Introduction
The hierarchical paranoia continuum model provides a framework to portray the relationships and overlapping constructs between persecutory paranoia and social anxiety fears (Freeman et al., Reference Freeman, Garety, Bebbington, Smith, Rollinson, Fowler, Kuipers, Ray and Dunn2005). Persecutory delusions, prevalent in psychosis, refer to the belief of imminent harm and malevolent intentions from others (Freeman and Garety, Reference Freeman and Garety2000). Individuals with paranoia or social anxiety often report fear regarding perceived threat and their reactions while under threat (Stopa et al., Reference Stopa, Denton, Wingfield and Taylor2013). However, those with paranoia tend to have more difficulties with perceptual experiences, distancing themselves from the threat, and narrative coherence (Stopa et al., Reference Stopa, Denton, Wingfield and Taylor2013). The complexity of paranoia and social anxiety requires further elucidation (Stopa et al., Reference Stopa, Denton, Wingfield and Taylor2013) to improve psychological interventions. It is recommended to identify manipulable mechanisms underlying the relationship between social anxiety and persecutory delusions (Michail et al., Reference Michail, Birchwood and Tait2017).
According to cognitive models, maladaptive self-beliefs and assumptions (e.g. I am incompetent) aggravate negative interpretations and feelings, leading to counter-productive safety behaviours aimed at preventing failure and embarrassment (Beck et al., Reference Beck, Emery and Greenberg2005). Individuals with social anxiety perceive themselves as social targets, which induces fear in social events and in public activities (Clark and Wells, Reference Clark and Wells1995). Negative self-evaluations in society can spiral into worrying processes, escalating plausible ideas (e.g. others talk about me) to implausible ideas (e.g. people hate me and threaten me) (Freeman and Garety, Reference Freeman and Garety1999; Sun et al., Reference Sun, So, Chan, Chiu and Leung2019). Our focus initially lies on appraisals regarding the loss of social role, feeling different from others, and enforced low social status that may worsen social fears (Aunjitsakul et al., Reference Aunjitsakul, Mcguire, Mcleod and Gumley2021; Iqbal et al., Reference Iqbal, Birchwood, Chadwick and Trower2000). Consequently, individuals may develop shame and stigma-related cognitions if they perceive themselves as falling short of social standards (Welten et al., Reference Welten, Zeelenberg and Breugelmans2012). Several studies have found that both shame and stigma are higher in socially anxious people (Michail and Birchwood, Reference Michail and Birchwood2013), and that these factors help predict social anxiety amongst those with psychosis (Aherne, Reference Aherne2014; Birchwood et al., Reference Birchwood, Trower, Brunet, Gilbert, Iqbal and Jackson2006; Lysaker et al., Reference Lysaker, Yanos, Outcalt and Roe2010). Negative appraisals, particularly shame and stigma cognitions, could be a crucial factor in identifying the relationship between social anxiety and paranoid ideation (Aunjitsakul et al., Reference Aunjitsakul, Mcleod and Gumley2022a; Aunjitsakul et al., Reference Aunjitsakul, Mcleod and Gumley2022b).
Considering the behavioural aspect, safety behaviours (e.g. avoiding eye contact, or speaking softly) are often used by people with social anxiety to deal with feared social events (Haghighat, Reference Haghighat2001; Smart and Wegner, Reference Smart and Wegner1999). Moreover, people with psychosis frequently use safety behaviours (e.g. avoidance, in-situation behaviours, or escape) to deal with persecutory threats (Freeman et al., Reference Freeman, Garety, Kuipers, Fowler, Bebbington and Dunn2007). These safety behaviours often hinder the processing of conflicting evidence, thereby impeding individuals with social anxiety from reassessing their false assumptions, which may perpetuate negative self-beliefs in society. Using safety behaviours hinders the ability to assimilate new evidence (Clark and Wells, Reference Clark and Wells1995), fostering interpersonal anxiety and potentially leading to paranoia (Sun et al., Reference Sun, So, Chan, Chiu and Leung2019). Given their role in perpetuating social anxiety (Smart and Wegner, Reference Smart and Wegner1999) and delusional thinking (Freeman et al., Reference Freeman, Garety, Kuipers, Fowler, Bebbington and Dunn2007), targeting safety behaviours holds significance in psychological interventions.
