Depression Mediates the Relationships between Hallucinations, Delusions, and Social Isolation in First-Episode Psychosis

Abstract

Social isolation is common among individuals with schizophrenia spectrum and other psychotic disorders. Research indicates that social isolation relates to poorer mental health outcomes, depression, and negative symptoms, with less known about its relationship with positive symptoms. This study examined depression as a mediator in the relationships between positive symptoms (i.e., hallucinations and delusions) and social isolation among an early treatment phase sample in the United States. Data were obtained from the Recovery After an Initial Schizophrenia Episode project of the National Institute of Mental Health’s Early Treatment Program. Participants (N = 404) included adults between ages 15 and 40 in a first episode of psychosis. Data were analyzed using structural equation modeling in Mplus (Version 8). The study showed that delusions (b = .095, SE = 0.04, p < .05) and hallucinations (b = .076, SE = 0.03, p < .01) were directly related to depression, and that both delusions (b = .129, SE = 0.06, p < .05) and depression (b = .254, SE = 0.09, p < .05) were directly related to social isolation. Findings of this study determined that depression functioned as a mediator in the relationships between positive symptoms and social isolation. Targeting psychosis symptomatology and depression in treatment, improving social skills and social support networks, and considering the role of stigma in social isolation are of great importance in the prevention of poorer mental health outcomes.

Social isolation, defined as having no or minimal social interactions, relationships, or connections with family, friends, neighbors, and society at large (Wang et al., 2017), is strongly linked to poor mental health outcomes (Coyle & Dugan, 2012; Leigh-Hunt et al., 2017) and lower quality of life (Portugal et al., 2016; Thompson & Heller, 1990). Studies have also shown that individuals who are socially isolated experience more stress than those who are not isolated, and it is theorized that emotion regulation and expression become impaired due to lack of social support (Cacioppo & Hawkley, 2003). Given the prominent findings in the literature, social isolation is an important factor to consider when assessing for and treating mental illness.

Among individuals with schizophrenia spectrum and other psychotic disorders (SSPDs), social isolation is a particularly relevant concern (Lee et al., 2021; White et al., 2020). Data show individuals with SSPDs are 2.3 to six times more likely to self-report loneliness (the perception of social isolation; Lim et al., 2018; Stain et al., 2012) than those in the general population (Lamster et al., 2017; Stain et al., 2012). Individuals with SSPDs often face challenges in the development and maintenance of social relationships (Michalska da Rocha et al., 2018), are less integrated within their communities (Abdallah et al., 2009), and report lower rates of social connectedness, networks, and supports as compared with general population controls (Lim et al., 2018). Stain and colleagues (2012) report that loneliness among individuals with SSPDs is identified as one of the top three challenges faced in recovery. However, despite the prevalence of social isolation and knowledge of its relationship to the experience of loneliness as a potential barrier to recovery, many psychosocial interventions for SSPDs do not have a primary focus on reducing social isolation (Kurtz & Mueser, 2008; Lim et al., 2018; Mueser et al., 2013).

Social Isolation and Symptoms of Depression

Depression and hopelessness both relate to the experience of social isolation among individuals with SSPDs (Bornheimer & Jaccard, 2017; Bornheimer et al., 2020; Cacioppo et al., 2006; Goldsmith et al., 2002; Hor & Taylor, 2010; Wasserman et al., 2009). In our prior research, we found greater depression relates to increased social isolation, and more specifically a tendency to keep to oneself, feel awkward in social settings, and have a preference for being alone (Bornheimer et al., 2020). Further, data show social isolation is strongly associated with earlier mortality rates (Holt-Lunstad et al., 2015; Rico-Uribe et al., 2018), higher rates of depression (Cacioppo et al., 2006), and increased risk for suicide (Bornheimer et al., 2020; Goldsmith et al., 2002), as compared with individuals who are not socially isolated.

