Introduction
Bipolar disorder (BD) is a mood disorder that presents with alternating periods of depressive or manic symptoms. In addition, cognitive impairment is present in one-third of BD patients even during euthymia, thereby decreasing their quality of life [Reference Aparicio, Sánchez-Morla, Santos and Mateo1–Reference Samamé, Martino and Strejilevich3]. Both genetic [Reference Scott and McClung4] and environmental [Reference Bortolato, Köhler, Evangelou, León-Caballero, Solmi and Stubbs5] factors have been identified as causes of BD.
Childhood maltreatment (CM) represents one of the most relevant environmental factors involved in BD development [Reference Rokita, Dauvermann and Donohoe6–Reference Fares-Otero and Seedat8]. These CM experiences are known to modulate the response to stress [Reference Davies, Thompson, Hentges, Parry and Sturge-Apple9–Reference Montoro, Moreno, Mulet, Miró, Leunda and Llaurador-Coll12] and, in turn, are linked to microanatomical changes in the brain [Reference McLaughlin, Weissman and Bitrán13]. Specifically, impacts in the hippocampus and amygdala, two regions involved in mood disorders, were differentially altered between patients with BD and nonaffected subjects [Reference Janiri, Sani, De Rossi, Piras, Banaj and Ciullo14, Reference Janiri, Sani, De, Piras, Iorio and Banaj15]. Social cognition is a multidimensional cognitive domain that involves complex processes that are used in socially relevant situations [Reference Harvey and Penn16] and includes emotional intelligence (EI). [Reference Green, Penn, Bentall, Carpenter, Gaebel and Gur17]. EI is defined as the ability to understand how we feel as people, alongside the ability to understand the emotional language of others and the ability to use our emotions appropriately in the social context in light of the specific situation and environment at hand [Reference Salovey and Mayer18]. Mayer and colleagues developed an instrument to measure EI known as the Mayer–Salovey–Caruso Emotional Intelligence Test (MSCEIT) [Reference Mayer, Salovey and Caruso19], which has good reliability in both clinical and nonclinical populations [Reference Aparicio, Santos, Jiménez-López, Bagney, Rodríguez-Jiménez and Sánchez-Morla2]. Impairments in EI assessed by means of the MSCEIT have been observed in patients with BD [Reference Lahera, Ruiz-Murugarren, Iglesias, Ruiz-Bennasar, Herrería and Montes20–Reference Varo, Jiménez, Solé, Bonnín, Torrent and Lahera23]. Recent studies have demonstrated a direct relationship between EI and psychosocial functioning (PSF) in euthymic BD patients [Reference Aparicio, Sánchez-Morla, Santos and Mateo1, Reference Aparicio, Santos, Jiménez-López, Bagney, Rodríguez-Jiménez and Sánchez-Morla2, Reference Varo, Jimenez, Solé, Bonnín, Torrent and Valls24]. Apparently, this relationship is not mediated by general cognitive function [Reference Ospina, Nitzburg, Shanahan, Perez-Rodriguez, Larsen and Latifoglu25].
Other studies have explored the impact of CM on social cognition and PSF in BD patients. Facial emotion recognition has been shown to be impaired in patients with BD who have experienced CM [Reference Catalan, Díaz, Angosto, Zamalloa, Martínez and Guede26–Reference Yenilmez, Atagün, Altun, Tunç, Uzgel and Altinbaş28]. Interestingly, Morán-Kneer and colleagues [Reference Moran-Kneer, Rios, Costa Cordella, Barría, Carvajal and Valenzuela29] reported that the relationship between CM and social cognition in BD patients was mediated by the “attachment anxiety” relationship style. Despite these results, two recent meta-analyses in patients with BD did not reveal conclusive evidence of associations between CM and social cognition [Reference Fares-Otero, Pérez-Ramos, Lopez-Escribano, Martin-Parra, Alameda and Halligan30, Reference Fares-Otero, De Prisco, Oliva, Radua, Halligan and Vieta31]. To our knowledge, no study has addressed the possible relationship between CM and PSF via social cognition, particularly EI, in euthymic patients with BD. If such a mediation was identified, it would open new therapeutic strategies.
In this context, we explored the relationships among CM, EI and PSF in adult euthymic patients with BD and matched healthy controls (HCs). We compared both groups of subjects and investigated whether EI served as a mediator of the impact of CM on PSF.
Methods
Subjects
A total of 248 adult participants (124 euthymic BD patients and 124 matched HCs) from a larger sample [Reference Garcia-Ruiz, de Moura, Muntané, Martorell, Bosch and Esteller32] were included in this study. The inclusion and exclusion criteria, including the criterion of euthymia, have been previously described [Reference Garcia-Ruiz, de Moura, Muntané, Martorell, Bosch and Esteller32]. All participants provided written informed consent, and the study was approved by all the local ethics committees.
