Perceived Stress Influences Anhedonia and Social Functioning in a Community Sample Enriched for Psychosis-Risk

Abstract

Existing animal and human research support the causal role of stress in the emergence of anhedonia, and in turn, the influence of anhedonia in social functioning. However, this model has not been tested in relation to psychosis-risk; this literature gap is notable given that both anhedonia and declining social functioning represent key markers of risk of developing a psychotic disorder such as schizophrenia. The current research tested the evidence for this model using structural equation modeling in 240 individuals selected based on a range of psychosis-risk symptomatology from the general community. Results supported this model in comparison with alternative models, and additionally emphasized the direct role of perceived stress in social functioning outcomes. Findings suggest the clinical relevance of targeting early perceptions of stress in individuals meeting psychosis-risk self-report criteria in an effort to prevent subsequent anhedonia and declines in social functioning.

Introduction

Stress, anhedonia, and impaired social functioning are key markers of risk for the development of psychotic disorders such as schizophrenia (Addington et al., 2008; Cornblatt et al., 2011; Piskulic et al., 2012; Walker et al., 2010; Walker et al., 2013; Zhang et al., 2019). Existing animal and human research suggest a model whereby stress exposure plays a causal role in the emergence of anhedonia and the subsequent influence of anhedonia in manifesting poorer social functioning (Dodell-Feder et al., 2014; Lincoln et al., 2019; Pizzagalli, 2014; Rassovsky et al., 2011; Scheggi et al., 2018; Treadway et al., 2013). Despite these converging lines of research, this model has yet to be tested in relation to psychosis-risk. Identifying whether this model exists in individuals at risk for psychotic illness has the potential to identify a pathway(s) to declines in social functioning - a central risk factor for transition to psychosis (Addington et al., 2008; Carrion et al., 2019; Carrion et al., 2013; Cornblatt et al., 2011).

Stress has long been considered a catalyst in the onset of psychotic symptoms (Corcoran et al., 2003; Mittal and Walker, 2019) and is included in most etiological models as a causal factor, perhaps most seminally in the diathesis-stress model of schizophrenia (Walker and Diforio, 1997). This model postulates that, along with biological vulnerability, stress introduces a cascade of events that give rise to the initial risk symptoms of psychosis and eventually propel an individual into the first psychotic episode. Within this framework, research emphasizes the psychosis-risk state as a means of further elucidating the impact of stress exposure on symptomatology. The psychosis-risk period can be conceptualized to include adolescents and young adults with schizotypal personality disorder and/or the attenuated psychosis syndrome (often labeled clinical high-risk or CHR) (McGlashan et al., 2001; Woods et al 2014). Recent efforts to comprehensively identify risk symptoms also include individuals with psychotic-like experiences (PLE, i.e. subclinical psychotic experiences) from the community (van Os and Linscott, 2012). These experiences occur in roughly 7% of the general population, and are associated with elevated risk for transitioning to a psychotic disorder (van Os and Reininghaus, 2016). The value of emphasizing psychosis risk on a continuum, and including individuals from the general population, is an ability to capture a wider and continuous range of symptoms. Studies show that psychosis-risk youth experience more stressful life events such as bullying and physical illness (Mayo et al., 2017; Ristanovic et al., 2020; Tessner et al., 2011; Vargas et al., 2019) and related increases in perceived stress and/or threat (Fusar-Poli et al., 2017; Millman et al., 2018; Palmier-Claus et al., 2011; Pruessner et al., 2011; Reininghaus et al., 2016) and stress sensitization (Chun et al., 2017; Cristobal-Narvaez et al., 2016; Howes and Murray, 2014; Mizrahi, 2016; Mizrahi et al., 2012; Trotman et al., 2014; van der Steen et al., 2017; Walker et al., 2008).

