The influence of illness-related variables, personal resources and context-related factors on real-life functioning of people with schizophrenia

Despite significant advances in pharmacological and psychological treatments, schizophrenia still ranks among the first ten leading causes of disability worldwide. It has a direct negative influence on the real life of about 26 millions of people and an indirect negative impact on more than twice this number when considering patients' relatives and caregivers 1. Major areas of everyday life are impaired, including independent living, productive activities and social relationships 2.

The lessening of symptoms and the reduction of relapse rate contributes to improve real-life functioning, but is not sufficient to attain functional recovery 3-6. Enhanced understanding of factors that hinder functioning in schizophrenia is vital for treatments to translate into more positive outcomes 7. Studies carried out so far have led to partial and sometimes discrepant findings. Furthermore, they usually focused on neurocognitive deficits, negative symptoms and depression, and often failed to simultaneously consider several relevant variables 8-10.

For neurocognitive impairment, small to large correlations with global indices of functioning have been reported, depending on the investigated domains, the use of a composite score (higher correlations for the composite score than for individual cognitive domains) and the rater of patient functioning (higher correlations for clinician rating than for patient self-report) 11, 12. Recent studies have shown that the impact of neurocognitive impairment may be mediated by functional capacity, i.e., the ability to perform tasks relevant to everyday life in a structured environment, guided by an examiner. In some studies, the impact of cognitive impairment on real-life functioning was negligible when functional capacity was included in the model 13, 14. Discrepant results have also been reported, i.e., no correlation between neurocognitive indices or functional capacity and self-reported real-life functioning 15, or a significant influence of neurocognitive dysfunction on everyday-life functioning in the presence of no influence of functional capacity 16.

Social cognition is currently considered a domain partly independent of neurocognitive functions and encompassing a large array of mental functions, from social perception to affect recognition, to theory of mind 17, 18. Relationships between deficits in social cognition and impaired social and occupational functioning have been reported 17, 19, 20. According to some studies, social cognition mediates the effect of neurocognitive impairment on real-life functioning 21, 22.

Negative symptoms have also been associated with patients' functional outcome 9, 23, 24. Both direct and indirect relationships between this psychopathological domain and real-life functioning have been reported 7, 13. Evidence of the role of negative symptoms as mediators of the impact of other variables (i.e., neurocognition or functional capacity) on real-life functioning has also been provided 9, 25. However, several limitations of previous studies might prevent solid conclusions and generalizability. In fact, negative symptoms have generally been regarded as a unitary construct, while the most recent literature suggests that these symptoms are heterogeneous and include at least two factors, “avolition” and “poor emotional expression”, that might be underpinned by different pathophysiological substrates 26, 27 and show different relationships to functional outcome 23. Moreover, largely used scales for the assessment of negative symptoms have been criticized for the inclusion of items assessing neurocognition and the focus on behavioral aspects, as opposed to internal experience, which may lead to artefactual associations with functional outcome measures 28, 29.

An impact of depressive symptoms on real-life functioning in schizophrenia has also been reported 30-32. However, an association has generally been found only when studies examined subjective indicators of real-life functioning 13, 33, suggesting that depression affects person's self-evaluation of functioning but not “real” functioning. The symptoms of depression may also affect functioning by interfering with subject's motivation and ability to properly organize him/herself in daily living activities. In this respect, the simultaneous evaluation of negative and depressive symptoms is important to clarify the relative contribution of these two psychopathological domains to real-life functioning.

Besides the variables summarized above, some studies have reported that patients with comparable severity of psychopathology may differ in their real-life functioning as a result of differences in personal resources 34-36. Resilience is a construct encompassing several aspects of personal resources. It is variously defined as a personal trait protective against mental disorders and a dynamic process of adaptation to challenging life conditions 37-40. In patients with schizophrenia, a significant correlation between resilience and psychosocial functioning has been reported 41; furthermore, lower baseline resilience was found among ultra-high risk subjects who converted to frank psychosis than among those who did not 42. Resilience is also related to patterns of mental health service engagement of patients with schizophrenia, which can affect real-life functioning 43.

Although it is obvious that real-life functioning is also influenced by the societal context, which includes disability compensation, job or housing opportunities, residential support, and various elements of attitudes and stigma 2, the identification of the most appropriate indices to capture the complexity of these variables is not an easy task. The evaluation of subjects' functioning with respect to employment or housing, for example, must take into account the offer of employment or housing in the place where the patient lives and the availability of social support, such as a disability pension. According to a recent study, differences in residential outcomes are likely based on differences in social services systems 44. Similarly, in a two-year follow-up of people with schizophrenia after their first episode, only those who were receiving disability compensation or were supported by their families were living independently 45. Indeed, it appears likely that interventions which modify the level of social support have an impact on real-life functioning in people with schizophrenia.