Assessment
The participants’ assessments have been previously described [Reference Garcia-Ruiz, de Moura, Muntané, Martorell, Bosch and Esteller32]. Briefly, manic symptoms were assessed through the Young Mania Rating Scale (YMRS) [Reference Young, Biggs, Ziegler and Meyer33], depressive symptoms were identified via the Hamilton Depression Rating Scale (HDRS) [Reference Hamilton34], and the estimated premorbid intelligence quotient (IQ) was determined using the vocabulary subtest of the Wechsler Adult Intelligence Scale-Third Edition (WAIS-III) [Reference Wechsler35].
CM was evaluated via the Spanish validated version [Reference Hernandez, Gallardo-Pujol, Pereda, Arntz, Bernstein and Gaviria36] of the Childhood Trauma Questionnaire-Short Form (CTQ-SF), which is a self-report questionnaire containing 25 items to explore the presence of different types of maltreatment experienced during childhood (emotional abuse, sexual abuse, physical abuse, emotional neglect, and physical neglect) plus 3 items to detect response bias that minimizes the extent of childhood trauma experienced. From the 3 minimization/denial items, we detected 17 subjects; however, only 3 of these subjects obtained scores for abuse in one dimension, and they were not removed from the study. The total score ranges from 25 to 125, in which higher scores indicate more severe CM. A detailed description of the four levels of CM (no abuse, mild abuse, moderate abuse, and severe or high abuse) was described by Bernstein and colleagues [Reference Bernstein and Fink37]. CM can display a U-shaped graph when plotted against stress responses [Reference Shakiba, Ellis, Bush and Boyce10, Reference Benjamini, Drai, Elmer, Kafkafi and Golani38], and nonlinear models or the conversion of a continuous variable into a categorical one are needed. Therefore, to maximize statistical power, we created the variable “gradual CM” and used the following three categories: no-CM, low/moderate-CM (including both mild and moderate abuse from the original scale), and high-CM.
EI was evaluated via the Spanish version [Reference Extremera, Fernández-Berrocal and Salovey39] of the MSCEIT, which is divided into two areas: the experiential area, with Branch 1 involving perceiving emotions and Branch 2 involving using emotions, and the strategic area, with Branch 3 involving understanding emotions and Branch 4 involving managing emotions. The average score for this assessment method is 100, with a standard deviation of 15 [Reference Extremera, Fernández-Berrocal and Salovey39]; moreover, lower scores indicate poorer performance in terms of EI.
PSF was assessed via the Functioning Assessment Short Test (FAST) [Reference Rosa, Sánchez-Moreno, Martínez-Aran, Salamero, Torrent and Reinares40], which is an interviewer-administered test that evaluates 6 functional domains (autonomy, occupational functioning, cognitive functioning, financial issues, interpersonal relationships, and leisure time). In this context, higher scores indicate a greater degree of functional impairment.
Statistical analysis
Descriptive analyses were conducted to compare sociodemographic and clinical characteristics according to diagnostic group, in which context chi-square tests were used for categorical variables, and t tests and analyses of variance (ANOVAs) were used for continuous variables. To identify the variables influencing the outcome (EI and PSF), bivariate test analyses (including Pearson’s correlations) were performed to identify which variables were significantly correlated. These variables (including center, age, diagnosis, IQ, education level, YMRS score, and HDRS score) were included as independent variables in the regression models. Linear regression models were used to analyze the relationships among gradual CM, EI, and PSF, in which context EI was the dependent variable and center, age, diagnosis, IQ, education level, FAST score, YMRS score, and HDRS score were independent variables. To explore the relationships among these variables in the patient group in further depth, similar models were used to stratify the sample by diagnosis. The η² (calculated through one-factor ANOVA) was used to measure the variance of the dependent variable (MSCEIT) explained by the independent variables and subsequently to estimate the sample size to have enough power. Finally, we performed a mediation analysis to assess whether EI mediated the effect of CM on PSF (Supplementary Figure S1) by using the multivariate additive regression tree approach implemented via the R package “mma” [Reference Yu and Li41]. The gradual CM category was used as a predictor; the MSCEIT (MSCEIT total score including each area and each branch) score was used as a mediator, and the FAST score was used as the outcome. Center, age, diagnosis, IQ, education level, YMRS score, and HDRS score were included as covariates. Diagnosis was not included in the subgroup analyses (patients and HCs). Only potential mediators with marginal contributions to the indirect effect (p < 0.1) were evaluated. Total, direct, and indirect effects were calculated on the basis of 1000 bootstrapped samples. Outputs are provided as the logarithm of the odds ratio (logOR) with a 95% CI, and p < 0.05 was considered to indicate statistical significance. To provide a more detailed explanation of the contribution of CM and EI to PSF, the coefficients of the linear regression models that explain the mediating effect were also calculated. Furthermore, a mediation analysis was performed using CM (CTQ total score) as a continuous variable.
