Table 1.
Main characteristics of the reviewed studies.
| Authors | Sample and duration of the study | Focus of the study | Study quality | Results |
|---|---|---|---|---|
| Leeson et al. (22) | 129 FEP patients in subgroups of different IQ levels and 27 controls. 3-years | IQ Trajectory and Cognitive Reserve | Covered comparability among groups. Appropriate and reproductible statistics: Schwarz’s Bayesian Criterion (BIC) | Better IQ at the beginning of the disease predicted better social functioning with regards to employment/occupation. Premorbid IQ did not |
| Bachman et al. (23) | 35 FEP patients with adolescent onset psychosis (12-20y) and 31 controls. 1-year | Cognition and social function in Teenage Onset | Covered comparability among groups. Appropriate and reproductible statistics: Step-wise linear regression | Worse cognitive processing speed performance predicted worse social functioning |
| Torgalsbøen et al. (24) | 28 FEP patients and 28 pairwise matched controls for age, gender and education. 6-months | Influence of neurocognition on clinical remission and real-life functioning | Covered comparability among groups. Appropriate and reproductible statistics: logistic and linear regression analyses. ROC analyses. | The ROC analyses showed attention was significantly predictive of remission. Attention/Vigilance and Working Memory at baseline significantly predicted social functioning 6 months later. |
| Torgalsbøen et al. (25) | 28 FEP patients and 28 pairwise matched controls for age, gender and education. 2-years | Neurocognitive predictors of recovery. Rates of remission | Covered comparability among groups. Appropriate and reproductible statistics: Linear regression analyses for predictions, with Bonferroni corrections for multiple testing | Attention/vigilance and more years of education at baseline were predictors of better social and role functioning at follow-up, with 44–48% of the variance explained by the model |
| Uren et al. (26) | 133 FEP patients and 46 controls. 6-months | To identify cognitive profiles within FEP patients, through predictive validity of cognitive clusters in relation to symptoms and functioning | Appropriate and reproductible statistics: Ward’s hierarchical agglomerative with k-means verification for the cluster analysis. Clusters were externally validated and 6-month predictive validity was also examined | Three-cluster solution according general cognition indexes based in units of SD below the means. Poorer cognitive functioning (cluster 1) was associated with poorer participation and functioning in educational, occupational, and social roles. Premorbid IQ was a significant predictor of group membership |
| Lindgren et al. (27) | 66 FEP patients and 62 controls. 2-months | Theory of mind and social functioning. Whether ToM impairments are explained by cognitive deficits or ToM is impaired independently | Covered comparability among groups. Appropriate and reproductible statistics: ROC, including AUC and bootstrap CI. Linear regression models with adjusted R2 | Level of social functioning was associated with ToM when age and gender were controlled for, and this association remained significant after controlling for general cognition |
| Griffiths et al. (28) | 98 FEP patients and 30 controls. 1-year | Find predictors for poor global functional -GF- outcome: exploring clinical symptomatology, neurocognition, and social cognition | Covered comparability among groups. Appropriate and reproductible statistics: Separate backward logistic regressions to predict dichotomy in GF cut-off for social and role functioning. Nagelkerke R2. Exploratory mediation analyses (Sobel test) | When considered alone, better verbal memory was a significant predictor for less likelihood to have poor role functioning, and social knowledge significantly predicted social functioning. Attribution style, emotion perception and visuospatial processing were not significant. When including symptoms, the negatives were the only significant predictor for 12-month social (OR = 1.12) and role (OR = 1.16) functioning. Negative s. mediated the relationship between baseline social knowledge and social functioning, and between baseline memory and role functioning at follow-up |
| Pelizza et al. (29) | 141 FEP (13–35y) patients and 98 non-FEP patients. 2-years | Social Cognition and its association with psychopathology and community functioning (work and social) | Appropriate and reproductible statistics: Multiple linear regression analysis | The social cognition impairment sub-score (from a pooled measure combined with a sub-score of classic cognitive domains and a sub-score of social cognition per se) was a significant predictor of cognitive change and depression at baseline (not tested at follow-up) |
| Woolverton et al. (30) | 71 FEP. 6-months | Social Cognition (divided in 5 domains: attributional style, emotion recognition, social knowledge, social perception and theory of mind) and whether is associated with social functioning | Appropriate and reproductible statistics: Cross-sectional Pearson and linear regression for predictive value of SC at follow-up | There were no longitudinal associations between social cognition at baseline and any measure of social functioning at 6-month follow-up. These results held both with and without corrections for multiple comparisons |
