A cross-sectional study of auditory verbal hallucinations experienced by people with a diagnosis of borderline personality disorder

Background: The presence of auditory verbal hallucinations (AVHs) does not currently feature in the main diagnostic criteria for borderline personality disorder (BPD). However, there is accumulating evidence that a high proportion of BPD patients report longstanding and frequent AVHs which constitute a significant risk factor for suicide plans and attempts, and hospitalization. Aim: This study addressed questions about the validity and phenomenology of AVHs in the context of BPD. The longer-term aim is to facilitate the development and translation of treatment approaches to address the unmet need of this population. Method: This was a cross-sectional study, combining phenomenological and psychological assessments administered in person and online. We explored the experiences of 48 patients with a diagnosis of BPD who were hearing AVHs. Results: Participants gave ‘ consistent ’ reports on the measure of AVH phenomenology, suggesting that these experiences were legitimate. Similar to AVHs in a psychosis context, AVHs were experienced as distressing and appraised as persecutory. AVHs were found to be weakly associated with BPD symptoms. AVHs were also rated highly as a treatment priority by the majority of participants. Conclusion: The findings suggest that AVH is a legitimate and distressing symptom of BPD and a treatment priority for some patients. The relative independence of AVHs from other BPD symptoms and emotional states suggests that psychological treatment may need to be targeted specifically at the symptom of AVHs. This treat-ment could be adapted from cognitive behaviour therapy, the psychological intervention that is recommended for the treatment of AVHs in the context of psychosis.

symptoms and emotional states suggests that psychological treatment may need to be targeted specifically at the symptom of AVHs.
• Psychological treatment of AVHs in the context of BPD could be adapted from cognitive behaviour therapy.

| INTRODUCTION
Borderline personality disorder (BPD) affects 1-3% of the general population (Lenzenweger et al., 2007).Patients given a diagnosis of BPD characteristically experience emotional instability, repeated selfinjury, suicidal tendencies, reactive aggression and high rates of comorbid mental disorders (Leichsenring et al., 2011).BPD markedly affects employment and, in combination with high use of mental health care, results in high societal costs (van Asselt et al., 2007).
The presence of auditory verbal hallucinations (AVHs) does not currently feature in the main diagnostic criteria for BPD (e.g., Diagnostic and Statistical Manual of Mental Disorders-fifth edition [DSM-5;American Psychiatric Association, 2013]).However, there is accumulating evidence that a high proportion of BPD patients (50-90%) report longstanding and frequent AVHs (Kingdon et al., 2010;Yee et al., 2005).
AVHs, defined as an auditory or speech experience in the absence of an actual stimulus (David, 2004), have typically been studied in the context of psychosis, where these experiences often involve hearing one or more AVHs.These AVHs are typically negative and critical in content and may comprise commands to harm oneself or others (McCarthy-Jones, Trauer, et al., 2014 ).When experienced in the context of BPD, AVHs constitute a significant risk factor for suicide plans and attempts, and hospitalization (Miller et al., 1993;Slotema et al., 2016).
Studies directly comparing the experiences of AVH in the contexts of psychosis and BPD have found no differences in terms of perceived location (Kingdon et al., 2010;Slotema et al., 2012;Tschoeke et al., 2014), content (including negativity, Kingdon et al., 2010;Slotema et al., 2012), frequency or duration (Kingdon et al., 2010;Slotema et al., 2012) or emotional impact (Slotema et al., 2012).Despite these recent advances, the nature and legitimacy of AVHs in the context of BPD is still being questioned (Merrett et al., 2016).
Questions that have repeatedly arisen over past decades in relation to the experience of AVHs in BPD include the following: (1) to what extent are the AVHs experienced by patients with BPD 'true' hallucinations in the sense of those described in the context of psychosis; and (2) to what extent do these experiences reflect 'malingering' on the part of the patient (Yee et al., 2005)?Where AVHs are accepted as a legitimate experience, some authors purport that AVHs in the context of BPD are qualitatively different to those experienced in the context of psychosis (Zanarini et al., 2013).Furthermore, the body of research exploring AVH phenomenology in the context of BPD remains small (Merrett et al., 2016;Niemantsverdriet et al., 2017) and lags behind developments in the psychosis literature, in which comprehensive quantitative approaches and analytic methods have been applied in order to further characterize the nature of these experiences.These studies have suggested the presence of different AVH 'subtypes' (McCarthy-Jones, Trauer, et al., 2014;Stephane et al., 2003), which may reflect different underlying cognitive and neural mechanisms (Jones, 2010;McCarthy-Jones, Thomas, et al., 2014).There have been calls for the clear characterization of AVH phenomenology within individual studies so that these features can be applied to neuroimaging data (Allen et al., 2012).Similarly, it has been suggested that the application of these approaches to populations other than psychosis is critical for developing understanding of transdiagnostic mechanisms of AVHs (Davies et al., 2020;Waters & Fernyhough, 2017).The present study drew upon neuroimaging and phenomenological methods to address these gaps in the literature and inform discussions about the validity and phenomenological profile of AVHs in the context of BPD.
A range of evidence-based intervention options exist for the treatment of distressing AVHs occurring in the context of psychosisincluding antipsychotic medication and cognitive behaviour therapy (CBT; National Collaborating Centre for Mental Health, 2014).However, these approaches are not currently recommended for patients experiencing AVHs in the context of BPD (National Collaborating Centre for Mental Health, 2009).A further aim of this study was to explore the relevance of the cognitive model of voices (Birchwood & Chadwick, 1997) to AVHs in the context of BPD.Studies have suggested that beliefs about self and AVHs, and associated behavioural and emotional responses to AVHs do not differ across psychosis and BPD contexts (e.g., Hepworth et al., 2013).If these findings can be corroborated, CBT may also be an appropriate treatment for distressing AVHs in the context of BPD.Exclusion criteria included criteria related to the neuroimagingbeing afraid of small, closed spaces or loud noises, non-removable metal in or on their body; pregnancy; having a diagnosed neurological or neurodegenerative disorder; and meeting DSM-5 diagnostic criteria for schizophrenia or schizoaffective disorder.