A higher level of internalized stigma – the process whereby people with severe mental disorders anticipate social rejection and consider themselves as devalued members of society 46-48 – has been found in association with lower levels of hope, empowerment, self-esteem, self-efficacy, quality of life, social support and adherence to treatment 49, suggesting an impact of this variable on real-life functioning, and the need to consider it in relevant studies.

From the summarized evidence, it is clear that real-life functioning of people with schizophrenia depends on a number of variables, some related to the disease, others to personal resources, and some more to the context in which the person lives. In the light of this complexity, it is crucial to consider all these aspects in order to explore their relative contribution.

In this paper we report on a large Italian multicenter study aimed to identify factors affecting real-life functioning of patients with schizophrenia and to assess their relative contribution. Factors to be included were chosen in the light of the literature review briefly summarized and of clinical experience, and were grouped into three categories: illness-related variables, personal resources and context-related variables. We predicted a significant association between the impairment of real-life functioning and the severity of negative symptoms and cognitive deficits, such as the more severe the negative symptoms and cognitive deficits, the more impaired the everyday functioning. As to negative symptoms, we expected that avolition would show a stronger association with real-life functioning than poor emotional expression. An association between neurocognition and real-life functioning was also expected, partly or entirely mediated by functional capacity and social cognition. We also hypothesized that the variables included among personal resources mediate the impact of symptoms and cognitive impairment on real-life functioning. Due to the paucity of literature data, it was more difficult to predict which context-related variables would show a direct or indirect association with real-life functioning. Nevertheless, we anticipated that a large social network and having access to social and family incentives would have a favorable impact on functioning, and that internalized stigma would mediate the influence of symptoms and cognitive deficits on functioning.

Several limitations of previous studies were addressed in the present investigation. The Brief Negative Symptom Scale (BNSS, 28) was used to assess negative symptoms; this is a recently developed instrument designed to overcome the above-mentioned limitations of other largely used scales for the assessment of negative symptoms. Depressive and extrapyramidal symptoms were evaluated to ascertain their possible influence on negative symptoms and real-life functioning. The MATRICS (Measurement and Treatment Research to Improve Cognition in Schizophrenia) Consensus Cognitive Battery (MCCB) was chosen for cognitive assessment, as it is regarded as the “state of the art” neuropsychological battery for research purposes in schizophrenia 50, 51. A full assessment of different aspects of social cognition was carried out, including emotional intelligence, emotion recognition and theory of mind. Personal resources and context related factors were included in the study.

In the light of difficulties in defining and measuring real-life functioning 2, the Specific Levels of Functioning Scale (SLOF) was selected to measure social, vocational, and everyday living outcomes 52. This instrument was endorsed by the panel of experts involved in the Validation of Everyday Real-World Outcomes (VALERO) initiative as a suitable measure of real-life functioning 12, 53.

Due to the high number of included variables, a large multicenter study was designed to be carried out in 26 university psychiatric clinics and/or mental health departments, recruiting up to 1000 subjects with schizophrenia.

METHODS

RESULTS

Out of 1691 screened patients, 1180 were eligible; of these, 202 refused to participate, 57 dropped out before completing the procedures and 921 were included in the analyses (641 males, 280 females). Data on demographic and illness-related variables are provided in Tables 1 and 2. Almost all patients were treated with antipsychotics, mostly second-generation drugs, and about one quarter received an integrated treatment, i.e., psychosocial interventions in addition to pharmacotherapy (including cognitive rehabilitation, psychoeducation, social skills training, self-help groups or sheltered employment).

Data on personal resources, context-related factors and functional capacity, and SLOF scale scores are provided in Table 3. Overall, study participants showed a modest degree of functional impairment. SLOF domains showing a moderate degree of impairment included interpersonal relationships, community activities and working abilities.

Inspection of the bivariate correlation matrix revealed that the socio-economic status (Hollingshead index), the social network and extrapyramidal symptoms were unrelated to all mediators and SLOF; therefore, these variables were not used in further analyses.

In our first SEM model (Figure 1), PANSS positive, PANSS disorganization, BNSS avolition, BNSS poor emotional expression, depression, neurocognition, and incentives were used as independent predictors; social cognition, functional capacity, internalized stigma, resilience and service engagement were used as mediators, and SLOF was the dependent variable.

