Introduction
People who were forcibly displaced from their homes have often been exposed to severe trauma, which increases their risk of developing trauma-related mental health problems, including posttraumatic stress disorder (PTSD), depression and anxiety (Scharpf et al., Reference Scharpf, Kaltenbach, Nickerson and Hecker2020; Steel et al., Reference Steel, Chey, Silove, Marnane, Bryant and Van Ommeren2009). Accumulating evidence from various cultural contexts suggests that parents’ own experiences of adversity, for example, war-related violence, childhood maltreatment and poverty, put their offspring at risk for behavioral and socio-emotional difficulties (Cooke et al., Reference Cooke, Racine, Plamondon, Tough and Madigan2019; Field et al., Reference Field, Muong and Sochanvimean2013; Madigan et al., Reference Madigan, Wade, Plamondon, Maguire and Jenkins2017; Palosaari et al., Reference Palosaari, Punamäki, Qouta and Diab2013; Rieder et al., Reference Rieder, Roth, Musyimi, Ndetei, Sassi, Mutiso, Hall and Gonzalez2019; Song et al., Reference Song, Tol and de Jong2014; Sroufe et al., Reference Sroufe, Egeland, Carlson, Collins, Grossmann, Grossmann and Waters2005). Childhood mental health problems can have long-lasting negative consequences for individuals, for example, physical and psychological health problems, delinquency and unemployment (Aebi et al., Reference Aebi, Giger, Plattner, Metzke and Steinhausen2013; Egan et al., Reference Egan, Daly and Delaney2015; Goodman et al., Reference Goodman, Joyce and Smith2011), thereby also affecting families, communities and society as a whole. Thus, to prevent negative and further destabilizing effects for people living in conflict-ridden regions, it is crucial to understand the mechanisms underlying the intergenerational transmission of trauma. In humanitarian settings, previous exposure to war trauma and ongoing hardships including violence, material deprivation and social conflict pose accumulating risks to individuals’ mental health (Miller & Rasmussen, Reference Miller and Rasmussen2017). Notwithstanding, studies that examine how parents’ experiences of trauma affect their children’s well-being in such contexts are particularly scarce.
An extant body of research mainly conducted in high-income countries has documented an association between parents’ mental health and their children’s mental health. For instance, a meta-analytic review by Goodman et al. (Reference Goodman, Rouse, Connell, Robbins Broth, Hall and Heyward2011) reported small, yet consistent links between maternal depression and higher levels of children’s internalizing, externalizing and general psychopathology. Another meta-analysis revealed a positive association between parents’ PTSD symptom severity and children’s psychological distress (Lambert et al., Reference Lambert, Holzer and Hasbun2014). Although there are much fewer studies on the association between parental and child mental health from low- and middle-income countries, available evidence points into a similar direction. For instance, caregivers’ mental health problems predicted children’s internalizing and externalizing problems in war-affected families in Afghanistan (Panter-Brick et al., Reference Panter-Brick, Grimon and Eggerman2014) and Ethiopia (Betancourt et al., Reference Betancourt, Yudron, Wheaton and Smith-Fawzi2012). Considering these findings and the well-established link between trauma exposure and the development of psychopathology (Neuner et al., Reference Neuner, Schauer, Karunakara, Klaschik, Robert and Elbert2004; Steel et al., Reference Steel, Chey, Silove, Marnane, Bryant and Van Ommeren2009), parents’ mental health has been attributed a key role in the intergenerational transmission of mental health risk. In fact, there is emerging evidence that mothers’ adverse childhood experiences are indirectly related to their young children’s well-being through mothers’ mental health problems (Cooke et al., Reference Cooke, Racine, Plamondon, Tough and Madigan2019; Madigan et al., Reference Madigan, Wade, Plamondon, Maguire and Jenkins2017; Rieder et al., Reference Rieder, Roth, Musyimi, Ndetei, Sassi, Mutiso, Hall and Gonzalez2019; Wurster et al., Reference Wurster, Sarche, Trucksess, Morse and Biringen2020). With regard to parents exposed to war-related trauma, recent systematic reviews note that parents’ trauma-related mental health problems were linked to their unexposed children’s mental health and behavioral problems (Flanagan et al., Reference Flanagan, Travers, Vallières, Hansen, Halpin, Sheaf, Rottmann and Johnsen2020; Sangalang & Vang, Reference Sangalang and Vang2017).
