Introduction

Confrontation with the death of a parent, sibling, or other relative is one of the most frequent and impactful life-events that children may experience. Like adults, most children do not experience persisting psychological problems following loss [1, 2]. Yet, for some, acute grief reactions develop into persistently distressing and disabling grief reactions or other mental health problems. Recently, the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision [DSM-5-TR; 3] and the Eleventh Revision of the International Statistical Classification of Diseases and Related Health Problems [ICD-11; 4] included prolonged grief disorder (PGD) as a novel disorder, encompassing persistent separation distress and accompanying symptoms (e.g., anger, disbelief) present to a distressing and disabling agree, at least six months after the loss (12 months for adult PGD as defined in DSM-5-TR).

Considering increasing recognition that in adults and children alike, not all grief is healthy, there is a need to understand why some children develop pervasive grief whereas others recover easily following a period of transitory grief. There is still a scarcity of knowledge about factors affecting grief in children. Regarding sociodemographic factors, it is uncertain whether boys and girls have a different risk of unhealthy grief [5]. As for situational variables, studies have found that loss of close others is associated with more intense grief compared to losing more remotely related relatives [6]. Research findings on cause of death are inconclusive with some [7] but not other [8] studies showing that confrontation with unnatural loss renders children more vulnerable to persistent grief [see also 9]. With respect to psychological processes mediating bereavement outcomes, there is evidence that persistent grief is associated with negative cognitions [10] and maladaptive coping behaviours [5]. Despite this research, knowledge about factors influencing grief in children is still limited. In addition, methodological differences between studies in terms of samples and measurements of unhealthy grief make it difficult to assess the relative importance of the different factors studied.

Notable too is that knowledge on the role of systemic variables that may affect the mental health of bereaved children is still scarce [9, 11]. Two relevant variables in this regard are caregiver’s mental health and the caregiver’s parenting behaviours [12]. The death of a parent or child within a family elicits grief and mourning in caregivers that may influence the grief of the remaining children, both directly and indirectly. For instance, witnessing the caregiver’s pain or mimicking his/her unhelpful coping strategies may strengthen the children’s grief directly. Moreover, the pain of caregivers may interfere with their responsiveness to the children’s needs, thereby indirectly increasing the children’s mental health issues. With respect to parenting behaviours, it is conceivable that positive parenting (e.g., providing support and involvement and granting autonomy) mitigates children’s grief, whereas a lack of positive parenting exacerbates this grief. Indeed, work from Sandler and colleagues [e.g., 13] has shown that increased positive parenting was associated with lower children’s mental health [14, see also 11]. Moreover, interventions increasing positive parenting successfully reduced bereaved children’s mental health problems [15].

These earlier studies are important but some of them are limited in that they focused on parentally bereaved children only and did not examine PGD symptoms. In the neighbouring field of traumatic stress studies, a growing literature has examined caregiver’s mental health and caregiver’s parenting behaviours in children’s posttraumatic stress (PTS) following traumatic events. Research has confirmed that parental and children’s PTS after a child’s exposure to a traumatic event are associated, with associations being moderated by factors such as trauma type and the method used to assess PTS [16]. Evidence also confirms that children’s PTS is associated with (increased) positive and (decreased) negative parenting, albeit that associations are generally small [17]. These findings tell us something about how caregiver’s mental health and parenting behaviours may affect children’s responses to adverse events. However, considering that traumatic events and loss events can have a very different impact on the family system, these findings cannot simply be generalized to consequences of childhood bereavement.

There is still a need to advance our understanding about factors accounting for the variability in children’s emotional distress following the death of a parent, sibling, or other close person. It seems particularly important to further our knowledge about the relative impact of psychological (individual-level) variables and systemic (family-level) variables in affecting children’s responses to loss. This knowledge bears theoretical relevance to our understanding of factors affecting children’s grief and may inform the refinement of treatment interventions to mitigate this grief.

