The Impact of Childhood Trauma, Hallucinations, and Emotional Reactivity on Delusional Ideation

Abstract

Background

A large body of evidence has demonstrated that childhood trauma increases the risk for developing a psychotic disorder. However, the path by which psychotic symptoms develop following trauma is still debated. Emotional reactivity, a tendency to experience more intense and enduring emotional responses, and hallucinations are sequelae of childhood trauma that may predict the emergence of delusional ideation. This study aimed to examine evidence for this hypothesis in a college student sample.

Methods

Self-report measures were used to cross-sectionally assess childhood trauma (Childhood Trauma Questionnaire), delusional ideation (Peters et al Delusions Inventory), hallucinations (Launay-Slade Hallucinations Scale), and emotional reactivity (Emotion Reactivity Scale) in 1703 US college students.

Results

Hallucinations and emotional reactivity fully mediated the relationship between childhood trauma and delusional ideation, after controlling for covariates (gender, age, levels of current stress, stressful life events, and resilience).

Discussion

These results suggest that hallucinations and emotional reactivity associated with childhood trauma may represent quantifiable targets for early intervention.

psychosis, early trauma, emotional reactivity, hallucinations, delusions

Introduction

Psychosis is currently understood within a dimensional framework as a highly heterogeneous continuum that spans from psychotic experiences in the general population, without distress or need for intervention, to frank psychotic symptoms requiring coordinated specialty care.¹ Typically, psychosis emerges in late adolescence and early adulthood,^2,3 which may be a particularly vulnerable period, as the organization of the brain is undergoing accelerated refinement.⁴ Many environmental and interpersonal stressors, such as leaving home, changes in relationships, and beginning new roles, commonly occur during this developmental time window.

Subclinical symptoms of psychotic phenotypes (eg, a clustering of “soft” delusional ideation or hallucinations), which have phenomenological similarities to clinical psychosis, are often referred to as psychotic experiences (PEs). The prevalence of PEs in the general population is around 7% when using a conservative measure.⁵ When PEs are defined more broadly (eg, when experiences that are in line with non-Western cultural norms are included), 13% of the general population endorse PEs as occurring “often”.⁶ There has been growing interest in studying these milder expressions of psychosis, as they may represent phenotypical markers of primary psychotic disorders (ie, persistent PEs confer at least a 10-fold increased risk for clinical psychosis⁷). As such, identifying etiological mechanisms of PEs is important for developing early detection strategies and identifying novel intervention targets.

It has been well-established that a history of early and cumulative trauma is common in individuals with psychosis,^8–10 especially interpersonal victimization.¹¹ Among those with a clinical psychotic disorder, 75% have experienced a traumatic experience during childhood¹² and approximately 12.5% display symptoms that meet full diagnostic criteria for Posttraumatic Stress Disorder (PTSD).^13–15 Keen, Hunter, and Peters¹⁴ have proposed linkages between the re-experiencing symptoms of PTSD and hallucinations, as well as connections between trauma and paranoia. However, while the relationship between early trauma and the full spectrum of psychosis (from subclinical to clinical expressions) is well-supported,¹⁶ the cognitive mechanisms underlying this association remain poorly understood.

Several pathways have been proposed to explain the associations between early traumatic experiences and psychosis. Trauma may lead to dissociation (ie, disruptions in the synthesis of thoughts, feelings, and experiences into the stream of consciousness and memory) as a coping mechanism for avoiding traumatic memories, which predict the presence of hallucinations^8,17,18 and delusions.¹⁹ Additionally, attachment difficulties, self-disturbances, aberrant salience, and other cognitive biases may also be important factors that contribute to the association between trauma and PEs.^20–22 Importantly, cognitive models have proposed that the unusual experience of a hallucination requires an appraisal by the individual, which could be negative or threatening, and this can lead to the development of delusional beliefs.^23–27 Thus, trauma may lead to hallucinations, and the interpretation of these hallucinations may subsequently give rise to delusional ideation.

