shelter, been sick and unable to access medical care, been attacked or assaulted, seen family member murdered, family member sick and unable to access medical care, serious injury, seen stranger murdered, torture, been forcibly separated from family, seen death from flood/earthquake, been imprisoned
Startup, 1999	224	Not reported	Female	Self-report questionnaire	O-LIFE	Self-report questionnaire	2 questions about sexual/physical abuse	Sexual abuse, physical abuse
Steenkamp et al., 2021	3437	White	50/50	Self-report questionnaire	YSR (2 items)	Parent interview	Custom Questions	Physical or sexual maltreatment, parental divorce/separation, moving, neighborhood problems, and others
Strachan et al., 2023	528	White	Female	Self-report questionnaire	MUSEQ	Self-report questionnaire	PDS-5	DSM-5 criterion A traumas plus other domestic/family violence (verbal/emotional/financial intimidation, abuse and/or violence within a family/intimate relationship) and any other event that was upsetting/distressing (argument, upsetting interpersonal exchange, embarrassing event, humiliating event, etc)
Sun et al., 2017	6623	Han	50/50	Self-report questionnaire	CAPE	Self-report questionnaire	THQ	Serious accident, severe illness or injury, death of someone close, separation from significant others, suicide of someone close, physical assault or threatening, victim or witness of mugging, attacked by a dog or other animal, witnessing physical violence, family member arrested or in jail, victim or witness of sexual activities
Sun et al., 2021	939	Han	Female	Self-report questionnaire	CAPE-P15	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Tang et al., 2020	5505	Not reported	50/50	Self-report questionnaire	SCL-90-R (Chinese version)	Self-report questionnaire	CTQ (no sexual abuse questions) since the earthquake, ASLEC since the earthquake, earthquake exposure	Emotional and physical abuse and emotional and physical neglect
Toutountzidis et al., 2018	320	White	Female	Self-report questionnaire	FFSI	Self-report questionnaire	3 subscales of ETISR-SF	Physical punishment, emotional abuse, sexual abuse
Vargas & Mittal, 2021	3839	White	50/50	Self-report questionnaire	PQ-BC	Self-report questionnaire	Items from 2 measures related to discrimination	Discrimination
Wang et al., 2022	533	Not reported	Male	Self-report questionnaire	CAPE-P8	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Wang et al., 2023	3234	Han	Female	Self-report questionnaire	CAPE-15	Self-report questionnaire	CTQ-SF	Emotional, physical, sexual abuse and emotional and physical neglect
Wei et al., 2022	1493	Zhuang	50/50	Self-report questionnaire	Chinese CAPE-8	Self-report questionnaire	CTQ-SF	Emotional, physical, sexual abuse and emotional and physical neglect
Wolke et al., 2014	4101	Not reported	50/50	Interview	PLIKSi	Interview	BFIS, SDQ	Bullying
Wong et al., 2021	9873	Not reported	Female	Self-report questionnaire	Custom questions (ideas of reference only)	Self-report questionnaire	Custom questions	Parents absent from home, bullied at school, loss of parent, physical abuse, emotional abuse, sexual abuse, and “others”; related to social unrest, related to COVID-19 pandemic, personal stressful life events
Yamasaki et al., 2016	4277	Not reported	50/50	Parent report	2 items of CBCL	Parent report	Questions adopted from the OBVQ	Bullying
Yang et al., 2024	333	Han	Male	Self-report questionnaire	CAPE-P8	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Zhang et al., 2022	5873	Not reported	50/50	Self-report questionnaire	CAPE-P15	Self-report questionnaire	CTQ	Emotional, physical, sexual abuse and emotional and physical neglect
Zhao et al., 2022	1258	Not reported	Female	Self-report questionnaire	CAPE-P15	Self-report questionnaire	C-BSA, CTQ (Chinese version)	Emotional, physical, sexual abuse, emotional and physical neglect, bullying

The second author independently screened and coded a subset of the identified articles (50%) to assess reliability of data extraction. The two-way mixed effects, consistency, single measurement intraclass correlation was good, ICC(C,1)=0.884, 95% CI [.854, .908].

The Risk Of Bias In Non-randomized Studies – of Exposures (ROBINS-E; Higgins et al., 2024) was used to assess the risk of bias in each included manuscript. Sensitivity analyses were conducted excluding studies which were assessed to have high risk of bias or higher. Results for these analyses can be found in the supplemental materials.

Analytical Strategy.

Results were graphed using the viz_thickforest function of the metaviz package (version 0.3.1) in RStudio (version 2023.12.1+042 for Windows). Samples were then grouped by variables thought to be conceptually meaningful, including trauma type, predominant sex of the sample, predominant race of the sample, and assessment method of trauma and of PLEs (Croft, 2019; Danese & Widom, 2020; Jones et al., 2019, 2021; Olff, 2017; Polanco-Roman, 2016; Stanton et al., 2020; Thomas et al., 2022).