There is significant impact on mental health by cultural factors, for example expression and functional outcomes, health seeking behaviours, attitudes of patients, and the practitioners and mental health systems (Hernandez et al., Reference Hernandez, Nesman, Mowery, Acevedo-Polakovich and Callejas2009). Asian individuals, for instance, are more likely to report somatic symptoms rather than emotions because they believe that mental weakness is not acceptable but physical illness is. Meanwhile, American individuals readily express their emotions to clinicians (Gopalkrishnan, Reference Gopalkrishnan2018; Lin and Cheung, Reference Lin and Cheung1999). As values and norms from each culture may reflect individual interpretation, the sociocultural context may be an essential modulator of negative appraisals, such as stigma or shame reactions. Individuals with mental illness are judged and treated differently in several societies (Haghighat, Reference Haghighat2001), where socio-cultural factors play an additional crucial role in the expression of psychopathology (Tseng, Reference Tseng2001). Various studies related to paranoid thinking have mainly been conducted in Western settings (Freeman et al., Reference Freeman, Garety, Bebbington, Smith, Rollinson, Fowler, Kuipers, Ray and Dunn2005; Johns et al., Reference Johns, Cannon, Singleton, Murray, Farrell, Brugha, Bebbington, Jenkins and Meltzer2004; Kaymaz and van Os, Reference Kaymaz and van Os2010; Linscott and van Os, Reference Linscott and van Os2010), whereas there is insufficient information from Asian perspectives. Furthermore, culture is an important influence on mental health and social evaluation concerns, for example belief contents affecting persecutory delusions (Skodlar et al., Reference Skodlar, Dernovsek and Kocmur2008), levels of social discrimination associated with mental illness (Moleiro, Reference Moleiro2018), each contextual norm and value causing different shameful experiences (Ha, Reference Ha1995), or each social interaction context leading to different safety behaviours (Piccirillo et al., Reference Piccirillo, Taylor Dryman and Heimberg2016). In such individuals, the expression of social anxiety or paranoia in psychosis and its underlying mechanisms are likely to be affected by the cultural context. This study was conducted in Thailand, where individuals diagnosed with schizophrenia primarily receive medication-based treatment, with psychological interventions being less accessible due to limited personal healthcare resources (Wannasewok et al., Reference Wannasewok, Suraaroonsamrit, Jeungsiragulwit and Udomratn2022). Considering cultural background, although stigma causes social difficulties for individuals with mental illness, Thai people now have a more positive attitude towards them and tend to support if those with mental illness are their friends/family (Aunjitsakul et al., Reference Aunjitsakul, Mcleod and Gumley2022b).
Social anxiety disorder (SAD) is a common psychological co-morbidity in schizophrenia (McEnery et al., Reference McEnery, Lim, Tremain, Knowles and Alvarez-Jimenez2019b). Individuals with schizophrenia and co-morbid SAD experience low functioning, low self-esteem (Karatzias et al., Reference Karatzias, Gumley, Power and O’Grady2007), and experience difficulties in engaging with social activities and relationships (Agid et al., Reference Agid, McDonald, Siu, Tsoutsoulas, Wass, Zipursky, Foussias and Remington2012). Cognitive behavioural therapy is the treatment of choice for SAD (Acarturk et al., Reference Acarturk, Cuijpers, van Straten and de Graaf2009; Mayo-Wilson et al., Reference Mayo-Wilson, Dias, Mavranezouli, Kew, Clark, Ades and Pilling2014) and can be used to reduce psychotic symptoms in psychosis patients (Taylor and Perera, Reference Taylor and Perera2015; Wykes et al., Reference Wykes, Steel, Everitt and Tarrier2008). Although there is evidence to treat social anxiety in psychosis (Hicks, Reference Hicks2019; McEnery et al., Reference McEnery, Lim, Knowles, Rice, Gleeson, Howell, Russon, Miles, D’alfonso and Alvarez-Jimenez2019a), little is known about the mechanism of SAD in individuals with psychosis (Michail et al., Reference Michail, Birchwood and Tait2017).
As cognitions and behaviours could play different roles in the relationship between social anxiety and paranoia, we aim to investigate separately whether the relationship is mediated by negative appraisals (shame or stigma) and safety behaviours factors (anxious avoidance and in situ safety behaviour) in individuals with schizophrenia. We first hypothesised that cognitions, shame or stigma, may contribute to be a key mediator of the social anxiety and paranoia relationship after controlling for depression. Second, regarding this relationship, we also tested whether behavioural strategies, anxious avoidance or in situ safety behaviours, could be a significant mediator.
