It is estimated that around 25 percent of children live in a household with at least one mentally ill parent (1–4). Children of parents with a mental illness (COPMI) are considered a high-risk population for the development of psychological disorders. Compared to children of parents without mental illness (COPWMI), COPMI have not only an increased lifetime risk to develop a mental illness themselves (5–7), but they are also at risk for multiple psychological and developmental disadvantages. In particular, COPMI have more subclinical internalizing and externalizing symptoms (8, 9). Thus, a transgenerational transmission of mental disorders (TTMD) can be assumed. That makes COPMI a target group for selective prevention programs.

In the TTMD model (5), different transmission factors and mechanisms and their interplay are assumed to underlie the transmission of mental disorders. Parent- and child-related factors display a promising target for preventive measures. However, the impact of the single factors is not sufficiently tested yet. In this context, empathy is of special interest, as besides the positive role it displays in many fields of interpersonal interaction, it has been shown to be a risk factor for psychopathology under certain conditions (10).

Empathy broadly refers to reactions of an individual to another person’s experiences (11). Research has shown that empathy is a multidimensional construct (11), consisting of both cognitive and affective facets (12). Cognitive empathy has been conceptualized as the ability to take over another person’s perspective and require an understanding of affect related motives, thoughts and feelings of a person [Birnie et al., (13)]. In contrast, affective empathy includes the ability to connect with the emotional state of another person (13) and sharing of affective states and feeling of concern for others (11, 14). A special subset of empathy is sympathy (15). Baron-Cohen and Wheelwright (15) define sympathy as feeling an emotion after seeing/learning another person’s distress which in turn moves one to alleviate the suffering of the other. It can involve elements of both cognitive and affective empathy (15). In contrast to sympathy, personal distress is a self-focused aversive affective reaction accompanied by the motivation to reduce the own distress (16).

Typically, empathy is seen as a strength and various studies reveal relations between empathy and positive outcomes in various functional areas in children. For example, responding empathically to others is associated with adaptive functioning and social competence (17, 18), popularity among peers (19, 20), and is linked to children’s academic success (21). In addition, previous studies have shown a negative relationship between empathy and aggression or externalizing problems (22, 23).

However, recent theoretical and empirical literature suggests that empathy might also be associated with certain risks (10, 24, 25). Particularly, affective empathy is positively associated with internalizing symptoms. This association seems to apply not only to children and adolescents in clinical (26, 27) but also to non-clinical (28–30) samples. In contrast, studies investigating cognitive empathy are less consistent. Whereas studies with healthy children (29) indicated that low cognitive empathy is associated with more psychopathology, studies investigating clinical samples of children did not find evidence for a relationship (26, 27). Empirical data on empathy and its effect on psychopathology in COPMI is limited. Studies investigating whether higher psychopathological symptoms in COPMI (8, 9, 31) are associated with higher or lower empathy levels are inconsistent. On the one hand, children of depressed mothers had both higher prevalence of psychopathology and lower affective empathy levels than children of mothers without depression (32). However, it must be noted that in this study more than half of the children had a mental illness themselves. In contrast, in the study conducted by Tully and Donohue (24) children of chronically depressed mothers and healthy mothers did not differ significantly in cognitive and affective empathy. Interestingly, higher levels of empathy (cognitive and affective) were related to greater internalizing problems in children of chronically depressed mothers (depressed for 36 months) only (24). At the same time, affective and cognitive empathy and internalizing symptoms were unrelated in children of mothers with shorter (12 or 24 months) depression and in children of mothers without depressions.

Zahn-Waxler and Van Hulle (33) suggested distinct pathways through which empathy can be adaptive or maladaptive in children: Unfavorable conditions in the early family environment contribute to a maladaptive pathway. For example, in case of parental depression, empathy can lead to anxiety, sadness and guilt because the child develops self-blame cognitions followed by pathogenic guilt. Pathogenic guilt in children, in turn, heightens the risk for developing depression (33). In this line, Tone and Tully (10) proposed that different moderators affect the development and impact of affective and cognitive empathy. These moderators in conjoint with emotion regulation difficulties lead to personal distress and guilt, resulting in an increased risk for internalizing problems. Parental mental depression is considered to be such a moderator. Thus, the tendency being cognitively empathic could be accompanied by attempts to understand the mother’s emotions and their fluctuations (24). Subsequent inaccurate assumptions of responsibility for the sadness of the parents, in turn, increases the risk of guilt and self-blame if the child is not able the regulate it (24). In regard to affective empathy, Tully and Donohue (24) suggested that affective empathic sensitivity in COPMI can lead to internalizing symptoms via deficits in emotion regulation (ER) strategies and personal distress, as well. Children with high affective empathy tendencies may have unregulated arousal, hypervigilance and distress in response to the depression of the mother if they cannot regulate their emotions effectively. The authors further argue that affective and cognitive empathy, while related, seem to function independently. Thus, affective empathy does not predict cognitive empathy or vice versa (24). While the models and assumptions described above are limited to parental depression only, it could apply to other parental mental disorders as well, as COPMI suffer from self-blame, misplaced responsibility or other dysfunctional cognitions regardless of the parent’s specific diagnosis (34, 35).