Accumulating evidence indicates that anhedonia and social functioning are likely consequences of this heightened stress sensitivity. Anhedonia findings within psychosis-risk samples are mixed, with some studies showing solely impaired anticipation of pleasure/rewards (Cooper et al., 2018; Pelletier-Baldelli et al., 2018) and others suggesting deficits in both anticipation and consumption of rewards (Gerritsen et al., 2019; Schlosser et al., 2014; Strauss et al., 2018). One study evaluated the link between stress and psychosis-risk symptoms (assessed quarterly over 4 years), and shows that impaired tolerance to stressors predicted increases in total negative symptoms (including anhedonia) (Devylder et al., 2013). A more recent study reports that perceived lack of control is causal to both anticipatory and consummatory anhedonia in psychosis-risk youth (Gerritsen et al., 2019). Both impaired tolerance to stress and perceived lack of control can be subsumed under the broad construct of perceived stress, which is defined as feelings of unpredictability, uncontrollability, and being overwhelmed by one’s life (Cohen et al., 1983). Thus, perceived stress may play a particularly strong role in the manifestation of anhedonia.

Longstanding animal research indicates that stressors (often represented by shock, deprivation of food/water, or maternal separation) are causal to anhedonic behaviors; these changes include the diminished seeking out of pleasurable experiences (driven by deficits in wanting a reward - anticipatory anhedonia) (Enman et al., 2015; Riga et al., 2015) and diminished reactivity to rewards (driven by deficits in liking a reward in-the-moment - consummatory anhedonia) (Scheggi et al., 2018; Scheggi et al., 2011). Transdiagnostic human investigations support this animal literature with studies showing that stress (e.g. introduction of threat-of-shock, social rejection, cold presser, perceived stress) alters neural anticipation and response to reward (Berghorst et al., 2013; Bogdan and Pizzagalli, 2006; Kumar et al., 2014; Lincoln et al., 2019; Porcelli et al., 2012; Treadway et al., 2013). Importantly, there is some indication that stress exposure has a stronger impact on reward processing in individuals reporting higher perceived stress (Bogdan and Pizzagalli, 2006; Dickerson and Kemeny, 2004; Pizzagalli, 2014; Treadway et al., 2013). This finding is in accordance with a recent study that showed aspects of perceived stress impacting anticipatory and consummatory anhedonia (Gerritsen et al., 2019).

These existing human and animal models of the stress-anhedonia relationship show disrupted effects on the processing of rewards, which in turn, are likely to impact social functioning. More specifically, existing research suggests a link between stress and disrupted social functioning (Sandi and Haller, 2015), with one study indicating a unique role of heightened perceived stress in altering social interactions (Dissing et al., 2019) The social deafferentation hypothesis of schizophrenia supports the link between anhedonia and social functioning by proposing that certain factors, such as social anhedonia, create a deficit of incoming social information and social withdrawal, ultimately contributing to psychotic symptoms (Hoffman, 2007). Indeed, higher social and non-social anhedonia is related to impaired social functioning across the developmental trajectory of psychosis (Blanchard et al., 1998; Cressman et al., 2015; Dodell-Feder et al., 2020; Fervaha et al., 2014) and impairments in social functioning exist across the psychosis continuum (Addington et al., 2008; Carrion et al., 2013; Dodell-Feder et al., 2020; Gayer-Anderson and Morgan, 2013; Robustelli et al., 2017).

In sum, although prior work suggests causal links between stress and anhedonia, and anhedonia and social functioning, there are no studies across the psychosis continuum that investigate all three domains using the SEM framework. Further, stress, anhedonia, and social functioning are complex, multifaceted constructs, and various studies use differing definitions. For the purposes of this study, perceived stress, non-social anticipatory/consummatory anhedonia, and broad social functioning (to include prosocial behavior, interpersonal communication, and social engagement) were investigated. These constructs were selected based on the unique role of perceived stress in non-social anhedonia (Gerritsen et al., 2019; Pizzagalli, 2014; Treadway et al., 2013), and prior work reporting impairment in selected social functioning domains within adolescents at risk for psychosis (Addington et al., 2008). Unique from existing research, the current study uses structural equation modeling (SEM) to test the following path within a community sample enriched for psychosis-risk: perceived stress → anhedonia → social functioning. Based on the aforementioned research, it was hypothesized that heightened perceived stress would relate to anhedonic experience, which would link with greater impairments in social functioning, consistent with prior models positing negative symptoms as preceding functional impairments (Devoe et al., 2020; Rassovsky et al., 2011).