However, it is less clear how parents’ mental health problems stemming from trauma and early adversity may increase children’s risk of developing mental health problems. Previous research has explored multiple direct and indirect modes of trauma transmission. The former include epigenetic mechanisms (Yehuda & Lehrner, Reference Yehuda and Lehrner2018) as well as psychodynamic processes (Dekel & Goldblatt, Reference Dekel and Goldblatt2008). The latter refer to indirect effects of parents’ traumatization on children through its impact on family processes and relationships, particularly the parent–child relationship (van Ee et al., Reference van Ee, Kleber and Jongmans2016). Research grounded in attachment theory suggests that traumatized mothers have more negative and distorted mental representations of their child, which impact their caregiving behaviors (Schechter et al., Reference Schechter, Coots, Zeanah, Davies, Coates, Trabka, Marshall, Liebowitz and Myers2005, Reference Schechter, Coates, Kaminer, Coots, Zeanah, Davies, Schonfeld, Marshall, Liebowitz, Trabka, McCaw and Myers2008). Accordingly, mothers with stronger symptoms of PTSD and depression were less sensitive and responsive, but more avoidant, overprotective, intrusive, hostile and controlling in the interaction with their young children (Erickson et al., Reference Erickson, Julian and Muzik2019; van Ee et al., Reference van Ee, Kleber and Jongmans2016). Children confronted with such behaviors are likely to develop an insecure attachment to their mother, which is characterized by anxious and avoidant behaviors (Bosquet Enlow et al., Reference Bosquet Enlow, Egeland, Carlson, Blood and Wright2014; Lyons-Ruth & Block, Reference Lyons-Ruth and Block1996). As the child’s insecure reactions and distress may remind the mother of her traumatic experiences, they may elicit further insensitive maternal behaviors and thus exacerbate the interactional dynamic (De Haene et al., Reference De Haene, Grietens and Verschueren2010). Insecure mother–child relationships, in turn, increase children’s risk of developing poor mental health outcomes including internalizing and externalizing problems (Badovinac et al., Reference Badovinac, Pillai Riddell, Deneault, Martin, Bureau and O’Neill2021; Brumariu & Kerns, Reference Brumariu and Kerns2010; Bureau et al., Reference Bureau, Easterbrooks and Lyons-Ruth2009; Ranson & Urichuk, Reference Ranson and Urichuk2008).
Increasing evidence suggests that parents’ war exposure and trauma-related psychopathology indirectly affect their children’s adjustment through dysfunctional parenting behaviors, for example more harshness and less warmth (Bryant et al., Reference Bryant, Edwards, Creamer, O'Donnell, Forbes, Felmingham, Silove, Steel, Nickerson, McFarlane, Van Hooff and Hadzi-Pavlovic2018; Eltanamly et al., Reference Eltanamly, Leijten, Jak and Overbeek2019; Sim et al., Reference Sim, Bowes and Gardner2018). However, a major limitation of most of these studies is that they rely on parents’ self-reports of their parenting behaviors, which are prone to reporting biases. For example, being in a depressive state may lead parents to selectively remember difficult interactions with their children. It would be paramount to assess parental behavior independently.
Furthermore, the assessment of specific isolated parenting behaviors provides little information about the general quality of the parent–child relationship, which is influenced by both parent and child. The concept of emotional availability (EA) is particularly suitable to conceptualize the quality of the parent–child relationship. EA refers to the caregiver’s responsiveness and attunement to the child’s emotional signals, which enables the caregiver to behave in a way that supports the fulfillment of the child’s needs and goals (Biringen et al., Reference Biringen, Derscheid, Vliegen, Closson and Easterbrooks2014). EA strongly builds on attachment theory (Ainsworth et al., Reference Ainsworth, Blehar, Waters and Wall1978) and embraces especially the importance of its hallmark concept, namely caregivers’ sensitivity to the infant’s cues and communications. However, EA also expands the attachment framework by emphasizing the emotional and dyadic nature of the caregiver–child relationship, providing a multidimensional set of features on the caregiver and child side to capture the emotional feedback loop in parent–child interactions in a wide range of contexts (Biringen et al., Reference Biringen, Derscheid, Vliegen, Closson and Easterbrooks2014). Thus, to get a comprehensive picture of the quality of the parent–child relationship, it is important to consider the affect and behavior of both interaction partners. The EA of parent and child can be assessed through observations of interactional situations using the rating system Emotional Availability Scales (EAS), thereby overcoming the limitations of using self-reports of parenting and the parent–child relationship.