Inconsistent findings regarding the association between empathy and psychopathology could be explained, inter alia, by mediating factors such as ER. Thus, empathy and psychopathological symptoms may not be directly related, but rather indirectly through ER. ER comprises processes that influence the incidence, kind, intensity, and duration of emotions as well as their effects on feelings and behaviors (36, 37). Strategies of ER can be adaptive (e.g. cognitive re-appraisal, problem solving, acceptance, distraction) if they increase positive or decrease negative emotions, or be maladaptive (e.g. rumination, suppression), having the opposed effect (38).

In addition to empathy, ER has been shown to be associated with psychopathology in children (39–42). In particular, the increased use of maladaptive ER strategies is associated with psychopathology in mental disorders of many kinds (39, 43). Studies investigating ER in COPMI revealed deficits in ER in COPMI versus COPMWI (8, 44–46).

It has been theoretically proposed that ER play a crucial role in the impact of empathy (47). By perceiving another individual’s state, an emotion state in the observer is generated. The latter is a function of the observer’s level of cognitive and affective empathy and is subject to the emotion regulatory process of the observer (48). The assumption is, that deficits in ER lead to higher levels of personal distress and lower levels of sympathy when confronted with another individual’s negative emotional state. Empirical studies further indicate that cognitive and affective empathy may be differentially related to ER. In adult community samples it has been shown that maladaptive ER is negatively related to cognitive empathy (25, 49, 50) and positively related to affective empathy (25, 49, 51). Studies investigating empathy and ER in healthy children are lacking to date. However, the results with adult community samples are expanded through single studies on either ER or empathy investigating clinical samples. Children with autism spectrum disorder show impairments in cognitive empathy (52), and also an increased use of maladaptive regulation strategies (53). Adult patients with borderline personality disorder are also characterized by having difficulties in ER (54), but accompanied by increased affective empathy (55).

In summary theoretical and empirical literature mainly support the association between high affective and low cognitive empathy and maladaptive ER on the one hand and high affective and low cognitive empathy and psychopathology on the other hand. However, the mediating effect of maladaptive ER on this relationship is barely examined. To the best of our knowledge, there is only one study that has investigated this relationship in an adult community sample (25). MacDonald and Price (25) showed that maladaptive ER mediated the relationship between affective empathy and internalizing symptoms. However, cognitive empathy and internalizing symptoms were not related.

It can be stated that there is a lack of studies investigating the role of parental mental illness as risk factor for the maladaptive pathway from empathy to psychopathological symptoms in children. In addition, the pathway by which empathy might contribute to psychopathological symptoms in COPMI has not yet been empirically studied. It is important to note, that all studies and models in this research area refer to internalizing symptoms. However, COPMI have an increased risk to develop not only internalizing but also externalizing symptoms (8, 9, 31). Further, the use of maladaptive ER strategies in children and adolescents is associated with both internalizing and externalizing symptoms (56). Therefore, the tendency being highly empathic could also be associated with externalizing symptoms in COPMI. Identifying the mechanisms of risk is of clinical importance since the reduction of maladaptive ER strategies or self-blaming thoughts and guilt could be targeted in preventive interventions and buffer the impact of parental mental illness on children.

In line with the theoretical background and it´s gaps, the first aim of the current study is to examine the moderating role of parental mental illness on the relationship between both cognitive and affective empathy and psychopathological symptoms in children. According to the results of the only existing study on this topic of Tully and Donohue (24), we hypothesize that empathy (cognitive & affective) is positively related to internalizing symptoms in COPMI and unrelated in COPWMI. We assume the same pattern for externalizing symptoms (see Figure 1 for demonstration).

