Introduction
Negative symptoms, characterized by impaired motivation and emotional expression, are core features of schizophrenia spectrum disorders (SSDs) and are among the most reliable predictors of poor functional outcomes in affected individuals (Foussias & Remington, Reference Foussias and Remington2010). Nearly 20 years ago, Rector, Beck, and colleagues published seminal works on the cognitive model of negative symptoms, theorizing from a biopsychosocial perspective that negative symptoms represent maladaptive responses to dysfunctional beliefs, thought to emerge following life difficulties and setbacks that occur throughout the evolution of the disorder (Beck & Rector, Reference Beck and Rector2005; Rector, Beck, & Stolar, Reference Rector, Beck and Stolar2005). Briefly, the model posits a set of unhelpful or dysfunctional belief systems that are underscored by common themes of ‘What’s the point?,’ ‘It’s not worth it,’ and ‘I can’t handle it.,’ including defeatist performance beliefs, asocial beliefs, low expectancies for success, low expectancies for pleasure, internalized stigma, and perception of limited resources. More specifically, defeatist performance beliefs refer to overgeneralized or rigid negative beliefs about performing goal-directed activity (e.g. ‘If you cannot do something well, there is little point in doing it at all’ (Weissman & Beck, Reference Weissman and Beck1978); asocial beliefs are aversive and/or apathetic attitudes towards social engagement and affiliation (e.g. ‘Having close friends is not as important as most people say’ (Grant & Beck, Reference Grant and Beck2010)); low expectancies for success and pleasure, are characterized by generally negative expectations and low confidence for experiencing success or pleasure, respectively; internalized stigma in this context refers to self-limiting beliefs due to identification with negative stigmatizing views of mental illness (e.g. ‘What do you expect, I’m mentally ill’ (Rector et al., Reference Rector, Beck and Stolar2005); and perception of limited resources reflects the subjective sense of having diminished or insufficient physical and/or mental internal resources to expend effort or complete tasks. Importantly, these beliefs are putatively related to the development, expression, and maintenance of negative symptoms, and ostensibly represent mechanistic targets for psychological interventions such as cognitive behavioral therapy (CBT), which has thus far demonstrated fairly mixed results for its efficacy in treating negative symptoms (Cella et al., Reference Cella, Roberts, Pillny, Riehle, O’Donoghue, Lyne and Preti2023).
As research on the cognitive model of negative symptoms has proliferated and evolved over the past two decades, it is imperative to synthesize this clinically rich body of work, and for the first time since its inception, comprehensively validate the hypothesized model. This approach may also serve to highlight a broad range of potential psychotherapeutic targets that can be translated into optimized CBT interventions for negative symptoms. Thus, a systematic review and meta-analysis were undertaken to empirically quantify the relationships between negative symptoms and each of the six primary dysfunctional beliefs postulated within the model: defeatist performance beliefs, asocial beliefs, low expectancies for success, low expectancies for pleasure, internalized stigma, and perception of limited resources.
Methods
This review was registered with PROSPERO (CRD42021245467), completed in Covidence, and followed PRISMA guidelines (Moher, Liberati, Tetzlaff, Altman, & Group, Reference Moher, Liberati, Tetzlaff, Altman and Group2009).
Search strategy
An electronic search was conducted on MEDLINE and PsycINFO, from inception through January 2024. A broad search strategy was developed in consultation with an academic librarian (see supplementary methods). We also conducted a manual review of reference lists and Google Scholar for additional articles.
Selection criteria
Titles and abstracts were independently screened for relevance (S.S., J.P., G.F.), after which full-text articles were assessed for eligibility using the following inclusion criteria: 1) written in English; 2) majority of sample (≥85%) diagnosed with non-affective SSDs, excluding schizotypal personality disorder or clinically high risk/prodrome (if <85%, study authors were contacted to request data with ineligible patients excluded); 3) includes valid measure of negative symptoms (global or subdomains); 4) includes measure of at least one relevant dysfunctional belief; 5) original data published in a peer-reviewed journal. For studies containing overlapping samples, the study with data for the larger sample size was included. Studies were also required to report statistics on the direct relationship between negative symptoms and dysfunctional beliefs, with sufficient data to estimate an effect size. Where otherwise eligible studies did not report this data, the author was contacted to request it. Only baseline cross-sectional data was included.
To enable a comprehensive review, we considered all instruments that appeared to evaluate relevant beliefs. Measures were only included if they assessed cognitions (i.e. beliefs, attitudes, expectancies); in the case of asocial beliefs and low expectancies for pleasure, we excluded instruments tapping into hedonic experiences (i.e. social anhedonia and anticipatory pleasure, respectively). Measures of stigma experiences or perceived societal stigma were also excluded. The final decision for inclusion was made by S.S. and G.F. based on consideration of the instrument’s use in other studies, as well as face validity in relation to the theoretical model.
Data extraction
In addition to effect size statistics, study characteristics including sample size, country of origin, mean age, % male, diagnoses, illness duration, and negative symptom severity were extracted. Study authors were contacted to request this data if unpublished.
Quality assessment
The methodological quality of studies was assessed using an adapted version of the Quality Assessment Tool for Quantitative Studies (Thomas, Ciliska, Dobbins, & Micucci, Reference Thomas, Ciliska, Dobbins and Micucci2004), with sections pertaining to randomized controlled trials removed and only relevant indicators of quality included (see supplementary methods).
Data analysis
Six meta-analyses were conducted to quantify the relationship between overall negative symptoms and each of the following dysfunctional beliefs: defeatist performance beliefs, asocial beliefs, low expectancies for success, low expectancies for pleasure, internalized stigma, and perception of limited resources. The direction of the reported correlation coefficient was reverse-coded, where necessary, so that an association between more severe negative symptoms and greater endorsement of dysfunctional beliefs is indicated by: 1) a positive correlation for defeatist performance beliefs, asocial beliefs, internalized stigma, and perception of limited resources; and 2) a negative correlation for expectancies for success and expectancies for pleasure to reflect more negative (i.e. lower) expectations.
All analyses were performed using the ‘metafor’ package in R software for statistical computation (R Core Team, 2013; Viechtbauer, Reference Viechtbauer2010). Pearson’s r correlation coefficient was used for effect sizes. Two studies (Beck, Grant, Huh, Perivoliotis, & Chang, Reference Beck, Grant, Huh, Perivoliotis and Chang2013; Bentall et al., Reference Bentall, Simpson, Lee, Williams, Elves, Brabbins and Morrison2010) used t-tests to compare dysfunctional beliefs in participants with high versus low negative symptoms, which were converted from Cohen’s d to r (Rosenthal, Reference Rosenthal1991). Spearman correlations were included in analyses when Pearson’s correlations could not be obtained.
Random-effects models with restricted maximum-likelihood estimators were used to pool effect size estimates of z-transformed Pearson’s r correlations. The primary analyses assessed the relationships between dysfunctional beliefs and global negative symptoms. If only subscale correlations were reported for either negative symptoms or dysfunctional beliefs instead of their total scores, they were averaged using Fisher’s r-to-z transformation to estimate an overall effect size. Missing data for nonsignificant correlations were imputed with a zero. Subset analyses were also conducted for the negative symptom subdomains of diminished motivation (i.e. avolition/apathy, anhedonia, and asociality) and diminished emotional expression (i.e. blunted affect and alogia), as well as the different measurement instruments used for both negative symptoms and dysfunctional beliefs. Separate meta-regression analyses were also completed to examine the moderating role of age, sex, illness duration, negative symptom severity, and study quality, respectively. Given the variability in negative symptom scales used across studies, we applied a min-max normalization, resulting in a standardized scale of 0–10, with higher scores reflecting greater symptom severity. A minimum three-study threshold was used for all analyses.