Studies that have examined EA as a potential mechanism underlying the intergenerational transmission of trauma have produced equivocal results. German mothers who experienced physical or sexual abuse during their childhood (n = 58) showed significantly lower EA in interaction with their infant children than a group of mothers (n = 61) without a history of abuse (Moehler et al., Reference Moehler, Biringen and Poustka2007). In a sample of mothers (n = 52) exposed to different types of trauma including community violence, intimate partner violence and childhood sexual abuse, and their 7- to 12-year-old children in the United States, maternal EA did not mediate the association between mothers’ PTSD symptoms and children’s emotion regulation, internalizing and externalizing problems (Samuelson et al., Reference Samuelson, Wilson, Padrón, Lee and Gavron2017). Only one study so far examined the role of EA in the intergenerational transmission of trauma in refugee families. Among 49 refugee mothers resettled in the Netherlands, higher levels of mothers’ PTSD symptoms were significantly related to lower observed EA toward their infants and to lower levels of infants’ psychosocial functioning as measured through mother–rated internalizing and externalizing problems (van Ee et al., Reference van Ee, Kleber and Mooren2012). However, as maternal EA was not related to children’s psychosocial functioning, no mediation by EA could be inferred.
Several gaps in the available evidence on the role of mother–child EA in the transmission of mental health risk are noteworthy. First, there are currently no studies with displaced families living in humanitarian settings in low- and middle-income countries, where traumatic experiences and other risk factors for mental health problems, for example, poverty and ongoing violence, are widespread. Second, existing findings are based on a small number of mother–child dyads (n ∼ 50), which limits the power to detect meaningful effects and hampers the interpretability of findings. Third, apart from the study by Samuelson et al. (Reference Samuelson, Wilson, Padrón, Lee and Gavron2017), available studies have been conducted with children of preschool age or younger. However, as children grow older, they undergo significant developmental changes which require children and parents to constantly adapt their interactions to meet their relational goals and needs (Branje, Reference Branje2018). Considering the dyadic nature of EA, the interactions of mothers with older children and adolescents may show unique characteristics. Therefore, there is a need for studies examining the role of the mother–child relationship for the well-being of older children and adolescents. Fourth, previous studies do not allow to disentangle the independent effects of mothers’ trauma exposure and mental health on their EA as they either did not consider maternal mental health (Cohen & Shulman, Reference Cohen and Shulman2019; Moehler et al., Reference Moehler, Biringen and Poustka2007) or mothers’ trauma exposure (Samuelson et al., Reference Samuelson, Wilson, Padrón, Lee and Gavron2017; van Ee et al., Reference van Ee, Kleber and Mooren2012) in their analyses.
The present study aimed to shed light on possible pathways through which mothers’ traumatic experiences may impact their children’s mental health in a high-risk sample of displaced families living in proximity to ongoing conflict. The study was part of a comprehensive research project that examined the mental health and well-being of Burundian refugee families living in refugee camps in Tanzania using multiple methods (structured interviews, behavioral observations, standardized tests, biomarkers) and informants (fathers, mothers, children). The key variables in the present study were mothers’ trauma exposure and psychopathology, mother–child EA and children’s mental health problems as reported by both mothers and fathers. Our hypotheses were informed by previous research on the role of attachment and mental health in the intergenerational transmission of the effects of trauma and adversity (e.g., Bureau et al., Reference Bureau, Easterbrooks and Lyons-Ruth2009; Madigan et al., Reference Madigan, Wade, Plamondon, Maguire and Jenkins2017; van Ee et al., Reference van Ee, Kleber and Jongmans2016). We hypothesized that higher levels of mothers’ trauma exposure would be related to higher levels of mothers’ psychopathology, which would be directly related to higher levels of children’s emotional and behavioral problems as reported by mothers and fathers. We also expected that higher levels of mothers’ psychopathology would be related to lower EA toward their children, which would in turn be related to higher levels of children’s emotional and behavioral problems as reported by mothers and fathers. Lastly, we hypothesized that higher levels of mothers and children’s trauma trauma exposure would also be directly related to lower maternal and child EA, respectively.
Methods
Study setting
The small East African country of Burundi has suffered several waves of extreme interethnic and political violence in its younger history, for example in 1972, 1988 and during the long-lasting civil war from 1993 until 2005 (Irankunda et al., Reference Irankunda, Heatherington and Fitts2017; Uvin, Reference Uvin2009). In April 2015, Burundi plunged into the latest crisis, when the then president announced to stay in power for an illegitimate third term. Violence and atrocities committed by members of the ruling party, including abductions, extrajudicial killings and torture, caused more than 400 000 Burundians to flee to neighboring countries, making Burundi the 10th largest source country for refugees worldwide at the end of 2017 (Human Rights Watch, 2017; UNHCR, 2018a). Tanzania hosted the largest number of Burundian refugees with over 250 000 people as of October 2017, 58% of whom were children (UNHCR, 2018b). The refugees were resettled in three refugee camps, Nyarugusu, Nduta and Mtendeli, in the Kigoma region in Western Tanzania close to the border to Burundi.