One potentially related pathway between childhood trauma and psychosis is suggested by the large body of evidence showing that experiences of childhood trauma, or adversity, are associated with subsequent impairments in emotion regulation.^28–31 Emotion regulation, consisting of processes responsible for monitoring, evaluating and modifying emotional reactions,³² is highly influenced by early childhood attachments and responsiveness of caregiving.³¹ Childhood trauma can disrupt the development of appropriate emotional regulation capacities, via alterations in stress hormone systems, which can lead to neuronal loss and changes in brain development.³³

In support of a potential link between early trauma and emotion dysregulation, a recent meta-analysis found that, in comparison to children who were not maltreated, maltreated children experience higher levels of negative emotions and behavioral expressions of emotional dysregulation.³⁴ However, specific disruptions in emotional reactivity were shown to be even greater than levels of overall emotion dysregulation in the maltreated children.³⁴ Emotional reactivity is defined as the extent to which an individual experiences emotions (1) in response to a wide array of stimuli (ie, sensitivity), (2) strongly or intensely (ie, intensity), and (3) for a prolonged period before returning to a baseline level of arousal (ie, persistence).³⁵ Those with high emotional reactivity have the tendency to experience more intense and long-lasting emotional responses from stressors, which may predispose these individuals to impairments in emotion regulation.³⁵

Research has highlighted the role of emotional regulation (specifically emotional reactivity or stress sensitivity) as a potential mediating factor between trauma or attachment and psychosis.^36–40 Experimentally manipulating stress has been shown to lead to an increased vulnerability to psychotic symptoms, including paranoia.⁴¹ Similarly, when assessing psychotic experiences within daily life using Ecological Momentary Assessment, stress sensitivity was shown to be associated with more intense psychotic experiences.^42,43 Consistent with prior research indicating that stress sensitization likely plays a key role in the etiology and persistence of PEs,^41,44–46 emotional reactivity has been linked to trauma history,^47–49 psychopathology,⁵⁰ and, for those with schizophrenia, an increase in psychotic symptoms.⁵¹ Thus, an excess of emotional reactivity, arising from childhood trauma, may underlie the increase in stress sensitization associated with PEs.

In support of this model, prior studies have reported that affective difficulties mediated the relationship between childhood trauma and psychotic symptoms in children,⁵² and affect dysregulation (eg, avoidance, numbing, and hyperarousal symptoms) mediated the relationship between childhood trauma and psychotic symptoms in schizophrenia.⁵³ However, no study has yet assessed the specific role of heightened emotional reactivity (one type of emotion dysregulation) in the relationship between childhood trauma and PEs in young adults. Based on this literature, the current study aimed to test the hypothesis that associations between self-reported PEs (hallucinations and delusional beliefs) and self-reported childhood trauma in a young adult population can be attributed in part to heightened emotional reactivity.

Specifically, we hypothesized that, while controlling for potential confounds (ie, age, gender, recent stressful life events, current perceived stress in life, and resilience levels, to assess the independent contribution of trauma, emotional reactivity and hallucinations on delusional ideation), higher levels of self-reported childhood trauma would be associated with increased delusional ideation and hallucinations in a sample of young adults. We also hypothesized that emotional reactivity and hallucinations would mediate the relationship between childhood trauma and delusional ideation. Moreover, to further understand these associations, we conducted exploratory factor analyses (EFA) to determine whether different types of delusional ideation are differentially predicted by hallucinations, emotional reactivity, and childhood trauma.

Methods

Participants

1703 ethnically diverse participants were recruited as part of a larger study of psychopathology in college students conducted in the greater Boston area from 2010 and 2017.^54–57Table 1 provides participant demographics and descriptive statistics. To be eligible for this study, participants needed to be enrolled as an undergraduate student at a local university and at least 18 years of age.

Table 1.

Participant Demographics and Descriptive Statistics

Categorical Variables	Level	Total (n = 1703)
Gender	Female	1220 (71.6%)
Race/ethnicity	Caucasian	690 (41%)
	Asian	530 (31.1%)
	Hispanic	199 (12%)
	African American	52 (3%)
	Other/biracial	81 (5%)
	No data	143 (8%)
Year in school	Freshman	487 (29%)
	Sophomore	445 (26%)
	Junior	382 (22%)
	Senior	346 (20%)
	Other	26 (2%)
	No data	17 (10%)
Marital status	Never married	1659 (97.4%)
Continuous Variables	Mean (SD)	Range
Age (y)	19.55 (1.30)	18–28
Perceived stress total (PSS)(0–40)	16.33 (7.11)	0–40
Emotion reactivity (ERS)(0–84)	28.03 (19.12)	0–84
PDI total (number of endorsed beliefs)(0–21)	5.29 (3.52)	0–21
PDI distress subscale(0–105)	12.84 (11.2)	0–79
PDI preoccupation subscale(0–105)	13.85 (11.3)	0–78
PDI conviction subscale(0–105)	15.83 (11.7)	0–77
Summed PDI score(0–336)	47.88 (36.5)	0–251
Hallucinations (LSHS)(0–64)	11.33 (10.49)	0–63
Childhood traumatic experiences (CTQ)(28–140)	34.96 (11.34)	24–101
Stressful life events (SEC)(0–1285)	334.95 (190.26)	0–1101
Resilience total score (CD-RISC)	70.72 (15.91)	10–100