Method
This cross-sectional study included individuals diagnosed with schizophrenia, who were followed up at the out-patient department (OPD) of University Hospital, Thailand.
Participants
Individuals diagnosed with schizophrenia via the diagnostic code F20 according to ICD-10 (World Health Organization, 2016) were recruited. The inclusion criteria required that participants were aged at least 18 years and had no hospital admission or medication changes in the previous three months. Recruited participants were people with long-standing psychosis who regularly followed up at the OPD. We included individuals of any severity level (experiencing any level of social anxiety/paranoia), as evaluated by a psychiatrist or a qualified health professional who was independent of the research team, had the capacity to provide informed consent, and expressed their willingness to participate in the study. Individuals who were unable to cooperate well and communicate meaningfully in Thai language were excluded.
Measurements
Five instruments were used in this study; one was the Thai version of the Depression Anxiety Stress Scale (DASS; Webster et al., Reference Webster, Hawley and Lopez2013). The other four English language instruments employed were forward and backward translated following the guidelines for the cross-cultural adaptation of self-report measures (Beaton et al., Reference Beaton, Bombardier, Guillemin and Ferraz2000; Van Ommeren et al., Reference Van Ommeren, Sharma, Thapa, Makaju, Prasain, Bhattarai and De Jong1999). The translation process for the Thai versions of these instruments started with their translation by an author and a PhD student in another field, from English to Thai. Subsequently, two independent professional translators, who were unaware of the outcome measurements at play, performed their back translation from Thai to English. Finally, experts in the field (co-author team) reviewed and validated the translations.
Measurement tools
Paranoia
The Revised Green et al. Paranoid Thought Scales (R-GPTS) is an 18-item questionnaire with an ‘ideas of social reference’ subscale (8 items) and a ‘paranoia’ subscale (10 items) (Freeman et al., Reference Freeman, Loe, Kingdon, Startup, Molodynski, Rosebrock, Brown, Sheaves, Waite and Bird2019). Items are scored on a 5-point Likert scale anchored by 0 (not at all) and 4 (total), giving a range of 0–32 for social reference scores and paranoia scores between 0 and 40. Higher scores indicate greater levels of paranoid thinking. R-GPTS has shown excellent psychometric properties, that is, a Cronbach’s alpha of .90 (Freeman et al., Reference Freeman, Loe, Kingdon, Startup, Molodynski, Rosebrock, Brown, Sheaves, Waite and Bird2019); in this study, it was calculated at .94.
Social anxiety
The Social Interaction Anxiety Scale (SIAS) is a 20-item questionnaire using a 5-point Likert scale from 0 (not at all) to 4 (extremely) (Mattick and Clarke, Reference Mattick and Clarke1998). It yields a total score of 80, with higher scores indicating higher levels of social anxiety. This scale has been shown to have excellent reliability (test–retest correlations 0.92), internal consistency (Cronbach’s alpha .94), and validity (Mattick and Clarke, Reference Mattick and Clarke1998). Our calculated Cronbach’s alpha was at a value of .88. Consistent with previous studies, scores over 36 were used to determine the presence of significant social phobia in our research (Peters, Reference Peters2000).
Shame and stigma
The Personal Beliefs about Illness Questionnaire-Revised (PBIQ-R) (Birchwood et al., Reference Birchwood, Jackson, Brunet, Holden and Barton2012) was used to assess shame and stigma based on the patients’ appraisals of their post-psychotic experiences. It is a 20-item rating using a 4-point Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). The PBIQ-R contains five subscales: shame, loss, entrapment, control over illness, and social marginalisation/group fit (or stigma). The subscales of shame (PBIQ-R Shame) and stigma (PBIQ-R Stigma) factors were used as mediators between the link of social anxiety and paranoia. The test–retest reliability scores of shame (0.84) and stigma (0.64) have been reported to range from acceptable to good. Likewise, the Cronbach’s alpha values of shame and stigma have also been reported to be adequate: –.73 and .78, respectively (Birchwood et al., Reference Birchwood, Jackson, Brunet, Holden and Barton2012); in our study, these were .84 and .83, respectively.