We used IBM SPSS Statistics software version 22 (IBM, USA) to perform all of the statistical analyses, except for the mediation analyses and the generation of the corplots, which were conducted in R. Significance was established at p < 0.05, and the Benjamini–Hochberg method [Reference Benjamini, Drai, Elmer, Kafkafi and Golani38] was applied to correct for multiple tests when necessary.
Results
Sample description
Table 1 presents the descriptive statistics and comparisons between patients with BD and HCs. Patients with BD had significantly higher CTQ total scores than HCs did (Table 1). Grouping by CM level revealed that 23.4% of the patient group was classified as high-CM (versus 8.9% of HCs, p = 0.002) (Table 1). Compared with HCs, patients with BD exhibited worse EI performance, although within the “normality” cutoff. The mean scores for the patient group were significantly lower with respect to the MSCEIT total score and the strategic area score, as well as within this area, including both the understanding emotions and managing emotions branches (Table 1). Compared with HCs, patients with BD had greater functional impairment (with higher total FAST and all subdomain scores above the “impaired” threshold).
Sample description stratifying according to diagnostic group
Table 1 Long description
The table is organized into four columns: Variables, B D patients (N = 124), Healthy controls (N = 124), and p-value.
Sociodemographic Variables:
* Female (%): 58.1 for B D, 58.9 for controls (n.s.).
* Age (mean plus or minus S D): 45.7 plus or minus 8.3 for B D, 45.0 plus or minus 8.3 for controls (n.s.).
* Cohabiting (%): 33.1 for B D, 63.4 for controls (p = 2.1 times 10 super -6).
* Working (%): 55.6 for B D, 87.0 for controls (p = 5.3 times 10 super -8).
* Estimated I Q (W A I S): 106.7 for B D, 110.2 for controls (p = 0.007).
Clinical Variables:
* Manic symptoms (Y M R S): 0.9 for B D, 0.8 for controls (n.s.).
* Depressive symptoms (H D R S): 2.4 for B D, 1.6 for controls (p = 0.002).
* B D specific data: Age at onset 26.6, Illness duration 18.8 years, B D type I 71.8%, Lithium treatment 53.7%.
Childhood maltreatment (C M):
* C T Q score: 38.8 for B D, 32.2 for controls (p = 1.2 times 10 super -5).
* High C M (%): 23.4 for B D, 8.9 for controls (p = 0.002).
Emotional intelligence (M S C E I T):
* Total score: 110.8 for B D, 115.7 for controls (p = 0.024).
* Strategic area: 100.1 for B D, 105.7 for controls (p = 0.001).
Psychosocial functioning (F A S T):
* Total score: 15.5 for B D, 6.4 for controls (p = 5.2 times 10 super -12).
* Occupational functioning: 6.1 for B D, 1.8 for controls (p = 2.2 times 10 super -10).
* Interpersonal relationship: 2.8 for B D, 1.2 for controls (p = 1.1 times 10 super -6).
Abbreviations: AP, antipsychotic dose; BD, bipolar disorder; CM, childhood maltreatment; CTQ, Childhood Trauma Questionnaire; FAST, Functioning Assessment Short Test, HDRS, Hamilton Depression Rating Scale; IQ, intelligence quotient; MSCEIT, Mayor-Salovey-Caruso Emotional Intelligence Test; SD, standard deviation; WAIS, Wechsler Adult Intelligence Scale; YMRS, Young Mania Rating Scale.
a Chi squared test was used for categorical variables and T-test for continuous variables as described in Materials and Methods section.
b AP medication dose in olanzapine mg equivalents as described in Materials and Methods section.
c Gradual CM, was recoded from the CTQ as described in Materials and Methods section.
Relationships among CM, EI and psychosocial functioning
The correlation plots in Figure 1 (Panel A) show that the CTQ total scores were positively correlated with age and the YMRS, HDRS and FAST total scores and were negatively correlated with MSCEIT total scores in the whole sample. However, when stratified by group (HCs and patients with BD), the inverse correlation between CM and EI was significant only in BD patients. Moreover, a higher reported CM for the subject corresponds to a higher clinical symptom score (YMRS and HDRS) and decreased EI and PSF.