I2 and τ2 statistics are reported as metrics of between-study heterogeneity. Several leave-one-out diagnostics (e.g. externally standardized residual, Cook’s distance, and covariance ratio) were computed using the ‘influence’ function in R to identify influential outlier studies. Publication bias was evaluated using Rosenthal’s fail-safe N and visual inspection of funnel plots.
Results
Selection of studies
The PRISMA flowchart is illustrated in Figure 1. A total of 176 studies were included across meta-analyses. Study sample characteristics are summarized in Table 1 and fully described in the supplement.
Flowchart illustrating the process of screening and selecting studies for meta-analysis.
*Note: Reflects the number of studies only for which effect size data was requested, and without which the study would not be eligible for inclusion.
Summary of sample characteristics for studies included in the primary meta-analysis for each dysfunctional belief system
Meta-analyses
A summary of the primary pooled effects is illustrated in Figure 2. Subset analyses for all measures included in each primary meta-analysis, with k ≥ 3, are presented in Table 2. A full description of belief measures is presented in Table S1.
Summary of pooled results for primary random-effects meta-analyses evaluating relationships between overall negative symptoms and dysfunctional belief systems. An association of more severe negative symptoms with greater dysfunctional beliefs is represented by positive effect sizes for defeatist performance beliefs, asocial beliefs, internalized stigma, and perception of limited resources, and negative effect sizes for low expectancies for success and low expectancies for pleasure to reflect more negative (i.e. lower) expectations.
Summary of subset analyses for different measures used to evaluate dysfunctional beliefs and negative systems, with k ≥ 3, in each primary meta-analysis
Abbreviations: ABS: Asocial Beliefs Scale; BNSS: Brief Negative Symptom Scale; BPRS: Brief Psychiatric Rating Scale; CAINS: Clinical Assessment Interview for Negative Symptoms; DAS-DPB: Dysfunctional Attitudes Scale – Defeatist Performance Beliefs; ES: Empowerment Scale – Self-Efficacy; GSES: General Self-Efficacy Scale; ISMI: Internalized Stigma of Mental Illness Scale; MCQ: Metacognitions Questionnaire – Cognitive Confidence; PANSS: Positive and Negative Syndrome Scale; Personal Beliefs about Illness Questionnaire – Entrapment/Expectations; PCS: Perceived Competency Scale; SANS: Scale for the Assessment of Negative Symptoms; RAS: Recovery Assessment Scale – Success and Goal Orientation; RSA: Resilience Scale for Adults – Personal Strength; RSES: Revised Self-Efficacy Scale; SARA-Q: Success and Resources Appraisals Questionnaire; SSS-S: Self-Stigma Scale – Short.
* Note: k < 3 for all measures of low expectancies for pleasure.
Defeatist performance beliefs
Study characteristics
Forty studies were included in the meta-analysis evaluating the relationship between negative symptoms and defeatist performance beliefs (Beck et al., Reference Beck, Grant, Huh, Perivoliotis and Chang2013; Bennett, Brown, Fang, & Blanchard, Reference Bennett, Brown, Fang and Blanchard2023; Berry & Greenwood, Reference Berry and Greenwood2018; Buchanan et al., Reference Buchanan, Kelly, Strauss, Gold, Weiner, Zaranski and Granholm2021; Clay, Raugh, Bartolomeo, & Strauss, Reference Clay, Raugh, Bartolomeo and Strauss2020; Couture, Blanchard, & Bennett, Reference Couture, Blanchard and Bennett2011; Décombe et al., Reference Décombe, Salesse, Jourdan, Laraki, Capdevielle and Raffard2021; Ebrahimi et al., Reference Ebrahimi, Poursharifi, Dolatshahi, Rezaee, Hassanabadi and Naeem2021; Fisher et al., Reference Fisher, Etter, Murray, Ghiasi, LaCross, Ramsay and Vinogradov2023; Granholm et al., Reference Granholm, Holden, Dwyer, Mikhael, Link and Depp2020; Granholm et al., Reference Granholm, Twamley, Mahmood, Keller, Lykins, Parrish and Holden2022; Granholm, Holden, Dwyer, & Link, Reference Granholm, Holden, Dwyer and Link2020; Granholm, Holden, Link, & McQuaid, Reference Granholm, Holden, Link and McQuaid2014; Granholm, Holden, Link, McQuaid, & Jeste, Reference Granholm, Holden, Link, McQuaid and Jeste2013; Grant & Beck, Reference Grant and Beck2009; Green et al., Reference Green, Wynn, Gabrielian, Hellemann, Horan, Kern and Sugar2022; Green, Hellemann, Horan, Lee, & Wynn, Reference Green, Hellemann, Horan, Lee and Wynn2012; Kiwanuka, Strauss, McMahon, & Gold, Reference Kiwanuka, Strauss, McMahon and Gold2014; Lee & Yu, Reference Lee and Yu2023; Lincoln et al., Reference Lincoln, Mehl, Ziegler, Kesting, Exner and Rief2010; Luther et al., Reference Luther, Westbrook, Ayawvi, Ruiz, Raugh, Chu and Strauss2023; McGovern, Reddy, Reavis, & Green, Reference McGovern, Reddy, Reavis and Green2020; Park, Bennett, Couture, & Blanchard, Reference Park, Bennett, Couture and Blanchard2013; Paul, Strauss, Gates-Woodyatt, Barchard, & Allen, Reference Paul, Strauss, Gates-Woodyatt, Barchard and Allen2023; Pillny, Krkovic, & Lincoln, Reference Pillny, Krkovic and Lincoln2018; Pillny & Lincoln, Reference Pillny and Lincoln2016; Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019; Raugh & Strauss, Reference Raugh and Strauss2024; Rector, Reference Rector2004; Reddy et al., Reference Reddy, Horan, Barch, Buchanan, Gold, Marder and Green2018; Romanowska & Best, Reference Romanowska and Best2023; Saperia et al., Reference Saperia, Da Silva, Siddiqui, Agid, Daskalakis, Ravindran and Foussias2019; Shaheen & Amin, Reference Shaheen and Amin2016; Staring, ter Huurne, & van der Gaag, Reference Staring, ter Huurne and van der Gaag2013; Strauss & Gold, Reference Strauss and Gold2016; Strauss, Morra, Sullivan, & Gold, Reference Strauss, Morra, Sullivan and Gold2015; Takeda et al., Reference Takeda, Nakataki, Ohta, Hamatani, Matsuura, Yoshida and Ohmori2019; Thonon, Levaux, Della Libera, & Larøi, Reference Thonon, Levaux, Della Libera and Larøi2020; Ventura et al., Reference Ventura, Subotnik, Ered, Gretchen-Doorly, Hellemann, Vaskinn and Nuechterlein2014; Zhang, James, & Strauss, Reference Zhang, James and Strauss2023), with 38 studies used in the main analysis with global negative symptoms (Table S2) and 31 studies used in the subdomain analyses (Table S2.1).