Sampling and participants
We combined a systematic and a random sampling strategy to recruit the study sample. Each camp was divided into camp zones labeled for example A to L. We randomly selected two zones in each camp. Based on geographical maps of the camps, we determined the centers of these zones. A screening team consisting of members of the research team and NGO workers from the refugee community went to the centers of the selected zones and randomly chose a sampling direction by spinning a pen. Every sixth hut or tent in this direction was approached and the family residing there was invited to participate in the study on the following day at a specific time. For the research project, we defined a family as a triad consisting of the father or primary male caregiver, the mother or primary female caregiver and the oldest child within primary school age in Tanzania, that is, between 6 and 15 years. If a family was absent during the recruitment or did not meet our inclusion criteria, for example when a male or female caregiver was permanently absent, the next household in the determined sampling direction was approached. When the end of the assigned zone was reached, a new sampling direction was randomly determined by spinning a pen and the procedure was repeated until the necessary number of families for the next day had been recruited.
This sampling strategy resulted in a total of 230 family triads, that is, 460 caregivers and 230 children. As the majority of the caregivers were the children’s biological parents (84.3% of the mothers and 83% of the fathers), we always refer to caregivers as mothers and fathers in the following. More than 60% of the families had arrived at the camps in 2015 after the outbreak and peak of political violence in Burundi, but over 30% fled only in 2016. Most families had lived in Burundian provinces directly bordering Tanzania before their flight. About 80% of the mothers and fathers identified the political violence in their country as the main reasons for their flight, while approximately 15% indicated conflicts with their extended family, often about land and properties, or other factors, for example extreme poverty or ethnic discrimination, as main reasons. Other sociodemographic characteristics of the sample for the present study can be found in Table 1.
Note. TZS = Tanzanian Shilling (1000TZS = ca. $0.4); EAS = Emotional Availability Scales, PCL-5 = PTSD Checklist for DSM-5, BSI-18 = Brief Symptoms Inventory, SDQ = Strengths and Difficulties Questionnaire.
Procedures
The core research team consisted of three psychologists from Tanzania who all had a master’s degree and two German psychologists, one with a PhD and one with a master’s degree.
This core team was supported by research assistants who had been recruited from the refugee community in each camp with the help of collaborating organizations in the camps and community leaders. The research assistants participated in a one-week training workshop led by the core team which provided them with the necessary knowledge and skills to conduct structured clinical interviews and other study procedures. More details about the selection and training of the research assistants can be found elsewhere (Scharpf, Mkinga, et al., Reference Scharpf, Mkinga, Neuner, Machumu and Hecker2020).
Data collection took place on the compound of a collaborating organization within the camp. Upon arrival, invited families received a detailed oral and written explanation of the purpose of the study, the procedure, associated risks, their right to withdraw from participation at any time and confidentiality of their data. Parents gave their informed consent for their own and their child’s participation by signing with their names or fingerprints. In addition, children aged 11 or older also provided their own consent. All but two invited families were willing to participate in the study. Then the child, mother and father of each family were interviewed individually by the Tanzanian psychologists and Burundian research assistants. The interviews were conducted either in Kiswahili, the lingua franca in Tanzania and the refugee camp, or in Kirundi, the native language of Burundians. If necessary, the Tanzanian interviewers were supported by a research assistant from the refugee community as interpreter. After the interviews children and mothers participated in a behavioral observation to assess the quality of mother–child interaction. Finally, families were thanked for their participation and received a material compensation of 20,000 Tanzanian Shillings (ca. $8).
Measures
The interview guides for children and parents consisted of individual questionnaires that were administered in the form of a structured clinical interview to ensure standardized and comparable responses. The instruments making up the interview guides were translated from English to Kiswahili, or existing Kiswahili versions were used, according to scientific guidelines using blind back translation procedures (Brislin et al., Reference Brislin, Lonner and Thorndike1973). Discrepancies in translations were discussed and resolved by the Tanzanian project team members. Qualitative focus group discussions during the trainings of research assistants from the Burundian refugee community supported the applicability of the instruments and underlying scientific concepts in Burundian culture. A pilot assessment of eight families in Mtendeli camp provided further evidence for the cultural and contextual applicability of the measures.
Sociodemographic characteristics
In the beginning of the interviews children and parents provided basic sociodemographic information, for example, about their age and education.
Mothers’ trauma exposure
Mothers’ lifetime exposure to traumatic events was measured with a checklist adapted from Ertl et al. (Reference Ertl, Pfeiffer, Saile, Schauer, Elbert and Neuner2010). The checklist contained 38 items related to directly experiencing and witnessing specific war-related and non-war-related events, for example, physical injury, sexual assault or accidents. Mothers indicated whether they had ever been exposed to each event during their lives on a dichotomous Yes/No scale. For the analysis, we created a sum score of mothers’ trauma exposure (range 0–38). The checklist has also been successfully applied in studies in the Great Lakes region (e.g., Hermenau et al., Reference Hermenau, Hecker, Schaal, Maedl and Elbert2013) and with Congolese refugees in Nakivale refugee camp (Hecker et al., Reference Hecker, Fetz, Ainamani and Elbert2015).