Questions about the validity and phenomenology of
The presence of both BPD and schizophrenia/schizoaffective disorder were confirmed during screening using the Borderline Personality Disorder section of the Structured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD) (First et al., 2016) and the Schizophrenia Spectrum or Other Psychotic Disorders section of the Structured Clinical Interview for DSM-5 (SCID-5) (First et al., 2016).The presence and duration of AVHs were confirmed using participant responses to observer-rated questions derived from the SCID-5.

| Measures
Beliefs About Voices Questionnaire-Revised (BAVQ-R; Chadwick et al., 2000) is a 35-item self-report questionnaire that measures beliefs about the malevolence, benevolence and omnipotence of voices, alongside behavioural and emotional responses to voices.A recent factor analytic study (Strauss et al., 2018) identified four subscales: two relating to beliefs about voices (persecutory and benevolent) and two relating to responses to voices (resistance and engagement).The measure shows good internal consistency and validity within a psychosis context (Chadwick et al., 2000).
Brief Core Schema Scale (BCSS; Fowler et al., 2006) is a 24-item self-report questionnaire assessing both negative and positive schemas about self and others.This measure consists of four subscales: 'negative self-schema', 'positive self-schema', 'negative otherschema' and 'positive other-schema'.The measure shows good internal consistency, test-retest reliability and validity within a psychosis population (Fowler et al., 2006;Smith et al., 2006).
Brief Symptom Impact Scale (BSIS) is a 10-item measure of impact of AVHs relative to other symptoms.This aimed to identify whether AVH should be a treatment priority.The questionnaire contains items corresponding to the nine diagnostic criteria for BPD (e.g., identify disturbance, impulsivity, etc.) alongside one item relating to AVHs.Participants ranked these 10 in order of their current negative impact (in terms of distress and/or ability to function) where 1 = generally affects me most and 10 = generally affects me least.
Childhood Trauma Questionnaire-Short Form (CTQ-SF; Bernstein & Fink, 1998) is a 28-item self-report questionnaire with strong internal consistency and test-retest reliability.Factor analytic studies have indicated the presence of five subscales: emotional abuse, physical abuse, sexual abuse, emotional neglect and physical neglect.
Computerized binary scale of auditory speech hallucinations (cbSASH; Stephane et al., 2006) is a 159-item computerized assessment providing a fine-grained assessment of AVH phenomenology along with measures of the reliability and consistency of the patient report.The reliability and consistency subscales have demonstrated good convergent validity with similar measures, and the phenomenology and reliability subscales have demonstrated high internal consistency (internal consistency does not apply for the consistency subscale).Furthermore, the phenomenology subscale has high testretest reliability (Stephane et al., 2006).
State-Trait Anxiety Inventory (STAI; Spielberger et al., 1983) is a commonly used measure of trait and state anxiety, comprising 20 items for assessing trait anxiety and 20 for state anxiety.Internal consistency coefficients for the scale have ranged from 0.86 to 0.95; test-retest reliability coefficients have ranged from 0.65 to 0.75 over a 2-month interval (Spielberger et al., 1983).
The Positive and Negative Syndrome Scale (PANSS) for Schizophrenia-Positive Syndrome subscale (PANSS-P; Kay et al., 1987) is one of the most commonly used clinician-rated scales to assess for the presence and severity of positive symptoms.The PANSS scales have demonstrated excellent internal consistency (coefficient alpha and split-half reliability) and consistency over time (testretest reliability) while still being sensitive to change.
Zanarini Rating Scale for Borderline Personality Disorder: Selfreport Version (ZAN-BPD: SRV; Zanarini et al., 2015) is a nine-item self-report measure assessing the severity of BPD symptoms over the past week.It consists of a five-level set of anchored rating points for each of the nine criteria for BPD found in the DSM-5.This scale has demonstrated good convergent validity with interview-based measures, good internal consistency and excellent test-retest reliability (Zanarini et al., 2015).

