Introduction

Currently, there is an active scientific research to determine the nature and degree of mutual influence of clinical manifestations of schizophrenia, psychological, socio-demographic characteristics of a patient, his quality of life (QoL) and social functioning (SF).

The level of SF in patients with schizophrenia is considered to depend directly on the psychopathological profile of the disease, the age at which the disease manifested itself, the number of psychotic episodes, and the rate of the disease process progression [1—3]. Schizophrenia already in the early stages of the disease has a pronounced negative impact on the functioning of patients in all spheres of life. In the group of patients with delayed diagnosis of schizophrenia, clinical and socio-demographic indicators are lower than in patients who were diagnosed in time. The literature [4] draws attention to the fact that delayed treatment significantly reduces the rehabilitation potential of patients. It concerns many spheres of functioning of patients with schizophrenia in remission — not only socially useful activity, including work and study, but also the sphere of relations with relatives. The results of a study by K. Jaracz et al. [5] showed that the risk of a negative outcome of the disease can be reduced by appropriate interventions at an early stage.

Patients with paranoid schizophrenia have more pronounced cognitive deficits than patients with schizotypal disorder, although the psychological adjustment profiles of patients with these diseases do not differ significantly [6—8]. G. van Rooijen et al. [9], using a special mathematical model, noted that the presence of depression has a direct influence on QoL in schizophrenic patients, while the symptoms of the schizophrenic spectrum itself affect QoL through SF. In a meta-analysis conducted by T. Halverson et al. [10] studied the relations between SF disorders, cognitive function indicators and social cognition in schizophrenic spectrum disorders. Social cognition was found to explain more significant differences in functioning than cognitive impairment, but a large proportion of these differences remained unexplained. Nevertheless, there is agreement that comprehensive interventions aimed at improving cognitive function and social cognition can improve NF in patients with different illnesses [10, 11]. In the paper presented by A. Nevarez-Flores et al. [12] meta-analysis, along with descriptive characteristics, used correlation coefficients between SF and QoL indicators, but taking into account heterogeneity in methodological approaches. The correlation of social functions with objective measures of QF was stronger than with subjective measures. The highest correlations were found for the social domains. The authors concluded that SF "predicts" overall QoL and, crucially, the severity of depressive and negative symptoms of schizophrenia. The essential role in the adaptation of the patient is given to the subjective assessment of CS. The discrepancy between a decreased level of EF and high satisfaction with one's psychosocial status leads to additional social disadaptation [13]. Low quantity and low quality of interpersonal communications together with severe symptoms of the disease have a negative effect on treatment outcomes and limit functional recovery in patients with schizophrenia [14, 15]. According to D. Fulford et al. [16], social functions can provide the necessary conditions for successful social communication of patients. In the study of N. Germain et al. [17] revealed the connection between clinical manifestations, SF and QoL taking into account the use of health care resources by the methodology having measurable indicators (HCRU methodology). The authors made an important conclusion that treatment outcomes assessed by the PANSS scale are associated with HCRU and QoL.

The age of patients is considered to be an important predictor of successful social adaptation of patients with schizophrenia [18]. In young patients, factors contributing to preservation of high social adaptation are qualitative remission, professional education, late debut of schizophrenia, rare hospitalizations, preserved social connections, sufficient level of self-esteem, constructive type of adaptive behavior [19]. QoL and SF are also significantly influenced by family factors [20, 21]. It has been shown that the best SF was observed in patients receiving treatment outside hospital, who had sufficient support of the immediate environment and absence of other family members with mental disorders [22, 23]. A number of researchers believe that patients living in families whose members are unfriendly, hostile, and aggressive toward the patient relapse more often [24, 25]. The main modern method of correction of these family communication disorders all over the world is psychoeducation, which, despite its proven effectiveness, needs more targeting of exposure factors [26].