The main analysis (k = 38) consisted of 2808 unique participants, most of whom were outpatients, and in approximately half of the studies were described as clinically stable. The majority of these participants were male (64.8%, k = 37), and the mean age was 41.2 years (k = 37). Two studies selected younger participants between the ages of 18–36 as part of their inclusion criteria (Berry & Greenwood, Reference Berry and Greenwood2018; Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019), while another two studies only included older participants over the age of 45 (Granholm et al., Reference Granholm, Holden, Link, McQuaid and Jeste2013; Granholm, Holden, Dwyer, & Link, Reference Granholm, Holden, Dwyer and Link2020). Duration of illness data was available for 24 studies (M = 17.2 years). While most of these publications reported on patients experiencing chronic illness, three studies investigated early episode psychosis (Berry & Greenwood, Reference Berry and Greenwood2018; Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019; Ventura et al., Reference Ventura, Subotnik, Ered, Gretchen-Doorly, Hellemann, Vaskinn and Nuechterlein2014). With respect to negative symptom severity, data was obtained for 32 studies (M = 3.5 on a standardized scale from 0 to 10). Nine studies specifically recruited patients with negative symptoms, with three studies requiring at least mild severity (Buchanan et al., Reference Buchanan, Kelly, Strauss, Gold, Weiner, Zaranski and Granholm2021; Pillny et al., Reference Pillny, Krkovic and Lincoln2018; Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019), five studies with moderate–severe cut-offs (Bennett et al., Reference Bennett, Brown, Fang and Blanchard2023; Ebrahimi et al., Reference Ebrahimi, Poursharifi, Dolatshahi, Rezaee, Hassanabadi and Naeem2021; Granholm et al., Reference Granholm, Twamley, Mahmood, Keller, Lykins, Parrish and Holden2022; Granholm, Holden, Dwyer, Mikhael, et al., Reference Granholm, Holden, Dwyer, Mikhael, Link and Depp2020; Staring et al., Reference Staring, ter Huurne and van der Gaag2013), and one study including a subgroup of patients with deficit syndrome (Beck et al., Reference Beck, Grant, Huh, Perivoliotis and Chang2013). Additionally, five studies excluded participants if they were in the acute phase of illness (Buchanan et al., Reference Buchanan, Kelly, Strauss, Gold, Weiner, Zaranski and Granholm2021; Ebrahimi et al., Reference Ebrahimi, Poursharifi, Dolatshahi, Rezaee, Hassanabadi and Naeem2021; Lee & Yu, Reference Lee and Yu2023) or experienced positive symptoms greater than moderate severity (Granholm et al., Reference Granholm, Twamley, Mahmood, Keller, Lykins, Parrish and Holden2022; Granholm, Holden, Dwyer, Mikhael, et al., Reference Granholm, Holden, Dwyer, Mikhael, Link and Depp2020). Only one study selected patients based on the severity of defeatist performance beliefs (Granholm, Holden, Dwyer, Mikhael, et al., Reference Granholm, Holden, Dwyer, Mikhael, Link and Depp2020).
Meta-Analysis
The random-effects meta-analysis for the relationship between overall negative symptoms and defeatist performance beliefs (k = 38, n = 2808) revealed a significant correlation, r = 0.23 (95% CI, 0.18–0.27, Z = 9.9, p < 0.0001; Figure S1). Heterogeneity across studies was relatively small (I2 = 25.3%, τ2 = 0.005) and there were no significant moderators (Table S4). Visual inspection of the funnel plot (Figure S2), as well as Rosenthal’s fail-safe N = 1702, suggests publication bias is unlikely. Subset analyses for negative symptom subdomains (Table S3) revealed significant effects for diminished motivation (k = 31, n = 2452), r = 0.19 (95% CI, 0.14–0.23, Z = 8.8, p < 0.0001) and diminished expression (k = 24, n = 2053), r = 0.19 (95% CI, 0.12–0.25, Z = 5.8, p < 0.0001), with substantially more heterogeneity observed across studies in the diminished expression analysis (I2 = 45.3%, τ2 = 0.01) compared to diminished motivation (I2 = 4.3%, τ2 = 0.0006).
Asocial beliefs
Study characteristics
Eight studies (Beck et al., Reference Beck, Grant, Huh, Perivoliotis and Chang2013; Buchanan et al., Reference Buchanan, Kelly, Strauss, Gold, Weiner, Zaranski and Granholm2021; Granholm et al., Reference Granholm, Twamley, Mahmood, Keller, Lykins, Parrish and Holden2022; Granholm, Holden, Dwyer, Mikhael, et al., Reference Granholm, Holden, Dwyer, Mikhael, Link and Depp2020; Grant & Beck, Reference Grant and Beck2010; Le, Holden, Link, & Granholm, Reference Le, Holden, Link and Granholm2018; Pillny et al., Reference Pillny, Krkovic and Lincoln2018; Zhang et al., Reference Zhang, James and Strauss2023) were included in the primary meta-analysis assessing the relationship between negative symptoms and asocial beliefs (Table S7), with five of these studies also being used for the subdomain analyses (Table S7.1). The 578 participants across the eight studies were all outpatients, primarily male (60.7%), with a mean age of 43 years, 23.2 years of illness (k = 5), and an average negative symptom severity score of 3.7. Five studies specifically recruited participants who endorsed negative symptoms (Beck et al., Reference Beck, Grant, Huh, Perivoliotis and Chang2013; Buchanan et al., Reference Buchanan, Kelly, Strauss, Gold, Weiner, Zaranski and Granholm2021; Granholm et al., Reference Granholm, Twamley, Mahmood, Keller, Lykins, Parrish and Holden2022; Grant & Beck, Reference Grant and Beck2010; Pillny et al., Reference Pillny, Krkovic and Lincoln2018).
Meta-Analysis
The meta-analysis revealed a significant correlation between overall negative symptoms and asocial beliefs (k = 8, n = 578), r = 0.21 (95% CI, 0.12–0.28, Z = 4.9, p < 0.0001; Figure S4). There was no heterogeneity across studies (I2 = 0%, τ2 = 0) and no significant moderators (Table S9). Publication bias is unlikely based on the funnel plot’s relative symmetry (Figure S5), and Rosenthal’s fail-safe N = 53. Subdomain analyses (Table S8) revealed a significant effect for diminished motivation (k = 5, n = 215), r = 0.15 (95% CI, 0.02–0.28, Z = 2.2, p = 0.03), but not diminished expression (k = 4, n = 184), r = 0.09 (95% CI, −0.06 – 0.24, Z = 1.2, p = 0.2). No heterogeneity was observed for either subdomain.