| Procedure
The study involved three separate assessment phases, which took approximately 5.5 h in total.In phase 1a, participants completed a series of clinical outcome measures (observer rated and self-report) during a face-to-face meeting with a researcher.Phase 1b involved the completion of self-report clinical outcome measures online, with remote assistance offered as necessary.In phases 2 and 3, participants attended a scanning centre to complete an experimental task (signal detection task) and neuroimaging procedures.

| Statistical analysis
Descriptive summaries were conducted for the participant characteristics in terms of mean (μ), standard deviation (SD), median, 25th/75th percentile, interquartile range, minimum and maximum for continuous data, and count (n) and percentage (%) for categorical data.Pearson correlations (ρ n ) were used to explore relationships between pairs of measures.Statistical tests were significant if p < 0.05.As these analyses are exploratory, and we do not intend to recommend direct changes to clinical practice based on our results, there will be no adjustment for multiple testing.All analyses were calculated with SPSS V25.Missing data were summarized but not treated.

| Participant data and characteristics
A total of 52 BPD patients participated in the study.Three participants withdrew before taking part in the first phases and so did not provide any data.Forty-eight participants took part in phases 1a and 1b and one person took part in phase 1b only.In terms of data completeness, participant characteristics were available for all 48 participants; for individual clinical measures, the level of completeness ranged from 62% to 92% and can be deducted from Tables 1-3.
Patients were mainly white British (96%) females (81%) with an average age of 34 years (SD = 10.9;range 19 to 56).Almost two thirds (65%) were single.Over half (54%) were not working due to long-term sickness or disability, 15% were unemployed, 10% were not working for other reasons (e.g., student, retired and looking after family) and 21% were in paid full-time employment.Six percent had no formal qualificaitons, around a third (33%) left school at 16, 38% went to college and just over 1 in 5 (23%) went to university.The majority of participants had experienced childhood trauma in the form of Emotional Abuse (94%), Physical Abuse (63%), Sexual Abuse (67%), Emotional Neglect (85%) and Physical Neglect (71%) (see Table 1 for a summary of the measures).These high levels of reported trauma were in line with expectations.All but two participants (a total of 96%) reported experiencing PTSD.

