Trauma, dissociation and hallucinations

The PTSD and SCZ+PTSD groups showed increased dissociation, specifically amnesia, linked to an increase in severity of hallucinations, supporting the mediating role of dissociation in individuals who have been exposed to trauma. There is now evidence that childhood adversity has occurred at significantly higher rates in people with psychosis. Studies have reported a 47% rate of childhood sexual abuse in populations with psychotic illness^{Reference Bendall, Jackson and Hulbert16} and that early trauma exposure significantly increases the risk for psychosis.^{Reference Morrison, Frame and Larkin17} However, although trauma can never been ruled out, not all individuals who experience hallucinations report trauma.^{Reference Luhrmann, Alderson-Day, Bell, Bless, Corlett and Hugdahl3}

In contrast, the subset of participants with schizophrenia without trauma in our study showed a lack of correlation between dissociation and hallucination scores, suggesting that trauma (and the process of dissociation) is not always necessary for auditory hallucinations to occur, as has been previously suggested.^{Reference Longden, Madill and Waterman4} This result provides potential support for the theory that there are multiple pathways to hearing voices.^{Reference Brewin and Patel2} Outside of the psychosocial stressors and personal narratives of voice hearers, neurobiological mechanisms in schizophrenia clearly link to auditory hallucinations, including frontotemporal connectivity alterations affecting language processing networks and loss of sensory anchors acting to increase dependence on prior expectations.^{Reference Allen, Modinos, Hubl, Shields, Cachia and Jardri18}

Comorbid schizophrenia and PTSD

A substantial proportion of our participants had comorbid schizophrenia and PTSD (SCZ+PTSD). This is consistent with the literature, which suggests that prevalence rates of PTSD in schizophrenia spectrum disorders are fairly high, varying between 0 and 55%, with 78% of papers reporting >10%.^{Reference Dallel, Cancel and Fakra19} This group had significantly higher scores on the PANSS and its three subscales, in line with studies showing that the severity of trauma and illness factors can result in increased symptom severity and increased comorbidity.^{Reference Dallel, Cancel and Fakra19} Yet this group had intermediary results in the severity of dissociation and hallucinations. This introduces the interesting concept of a subgroup of participants who experienced trauma-linked dissociative hallucinations, as well as auditory hallucinations that directly link to their psychotic illness. This concept supports the research in the field suggesting that both biological and psychological treatment strategies are clinically useful in schizophrenia.^{Reference Longden, Madill and Waterman4}

CADSS dissociation subscales: amnesia, derealisation and depersonalisation

The results from our analyses of the dissociation subscales of the CADSS were less clear-cut, which was possibly due to limited statistical power, particularly in the schizophrenia group. In the PTSD and SCZ+PTSD groups, however, there was a significant correlation between hearing voices and amnesia subscale scores. Regression analysis suggested that dissociative hallucinations were best predicted by symptoms of amnesia.

There was a non-significant trend with depersonalisation. In the literature, depersonalisation symptoms have been linked to auditory hallucinations^{Reference Longden, Madill and Waterman4,Reference Moskowitz and Corstens5} in which the voice is seen to represent a component of self-identity which has been dissociated owing to failure of integration of a traumatic experience.^{Reference Longden, Madill and Waterman4} Depersonalisation also plays a central role in the concept of loss of internal anchors described by Allen et al.^{Reference Allen, Coyne and Console20}

Nevertheless, in this study the evidence of the role of amnesia in mediating voice hearing was considerably stronger, particularly given that mediation analysis is a statistical tool that attempts to describe causality.^{Reference Kosuke, Keele and Tingley21} Steel et al^{Reference Steel, Fowler and Holmes6} have developed a contextual integration model to explain the experience of involuntary intrusion in trauma. This model provides the best fit for the results in this study, where trauma fundamentally disrupts the ability to encode and integrate incoming sensory information with co-occurring data stored in the hippocampus. Clinically, this weakened ability to integrate information would present as amnesia and the sensory involuntary intrusions could present as hallucinations. The dissociative amnesia may represent an attempt of the unconscious to forget, and the potential failure of this suppressive defence process invites the experience of hallucinations as involuntary sensory intrusions.