| Pennou et al. (31) | 37 FEP. 3-months | To evaluate the prognostic value of theory of mind, neurocognition and negative symptoms | Appropriate and reproductible statistics: multiple hierarchical linear regressions (in 3 steps and Intro) | The addition of the theory of mind in STEP3 of the regression model to predict independent living skills and family was associated with a significant increase in the variance explained by the model |
| González-Blanch et al. (32) | 131 FEP in two groups: functional deficits (97) vs. functional recovery (34). 1-year | Whether global functional recovery (integrating social and occupational outcomes) after a FEP can be predicted by cognitive variables | Appropriate and reproductible statistics: Binary logistic regression with forward stepwise based on likelihood ratio, with recovery as the dependent variable | Sustained attention was a predictor of social functioning. Prediction accurate (97%) for the disabled but not for the recovered group |
| Popolo et al. (33) | 15 FEP of age 18–30. 1-year | Cognitive assessment, social functioning and quality of life | Statistical analyses with low power to address predictions: t-test and Pearson correlations | Social functioning and quality of life found to be mostly independent of cognitive impairment (not correlations among variables found, except for attention and some subscales of social functioning) |
| Horan et al. (34) | 55 FEP. 1-year | Social Cognition and functional correlates | Appropriate and reproductible statistics: 2-wave panel design (0 and 12 months) to evaluate associations using a cross-lagged panel correlational analyses | At the 12-month follow-up, better social cognition - both theory of mind and emotional processing - was related to better real-world functioning (work, independent living, and social functioning). The pattern of cross-lagged correlations suggested that the direction of causality is from baseline social cognition leading to 12-month functioning |
| Vestarger et al. (35) | 117 FEP (18-34y). 10-months | Cognitive and clinical predictors of functionality. The associations between functional capacity and measures of real-world functioning | Linear and logistic regression analyses with backward exclusion based on Wald-test were adjusted for age, gender, and site | Verbal learning (30% variance explained), working memory (further 9%), and negative symptoms (4%) predicted 4-month functional capacity, accounting for 47% of the variance. Working memory (24%) and visual learning (6%) predicted 10-month functional capacity, accounting for a 30% of the variance |
| Stouten et al. (36) | 153 FEP. 1-year | Psychotic symptoms and cognition as predictors of functioning | Appropriate and reproductible statistics: Backward regression models with post-hoc analyses (Holm–Bonferroni correction) | Some concrete cognitive deficits (visual learning) were stronger predictors for functionality in the areas of vocational/academic performance and social relationships problems at the follow-up. Theory of Mind was also a predictor for problems in social relationships at the follow-up |
| Ventura et al. (37) | 48 FEP patients and 21 controls. 6-months | Theory of mind. To test whether the relation-ship between neurocognition (MATRICS) and functional outcome was mediated by ToM | Covered comparability among groups. Appropriate and reproductible statistics: Mediation analyses in consecutive model with the Sobel test | Longitudinal analysis indicated that neurocognition was a significant mediator of the significant relationship between both of the ToM variables (intentionality and appropriateness) and role functioning at 6 months |
| Norman et al. (38) | 79 FEP. 5-years | Cognition and functional prediction. IQ | Appropriate and reproductible statistics: Hierarchical linear regression and logistic regression to assess the role of cognitive indices in predicting outcomes. | No relationship was detected between cognitive function and the social functioning score (SOFAS) at 5 years. However, robust correlations appeared between education at onset, premorbid academic adjustment, and overall IQ with full-time occupation and use of disability pension at 5 years |
| Faber et al. (39) | 51 FEP. 2-years, with assessments at baseline, 6, 15, and 24 months | Predictive value of neurocognitive domains and clinical variables for symptomatic and functional outcome and clinical recovery | Appropriate and reproductible statistics: Binary logistic regression analysis with forward selection (likelihood ratio) was applied, with recovery as the dependent variable | Processing speed and working memory predicted recovery (recovery defined as the combination of both symptomatic and social functioning remission) |
| Bodén et al. (40) | 46 FEP. 5-years | The predictive value of a battery of cognitive tests that assess global cognitive function, psychomotor function, processing speed, and verbal learning | Appropriate and reproductible statistics: logistic regression models with the Z-scores from the cognitive test variables as independent variables (two models: unadjusted and adjusted for antipsychotics) | Low psychomotor speed was associated with poor social functioning (OR = 3.37 adjusted for antipsychotic drug use) |