Assessment of the influence of different factors on the trajectory of the disease process in schizophrenia is very difficult because of their multiplicity, heterogeneity, and imperfection or lack of valid assessment tools. For this reason, mechanisms of mutual influence of dynamics of clinical manifestations of disease, personality of the patient, his psychological, social and demographic characteristics and social environment factors remain poorly studied. The threshold levels of cognitive and other mental disorders necessary for predicting the severity of functional deficits have not been sufficiently investigated [6]. It is important that modern authors point not only to the necessity of using indicators of disease symptoms and cognitive functions in predicting SF, but also consider indicators of functioning as prerequisites for making a diagnosis. Moreover, T. Suzuki [25] in his work describes a new paradigm in psychiatry, according to which patients with schizophrenia can be classified using simple and pragmatic indicators — both clinical and SF. In this connection, the search for a new valid psychometric tool becomes urgent. A.A. Marchenko et al. [27] proposed an "autonomy scale" to measure a schizophrenia patient's ability to function independently (autonomously), based on the study of correlations between the PANSS, PSP, and EQL scores.

In the field under consideration, much attention is currently paid to the methods of mathematical modeling, which allow revealing not only direct, but also indirect relations [9, 17]. This approach makes it possible to establish a hierarchical relationship in which individual factors are assigned the role of moderators, or facilitators, i.e., factors that only indirectly ensure the success of SF [16]. Accordingly, it seems extremely relevant to search for approaches to the development of appropriate new constructs. At the same time, there is reason to believe that the development of this direction can potentially find its application in justification of targets, volume and sequence of organization of a modern modular system of providing medical and rehabilitation and socio-psychological care to patients [28].

The aim of the present study was to investigate the relationship between the indexes of QoL and SF of patients with paranoid schizophrenia, clinical-dynamic and socio-demographic characteristics, psychometric indexes of cognitive functions and severity of disease symptoms, using modern mathematical modeling methods.

Material and methods

A total of 300 patients with paranoid schizophrenia were selected by a blanket method.

Inclusion criteria for patients: age 18-50 years, diagnosis of "schizophrenia, paranoid form" (section F20.0 of ICD-10), duration of illness at least 5 years.

Non-inclusion criteria: acute psychosis (total PANSS score >120), marked somatic disorders that impede daily functioning, intensive sedative pharmacotherapy that reduces attention and ability to absorb new information.

The sample included 245 inpatients and 55 outpatients. In the inpatient and outpatient phases, patients received medical care in accordance with current standards of diagnosis and treatment of schizophrenia. Medical care included observation of patients by qualified psychiatrists, access to all resources of the dispensary: a day hospital, a medical rehabilitation unit, a medico-social and psychological assistance office, and free medication with modern psychopharmacological agents. In all cases, hospitalization was due to an exacerbation of the disease.

The diagnosis of schizophrenia was verified using the Structured clinical interview for DSM-5, Clinical Trials Version (SCID-5-CT).

The research protocol included the following modules: main psychopathological manifestations and clinicodynamic characteristics of schizophrenia (schizophrenia syndromes, duration of the disease, number of years since manifestation, number and duration of hospitalization, etc.); sociodemographic characteristics (age, gender, education, employment, disability, family characteristics, etc.); data from clinical and psychological examination.); data from clinical-psychological and experimental-psychological research; psychometric scores on the PANSS scale [29], the BACS Short Assessment of Cognitive Functions scale [30, 31], the WHO quality of life questionnaire (special module for schizophrenia patients "QoL-SM") [32], the PSP socially oriented and social functioning scale [33].

Statistical analysis methods were also used in the study. Categorical variables were presented as absolute values (n) and fractions of the whole (%), continuous variables as mean and standard deviation (M±SD), discrete and ordinal variables as median (Me) and 1-3rd quartiles ([Q₁; Q₃]).