Safety behaviours (specific to paranoia)
The Oxford Cognitions and Defences Questionnaire (O-CDQ) is a 46-item measure rated on a 4-point Likert scale with answers ranging from 0 (never) to 3 (always) (Rosebrock et al., Reference Rosebrock, Lambe, Mulhall, Petit, Loe, Saidel, Pervez, Mitchell, Chauhan, Prouten, Chan, Aynsworth, Murphy, Jones, Powling, Chapman, Dudley, Morrison, O’regan, Clark, Waite and Freeman2022). Its questions include three main factors related to (1) threat cognitions (O-CDQ Threat cognitions); (2) anxious avoidance (O-CDQ Avoidance); and (3) putting up defence in the presence of outside or in situ (within-situation) safety behaviours (O-CDQ In situ behaviours). The latter two factors – O-CDQ Avoidance and O-CDQ In situ behaviours – were the safety behaviour factors used as mediators between social anxiety and paranoia. O-CDQ has shown excellent psychometric properties. The reported Cronbach’s alpha values are as follows: threat cognitions, .93; avoidance, .94; and in situ behaviours, .93; and those for the test–retest reliability are as follows: threat cognitions, 0.88; avoidance, 0.92; and in situ behaviours, 0.89 (Rosebrock et al., Reference Rosebrock, Lambe, Mulhall, Petit, Loe, Saidel, Pervez, Mitchell, Chauhan, Prouten, Chan, Aynsworth, Murphy, Jones, Powling, Chapman, Dudley, Morrison, O’regan, Clark, Waite and Freeman2022). From our analyses, the Cronbach’s alpha coefficients of O-CDQ Threat cognitions, Avoidance and In situ behaviours were .91, .89, and .85, respectively.
Depression
The DASS-42 (Webster et al., Reference Webster, Hawley and Lopez2013) measures general negative affect and distress in the domains of depression, anxiety and stress. We employed only a 14-item depression sub-scale as a co-variance factor; it consisted of 4-point scaled items with answers ranging from 0 (did not apply to me at all) to 3 (applied to me very much). The DASS scale has shown excellent psychometric properties for depression (Cronbach’s alpha .91) (Lovibond and Lovibond, Reference Lovibond and Lovibond1995; Webster et al., Reference Webster, Hawley and Lopez2013), and has been validated across Asian samples, including Thailand, with a Cronbach’s alpha coefficient range of .70–.86 (Oei et al., Reference Oei, Sawang, Goh and Mukhtar2013); in this study, it was .94.
Data collection
A convenience sample of patients was invited to participate by a nurse at the OPD; the nurse was not a part of the research team. After obtaining their consent, the questionnaires were given to participants by either the researcher or the research assistant. The participants were asked to complete the Thai version of the five instruments in full. Brief demographics including age, gender, ethnicity, religion, income and academic qualification were also collected. Participants were informed that they could request the researcher or research assistant to help with reading and/or filling in the questionnaires.
Data analysis
IBM SPSS Statistics for Windows (version 27.0) was used for our data analyses. Pearson’s correlation coefficients were employed to calculate inter-variable associations. Considering the hypotheses, we checked for assumptions to interpret the mediation analyses as well as for linearity and multi-collinearity. Regarding the linear associations of social anxiety with paranoia, we used a linear regression model to investigate the associations. A stepwise multiple regression analysis was also used to confirm the final model of the social anxiety–paranoia association. Regarding the stepwise regression, the factors were added to the model according to Hypothesis 1 (e.g. Shame, Stigma, Depression) and Hypothesis 2 (e.g. Anxious avoidance, In situ safety behaviours, Depression). Multi-collinearity was also checked in the regression model (factors with a variance inflation factor >5 and tolerance <0.2 were excluded) (Christopher and Odum Institute, 2019). There were no assumption violations related to linearity and multi-collinearity; this allowed us to continue using the mediation analyses.
The mediation analysis was addressed to test which variable(s) mediate(s) the association between social anxiety and paranoia. Simple and parallel multiple mediation analyses with co-varying depression were established, using shame and stigma (Hypothesis 1), and anxious avoidance and in situ safety behaviour (Hypothesis 2) as mediators. The PROCESS macro for SPSS (version 3.4) was used to conduct the mediation analyses (Hayes, Reference Hayes2018). The analyses of 10,000 bias-corrected bootstrap samples were conducted to estimate the 95% confidence intervals of the indirect effect.