Maternal psychopathology
Mothers’ PTSD symptoms were assessed using the PTSD Checklist for DSM-5 (Weathers et al., Reference Weathers, Litz, Keane, Palmieri, Marx and Schnurr2013). Mothers reported on how much they had been bothered by each of the 20 DSM-5 symptoms of PTSD in the past month on a 5-point Likert scale from 0 (not at all) to 4 (extremely). The responses were summed up to obtain a total score of PTSD symptom severity ranging from 0 to 80. This score had high internal consistency in our sample (Cronbach’s alpha = 0.92). The PCL-5 has shown good psychometric properties in various cultural settings (Ibrahim et al., Reference Ibrahim, Ertl, Catani, Ismail and Neuner2018; Sveen et al., Reference Sveen, Bondjers and Willebrand2016; Verhey et al., Reference Verhey, Chibanda, Gibson, Brakarsh and Seedat2018). Mothers’ symptoms of psychological distress were assessed using the 18-item Brief Symptoms Inventory (BSI-18; Derogatis, Reference Derogatis2000), which covers symptoms of depression, anxiety and somatization with six items each. Mothers indicated how much they had been distressed or bothered by each symptom in the past seven days using a 5-point Likert scale ranging from 0 (not at all) to 4 (extremely). For the analyses, the Global Severity Index (range 0–72) was created as the sum of all items, with higher scores representing higher levels of psychological distress (Cronbach’s alpha = 0.90). The BSI-18 has demonstrated cross-cultural applicability (Asner-Self et al., Reference Asner-Self, Schreiber and Marotta2006) and it has been used with refugee populations in different settings (Kaltenbach et al., Reference Kaltenbach, Härdtner, Hermenau, Schauer and Elbert2017; Rasmussen et al., Reference Rasmussen, Nguyen, Wilkinson, Vundla, Raghavan, Miller and Keller2010).
Maternal and child EA
We assessed mothers and children’s EA in their interaction with each other using the middle childhood/youth version (age six years and above) of the EAS (Biringen, Reference Biringen2008). The EAS consist of six scales, four of which assess interactional features on the side of the caregiver and two on the side of the child. Caregiver dimensions are Sensitivity, Structuring, Non-Intrusiveness and Non-Hostility (Saunders et al., Reference Saunders, Kraus, Barone and Biringen2015). Sensitivity refers to the adult’s ability to create a positive, healthy emotional connection with the child through behaviors and emotions. Structuring describes the extent to which an adult can support a child’s learning and guide the child to acquire a higher level of understanding. Non-intrusiveness refers to the adult’s capacity to follow the child’s lead during play without interfering physically or verbally. Non-hostility refers to an adults’ ability to regulate their negative emotions in interaction with the child. Child dimensions are Responsiveness, that is, the extent to which the child enjoys interacting with the adult and responds to their cues, and Involvement, that is, the extent to which the child engages the adult in joint play through verbal and non-verbal means (Saunders et al., Reference Saunders, Kraus, Barone and Biringen2015). With EA being a dyadic construct, scores for the dimensions of one individual also reflect the interaction partner’s complementary interaction features (Biringen et al., Reference Biringen, Derscheid, Vliegen, Closson and Easterbrooks2014). In the EAS, each dimension is rated on a scale from 1 to 7 (half points possible), with higher scores reflecting higher levels of the respective interactional quality. Based on the scores in the respective key dimensions adult sensitivity and child responsiveness, adults and children can be assigned to one of four zones: emotionally available (5.5–7), complicated (4–5), detached (2.5–3.5) and problematic (1–2). The EAS have been validated for children from 0 to 14 years. Studies found acceptable interrater reliabilities in diverse age groups (e.g., Bornstein et al., Reference Bornstein, Suwalsky, Putnick, Gini, Venuti, de Falco, Heslington and Zingman de Galperín2010; Easterbrooks et al., Reference Easterbrooks, Bureau and Lyons-Ruth2012). Significant relationships between EA and child attachment, as measured through behavior observation (Ziv et al., Reference Ziv, Aviezer and Koren-Karie2000) and mothers’ judgment (Altenhofen et al., Reference Altenhofen, Clyman, Little, Baker and Biringen2013) are considered evidence for the convergent validity of the EAS. The EAS have been used in a variety of different countries, demonstrating their cross-cultural applicability (Biringen et al., Reference Biringen, Derscheid, Vliegen, Closson and Easterbrooks2014).