A funnel plot was created using the funnel function of the metafor package (version 4.4–0). A regression test for normality in the funnel plot was conducted using the regtest function of the metafor package.

Transparency and Openness.

We adhered to the PRISMA 2020 guidelines for systematic reviews (Page et al., 2021). We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study. All data, analysis code, research materials, and bias assessment templates are available at https://osf.io/47cxj/overview?view_only=b74759d534304e21a616cfaf1656a1b6. This study was not preregistered.

Results

Question 1: Overall Association Between Trauma Exposure and PLEs.

The literature search identified 2,396 manuscripts to be screened. These manuscripts were screened by title, resulting in 224 potentially eligible manuscripts. The 168 potentially eligible manuscripts after screening by abstract were screened by the full text. After coding, 92 manuscripts comprising 104 individual samples (171,828 participants) were included. Of the 76 manuscripts which reached full-text screening but were not included, exclusion was due to not reporting a direct comparison between trauma exposure and presence of PLEs (n=18), not reporting all information necessary to convert the effect size to a correlation coefficient (e.g., not reporting the standard deviations of both independent and dependent variables; n=17), reporting on the same sample and relationship as another study which was included (i.e., the effect size was already represented in the meta-analysis; n=9), collapsing across clinical and non-clinical psychotic experiences (n=9), not assessing trauma exposure (e.g., assessed only trauma-related symptoms; n=5), collapsing across clinical and non-clinical samples (n=4), using non-quantitative analyses (n=3), not assessing PLEs in the non-clinical sample (n=3), analyzing trauma exposure or PLEs only as a mediator for a different comparison (n=2), not reporting on psychological trauma (e.g., traumatic brain injury; n=2), not reporting sufficient information to determine whether reports were on eligible types of trauma/PLEs (e.g., collecting scales which do and do not require clinical relevance for a psychotic-like experience to be scored as “present” and not specifying which scale was used in analyses; n=2), reporting on clinically relevant psychosis (n=1), and collapsing across PLEs and other domains of psychopathology (n=1). This information is represented in a cohort diagram in Supplementary Figure 1.

The results of the aggregated analysis are presented in Figure 1. Trauma exposure was associated with an increased likelihood of PLEs, k=104, Q(103)=1918.81, p<.0001, tau=0.107, r=.23 [95% CI = 0.204–0.228]. A random-effects meta-regression model testing for funnel plot asymmetry was significant, z=2.039, p=.04, which suggests publication bias may have been observed in this sample. Supplementary Figure 2 presents a funnel plot for included samples. A fail-safe n calculation indicated that 7,881 analyses with null effects would need to be added to the current meta-analysis for the effect to be non-significant (p≥.05), which indicates that there would need to be substantially more unpublished studies with null effects than published studies for the effect to be non-significant. The fail-safe n indicates that there would need to be 78 times as many unpublished studies with null results as published analyses included in this meta-analysis before the overall average effect size would not be significantly different from zero, which suggests that the current results are robust even if they are influenced by publication bias.

Figure 1.

Forest plot of the effect sizes for all included samples. Thicker bars represent a larger sample size.

Question 2: Effects by Type of Trauma

To examine whether the strength of the relationship between trauma and PLEs varied by trauma type, effect sizes were examined for abuse, neglect, emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, bullying, and discrimination separately. Effect sizes were similar for all types of abuse (proxy for “threat”), r=.21 [95% CI = 0.177–0.248], and all types of neglect (proxy for “deprivation”), r=.18 [95% CI = 0.127–0.229]. Within types of abuse, effect sizes for emotional abuse, r=.26 [95% CI = 0.209–0.312], were larger than those for physical abuse, r=.17 [95% CI = 0.129–0.202], or sexual abuse, r=.16 [95% CI = 0.140–0.189]. Effect sizes were comparable between emotional neglect, r=.18 [95% CI =0.155–0.214], and physical neglect, r=.17 [95% CI =0.096–0.236]. Notably, effect sizes for the small number of studies reporting on bullying, r=.28 [95% CI = 0.185–0.367], and discrimination, r=.23 [95% CI = 0.143–0.313], (proxies for “discrepancy”) were of similar sizes to other trauma types. Effect sizes grouped by trauma type are presented in Figure 2 a and b.

Figure 2.

A) Forest plot of effect sizes grouped by abuse or neglect for samples with sufficient detail reported to calculate these effect sizes. B) Forest plot of effect sizes grouped by emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, bullying, and discrimination for samples with sufficient detail reported to calculate these effect sizes.