Relationships among CM, EI, and psychosocial functioning. (A) Corplots for the whole sample, the healthy control group, and the bipolar disorder group. Pearson correlation tests were performed as described in the Materials and Methods section. *AdjP<0.01, **AdjP<0.001 and ***AdjP<0.0001. (B) Total EI, branches, and areas according to gradual CM. ANOVA was performed, and none of the comparisons among the CM categories were statistically significant. (C) Total psychosocial functioning (FAST) according to gradual CM. ANOVA was performed, and none of the comparisons among the CM categories were statistically significant.
Figure 1 Long description
Panel A contains three triangular correlation matrices labeled Whole sample, H C s, and B D. The vertical axes list Age, I Q, Y M R S, H D R S, M S C E I T, F A S T, and C T Q. The horizontal axes mirror these labels. A color scale at the bottom ranges from dark red negative 0.6 to dark blue positive 0.6. Blue circles indicate positive correlations, and red circles indicate negative correlations, with asterisks denoting significance levels.
Panel B shows a hierarchical breakdown of M S C E I T scores. On the far left, a Total score line graph shows scores for H C in black and B D in blue across three categories on the x-axis: N, L M, and H. This branches into two Area graphs: Experiential and Strategic. These further branch into four Branch graphs: Perceiving emotions, Using emotions, Understanding emotions, and Managing emotions. All graphs show Score on the y-axis from 90 to 130.
Panel C is a single line graph titled F A S T Total. The y-axis shows Score from 0 to 40. The x-axis shows categories N, L M, and H. The H C group is represented by a black line with circular markers, and the B D group is represented by a blue line with square markers. Error bars are present for all data points in panels B and C.
Because a curvilinear U-shaped relationship between CM and stress responses has been reported in the literature [Reference Shakiba, Ellis, Bush and Boyce10, Reference Boyce and Ellis42], we used the gradual CM variable to plot the outcome variables (i.e., EI and social PSF) against CM. The plots of the EI total, areas, and branches against CM and the corresponding stratification by group are shown in Figure 1 (Panel B). The plot of the FAST scores against CM is shown in Figure 1 (Panel C). Although the BD patients tended to exhibit a U-shaped curve, the comparison of the means (ANOVA) for each CM exposure grade (no-CM, low/moderate-CM, and high-CM) did not reveal a statistically significant result.
Table 2 shows the linear regression analysis results for each statistically significant model and presents those dependent variables that were significant in the equations. Notably, in all of the models, the low/moderate-CM and high-CM categories were used as the CM exposure variables, whereas the category “no-CM” was used as the reference. With respect to the whole sample, in Model I (where total MSCEIT was used as the dependent variable), age was inversely associated with EI (β = −0.38) and was directly associated with IQ (β = 0.56). In Model II, which explored the strategic area and explained 21% of the EI variability, EI was inversely associated with age, PSF, and low/moderate-CM (β = −0.28, β = −0.21 and β = −3.64, respectively) and was directly associated with IQ (β = 0.50). Model III revealed associations between the understanding emotions branch and age (β = −0.35), IQ (β = 0.45), and PSF (β = −0.22). In Model IV (in which the managing emotions branch was assessed), IQ (β = 0.37) was the only in dependent variable that exhibited a direct relationship. The splitting of the sample according to the diagnostic group and the construction of the same models revealed that the impact of gradual CM on EI is driven mainly by the BD patient group (Models VII and IX). Notably, the relationships between CM and the MSCEIT total score (Model VII) were negatively related to age and low/moderate-CM (β = −0.50 and β = −9.95, respectively) and were directly related to IQ (β = 0.66). In Model VIII, the MSCEIT strategic area score was negatively associated with age (β = −0.42), PSF (β = −0.24), and low/moderate-CM (β = −5.55) and was positively associated with IQ (β = 0.66). Model IX explored the MSCEIT understanding emotions branch and revealed indirect associations between EI and both age (β = −0.57) and PSF (β = −0.22), as well as a direct association with IQ (β = 0.56). Notably, Models VIII and IX explained the greatest percentage of the EI variability (27% and 29%, respectively). In the HCs (Models V and VI), the strategic area and understanding emotions branch were associated with only IQ (β = 0.40 and β = 0.39, respectively). Notably, the relationship between CM and the category of high-CM was not statistically significant, although a trend was observed, indicating that participants with high-CM exhibited EI scores similar to those of participants with no-CM. The effects of CM on the MSCEIT total score were small (η2 = 0.024, 95% CI 0.00–0.069 and η2 = 0.025, 95% CI 0.00–0.080 when low/moderate-CM was compared to no-CM and all three groups, respectively), and the statistical power of the present analyses was moderate (approximately 50%). In addition, these models demonstrate that, in our sample, patients with BD with low/moderate-CM exhibited worse EI performance than patients with no-CM or high-CM did. These results are in line with the U-shaped plots shown in Figure 1 (Panel B).