Note: CD-RISC, Connor-Davidson Resilience Scale; CTQ, Childhood Trauma Questionnaire; ERS, Emotion Reactivity Scale; LSHS, Launay-Slade Hallucination Scale-Revised; PDI, Peters et al Delusions Inventory.

Procedures

In-person mental health screening events were conducted at 4 Boston area campuses for 1 to 2 days. Participants completed the informed consent procedure and, subsequently completed self-report questionnaires assessing various aspects of mental health (see refs.^54–56 for additional details and above).

Measures

This study assessed psychotic experiences (hallucinations and delusional ideation) on a continuum and has, therefore, included measures applicable for use in the general population.

Peters et al Delusions Inventory.

The Peters et al Delusions Inventory (PDI)⁵⁸ is a 21-item self-report questionnaire designed to assess delusional ideation in the general population. Participants answer “yes” or “no” to 21 statements. If they answer “yes,” then they are asked to rate from 1 to 5 the degree of (1) conviction, (2) preoccupation, and (3) distress of the endorsed experience on a Likert scale. This revised version of the PDI, the PDI-21, has demonstrated internal consistency, test-retest reliability, and validity.⁵⁹ PDI has been shown to be appropriate for use in clinical and non-clinical groups,^60,61 including college students.⁶² The number of endorsed beliefs is added to the scores of the 3 subscales to generate the summed total PDI score. Internal consistency for this measure was high (α = 0.77).

Launay-Slade Hallucination Scale 16-Item Version.

The Launay-Slade Hallucination Scale-Revised (LSHS) is a 16-item self-report questionnaire designed to assess perceptual aberrations and hallucinations,⁶³ which was derived from an earlier, longer version.⁶⁴ Participants rate the 16 statements on a 5-point scale to indicate how much each statement relates to their own experiences from 0 (certainly does not apply to me) to 4 (certainly applies to me). Scored were summed to produce a total score was used. This measure has shown good reliability⁶⁵ and is suitable for use in non-clinical groups.^63,65 Internal consistency was α = 0.88 in the current study.

Childhood Trauma.

The Childhood Trauma Questionnaire-Short Form (CTQ-SF)⁶⁶ is a 28-item measure designed to retrospectively assess experiences of abuse or neglect during childhood. CTQ includes 5 subscales: physical abuse, emotional abuse, sexual abuse, physical neglect, and emotional neglect. The items are summed to produce a total score from 28 to 140. The CTQ has shown excellent test-retest reliability and construct validity⁶⁷ and good internal consistency in community samples.⁶⁸

Current Stress.

The Perceived Stress Scale (PSS)⁶⁹ is a commonly used tool for measuring levels of current psychological stress. The scale consists of 10 items, each designed to evaluate the degree to which a person finds their life to be unpredictable and uncontrollable in the past month.⁶⁹ Internal consistency was α = 0.97 in the current study.

Past and Current Stressful Events.

The College Student’s Stressful Events Checklist (SEC; retrieved from ASU Wellness asu.edu/wellness) includes 32 life event items relevant to college-age students including death of a close friend, having to repeat a course, pregnancy, and change in social habits, which are each assigned a value.

Emotional Reactivity.

The Emotion Reactivity Scale (ERS)³⁵ was used to measure self-reported, trait-like tendencies towards particular levels of intensity of emotional responses to events. The ERS consists of 21 items used to assess one’s emotional sensitivity (eg, “I tend to get emotional very easily”), intensity (eg, “I experience emotions very strongly”), and persistence (eg, “When I am angry/ upset, it takes me longer than most people to calm down”). Scores are summed to produce a total score, with higher scores indicating higher levels of self-reported emotional reactivity. The ERS has demonstrated good convergent and discriminant construct validity, and criterion-related validity.^35,70 Internal consistency was α = 0.95 in the current study.