As this study was conducted between January and April 2020, that is amid the SARS-COVID-19 outbreak, it is plausible that some of the anxious avoidance and in situ safety behaviour detected was attributable to the fear of COVID-19 infection. Hence, some increase in negative appraisals, threat cognitions related to social interaction, decreased socialisation, or avoidance of public places may have been a part of a normal reaction to a legitimate health threat. To explore this, we used 11 March 2020, the date when the World Health Organization announced COVID-19 as a pandemic, to categorise our study participants into two groups: (1) those who provided data pre-pandemic (1 January–11 March 2020); and (2) those who completed the questionnaires after the pandemic was declared (12 March–30 April 2020). We performed a post-hoc analysis to first compare the data profiles between pre- and post-pandemic declaration groups. Second, although the O-CDQ threat cognitions were not in our original hypotheses, we deemed it necessary to add this outcome using linear regression and mediation analyses because we reasoned that these worrying thoughts could have been theoretically affected by the pandemic and that the COVID-19 pandemic could create threat and mistrust amongst community (Phillips et al., Reference Phillips, Gillespie, Hallingberg, Evans, Taiyari, Torrens-Burton, Cannings-John, Williams, Sheils, Ashfield-Watt, Akbari, Hughes, Thomas-Jones, James and Wood2022). Furthermore, we created the pre/post-COVID-19 pandemic variable to adjust the mediation analyses (along with depression); this was done to test whether the mediator outcomes (in Hypotheses 1 and 2) were affected by the duration of this pandemic or not. Post-hoc analyses outcomes are presented in the Supplementary material.
Results
Sample characteristics
We approached 130 individuals with schizophrenia; 113 respondents (86.9%) completed the questionnaires, nine (6.9%) declined to participate, and eight (6.2%) were excluded owing to language barrier or illness factors (e.g. severe disorganisation). Of the included responders, 59.3% were male, and the mean age of our samples was 44.2 years. All but one were educated at least at primary school level. The mean SIAS score was 21.6, and 9.7% of the participants met the threshold for social phobia. The details concerning the other factors are described in Table 1.
Table 1. Demographic and psychological factors of people with schizophrenia (N=113)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20240828211518928-0654:S1352465824000225:S1352465824000225_tab1.png?pub-status=live)
O-CDQ, Oxford Cognitions and Defences Questionnaire; DASS, Depression Anxiety Stress Scales; PBIQ-R, Personal Beliefs about Illness Questionnaire-Revised; R-GPTS, Revised Green et al. Paranoid Thought Scales; SIAS, Social Interaction Anxiety Scale. Data are presented as means±SD unless otherwise indicated. *1GBP=40.59Baht (source: Exchange Rates UK as of 29 October 2020; retrieved from: https://www.exchangerates.org.uk/Pounds-to-Baht-currency-conversion-page.html).
Inter-correlation of potential variables
SIAS and R-GPTS Persecutory were significantly correlated with variables (e.g. R-GPTS Reference, PBIQ-R [Control over illness; Shame; Entrapment; Loss; Stigma], O-CDQ [Threat cognitions; Anxious avoidance; In situ safety behaviours], and DASS Depression). The highest correlation coefficients of SIAS and R-GPTS Persecutory were related to O-CDQ Threat cognitions (r=0.73, p < .01 and 0.74, p < .01). The other bivariate correlation coefficients are shown in Table 2.
Table 2. Intercorrelations of potential variables of people with schizophrenia (N=113)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20240828211518928-0654:S1352465824000225:S1352465824000225_tab2.png?pub-status=live)
O-CDQ, Oxford Cognitions and Defences Questionnaire; DASS, Depression Anxiety Stress Scales; PBIQ-R, Personal Beliefs about Illness Questionnaire-Revised; R-GPTS, Revised Green et al. Paranoid Thought Scales; SIAS, Social Interaction Anxiety Scale. *p<.01.
Social anxiety and potential mediators in association with persecutory delusions
In Table 3, SIAS was significantly associated with R-GPTS Persecutory (B .33, p < .001) (Model 1); however, after controlling for DASS Depression, this relationship was no longer significant (Model 4). When controlling for DASS Depression in the social anxiety and paranoia relationship, PBIQ-R Shame (B –.75, p < .05) and PBIQ-R Stigma (B .88, p=.022) were significantly associated with R-GPTS Persecutory (Model 5); meanwhile, only O-CDQ In situ behaviours was significantly associated with R-GTPS Persecutory (B .47, p < .01) (Model 6).