Low expectancies for success
Study characteristics
Sixty studies in total examined the relationship between negative symptoms and low expectancies for success (Avery, Startup, & Calabria, Reference Avery, Startup and Calabria2009; Beaudette, Cruz, Lukachko, Roché, & Silverstein, Reference Beaudette, Cruz, Lukachko, Roché and Silverstein2020; Bentall et al., Reference Bentall, Simpson, Lee, Williams, Elves, Brabbins and Morrison2010; Best, Milanovic, Iftene, & Bowie, Reference Best, Milanovic, Iftene and Bowie2019; Caqueo-Urízar, Ponce-Correa, Semir-González, & Urzúa, Reference Caqueo-Urízar, Ponce-Correa, Semir-González and Urzúa2022; Cardenas et al., Reference Cardenas, Abel, Bowie, Tiznado, Depp, Patterson and Mausbach2013; Cavelti, Wirtz, Corrigan, & Vauth, Reference Cavelti, Wirtz, Corrigan and Vauth2017; Chang et al., Reference Chang, Cheung, Hui, Lin, Chan, Lee and Chen2015, Reference Chang, Kwong, Hui, Chan, Lee and Chen2017; Cheng, Nadin, Bohonis, Katt, & Dewa, Reference Cheng, Nadin, Bohonis, Katt and Dewa2023; Chino, Nemoto, Fujii, & Mizuno, Reference Chino, Nemoto, Fujii and Mizuno2009; Choi, Fiszdon, & Medalia, Reference Choi, Fiszdon and Medalia2010; Choi, Saperstein, & Medalia, Reference Choi, Saperstein and Medalia2012; Chrostek, Grygiel, Anczewska, Wciórka, & Świtaj, Reference Chrostek, Grygiel, Anczewska, Wciórka and Świtaj2016; Clari et al., Reference Clari, Headley, Egger, Swai, Lawala, Minja and Baumgartner2022; Cowan, Lundin, Moe, & Breitborde, Reference Cowan, Lundin, Moe and Breitborde2023; Fisher et al., Reference Fisher, Etter, Murray, Ghiasi, LaCross, Ramsay and Vinogradov2023; Fiszdon, Kurtz, Choi, Bell, & Martino, Reference Fiszdon, Kurtz, Choi, Bell and Martino2016; Galliot et al., Reference Galliot, Sanchez-Rodriguez, Belloc, Phulpin, Icher, Birmes and Gozé2022; García-Mieres, Lysaker, & Leonhardt, Reference García-Mieres, Lysaker and Leonhardt2022; González-Domínguez, González-Sanguino, & Muñoz, Reference González-Domínguez, González-Sanguino and Muñoz2019; Gruber et al., Reference Gruber, Rumpold, Schrank, Sibitz, Otzelberger, Jahn and Unger2020; Haugen, Stubberud, Ueland, Haug, & Øie, Reference Haugen, Stubberud, Ueland, Haug and Øie2021; Hayward et al., Reference Hayward, David, Green, Rabe-Hesketh, Haworth, Thompson and Wykes2009; Herpertz et al., Reference Herpertz, Richter, Barkhau, Storck, Blitz, Steinmann and Opel2022; Hill & Startup, Reference Hill and Startup2013; Huddy, Drake, & Wykes, Reference Huddy, Drake and Wykes2016; Izydorczyk, Sitnik-Warchulska, Kühn-Dymecka, & Lizińczyk, Reference Izydorczyk, Sitnik-Warchulska, Kühn-Dymecka and Lizińczyk2019; Keefe et al., Reference Keefe, Vinogradov, Medalia, Buckley, Caroff, D’Souza and Stroup2012; Kinoshita, Hashimoto, Nishimura, & Yotsumoto, Reference Kinoshita, Hashimoto, Nishimura and Yotsumoto2023; Kukla, Strasburger, Salyers, Rollins, & Lysaker, Reference Kukla, Strasburger, Salyers, Rollins and Lysaker2021; Kurtz, Olfson, & Rose, Reference Kurtz, Olfson and Rose2013; Laxmi, Sahoo, Grover, & Nehra, Reference Laxmi, Sahoo, Grover and Nehra2023; Lee et al., Reference Lee, Jang, Lee, Park, Medalia and Choi2017, Reference Lee, Lawrence, Bryce, Ponsford, Tan and Rossell2021; Li, Wu, & Chen, Reference Li, Wu and Chen2023; Lim, Li, Xie, Tan, & Lee, Reference Lim, Li, Xie, Tan and Lee2021; Luther et al., Reference Luther, Fukui, Firmin, McGuire, White, Minor and Salyers2015; Lysaker, Clements, Wright, Evans, & Marks, Reference Lysaker, Clements, Wright, Evans and Marks2001; Markiewicz & Dobrowolska, Reference Markiewicz and Dobrowolska2021; Melau et al., Reference Melau, Harder, Jeppesen, Hjorthøj, Jepsen, Thorup and Nordentoft2015; Morgades-Bamba, Fuster-Ruizdeapodaca, & Molero, Reference Morgades-Bamba, Fuster-Ruizdeapodaca and Molero2019; Murphy et al., Reference Murphy, Hogarth, Reynolds, Wood-Ross, Vanrooy and Bowie2023; Norman, Windell, Lynch, & Manchanda, Reference Norman, Windell, Lynch and Manchanda2013; Pratt, Mueser, Smith, & Lu, Reference Pratt, Mueser, Smith and Lu2005; Priebe et al., Reference Priebe, Kelley, Omer, Golden, Walsh, Khanom and McCabe2015; Raffard et al., Reference Raffard, Bortolon, Rolland, Capdevielle, Boulenger, Gely-Nargeot and Stephan2014; Rossi et al., Reference Rossi, Galderisi, Rocca, Bertolino, Rucci, Gibertoni and Maj2017; Santosh & Kundu, Reference Santosh and Kundu2023; Şenormancı et al., Reference Şenormancı, Korkmaz, Şenormancı, Uğur, Topsaç and Gültekin2021; Şenormanci, Güçlü, & Şenormanci, Reference Şenormanci, Güçlü and Şenormanci2022; Song et al., Reference Song, Kang, Kim, Lee, Lee and An2013; Strauss et al., Reference Strauss, Morra, Sullivan and Gold2015; Thonon et al., Reference Thonon, Levaux, Della Libera and Larøi2020; Vaskinn, Ventura, Andreassen, Melle, & Sundet, Reference Vaskinn, Ventura, Andreassen, Melle and Sundet2015; Vauth, Kleim, Wirtz, & Corrigan, Reference Vauth, Kleim, Wirtz and Corrigan2007; Ventura et al., Reference Ventura, Subotnik, Ered, Gretchen-Doorly, Hellemann, Vaskinn and Nuechterlein2014; Watanabe, Taniguchi, & Sugihara, Reference Watanabe, Taniguchi and Sugihara2022; Wciórka, Świtaj, & Anczewska, Reference Wciórka, Świtaj and Anczewska2015; Wright et al., Reference Wright, Browne, Cather, Pratt, Bartels and Mueser2021). Fifty-five of these studies were used for the primary analysis for the relationship with overall negative symptoms (Table S12), and 14 were included in the subdomain analyses (Table S12.1).
Within the 55 studies included within the primary analysis were 5664 unique participants who were mostly male (65%), outpatients, and with a mean age of 37.6 years. Five studies exclusively included younger participants (Cheng et al., Reference Cheng, Nadin, Bohonis, Katt and Dewa2023; Chino et al., Reference Chino, Nemoto, Fujii and Mizuno2009; Melau et al., Reference Melau, Harder, Jeppesen, Hjorthøj, Jepsen, Thorup and Nordentoft2015; Murphy et al., Reference Murphy, Hogarth, Reynolds, Wood-Ross, Vanrooy and Bowie2023; Song et al., Reference Song, Kang, Kim, Lee, Lee and An2013), while another two studies only recruited older participants (Cardenas et al., Reference Cardenas, Abel, Bowie, Tiznado, Depp, Patterson and Mausbach2013; Wright et al., Reference Wright, Browne, Cather, Pratt, Bartels and Mueser2021). The average duration of illness was 12.8 years (k = 43), with nine studies specifically selecting first/early episode patients (Chang et al., Reference Chang, Cheung, Hui, Lin, Chan, Lee and Chen2015; Cheng et al., Reference Cheng, Nadin, Bohonis, Katt and Dewa2023; Cowan et al., Reference Cowan, Lundin, Moe and Breitborde2023; Laxmi et al., Reference Laxmi, Sahoo, Grover and Nehra2023; Melau et al., Reference Melau, Harder, Jeppesen, Hjorthøj, Jepsen, Thorup and Nordentoft2015; Murphy et al., Reference Murphy, Hogarth, Reynolds, Wood-Ross, Vanrooy and Bowie2023; Norman et al., Reference Norman, Windell, Lynch and Manchanda2013; Song et al., Reference Song, Kang, Kim, Lee, Lee and An2013; Ventura et al., Reference Ventura, Subotnik, Ered, Gretchen-Doorly, Hellemann, Vaskinn and Nuechterlein2014). The mean negative symptom severity score was 2.9 (k = 46). Samples were characterized as clinically stable in nearly half of the included studies.