PANSS positive and PANSS disorganization, as well as BNSS avolition, proved to have significant direct and indirect effects; depression had no significant effect on SLOF, and BNSS poor emotional expression was only indirectly and weakly related with SLOF (Table 4, Figure 1). Neurocognition had only indirect effects on SLOF. Incentives proved to be a significant predictor, and social cognition, functional capacity, resilience, internalized stigma and service engagement served as mediators, as hypothesized.

CFI and TLI indices for this model were 0.925 and 0.916, respectively, and the RMSEA index was 0.047, denoting a good fit to the data. The included variables explained 53.5% of SLOF variance.

After trimming non-significant paths (from neurocognition, CDSS, BNSS poor emotional expression to SLOF, and from social cognition to internalized stigma), a final model was obtained with five predictors and five mediators (Figure 2). This model accounted for 53.8% of variance of the SLOF, was more parsimonious and proved to have a better fit compared with the initial model (CFI=0.940, TLI=0.932, RMSEA=0.044). Comparison of the AIC indices for the initial and the final model (84686.906 and 80400.015, respectively) further supported the choice of the latter to represent the relationships among variables without loss of information.

In this final model, PANSS positive, PANSS disorganization and BNSS avolition showed a negative direct effect on SLOF (Table 4), indicating that higher levels of psychopathology are associated with poorer functioning. Several indirect effects on SLOF were also observed: PANSS positive through service engagement, PANSS disorganization through functional capacity, and BNSS avolition through service engagement and resilience. BNSS avolition had also an effect on internalized stigma that, in its turn, was indirectly associated with SLOF through resilience. Neurocognition showed indirect effects on SLOF through four different mediators: service engagement, functional capacity, internalized stigma (through resilience) and social cognition, and when compared with other predictors of the same mediator, it always showed the strongest effect (Figure 2).

Table 4 provides a summary of direct, indirect and total effects on SLOF of variables included in the final model. Neurocognition showed the strongest total effect, followed by disorganization, avolition, functional capacity, service engagement, social cognition, positive symptoms, incentives, resilience and internalized stigma.

DISCUSSION

To our knowledge, this is the largest study carried out so far on factors associated with real-life functioning in people with schizophrenia, in terms of both sample size and number of investigated domains. According to our findings, variables relevant to the disease, personal resources and social context explain 53.8% of real-life functioning variance in patients with schizophrenia living in the community and treated with antipsychotics, mainly second-generation drugs.

Neurocognition exhibited the strongest association with real-life functioning. This result corroborates previous findings of moderate to large correlations of neurocognition with everyday functioning measures 75-77. Our use of neurocognition as a latent variable, to reduce its measurement error, further supports the robustness of the finding. In line with previous research 77, 78, neurocognition proved to be a distal variable with respect to real-life functioning, since its relationship with SLOF was mediated by functional capacity, social cognition, service engagement and internalized stigma.

Associations between functional capacity and neurocognition have previously been reported in cross-sectional studies 79-81. In our model neurocognition was the strongest predictor of functional capacity, which reflects its important contribution to the latter construct. Actually, both measures assess the individuals' capability of performing tasks and/or behaviors in a standardized setting, but tests of functional capacity do so in more “ecological” way, i.e., simulating everyday life tasks, though not carried out in the community 15, 82. The role of functional capacity as mediator of the impact of neurocognition on real-life functioning reported in our study has also been previously observed 13, 14.

The correlation between social cognition and real-life functioning ranged from small to large in previous studies, mainly depending on the examined aspect of social cognition, with largest effects observed for theory of mind tasks 17, 20. Our findings confirm that social cognition accounts for a unique proportion of functioning variance, independent of neurocognition 83-87, support its independence from negative symptoms 18, 22, and do not support its role as mediator of the impact of negative symptoms on real-life functioning 7.

To our knowledge, the role of service engagement and internalized stigma as mediators between neurocognition and real-life functioning has not been investigated in previous studies. In our model, service engagement was directly associated with SLOF, while internalized stigma showed, in its turn, an indirect association with SLOF, mediated by resilience. Service engagement, as assessed in the present study, reflects subject's degree of collaboration with mental health services (e.g., active participation in defining treatment plans, ability to seek service help if needed and to show up on time for the appointments) and adherence to prescribed treatments. According to our results, impaired cognitive functioning interferes with subject's collaboration to treatment and therefore it is an obstacle to successful outcome.

