Auditory Hallucinations, Depressive Symptoms, and Current Suicidal Ideation or Behavior Among Patients with Acute-episode Schizophrenia

Abstract

Suicide risk and auditory hallucinations are common in schizophrenia, but less is known about its associations. This cross-sectional study aimed to determine whether the presence and severity of auditory hallucinations were associated with current suicidal ideation or behavior (CSIB) among patients with schizophrenia. We interviewed 299 individuals with schizophrenia and acute symptoms and reviewed their medical records. Measurement included the Psychotic Symptom Rating Scale (PSYRATS-AH), the Calgary Depression Scale for Schizophrenia (CDSS), and the Positive and Negative Syndrome Scale. Logistic regression and path analysis were used. The CSIB prevalence was higher among patients with current auditory hallucination than those without (19.5% vs. 8.6%, crude odds ratio = 2.58, p = .009). Lifetime auditory hallucination experience (adjusted odds ratio [AOR] = 3.81; 95% CI: 1.45–10.05) or current auditory hallucination experience (AOR = 3.22; 95% CI: 1.25–8.28) can elevate the likelihood of CSIB while controlling for depressive symptoms and lifetime suicide-attempt history. Among those with auditory hallucinations, the emotional score of the PSYRATS-AH was positively associated with the CDSS score and there was a small indirect effect of the CDSS score on the association between the emotional domain score and CSIB (bias-corrected 95% CI, 0.02–0.20). In conclusion, the presence of auditory hallucinations was strongly associated with CSIB, independent of depressive symptoms and lifetime suicide attempts. Suicide risk assessment should consider auditory hallucination experience and patients’ appraisal of its emotional characteristics. Future cohort studies are necessary to provide more conclusive evidence for the mediating pathways between auditory hallucinations and CSIB.

Globally, approximately one in four individuals with schizophrenia have attempted suicide, and 4%–20% die by suicide (Ko et al., 2018; Lu et al., 2019; Ran, Mao, Chan, Chen, & Conwell, 2015), making suicide prevention a key component for the treatment and recovery of schizophrenia (Foster, 2015). Finding those with high suicide risk and exploring the risk factors is important. Previous studies have demonstrated that psychotic experience often co-occurs with suicidal ideation (SI) or suicide attempt (SA) not only in the general population (Bromet et al., 2017; Capra, Kavanagh, Hides, & Scott, 2015; DeVylder, Lukens, Link, & Lieberman, 2015; Hielscher, DeVylder, Saha, Connell, & Scott, 2018; Nishida et al., 2010; Yates et al., 2019) but also in clinical samples (Bornheimer, Wojtalik, Li, Cobia, & Smith, 2021). Hallucinations in schizophrenia spectrum disorders, including visual and other hallucinations, are also related with recent SI and suicidal plans (Bornheimer, 2019; L. A. Bornheimer & Jaccard, 2017; Kjelby et al., 2015; Madsen & Nordentoft, 2012; Wong et al., 2013).

However, increased understandings are needed about the association between auditory hallucinations and suicide risk in schizophrenia. Auditory hallucinations are sensory experiences of voices or non-verbal sounds (e.g., ringing) without external stimulation (de Leede-Smith & Barkus, 2013). As a prominent psychotic symptom (Lim, Hoek, Deen, & Blom, 2016), auditory hallucinations often contain negative content, including devaluation and commands to harm oneself or others (Laroi et al., 2012). Command auditory hallucinations, which are unique in psychosis, are associated with an increased risk of recent SI or SA (Wong et al., 2013). In community settings, adolescents with auditory hallucinations also were found to have an increased risk for future non-suicidal self-injury (NSSI) and SA within the next two years (Hielscher et al., 2020). However, these previous studies had limitations in controlling confounding factors such as depressive symptoms, history of suicide attempts, and the severity of other positive symptoms. In addition, researchers summarized several characteristics of auditory hallucinations, including negative content, frequency, and emotional valence (Haddock, McCarron, Tarrier, & Faragher, 1999). Differentiating the features of auditory hallucinations according to the above criteria may advance our understanding about which characteristic of auditory hallucinations can elevate the suicide risk. However, only one UK team reported that the severity of recent SI was positively correlated with negative emotional reactions to auditory hallucinations among those with psychosis (Fialko et al., 2006). The limitations of this study were that they used secondary data from a clinical trial cohort with high potential recruitment bias and did not consider other explanatory variables such as depressive symptoms. Then, findings among those without schizophrenia also suggest the value to explore the features of auditory hallucinations and suicide risk. One community study showed that only distressed voice-hearing experience was associated with adolescents’ lifetime self-harm, but the non-distressing voice was not (Løberg, Gjestad, Posserud, Kompus, & Lundervold, 2019).