Meta-analysis
The meta-analysis (k = 55, n = 5664) revealed a significant inverse correlation, r = −0.21 (95% CI, −0.15 – −0.26, Z = −7.3, p < 0.0001; Figure S7), such that more severe overall negative symptoms were associated with lower expectancies for success. There was substantial heterogeneity across studies (I2 = 74.2%, τ2 = 0.03), with age (ß = 0.007, p = 0.03) and illness duration (ß = 0.01, p = 0.02) as significant moderators (Table S14). One outlier study was identified (Santosh & Kundu, Reference Santosh and Kundu2023) (Figure S9); though results were mostly unchanged when excluded (Tables S17–S19), but with reduced heterogeneity (I2 = 67.3%) and only non-significant trends for moderation (age (p = 0.052), duration of illness (p = 0.07)). Publication bias is unlikely based on the funnel plots (Figures S8, S10), and Rosenthal’s fail-safe N = 3936. Subdomain analyses (Table S13) revealed significant effects for diminished motivation (k = 14, n = 2297), r = −0.33 (95% CI, −0.22 – −0.42, Z = −5.9, p < 0.0001) and diminished expression (k = 9, n = 1898), r = −0.21 (95% CI, −0.06 – −0.34, Z = −2.8, p = 0.005). Substantial heterogeneity was observed for both diminished motivation (I2 = 81.4%, τ2 = 0.03) and diminished expression (I2 = 85.7%, τ2 = 0.04).
Low expectancies for pleasure
Study characteristics
Five studies investigated the relationship between negative symptoms and low expectancies for pleasure (Beck et al., Reference Beck, Grant, Huh, Perivoliotis and Chang2013; Hartmann et al., Reference Hartmann, Hager, Reimann, Chumbley, Kirschner, Seifritz and Kaiser2015; Hu et al., Reference Hu, Jiang, Shan, Chu, Lv, Yi and Chan2022; Strauss et al., Reference Strauss, Morra, Sullivan and Gold2015; Yang et al., Reference Yang, Yang, Zou, Shi, Wang, Xie and Chan2018). One study (Hartmann et al., Reference Hartmann, Hager, Reimann, Chumbley, Kirschner, Seifritz and Kaiser2015) evaluated this association only with the amotivation subdomain and not global negative symptoms, but it was nonetheless included in the primary analysis in an attempt to maximize power (see Table S21 for sensitivity analysis).
The five studies consisted of 249 primarily outpatient participants, who were mostly male (66.4%), with a mean age of 35 years, 12.5 years of illness (k = 3), and an average negative symptom severity score of 2.7 (k = 4). One study included a subset of participants who were classified as having deficit syndrome (Beck et al., Reference Beck, Grant, Huh, Perivoliotis and Chang2013).
Meta-analysis
A significant inverse correlation emerged (k = 5, n = 249), r = −0.19 (95% CI, −0.06 – −0.31, Z = −2.9, p = 0.003; Figure S11), such that more severe negative symptoms were associated with lower expectancies for pleasure. There was no heterogeneity across studies and no significant moderators (Table S22). One study emerged as an outlier (Beck et al., Reference Beck, Grant, Huh, Perivoliotis and Chang2013) (Figure S13), and its exclusion led to a somewhat larger effect size (r = −0.23) (Tables S24–S25). Based on inspection of funnel plots (Figures S12, S14), Rosenthal’s fail-safe N = 13, and the small number of studies included, these results are potentially susceptible to publication bias. There was no effect for the diminished motivation subdomain analysis (k = 3, n = 110), r = −0.21, (95% CI, 0.10 – −0.48, Z = −1.3, p = 0.2; Table S21).
Internalized stigma
Study characteristics
The meta-analysis for the relationship between negative symptoms and internalized stigma included 83 studies in total (Acosta, Navarro, Cabrera, Ramallo-Fariña, & Martínez, Reference Acosta, Navarro, Cabrera, Ramallo-Fariña and Martínez2020; Barlati et al., Reference Barlati, Morena, Nibbio, Cacciani, Corsini, Mosca and Vita2022; Berry & Greenwood, Reference Berry and Greenwood2018; Campellone, Caponigro, & Kring, Reference Campellone, Caponigro and Kring2014; Capatina & Miclutia, Reference Capatina and Miclutia2018; Caqueo-Urízar et al., Reference Caqueo-Urízar, Ponce-Correa, Semir-González and Urzúa2022, Reference Caqueo-Urízar, Urzúa, Loundon, Boucekine, Fond and Boyer2019; Chan et al., Reference Chan, Kao, Leung, Hui, Lee, Chang and Chen2019; Chen et al., Reference Chen, Chang, Hui, Chan, Lee and Chen2016; Chrostek et al., Reference Chrostek, Grygiel, Anczewska, Wciórka and Świtaj2016; Chu et al., Reference Chu, Ng, Chu, Lui, Lau, Chan and Chang2023; Clari et al., Reference Clari, Headley, Egger, Swai, Lawala, Minja and Baumgartner2022; Cloutier et al., Reference Cloutier, Lecomte, Diotte, Lamontagne, Abdel-Baki, Daneault and Perrine2023; Degnan, Berry, Vaughan, Crossley, & Edge, Reference Degnan, Berry, Vaughan, Crossley and Edge2022; DeLuca et al., Reference DeLuca, Akouri-Shan, Jay, Redman, Petti, Lucksted and Schiffman2021; DeTore et al., Reference DeTore, Balogun-Mwangi, Tepper, Cather, Russinova, Lanca and Mueser2022; Fekih-Romdhane, Hajje, Haddad, Hallit, & Azar, Reference Fekih-Romdhane, Hajje, Haddad, Hallit and Azar2023; Feldhaus et al., Reference Feldhaus, Falke, von Gruchalla, Maisch, Uhlmann, Bock and Lencer2018; Firmin et al., Reference Firmin, Lysaker, Luther, Yanos, Leonhardt, Breier and Vohs2019; González-Domínguez et al., Reference González-Domínguez, González-Sanguino and Muñoz2019; Grover et al., Reference Grover, Avasthi, Singh, Dan, Neogi, Kaur and Behere2017; Grover, Sahoo, Chakrabarti, & Avasthi, Reference Grover, Sahoo, Chakrabarti and Avasthi2018; Gruber et al., Reference Gruber, Rumpold, Schrank, Sibitz, Otzelberger, Jahn and Unger2020; Hill & Startup, Reference Hill and Startup2013; Hofer et al., Reference Hofer, Mizuno, Frajo-Apor, Kemmler, Suzuki, Pardeller and Uchida2016, Reference Hofer, Post, Pardeller, Frajo-Apor, Hoertnagl, Kemmler and Fleischhacker2019; Horsselenberg, Busschbach, Aleman, & Pijnenborg, Reference Horsselenberg, Busschbach, Aleman and Pijnenborg2016; Huang, Liu, & Yang, Reference Huang, Liu and Yang2023; Ipci et al., Reference Ipci, yıldız, İncedere, Kiras, Esen and Gürcan2020; Jian et al., Reference Jian, Wang, Lin, Huang, Yeh, Liu and Yen2022; Karidi et al., Reference Karidi, Vasilopoulou, Savvidou, Vitoratou, Rabavilas and Stefanis2014, Reference Karidi, Vassilopoulou, Savvidou, Vitoratou, Maillis, Rabavilas and Stefanis2015; Khalaf, Fathy, Ebrahim, & Samie, Reference Khalaf, Fathy, Ebrahim and Samie2023; Koçak et al., Reference Koçak, Rıfat Şahin, Güz, Böke, Sarısoy and Karabekiroğlu2022; Konsztowicz, Gelencser, Otis, Schmitz, & Lepage, Reference Konsztowicz, Gelencser, Otis, Schmitz and Lepage2021; Krzyzanowski, Agid, Goghari, & Remington, Reference Krzyzanowski, Agid, Goghari and Remington2021; Laxmi et al., Reference Laxmi, Sahoo, Grover and Nehra2023; Li et al., Reference Li, Guo, Huang, Liu, Chen, Zhang and Thornicroft2017; Lien et al., Reference Lien, Chang, Kao, Tzeng, Lu and Loh2018; Lim et al., Reference Lim, Li, Xie, Tan and Lee2021; Lo et al., Reference Lo, Huber, Meyer, Weinmann, Luethi, Dechent and Moeller2022; Luciano et al., Reference Luciano, Sampogna, Del Vecchio, Giallonardo, Palummo, Andriola and Fiorillo2021; Lv, Wolf, & Wang, Reference Lv, Wolf and Wang2013; Lysaker, Davis, Warman, Strasburger, & Beattie, Reference Lysaker, Davis, Warman, Strasburger and Beattie2007; Lysaker, Roe, & Yanos, Reference Lysaker, Roe and Yanos2007; Ma et al., Reference Ma, Ju, Xia, Pan, Gao, Zhang