Table 3. Linear regression analysis of R-GPTS persecutory (a dependent variable) testing Hypotheses 1 and 2, in Model 5† and Model 6‡, respectively (N=113)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20240828211518928-0654:S1352465824000225:S1352465824000225_tab3.png?pub-status=live)
O-CDQ, Oxford Cognitions and Defences Questionnaire; DASS, Depression Anxiety Stress Scales; PBIQ-R, Personal Beliefs about Illness Questionnaire-Revised; R-GPTS, Revised Green et al. Paranoid Thought Scales; SIAS, Social Interaction Anxiety Scale. †Regarding the stepwise regression analysis, the final model of R-GPTS Persecutory included only Depression (B .65, p<.001) with an adjusted R 2 of 37.1%. ‡Regarding the stepwise regression analysis, the final model of R-GPTS Persecutory included Depression (B .47, p<.001) and O-CDQ In situ safety behaviours (B .57, p<.001) with an adjusted R 2 of 47.1%.
To confirm these relationships, we used stepwise regression analyses. Concerning social appraisals, PBIQ-R Shame and PBIQ-R Stigma were not significantly associated with R-GTPS Persecutory; only DASS Depression (B .65, p < .001) reached the level of significant in the final model, accounting for 37.1% of variance. In the safety behaviours model, accounting for 47.1% of variance, O-CDQ In situ behaviours (B .57, p < .001) and Depression (B .47, p < .001) were statistically significant.
Mediation analysis testing theoretical hypotheses with potential factors
We investigated the relationship between social anxiety and paranoia with potential mediators. After the simple mediation analysis, SIAS was found to relate to R-GPTS Persecutory through its relationship with O-CDQ Avoidance and O-CDQ In situ behaviours; the indirect effect based on 10,000 bootstrapped samples was ab=.07 (95% CI=.021, .149; a=.30, p < .001; b=.24, p < .01) and .12 (95% CI=.053, .218; a=.21, p < .001; b=.59, p < .001), respectively. The details regarding the other effects are presented in Table 4.
Table 4. Results of simple and parallel multiple mediation analyses examining direct, indirect and total effects of independent (social anxiety) and dependent variables (R-GPTS persecutory) with co-variances (DASS Depression) through mediators (N=113)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20240828211518928-0654:S1352465824000225:S1352465824000225_tab4.png?pub-status=live)
O-CDQ, Oxford Cognitions and Defences Questionnaire; DASS, Depression Anxiety Stress Scales; PBIQ-R, Personal Beliefs about Illness Questionnaire-Revised; R-GPTS, Revised Green et al. Paranoid Thought Scales; SIAS, Social Interaction Anxiety Scale. *p<.05, **p<.01, ***p<.001.
To test the two priori hypotheses, with mediators being shame and stigma (Hypothesis 1), and anxious avoidance and in situ safety behaviours (Hypothesis 2), we used the multiple parallel mediation analysis controlling for depression. Considering Hypothesis 1, we analysed whether the social anxiety–paranoia relationship was mediated by stigma and shame (see Fig. 1A). We found that PBIQ-R Shame and PBIQ-R Stigma did not show significant indirect effects. Only PBIQ-R Shame (b=–.69, p < .05) and PBIQ-R Stigma (b=.80, p < .05) were detected to have a significant direct effect on RGTPS Persecutory.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20240828211518928-0654:S1352465824000225:S1352465824000225_fig1.png?pub-status=live)
Figure 1. Multiple mediation analyses of the relationship between social anxiety and paranoia with shame and stigma (Hypothesis 1, panel A), and anxious avoidance and in situ safety behaviours (Hypothesis 2, panel B) considered as mediators. †Value of indirect effect of the in situ safety behaviours (M3): ab=.105, 95% CI=.038, .201. n.s., not significant.
Regarding Hypothesis 2, we analysed whether the social anxiety–paranoia relationship was mediated by anxious avoidance or in situ safety behaviours (see Fig. 1B). We found that O-CDQ In situ behaviours showed a significant indirect effect through the relationship of SIAS with R-GPTS Persecutory when controlling for DASS Depression. The direct effect of SIAS on O-CDQ In situ behaviours was a=.21 (p < .001), and the direct effect of O-CDQ In situ behaviours on RGTPS Persecutory was b=.50 (p < .01); meanwhile, the indirect effect was ab=.11 (95% CI=.038, .201) based on 10,000 bootstrapped samples.