Accordingly, the overarching aim of this study was to examine the associations of different bereavement outcomes in children (including levels of self-rated PGD, depression, PTS, and functional impairment, and caregiver-rated internalizing and externalizing) with sociodemographic, loss-related, individual-level, and family-level variables. In so doing, we used data from children and one of their caregivers participating in a randomized controlled trial (RCT) examining the effects of cognitive behavioural therapy (CBT) for PGD [18]. Two theoretical frameworks guided this research. The first was the cognitive behavioural conceptualization of prolonged grief [PG; 19]; this approach postulates that acute grief may develop into PG and associated symptoms under the influence of, e.g., negative cognitions (about the self, life, the future) and maladaptive avoidance strategies (including depressive avoidance of activities that could foster adjustment and anxious avoidance of stimuli reminding of the loss). The second was the theoretical approach underpinning Sandler et al.’s family bereavement program [12, 13] proposing that psychological functioning of caregivers, plus aspects of positive parenting (including warmth and involvement, reasoning and induction, and autonomy-granting) significantly impact children’s adjustment to bereavement.

The aims of this study were fourfold. First, we examined the associations of children’s bereavement outcomes with three individual-level variables, namely children’s negative cognitions and anxious and depressive avoidance behaviours.

Second, we examined the associations of children’s bereavement outcomes with the severity of caregiver’s symptom-levels of PGD, depression, and anxiety.

Third, we examined the associations of children’s outcomes with parenting behaviours, specifically warmth and involvement, reasoning and induction, and autonomy-granting, both as perceived by the child as well as from the perspective of the caregiver.

Last, we considered all individual-level and family-level variables significantly correlated with one or more of the children’s bereavement outcomes together, to examine which of these were most important correlates of these outcomes. We anticipated that both individual-level and family-level variables would be associated with PGD and some of the other indices of children’s mental health. Given the lack of research in this area, we had no specific hypotheses about the relative importance of the variables in explaining variance in children’s mental health outcomes.

Method

Participants and procedure

Participants were all recruited from eight outpatient clinics in the Netherlands (from 2010–2015) for a study evaluating CBT for PGD [18, 20]. They were all self-referred or referred by local professionals. Following regular screening at the outpatient clinic, eligible children and their caregivers (i.e., biological, step, or foster parents) were informed about the study, using written information and verbal explanation. To be included, children had to be aged between 8 and 18 years, to be confronted with the loss of a close person (i.e., (grant)parent, sibling, friend, someone else considered close), and to be experiencing pervasive PGD symptoms as primary complaint and reason for seeking help. Reasons for exclusion were mental disability, severe conduct or developmental disorders, undergoing concurrent psychological treatment (other than the treatments evaluated in the RCT) or pharmacological treatment, reporting substance abuse or dependence, psychotic symptoms, severe depression, and suicidality. Written informed consent was obtained from caretakers and—in keeping with ethical regulations—verbal assent was obtained from children <11 years and written consent from children ≥12 years. Next, pre-treatment measures were administered. Children completed those in the presence of a therapist. Caregivers completed these on their own. Ethical approval for the study procedures was obtained from an independent medical ethics committee (CCMO number NL30528.041.09). In total, 161 children completed the pre-treatment assessment. Data of this group were analysed except for data from two children, for whom no measures completed by caregivers were available. Authors had access to information that could identify individual participants during data collection, because they were responsible for storage of the data in accord with ethical regulations. Anonymized data were stored on a secured server, using pass-word protection. Table 1 summarizes characteristics of the sample for the current study (N = 159). (Note that in our RCT, 134 participants were included. Of all 161 children completing the pre-treatment assessment, 27 were excluded because PGD-symptoms were not the primary problem and reason to seek therapy in that study [18]. As described in the manuscript on our RCT [18], these 134 participants were randomly allocated to a condition with CBT (called CBT Grief-Help) or supportive counselling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic characteristics, loss-related characteristics, and scores on children’s bereavement outcomes (N = 159).