The calculation of the latent factor was performed using the Gifi package [34], which is designed to solve the problems of optimal scaling with the subsequent calculation of the principal components [35]. A latent factor was obtained — a general index containing an integral assessment of QoL and SF, which in the present study was named the "disease burden factor" (factor "B"). Hierarchical cluster analysis by Ward's method was performed for primary analysis on the outcome coordinates [36]. Intergroup analysis was performed using the Mann-Whitney test (U-statistics). Spearman correlation coefficient was used to assess the relationship between the obtained factor and interval-scale estimates. The most informative variables for predicting the value of factor "B" were selected using the "random forest" method (1000 oversamples, probability of error of the first kind, p=0.01) [37]. The selected variables were included in Bayesian network analysis (rsmax2 algorithm), the results of which are presented as linear model coefficients [38]. The null hypotheses were rejected at p<0.05. Calculations were performed in the programming language R v3.6.3 [39].

Results

When studying the data of descriptive statistics, a large number of coincidences between the results of expert assessments according to the PSP scale and self-reports according to the QoL-SM scale were found. This observation served as a basis for calculating, on the basis of rating scales, a common computable index containing an integral assessment of QoL and SF — factor "B". A linear model was used to assess the influence of qualitative variables on the distribution of factor "B".

The distribution of QoL (QoL-SM scale) and SF (PSP scale) values by the Gifi method showed a non-linear association of QoL and SF indicators with the explained variance of the principal component in the studied sample (61.4%), which allows us to recognize its correspondence to the calculable indicator — factor "B". Significant correlations were revealed between the value of the principal component and the initial and transformed scores of the QoL-SM and PSP scales. The results of the factor "B" test indicated its validity and reliability.

The correlations between factor "B", sociodemographic and morbid characteristics, indicators of clinical-psychological and experimental-psychological examinations (145 variables) were studied. As a result, we found statistically significant correlations of factor "B" with age (p=0.001), age at manifestation (p<0.001), age at seeking psychiatric help (p<0.001) and diagnosis (p<0.001); with duration of hospitalization in the last 12 months (p=0.022), and with indicators of "positive symptoms" (p<0.001), "negative symptoms" (p<0.001), "other symptoms" (p<0.001), "composite index" (p<0.001), "total score" (p<0.001) of the PANSS scale; "verbal learning" (p<0.001), "working memory" (p=0.037), "motor skills" (p<0.001), "fluency" (p<0.001), "coding" (p<0.001), "Tower of London" (p=0.014), "composite index" (p<0.001), "total score" (p<0.001) BACS scale (Table 1).

Table 1. Statistically significant correlations of factor "B" with indicators of the PANSS, BACS, QoL-SM, PSP scales, clinical and dynamic characteristics, and age

Note. r — Spearman’s rank correlation coefficient.

The study did not find any associations of factor "B" with other categorical, quantitative and ordinal variables of the modules "main psychopathological manifestations and clinical and dynamic characteristics of schizophrenia" and "socio-demographic characteristics" included in the descriptive statistics. In other words, there was no significant association between the index of integral assessment of QoL and SF with such groups of characteristics as "sex", "education", "marital status", "residence", "presence of children", "type of care", "comorbid alcohol dependence", "main psychopathological syndromes", etc.

Division of patients into groups by cluster analysis relative to the indices of factor "B" allowed us to distinguish two categories of patients (see figure). The calculated principal component PC1 explained 61.4% of the variance in the data. Based on the theoretical assumption that PC1 is a numerical reflection of factor "B," the extracted integral index connects SF and QoL in an optimal way. Cluster 1 relative to the main component (factor "B") includes conditionally "well" patients; cluster 2 includes conditionally "unfavorable" patients.

Distribution of patients with schizophrenia into clusters, relative to the factor "B" index.

PSP — the Personal and Social Performance Scale; QoL — World Health Organization Quality of Life Questionnaire, a special module for patients with endogenous psychoses; Cluster 1 ("prosperous"); Cluster 2 ("unprosperous"); PC — principal component.