Discussion
To investigate the mechanisms of the relationship between social anxiety and paranoia in schizophrenia in an Asian sample, we tested whether negative social appraisals (shame or stigma) and safety behaviours (anxious avoidance or in situ safety behaviours) fully mediate the social anxiety–paranoia relationship. We discovered a linear relationship between social anxiety and paranoia. This finding is similar to Western populations (Freeman et al., Reference Freeman, Garety, Bebbington, Smith, Rollinson, Fowler, Kuipers, Ray and Dunn2005; Linscott and van Os, Reference Linscott and van Os2010). However, after controlling for depression, the relationship between social anxiety and paranoia was no longer significant. This is because depression was highly correlated with all factors in the study. Multiple regression analyses controlling for age, gender and depression found that shame, stigma, and in situ safety behaviours were significantly associated with paranoia. In the stepwise regression analyses, only in situ safety behaviours was found to be a significant factor. We then conducted mediation analyses and found that stigma and shame were not significant mediators, whereas in situ safety behaviours was a full mediator of the social anxiety–paranoia relationship after co-varying for depression.
As our study was conducted during the COVID-19 pandemic, potentially impacting negative appraisals and safety behaviours, post-hoc analyses were also performed. No significant differences in terms of sociodemographic and potential variables were detected between the pre- and post-pandemic declaration groups. In the mediation analyses, controlling for the pre/post-COVID-19 pandemic variable, no differences in mediating effect were observed. While the pandemic might have influenced the negative appraisals or safety behaviours of people with psychosis in this study, definitive claims from post-hoc analyses are limited owing to study power constraints and misalignment with research objectives.
Contrary to our expectations, stigma and shame were not significant mediators of the social anxiety–paranoia relationship. One explanation for this is that depression could confound mediator outcomes of this relationship, because from the overall result depression is significant with social anxiety, paranoia, and other study factors. Depression also leads to negative appraisals in psychosis (Birchwood et al., Reference Birchwood, Mason, Macmillan and Healy1993; Karatzias et al., Reference Karatzias, Gumley, Power and O’Grady2007) and also links with social discrimination or unattractiveness concerns (Gumley et al., Reference Gumley, O’Grady, Power and Schwannauer2004; Karatzias et al., Reference Karatzias, Gumley, Power and O’Grady2007), including interpersonal worry and threat (mis)interpretation (Freeman et al., Reference Freeman, Gittins, Pugh, Antley, Slater and Dunn2008).
Additionally, there were a small number of individuals with social anxiety (only 11) according to the SIAS cut-off score, indicating that their anxiety does not meet the threshold levels of distress in social evaluation. Given that negative social appraisals like stigma and shame could theoretically explain and have been suggested to be targeted in the treatment of social anxiety in psychosis (Aunjitsakul et al., Reference Aunjitsakul, Mcguire, Mcleod and Gumley2021), further clinical work with a larger study sample is required to develop a comprehensive understanding of the role of stigma and shame in alleviating social fears or persecutory paranoia in psychosis. Furthermore, regarding the context or environment affecting negative interpretation in others, stigma or shame might be a moderator rather than a mediator of the social anxiety and paranoia association.
Safety behaviours were found to be a full mediator between social anxiety and paranoia in this study. It could be that when individuals with schizophrenia perceive threats as a misperception which is influenced by social anxiety or paranoia cognitions, they may feel different or fear being judged (Haghighat, Reference Haghighat2001) or unattractive (Trower and Gilbert, Reference Trower and Gilbert1989). Such individuals may design actions or safety behaviours to prevent their feared catastrophe from occurring (Salkovskis et al., Reference Salkovskis, Clark and Gelder1996). This, in turn, results in the persistence of social anxiety (Smart and Wegner, Reference Smart and Wegner1999), delusional thinking (Freeman et al., Reference Freeman, Garety, Kuipers, Fowler, Bebbington and Dunn2007), and emotional distress (Tully et al., Reference Tully, Wells and Morrison2017). Our data support the possibility that safety behaviours could be a crucial factor to be targeted in treating social anxiety or paranoia in individuals with psychosis.
Moreover, the results of this research highlight the crucial role behavioural strategies play in cognitive behavioural approaches (Wykes et al., Reference Wykes, Steel, Everitt and Tarrier2008). Regarding research implication, causal interventionist treatment trials (Kendler and Campbell, Reference Kendler and Campbell2009) are required to test the effectiveness of modifying safety behaviours in treating paranoia, but also social anxiety, in individuals with psychosis. Further research is required to confirm this theoretical concept and develop a brief intervention targeting safety behaviours. Thus, clinical interventions targeting safety behaviours may help alleviate anxious distress for individuals with psychosis.