Question 3: Cumulative Effect

For studies with effects reported by level of trauma exposure, independent effect sizes were recorded for each subsample reported. The visualization of the relationship between approximate number of trauma exposures and effect sizes is shown in Figure 3. As represented in this figure, effect sizes between number of trauma exposures and PLEs almost uniformly increase as the number of trauma exposures increases. Numerically, the change in effect size ranged from a decrease of r=−.04 to an increase of r=.312 and averaged an increase of r=.118.

Figure 3.

Effect size of the relationship between number of trauma exposures and PLEs ordered by approximate number of trauma exposures. Points which share a shape and color represent effect sizes derived from independent subsamples of the same overall sample.

Question 4: Impact of Methodological Variation

Most studies used the participant’s self-reported trauma exposure, r=.24 [95% CI = 0.213–0.265]. A smaller subset of studies using parent-reported trauma exposure demonstrated smaller effect sizes, r=.17 [95% CI = 0.116–0.217]. Although the small cell size limits comparisons, the 2 studies which used child protective services involvement to assess trauma also reported a smaller association, r=.09 [95% CI = −0.204–0.378]. Relationships were similar when PLEs were self-reported, r=.23 [95% CI = 0.205–.259], or when obtained through interviews, r=.22 [95% CI = 0.161–0.272].

Question 5: Influence of Sample Characteristics

The proportion of the sample which was male was calculated for included samples, and this percentage was included in a meta-regression model as a moderator. A mixed-effects meta-regression indicated that there was not a significant difference in results based on proportion of the sample which was male, k=102, QM(1)=0.163, p=.687, tau=0.108. Predominantly female samples (50% of samples, 51,215 participants) had an average effect of r=.23 [95% CI = 0.193–0.263]. Results were similar for samples balanced on sex (46.2% of samples, 119,011 participants), r=.22 [95% CI = 0.186–0.252], and potentially larger for the small number of predominantly male samples (3.6% of samples, 1,602 participants), r=.34 [95% CI = 0.090–.589], although the wide confidence intervals limit interpretability. Effect sizes grouped by sample sex are presented in Figure 4a. There were no notable differences in the pattern of effect sizes for different trauma types when grouped by predominant sample sex.

Figure 4.

A) Forest plot for all included samples grouped by the majority sex of the sample. B) Forest plot for all included samples grouped by the majority race/ethnicity of the sample.

Similarly, samples were grouped by predominant race/ethnicity. Effects were similar for predominantly white samples (29.8% of samples, 30,139 participants), r=.24 [95% CI = .189–0.282], predominantly non-white samples (17.3% of samples, 34,539 participants), r=.21 [95% CI = 0.142–.274], and for samples which did not report race/ethnicity (52.9% of samples, 107,150 participants), r=.23 [95% CI = 0.199–.262]. Effect sizes grouped by predominant sample race/ethnicity are presented in Figure 4b.

Discussion

This meta-analysis is, to the authors’ knowledge, the first meta-analytic summary of the relationship between trauma exposure and PLEs. The relationship between higher trauma exposure and higher PLEs is consistent with studies examining trauma exposure and clinically-relevant psychotic symptoms, and observed effect sizes are a remarkably similar size to those observed in clinical psychosis (Bailey, 2018; Pastore et al., 2022). This consistency suggests that PLEs might be a useful analog for psychotic symptoms or disorder when conducting preliminary research on the relationship between trauma exposures and psychosis spectrum experiences. In addition, the results of the current analyses reinforce the importance of a comprehensive assessment for trauma in clinical care in order to have a complete understanding of factors influencing current clinical presentation. However, the rigor of these results is limited by the possibility of publication bias within the analyses included.

The proneness-persistence-impairment model discusses how, among those who experience PLEs, the experiences will often be transient, with only some individuals developing persistent PLEs and a smaller group developing clinically-relevant distress and impairment. The current analysis provides evidence that stress exposure contributes at the proneness stage. Importantly, this finding emphasizes the potential richness of examining the interplay between environment and biology in individuals experiencing transient, persistent, or impairing PLEs to develop models of the relative contributions of each factor in increasing psychosis proneness and the likelihood of clinical relevance and/or disorder. In particular, evidence from this and other studies (Dominguez, 2011; Lindgren et al., 2022; Zammit et al., 2013) suggest that PLEs can be considered a low-level presentation of experiences along a spectrum of psychosis and that they may be a useful analog population in which to augment research studying factors which contribute to the development of psychosis. This can move the field forward faster given that it is substantially easier to recruit participants experiencing PLEs than those experiencing clinical psychosis due to higher prevalence and the ability to use self-report measures instead of requiring interview-based measures (Bourgin et al., 2020; Waters & Stephane, 2014), and there are likely fewer confounds complicating the presentation (e.g., the side effects of antipsychotic medications).