Moreover, studies have demonstrated that the severity of auditory hallucinations is positively associated with depressive symptoms among patients with schizophrenia (Steel et al., 2007; Wang, Beckstead, & Yang, 2019). The cognitive theory of auditory hallucinations also argues that individuals hearing voices during psychotic episodes are vulnerable to elevated depressive moods (Chadwick & Birchwood, 1994; Mawson, Cohen, & Berry, 2010). Meanwhile, individuals with depressive symptoms frequently use negative cognitive appraisal in their social life and this social schema may be applied to the appraisal of voices (Paulik, 2012), generating negative emotional reactions to auditory hallucinations.

Since the depressive symptom also strongly predicts suicide and SA (Cassidy, Yang, Kapczinski, & Passos, 2018), it is necessary to consider its function in the association between psychotic experience and suicide thoughts (Honings, Drukker, Groen, & van Os, 2016). Previous community-based studies reported that depressive symptoms mediated or moderate psychotic experience and SI (Bornheimer et al., 2019; Jang et al., 2014), subsequent SA (Kelleher et al., 2013), or self-harm (Hielscher et al., 2019). Especially, adolescents with auditory hallucinations have more depressive symptoms (Hielscher, Connell, et al., 2018). Although the issue has been examined, there remain limited understandings of the relationships between suicide and schizophrenia. Especially, none study discusses the mediation role of depressive symptoms between auditory hallucinations and suicidality among individuals with schizophrenia.

Due to the high suicide rate and limited understanding of suicide risk among individuals diagnosed with schizophrenia, it is needed to examine suicide risk to develop targeted suicide prevention and reduce suicide risk in schizophrenia. To fill the research gap, we conducted a cross-sectional study to test: (1) whether auditory hallucinations (lifetime presence/ current presence/ severity) were significantly associated with current suicidal ideation or behavior (CSIB) in schizophrenia; (2) whether auditory hallucinations (lifetime presence/ current presence/ severity) were significantly associated with depressive symptoms and whether depressive symptoms mediated associations between auditory hallucinations and CSIB among individuals with schizophrenia if there were. Findings would help identify high-risk individuals with schizophrenia for early suicide intervention.

Methods

Subjects were recruited from inpatients at Beijing HuiLongGuan Hospital, a psychiatric hospital with approximately 1,300 beds in a northern city of China.

Patients and/or their guardians (for those under 18 years old or lack of capability to consent) provided written consent. This study was approved by the Institutional Review Boards of the Beijing HuiLongGuan Hospital, Beijing, China.

Participants

Psychiatrists recruited potential participants and their guardians, presented information about the study, and explained the benefits and risks. Psychiatrists screened the patients who had a wish to participate. Trained psychiatrists in the research team confirmed the inclusion and exclusion criteria. Patients were included if they were between 16 and 60 years old, diagnosed with schizophrenia based on the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV ;American Psychiatric Association, 1994), currently experiencing psychosis symptoms, and if they had exposure to antipsychotic medication for less than two weeks in this episode. Currently experiencing psychosis symptoms was defined as any symptom of participants was rated more than three in any item on the positive symptom subscale of the Positive and Negative Syndrome Scale (PANSS; He & Zhang, 2000; Kay, Fiszbein, & Opler, 1987). Exclusion criteria included: a diagnosis of organic brain disorders, organic auditory or visual perception disorders, current comorbidity with other Axis I mental disorders, unstable major medical conditions (e.g., cancer), significant adverse reactions to medicines, pregnancy, or lactation.