Material and Methods

Participants

Participants included adolescent and young adult individuals selected from the Multisite Assessment of Psychosis-Risk (MAP) study across 3 sites: Northwestern University, University of Maryland Baltimore Country, and Temple University. The MAP study is a larger investigation designed to improve identification and evaluation of psychosis-risk within the general community. Study recruitment entailed a broad outreach to the local community and universities through flyers and online sources (e.g., social media, SONA, craig’s list). Participant selection had two steps: first, participants (n = 3,460) filled out an online survey that included a battery of questionnaires including two self-report scales designed to screen for psychosis-risk. Based on their score on these two scales, participants were identified as either (1) scoring above the pre-determined psychosis-risk thresholds (Questionnaire High-Risk - QHR) or (2) scoring below the pre-determined psychosis-risk thresholds (Questionnaire Low-Risk - QLR). Next, all QHR and randomly selected QLR groups were invited for an in-person interview. The eligible and invited sample included 651 QHR and 656 QLR participants. Of the 651 QHR invited, 243 completed an in-person interview at the time of this analysis. The QHR sample was the focus of the current investigation, as these individuals reflect a sample enriched for psychosis-risk using questionnaire assessment

Inclusion criteria required proficiency in the English language, being within the ages of 16–30, and having normal or corrected vision (necessary for other portions of the larger study). Participants were excluded if they did not meet inclusion criteria or reported inability or unwillingness to give consent and attend an in-person visit. Given the goal of the larger project to ascertain and assess psychosis-risk from the broader population without imposing additional constraints, there were no diagnostic exclusionary criteria. For the purposes of the current study, invited participants who were diagnosed with a DSM-5 psychotic illness at the interview were excluded (n = 3), resulting in a final sample of 240 QHR adolescents and young adults analyzed in the present study. Each site’s Institutional Review Board approved the study protocol and all procedures were carried out according with provisions outlined in the Declaration of Helsinki.

Measures

Stage I. Online Screening

Participants completed a battery of questionnaires online using Qualtrics as part of the overall study. The following questionnaires were used for the present study.

Psychosis-Risk Screening Questionnaires.

Two separate psychosis-risk scales were used to ascertain QHR status, including the Prodromal Questionnaire (PQ) (Loewy et al., 2005; Loewy et al., 2007) and the PRIME Screen (Miller et al., 2004). Established thresholds were used in the present study and include having 8 or more attenuated positive symptoms identified as distressing on the PQ and/or having two or more items endorsed on the PRIME-Screen with responses of “somewhat” or “definitely agree” (Loewy et al., 2005; Miller et al., 2004).

Stress.

Participants also completed the 14-item Perceived Stress Scale (PSS) (Cohen et al., 1983), which evaluates the extent to which daily situations are perceived as stressful and uncontrollable in the past month, with higher scores indicating higher perceived stress. The PSS scores are able to discriminate between psychosis populations and controls, and are associated with psychosis-risk presentations (Horan et al., 2007; Mondelli et al., 2010; Palmier-Claus et al., 2011).

Anhedonia.