and Zhu2023; MacDougall, Vandermeer, & Norman, Reference MacDougall, Vandermeer and Norman2015; Margetić, Jakovljević, Ivanec, Margetić, & Tošić, Reference Margetić, Jakovljević, Ivanec, Margetić and Tošić2010; Morgades-Bamba et al., Reference Morgades-Bamba, Fuster-Ruizdeapodaca and Molero2019; Murphy et al., Reference Murphy, Hogarth, Reynolds, Wood-Ross, Vanrooy and Bowie2023; Nabors et al., Reference Nabors, Yanos, Roe, Hasson-Ohayon, Leonhardt, Buck and Lysaker2014; Ng, Yu, & Leung, Reference Ng, Yu and Leung2024; O’Connor, Yanos, & Firmin, Reference O’Connor, Yanos and Firmin2018; Ordóñez-Camblor, Paino, Fonseca-Pedrero, & Pizarro-Ruiz, Reference Ordóñez-Camblor, Paino, Fonseca-Pedrero and Pizarro-Ruiz2021; Park et al., Reference Park, Bennett, Couture and Blanchard2013; Pérez-Aguado et al., Reference Pérez-Aguado, Lacamara, Ruiz, Dasí, Soldevila-Matias and Fuentes-Durá2024; Pishdadian et al., Reference Pishdadian, Martins, Milanovic, Doell, Kidd and Grossman2023; Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019; Prouteau, Roux, Destaillats, & Bergua, Reference Prouteau, Roux, Destaillats and Bergua2017; Reneses et al., Reference Reneses, Sevilla-Llewellyn-Jones, Vila-Badia, Palomo, Lopez-Micó, Pereira and Ochoa2020; Rossi et al., Reference Rossi, Galderisi, Rocca, Bertolino, Rucci, Gibertoni and Maj2017; Schrank, Amering, Hay, Weber, & Sibitz, Reference Schrank, Amering, Hay, Weber and Sibitz2014; Schwarzbold et al., Reference Schwarzbold, Kern, Novacek, McGovern, Catalano and Green2021; Sen, Nehra, & Grover, Reference Sen, Nehra and Grover2020; Shaheen & Amin, Reference Shaheen and Amin2016; Shin, Joo, & Kim, Reference Shin, Joo and Kim2016; Singh, Mattoo, & Grover, Reference Singh, Mattoo and Grover2016; Singla, Avasthi, & Grover, Reference Singla, Avasthi and Grover2020; Staring et al., Reference Staring, ter Huurne and van der Gaag2013; Styła & Świtaj, Reference Styła and Świtaj2024; Suman, Nehra, Sahoo, & Grover, Reference Suman, Nehra, Sahoo and Grover2023; Swanson et al., Reference Swanson, Schwannauer, Bird, Eliasson, Millar, Moritz and Griffiths2022; Świtaj, Grygiel, Anczewska, & Wciórka, Reference Świtaj, Grygiel, Anczewska and Wciórka2014; Tao et al., Reference Tao, Hui, Ho, Hui, Suen, Lee and Chen2022; Tu, Liu, & Huang, Reference Tu, Liu and Huang2023; Villagonzalo et al., Reference Villagonzalo, Arnold, Farhall, Rossell, Foley and Thomas2019; Vrbova, Prasko, Holubova, Slepecky, & Ociskova, Reference Vrbova, Prasko, Holubova, Slepecky and Ociskova2018; White, Haddock, Haarmans, & Varese, Reference White, Haddock, Haarmans and Varese2023; White, McCleery, Gumley, & Mulholland, Reference White, McCleery, Gumley and Mulholland2007; Yanos et al., Reference Yanos, West, Gonzales, Smith, Roe and Lysaker2012; Yanos, Roe, Markus, & Lysaker, Reference Yanos, Roe, Markus and Lysaker2008; Yıldız, Kiras, İncedere, & Abut, Reference Yıldız, Kiras, İncedere and Abut2019; Zhang et al., Reference Zhang, Wong, Yu, Ni, He and Bacon-Shone2019), with 81 of these studies used for the primary analysis with global negative symptoms (Table S26) and 23 studies included in the subdomain analyses (Table S26.1).
The 81 studies included in the main analysis consisted of 9766 participants. Most of the participants were male (62.9%, k = 79) and outpatients and were described as clinically stable in slightly less than half of the studies. The mean age of these participants was 38.4 years (k = 79), with six studies specifically recruiting young people (Chen et al., Reference Chen, Chang, Hui, Chan, Lee and Chen2016; Cloutier et al., Reference Cloutier, Lecomte, Diotte, Lamontagne, Abdel-Baki, Daneault and Perrine2023; DeLuca et al., Reference DeLuca, Akouri-Shan, Jay, Redman, Petti, Lucksted and Schiffman2021; Khalaf et al., Reference Khalaf, Fathy, Ebrahim and Samie2023; Murphy et al., Reference Murphy, Hogarth, Reynolds, Wood-Ross, Vanrooy and Bowie2023; Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019). The average duration of illness was 13.9 years (k = 63), with 11 studies evaluating primarily first/early episode psychosis patients (Berry & Greenwood, Reference Berry and Greenwood2018; Chen et al., Reference Chen, Chang, Hui, Chan, Lee and Chen2016; Chu et al., Reference Chu, Ng, Chu, Lui, Lau, Chan and Chang2023; Cloutier et al., Reference Cloutier, Lecomte, Diotte, Lamontagne, Abdel-Baki, Daneault and Perrine2023; DeLuca et al., Reference DeLuca, Akouri-Shan, Jay, Redman, Petti, Lucksted and Schiffman2021; DeTore et al., Reference DeTore, Balogun-Mwangi, Tepper, Cather, Russinova, Lanca and Mueser2022; MacDougall et al., Reference MacDougall, Vandermeer and Norman2015; Murphy et al., Reference Murphy, Hogarth, Reynolds, Wood-Ross, Vanrooy and Bowie2023; Ng et al., Reference Ng, Yu and Leung2024; Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019; Sen et al., Reference Sen, Nehra and Grover2020), seven studies with minimum illness or treatment duration inclusion cut-offs (Fekih-Romdhane et al., Reference Fekih-Romdhane, Hajje, Haddad, Hallit and Azar2023; Grover et al., Reference Grover, Avasthi, Singh, Dan, Neogi, Kaur and Behere2017, Reference Grover, Sahoo, Chakrabarti and Avasthi2018; Huang et al., Reference Huang, Liu and Yang2023; Konsztowicz et al., Reference Konsztowicz, Gelencser, Otis, Schmitz and Lepage2021; Singh et al., Reference Singh, Mattoo and Grover2016; Singla et al., Reference Singla, Avasthi and Grover2020), and one study recruiting both early phase and prolonged illness patients (Firmin et al., Reference Firmin, Lysaker, Luther, Yanos, Leonhardt, Breier and Vohs2019). The mean negative symptom severity score was 2.6 (k = 74), with two studies specifically selecting participants endorsing at least mild or greater severity of negative symptoms (Pos et al., Reference Pos, Franke, Smit, Wijnen, Staring, Van der Gaag and Schirmbeck2019; Staring et al., Reference Staring, ter Huurne and van der Gaag2013). Three studies excluded participants with acute or high levels of psychotic symptoms (González-Domínguez et al., Reference González-Domínguez, González-Sanguino and Muñoz2019; Horsselenberg et al., Reference Horsselenberg, Busschbach, Aleman and Pijnenborg2016; Koçak et al., Reference Koçak, Rıfat Şahin, Güz, Böke, Sarısoy and Karabekiroğlu2022), and seven studies classified their samples as being in some degree of symptomatic remission (Grover et al., Reference Grover, Avasthi, Singh, Dan, Neogi, Kaur and Behere2017; Karidi et al., Reference Karidi, Vassilopoulou, Savvidou, Vitoratou, Maillis, Rabavilas and Stefanis2015; Krzyzanowski et al., Reference Krzyzanowski, Agid, Goghari and Remington2021; Sen et al., Reference Sen, Nehra and Grover2020; Singh et al., Reference Singh, Mattoo and Grover2016; Singla et al., Reference Singla, Avasthi and Grover2020; Suman et al., Reference Suman, Nehra, Sahoo and Grover2023). Two studies also specifically selected participants with elevated internalized stigma scores as part of their inclusion criteria (González-Domínguez et al., Reference González-Domínguez, González-Sanguino and Muñoz2019; Yanos et al., Reference Yanos, West, Gonzales, Smith, Roe and Lysaker2012), with one additional study including a subsample of patients with at least moderate to high levels of internalized stigma (O’Connor et al., Reference O’Connor, Yanos and Firmin2018).