https://doi.org/10.1371/journal.pone.0302725.t001

Results

Associations of children’s bereavement outcomes with cognitive behavioural variables

Table 2 shows zero-order correlations of the six children’s bereavement outcomes with indices of negative cognitions, anxious avoidance, and depressive avoidance. All correlations were statistically significant, except correlations with externalizing. Outcomes of the multiple regression analyses, with all children’s outcomes subsequently regressed on the three cognitive behavioural variables (entered simultaneously) are summarized in S1 Table. All five regression models were statistically significant. In the regression analyses with PG, depression, and PTS-related functional impairment as dependent variables, negative cognitions and depressive avoidance explained unique variance. In the regression predicting PTS total scores, all three cognitive behavioural variables explained unique variance. In the regression predicting internalizing, depressive avoidance explained unique variance. Notice that we did not control for any sociodemographic or loss-related variables in these regression analyses, nor in any of the other regression analyses we conducted, because none of the sociodemographic or loss-related variables were associated with the six children’s bereavement outcomes that we considered.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Correlations between children’s bereavement outcomes and cognitive behavioural variables, caregiver’s mental health, and parenting behaviours.

https://doi.org/10.1371/journal.pone.0302725.t002

Discussion

When children are confronted with the death of a mother, father, sibling, or other close person, they face the challenge of adjusting to the loss in a situation where people close are also grieving. In this context, children are mourning their own loss and are also exposed to the pain of their family members. Conceivably, the children’s grief reactions are affected both by their individual way of coping and processing as well as by the emotional responses of caregivers and other systemic processes [cf. 9, 11–15, 41]. The current study sought to enhance knowledge about the relative importance of psychological (individual-level) and systemic (family-level) variables in affecting the severity of PGD symptoms and other outcomes in children confronted with the death of a loved one. In so doing, we focused specifically on negative cognitions and maladaptive avoidance behaviours [key mechanisms underlying PG from the perspective of cognitive behavioural grief theories; 10, 19] and emotional functioning and parenting behaviours of caregivers [central to system-based perspectives on grief, including Sandler’s et al.’s family bereavement program; 12, 13].

We first considered association of children’s bereavement outcomes with negative cognitions, anxious avoidance of loss-related cues, and depressive avoidance of potentially adaptive activities. All children’s self-reported bereavement outcomes (PG, depression, PTS, and functional impairment associated with PTS) and caregiver-rated internalizing were significantly correlated with these three cognitive behavioural variables. When we controlled for the shared variance between these variables, negative cognitions and depressive avoidance turned out to have the strongest relationship with most of these outcomes. That negative bereavement-related cognitions were associated with outcomes accords with prior studies [10]. That both forms of avoidance and particularly depressive avoidance were associated with bereavement outcomes is consistent with studies in adults [42, 43] and supports key principles from cognitive behavioural grief theories [19]. Interestingly, the cognitive behavioural variables were associated with caregiver-rated internalizing but not externalizing problems. Thus, to the extent that children display aggressive or oppositional behaviour following loss, this seems unrelated to their propensity to engage in negative thinking and avoidant coping.

We also examined associations of children’s PG and other bereavement outcomes with the severity of caregiver’s PG, depression, and anxiety. Strikingly, all associations were close to zero. Apparently, if caregivers experience more grief, depression, and anxiety, this does not mean that their children experience more emotional problems (and vice versa). These results are broadly in line with different studies in psychotraumatology showing that the intensity of parental distress and children’s distress in the face of traumatic life events are not, or only to a small degree, associated [16], especially when (as in our study) the distress was measured separately in parents and children and not reported exclusively by the parent. The absence of a linkage between children’s PG, depression, and PTS and caregiver’s PG, depression and anxiety is interesting. It may be interpreted as indicating that these emotional experiences in the face of loss have relatively independent, separate courses and determinants. Apparently, no transfer has occurred, where caregiver’s reactions spread to the children or vice versa [but see 41]. At the same time, it is possible that the association between children and caregiver responses is moderated by family characteristics (e.g., greater enmeshment) and other non-assessed variables or that transfer occurs for some, but not other outcomes. Indeed, we did observe two significant but small associations between children’s functional impairment and caregiver depression and between children’s internalizing and caregiver anxiety. The interaction between children and caregiver emotions in the face of loss is not well understood and warrants further scrutiny.