The Temporal Experience of Pleasure Scale (TEPS) (Gard et al., 2006) was selected to measure anhedonia given its prominent use in the field, allowing us to interpret anhedonia results in the context of a larger body of literature. The TEPS is comprised of anticipatory (TEPS-ANT) and consummatory (TEPS-CON) subscales, with higher scores indicating intact experience. The TEPS is extensively used in the psychosis literature, (Da Silva et al., 2017; Mote et al., 2014; Pelletier-Baldelli et al., 2018; Schlosser et al., 2014; Strauss et al., 2011), and is associated with increased scores on the PQ (Cooper et al., 2018).

Social Functioning.

Participants also completed the Social Functioning Scale (SFS) (Birchwood et al., 1990), which has been successfully implemented with psychosis-risk samples (Addington et al., 2008). The SFS was selected based on the range of domains evaluated in the measure, as this would allow us to create a latent social functioning construct, ultimately modeling error more effectively. Higher scores indicate higher functioning. The overall study assessed 5 subdomains of the SFS; given that recreation and employment/occupation are more often assessed as distinct from social functioning, albeit related, only the prosocial, engagement, and interpersonal communication subscales were investigated in this study.

Stage II. In-Person Assessment

As part of a larger battery, the Structured Clinical Interview for DSM-5 was administered to assess for co-occurring clinical diagnoses (First et al., 2015). Attenuated positive symptoms were assessed using the Structured Interview for Psychosis-Risk Syndromes (SIPS) (McGlashan et al., 2001) to determine whether participants exhibited a clinical high-risk (CHR) presentation. The SIPS was administered by trained graduate students and clinicians across the 3 sites. All SIPS administrators were trained by a SIPS certified trainer, which required an extensive training process including workshop attendance, rating of videos, and achieving a high level of reliability over at least two videos before being cleared for SIPS interviewing (ICC > 0.80 on positive symptom ratings). Additionally, participants’ clinical presentations were reviewed during weekly cross-site meetings to provide supervision and agreement in ratings across sites led by the SIPS certified trainer. The SIPS determines CHR status through three ways: attenuated positive symptoms (APS), brief intermittent psychotic symptoms (BIPS), and genetic risk and functional deterioration (GRD). In accordance with recent CHR conceptualization (Woods et al., 2014), both a progressive presentation and a broader CHR persistent presentation were included. Roughly 30% of the QHR sample met CHR criteria (74 APS, 1 GRD persistent, 0 BIPS). Within the CHR APS subgroup, 42 met criteria for CHR persistence and 33 for CHR progressive.

Statistical Analysis

An SEM analysis was conducted using the lavaan package (Rosseel, 2012) in R (v. 3.6.1) (R Development Core Team, 2019) to identify latent constructs using a confirmatory factor analysis and simultaneously test statistical relationships among the latent factors and manifest (observed) variables in the model. Model estimation was conducted using robust full information maximum likelihood (FIML), which accounts for missing data (see supplementary materials) by weighting cases with complete data more heavily, and addresses issues with non-normality that are often present in clinical and psychosis-risk evaluations (i.e. symptoms are often positively skewed), as was the case in the QHR sample. Fit indices align with current standards in the field (Bentler, 1990; Hu and Bentler, 1999), and included robust root mean square error of approximation (rmsea), robust comparative fit index and tucker-lewis index (cfi and tfi), and standardized root mean square residual (srmr). Importantly, prior work suggests not reporting rmsea for models with small df (as in the current model), given that rmsea may erroneously indicate a poor fitting model (Kenny et al., 2014). In light of this fact, rmsea was not emphasized when interpreting results. Modification indices were inspected after evaluating model fit indices.

Finally, due to the cross-sectional nature of the present study, secondary analyses were conducted to determine whether several alternative, theoretically-based models better represented the data: (A) stress→social functioning→anhedonia, (B) anhedonia→social functioning→stress, (C) social functioning→anhedonia→stress. The proposed model was then tested against these alternative structures individually using the likelihood ratio test function in lavaan (Rosseel, 2012), which identifies whether a significant difference in Chi-Square values exist between models.