Meta-analysis
The meta-analysis revealed a significant correlation between overall negative symptoms and internalized stigma (k = 81, n = 9766), r = 0.17 (95% CI, 0.12–0.22, Z = 6.5, p < 0.0001; Figure S15). Heterogeneity across studies was high (I2 = 83.7%, τ2 = 0.04), with a non-significant trend for moderation by study quality (Table S28). Excluding the two outlier studies (Khalaf et al., Reference Khalaf, Fathy, Ebrahim and Samie2023; Shaheen & Amin, Reference Shaheen and Amin2016) (Figure S17) resulted in reduced heterogeneity (I2 = 64%, τ2 = 0.02), and significant moderating effects of negative symptom severity (β = 0.04, p = 0.004) and study quality (β = 0.05, p = 0.02 (Tables S31–S33). Based on funnel plots (Figures S16, S18) and Rosenthal’s fail-safe N = 7399, publication bias is unlikely. Significant effects emerged for diminished motivation (k = 23, n = 3255), r = 0.21 (95% CI, 0.15–0.28, Z = 6.1, p < 0.0001), but only trend-level for diminished expression (k = 16, n = 2256), r = 0.08 (95% CI, −0.01 – 0.16, Z = 1.8, p = 0.07) (Table S27). There was substantial heterogeneity for both diminished motivation (I2 = 68.9%, τ2 = 0.02) and diminished expression (I2 = 66.6%, τ2 = 0.02).
Perception of limited resources
Study characteristics
A total of 11 studies were included for the meta-analysis examining the relationship between negative symptoms and perception of limited resources (Bennett et al., Reference Bennett, Brown, Fang and Blanchard2023; Bortolon et al., Reference Bortolon, Larøi, Stephan, Capdevielle, Yazbek, Boulenger and Raffard2014; Bröcker et al., Reference Bröcker, Bayer, Stuke, Giemsa, Heinz, Bermpohl and Montag2017; Brüne, Drommelschmidt, Krüger-Özgürdal, & Juckel, Reference Brüne, Drommelschmidt, Krüger-Özgürdal and Juckel2019; Couture et al., Reference Couture, Blanchard and Bennett2011; Gesraha, Shalaby, & Harfush, Reference Gesraha, Shalaby and Harfush2023; Minor et al., Reference Minor, Marggraf, Davis, Mickens, Abel, Robbins and Lysaker2022; Moritz, Peters, Larøi, & Lincoln, Reference Moritz, Peters, Larøi and Lincoln2010; Østefjells et al., Reference Østefjells, Melle, Hagen, Romm, Sönmez, Andreassen and Røssberg2015; Popolo et al., Reference Popolo, Smith, Lysaker, Lestingi, Cavallo, Melchiorre and Dimaggio2017; Strauss et al., Reference Strauss, Morra, Sullivan and Gold2015), with 10 studies included in the primary analysis (Table S34) and three in the subdomain analysis (Table S34.1).
Across the 10 studies were 463 unique participants, 66% of whom were male, with a mean age of 38.8 and negative symptom severity score of 3.2 (k = 9). Duration of illness was only available for three studies (M = 21.4 years), with two studies specifically recruiting first/early episode psychosis patients (Brüne et al., Reference Brüne, Drommelschmidt, Krüger-Özgürdal and Juckel2019; Østefjells et al., Reference Østefjells, Melle, Hagen, Romm, Sönmez, Andreassen and Røssberg2015).
Meta-analysis
A non-significant correlation, with high levels of heterogeneity across studies, emerged between overall negative symptoms and perception of limited resources (k = 10, n = 463), r = 0.08 (95% CI, −0.13 – 0.27, Z = 0.7, p = 0.5, I2 = 75%, τ2 = 0.07; Figure S19). There was, however, a significant effect specifically for the diminished motivation subdomain analysis (k = 3, n = 266), r = 0.29 (95% CI, 0.18–0.40, Z = 4.8, p < 0.0001), with no heterogeneity observed (Table S35). The funnel plot is presented in Figure S20.
Discussion
The theorized cognitive model has become an increasingly popular psychological framework for conceptualizing negative symptoms in SSDs. The present study sought to critically appraise the extant empirical evidence related to the model by way of the largest and most comprehensive meta-analysis to date.
Defeatist performance beliefs
In line with an older meta-analysis by Campellone, Sanchez, and Kring (Reference Campellone, Sanchez and Kring2016), our updated findings revealed a small, significant relationship between overall negative symptoms and defeatist performance beliefs. Interestingly, our subdomain analyses showed no differences in effect sizes between diminished motivation and diminished expression, which were also both smaller in magnitude than the relationship with global negative symptoms. This is somewhat surprising given the conceptually stronger link between overgeneralized negative beliefs about performing goal-directed activity and diminished motivation in particular (Couture et al., Reference Couture, Blanchard and Bennett2011), which may call into question the specificity of this relationship. Nonetheless, our results provide further support for the importance of addressing self-defeating cognitions as treatment targets for negative symptoms.
Asocial beliefs
Asocial beliefs, or negative and/or apathetic attitudes towards social engagement and affiliation, have been hypothesized to represent key processes underlying reduced motivation for interpersonal relationships in SSDs (Grant & Beck, Reference Grant and Beck2010; Rector et al., Reference Rector, Beck and Stolar2005). While only a few studies were identified, our results offer promising preliminary support for this hypothesis, with findings of a small, significant correlation between asocial beliefs and overall negative symptom severity, as well as diminished motivation more specifically. With asocial beliefs emerging as potential mechanisms of change in CBT trials for negative symptoms (Granholm, Holden, & Worley, Reference Granholm, Holden and Worley2018), further research into this belief system and its relationship with the asociality features of negative symptoms may provide important insights for psychosocial intervention.