Social Isolation and Symptoms of Psychosis

Research has demonstrated a relationship between social isolation and the experience of greater negative symptoms among individuals with SSPDs (i.e., reduction in emotional experience and loss of volition; Harrop et al., 2015; Mucheru et al., 2017; Piskulic et al., 2012). Literature suggests that socially isolated individuals more often experience asociality (reduced motivation to engage in social interaction and a preference for being alone), anhedonia (reduced motivation or ability to feel pleasure), and blunted affect as compared with individuals who are not as socially isolated (Harrop et al., 2015). Due to the nature of negative symptoms, it is common for individuals to be less involved in activities with others, and rather, engage more in solitary behaviors or avoid others (Baek, 2014). Less, however, is known about the relationships between social isolation and positive symptoms (hallucinations and delusions in particular). Loneliness, on other hand, has been examined in relation to positive symptoms with also limited (Lim et al., 2018) yet slightly greater understandings to date in comparison with social isolation. Similar to social isolation, the experience of loneliness can function as a barrier to treatment and recovery for individuals with schizophrenia spectrum disorders (Ludwig et al., 2020). Angell and Test (2002) examined 87 young adults with schizophrenia and found the experience of greater positive symptoms over a six-month period associated with the loss of reciprocal network ties, more dissatisfaction with social relationships, and an increase in loneliness.

Other studies have examined positive symptoms in greater detail to find increased hallucinations (Michalska da Rocha et al., 2018; Myin-Germeys et al., 2001) and paranoia relate to greater self-reported loneliness (Jaya et al., 2017; Lamster et al., 2017; Sündermann et al., 2014). Lamster and colleagues (2017) conducted a study in which loneliness was experimentally manipulated using a false-feedback paradigm and results showed the induction of loneliness related to an increase in paranoid beliefs. Similarly, our prior research found greater hallucinations and delusions to relate to increased social isolation (Bornheimer et al., 2020). Research also supports the reverse of the relationships described earlier, in which loneliness predicts symptoms of psychosis (Lim et al., 2014; Michalska da Rocha et al., 2018). For example, Jaya and colleagues (2017) found greater loneliness related to increased positive symptoms within a community sample. Thus, a bidirectional relationship between symptoms of psychosis and the experience of loneliness and/or social isolation is most likely at play.

First-Episode Psychosis (FEP)

Much of what is known to date regarding psychosis and social isolation is within the context of samples with chronic SSPDs (i.e., Barut et al., 2016; Hor & Taylor, 2010; Lamster et al., 2017). Literature on these relationships within an early stage or FEP are more recently emerging with a focus on social networks, relationships, support, isolation, and loneliness (Lamster et al., 2017; Reed, 2008; Robustelli et al., 2017; Sündermann et al., 2014). Findings indicate poorly perceived social support and loneliness in FEP samples relate to greater depression (Sündermann et al., 2014), presence and severity of symptoms of psychosis, and lower overall functioning (Robustelli et al., 2017; Sündermann et al., 2014). During the early and acute stage of SSPD onset, it is theorized that social networks and supports are impacted by an individual’s potential experience of confusion or fear regarding positive and negative symptoms, feeling “out of control,” and having self-concept disrupted (MacDonald et al., 2005; Reed, 2008). These factors, combined with potential experiences with stigma (Corcoran et al., 2007; Goffman, 1963; Grant et al., 2017), relate to subsequent isolation experiences (Robustelli et al., 2017). Although literature among early stage or FEP populations on these topics are emerging, they have primarily examined the construct of loneliness as opposed to social isolation and have not examined mechanisms in the relationship between psychosis symptomatology and isolation, to date.

Given symptoms of psychosis and depression independently relate to social isolation (Bornheimer & Jaccard, 2017; Cacioppo et al., 2006; Harrop et al., 2015) and as these relationships are less understood in early phases of SSPD illness, the current study examined a model in which depression mediates the relationships between positive symptoms and social isolation within a sample of participants in an early treatment phase of illness. It is hypothesized that (a) hallucinations, delusions, and depression will directly relate to social isolation, and (b) hallucinations and delusions will directly relate to depression; thus, depression will mediate the independent relationships between hallucinations, delusions, and social isolation.

Method

Data were obtained from the Recovery After an Initial Schizophrenia Episode (RAISE) project of National Institute of Mental Health’s early treatment program (ETP). The program aimed to change the trajectory and prognosis of FEP and compared two ETPs to improve functional outcomes and quality of life between 2010 and 2012 (Kane et al., 2015). Community mental health clinics (n = 34) across 21 states were randomized to offer one of the two programs: (1) early treatment intervention (n = 223) or (2) standard community care (n = 181). The ETP, entitled NAVIGATE, included services such as medication management, psychoeducation, resilience-focused 1:1 therapy, education, and supported employment (Kane et al., 2016).