Results

Sample Characteristics

Inspection of QHR demographics revealed the sample to be majority White (n = 125) followed by African American/Black (n = 42), and Asian (n = 59), with 23 participants identifying as Hispanic. Average maternal education (as a proxy for socioeconomic status (Braveman et al., 2001)) was at the level of a Bachelors degree. The QHR sample was also predominately female (n = 185), had an average age of 20.14 years (SD = 1.83), and displayed heterogeneous psychosis-risk and diagnostic symptomatology, with 30% of the sample seeking current mental health treatment (n = 74). Specifically, 8% met DSM-5 criteria for a stress (e.g. post-traumatic stress disorder) or adjustment disorder (n = 20), 57% for anxiety (n = 136), 38% for a mood disorder (n = 91), 8% for an eating disorder (n =19), and 22% for a substance use disorder (n = 53). Study specific clinical characteristics are presented in Table 1, and sample information by site is available online in supplemental Table S1. As anticipated, scores on primary outcome measures indicated, in general, somewhat less impairment than studies solely investigating CHR individuals (supplemental Table S1).

Table 1.

Sample Clinical Characteristics

Measure	N	Mean	Standard Deviation	Range	Skew	Kurtosis
SFS: Engagement	236	9.82	2.29	2–15	−0.39	0.04
SFS: Prosocial Behavior	239	22.52	9.25	3–50	0.13	−0.17
SFS: Interpersonal Communication	232	7.59	1.56	2–9	−1.27	1.27
TEPS Anticipatory	240	4.26	0.7	2.3–6	−0.45	0
TEPS Consummatory	239	4.15	0.7	2.12–5.88	−0.29	−0.08
PSS Total	238	32.41	7.5	10–56	−0.04	0.71
SIPS Positive Total	240	6.22	3.65	0–17	0.68	0.42

Note: SFS (Social Functioning Scale), TEPS (Temporal Experience of Pleasure Scale), PSS (Perceived Stress Scale), and SIPS (Structured Interview for Psychosis-Risk Syndromes). Aside from the TEPS and SFS, higher summary scores indicate greater impairment. Discrepancies in total cases for all variables (i.e. missing data) were handled using the full information maximum likelihood approach in all models.

Model Evaluation

Correlations among all observed variables are presented in Table 2. Although significant correlations are evident in Table 2, multicollinearity issues were not observed in the model, and no problems with model identification were noted.

Table 2.

Correlations Among Observed Variables for CFA and SEM analysis

	Social Engagement	Interpersonal Communication	Prosocial Behavior	Stress	TEPS-ANT	TEPS-CON
Social Engagement	1	–	–	–	–	–
Interpersonal Communication	0.49	1	–	–	–	–
Prosocial Behavior	0.46	0.38	1	–	–	–
Stress	−0.34	−0.48	−0.39	1	–	–
TEPS-ANT	0.29	0.18	0.27	−0.14	1	–
TEPS-CON	0.23	0.20	0.26	−0.24	0.60	1

All correlations are significant at p < 0.05. Stress is measured by the Perceived Stress Scale total score. Prosocial Behavior, Interpersonal Communication, and Social Engagement are subscales derived from the Social Functioning Scale. TEPS-ANT and TEPS-CON are the anticipatory and consummatory subscales anhedonia from the Temporal Experience of Pleasure Scale. Higher Stress scores indicate greater perceived stress; higher TEPS scores indicate intact hedonic experience; higher SFS scores indicate better functioning.

CFA results showed that both TEPS indicators loaded well onto the anhedonia factor, with standardized coefficients of 0.74 and 0.78. All three SFS indicators also exhibited high loadings onto the SFS construct (0.63 – 0.76). Evaluation of unstandardized coefficients revealed that higher perceived stress significantly predicted greater anhedonia (B = −0.02, SE = 0.005, z = −4.06 p < 0.001) and greater anhedonia significantly predicted poorer social functioning (B = 1.72, SE = 0.37, z = 4.67, p < 0.001). However, robust model fit indices revealed the overall model structure to be of limited fit for the data (χ²(8) = 67.65, p < 0.001; srmr = 0.11; rmsea = 0.18 [90% Confidence Interval (CI) = 0.14,0.22]; cfi = 0.82; tli = 0.66).