Low expectancies for success
In classic theories of motivation, self-efficacy beliefs (i.e. the extent to which an individual believes in their ability to succeed) are central to engagement in goal-directed activity (Bandura, Reference Bandura1977). To this end, low expectancies for success, or reduced self-efficacy, have been positioned within the cognitive model as a critical belief system underlying negative symptoms, particularly amotivation. According to our meta-analysis, there was indeed a significant relationship between expectancies for success and overall negative symptom severity, with an even more pronounced effect for diminished motivation. Further, this relationship was moderated by age and illness duration, which may highlight the importance of early intervention to address and protect self-efficacy beliefs.
Low expectancies for pleasure
Few studies were identified that examined low expectancies for pleasure. While this may be surprising given the large body of research dedicated to anhedonia in SSDs, our literature review revealed that most studies purporting to examine beliefs about pleasure utilized the TEPS anticipatory pleasure subscale (Gard, Gard, Kring, & John, Reference Gard, Gard, Kring and John2006), which does not assess the beliefs or expectancies one holds regarding their likelihood of experiencing pleasure in the future (i.e. pleasure anticipation) but rather appears to measure one’s experience of pleasure in-the-moment while anticipating future events (i.e. anticipatory pleasure). As a result, only five studies, with questionable psychometric properties, were included in our meta-analysis, which revealed a significant, but tenuous correlation between expectancies for pleasure and negative symptom severity. While promising, the lack of well-established measures for this cognition necessitates further research and development, along with further delineation between pleasure expectancies versus anticipatory pleasure.
Internalized stigma
Also known as self-stigmatization, internalized stigma refers to a process by which affected individuals self-identify with negative beliefs about their mental illness, incorporate stigmatizing views of the illness into their self-concept, and accept low expectations for themselves and their future as a result (Corrigan & Rao, Reference Corrigan and Rao2012). Accordingly, the cognitive model posits that certain maladaptive beliefs endorsed by patients with negative symptoms are driven by internalized stigma (Rector et al., Reference Rector, Beck and Stolar2005). The results of our meta-analysis align with a smaller one by Sarraf, Lepage, and Sauvé (Reference Sarraf, Lepage and Sauvé2022) and lend further support to this hypothesis in finding a significant relationship between negative symptom severity and internalized stigma, though the effect was relatively small and characterized by high levels of heterogeneity. There was, however, a more pronounced relationship with diminished motivation specifically, suggesting a role for demoralization in patients’ withdrawal from goal-directed activity and aligning with the ‘why try’ model of self-stigma (Corrigan, Bink, Schmidt, Jones, & Rüsch, Reference Corrigan, Bink, Schmidt, Jones and Rüsch2016).
Perception of limited resources
According to Rector et al. (Reference Rector, Beck and Stolar2005), patients endorse statements such as ‘It’s too much’ and ‘I can’t handle it’ when asked to perform different tasks, and it is this subjective account of limited internal resources that contributes to a pattern of disengagement, passivity, and avoidance. Withdrawal from effortful activities and goal-directed behavior is therefore hypothesized to be partly due to beliefs characterized by underestimation of abilities and available resources, exaggeration of limitations, and overestimation of personal costs of expending energy. In the present meta-analysis, we identified very few studies that utilized measures that mapped onto this construct, despite including measures of both perception of cognitive resources and physical and psychological resources. Further, the effect was non-significant, though the inclusion of apparently disparate constructs potentially obfuscated the results, with the non-significant effect seemingly driven by the MCQ (i.e. cognitive resources). Thus, the available evidence as it currently stands does not support the relationship between perception of limited resources and overall negative symptoms; however, there remains a need to better define and evaluate this belief system, especially as it is often observed in patients clinically.
Implications and future directions
It is important to acknowledge that the magnitude of the effect sizes reported above was mostly modest, though this is not entirely surprising given the extensive methodological and sample variability across studies, as well as the multifaceted nature of negative symptoms themselves. Indeed, negative symptoms are inherently complex clinical phenomena that are likely best explained by a multitude of interacting mechanisms, at both the individual and environmental levels (Strauss, Reference Strauss2021), but with potentially only small additive effects when examined on their own. Furthermore, our findings reflect population effect sizes amongst a heterogeneous sample, and therefore, it is not necessarily that dysfunctional beliefs exert only small effects for all patients with negative symptoms, nor is it the case that the five significant beliefs identified in this meta-analysis each contribute equally for every patient with negative symptoms. Instead, a more clinically meaningful approach going forward would be to focus on identifying the specific subgroups of patients for whom dysfunctional beliefs do relate to their negative symptoms, and then further subtype them based on their underlying belief profile. This would then allow for a more personalized approach to treatment planning and delivery, and may also pave the way for the development of a modularized treatment for negative symptoms, similar to what has been done for obsessive-compulsive disorder (Steketee et al., Reference Steketee, Siev, Fama, Keshaviah, Chosak and Wilhelm2011; Wilhelm et al., Reference Wilhelm, Steketee, Fama, Buhlmann, Teachman and Golan2009), whereby separate self-contained therapeutic modules are created for each dysfunctional belief system and then selectively matched with the patient based on their individual presentation and case conceptualization. This could serve to optimize the effectiveness and efficiency of future CBT interventions for negative symptoms.
Limitations
The current review only evaluated the cross-sectional associations between negative symptoms and dysfunctional beliefs. Accordingly, it is not possible to determine causality or speak to the direction of the reported relationships. As few studies to date have utilized longitudinal designs, the extent to which different dysfunctional beliefs precede and predict the emergence and/or worsening of negative symptoms remains unknown. We were also unable to systematically ascertain whether negative symptoms reported across studies were primary or secondary to other factors, such as depressive or positive symptoms, which could also contribute to dysfunctional beliefs. Some of the meta-analyses also exhibited high levels of heterogeneity, potentially attributable to the wide range of included instruments for each belief system with possibly different operationalizations. Finally, while each of the dysfunctional beliefs examined in this meta-analysis was proposed as separate processes, it is highly possible that overlap exists among them. Unfortunately, however, factor analyses are limited in this regard, with the only potentially relevant analysis in this population, to our knowledge, demonstrating a relatively clear separation between defeatist performance beliefs and asocial beliefs (Pillny et al., Reference Pillny, Krkovic and Lincoln2018). Across studies included in this meta-analysis (k = 176), there were comparatively few that simultaneously evaluated more than one belief system (k = 26), and even fewer with available intercorrelation data (k = 10), with results that ranged widely from rs = |0.009–0.71|.| It will therefore be important for future research to conduct more fulsome concurrent examinations in order to identify the factor structure of these dysfunctional beliefs and their interrelationships.
Summary
This comprehensive meta-analysis provides strong support and validation for the cognitive model of negative symptoms. Significant cross-sectional correlations with overall negative symptoms emerged for defeatist performance beliefs, asocial beliefs, low expectancies for success, low expectancies for pleasure, and internalized stigma, but not perception of limited resources. Subdomain analyses further suggested that many of these relationships appeared to be linked to diminished motivation specifically, which is particularly noteworthy from a clinical perspective given its association with poor outcomes. Longitudinal investigations of these relationships, along with improved measurement strategies, are important next steps to further delineate how dysfunctional beliefs contribute to negative symptoms in SSDs. The identification of these belief systems as clinically meaningful treatment targets may afford opportunities for the development of more optimized and precision-based CBT interventions for this critically unmet therapeutic need.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291724003325.
Author contribution
Konstantine Zakzanis and George Foussias have shared senior authorship.
Financial support
This work was supported by a Canada Graduate Scholarships Doctoral Award (CGS-D) from the Canadian Institutes of Health Research (CIHR) to S. Saperia. The funding source had no contribution to the design or conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Competing interest
Dr. Best receives consulting and speaker fees from Boehringer Ingelheim. The other authors report no biomedical financial interests or potential conflicts of interest.
Open access