The second aim is to investigate the mediating effect of maladaptive ER strategies on the relationship between cognitive/affective empathy and psychopathological symptoms in children. We examined maladaptive ER strategies because psychopathology is rather associated with maladaptive ER (39). The use of maladaptive ER strategies in turn is positively associated with personal distress (57, 58). Both are suggested to underlie the maladaptive pathway of empathy (10). If the moderating effect of parental mental illness is confirmed, we will test the mediations separately for COPMI and COPWMI. We assume, that maladaptive ER mediate the relationships in COPMI only. We hypothesize that higher affective/cognitive empathy is related to more maladaptive ER strategies and more maladaptive ER strategies are associated with more psychopathological symptoms in this specific group. We will test all hypotheses separately for cognitive and affective empathy (see Figure 2 for demonstration).

Two hundred parents with mental illness signed informed consent for participating in the add-on project COMPARE-Emotion. However, complete data sets were only available from 122 independent parent-child dyads. Of these dyads, in turn, other information from the COMPARE-Family project such as psychopathology of parents or children were missing (N=22). Eighty-seven parents without mental illness signed informed consent for COMPARE-Emotion. In the end, for the current study complete data sets of n = 187 independent parent-child dyads including 100 COPMI and 87 COPWMI were available. Children ranged from six to sixteen years (M = 10.42, SD = 2.53) in age and included 82 males (44%). COPMI and COPWMI groups did not differ in child age, child gender and parents’ age. Children’s age was evenly distributed across males and females, t(185) = 1.31, p = .190, Cohen’s d = .19. Furthermore, the socioeconomic status (SES) of COPMI (M = 4.69, SD = .99) was lower than of COPWMI (M = 6.06, SD = .85). For demographic characteristics of participants separately for COPMI and COPWMI see Table 1 . As noted in Table 2 , 45% of mentally ill parents had a Depressive Disorder as primary diagnosis. The number of comorbid diagnoses in parents with mental illness ranged between 0 - 5 (M = 1.12, SD = 1.19), and average severity of the primary diagnosis was six (SD = 1.03, range from 3 - 8).

COPMI were recruited as part of a randomized controlled multicenter RCT-study for COPMI in Germany (COMPARE-Family) (1, 59). Patients were primarily recruited from the University outpatient clinics at each study site. In the study center in Giessen patients were recruited in addition by mailings of randomly picked addresses of families with children in the corresponding age range provided by the local registry office, public advertisement (flyer, newspaper), inpatient psychiatric clinics (COPMI) and the University’s internal mailing list. COPWMI were recruited as part of the add-on project COMPARE-Emotion in addition via the research group’s database of former study participants. Inclusion criteria for COMPI were: (a) between 6-16 years of age, (b) parent with a mental illness according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (60). For COPWMI inclusion criteria were (c) parents without mental disorders and without psychotherapeutic treatment in the last 5 years or after the child was born. Exclusion criteria were (a) insufficient German language skills of children and the parents, (b) severe impairment of the children requiring comprehensive treatment, (c) parental outpatient or inpatient treatment while participating in the study, or continuous use of benzodiazepines.

The study was approved by the local ethics committee. All participants and their parents gave written informed consent. While the families of the COPWMI group only took part in the add-on project once, the assessment was repeated for the families of the COPMI group at three measurement points (59). From the COPMI group the data of the first assessment point of the study were analyzed.

All statistical analyses were performed using SPSS version 27 (71). For the moderation and mediation analyses, the PROCESS tool was used (72). The analytical strategy included preliminary analyses of possible differences between groups (COPMI vs. COPWMI) in the study variables according to demographic characteristics to address the need for potential confounding variables in the subsequent analyses. For both moderation and mediation analyses unstandardized path coefficients are reported.

Aim 1: Moderation analyses were run to determine whether the relationship between child empathy and child psychopathological symptoms is moderated by parental mental illness [model 1 of the PROCESS Tool, Hayes (72)]. Separate moderation analyses for affective/cognitive empathy and internalizing/externalizing symptoms were calculated. The relationships of all variables involved in the moderation analyses were approximately linear, as assessed by visual inspection of the scatterplots after LOESS smoothing. Further, observations were independent. Since we used a robust method for the analyses, we dispense with checking normal distribution and heteroscedasticity (72).

Aim 2: Whether child maladaptive ER strategies mediate the relationship between child empathy and child psychopathological symptoms was analyzed by moderation analyses [model 4 of the PROCESS Tool, Hayes (72)]. Bivariate correlations were calculated to determine the relations between the study variables. We calculated mediation analyses with maladaptive ER strategies as mediator separately for internalizing and externalizing symptoms. The relationships of all variables involved in the mediation analysis were approximately linear, as assessed by visual inspection of the scatterplots after LOESS smoothing. Further, observations were independent. Since we used a robust method for the analyses, we dispense with checking normal distribution and heteroscedasticity (72). Indirect effects were estimated using the bootstrapping technique with 5000 bootstrap samples and 95% BC confidence intervals. The mediation model was determined to be significant if the 95% BC confidence interval did not contain zero. Since the groups differed in SES (p <.001), this variable was included as a covariate in each analysis conducted with the total sample.