Linear regression models assessing the relationship between CM, EI, and psychosocial functioning
Table 2 Long description
The table is organized into seven columns: Model, M S C E I T, Variables into the equation, Variable statistics (beta and p), and Model statistics (Adjusted R super 2 and p).
1. Whole sample group:
- Model I (Total M S C E I T): Significant variables are Age (beta -0.38) and I Q (beta 0.56). Model Adjusted R super 2 is 0.09.
- Model II (Strategic area): Significant variables include Age, I Q, Psychosocial functioning, and Low/moderate-C M. Model Adjusted R super 2 is 0.21.
- Model III (Understanding emotions branch): Significant variables are Age, I Q, and Psychosocial functioning. Model Adjusted R super 2 is 0.18.
- Model IV (Managing emotions branch): Only I Q is significant (beta 0.37). Model Adjusted R super 2 is 0.1.
2. Healthy controls group:
- Model V (Strategic area): Significant variable is I Q (beta 0.40). Model Adjusted R super 2 is 0.10.
- Model VI (Understanding emotions branch): Significant variable is I Q (beta 0.39). Model Adjusted R super 2 is 0.07.
3. Bipolar disorder patients group:
- Model VII (Total M S C E I T): Significant variables are Age, I Q, and Low/moderate-C M (beta -9.95). Model Adjusted R super 2 is 0.11.
- Model VIII (Strategic area): Significant variables are Age, I Q, Psychosocial functioning, and Low/moderate-C M. Model Adjusted R super 2 is 0.27.
- Model IX (Understanding emotions branch): Significant variables are Age, I Q, and Psychosocial functioning. Model Adjusted R super 2 is 0.29.
Abbreviations: CM, childhood maltreatment; EI, emotional intelligence; FAST, Functioning Assessment Short Test; HDRS, Hamilton Depression Rating Scale; IQ, intelligence quotient; MSCEIT, Mayer–Salovey–Caruso Emotional Intelligence Test; WAIS, Wechsler Adult Intelligence Scale; YMRS, Young Mania Rating Scale.
Note: The dependent variables in each model were Total, Areas, or branches MSCEIT scores. Gradual CM, diagnostic, center, education level, IQ (WAIS), age, YMRS, HDRS and FAST were independent variables in models I–IV. In models V–IX, the diagnostic variable was not included into the equation. Only models and variables in each model showing p < 0.05 are shown.
Mediation of CM between EI and psychosocial functioning
For the whole sample, the mediation analysis revealed the MSCEIT strategic area score and understanding emotions score as potential mediators between CM and the FAST total variable. However, when the no-CM and low/moderate-CM groups were compared, neither variable showed significant indirect effects (Supplementary Table S1). Similarly, when the no-CM and high-CM groups were compared, the indirect effects remained nonsignificant for both the strategic area and understanding emotions (Supplementary Table S1). In the HC group, the analysis did not identify any possible mediators, and in the BD group, the strategic area and understanding emotions were identified as possible mediators. However, the next step in the analysis did not show any significant indirect effect between the no-CM group and the low/moderate-CM group (Supplementary Table S1). A comparison of the no-CM and high-CM groups revealed that the indirect effect remained nonsignificant for both the strategic area and understanding emotions (Supplementary Table S1). Therefore, EI cannot be viewed as a mediator between CM and PSF. These analyses were also performed using the total CTQ score as a continuous variable (Supplementary Table S2), and the results were similar.