Measurements

Trained psychiatrists administrated all the interviews.

Current experience of auditory hallucinations was defined as hearing voices or non-verbal sounds when conscious and not under the influence of external stimulations, drugs, alcohol, physical illness, or hypnagogic therapy within the week of the interview (Kelleher et al., 2014). First, psychiatrists asked participants whether they had lifetime auditory hallucinations and when the auditory hallucinations occurred (within seven days/ seven days before/ never). Second, the subjective severity of auditory hallucinations within seven days was assessed using the auditory hallucinations rating subscale of the Psychotic Symptom Rating Scale (PSYRATS-AH ;Haddock et al., 1999; Xu, Li, Guo, Chen, & Zhang, 2012). The subscale contains 11 items covering 3 domains: physical characteristics (frequency, duration, location, loudness), emotional characteristics (amount and degree of negative content, amount and intensity of distress), and cognitive interpretation (beliefs about the origin of voices, disruption to life caused by voices, controllability of voices; Haddock et al., 1999). All items are rated with a five-point ordinal scale (0–4); higher scores indicate more severe symptoms. In the current study, Cronbach’s α for each domain was .53, .82, and .69. Moreover, the presence of command auditory hallucinations was recoded from the item 7 “degree of negative content” of the PSYRATS-AH (1 to 3 = absence; 4 “personal threats to self, e.g., threats to harm self or family, extreme instructions or commands to harm self or others” = presence).

Depressives symptoms were evaluated through the Calgary Depression Scale for Schizophrenia (CDSS; Addington, Addington, & Maticka-Tyndale, 1993; Liu et al., 2009). The scale measures depressive symptoms in the last two weeks using nine items, each with ratings from 0 (absent) to 3 (severe). This study calculated the total CDSS score as the sum of all items except item 8, which measured suicidality. Higher depression scores mean greater symptoms of depression. In our study, the Cronbach’s α was .88.

CSIB was assessed through item 8 of the CDSS, which measures suicide thoughts and behavior within two weeks. Scores were 0 for no suicidality, 1 for the presence of SI, 2 for a suicidal plan, and 3 for a SA. The presence of CSIB was coded as 1 (yes) if the item score was >0 and 0 (no) if the item score was 0.

Lifetime SA was defined as a previous nonfatal act of deliberate self-harm with intent to die regardless of the outcome or treatments. The information was collected through interviewing with the patients and family informants or extracting from suicide risk management records and other medical records.

Positive symptoms were assessed using the PANSS (He & Zhang, 2000; Kay et al., 1987). The seven items in the positive symptom subscale, rated on a 7-point scale (1 = absent; 7 = extreme), measure delusion, conceptual disorganization, hallucinatory behavior, excitement, grandiosity, suspiciousness/persecution, and hostility. The inter-observer correlation coefficient was > .80. In this study, we used the sum score of the six items except “P3 hallucinatory behavior” as a confounder. The total subscale score ranges from 6 to 42 with a higher score indicating greater severity of positive symptoms.

Statistical Analysis

The group differences in demographic and clinical characteristics were determined using χ² tests or Fisher’s exact tests for categorical variables. For variables that violated normality, Mann-Whitney U rank-sum test was performed to determine trends between the two groups.