Another primary conclusion from this analysis is that the assessment of trauma and stressors in psychosis research urgently needs to be expanded. Although there is often recognition in clinical contexts that trauma-related symptoms, and the associated need for support and potentially clinical care, may arise from experiences which would not be classified as sufficient to diagnose post-traumatic stress disorder (PTSD), research has more typically focused on a narrow definition of trauma which excludes these experiences. Bullying and discrimination, which have only recently begun to gain recognition in the research domain as important risk factors for developing psychosis, are shown here to have preliminary evidence that they may be just as influential as experiences more traditionally considered trauma (e.g., sexual abuse). Research examining the strength of this relationship and the mechanisms underlying the influence of stressors such as bullying and discrimination on psychosis risk should evaluate whether these exposures function in the same or in distinct ways relative to stringently defined trauma exposures. Doing so could elucidate the specific characteristics of exposures (e.g., threat to life, biological stress response) which are driving the relationship with psychosis risk and inform nuanced risk assessment and treatment targets in the future. One such characteristic which should be carefully considered is the role of social context in stressful or traumatic exposures, given that previous research has demonstrated that psychopathology is more likely following interpersonal traumas such as sexual abuse than impersonal traumas such as car accidents (Furr et al., 2018; McLaughlin et al., 2020).

The importance of considering and reporting gender, race/ethnicity, and intersectionality (e.g., reporting on the number of Black women in the sample rather than the number of Black participants and the number of female participants separately) is underscored with the recognition that stressful experiences do not occur at equal rates across demographic groups. In doing so, research practices may become more aligned with clinical application – a critical goal if empirically-derived indicators of risk and understanding of the mechanisms driving symptom development will benefit real-world implementation of prevention or early intervention models of care. Similarly, the smaller effect sizes observed when trauma exposure was assessed through agency involvement suggest that, as may be expected, agency reports likely do not capture all exposures with meaningful impact on psychosis risk. Furthermore, systemic factors like the tendency to consider the study of white participants to act as a “universal” population, lack of sensitivity to culturally relevant practices, and the degree of comfort participants have participating in and reporting to research studies (or institutional authorities more broadly) mean it is likely that the nature and extent of relevant experiences is not yet known. This is particularly of note in relation to the surprising finding that there were not significant differences in the strength of the relationship between trauma exposures based on the majority race/ethnicity of the sample despite a significant relationship between discrimination and PLEs. It is possible that heterogeneity in the location of studies may obscure any differences given that the likelihood of experiencing discrimination is context-dependent (e.g., non-White is not the same as marginalized); however, it is also possible that studies have not adequately captured factors related to resilience which may buffer the effects of discrimination and trauma. The probability of missing important pathways to psychosis because of these underlying issues in information gathering raises critical considerations in the design and interpretation of studies, as well as how studies may inform treatment and policy recommendations. Significantly, it is necessary to consider the contexts and cultures in which stress and resilience occur when determining the appropriate time to recommend interventions and which interventions to recommend, such that the strengths of a community can be leveraged and bolstered as effectively as possible. Additional community-engaged and qualitative work would be most likely to support considered and impactful decision making in these areas.

Additionally, assessment would benefit from increased nuance in describing the intensity, chronicity, and conceptual category (i.e., threat, deprivation) of trauma. Even with incredibly high variability in defining “dose” of trauma, more trauma exposure results in a notably stronger relationship with PLEs. In studies reporting effects for different levels of trauma exposure, the definition of “low trauma exposure” ranged from no reported traumas to 5 or fewer reported traumas. These groupings also most often relied on the number of different trauma types reported without accounting for intensity or chronicity (e.g., someone reporting based on one physical assault and someone reporting based on frequent physical abuse would be considered to have the “same” exposure). Assessment also rarely includes information relevant to exploring the contributions of the different mechanisms implicated by threat-, deprivation-, and discrepancy-based exposures. Improving the granularity of assessments will enable more precise information about what components of trauma exposure are most influential for increasing psychosis risk as well as inform treatment development and selection based on underlying mechanisms.

Conclusion

Trauma exposure has a clear, well-documented influence on the likelihood of experiencing PLEs even if these experiences do not become clinically relevant, which supports the conceptualization of a psychosis continuum. Differences in average effect sizes based on trauma type, number of exposures, predominant sex of the sample, and assessment methods highlight key areas of consideration when designing studies. Finally, this meta-analysis underscores the need to expand the definitions and granularity used in trauma assessment to capture experiences influencing psychosis risk more fully.

Public health significance statement:

Psychosis is a type of mental health symptom which leads to considerable disability burden. Early intervention has been shown to improve long-term outcomes for people experiencing psychosis. The current paper summarizes across many studies to shed light on the role played by psychological trauma exposure (e.g., abuse, neglect) which elevates later risk for psychosis and which may guide future preventative efforts in the field.