Another study aim was to examine to what extent children’s levels of PG and other bereavement outcomes were associated with parenting behaviours, as rated by children and by caregivers. Here, several notable findings emerged. First, considering the caregiver’s own ratings of their degree of warmth and involvement, reasoning and induction, and autonomy granting, we found that these ratings were all unrelated with children’s bereavement outcomes. There was one exception: there was a small but significant association between higher PTS-related functional impairment and lower caregiver-rated reasoning/induction. Considering the many correlations we considered, this finding should not be overrated. Yet it does provide a tentative indication that when children struggle with everyday activities (school, friends, home life) after loss, this may relate to caregivers being slightly less inclined to explain the consequences of the child’s behaviours to children. What is more striking is that the other indicators of caregiver-rated parenting behaviours were unrelated to children’s PG and the other outcomes. This finding is inconsistent with some system-based perspectives on grief, including Sandler’s et al.’s family bereavement program [12, 13] and is somewhat unexpected considering prior research among bereaved children indicating that higher positive parenting—which has some overlap with the parenting indices examined in this study—was indeed associated with lower mental health problems in parentally bereaved children [e.g., 44].

There are several possible explanations for differences between our and prior research. First, prior studies exclusively focused on parentally bereaved children, whereas our study included a more heterogenous group of children. Possibly, associations of parenting and children’s distress are moderated by who died within the family. Second, prior studies examined caregiver-rated internalizing and externalizing [e.g., 44] but not PG, depression, and PTS. It is possible that parenting has a different influence on different children’s outcomes. Third, it is possible that relatively well-functioning family systems were overrepresented in our sample; associations between parenting and children’s bereavement outcomes may be stronger in less healthy functioning family systems [cf. 9, 45]. Fourth, parenting behaviours are possibly more validly assessed using interview-based assessment, rather than questionnaire measures employed in this study. All this notwithstanding, the absence of a linkage of parenting and bereavement outcomes is not entirely unexpected considering research in PTS, showing that parenting generally shows a very weak association with children’s PTS following all kinds over adverse events [17]. Taken together, it seems imperative to further examine the role of parenting behaviours in children’s responses to bereavement, as well as possible moderators of this associations.

We also considered associations of children’s outcomes with their own observations of their caregiver’s warmth and involvement, reasoning and induction, and autonomy granting. Notably, when we considered correlations for the entire group, with some children reporting about their mother and others about their father, none of the children’s bereavement outcomes were associated with the parenting behaviours. For exploratory reasons, we examined the moderation effect of the caregiver the children reported about, considering that mothers and fathers may have different parenting behaviours with different impacts on their offspring. Associations of outcomes with reasoning and induction were not moderated. For warmth and involvement and for autonomy granting, interesting findings emerged. With respect to children-rated warmth/involvement, for children reporting about their mother, more warmth/involvement was associated with more PG. For children reporting about their father, more warmth was associated with less PG. Associations with depression and PTS also differed, with more warmth/involvement being associated with less depression and PTS among children reporting about their father, but not those reporting about their mother. With respect to children-rated autonomy granting, a similar significant moderation was found, with more autonomy granting being associated with less depression among children reporting about their father, but not those reporting about their mother. Again, in light of the many correlations we examined, we should be cautious not to overinterpret these outcomes. Broadly, the findings do, however, suggest that when—in the eyes of children—fathers display more warmth/involvement and grant more autonomy, this coincides with children experiencing lower grief complaints, whereas for mothers no such linkages exist. We can only speculate about the direction of causality in this association. It is possible that father’s display of positive parenting mitigated their children’s suffering. Alternatively, it is possible that when children experience less severe suffering, this encourages fathers to show more warmth and involvement and to grant more autonomy. It is also possible that the differential impact of father’s and mother’s parenting behaviours on children is affected by who died. In many cases that children reported about their mother’s parenting, the father had died, whereas in many cases children reported about their father’s parenting, the mother had passed away. Speculatively, it is possible that the connection between the remaining parent and the child for emotional support and parental advice changes much more strongly when mothers die than when fathers pass away.