To our knowledge, this is the first study to investigate the mediators that affect the relationship between social anxiety and paranoia in an Asian clinical population. The strengths of this study were as follows. First, consistent with reports from Western settings (Aherne, Reference Aherne2014; Matos et al., Reference Matos, Pinto-Gouveia and Gilbert2013; Newman Taylor and Stopa, Reference Newman Taylor and Stopa2013; Piccirillo and Heimberg, Reference Piccirillo and Heimberg2016), our findings offer a cross-cultural confirmation of a significant association between social anxiety and paranoia in Thailand. Second, we identified a potential mediator, namely safety behaviours, which is not only relevant to the theoretical understanding, but was also found to play a role in affecting the link between social anxiety and persecutory delusions. Therefore, these findings help shed some light on the intricacies involved as well as indicate the potential development of improved treatments for people with psychosis. However, this study has a few limitations. The COVID-19 pandemic during the data collection period could have affected the mental health and social functioning of our sample. Nevertheless, to avoid possible confounding, data were analysed by comparing the pre- and post-pandemic declaration groups and verified the outcomes by adjusting for the pre/post-COVID-19 pandemic variable during the data analyses. This identified only anxious avoidance as being significantly affected; no significant differences were observed between the pre- and post-pandemic group data after controlling for the pre/post-COVID-19 pandemic variable. Therefore, we consider it to be unlikely that the number of participants, which were recruited after the pandemic, has confounded our mediation outcomes. Additionally, we did not measure severity or duration of illness, which limits our understanding of the sample, including the proportion of people with long-standing psychosis.
Moreover, the use of the convenience sampling method might mean that our study participants may not be fully representative of this patient population owing to possible selection bias. Hence, further longitudinal work is warranted to support our findings, given this study’s limited ability to offer a causal explanation. Furthermore, the usefulness of safety behaviours relevant to the social anxiety–paranoia relationship remains to be elucidated in experimental studies. Larger clinical investigations examining stigma and shame related to cognitions within this relationship along with developing novel intervention designs should be undertaken.
Conclusion
Safety behaviours play a crucial role in the relationship between social anxiety and persecutory thoughts in individuals experiencing psychosis. The in situ safety behaviours was found to be a full mediator of this relationship. We also found that negative social appraisals (shame and stigma) as well as safety behaviours (in situ safety behaviours) were associated with paranoia. Emphasising causal and mechanistic approaches could produce robust findings related to safety behaviours facilitating the development of tailored interventions for addressing social anxiety and paranoia among individuals with psychosis.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1352465824000225
Data availability statement
Data will be available on reasonable request from the authors.
Acknowledgements
This study was conducted primarily by W.A. as part of a PhD program supervised by H.M. and A.G. The authors would like to thank S.L. and D.F. for their role in providing the O-CDQ tool, and all participants for their invaluable contributions to this study. They are also grateful to their co-author team for their excellent collaboration in various parts of this project.
Author contributions
Warut Aunjitsakul: Conceptualization (lead), Data curation (lead), Formal analysis (lead), Funding acquisition (lead), Investigation (lead), Methodology (lead), Project administration (lead), Resources (lead), Validation (lead), Visualization (lead), Writing – original draft (lead), Writing – review & editing (lead); Kreuwan Jongbowonwiwat: Data curation (lead), Funding acquisition- (supporting), Investigation- (supporting), Writing – review & editing- (supporting); Sinead Lambe: Investigation- (supporting), Methodology- (supporting), Validation- (supporting), Writing – review & editing- (supporting); Daniel Freeman: Investigation- (supporting), Methodology- (supporting), Supervision- (supporting), Validation- (supporting), Writing – review & editing- (supporting); Hamish McLeod: Conceptualization (equal), Formal analysis- (supporting), Investigation- (supporting), Methodology- (supporting), Supervision (lead), Validation- (supporting), Visualization (equal), Writing – original draft- (supporting), Writing – review & editing- (supporting); Andrew Gumley: Conceptualization (equal), Formal analysis- (supporting), Investigation- (supporting), Methodology- (supporting), Supervision (lead), Validation- (supporting), Visualization (equal), Writing – original draft- (supporting), Writing – review & editing- (supporting).
Financial support
This study was supported by the Faculty of Medicine, Prince of Songkla University, Thailand.
Competing interests
The authors declare none.
Ethical standards
The study was approved with reference number REC.62-394-3-1 by the Ethics Committee of the Faculty of Medicine, Prince of Songkla University, in accordance with the ethical principles of the Declaration of Helsinki. Ethical approval included informed consent for the results to be published.
Comments
No Comments have been published for this article.