Regression analysis using item scores in the schizophrenia group showed that hallucinations correlated with a sense of detachment from the world (things moving in slow motion) and had an inverse relationship to depersonalisation (disconnected from their own body). Although changes in the sense of body including a loss of ‘bodily-self’ are a common finding in schizophrenia,^{Reference Stanghellini22} in our study this depersonalisation experience was negatively correlated with hearing voices. In contrast, seeing the world as distant or detached, consistent with the derealisation, was positively correlated. This finding is reflected in the clinical situation, where auditory hallucinations in schizophrenia are often described as ‘real’^{Reference Aggemaes23} and can be experienced with a clear sense of the self ‘having a conversation’, whereas the world is seen as distant and detached.

Dissociative hallucination process model

We would suggest that individuals undergo two psychological processes when experiencing dissociative hallucinations (Fig. 1). The first is a loss of integration of consciousness, with either detachment (the subjective experience of an altered mental state) or compartmentalisation (a failure to control that which is normally under voluntary control), as described in Cardena's original model of dissociation.^{Reference Cardena, Lynn and Rhue24} The second process might involve the involuntary intrusion of sensory, affective or cognitive content into conscious awareness, in order to describe the experience of dissociative hallucinations.^{Reference Steel, Fowler and Holmes6,Reference Krause-Utz, Frost, Winter and Elzinga25} These intrusions could be trauma specific, as in dissociative flashbacks, or general, as in dissociative hallucinations. We would see dissociative hallucinations as occurring on a continuum from the experience of internal voices to external, uncontrollable, ego-alien negative voices. Voices may begin with a link to negative self-talk and memories of abuse but can lose that link as anchors fade.

Fig. 1 Mechanisms of symptom experience in dissociation. Adapted from Cardena^{Reference Cardena, Lynn and Rhue24} and Steel et al^{Reference Steel, Fowler and Holmes6}.

Limitations

This research has limitations and methodological problems. The subjective experience of dissociation is difficult to operationalise, and despite our efforts to link the symptoms to hearing voices, it is potentially vulnerable to contamination. The use of the DSM-5 definition of trauma in PTSD set specific criteria but potentially leaves out many other trauma experiences. We asked participants specifically about their trauma memories, but recognise that retrospective descriptions of trauma are open to contamination. Individuals with schizophrenia are exposed to trauma, and dissecting comorbidity can be difficult in complex cases. Confounding variables in all participants may have included other diagnoses, such as major depression or generalised anxiety disorder, as only comorbid substance misuse was specifically excluded. It is also important to note that the PSYRATS examines the severity of existing hallucinations, but the process of development of these hallucinations cannot be directly measured.

Sample size was also a significant problem. The small number of participants, especially in the schizophrenia group, affected the statistical power of analyses, particularly when considering correlations among subscales. Small sample size means that sampling variability may affect inferences drawn about the represented population. For example, the participants with PTSD had PSSI-5 scores that represented a severe degree of PTSD symptomology. This increases the likelihood of multiple and complex trauma and means that the data are not necessarily representative of all individuals with PTSD. Group assignment was not random and required clinical accuracy regarding diagnosis, although this problem was mediated by the fact that all the psychiatrists involved had significant (between 10 and 30 years’) clinical experience and were dealing with patients who were clinically known to them.

Clinical and research implications

If experiences of auditory hallucinations have more than a single causative pathway, it is clinically valid to try to distinguish them. The description of hallucinatory experiences can be similar between individuals who have PTSD and schizophrenia, so it is essential to consider contextual variables such as dissociation symptoms. Dissociation is a universal defence mechanism in response to trauma but our groups described differing experiences of it. There is therefore a need to encourage further research into the role of dissociation, particularly amnesia, as a potential mediating factor between hallucinations and trauma.