Participants (N = 404) between the ages of 15 and 40 with a diagnosis of schizophrenia, schizoaffective disorder, schizophreniform disorder, brief psychotic disorder, or psychotic disorder not otherwise specified based on the DSM-IV were included in the study. All participants experienced a first episode of psychosis, spoke English, and had been on antipsychotic medications for six months or less. Standard care involved clinical care for psychosis as determined by providers and clinic capacities (Kane et al., 2016). Greater detail about RAISE and NAVIGATE can be found in Mueser et al. (2015) and Kane et al. (2015).

Measures

Depression was measured at baseline using the Calgary Depression Scale for Schizophrenia (CDSS; Addington et al., 1990). The CDSS is a widely used and well-validated scale to assess severity of depressive symptoms among individuals diagnosed with schizophrenia (Addington et al., 1993) and was administered by a trained interviewer in the RAISE project. Items pertain to depression, hopelessness, self-depreciation, guilty ideas of reference, pathological guilt, morning depression, early wakening, suicide ideation and attempt, and observed depression. Response categories for each item range from 0 to 3, with higher scores indicating greater symptoms of depression (Cronbach’s alpha = .81).

Hallucinations and delusions were measured at baseline using the Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987). The PANSS is widely used in clinical studies of psychosis with strong psychometrics and was administered by a trained interviewer in the RAISE project. In the current study, two positive symptom items were of focus: (1) hallucinations, defined as verbal report or behavior indicating perceptions that are not generated by external stimuli, and may occur in the auditory, visual, olfactory, or somatic realms; and (2) delusions, defined as beliefs that are unfounded and idiosyncratic. The hallucination and delusion item response categories range from 1 to 7, with higher scores indicating greater presence and severity of hallucinations and/or delusions.

Social isolation was measured at baseline using a single item of the PANSS (Kay et al., 1987). The isolation item measured social withdrawal, described as diminished interest and initiative in social interactions due to passivity, apathy, anergy, or avolition leading to reduced interpersonal involvements. Item response categories range from 1 to 7, with higher scores indicating greater isolation and withdrawal from social interactions. Studies demonstrate convergent evidence with other assessment methods (i.e., self-report) to support the validity of this PANSS item (Hansen et al., 2013).

Quantitative Modeling and Analysis

Data were analyzed using Mplus (Version 8). Confirmatory factor analysis was conducted to specify and test the fit of the nine CDSS items for a latent depression variable prior to mediation modeling. All items significantly loaded onto the latent depression construct (p < .001), and thus treating the items as unidimensional was determined to be reasonable. Structural equation modeling (SEM) was performed to examine the study aims using a robust Huber–White maximum likelihood algorithm to obtain robust standard error (SE) and deal with nonnormality and variance heterogeneity. Endogenous variables in the model included social isolation and a latent depression variable (mediator), while exogenous variables included hallucinations and delusions. Full information maximum likelihood methods were used, consistent with recommendations in the literature (Enders, 2001; Little et al., 2014) to handle missing data, though minimal (<1%). Model fit was evaluated using both global (chi square, comparative fit index [CFI], standardized root-mean-square residual [SRMR], root-mean-square error of approximation [RMSEA]) and focused (standardized residuals <|2| and modification indices <|4|) fit indices (Muthén & Muthén, 1998–2017). Acceptable fit was determined by a minimum cutoff of 0.90 for CFI, a maximum cutoff of 0.08 for RMSEA, and a maximum cutoff of 0.08 for SRMR (Muthén & Muthén, 1998–2017). Consistent with prior research of FEP populations and use of secondary RAISE data (Bornheimer, 2019; Bornheimer et al., 2021), the following covariates were included in the model: age, diagnosis based on the DSM-IV (schizophrenia, schizoaffective bipolar, schizoaffective depressive, schizophreniform provisional or definite, brief psychotic disorder, or psychotic disorder not otherwise specified), antipsychotic medication taken at baseline, and duration of untreated psychosis.