The coding was based on an interaction situation between mother and child that was recorded on video tape and took place in a separate tent (Mtendeli) or room (Nduta and Nyarugusu) provided by the collaborating organization. No one could disturb the interaction by looking into or entering the tent or room. The interaction protocol consisted of two parts, a free play situation and a challenging task, which was expected to induce mild stress. First, a research assistant asked mother and child to sit on a sitting mat and instructed them to play freely with the play materials spread out on the mat, for example, marbles, painting equipment and puppets. The research assistant then started the recording and left the scene. After five minutes, the assistant came back and collected the play materials. Then the assistant instructed mother and child to build a tower together out of small wooden blocks within five minutes. The assistant emphasized that they should build the tower as high as possible and that they should start afresh if the tower fell down. After five minutes, the assistant came back and stopped the video recording. The coding was done by three of the authors of this study (FS, NC and LB) and one master student in psychology. All coders had received training in coding the EAS by the developer of the system. Recordings of mother–child interactions were available for 226 dyads. Coding relied only on nonverbal cues because loud background noise in the refugee camps made it difficult to understand what mother and child were saying to each other. In general, there was rather little verbal communication within the dyads, thus, coding relied mainly on nonverbal cues in any case. The coder training included an additional component (Level II training), in which the developer of the system independently coded selected cases and provided supervision regarding challenges in coding the interactions in the specific study context, especially the focus on nonverbal cues. To establish sufficient interrater reliability among the four coders, recordings of 20 mother–child dyads were randomly selected and rated by all four coders following the guidelines of the EAS. Overall, intraclass correlation coefficients ranged between .89 for the dimension non-hostility and .98 for the dimension structuring, indicating high interrater reliability among the four coders.
Children’s trauma exposure
Children’s lifetime exposure to traumatic events was assessed using a checklist that consisted of 13 items from the UCLA Child/Adolescent PTSD Reaction Index (Pynoos & Steinberg, Reference Pynoos and Steinberg2015) and 22 items adapted from the checklist by Ertl et al. (Reference Ertl, Pfeiffer, Saile, Schauer, Elbert and Neuner2010). The items covered both war- and non-war-related events (e.g., natural disasters, accidents, injury, sexual assault). We calculated a sum score of children’s lifetime exposure to traumatic experiences (range: 0–35).
Child mental health problems
Children’s mental health problems were assessed through mothers and fathers’ reports on the Strengths and Difficulties Questionnaire (SDQ; Goodman, Reference Goodman1997). The SDQ is a widely used screening measure for children and adolescents’ emotional and behavioral problems that can be administered to the parents and teachers of 4- to 16-year-olds and to 11- to 16-year-olds themselves (Goodman et al., Reference Goodman, Ford, Simmons, Gatward and Meltzer2000; Goodman et al., Reference Goodman, Meltzer and Bailey1998). The SDQ consists of 25 statements, which spread evenly across the five subscales emotional problems, peer problems, conduct problems, hyperactivity and prosocial behavior. Parents indicate if each statement has been not true (0), somewhat true (1) or certainly true (2) for their child in the past six months. Five items are reverse coded. The responses to all items except the five items on prosocial behavior can be summed up to receive a total score of the child’s difficulties. In our sample, this score had an acceptable internal consistency for mothers (Cronbach’s alpha = 0.75) and fathers (Cronbach’s alpha = 0.76). The mother- and father-reported scores were significantly correlated with each other (r = .25, p <. 001). To consider both parents’ perspectives on children’s mental health and to reduce single reporter bias as well as the number of necessary analyses (Holmbeck et al., Reference Holmbeck, Li, Schurman, Friedman and Coakley2002), we followed the approach of previous studies (e.g., Colonnesi et al., Reference Colonnesi, Zeegers, Majdandžić, van Steensel and Bögels2019; Volling et al., Reference Volling, Oh, Gonzalez, Bader, Tan and Rosenberg2021) and created a composite score of children’s mental health (Cronbach’s alpha = 0.80) by averaging reports of mothers and fathers. Numerous studies in different cultural settings support the usefulness of the parent version of the SDQ as screening tool for children and adolescents’ mental health problems (Cartwright et al., Reference Cartwright, El-Khani, Subryan and Calam2015; Cluver et al., Reference Cluver, Gardner and Operario2007; Costa et al., Reference Costa, Biddle and Bozorgmehr2021).