Discussion
Our results revealed that compared with HCs, the euthymic BD group presented higher rates and more severe CM, which is consistent with our previous research in this sample using different instruments [Reference Montoro, Moreno, Mulet, Miró, Leunda and Llaurador-Coll12, Reference Garcia-Ruiz, Jiménez, Aranda, Verdolini, Gutiérrez-Zotes and Sáez43], as well as with findings reported by other researchers [Reference Palmier-Claus, Berry, Bucci, Mansell and Varese44–Reference Robinson, Ploner, Leone, Lichtenstein, Kendler and Bergen47]. We also observed that the BD group had lower EI scores than the HC group did, similar to previous case–control studies in euthymic BD patients. These studies reported that patients had lower total MSCEIT scores and lower scores in both areas and branches, except for the perceiving emotions branch [Reference Aparicio, Santos, Jiménez-López, Bagney, Rodríguez-Jiménez and Sánchez-Morla2, Reference Frajo-Apor, Kemmler, Pardeller, Huber, Macina and Welte48]. Other authors with similar study designs reported only differences in the understanding of emotions branch [Reference Varo, Jiménez, Solé, Bonnín, Torrent and Lahera23, Reference Varo, Amoretti, Sparacino, Jiménez, Solé and Bonnin49, Reference Liu, Tseng, Chang, Chang, Yang and Chen50] or both branches of the strategic area [Reference Yingying, Bo and Jun51]. Taken together, these findings indicate that the strategic area is the most impaired in patients with BD. However, some studies have reported differences only in total MSCEIT [Reference Varo, Jiménez, Solé, Bonnín, Torrent and Lahera23, Reference Varo, Amoretti, Sparacino, Jiménez, Solé and Bonnin49, Reference Yingying, Bo and Jun51]. With respect to PSF, our results are also in line with those reporting worse social functioning among BD patients than among HCs [Reference Varo, Solé, Jiménez, Bonnín, Torrent and Valls22, Reference Varo, Jiménez, Solé, Bonnín, Torrent and Lahera23, Reference Jiménez-López, Villanueva-Romero, Sánchez-Morla, Martínez-Vizcaíno, Ortiz and Rodriguez-Jimenez52–Reference Miskowiak, Hansen, Mariegaard and Kessing55]. The most interesting result of our study is that low/moderate-CM (in conjunction with age and PSF) is inversely associated with EI, specifically with the understanding emotions branch. In other words, older patients with less PSF and low/moderate-CM are those individuals who exhibit the worst EI. Inverse associations of EI with age and PSF have been previously reported among BD patients [Reference Aparicio, Santos, Jiménez-López, Bagney, Rodríguez-Jiménez and Sánchez-Morla2]. However, our original research contribution lies in the finding indicating that BD patients in the low/moderate-CM category (and not those individuals in the no-CM or high-CM categories) demonstrated the worst EI. This result could be viewed as counterintuitive since the generally accepted hypothesis suggests that CM is associated with poorer social cognition in individuals with BD [Reference Rokita, Dauvermann and Donohoe6] and psychosis [Reference Rodriguez, Aas, Vorontsova, Trotta, Gadelrab and Rooprai56]. However, this effect could be explained by the “biological sensitivity to context” theory, which states that, compared with children who are exposed to moderate adversity, children who are exposed to both protective environments (with no adversity) and highly stressful conditions are more prone to develop a stress-reactive profile that allows them to adapt to environmental stressors later in life [Reference Boyce and Ellis42]. Evidence obtained mainly from general population studies has revealed that compared with children who are not exposed or who have negligible exposure, children and adolescents who are exposed to mild levels of adversity exhibit decreased stress reactivity [Reference Davies, Thompson, Hentges, Parry and Sturge-Apple9, Reference Boyce and Ellis42]. This result (which was initially considered a negative result) was thoroughly examined from an evolutionary perspective by Boyce and Ellis, who reported evidence to support an adaptive response [Reference Boyce and Ellis42]. Interestingly, studies involving the personality and behavioral characteristics of individuals scoring high in “reactivity to stress” have demonstrated that these individuals tend to be more attentive, discriminating, and reflective [Reference Aron and Aron57]. These observations are in line with our finding that subjects with BD with no- or high-CM obtained better understanding emotions scores, which implies an understanding of the meaning of emotions and how to manage both their own emotions and those of others. This scenario also involves being aware of and using emotions to solve problems in a reflective manner. All of these functions correspond to those that are considered high-level cognitive functions [Reference Mortillaro and Schlegel58], and those functions in BD patients are altered and modulated by emotional dysregulation [Reference Lima, Peckham and Johnson59]. This result was not observed in the HC group. Interestingly, neuroimaging studies in subjects with BD exploring the impact of CM on brain structures have consistently revealed changes in the hippocampus and amygdala [Reference McLaughlin, Weissman and Bitrán13–Reference Janiri, Sani, De, Piras, Iorio and Banaj15], which are key regions involved in emotional regulation.
Research on the relationship between CM and social cognition remains limited [Reference Fares-Otero, Pérez-Ramos, Lopez-Escribano, Martin-Parra, Alameda and Halligan30, Reference Fares-Otero, De Prisco, Oliva, Radua, Halligan and Vieta31]. A recent study in healthy adults reported a U-shaped association between CM and performance on the “Reading the Mind in the Eyes” task [Reference Pilot, Daniels, Iron Rope, Pfingston, Powless and Bouvier60]. However, to our knowledge, no similar results have been published for BD. A linear and simple relationship between CM and EI in BD patients has been observed; however, the results reported by different studies are controversial. For example, when the “hinting task” was used in euthymic BD patients, Morán-Kneer and colleagues reported that higher scores on CM were associated with better social cognition performance [Reference Moran-Kneer, Rios, Costa Cordella, Barría, Carvajal and Valenzuela29]. Conversely, Russo and colleagues reported that patients with BD who were exposed to at least one type of maltreatment performed worse in the “emotion recognition task” [Reference Russo, Mahon, Shanahan, Solon, Ramjas and Turpin27]. A recent study that used the “Dokuz Eylul theory of mind index” reported that euthymic BD subjects who were exposed to more maltreatment experiences performed worse on social cognition tests [Reference Takım, Sarı and Gokcay61]. Conversely, other authors using the MSCEIT have reported no significant association between CM and emotion processing, including EI, in BD [Reference Fares-Otero, Pérez-Ramos, Lopez-Escribano, Martin-Parra, Alameda and Halligan30]. These inconclusive findings may stem from heterogeneity in both the assessment used for CM and social cognition, making it difficult to draw definitive conclusions.