| AVH experiences
The cbSASH was used to explore the extent to which patients' AVH experiences were atypical.Total scores ranged from 0 to 30 where >7 indicates malingering.Two (4.4%) patients were deemed to be malingering (both with scores of 8).Forty-six (93.9%) participants gave a 'consistent' report of their AVH experiences (i.e., not considered to be malingering).
T A B L E 2 Percentage of participants responding affirmatively to selected items from the computerized binary scale of auditory speech hallucinations (cbSASH) assessment of AVH phenomenology (N = 45)  which was unfamiliar (31% reported familiarity) and attributed the cause of this experience to having a mental health problem (69%).
The location of AVHs was perceived to be both inside (51%) and outside of the head (49%).AVHs often talked to the participant (93%) and told them what to do (78%), including commands to self-harm (76%).The participants responded by talking back to AVHs (69%) and engaging in reciprocal conversations (60%) but complied with AVH commands infrequently (29%).
Table 3 displays
Scores on the BCSS (Table 3) indicated that patients tended to have a higher degree of belief in the negative descriptions of themselves compared to the positive characteristics presented (BCSS scores were μ = 14.4,SD = 7.0 and μ = 11.6,SD = 9.2, respectively).
Patients also reported higher levels of negative characteristics in other people (μ = 12.2, SD = 8.8) and were less likely to endorse positive attributes in others (μ = 8.9, SD = 5.8).
Associations between the range of AVH characteristics (PSYRATS-AH sub-scores) and behavioural responses (BAVQ and BCSS sub-scores) were explored to gain a better understanding of potential distress-maintenance patterns (Table 4).Inspection of the relationships between AVH distress and the different dimensions of beliefs about AVHs revealed a strong positive correlation with persecutory beliefs (ρ 27 = 0.67; p < 0.001) and a weak positive relationship with resistance (ρ 27 = 0.45; p = 0.019).There was a strong relationship between believing AVHs were persecutory and resisting the AVHs (ρ 33 = 0.75; p < 0.001).Believing the AVHs were benevolent was strongly correlated with engagement with AVHs (ρ 33 = 0.83; p < 0.001).However, there was no evidence to suggest that either beliefs about AVH benevolence, engagement with AVHs or any of the domains of the BCSS were associated with AVH distress.

| Perceived impact of AVHs and priority for treatment
Using the BSIS, participants were asked to rank their symptoms, which included AVHs, in order of impact.Overall, 'hearing a voice/ voices that others could not hear' was ranked fourth highest of the 10 symptoms, behind feelings of moodiness, emptiness and angry feelings or acts.AVHs were seen as having the most impact by 17% of patients; over a third (35%) ranked AVHs among the top three impacts.Figure 1 displays the percentages of participants who ranked symptoms in their top five for impact; AVHs was ranked in the top five by more than two thirds (69%) of participants.