Data analyses
Recordings of mother–child interactions were available and coded for 226 mother–child dyads. Four dyads were excluded because they were foster families in which the female caregiver and child had just been united recently. Thus, the final sample consisted of 222 mother–child dyads. First, we calculated descriptive statistics of and zero-order correlations between all study variables. In order to examine the direct and indirect associations between mothers’ trauma exposure, mothers’ psychopathology, maternal EA and children’s mental health problems, we applied structural equation modeling using the R package lavaan (Rosseel, Reference Rosseel2012). The sum scores of mothers’ exposure to traumatic experiences and the composite score of mother- and father-reported emotional and behavioral problems were included as observed variables in the model. Maternal psychopathology was modeled as a latent variable with the sum scores of PTSD symptom severity and psychological distress as indicators. Similarly, maternal EA and child EA were modeled as latent variables represented by the scores in the respective dimensions. This aggregation had the advantage of reducing the multiple scales into two conceptually distinct measures (Oppenheim, Reference Oppenheim2012). All indicators significantly loaded on their respective latent variable at a level of p < .001. The variances of the latent variables were fixed to unity. Following Hu and Bentler (Reference Hu and Bentler1999), goodness of fit of the measurement model of the latent variables was indicated by a comparative fit index (CFI) ≥ 0.95, a nonsignificant root mean square error of approximation (RMSEA) ≤ 0.06 and a standardized root mean square residual (SRMR) ≤ 0.08. The initial measurement model had a reasonable fit according to the CFI (0.94) and SRMR (0.060), but the RMSEA (0.123) indicated a poor fit. Upon inspection of modification indices, we allowed covariances between the indicator non-intrusiveness and the indicators non-hostility and structuring within the latent variable maternal EA. The resulting measurement model showed a good fit (CFI = 0.99, RMSEA = 0.043 [p = .563], SRMR = 0.029].
In the structural model, we analyzed the direct and indirect associations between mothers and children’s trauma exposure, mothers’ psychopathology, maternal and child EA and children’s mental health problems. Statistical significance of the indirect associations was assessed through the Delta method as implemented in lavaan. Mothers’ age, children’s age and gender were included as control variables in the model. As our study hypotheses were directional in nature, all analyses used a one-tailed alpha of 0.05.
Results
Sociodemographic characteristics of the study sample and descriptive statistics of the study variables are displayed in Table 1. Zero-order correlations between the study variables are shown in Table 2.
*** p ≤ .001, ** p ≤ .01, * p ≤ .05.
The structural model depicting the hypothesized direct and indirect associations between mothers and children’s trauma exposure, maternal psychopathology, maternal and child EA and children’s mental health problems is shown in Figure 1. The model fit was acceptable (CFI = 0.95, RMSEA = 0.065, p = .091, SRMR = 0.061). Mothers’ lifetime trauma exposure was not significantly related to children’s mental health problems (β = −0.08, p = .293). Higher levels of mothers’ trauma exposure were significantly associated with higher levels of maternal psychopathology (β = 0.44, p < .001), which were in in turn significantly associated with higher levels of children’s emotional and behavioral problems (β = 0.26, p = .002). Thus, the indirect effect of mothers’ trauma exposure on children’s mental health problems via mothers’ psychopathology was significant (β = 0.12, p = .008). Higher levels of mothers’ trauma exposure were also significantly related to lower maternal EA (β = −0.07, p = .030). Lower maternal EA, in turn, was significantly related to higher levels of children’s emotional and behavioral problems (β = −0.41, p = .024). However, the corresponding indirect effect of mothers’ trauma exposure on children’s mental health problems via maternal EA did not reach the level of statistical significance (β = 0.03, p = .089). Given the lack of a meaningful association between mothers’ psychopathology and their EA (β = 0.04, p = .318), there was neither a significant indirect effect of mothers’ trauma exposure on maternal EA via mothers’ psychopathology (β = 0.02, p = .326) nor of mothers’ psychopathology on children’s mental health problems via maternal EA (β = −0.02, p = .385). Children’s own trauma exposure was not significantly related to child EA (β = 0.03, p = .344) and mental health problems (β = −0.04, p = .549). A positive nonsignificant association between children’s EA and their parent-reported emotional and behavioral problems (β = 0.12, p = .090) was observed. All direct, indirect, and total effects are displayed in Table 3. The effects of control variables in the model are shown in Supplementary Table 1.
Statistically significant effects are highlighted in bold. b, unstandardized regression weight; SE, standard error; z, z value; β, standardized regression weight; p, p value.
Discussion
It is well established that children of trauma-exposed parents are at an increased risk of developing mental health problems. However, the mechanisms underlying the intergenerational transmission of mental health risks are not yet well understood, especially in humanitarian settings. To contribute to the closure of the transmission gap, the present study examined the role of maternal psychopathology and EA in the association between mothers’ trauma exposure and children’s mental health problems among war-affected families living in refugee camps.