Depressive symptoms, even at subsyndromic presentation, may impact EI [Reference Lahera, Ruiz-Murugarren, Iglesias, Ruiz-Bennasar, Herrería and Montes20, Reference Ospina, Nitzburg, Shanahan, Perez-Rodriguez, Larsen and Latifoglu25, Reference Navarra-Ventura, Vicent-Gil, Serra-Blasco, Massons, Crosas and Cobo62–Reference Lahera, Herrera, Reinares, Benito, Rullas and González-Cases64]. Notably, although our participants were euthymic at the time of the assessment, YMRS and HDRS scores were included as covariates in the regression models; however, neither was significantly associated with MSCEIT performance. Furthermore, the persistence of social cognition deficits during remission phases of the disease highlights the need for interventions aimed at reversing or minimizing these symptoms.
CM has been defined as the greatest preventable cause of mental illness; moreover, despite calls from experts for changes in public policy [Reference Lieberman, Chu, Van Horn and Harris65], no generalized programs currently exist to prevent CM in children. With respect to BD, a systematic review revealed that a history of traumatic events was associated with earlier age of onset, longer episode duration, greater number of mood episodes, and greater likelihood of experiencing psychotic features and suicide attempts [Reference Rowe, Perich and Meade66]. Furthermore, the implementation of therapeutic plans to improve social skills and PSF after early detection of CM seems to be necessary to improve the lifetime quality of life of patients. Our results emphasize the need to carefully evaluate CM history, as low or moderate levels were associated with poorer EI and PSF, suggesting that mild or intermittent exposure can cause long-lasting alterations. Patients with BD who experienced CM and had poorer EI and PSF may constitute a vulnerable clinical subgroup that needs to be helped with social skills, problem-solving, and emotional management interventions. Finally, early detection of CM will aid in the planning of early targeted interventions for emotional and stress regulation to prevent functional deficits in adulthood. Given that CM impacts brain areas involved in emotion regulation, targeted psychological interventions could be prescribed on a more regular basis to children and adolescents who suffer from CM. A recently published trial comparing a behavioral intervention targeting a broad set of resilience factors against usual care demonstrated that a positive appraisal style (the tendency to evaluate stressors in a realistic to slightly unrealistically positive fashion) is modifiable, improving psychological resilience [Reference Petri-Romão, Mediavilla, Restrepo-Henao, Puhlmann, Zerban and Ahrens67]. More specifically, social cognition could be used as a target for specific treatments aimed at improving PSF [Reference Aparicio, Santos, Jiménez-López, Bagney, Rodríguez-Jiménez and Sánchez-Morla2, Reference Frajo-Apor, Pardeller, Kemmler, Mühlbacher, Welte and Hörtnagl68]. To our knowledge, no previous study has specifically reported treatments aimed at improving social cognition in BD patients with a history of CM; however, such studies have been conducted in the general population [Reference Canady, Zullig, Brumage and Goerling69, Reference Mattingly and Kraiger70] and in other psychiatric conditions. A recent review by Razavi and colleagues reported the effectiveness of cognitive interventions (including social cognition and rehabilitation) in BD patients [Reference Razavi, Fathi, Vahednia, Ardani, Honari and Akbarzadeh71].