First, we ran a multivariable binary logistic regression to explore whether auditory hallucinations (lifetime presence/absence; current presence/absence; three-domain scores) and the CDSS score predicted CSIB (with /without). Second, we examined associated factors for the CDSS score using multivariable linear regression. The model was controlled for age, sex, lifetime SA, use of antidepressants, positive symptoms, and duration of illness. Furthermore, we confirmed the mediation effect size through path analysis, using a generalized structural equation model (GSEM) where the outcome variable (CSIB) was dichotomous. The best-fitting model was determined using the Akaike information criterion (AIC) or the Bayesian information criterion (BIC). A lower AIC score or BIC score indicates a better-fitting model. Bias-corrected confidence intervals for all direct and indirect pathways were calculated through 500 replicated bootstraps. Multicollinearity was checked through a variance inflation factor (VIF) less than 3.0 for all variables.

Statistical analyses were performed in Stata 15.0 for Windows (Stata Corp, College Station, TX, USA). Significance level was set at two-side p-value < .05.

Results

Demographics and Clinical Characteristics

Of the original 323 participants, 22 were without positive symptoms now, and 2 did not finish every item in the CDSS. Thus, the statistical analysis about the experience of auditory hallucinations included 299 samples. For three participants with missing items (less than two) in the PSYRATS-AH, we imputed missing values with the median score.

We identified 14.3% of participants with CSIB (29 suicidal ideation, 8 suicidal plan, 6 attempted suicide). Those with or without CSIB had no difference in age, sex, education years, duration of illness, and antipsychotic use (Table 1). However, the proportion of antidepressant use was higher in those with CSIB than those without (18.6% vs. 2.3%, p < .001). Medication use was detailed in Supplemental Table S1.

Table 1.

The Characteristics of Patients with Schizophrenia by the Presence of Current Suicidal Ideation or Behavior (N = 299).

Variables	Current suicidal ideation or behavior		Statistics	p
	Without (n = 256)	With (n = 43)
	Median (IQR)	Median (IQR)	z
Age	31(21)	29(21)	0.20	.842
Education years	12(9)	11(9)	−0.08	.935
Duration of illness, years	1(4.5)	2(4.5)	−0.58	.561
PSYRATS-AH a
Physical characteristics	10(3)	10(4)	0.74	.459
Emotional characteristics	8(5)	9(6)	−0.23	.822
Cognitive interpretation	9(3)	8(4)	1.29	.198
CDSS	1(4)	9(9)	−8.21	< .001
PANSS-P	19(7)	18(5)	0.409	.683
	n (%)	n (%)	χ2
Male	129 (50.4)	26 (60.5)	1.50	.221
Lifetime auditory hallucinations	139(54.3)	33(76.7)	7.59	.006
Current auditory hallucinations	128(50)	31(72.1)	7.22	.007
Prior suicide attempts	18 (7.0)	15 (34.9)	29.09	< .001
				Fisher's exact
Current command auditory hallucinations	19(7.4)	3(7.0)		.608
Using antipsychotics	249(97.3)	41(95.4)		.622
Using antidepressants	6(2.3)	8(18.6)		< .001

Abbreviations: PSYRATS-AH = the auditory hallucinations subscale of the Psychotic Symptom Rating Scale; CDSS = Calgary Depression Scale for Schizophrenia (except the item 8 “Suicidal history”); PANSS-P = the Positive and Negative Syndrome Scale – Positive symptom subscale (except the item “P3 hallucinations”).

Note:

^a.The PSYRATS-AH score was only measured in those with current auditory hallucinations.

Results statistically significant at p < .05 level are in bold.

Auditory Hallucinations and CSIB

The bivariate analysis showed that higher proportions of lifetime auditory hallucination experience and current auditory hallucination experience were reported by patients with CSIB than those without (Table 1). CSIB prevalence was higher among patients with current auditory hallucination experience than those without (19.5% vs. 8.6%, crude odds ratio = 2.58, p = .009). The difference between those with or without CSIB in the three domain scores of the PSYRATS-AH was nonsignificant (p > .05, Table 1). Moreover, those with or without command auditory hallucinations reported similar rates of CSIB (3/22 vs. 28/137, Fisher’s exact p = .572).