Results

Demographic characteristics of participants at baseline are presented in Table 1. Participants were on average 23.6 years of age (SD = 5.06) and identified as male (n = 293, 73%), White (n = 218, 54%), and non-Hispanic/Latino (n = 331, 82%). Participants identified as being single/unmarried (n = 358, 89%), attending some high school or less (n = 145, 36%), not currently working/employed (n = 346, 86%), living with family (n = 287, 71%), and being uninsured (n = 192, 48%). Participants most often had a diagnosis of schizophrenia (n = 214, 53%) and reported the experience of untreated psychosis for on average six months (SD = 8.62). At the time of consent, 83% (n = 337) of participants reported current use of one or more antipsychotic medications. On average, social isolation scores were 3.53 (SD = 1.24), delusions were 3.63 (SD = 1.18), and hallucinations were 3.11 (SD = 1.58). Average depression scores are not reported, given the use of a latent depression variable in SEM comprising the nine depression items as its factor loadings.

Table 1:

Demographic Characteristics of the RAISE Sample at Baseline (N = 404)

Characteristic	n	%	M	SD
Age (years)			404	23.62 ± 5.06
Gender
Male	293	72.5
Female	111	27.5
Race
African American	152	37.6
White	218	54.0
American Indian or Alaska Native	31	5.2
Asian	12	3.0
Hawaiian or Pacific Islander	1	0.2
Ethnicity
Hispanic	73	18.1
Non-Hispanic	331	81.9
Marital Status
Married	24	5.9
Single/unmarried	358	88.6
Divorced, widowed, or separated	22	5.4
Education
Some high school or less	145	36.0
Completed high school	133	33.0
Some college or higher	125	31.0
Employment
Currently working	58	14.4
Not currently working	346	85.6
Insurance type
Private	82	20.4
Public	127	31.7
Uninsured	192	47.9
Residence
Independent living	72	17.8
Lives with family	287	71.0
Supported or structured housing	14	3.5
Homeless, shelter, other	31	7.7
Medication status
Using antipsychotics	337	83.4
Not using antipsychotics	67	16.6
Months of untreated psychosis			355	6.36 ± 8.62
Age of first psychiatric illness			398	16.52 ± 6.32
Age of first psychotic symptoms			392	19.15 ± 6.12
Number of psychiatric hospitalizations			314	1.94 ± 1.98
Diagnosis
Schizophrenia	214	53.0
Schizoaffective bipolar	24	5.9
Schizoaffective depressive	57	14.1
Schizophreniform provisional or definite	67	16.6
Brief psychotic disorder	2	0.5
Psychotic disorder, NOS	40	9.9

Notes: RAISE = Recovery After an Initial Schizophrenia Episode project; NOS = not otherwise specified.

Figure 1 presents the standardized parameter estimates for the measurement model and unstandardized parameter estimates for the structural model with SE in parentheses. Global fit indices all pointed to good model fit (χ² = 3.726, df = 65, p = .214; CFI = .991, RMSEA = 0.018, p for close fit = .992, SRMR = 0.035), and focused fit indices (standardized residuals and modification indices) revealed no points of stress on the model. Hallucinations, delusions, depression, and all covariates (age, diagnosis, antipsychotic medication, and duration of untreated psychosis) accounted for 8% of the variance in social isolation. Hallucinations, delusions, and all covariates accounted for 15% of the variance in depression.

Figure 1:

Model Results

Notes: Standardized parameter estimates are presented for the measurement model and unstandardized parameter estimates for the structural model. Standard errors are shown in parentheses. Calg = Calgary Depression Rating Scale for Schizophrenia.

*p < .05. **p < .01. ***p < .001.

Delusions related to social isolation both directly and indirectly through depression, thus serving as a partial mediator as indicated by the joint significance test (MacKinnon et al., 2002). As delusions increased, on average, there was an associated increase in social isolation holding covariates, depression, and hallucinations constant (b = .129, SE = 0.06, p < .05). Additionally, as delusions increased, on average, there was an associated increase in depression holding all other variables constant (b = .095, SE = 0.04, p < .05). Last, as depression increased, on average, there was an associated increase in social isolation holding all other variables constant (b = .254, SE = 0.09, p < .05). Given depression functioned as a mediator in the relationship between delusions and social isolation, the indirect effect was calculated. For every one-unit increase in delusions, there was an average associate .024 unit increase in social isolation holding all other variables constant. Taken together as a total effect, given the indirect and direct effect found, for every one-unit increase in delusions, there was an average associated .153 unit increase in social isolation holding all other variables constant.