Our findings revealed that EI did not mediate the relationship between CM and PSF. Although no a priori power analysis was performed, the study was explicitly exploratory in nature and aimed to identify patterns and estimate effect sizes rather than to test confirmatory hypotheses with high statistical power. The effects observed for the MSCEIT total score variable were small; consequently, the statistical power of the present analyses was moderate (approximately 50%) for detecting effects of this magnitude. To achieve greater statistical power (80%), we would need a sample size of 300–450 participants. This limitation may have reduced the ability to detect indirect effects, such as the mediation of EI between CM and PSF; thus, nonsignificant mediation results should be interpreted with caution. For this reason, we emphasize that new studies with larger sample sizes are needed. Despite this limitation, our findings align with previous research in nonaffective psychotic disorders, where CM was found to be significantly related to social cognition deficits but not through a mediating effect of EI [Reference Weijers, Fonagy, Eurelings-Bontekoe, Termorshuizen, Viechtbauer and Selten72]. In BD, nonlinear effects of CM on brain structures have been described [Reference Janiri, Sani, De Rossi, Piras, Banaj and Ciullo14, Reference Janiri, Sani, De, Piras, Iorio and Banaj15], a result that indirectly supports our findings. Overall, these results indicate that the relationships among CM, EI, and PSF are complex and require further exploration. The mechanism underlying the ability of CM to impact children’s cognition and behavior has been extensively studied [Reference Davies, Thompson, Hentges, Parry and Sturge-Apple9, Reference Lieberman, Chu, Van Horn and Harris65], as has the association between EI and PSF [Reference Aparicio, Santos, Jiménez-López, Bagney, Rodríguez-Jiménez and Sánchez-Morla2, Reference Ospina, Nitzburg, Shanahan, Perez-Rodriguez, Larsen and Latifoglu25]. However, the links among these three variables also require further investigation.
Our study has other limitations that should be noted. First, the CTQ is self-administered and retrospective, and responses to this measure may be influenced by the presence of a psychiatric disorder. Specifically, the person may have perceived and described situations from the past (with a subjectivity effect of the present moment) in an unintentional but biased manner [Reference Danese and Widom73]. In addition, the CTQ does not consider the multiplicity and chronicity of abuse, which are factors that can be measured via other instruments [Reference Teicher and Parigger74]. Second, although the MSCEIT has been recognized for its good psychometric capabilities, some criticisms have been reported, such as its long administration time [Reference Kuo, Chang, Wang, Tseng, Huang and Chen75], cross-cultural validity with respect to the possibility of comparing the results obtained in different countries in which the test has been validated (when considering the cultural differences of each location) [Reference Hellemann, Green, Kern, Sitarenios and Nuechterlein76], and low discriminant validity [Reference Fiori, Antonietti, Mikolajczak, Luminet, Hansenne and Rossier77]. Third, although psychotropic drug treatment was included in the covariates, we must consider the fact that psychotropic drugs can affect emotions, which, in turn, may influence the completion of tests at the individual level [Reference Szmulewicz, Samamé, Caravotta, Martino, Igoa and Hidalgo-Mazzei21]. Finally, owing to the limited size of the sample, stratification by sex was not possible [Reference De Prisco and Vieta78]. Conversely, our study has several strengths, such as the exhaustive inclusion and exclusion criteria used in this research, the comprehensive assessment of diagnostic criteria and symptomatology, and the fact that patients and controls were euthymic and matched in terms of age, sex, and education level, which are, in turn, three variables usually associated with EI.
In summary, our results demonstrate that the presence of a history of CM in patients with BD is associated with EI performance and PSF; however, no mediating effects were identified. Although limited by the sample size, our results highlight the importance of identifying CM history in the clinical setting to specifically target psychological therapies.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1192/j.eurpsy.2026.12251.
Data availability statement
The dataset used in this study will be available upon request from the corresponding author.
Financial support
This study was supported by grants from the Brain and Behavior Research Foundation (2017 NARSAD Independent grant, #25811 to El. V.), Instituto de Salud Carlos III (Research Project PI15/00852 and PI18/00954 to El. V., PI15/00283 to Ed. V.), and Agència de Gestió d’Ajuts Universitaris i de Recerca (2017SGR00444 to El. V., 2017SGR01365 to Ed. V., 2017SGR01271 to E. P-C.). Selena Aranda was the recipient of a predoctoral fellowship granted by the Instituto de Salud Carlos III (PFIS FI19/00268), and Norma Verdolini was the recipient of a “Río Hortega” contract (CM17/00258) from the Instituto de Salud Carlos III.
Competing interests
EV has received grants and served as a consultant, advisor, or CME speaker for the following entities: AB-Biotics, Abbott, AbbVie, Adamed, Adium, Alcediag, Angelini, Biogen, Beckley-Psytech, Biohaven, Boehringer-Ingelheim, Casen-Recordati, Celon Pharma, Compass, Dainippon Sumitomo Pharma, Esteve, Ethypharm, Ferrer, Gedeon Richter, GH Research, GlaxoSmithKline, HMNC, Intra-Cellular Therapies, Idorsia, Johnson & Johnson, Lundbeck, Luye Pharma, Medincell, Merck, Mitsubishi Tanabe Pharma, Newron, Novartis, Organon, Orion Corporation, Otsuka, Roche, Rovi, Sage, Sanofi-Aventis, Sunovion, Takeda, Teva, and Viatris, outside the submitted work. NV has received financial support for CME activities and travel funds from the following entities (unrelated to the present work): Angelini, Janssen, Lundbeck, and Otsuka.
Open access
Comments
No Comments have been published for this article.