After controlling for the CDSS scores (exclude item “suicidality”), lifetime suicide-attempt history, and other confounders, the multivariable logistic regression demonstrated that lifetime auditory hallucination experience (adjusted odds ratio [AOR] = 3.81; 95% CI: 1.45–10.05) or current auditory hallucination experience (AOR = 3.22; 95% CI: 1.25–8.28) can elevate the risk of CSIB (Table 2, Model 1 and Model 2). Lifetime SA history and higher CDSS scores were robustly associated with increased risk of CSIB.

Table 2.

The Presence and Severity of Auditory Hallucinations with Current Suicidal Ideation or Behavior Among Those with Schizophrenia.

	Model 1		Model 2		Model 3		Model 4		Model 5
Variables	AOR(95% CI)	p	AOR(95% CI)	p	AOR(95% CI)	p	AOR(95% CI)	p	AOR(95% CI)	p
Age, years	1.04(0.99–1.09)	.119	1.04(0.99–1.09)	.117	1.03(0.97–1.09)	.314	1.03(0.97–1.09)	.326	1.03(0.97–1.09)	.333
Male	2.77(1.13–6.79) b	.026	2.74(1.11–6.73)	.028	3(0.98–9.17)	.055	3.49(1.09–11.17)	.035	3.24(1.02–10.26)	.046
Life-time auditory hallucinations	3.81(1.45–10.05)	.007
Current auditory hallucinations			3.22(1.25–8.28)	.015
PSYRATS-AH score a
Physical domain					0.95(0.78–1.14)	.560
Emotional domain							0.87(0.76–1.00)	.055
Cognitive domain									0.83(0.68–1.00)	.052
CDSS	1.30(1.19–1.42)	<.001	1.29(1.18–1.42)	<.001	1.35(1.18–1.54)	< .001	1.40(1.21–1.61)	< .001	1.38(1.2–1.59)	< .001
Prior suicide attempts	4.98(1.91–13.03)	.001	4.73(1.81–12.32)	.001	3.94(1.24–12.57)	.020	4.63(1.4–15.33)	.012	4.39(1.34–14.39)	.015
PANSS-P	1.00(0.91–1.09)	.978	1(0.91–1.09)	.941	0.97(0.87–1.08)	.594	0.97(0.87–1.09)	.636	1.00(0.89–1.12)	.950
Antidepressants use	6.41(1.32–31.09)	.021	6.29(1.30–30.40)	.022	0.76(0.07–8.09)	.819	0.68(0.07–6.83)	.740	0.96(0.09–9.94)	.973
Illness duration, years	0.97(0.92–1.03)	.288	0.97(0.92–1.03)	.386	0.99(0.92–1.07)	.860	0.98(0.91–1.07)	.712	0.99(0.91–1.07)	.754
n	299		299		195		195		195

Abbreviations: PSYRATS-AH = the auditory hallucinations subscale of the Psychotic Symptom Rating Scale; CDSS = Calgary Depression Scale for Schizophrenia (except the item 8 “Suicidal history”); PANSS-P = the Positive and Negative Syndrome Scale – Positive symptom subscale (except the item “P3 hallucinations”); AOR = adjusted odds ratio, 95% CI = 95% confidence intervals.

Note:

^a.The PSYRATS-AH score was only measured in those with current auditory hallucinations.

^b.Results statistically significant at p < .05 level are in bold.

Then, we repeated the logistic regression and found that each domain score of the PSYRATS-AH was not associated with the presence of CSIB among those with current auditory hallucinations (Table 2, Models 3–5).