Resilience.

The Connor-Davidson Resilience Scale (CD-RISC)⁷¹ assesses an individual’s ability to cope with stress. Scored are summed for a total score with higher scores on this 25-item self-report measure indicating greater resilience. This scale is widely used and has observed good reliability, validity, and factor structure.⁷¹ Internal consistency was α = 0.94 in the current study.

Analytic Approach

Overall, missing data was assumed to be missing completely at random (MCAR) after conducting correlational and chi-squared analyses. For regression analyses, listwise deletion was used.⁷² SPSS 24 was used for all descriptive and regression analyses and a P value of < .05 was considered significant. For mediation analysis, full information maximum likelihood was used, which combines available information to estimate population parameters.^73,74 The mediation analyses were conducted using Mplus with Multiple Mediation Model (structural equation modeling) using Maximum Likelihood Estimation (MLE), bootstrapping and corrected confidence intervals, following Preacher and Hayes⁷⁵ causal steps of mediation.

First, analyses included obtaining descriptive statistics and running correlations to assess associations (and covariates) of delusional ideation and hallucinations.⁷⁶ We applied a Bonferroni-corrected P-value (P < .05 divided by number of correlations, 21 correlations/.05 = .0023) for the correlation analyses. Hierarchical regression models, controlling for covariates: age, gender, current perceived stress, stressful life events, and resilience, were then conducted to evaluate whether childhood trauma predicted both delusional ideation and hallucinations independent of individual differences in recent stressors, current stress level, and resilience factors.

A multiple mediation model was then developed to assess whether emotional reactivity and hallucinations mediate the relationship between childhood trauma and delusional ideation, while controlling for covariates. Then, in order to understand whether different types of delusions were differentially predicted by hallucinations, emotional reactivity and childhood trauma, we conducted an exploratory factor analysis (EFA) using all items of the PDI to assess loading onto separate factors. Following this, we conducted a confirmatory factor analysis (CFA) using a new dataset of students (N = 1138 recruited from the same Boston area college student population, see supplementary table 1 for additional details) to confirm these factors. We also re-ran the multiple mediation using the Abuse and Neglect subscales of the CTQ. See supplementary materials A for further details. Lastly, we re-ran the multiple mediation model to assess whether emotional reactivity and hallucinations mediate the relationship between trauma and the PDI factors produced from the EFA (eg, persecutory-referential-passivity, grandiosity and religiosity).

Results

Preliminary Findings

There were significant correlations between all study variables (correlation coefficient r = 0.11–0.6; supplementary table 2). Preliminary correlational analyses revealed that age (delusions [d]: r = −.1; P < .001; hallucinations [h]: r = −0.07; P = .01), gender (ERS, t(505.3) = 5.1, P < .001), current perceived stress (d: r = 0.39; P < .001; h: r = 0.44; P < .001), stressful life events (d: r = 0.3; P < .001; h: r = 0.22; P < .001), and resilience (d: r = −.2; P < .001; h: r = −.26; P < .001) were all significantly associated with delusional ideation and hallucinations. Therefore, to control for potential confounds as well as test the aforementioned hypotheses, these were included as covariates in the regression models.

Relationship Between Childhood Trauma and Psychotic Experiences

Controlling for age, gender, current perceived stress, recent stressful life events, and self-reported resilience, levels of childhood traumatic experiences (β = .14, P = .004) were positively associated with delusional ideation. Additionally, there was a direct and significant relationship between childhood trauma (β = .16, P = .001) and hallucinations, while controlling for these same factors. Table 2 provides statistics for each of these regression models.

Table 2.