For affective empathy as independent variable and internalizing symptoms as outcome variable, the overall model was significant, F(4, 182) = 8.95, p <.0001, predicting 16.43% of the variance. However, results did not show that parental mental illness moderates the effect between affective empathy and internalizing symptoms significantly. Following recommendations by Hayes (72), the interaction term and moderator was dropped from the model, resulting in a new linear regression model with the independent variable affective empathy. This new model revealed a significant relationship between affective empathy, B = .366, p <.001, for internalizing symptoms indicating that affective empathy predicts internalizing symptoms positively. For affective empathy as independent variable and externalizing symptoms as outcome variable, the overall model was significant, F(4, 182) = 3.38, p = .011, predicting 6.91% of the variance. However, parental mental illness did not moderate the effect between affective empathy and externalizing symptoms significantly. The followed linear regression model (see above) revealed a significant relationship between affective empathy, B = .217, p = .006, for externalizing symptoms indicating that affective empathy predicts externalizing symptoms positively.

The overall model for cognitive empathy and internalizing symptoms was also significant, F(4, 182) = 5.61, p <.001, predicting 17.65% of the variance. Results showed that parental mental illness moderated the effect between cognitive empathy and internalizing symptoms significantly, ΔR² = 1.78%. Whereas in COPMI cognitive empathy positively predicted internalizing symptoms, b = .249, 95% BCa CI [.032,.466], t = 2.723, p = .024, in COPWMI the relationship was not significant, b = -.092, 95% BCa CI [-.354,.171], t = -.688, p = .493. For cognitive empathy as independent variable and externalizing symptoms as outcome variable, the overall model was significant, F(4, 182) = 5.02, p <.001, predicting 9.94% of the variance. Results showed that parental mental illness moderated the effect between cognitive empathy and externalizing symptoms significantly, ΔR² = 3.23%. Whereas in COPMI cognitive empathy positively predicted externalizing symptoms, b = .299, 95% BCa CI [.090,.508], t = 2.826, p = .005, in COPWMI the relationship was not significant, b = -.125, 95% BCa CI [-.378,.129], t = -.971, p = .333. For regression coefficients, confidence intervals, standard errors, p-values, and test statistics see Table 3 (affective empathy) and Table 4 (cognitive empathy).

All variables of interest were correlated at the p < 0.01 level (see Table 5 ). Affective and cognitive empathy was positively moderately correlated with maladaptive ER and positively weak with internalizing/externalizing symptoms. Maladaptive ER was positively moderately associated with internalizing/externalizing symptoms. For demonstration of the results of the mediation analyses see Figures 3 , 4 (affective empathy) and Figures 5 , 6 (cognitive empathy).

The relationship between affective empathy and psychopathological symptoms was not moderated by parental mental illness and psychopathological symptoms were significantly predicted by affective empathy in the total sample. We therefore assumed that affective empathy and psychopathological symptoms are related regardless of parental mental illness indicating the same relationship pattern in both groups. Consequently, we analyzed the mediating effect of maladaptive ER strategies in the total sample. The relationship between cognitive empathy and psychopathological symptoms was moderated by parental mental illness. Thus, we calculated two mediation analyses, one with COPMI and one with COPWMI.

Indeed, there was a significant indirect effect of affective empathy on psychopathological symptoms (inter-/externalizing symptoms) through maladaptive ER strategies. This indicates that the relationship between affective empathy and psychopathological symptoms can be explained by maladaptive ER in COPMI and COPWMI. Because of the significant moderating effect of parental mental illness on the relationship between cognitive empathy and psychopathological symptoms, two separate mediation analyses were calculated, one with COPMI and one with COPWMI each with internalizing and externalizing symptoms. In COPMI, there was a significant indirect effect of cognitive empathy on psychopathological symptoms (inter-/externalizing symptoms) through maladaptive ER strategies. This indicates that the relationship between cognitive empathy and psychopathological symptoms can be explained by maladaptive ER in COPMI. In COPWMI, neither the direct nor the indirect effect was significant indicating that cognitive empathy is unrelated to psychopathological symptoms in COPWMI.