In a final series of regression analyses, we had planned to examine which of the individual-level and family-level variables were most important correlates of the children’s bereavement outcomes. Because, however, apart from the cognitive behavioural variables, the family-level variables were not or hardly correlated with these outcomes, this added little to the rest of our findings. Children’s PG, depression, and PTS were correlated with the individual-level cognitive behavioural variables, but not with the family-level variables. Greater functional impairment was associated with negative cognitions, depressive avoidance, and caregiver’s depression. Internalizing was associated with increased depressive avoidance and caregiver’s anxiety. Externalizing was not associated with any of the individual-levels and family-levels variables considered.

There are several limitations that should be considered when interpreting the present outcomes. First and foremost, this was a cross-sectional study and, consequently, no conclusions can be drawn about the direction of causality between variables. So, for instance, future research is needed to examine whether negative cognitions and avoidant coping are, like in adults [46], associated with outcomes assessed later in time. Moreover, as noted above, the temporal order of the association between father’s parenting behaviours and children’s outcomes that we observed warrants further scrutiny. Second, only few family-level variables were assessed. If would be interesting for future studies to examine to what extent other variables (e.g., family communication, cohesion) moderate and mediate the association between children’s and caregiver’s emotional responses to bereavement [cf. 45]. Third, data were all collected using questionnaire measures. Research in PTS indicated that associations between children’s and parent’s PTS are stronger when interview-based assessment is used [16]. Accordingly, associations we found between mental health and parenting as reported by children and caregivers may have been underestimated because of our use of questionnaires. Future research should preferably use interview-based assessments.

Conclusion

Notwithstanding these limitations, the present study adds to a gradually growing literature examining associations between children’s responses to loss and responses of their caregivers. Findings suggest that associations of children’s symptom-levels of PG and associated indices of functioning with responses of caregivers may be small, and that children’s responses are more strongly driven by individual cognitive behavioural processes. In terms of clinical implications, these findings indicate that, when helping children who are stuck in processing loss, it is important to pay close attention to the extent to which they have negative, rigid, unhelpful thinking patterns and to address these with cognitive restructuring (including challenging cognitions and behavioural experiments). Likewise, it is indicated to help them reduce anxious avoidance of loss-related stimuli and to redirect depressive avoidance into active behaviours that help incorporate the loss into life. However, also considering prior research and theorizing pointing at the relevance of family processes in children’s responses to loss, more research on the interaction between children’s grief and the emotional responses and parenting behaviours of their caregivers and on the interplay between individual cognitive behavioural processes and systemic influences is urgently needed. If future research shows that children’s reactions are indeed significantly affected by the responses of their caregivers as well as other family-level variables, that may have clinical implications. For instance, it would emphasize the importance of considering the influence of caregiver’s reactions, parenting styles, and family-characteristics (e.g., cohesion, communication) in the assessment of children’s grief reactions, in order to capture a broad range of factors affecting the children’s responses. In addition, it would suggest the importance of addressing systemic and family-level variables in the treatment of children’s PGD. For instance, in instances were caregiver’s grief-related distress is considered to interfere with the children’s grief, parental counselling could be offered parallel to individual treatment of children. Observations in our own clinical trial indicated that caregiver counselling was a welcome addition to the individual CBT for children. In cases where the children’s grief is hindered by problematic family functioning (e.g., poor communication, conflict regulation, cohesion) family-focused interventions may be used. More research is urgently needed to enhance knowledge about the interaction between grief reactions of children and their caregivers, to inform the development of interventions promoting adaptive grieving in children.

Supporting information