Hierarchical Regression Analyses Examining the Relationships Between Early Trauma, Delusional Ideation, and Hallucinations While Controlling for Age, Recently Perceived Stress, Recent Stressful Events, and Resilience

Variable	UnStd. β (SE)	Std. β	P-value	ΔR2	F-value	dfs	P-value
PDI
Block 1				0.18	17.64	(5, 399)	<.001
Age	−0.4 (0.12)	−0.15	.001
Gender	−0.49 (0.37)	−0.06	.19
CD Resilience Scale	−0.05 (0.01)	−0.2	.7
Stressful Life Events	0.04 (0.01)	0.21	<.001
Perceived Stress Scale	0.13 (0.03)	0.28	<.001
Block 2				.009	15.53	(6, 404)	.04
CTQ Total	0.03 (0.02)	0.1	.04
LSHS
Block 1				0.27	29.29	(5, 395)	<.001
Age	−1.03 (0.34)	−0.13	.002
Gender	0.34 (1.08)	0.01	.76
CD Resilience Scale	−0.04 (0.03)	−0.07	.23
Stressful Life Events	0.01 (0.01)	0.15	.001
Perceived Stress Scale	0.57 (0.08)	0.41	<.001
Block 2				0.02	26.89	(6, 394)	.001
CTQ Total	0.15 (0.04)	0.16	.001

Note: CTQ, Childhood Trauma Questionnaire; LSHS, Launay-Slade Hallucination Scale-Revised; PDI, Peters et al Delusions Inventory.

Mediation Models

A series of mediation models were conducted to identify the indirect mediating effect of hallucinations and emotional reactivity between childhood trauma and delusional ideation. Subsequently, a full multiple mediation model was developed.

First, the mediating effect of hallucinations on the relationship between childhood trauma and delusional ideation was measured. Hallucinations significantly, fully mediated the relationship between trauma and delusional ideation (β = .2, P < .001, ±95% CI [0.13,0.27]; supplementary figure 1). Second, the mediating effect of emotional reactivity on the relationship between childhood trauma and delusional ideation was measured. Emotional reactivity significantly, partially mediated the relationship between trauma and delusional ideation (β = .15, P < .001, ±95% CI [0.11,0.19]; supplementary figure 2).

Finally, a full multiple mediation model was conducted, assessing the mediating effects of emotional reactivity and hallucinations on the relationship between trauma and delusional ideation, controlling for covariates: age, gender, perceived stress, stressful events and resilience. Significant direct pathways were found between trauma and hallucinations (β = .35, P < .001) and trauma and emotional reactivity (β = .33, P < .001). Significant direct pathways were found between delusional ideation and hallucinations (β = .49, P = .001) and delusional ideation and emotional reactivity (β = .22, P < .001). Emotional reactivity significantly mediated the relationship between trauma and delusional ideation (β = .07, P < .001, ±95% CI [0.04,0.11]) and hallucinations also significantly mediated the relationship between trauma and delusional ideation (β = .17, P < .001, ±95% CI [0.1,0.23]). The direct pathway was nonsignificant suggesting a full mediation model (see figure 1; β = .24, P < .001, ±95% CI [0.17,0.31]). When omitting covariates, this model remained significant (β = .2, P < .001, ±95% CI [0.15,0.24]). Using the scores on the PDI distress subscale only, which is associated with increased risk for developing clinical psychosis, the model remained significant (β = .25, P < .001, ±95% CI [0.18, 0.32]).

Figure 1.

Multiple mediation model from childhood trauma to delusional ideation with 2 mediators: hallucinations and emotional reactivity.

Childhood Trauma Questionnaire Subscales

We re-ran the multiple mediation using the Abuse and Neglect subscales from the CTQ, demonstrating a full mediation with the Neglect subscale and a partial mediation with the Abuse subscale. See supplementary materials A for further details.

Full EFA

An EFA was conducted which included all items of the PDI to identify loadings onto separate factors within the full sample. A 1-factor solution was poor fit [χ ²(189) = 1985.93, P = .00, CFI = .7, TLI = .67, RMSEA = 0.08]. The scree plot (supplementary figure 3) and model fit suggested a 3-factor solution. A 3-factor solution was an appropriate fit to the data [χ ²(150) = 635.6, P = .00, CFI = .92, TLI = .89, RMSEA = 0.04]. While 2 of the factors had only 2 items loading, these factors were conceptual defensible as consistent with previous PDI factor analyses.^77–79Table 3 lists the factor loadings. Factor 1 represented “persecutory-referential-passivity” delusions (M = 34.5, SD = 30.5), factor 2 represented grandiosity (M = 9.13, SD = 8.2) and factor 3 represented religiosity (M = 3.15, SD = 6.5).

Table 3.