Auditory Hallucinations and Depressive Symptoms

Multivariable linear regression did not find that the lifetime or current presence of auditory hallucinations was significantly correlated with total CDSS score (except item “suicidality”) after adjusting for the age, sex, prior SAs, positive symptoms, antidepressants use, and the duration of illness (Table 3, Model 1 & Model 2). For each domain within the PSYRATS-AH, only the emotional characteristic domain score of the PSYRATS-AH was positively correlated with the CDSS score among those with current auditory hallucinations (adjusted B = 0.27, p = .003, Table 3 Model 4); however, the other two domain scores of the PSYRATS-AH score were not associated with the CDSS score.

Table 3.

The Presence and Severity of Auditory Hallucinations with the Score of the Calgary Depression Scale for Schizophrenia Among Those with Schizophrenia. ^a

Variables	Model 1		Model 2		Model 3		Model 4		Model 5
	B	p	B	p	B	p	B	p	B	p
Age, years	−0.03	(.272)	−0.03	(.288)	−0.02	(.534)	−0.02	(.516)	−0.02	(.544)
Male	−0.51	(.316)	−0.52	(.308)	−0.60	(.398)	−0.77	(.264)	−0.63	(.376)
Life-time auditory hallucinations	0.34	(.494)
Current auditory hallucinations			0.47	(.351)
PSYRATS-AH score b
Physical domain					−0.05	(.673)
Emotional domain							0.27	(.003)
Cognitive domain									0.04	(.779)
Prior suicide attempts	2.69 c	(.001)	2.66	(.001)	3.14	(.002)	2.79	(.005)	3.16	(.002)
PANSS-P	0.03	(.573)	0.03	(.606)	0.02	(.765)	−0.01	(.832)	0.01	(0.874)
Antidepressants use	4.48	(< .001)	4.48	(< .001)	6.62	(< .001)	6.96	(< .001)	6.63	(< .001)
Illness duration, years	−0.02	(.524)	−0.02	(.570)	−0.02	(.755)	0.01	(.858)	−0.01	(.782)
n	299		299		195		195		195

Abbreviations: PSYRATS-AH = the auditory hallucinations subscale of the Psychotic Symptom Rating Scale; PANSS-P = the Positive and Negative Syndrome Scale – Positive symptom subscale (except the item “P3 hallucinations”).

Note:

^a.The score of Calgary Depression Scale for Schizophrenia excluded the score of item 8 “Suicidal history”.

^b.The PSYRATS-AH score was only measured in those with current auditory hallucinations.

^c.Results statistically significant at p < .05 level are in bold.

Then, we used path analysis to explore the mediation effect of “Emotional characteristic domain score of the PSYRATS-AH → CDSS score → CSIB.” We compared models and the final model (with the smallest AIC and BIC) only controlled sex, history of suicide attempt, and use of antidepressants. The indirect effect of the CDSS scores was small and significant (indirect effect size, bias-corrected 95% CI, 0.02–0.20) among those with current auditory hallucinations. The total effect (bias-corrected 95% CI, −0.20–0.09) and direct effect (bias-corrected 95% CI, −0.30–0.002) of emotional characteristics domain score on the CSIB were nonsignificant.

Discussion

Prevention of possible SI or SA is critical in treating schizophrenia. This study demonstrated that patients with schizophrenia who experienced auditory hallucinations recently or in their whole life were approximately three times as likely to exhibit CSIB than those without such auditory hallucinations. This effect was independent of depressive symptoms and lifetime SA. Although we did not find a dose-response association between the severity of auditory hallucination and the possibility of CSIB, more negative emotion about auditory hallucinations was related to more severe depressive symptoms.