Hallucinations related to social isolation indirectly through depression, thus serving as a full mediator as indicated by the joint significance test (MacKinnon et al., 2002). As hallucinations increased, on average, there was an associated increase in depression holding all other variables constant (b = .076, SE = 0.03, p < .01). Furthermore, as depression increased, on average, there was an associated increase in social isolation holding all other variables constant (b = .254, SE = 0.09, p < .05). Given depression functioned as a mediator in the relationship between hallucinations and social isolation, the indirect effect was calculated. For every one-unit increase in delusions, there was an average associate .019 unit increase in social isolation holding all other variables constant.

Discussion

Study findings indicate delusions and depression directly related to social isolation, and hallucinations indirectly related to social isolation through depression. Thus, as hypothesized, depression served as a mediator in both relationships between (a) delusions and social isolation and (b) hallucinations and social isolation. While literature supports the relationships between depression, social isolation, and negative symptoms within psychosis (Bornheimer et al., 2020; Cacioppo et al., 2006; Harrop et al., 2015), this study gives support for the relationships between positive symptoms (i.e., hallucinations and delusions), depression, and social isolation in an FEP sample.

Though delusions directly related to depression and social isolation (partial mediation), hallucinations only indirectly related to social isolation through depression (full mediation). It was anticipated that social isolation would be increased among individuals experiencing greater hallucinations, such that isolation could be a potential coping mechanism in response to the distress of symptom experience, feeling unrelatable to others, and stigma (Lim et al., 2014; Velthorst et al., 2012). A potential rationale for the nonsignificant finding of social isolation relating to hallucinations in the current study pertains to the social deafferentation hypothesis. This hypothesis proposes that hallucinations may fulfill communication and interactional needs in which one may experience social relatedness to the hallucination experience (Hoffman, 2008). In particular, a study looking at hallucinations, loneliness, and social isolation describes hallucination experiences allowing socially isolated individuals to “escape from boredom,” emptiness, and affective deprivation due to increased stimuli (El Haj et al., 2016). Future research is needed to examine the social deafferentation hypothesis specifically within FEP populations.

As previously mentioned, it is also important to consider the likely bidirectional relationship of social isolation exacerbating positive symptoms and vice versa. Social isolation may reinforce the maintenance of symptoms of psychosis, namely delusions, by reducing opportunities to reality test in social interactions (Depp et al., 2016; Lim et al., 2018). Importantly, in the other direction, having symptoms of psychosis may result in greater isolation from others as a coping mechanism due to distress and feeling that symptoms are inimitable and unrelatable (Bornheimer et al., 2020; Velthorst et al., 2012). Greater isolation resulting from positive symptoms over time may also subsequently increase depression. As a result, future longitudinal research is needed to investigate the nature and directionality of these relationships with the goal of increasing understandings to inform targets of psychosocial interventions.

Study findings point toward several implications for clinical practice. First, it is important for interventions to target psychosis symptomatology. Evidence-based pharmacological and psychosocial treatments are well established in the literature with data supporting the value of early intervention following a first episode with use of antipsychotic medication (Dixon et al., 2018; Stafford et al., 2015), cognitive–behavioral therapy (Jackson et al., 2005; Jackson et al., 2008; Lewis et al., 2002; Tarrier et al., 2006; Tarrier & Wykes, 2004), and psychoeducation for individuals and their families (Leavey et al., 2004). Coordinated specialty care, such as within the RAISE ETP, has more recently emerged as a recovery-oriented approach with a focus on improving the course of schizophrenia (Dixon et al., 2015; Kane et al., 2016). Second, and given the mediation finding of depression in the current study, it is essential to evaluate for and treat symptoms of depression. Targeting affective states such as depression has the potential to buffer the effect of psychosis symptoms on the experience of social isolation. Also, given literature suggesting the role that stigma plays in social isolation, depression is likely a less stigmatizing and more acceptable treatment target for individuals in an early phase or first episode of illness (Bornheimer et al., 2021; Jaya et al., 2017).