| DISCUSSION
This study was an exploration of the phenomenological and psychological profile of AVHs for patients with BPD.With a better understanding of the legitimacy of AVHs, how this experience relates to other symptoms of BPD and the cognitive behavioural mechanisms of AVHs, further consideration can be given as to how the treatment needs of this population can be met.If BPD patients report being distressed by AVHs in ways that are similar to psychosis patients, then arguments can be made to explore the translation of the range of evidence-based intervention options.
The participants within this study tended to be relatively young females who were not in a relationship, and over half were not working due to long-term sickness or disability.The majority of participants had experienced significant childhood trauma, and almost all reported emotional abuse, PTSD and high levels of anxiety.In these respects, the participants seemed to be typical of patients given a diagnosis of BPD.This conclusion was corroborated by the profile of high scores relation to the high volume of commands and the perceived location of AVHs (both inside and outside of the head).
With regard to AVH characteristics and distress (measured by PSYRATS), scores were marginally lower within the current study, when compared to a large sample of psychosis patients (Craig et al., 2018).The opposite was the case for appraisals of AVH (using the BAVQ-R), as AVHs were appraised in the current study as persecutory to a degree that was marginally higher in comparison to a recently completed trial involving psychosis patients (Hayward et al., 2021).Appraisals of self differed more markedly as participants within the current study reported higher scores for both negative and positive appraisals, in comparison to the Hayward et al. (2021) psychosis sample.This pattern of some variation was evident when associations were explored between the characteristics, distress and beliefs about AVHs, and beliefs about self.Consistent with psychosis samples (Mawson et al., 2010), significant associations were found between persecutory beliefs about AVHs and AVH distress, and resistance to AVHs and AVH distress.However, contrary to findings from a transdiagnostic sample (Cole et al., 2017), no associations were found between beliefs about self and AVH distress.Studies exploring associations between beliefs about self and AVHs within specific diagnostic groups have foregrounded associations with depression (e.g., Cavelti et al., 2020;Fannon et al., 2009;Peters et al., 2012).The role of beliefs about self in the maintenance of AVH distress should be a focus of future research.
The exploration of associations between AVH and BPD symptoms and AVHs and anxiety generated only weak associations, suggesting that AVHs may be somewhat independent from the affective symptoms that are characteristic of this patient group.
Final consideration was given to the perceived impact of AVHs and the priority afforded by participants to the treatment of AVHs.
Consistent with the high level of distress reported on PSYRATS, AVHs were rated highly in terms of impact by the majority of participants.
Indeed, a higher level of impact was afforded to AVHs in comparison to some of the other symptoms which are typically the focus of inter- This study has limitations in several respects.Firstly, despite being larger than many existing studies in the literature, the sample was relatively small and was not adequately powered to detect all statistically meaningful findings.Secondly, there was a considerable amount of missing data for some variables, which may have been attributable to the study's ambitious focus upon multiple research questions.Thirdly, the correlational nature of the analyses does not facilitate exploration of the causal relationship between variables.Finally, the absence of a comparison group with a psychosis diagnosis prevented direct comparisons of AVH across diagnoses.
The findings from this study suggest that AVH is a legitimate and distressing symptom of BPD and a treatment priority for some patients.The relative independence of AVHs from other BPD symptoms and emotional states suggests that psychological treatment may need to be targeted specifically at the symptom of AVHs.
Single-symptom forms of evidence-based CBT are showing promise in the treatment of AVHs in the context of psychosis F I G U R E 1 BSIS ratings for impact of symptoms.Notes: Participants were asked to rate 10 symptoms in order of impact.Every time a symptom appeared in a top five, it was counted.The figure displays the proportion of participants for whom the symptom appeared in the top five (Lincoln & Peters, 2019).These interventions focus primarily upon the re-evaluation of persecutory beliefs about AVHs-a variable which this study found to be prominent and associated with AVH distress for BPD patients.CBT for AVHs in the context of psychosis additionally focuses upon variables which this study found were less relevant to AVHs in the context of BPD (e.g., beliefs about self -Hazell et al., 2018).Furthermore, variants of CBT for AVHs are targeting relational variables (e.g., Hayward et al., 2017) which may be pertinent to BPD patients but were not assessed within this study.These differences suggest that: (1) an adequately powered comparison of AVHs in the context of psychosis and BPD is required; and (2) CBT for AVHs may benefit from being adapted prior to evaluation with BPD patients.
AVHs in the context of BPD hinder the development and translation of treatment approaches to address the unmet need of this population.To address these questions, we conducted: (1) a comprehensive examination of the phenomenological and cognitive behavioural mechanisms of AVH in BPD; and (2) a neuroimaging-based exploration of the neural networks operating in real time during an AVH episode.Here, we focus upon the examination of the experiences of a group of BPD patients who reported hearing AVHs; the neuroimaging-based exploration will be published elsewhere.2 | METHODS 2.1 | Design and participants This was a cross-sectional study, combining neuroimaging methods and phenomenological and psychological assessments administered in person and online.Findings from the neuroimaging part of the study will be reported in a separate paper.Patients with a diagnosis of BPD were recruited from the mental health services of two Trusts within the UK National Health Service (NHS)-Sussex Partnership NHS Foundation Trust (including the Sussex Voices Clinic, https://www.sussexpartnership.nhs.uk/sussex-voices-clinic) and Kent and Medway NHS and Social Care Partnership Trust.Inclusion criteria required participants to be aged 18-65; be right-handed; be fluent in speaking and reading English; have normal to corrected vision and normal hearing (no clinical deafness or hearing impairment); have received a clinician-administered diagnosis of BPD; have heard AVHs in the past week; and have been experiencing persisting AVHs for the past 6 months.
Ethical approval was obtained from South Central Berkshire 'B' Research Ethics Committee via the National Research Ethics System ID 234904.The study sponsor was Sussex Partnership NHS Foundation Trust.The study was also approved by the Brighton and Sussex Medical School Research Governance and Ethics Committee.All data acquisition methods used were in accordance with international, national and institutional guidelines.All participants gave informed consent following Declaration of Helsinki guidelines.
vention, for example, self-harm and unstable relationships.The National Collaborating Centre for Mental Health (NICE, 2009) foregrounds the 'autonomy and choice' of BPD patients and encourage the consideration of different treatment options.Requests by patients for a focus upon the treatment of AVHs may be perceived as atypical but should nonetheless be respected.

Table 2
displays the percentage of patients who reported experiencing selected items from the cbSASH phenomenology subscale.Most participants reported hearing more than one AVH (56%)