Post-hoc inspection of modification indices revealed that a direct effect of perceived stress on social functioning would significantly enhance model fit, which is theoretically supported by animal studies showing causal influence of stress on social behavior (Berton et al., 2006; Scheggi et al., 2018). As such, a second model that included this direct effect was analyzed. The resulting final model performed significantly better than the original model (χ²(1)= 40.40, p < 0.001). Results showed a maintained significant effect of perceived stress on anhedonia (B = −0.02, SE = 0.005, z = −3.54, p < 0.001) and anhedonia on social functioning (B = 0.97, SE = 0.25, z = 3.9, p < 0.001), but also a significant effect of perceived stress directly on social functioning (B = −0.11, SE = 0.02, z = −7.23, p < 0.001) such that higher stress predicted poorer social functioning. This addition of the direct effect of stress on social functioning resulted in a partial mediation. In order to examine the significance of the indirect and total effect, the unstandardized effects were computed for each of the 10,000 bootstrapped samples, and the bias-corrected 95% CIs were computed. The bootstrapped unstandardized indirect effect was −0.02 (95% CI = −0.04, −0.005), and the bootstrapped unstandardized total effect was −0.13 (95% CI = −0.16, −0.10). Thus, both the indirect and total effects of perceived stress on social functioning were statistically significant. Lastly, a post-hoc model was analyzed to examine the influence of potential confounds including sex, site location, and age. Parameter estimates did not significantly change when accounting for these variables, and there was no evidence of a significant association involving sex, site, or age with stress, anhedonia, or social functioning (see supplemental material).

Analysis of residuals did not suggest model fit issues (residual covariance matrix and standardized residuals for this model are presented in supplemental Tables S2 and S3 available online). Fit indices showed this model to be an overall good fit for the data as indicated by several robust estimators of fit (χ²(7)= 20.2, p = 0.005; srmr = 0.03; rmsea = 0.09 [90% CI = 0.04, 0.14]; cfi = 0.96; tli = 0.92) (Figure 1).

Figure 1.

Final model showing factor loadings or standardized coefficients. Stress significantly predicts anhedonia, which in turn significantly predicts social functioning (SFS). Stress also has a significant direct influence on social functioning. Latent variables are depicted as circles and manifest variables as squares. Intercepts and residuals are not presented for viewing clarity. TEPS_ANT and TEPS_CON - Temporal Experience of Pleasure Scale Anticipatory and Consummatory; Engage - Social Engagement; Int-Comm - Interpersonal Communication; Prosocial - Prosocial Behavior.

Despite the strong theoretical and evidence-based foundation for the directionality of the final model, alternative models were evaluated due to the cross-sectional nature of the data. Significance testing of the difference in chi square between the final model and alternative models were as follows: (A) stress→social functioning→anhedonia (χ² (1) = 0.04, p = 0.85), (B) anhedonia→social functioning→stress (χ² (1) = −0.72, p = 1.00), (C) social functioning→anhedonia→stress (χ² (1) = 39.71, p < 0.001). Models A and B were not significantly better or worse than the final model; in contrast, Model C performed significantly worse. Therefore, none of the alternative models resulted in a significantly better fit for the data and have relatively less causal support in the literature, thus providing incremental evidence in favor of the final model presented here.

Discussion

This study reports the first modeling of stress effects on anhedonia and social functioning in a community-derived sample at heightened risk for psychosis. An SEM approach was used to test our theoretical model, allowing us to model error more effectively through establishing latent constructs. The final model indicates statistical directionality stemming from perceived stress to anhedonia and anhedonia to social functioning, along with a direct effect of stress on social functioning. While considering the limitations related to cross-sectional investigations, the current study extends existing stress-anhedonia models by adding a downstream social functioning factor in a community-derived sample exhibiting psychosis-risk symptomatology.