Auditory Hallucinations and CSIB

The presence of auditory hallucinations was associated with an increased possibility of CSIB while controlling for depressive symptoms and lifetime SA history. The association effect of the lifetime or current auditory hallucinations was similar in magnitude. Our findings further studied the association between the presence of hallucinations and recent suicide thoughts and plans (Bornheimer, 2019; Bornheimer & Jaccard, 2017; Kjelby et al., 2015; Madsen & Nordentoft, 2012; Wong et al., 2013) and differentiated auditory hallucinations. Compared to their research, our strength was that the lifetime SA history was controlled, and we found that the presence of auditory hallucinations raised the average risk of CSIB slightly lower than previous SA. Another novel feature was that we found the presence of auditory hallucinations directly increased the risk of CSIB rather than elevating depressive symptoms. Moreover, the association between the presence of auditory hallucinations and CSIB was in line with findings in general populations (Bromet et al., 2017; Yates et al., 2019) and individuals with other mental disorders (Slotema et al., 2017). Thus, the auditory hallucination experience is a marker for a higher risk of CSIB across groups.

There are several explanations for the association between the auditory hallucination experience and CSIB. First, experiencing psychosis symptoms and engaging with treatment can relate to stress (DeVylder & Hilimire, 2015), and stress may lead to CSIB (DeVylder & Hilimire, 2015). Also, auditory hallucinations can cause greater disruption in daily life (Badcock, Paulik, & Maybery, 2011). Second, command auditory hallucinations increase self-harm behavior (Harkavy-Friedman et al., 2003; Wong et al., 2013). However, the evidence is insufficient enough because both our study and previous studies had small numbers of persons suffering command auditory hallucinations. Moreover, we did not find a bivariate association between command auditory hallucinations and CSIB. Third, emotional childhood trauma predicts negative auditory hallucinations (Scott, Rossell, Meyer, Toh, & Thomas, 2020), and those with trauma always have a stronger risk for SA (Alli et al., 2019). Fourth, neuroimaging studies have identified individuals with schizophrenia and lifetime history of SA had thinning of cortical thickness and decreased gray matter in the temporal region, which is highly associated with auditory hallucinations, compared with patients without SA history (Aguilar et al., 2008; Besteher et al., 2016).

Additionally, we measured the severity of current auditory hallucinations using a structured scale. The findings from the subscale of the PSYRATS-AH did not show a dose-response effect. Especially, a direct association between auditory-hallucination-related negative emotions and CSIB was not found. This result was contrary to expectations and the findings of a study in psychosis (Fialko et al., 2006). The first explanation may be a small sample size because the p-value of emotion domain score in the regression was near 0.05. Second, we only included those with current experiencing auditory hallucinations while analyzing the association between auditory hallucination characteristics and the likelihood of CSIB. Third, self-reported distress related to auditory hallucinations may overlap with depressive symptoms, and our analysis added depressive symptoms as a confounder but previous research did not (Fialko et al., 2006).

Among those with current auditory hallucinations, a poor cognitive interpretation was not found to be associated with CSIB. Our findings support previous evidence on the nonsignificant bivariate association of cognitive-related item score in the PSYRATS-AH and suicide ideation (Fialko et al., 2006). However, there was a strong tendency towards statistical significance that the increased cognitive domain score of the PSYRATS-AH was a protective factor of the CSIB. Higher scores on the cognition interpretation domain of the PSYRATS-AH may suggest the patient has poorer ability in reality testing. Then, severe psychopathology status and cognition dysfunction may limit their thinking on ending their lives (Bredemeier & Miller, 2015). Moreover, according to the “insight paradox”(Lysaker, Roe, & Yanos, 2007), poor insight is associated with fewer depressive symptoms (Belvederi Murri et al., 2016; Belvederi Murri et al., 2015) and less CSIB (Barrett et al., 2015; Bornheimer et al., 2021). The reason may be that increased awareness of their illness at the early stage of illness may bring hopelessness for treatment and life, stigma, and identity loss (Lysaker et al., 2007).

Other Associated Factors of CSIB

Depressive symptoms, robust risk factors for suicide in schizophrenia (Cassidy et al., 2018; Popovic et al., 2014), were associated with an increased risk for CSIB. The association between depressive symptoms and CSIB was consistent with a previous study in outpatients with schizophrenia (Yan et al., 2013). We also showed that a history of SA can increase the risk of CSIB, in line with previous reports (Cassidy et al., 2018; Coentre, Talina, Gois, & Figueira, 2017).