Factor Loadings for Peters et al Delusions Inventory (PDI)

Item #	Description	Factor 1 (General Delusions)	Factor 2 (Grandiosity)	Factor 3 (Religiosity)
1	Hints/double meaning	0.52
2	Special messages on TV/Magazines	0.3
3	People are not what they seem to be	0.47
4	Being persecuted	0.5
5	Conspiracy	0.43
6	To be someone very important		0.74
7	To be a special or unusual person		0.59
8	To be especially close to God			0.88
9	Telepathic communication	0.4
10	Electric device influencing way of thinking	0.33
11	Having been chosen by God			0.7
12	Believing in the power of witchcraft	0.36
13	Worrying about one’s partner’s unfaithfulness	0.29
14	To have sinned more than the average person	0.36
15	People looking oddly at you	0.45
16	To have no thoughts in head	0.45
17	End of the world	0.38
18	Alien thoughts	0.59
19	Thought broadcasting	0.54
20	Thought echoing back	0.5
21	To be like a robot or zombie	0.46

PDI CFA

We aimed to confirm the 3-factor solution for the PDI using a confirmatory factor analysis on a new sample of 1138 college students. The CFA demonstrated that a 3-factor solution was an appropriate fit to the data [χ ²(186) = 575.33, P = .00, CFI = .86, TLI = .84, RMSEA = 0.05].

PDI Factors

Next, a full multiple mediation model was conducted, using the original sample, accounting for the mediating effects of emotional reactivity and hallucinations on the relationship between trauma and the persecutory-referential-passivity delusion factor, controlling for covariates. Significant direct pathways were found between persecutory-referential-passivity delusions and hallucinations (β = .5, P < .001) and persecutory-referential-passivity delusions and emotional reactivity (β = .24, P < .001). Emotional reactivity significantly mediated the relationship between trauma and persecutory-referential-passivity delusions (β = .08, P < .001), and hallucinations also significantly mediated the relationship between trauma and persecutory-referential-passivity delusions (β = .17, P < .001). The direct pathway was nonsignificant, suggesting a full-mediation model.

Next, a full multiple mediation model was conducted, accounting for the mediating effects of emotional reactivity and hallucinations on the relationship between trauma and the grandiosity factor, controlling for covariates. Significant direct pathways were found between grandiosity and hallucinations (β = .29, P < .001), but not between grandiosity and emotional reactivity (P = .15). Hallucinations also significantly mediated the relationship between trauma and grandiosity (β = .1, P < .001). The direct pathway was nonsignificant, suggesting a full mediation model.

Finally, a full multiple mediation model was conducted, accounting for mediating effects of emotional reactivity and hallucinations on the relationship between trauma and the religiosity factor. There was no relationship between religiosity and trauma, hallucinations, nor emotional reactivity.

Discussion

This study found that both hallucinations and emotional reactivity fully mediated the relationship between childhood trauma and delusional ideation in a large sample of US college students. The multiple mediation remained significant even when controlling for level of resilience, age, gender, current perceived stress, and recent stressful life events. This study supports prior research demonstrating associations among childhood trauma, emotional dysregulation, and psychotic symptoms in individuals with schizophrenia,⁵³ but further extends this work by demonstrating a specific relationship with emotional reactivity (a specific type of emotion dysregulation). Additionally, this study was conducted in college students, who are at the peak age of risk for the emergence of psychotic disorders.⁴ These findings highlight key psychological mechanisms via which delusional ideation could develop.

Childhood trauma was associated with emotional reactivity, consistent with prior studies.³⁴ This relationship between trauma and emotional reactivity may be linked to deficiencies in the caregiving environment, ie, a decreased availability of responsive caregiving and secure attachments in childhood.³¹ The presence of hallucinations was also associated with childhood trauma; such perceptual abberations may represent a compensatory, avoidant coping mechanism related to dissociation,¹⁸ a putative precursor to hallucinations.^80–83

Full mediation occurs when inclusion of the indirect paths (eg, childhood trauma to hallucinations/emotional reactivity, and hallucinations/emotional reactivity to delusional ideation) results in the direct path becoming nonsignificant (eg, from childhood trauma to delusional ideation). This full mediation supports the proposal that emotional reactivity and hallucinations may result from compensatory responses of the brain (eg, avoidance and threat response mechanisms) to the highly stressful experiences of childhood abuse and/or neglect. The emergence of hallucinations and emotional reactivity may signal the failure of these compensatory processes. Moreover, the development of delusional ideas in association with hallucinations and emotional reactivity may represent a subsequent compensatory process, aimed at providing an external explanation^84–86 or organizing framework for experiences of emotional distress and hallucinations.