| Jordan et al. (41) | 267 FEP (15-35). 2-years | Cognition and symptom remission in functional recovery (social and occupational functioning) | A series of stepwise regression models in consecutive blocks | Verbal memory was a significant predictor on functionality, but explained only 9% of the variance. Adding consecutive months in symptom remission explained 30% |
| Ayesa-Arriola et al. (42) | 202 FEP patients in subgroups with (114) or without functional disability. 1 and 3-years | Cognition, premorbid social adjustment and real-world functional disability prediction | Covered comparability among groups. Appropriate and reproductible statistics: Different logistic regression models with Nagelkerke’s R2 | At the end of the 3-year study, premorbid social adjustment (OR = 1.628) and years of education (OR = 1.117) were significant predictors of functional disability. Processing speed (OR = 1.575) and overall general cognitive functioning (OR = 1.665) were predictors at 1 year but not at the end of the 3-year study |
| Fu et al. (43) | 24 FEP. 6 follow-ups over 4-years | Level of cognitive alterations and exploration of the functional trajectory of the disease. To stratify the patient group for identifying a poor outcome group early in the course of illness | Appropriate and reproductible statistics: A series of multilevel growth curve models fitted using maximum likelihood and an unstructured covariance structure | When a time × baseline interaction was tested, social functioning was significantly predicted by attention, verbal learning, and working memory. Role functioning was significantly predicted by attention, working memory, and reasoning/problem solving. Education level at baseline was significantly associated with role functioning. The other cognitive domains of the MATRICS did not significantly predict functional outcome |
| Matsuda et al. (44) | 26 FEP. 5-years. Assessments every 6-months | Course of neurocognitive deficits. Find whether there is a generalized or specific domain of neurocognitive deficits that endure, improve, or deteriorate | Statistical analyses with low power to address predictions: t-test, ANOVA, Pearson correlations | Some cognitive domains worsen at 5 years (verbal memory and executive functions), but there was no correlation between any of them with social functioning |
| Peña et al. (45) | 95 FEP. 2-years | Cognition and other variables as predictors of functional outcome in schizophrenia and non-schizophrenia syndromes after a FEP | Appropriate and reproductible statistics: multiple stepwise regressions (forward) | Worse visuospatial processing at baseline predicted poorer functional outcome for pooled FEP patients. Processing speed predicted most functional outcome measures in schizophrenia, whereas visuospatial functioning was the only significant predictor of functional outcomes in the non-schizophrenia group |
| González-Ortega et al. (46) | 282 FEP. 2-years | To analyze the influence of social cognition as a mediator between cognitive reserve (IQ plus occupational attainment) and cognitive domains on functioning in FEP both at baseline and at 2 years | Appropriate and reproductible statistics: A three-steps path analysis for testing the mediating model, with linear regressions adjusted for potential confounders (stepwise and bootstrap) | In a first path, cognitive reserve and verbal memory were significantly related to functioning. A second path identified social cognition acting as a mediator between cognitive reserve and functioning, and between verbal memory and functioning, both of them at follow-up (not at baseline) |
| Oomen et al. (47) | 204 FEP and 40 controls. 1-year | Cognitive subtypes, characterized by differences in both clinical and functional outcomes at baseline and follow-up | Appropriate and reproductible statistics: A hierarchical clustering approach (HCA) with Ward method and K-means | General functioning (GAF) in the severely impaired cluster was significantly lower than in those with preserved cognition at baseline and showed trend-level effects at 6- and 12-month follow-up |
| Sullivan et al. (48) | 54 FEP. 1-year | To investigate the longitudinal association between both theory of mind and psychotic symptoms and social functioning outcome | Appropriate and reproductible statistics: Random effects multivariate linear regression models | Theory of mind was stable over time. Theory of mind at baseline was not correlated (therefore it was not considered a predictor) with social functioning at 6 and 12 months |
| Wright et al. (49) | 80 FEP evaluated but only 26 followed 3-years | Whether neurocognition, functional capacity and metacognitive ability predicted functional outcome | A series of single regression exploratory analyses and a stepwise regression model in consecutive blocks controlling for small size of the sample | Metacognitive ability was a significant predictor of change in functional outcome from baseline to follow-up (72% variance explained). Negative symptoms did not change the model |