Intergroup analysis of quantitative and ordinal characteristics between the cluster of "advantaged" (cluster 1) and "disadvantaged" (cluster 2) patients showed significantly higher frequency of exacerbations, which required a change in treatment regimen in "disadvantaged" patients, as well as longer duration of disease (p<0.001). Patients in this cluster also had significantly more pronounced symptomatology according to all PANSS scales (p<0.001). QoL and SF scores were expectedly higher in "advantaged" patients in cluster 2 (p<0.001) (Tables 2, 3). "Well-being" patients differed in higher cognitive scores on the subtests "motor test with chips" (p=0.003); "fluency" and "coding" (p<0.001), and on the total and composite values of the BACS technique (p<0.001).

Table 2. Comparative analysis of the indicators of the PANSS and BACS scales in the clusters patients with schizophrenia, relative to the distribution of the factor "B" index (significant differences)

Note. Here and in table 3: U — statistics of the Mann-Whitney criterion; p — probability of error of the first kind.

Table 3. Comparative analysis of indicators of the QoL-SM, PSP scales, duration of the disease and the number of hospitalizations in the clusters patients with schizophrenia, relative to the distribution of the factor "B" index (significant differences)

In the clusters of "disadvantaged" and "advantaged" schizophrenia patients, a comparative analysis of the remaining indicators of the modules "basic psychopathological manifestations and clinical and dynamic characteristics of schizophrenia" and "sociodemographic characteristics" showed no significant differences, except that there were more patients with disabilities in the "disadvantaged" cluster: 89 (55.6%) versus 45 (32.1%) in the "advantaged" cluster (p<0.001).

The results of the present study showed an association between the severity of clinical manifestations of the disease, QoL and SF of a schizophrenic patient, which is confirmed by the results of the study of A. Galuppi et al. [40]. Earlier T. Tominaga et al. [41] established a connection between the scores of the subtest "coding" of the BACS scale, symptom severity according to the PANSS scale and the SF and QoL scores. At the same time, the influence of symptom severity was assessed as stronger. Another study [42] found an association between verbal fluency and QoL. Data from a meta-analysis conducted by N. Germain et al. [17] suggest that symptom severity is the most significant predictor of QoL in patients with schizophrenia. Nevertheless, in the literature, the influence of symptom severity on SF is estimated by the authors less definitely.

The present study was based on a mathematical model based on a construct that integrally reflects QoL and SF of a patient. The proposed index, named "burden of disease factor" (factor "B"), has an explained variance of the main component in the studied sample (61.4%) that allows to consider it valid and reliable. Factor B was used to relate the indicators of clinical symptom severity, QoL, and SF to quantitative indicators: age, gender, age of patients at the time of diagnosis and manifestation, duration of the disease, etc., by logical relationships. Factor "B" found a significant association with the scores of all subtests of the PANSS and BACS methods. However, according to cluster analysis, stable differences between "advantaged" and "disadvantaged" patients regarding the indicator of factor "B" were revealed only in the subtests "motor test with chips", "fluency" and "coding" of the BACS scale, which indicates a significant role of the dynamic characteristics of the patient's mental activity in the maintenance of QoL and SF.

The data of the present study are consistent with current ideas about the mechanisms of cognitive deficits in schizophrenia, where one of the main problems is considered to be spontaneous integration of information, which prevents the performance of tasks requiring the involvement of active attention and working memory. Patients in the "disadvantaged" cluster had significantly worse indicators of disease duration, the number of hospitalizations and disability compared to "advantaged" patients, which indicates a tendency for deterioration of patients' condition over time, reduction of QoL and SF and correlates with the data from other national studies [43].

The study showed reliability and validity of the method developed by the authors to determine the factor "B", which allows integral estimation of SF and QoL of a schizophrenia patient. The mathematical model based on the proposed construct seems to be a reliable tool for studying the influence of different factors on the trajectory of schizophrenia. The authors suggest further development of the study, in particular, taking into account the contribution of molecular genetic factors [44].

The study was conducted within the RFBR project 17-29-02173 "Assessment of the influence of environmental pathoplastic factors on the clinical manifestations of schizophrenia, taking into account the patient's genotype.

The authors declare no conflicts of interest.