This interpretation of our findings is consistent with existing cognitive and Bayesian models of psychosis, which have proposed that hallucinations may be followed by delusional ideas that represent an attempt to make sense of hallucinatory or anomalous experiences.^{24,25,83–87} The role of trauma and emotional reactivity in this model is also consistent with evidence that delusions frequently reflect current emotional concerns.^88–91 The current study extends this literature by highlighting the role of childhood trauma and the contribution of hallucinations and emotional reactivity to the formation of a delusional idea.

It should be noted that this model incorporates two putative mechanisms (via hallucinations and emotional reactivity) of delusional ideation. There are several other mechanisms of delusions which have been studied previously as well, including a “worry thinking style,” negative beliefs about the self, and certain reasoning biases, such as “Jumping to Conclusions”.⁹²

In addition, this study demonstrated that delusional ideation may be comprised of at least three categories: (1) “persecutory-referential-passivity” delusional ideation, which includes a range of commonly experienced unusual beliefs such as ideas of reference, persecution and thought insertion and withdrawal, (2) religiosity, and (3) grandiosity. When the analysis was repeated using these three factors in turn, only the model with persecutory-referential-passivity delusional ideation replicated the results of the original multiple mediation model. These results suggest a specific relationship between childhood trauma and persecutory-referential-passivity delusions, that was not found for grandiosity or religiosity delusions. This result is aligned with studies which have previously demonstrated a relationship between persecutory delusions and traumatic experiences.⁹³ Bentall et al⁹⁴ additionally emphasized a symptom-specific relationship with trauma, as studies have demonstrated that persecutory or paranoia symptoms are associated with neglect, discrimination and physical abuse,^95–97 compared to hallucinations, which are more closely associated with childhood sexual abuse.⁹⁶ Other studies highlight that trauma predicts persecutory thinking, via affective components,^97,98 supporting the indirect path between trauma and persecutory-referential-passivity delusions, via emotional reactivity, reported here.

On the other hand, trauma has been shown to impact hallucinations directly and persecutory thinking indirectly, via affect⁹⁸. The indirect effect, via affective components, has been replicated in a large dataset,⁹⁷ supporting the indirect path between trauma and persecutory-referential-passivity delusions, via emotional reactivity and hallucinations, found in this study. When assessing the grandiosity factor, only hallucinations (not emotional reactivity) mediated the relationship between trauma and grandiosity, suggesting that grandiosity may develop via a specific pathway involving hallucinations only. This finding is consistent with the observed associations between grandiosity and positive, rather than negative, affect. Emotional reactivity is typically associated with higher levels of negative affect and thus may not play an important role in grandiose delusions.

There was no correlation between any variable and religiosity. While this lack of association may be due to a weaker factor structure (eg, only two items loaded onto this factor), evidence that religious beliefs, or even delusional religiosity, may reflect some aspects of adaptive psychological functioning in both schizophrenia⁹⁹ and non-clinical¹⁰⁰ samples, suggests that religiosity may be considered (mechanistically) distinct from other delusions.¹⁰¹

Limitations

This study has several limitations. First, as this was a combined dataset derived from several iterations of data collection, there was missing data. Missing data within this data set was rectified via MLE. Future analyses in independent datasets must be conducted to confirm these results. Second, all measures were collected concurrently, limiting inferences regarding causal directions of these paths. Longitudinal data will be required to further understand these associations. Third, childhood trauma was assessed with the CTQ which is a relatively brief, retrospective, self-report measure, which may fail to detect or accurately represent events that would be identified as maltreatment via prospective assessments.^102–105 Finally, this was a convenience, rather than a random sample of US college students. Thus, while this population may be at higher risk of developing psychosis, due to their age and various environmental stressors associated with attending college, these findings may be less applicable to a broader population of young adults. In addition, the sample was 70% female which is less representative of clinical samples with psychosis, which tend to be predominantly male.