Internalized stigma, according to a recent meta-analysis including 127 studies, is directly associated with severity of psychiatric symptoms and inversely related to levels of hope, empowerment, self-efficacy, quality of life and adherence to treatment 49. Our data confirm the previously reported association between internalized stigma and negative symptoms 88, and extend this finding to neurocognition. An association between social cognition and internalized stigma has been reported by other authors 89, but has not been observed in our model. A possible interpretation of this discrepancy could be that the relationship between social cognition and internalized stigma is spurious; in other words, it is possible that neurocognitive impairment underlies both these variables and therefore accounts for their relationship.

The illness-related variables disorganization, positive symptoms and avolition were both directly and indirectly related to functioning. The key role of disorganization observed in our study deserves comments. The PANSS items “conceptual disorganization”, “difficulties in abstract thinking” and “poor attention” are generally considered core aspects of the disorganization factor 90-92; in the present study, the structure of the PANSS disorganization dimension was defined according to the consensus 5-factor solution proposed by Wallwork et al 55, in which the three above-mentioned items load on the disorganization factor. The overlap of the items “difficulties in abstract thinking” and “poor attention” with neurocognitive impairment cannot be underestimated. As a matter of fact, in our data, neurocognition and disorganization showed a significant inverse correlation and a similar pattern of association with SLOF, through functional capacity and social cognition.

Avolition had both a direct and an indirect relationship with SLOF. In the indirect path, it has service engagement, resilience and internalized stigma (in its turn associated with SLOF through resilience) as mediators. In the initial model, both BNSS factors had been included, but BNSS poor emotional expression only showed an indirect and weak relation with SLOF, and its exclusion from the final model did not worsen the fit or reduce the explanatory power of the model.

A significant relationship between avolition and poor social outcome has been reported in previous studies 93-95. In a recent investigation on long-term stability and outcome of negative symptoms, Galderisi et al 23 found that avolition has a higher predictive value of functional outcome than poor emotional expression at 5-year follow-up. The scarce relevance of poor emotional expression to real-life functioning is in line with Foussias et al's findings 96.

The strong impact of avolition on real-life functioning might be due to the partial overlap between these two measures. However, the degree of overlap was most probably limited in our study, since avolition, as measured by the BNSS, provides an assessment of both behavioral (e.g., deficit in initiating and persisting in different activities) and inner experience aspects (e.g., lack of interest and motivation in different activities, impaired anticipation of rewarding outcome), while the real-life assessment provided by a caregiver mainly focuses on subject's performance and behavior in several types of everyday activities. Efforts aimed to improve our understanding of the pathophysiology of avolition represent a priority of research in schizophrenia, and the implementation of treatments targeting motivation is likely to be an important tool to enable people with schizophrenia to achieve a meaningful life.

At odds with previous literature 8, 13, 14, 53, 97, 98, no impact of depression on real-life functioning was observed in our study. The discrepancy with previous studies might be due to the use of different instruments for the assessment of depression (mostly the Beck Depression Inventory, a self-report measure, in prior investigations) and the low degree of severity of depression in our sample (mean score=4, with a cutoff of 6/7 for depression in the CDSS).

Extrapyramidal symptoms, family socio-economic status and social network were not included in the SEM, as they showed no or weak associations with SLOF and mediating variables. As to extrapyramidal symptoms, we cannot exclude that the prevalence of treatment with second-generation antipsychotics yielded a floor effect, making their impact on SLOF negligible, while for both social network and socio-economic status we cannot rule out the possibility of redundancy with other variables, or poor performance of the used instruments.

Access to family and social incentives had a negative association with SLOF, i.e., a higher number of incentives was associated with a poorer real-life functioning. Due to the cross-sectional study design, this association may be interpreted either way, i.e., access to incentives may be due to functional impairment or incentives may have a negative impact on real-life functioning. In fact, many patients would not renounce to the disability pension or the support received by the family for a job, as the latter is generally regarded as less stable and more effortful. Directly relevant to this point is the finding by Rosenheck et al 99 that disability compensation status, which is often linked to the individual's health insurance coverage, had the largest negative impact on vocational outcomes of all the measured predictors.

In conclusion, we found that some illness-related variables (neurocognition, disorganization, avolition and positive symptoms) and incentives predict real-life functioning either directly or through the mediation of resilience, stigma, social cognition, functional capacity and engagement with mental health services. The final SEM model explained about 54% of the SLOF variance, a higher percentage compared with those reported in similar studies that used neurocognition, social cognition, social competence and negative symptoms to predict real-life functioning (7-25%) 7, 22, 83, 87.