Auditory Hallucinations and Depressive Symptoms

Negative emotions linked to auditory-hallucination were positively associated with depressive mood. The emotional characteristics domain in the PSYRATS-AH, measuring the amount and intensity of negative content or distressing feels caused by auditory hallucinations, were the subjective affective response or “cognitive appraisal” (Mawson et al., 2010). Bivariate association of distress dimension of auditory hallucinations and depressive symptoms has been reported (Steel et al., 2007). New in our study is the finding that the significant association was independent of the history of SAs.

Contrary to previous research (Bornheimer et al., 2019; Jang et al., 2014), we did not find depressive symptoms to be a mediator between auditory hallucinations experience and CSIB. The indirect effect of depressive symptoms on the association between emotional distress of auditory hallucinations and CSIB highlights the importance of distress caused by auditory hallucination. This result was similar to the findings in community samples (Jang et al., 2014; Bornheimer et al., 2019). However, its effect may not be robust enough when we examine the variable's statistical significance and effect size. The mediation analysis was conducted in 159 patients with auditory hallucinations, which may lead to low statistical power. In addition, this preliminary finding using cross-sectional mediation suggests a cohort study be necessary to explore whether depressive symptoms are mediators in the association between emotional appraisal of auditory hallucinations and CSIB.

Implications

The data contributes a clearer understanding of psychotic symptoms and suicidal risk in schizophrenia. When evaluating suicide risk in patients with schizophrenia, we should consider both traditional suicide factors, such as depressive symptoms and SA history, and the occurrence of auditory hallucinations. Using the items related to the emotional characteristic domain in the PSYRATS-AH may therefore benefit suicide risk assessment. These results build on existing evidence of hallucinations and suicide risk among general populations (Kelleher et al., 2013). Moreover, preventing suicidality in patients with schizophrenia should involve shortening untreated durations and using antipsychotics to treat auditory hallucination symptoms. Using clozapine may reduce the risk for recurrent suicidal behavior (American Psychiatric Association, 2021). In addition, therapists can guide patients to change biased emotion appraisal to auditory hallucinations, which may decrease depressive symptoms and suicide risk.

Strengths and limitations

The first strength is that we removed suicidal items from the sum scores of the CDSS when adjusting for depressive symptoms in the association between psychotic experiences and suicidal outcomes. Second, we controlled other positive symptoms and SA history as confounders. Finally, a multi-domain scale was used to measure the severity of auditory hallucinations.

This study has several limitations. First, the cross-sectional design did not allow us to infer causation, although we used path analysis to infer directionality. The temporal sequence of the recent hallucination and CSIB can be inferred. Our preliminary mediation findings ignored other temporal precedence and variables’ change over time. To find unbiased estimates, a longitudinal examination of the relationships would be necessary (Maxwell & Cole, 2007; Maxwell, Cole, & Mitchell, 2011). Second, as the limited sample size, the confidence intervals were wide for some estimates. Third, patients came from one site in a routine clinical practice setting, potentially introducing selection biases and limiting generalization. Fourth, stigma may lead to possible bias in the disclosure of SI or suicidal plans.

Conclusions

The lifetime and current occurrence of auditory hallucinations among individuals with schizophrenia play important role in CSIB independent of depressive symptoms and lifetime SA history. The severity of auditory hallucination was not found to be associated with an increased risk of CSIB. Emotional distress caused by auditory hallucinations was positively related to depressive symptoms. It suggests that suicide risk assessment for patients with schizophrenia should consider the presence of the auditory hallucination experience. Cohort studies should further advance our understanding of the mechanisms that increase subsequent suicide risk in people with auditory hallucinations.

Highlights.

• The presence of auditory hallucinations was associated with current suicidality.

• Auditory hallucinations’ emotional severity was related to depressive symptoms.

• The severity of auditory hallucination was not directly associated with suicidality.