Perceived Stress and Anhedonia

The direct effect of perceived stress on anhedonia is consistent with existing animal models (Scheggi et al., 2018) and prior work in CHR youth reporting that both anticipatory and consummatory anhedonia are influenced by perceived uncontrollability (Gerritsen et al., 2019). Further, the stress-anhedonia relationship is widely emphasized in depression research (Pizzagalli, 2014), and these associations are identified in the current study for the first time in a community-derived sample with varying degrees of psychosis-risk symptomatology. The implications of this stress-anhedonia finding in the current study are two-fold: first, it may be that the mechanisms underlying these relationships are evident across the early psychosis-risk continuum and second (and relatedly), that the stress-anhedonia relationship includes perceived stress.

As one of the longstanding theories on schizophrenia etiology, the diathesis-stress model proposes that the hypothalamic-pituitary-adrenal (HPA) axis (as the biological stress response system) plays a mediating role in the relationship between underlying biological vulnerability and psychosis onset (Walker et al., 2008). A disrupted HPA response and resulting heightened stress sensitivity may be particularly evident in the psychosis-risk stage (Holtzman et al., 2013; Trotman et al., 2014). This altered HPA functioning is thought to disrupt neural reward functioning and lead to anhedonia (Stanton et al., 2019). Consistent with the proposal of others (Gomes and Grace, 2017), the current findings lend support to the idea that these important biology-stress interactions are transphasic in regard to early psychosis-risk presentations. In fact, it may be that the relationships found here exist regardless of psychiatric presentation, as studies report that acute stress increases anhedonia in non-psychiatric participants (Gerritsen et al., 2019), at least in part due to diminishing positive response to reward (Kumar et al., 2014; Lincoln et al., 2019; Treadway et al., 2013). Further, although perceived stress is investigated elsewhere in regard to altered reward-related processes (Pizzagalli et al., 2007; Treadway et al., 2013), and is known to be elevated in individuals with anhedonia (Horan et al., 2007), this is the first study to show the perceived stress - anhedonia link in a community-derived psychosis-risk sample. These findings point to the relevance of the stress-anhedonia interaction at all levels of psychosis-risk and suggest that clinicians (community-based and otherwise) should query into perception of stress and anhedonia experiences if either one is elevated, regardless of diagnostic presentation.

Anhedonia and Social Functioning

Anhedonia was significantly related to social functioning in the model, supporting similar findings in schizophrenia (Rassovsky et al., 2011) and early psychosis-risk samples (Dodell-Feder et al., 2020). This extension to the existing stress-anhedonia model is important given that 50% of psychosis-risk youth experience declines in social functioning (Carrion et al., 2013). The specific contributions to disrupted social functioning in the early stages of psychosis-risk remains unknown, and the current investigation suggests that perceived stress and anhedonia play a role. Thus, it may be that reduced capacity to experience pleasure leads to diminished seeking out of interpersonal situations. Altered social functioning is a critical indicator for prognostic outcome (Carrion et al., 2013; Cornblatt et al., 2011) and understanding the etiology of these changes in social behavior is essential to intervening with individuals at risk for psychosis.