Third, social isolation can be targeted in interventions with a focus on bolstering self-esteem and reducing guilt (Ludwig et al., 2020), in addition to improving social skills and strengthening support networks among individuals with symptoms of psychosis. Social skills training (SST) programs are structural psychosocial interventions aiming to promote social support among adults diagnosed with schizophrenia (Kasckow et al., 2011). Specifically, they focus on receiving skills (perception), processing skills (cognition), sending skills (behavioral), affiliative skills (expressing affection), instrumental role skills (everyday social activities), interaction skills (conversing), and behavior governed by social norms (scripted social activities; Kopelowicz et al., 2006). Techniques include teaching skills, goal setting, role modeling, behavioral rehearsal, positive reinforcement, and corrective feedback (Mueser et al., 2013). Data from meta-analyses provide support for improvements in social functioning, psychopathology, daily living skills, negative symptoms, and relapse among individuals who receive SST (Kurtz & Mueser, 2008). Clinical practice and empirical understandings of SSTs would benefit from examinations of treatment effects including positive symptom outcomes (hallucinations and delusions in particular).

Last, though not examined in the current study, stigma is an important treatment target among individuals with SSPDs, given its prevalence, negative impact on engagement in treatment (Corcoran et al., 2007; Goffman, 1963; Grant et al., 2017), and established relationships with depression and symptoms of psychosis, particularly in FEP (Bornheimer et al., 2021; Kane et al., 2016; Lysaker et al., 2007; Mueser et al., 2015). The experience of stigma can function as a chronic stressor and, combined with the knowledge that stress relates to greater symptoms of psychosis (van Winkel et al., 2008), stigma has the potential to significantly impact the course of one’s illness (van Zelst, 2009). Psychosocial interventions with a focus on building resilience and coping skills to manage experiences of stigma are greatly needed. Data also show stigma negatively impacts illness to a greater degree for individuals who use isolating coping skills as coping, compared with individuals who don’t self-isolate (González-Torres et al., 2007; Lysaker et al., 2007; Vauth et al., 2007). As a result, it is likely that interventions targeting resilience and coping in treatment would benefit from delivery within the context of shared experiences and/or peer support (van Zelst, 2009).

The current study must be considered in light of several potential limitations. First, the RAISE project was not conducted to address the aims of the current study, thus, measurement of social isolation was constrained. The social isolation measure was represented by a single item of the PANSS negative symptom scale and ideally the construct would be measured by its own standardized multi-item scale. Second, self-report and social desirability are common concerns in mental health research and should be considered in the current study, especially when considering the assessment of symptoms of psychosis and social isolation. Third, the sample was diagnostically heterogenous, given the RAISE project included subjects with schizophrenia, schizoaffective disorder, schizophreniform disorder, brief psychotic disorder, or psychotic disorder not otherwise specified based on the DSM-IV; thus, future research should investigate potential differences between SSPD diagnostic categories. Last, causality cannot be implied given the present study is cross-sectional in nature.

In sum, depression served as a mediator in the relationships between hallucinations, delusions, and social isolation in an FEP sample. These findings are consistent with the overarching literature of social isolation being impacted by mental health and, as a result, the following four areas of treatment are important to consider: (1) assessing for and treating psychosis symptomatology, (2) assessing for and treating symptoms of depression, (3) bolstering social skills and social networks of support, and (4) exploring the role of stigma in isolation experiences. Future longitudinal research is needed to investigate the nature and directionality of these relationships, particularly to inform early intervention efforts for FEP populations.

 

This work was supported by the National Institute of Mental Health contract [HHSN271200900019C], which included funds from the American Recovery and Reinvestment Act. Clinical trials registration [NCT01321177]: An Integrated Program for the Treatment of First Episode of Psychosis (RAISE ETP; http://www.clinicaltrials.gov/ct2/show/NCT01321177). The principal investigator of the Recovery After an Initial Schizophrenia Episode ETP study is John M. Kane, M.D. The contents of this article are solely the responsibility of the authors and do not necessarily represent the views of National Institute of Mental Health or the U.S. Department of Health and Human Services.