Clinical Implications and Future Directions

These findings have several clinical implications. This study further highlights the importance of early identification and intervention for individuals who have experienced trauma. Recent studies suggest that there may be a critical, vulnerable period early in life with respect to the effects of trauma on brain development. For example, one study found that participants who were first exposed to trauma during early childhood experienced levels of depression and PTSD symptoms that were twice as high as those exposed later.^106,107 Trauma-focused therapies for psychosis, including prolonged exposure or eye movement desensitization and reprocessing¹⁰⁸ and trauma-focused Cognitive Behavior Therapy in psychosis, aim to address the broader range of posttraumatic stress in psychosis.¹⁰⁹ These interventions may enable an individual to process traumatic experiences alongside any distressing responses to those experiences.

Second, this study suggests that emotional reactivity and hallucinations may predict delusional ideation and could represent important targets for preventive interventions—particularly for those with a history of trauma. Evidence-based treatments such as Cognitive and Dialectical Behavioral Therapy (CBT; DBT), CBT for Psychosis,^110,111 mindfulness,¹¹² and interventions which adapt and combine these modalities¹¹³ may reduce risk for subsequent psychopathology in youth with histories of trauma. These interventions systematically target emotional dysregulation and aim to develop an adaptive cognitive explanation of symptoms and other stressful life experiences. Early access to such mental health services (ie, via pediatric practices, schools) could potentially interrupt progression to more impairing symptoms in those who are particularly vulnerable.

Conclusion

This study demonstrated that both hallucinations and emotional reactivity mediate the relationship between childhood trauma and delusional ideation, in a sample of US college students. Individuals with trauma are more likely to experience high levels of emotional reactivity and hallucinations following stressors, both of which may predict delusional ideation. This study also identified initial evidence for the existence of separate pathways for developing distinct types of delusional beliefs. Both emotional reactivity and hallucinations can potentially serve as markers of those at-risk for delusional ideation, who may benefit from interventions to prevent further impairment.

Funding

This work was supported by internal funds provided by the Massachusetts General Hospital (MGH) Department of Psychiatry, the Sydney R. Baer, Jr Foundation and the Massachusetts Department of Mental Health to Center of Excellence for Psychosocial and Systemic Research at Massachusetts General Hospital.

Acknowledgments

We are grateful for the invaluable assistance of numerous staff members of the MGH Depression Clinical Research Program and the MGH Resilience and Prevention Program with data collection. M.F. has received research support from following: Abbott Laboratories; Acadia Pharmaceuticals; Alkermes, Inc.; American Cyanamid; Aspect Medical Systems; AstraZeneca; Avanir Pharmaceuticals; AXSOME Therapeutics; Biohaven; BioResearch; BrainCells Inc.; Bristol-Myers Squibb; CeNeRx Bio-Pharma; Cephalon; Cerecor; Clarus Funds; Clintara, LLC; Covance; Covidien; Eli Lilly and Company; EnVivo Pharmaceuticals, Inc.; Euthymics Bioscience, Inc.; Forest Pharmaceuticals, Inc.; FORUM Pharmaceuticals; Ganeden Biotech, Inc.; GlaxoSmithKline; Harvard Clinical Research Institute; Hoffman- LaRoche; Icon Clinical Research; i3 Innovus/Ingenix; Janssen R&D, LLC; Jed Foundation; Johnson & Johnson Pharmaceutical Research & Development; Lichtwer Pharma GmbH; Lorex Pharmaceuticals; Lundbeck Inc.; Marinus Pharmaceuticals; MedAvante; Methylation Sciences Inc; National Alliance for Research on Schizophrenia and Depression; National Center for Complementary and Alternative Medicine; National Coordinating Center for Integrated Medicine; National Institute of Drug Abuse; National Institute of Mental Health; Neuralstem, Inc.; NeuroRx; Novartis AG; Organon Pharmaceuticals; Otsuka Pharmaceutical Development, Inc.; PamLab, LLC.; Pfizer Inc.; Pharmacia-Upjohn; Pharmaceutical Research Associates., Inc.; Pharmavite LLC; PharmoRx Therapeutics; Photothera; Reckitt Benckiser; Roche Pharmaceuticals; RCT Logic, LLC (formerly Clinical Trials Solutions, LLC); Sanofi-Aventis US LLC; Shire; Solvay Pharmaceuticals, Inc.; Stanley Medical Research Institute; Synthelabo; Taisho Pharmaceuticals; Takeda Pharmaceuticals; Tal Medical; VistaGen; and Wyeth-Ayerst Laboratories. DJH has received research support from Forum Pharmaceuticals, Inc., and Janssen Scientific Affairs.

Author notes

These authors contributed equally to the article.