Perceived Stress and Social Functioning

There is surprisingly little research focused on the effects of perceived stress on social functioning in psychosis. However, the broader literature does support a direct relationship between these two domains; for example, animal models clearly show life stressors impairing social behavior (Berton et al., 2006; Scheggi et al., 2018), and human psychosis-risk research finds that childhood maltreatment and poor home environment are associated with impaired social functioning later in life (Schlosser et al., 2010; Stain et al., 2014). Baseline cortisol levels - an endocrine indicator of stress - are also linked with poorer social functioning in non-psychiatric adult samples (Corcoran et al., 2003). In light of these findings, it is not surprising that perceived stress directly impacted social functioning in the model, particularly given the intuitive idea that perceived uncontrollability would hinder the ability to successfully interact with others. It is notable, however, that the addition of this path significantly improved model fit for the data, suggesting that the direct effect of perceived stress is a significant precursor to altered social behavior, and may be the strongest contributor among our variables (based on parameter estimates). In total, perceived stress significantly impacted social functioning directly, as well as indirectly by way of anhedonia. This finding aligns well with the conceptualization of stress as a cascade in the development of psychosis (Corcoran et al., 2003; Klippel et al., 2018), whereby stress functions as a catalyst that sets in motion a widespread systemic change to precipitate or exacerbate the manifestation of psychosis symptomatology. Subsequent longitudinal studies should investigate the role of both perceived stress and anhedonia in altered social functioning within psychosis-risk samples to identify the unique contribution of these domains.

Limitations and Future Directions

The current investigation probed stress, anhedonia, and social functioning across the early psychosis-risk continuum in individuals from the general community. As such, results are generalizable to this very broad sample, and may be different in help-seeking populations. Further, although SEM is a statistically directional modeling technique, the cross-sectional data does not allow for a comprehensive investigation into causality, despite the current testing alternative models. Future studies would benefit from including a longitudinal component. Other work indicates that stress impacts reward response differently based on biological sex (Lighthall et al., 2012), which suggests that future investigations with more equal representation of males/female should prioritize probing sex as a moderating factor. In the present investigation, anhedonia was probed using the TEPS, primary due to the desire to compare findings to existing psychosis-risk research that uses this scale (Cooper et al., 2018; Pelletier-Baldelli et al., 2018; Schlosser et al., 2014) and the fact that it aligns with the National Institutes of Health Research Domain Criteria initiative. However, the TEPS may reflect appraisal of hedonic experience as opposed to actual anticipation or consumption of rewards (Strauss and Gold, 2012). As such, future studies may benefit from including additional measures of anhedonia, such as in-the-moment laboratory experiments. Stress, anhedonia, and social functioning are also vastly complex constructs, and the simplified model presented here does not seek to address the many facets and related factors that exist. However, future work should expound the present model by adding additional variables and relationships, including other aspects of these constructs such as stress exposure, social anhedonia, and social motivation, along with the bidirectional relationship among stress and reward sensitivity, (Vidal-Ribas et al., 2019) influence of additional clinical symptoms (Gerritsen et al., 2019), and the role of intellectual functioning (Allott et al., 2015), social relationships, (Cohen and Wills, 1985) and self-esteem (Jongeneel et al., 2018).

Conclusions

The findings suggest a model such that perceived stress influences anhedonia, which in turn, predicts social functioning; perceived stress also directly impacts social functioning. Although cross-sectional, these data support existing animal and human research that report a causal role of stress in anhedonia, and extends these findings to an important, and clinically-relevant, outcome - social functioning. These results also indicate that stress, anhedonia, and impaired social functioning relationships are evident earlier in a community-derived sample that spans different phases of the psychosis-risk continuum. Although only 30% of the sample investigated met CHR criteria, all participants are at elevated risk for psychosis given the endorsement of psychotic-like experience (van Os and Reininghaus, 2016). Furthermore, the majority of the sample met diagnostic criteria for a DSM-5 diagnosis (particularly mood and anxiety). Thus, clinical interventions with this general community/non-help seeking population may benefit from an acknowledgement of the interconnectedness among stress, anhedonia, and social functioning, as there is potential to impact important functional and diagnostic outcomes aside from just transition to psychotic illness.

Highlights.

• Perceived stress impacts social functioning, both directly and by way of anhedonia

• These paths are evident in a sample enriched for psychosis-risk

• Perceived stress represents a target for early psychosis-risk intervention