The strengths of this study include the large sample size and the use of state-of-the-art instruments to assess neurocognitive, psychopathological, social cognition, and personal resources domains. Some possible limitations include the restricted variability range of patients' clinical characteristics and functioning in the real life (most of the patients showed a mild/moderate degree of symptoms severity and functional impairment); the use of SLOF as a latent variable, which might have advantages, but might also prevent the identification of predictors of specific domains, and the cross-sectional design, which does not allow to test causal relationships.

Our findings can have important treatment and research implications. The impact of neurocognition and social cognition on real-life functioning suggests that training addressing neurocognitive and social cognition impairment should be part of integrated treatment packages for schizophrenia. A greater emphasis on social cognition than on neurocognition has been suggested, given the greater proximity and higher direct explanatory power of the former, with respect to the latter domain 100. However, there is no evidence that social cognitive training alone counteracts neurocognitive impairment, whose impact on other domains, i.e., functional capacity and service engagement, would probably persist, in spite of possible improvement in social cognition.

The complexity of the pathway from neurocognition to real-life functioning through internalized stigma suggests that, in order to enhance the impact of interventions targeting neurocognitive impairment on functioning in the real life, we also need to promote reduction of internalized stigma related to mental illness and minimize its negative effects. Data have been provided that anti-stigma interventions are effective at reducing internalized stigma 101-103, but the impact of such positive outcome on other dimensions of the disorder and on patients' functioning does need further investigation.

Our finding that avolition is an independent domain with respect to both neurocognition and social cognition suggests that the search for treatments with an impact on this domain should be a priority of mental health research strategies.

The contribution of resilience to real-life functioning highlights the importance of personalization when designing treatment plans and defining life goals together with our patients. Prejudicial optimistic or, more frequently, pessimistic attitudes should always be modulated by the awareness that individuals do vary a lot in terms of personal resources, and such a variability does not allow undue generalization.

Improved understanding of factors that hinder real-life functioning is vital for treatments to translate into more positive outcomes. Findings from the present study provide a valuable contribution in this direction; in particular, the observed complex associations among investigated predictors, mediators and real-life functioning strongly suggest that integrated and personalized programs should be provided as standard treatment to people with schizophrenia.

Appendix

Members of the Italian Network for Research on Psychoses include: Marcello Chieffi, Stefania De Simone, Francesco De Riso, Rosa Giugliano, Giuseppe Piegari, Annarita Vignapiano (University of Naples SUN, Naples); Grazia Caforio, Marina Mancini, Lucia Colagiorgio (University of Bari); Stefano Porcelli, Raffaele Salfi, Oriana Bianchini (University of Bologna); Alessandro Galluzzo, Stefano Barlati (University of Brescia); Bernardo Carpiniello, Francesca Fatteri, Silvia Lostia di Santa Sofia (University of Cagliari); Dario Cannavò, Giuseppe Minutolo, Maria Signorelli (University of Catania); Giovanni Martinotti, Giuseppe Di Iorio, Tiziano Acciavatti (University of Chieti); Stefano Pallanti, Carlo Faravelli (University of Florence); Mario Altamura, Eleonora Stella, Daniele Marasco (University of Foggia); Pietro Calcagno, Matteo Respino, Valentina Marozzi (University of Genoa); Ilaria Riccardi, Alberto Collazzoni, Paolo Stratta, Laura Giusti, Donatella Ussorio, Ida Delauretis (University of L'Aquila); Marta Serati, Alice Caldiroli, Carlotta Palazzo (University of Milan); Felice Iasevoli (University of Naples Federico II); Carla Gramaglia, Sabrina Gili, Eleonora Gattoni (University of Eastern Piedmont, Novara); Elena Tenconi, Valeria Giannunzio, Francesco Monaco (University of Padua); Chiara De Panfilis, Annalisa Camerlengo, Paolo Ossola (University of Parma); Paola Landi, Grazia Rutigliano, Irene Pergentini, Mauro Mauri (University of Pisa); Fabio Di Fabio, Chiara Torti, Antonino Buzzanca, Anna Comparelli, Antonella De Carolis, Valentina Corigliano (Sapienza University of Rome); Giorgio Di Lorenzo, Cinzia Niolu, Alfonso Troisi (Tor Vergata University of Rome); Giulio Corrivetti, Gaetano Pinto, Ferdinando Diasco (Department of Mental Health, Salerno); Arianna Goracci, Simone Bolognesi, Elisa Borghini (University of Siena); Cristiana Montemagni, Tiziana Frieri, Nadia Birindelli (University of Turin).