Our findings suggest two independent pathways through which mothers’ trauma exposure may affect children’s well-being. On the one hand, mothers’ higher trauma exposure was related to higher levels of psychopathology, which were in turn related to higher levels of children’s emotional and behavioral problems. This is in keeping with a growing body of evidence on parents’ psychopathology as a key factor linking parents’ exposure to trauma and adversity and children’s mental health (Cooke et al., Reference Cooke, Racine, Plamondon, Tough and Madigan2019; Flanagan et al., Reference Flanagan, Travers, Vallières, Hansen, Halpin, Sheaf, Rottmann and Johnsen2020; Greene et al., Reference Greene, Chan, McCarthy, Wakschlag and Briggs-Gowan2018). On the other hand, mothers’ higher lifetime exposure to traumatic experiences was related to lower EA in interactions with their child. Lower maternal EA, in turn, was related to higher levels of children’s emotional and behavioral problems as reported by their parents. These findings extend previous observational studies reporting negative associations between mothers’ trauma exposure and their EA toward their young children (Cohen & Shulman, Reference Cohen and Shulman2019; MacMillan et al., Reference MacMillan, Lewis, Watson, Jansen and Galbally2020; Moehler et al., Reference Moehler, Biringen and Poustka2007) and between mothers’ EA and children’s internalizing and externalizing problems in infancy, toddlerhood and middle childhood (Cohen & Shulman, Reference Cohen and Shulman2019; Easterbrooks et al., Reference Easterbrooks, Bureau and Lyons-Ruth2012; van Ee et al., Reference van Ee, Kleber and Mooren2012). As mothers’ psychopathology was not related to their EA, our findings indicate that mothers’ trauma exposure per se may affect their children’s well-being through their interactional behavior. This is noteworthy as people who do not develop increased levels of psychopathology following high trauma exposure are usually considered resilient and are thus not targeted by mental health and psychosocial services.
Our use of an observational measure of mother–child interaction corroborates previous evidence on the links between parents’ war exposure, self-reported parenting and children’s well-being (Eltanamly et al., Reference Eltanamly, Leijten, Jak and Overbeek2019; Scharpf, Kaltenbach, et al., Reference Scharpf, Kaltenbach, Nickerson and Hecker2020). At the same time, it is likely to explain the discrepancy with studies reporting associations between parents’ mental health problems including symptoms of PTSD and depression and self-reported deficits in their parenting behaviors and in the parent–child relationship (Christie et al., Reference Christie, Hamilton-Giachritsis, Alves-Costa, Tomlinson and Halligan2019; van Ee et al., Reference van Ee, Kleber and Jongmans2016). As certain psychopathological symptoms related to the sense of self, for example, negative cognitions about oneself and low self-worth, may influence parents’ evaluations of their own parenting (Banyard et al., Reference Banyard, Williams and Siegel2003; Schechter et al., Reference Schechter, Moser, Reliford, McCaw, Coates, Turner, Serpa and Willheim2015), parents’ self-reports may reflect more negative perceptions of their parental functioning rather than their actual behaviors. Although mothers suffering from mental health problems may have managed to be more sensitive and available toward their children in the short and videotaped interactional situation due to social desirability (Levendosky et al., Reference Levendosky, Huth-Bocks, Shapiro and Semel2003), the EA rating system considers maternal behaviors and affective expressions that do not seem natural or genuine as indicators of lower EA (Biringen et al., Reference Biringen, Derscheid, Vliegen, Closson and Easterbrooks2014). In a previous study using the same sample, we found higher maternal psychopathology to be related to children’s more insecure attachment representations of the mother–child relationship (Scharpf, Mkinga, et al., Reference Scharpf, Mkinga, Neuner, Machumu and Hecker2020). While this is inconsistent with the present finding, it points to the complexity of the parent–child relationship and underscores the need to combine different methodological approaches. Furthermore, the conceptualization and assessment of mothers’ psychopathology appear to be important with regard to EA. While we used a dimensional assessment of current symptoms of PTSD and psychological distress, a recent study comparing mothers with and without a lifetime diagnosis of major depression found significantly lower EA among mothers with lifetime depression (Aran et al., Reference Aran, Lewis, Watson and Galbally2022). Current depressive symptoms did not have a significant effect on EA when considering the presence of a diagnosis. Thus, effects of mothers’ psychopathology on EA may be more visible when grouping and comparing mothers based on lifetime diagnoses.
Notwithstanding, our finding that mothers’ psychopathology was unrelated to their observed interactional behaviors aligns with previous studies suggesting a more complex relationship between maternal mental health problems and their parenting (Dozio et al., Reference Dozio, Feldman, Bizouerne, Drain, Laroche Joubert, Mansouri, Moro and Ouss2020; Greene et al., Reference Greene, McCarthy, Estabrook, Wakschlag and Briggs-Gowan2020; Kumar et al., Reference Kumar, Franz, Brock and DiLillo2020; Letourneau et al., Reference Letourneau, Fedick and Willms2007). For instance, a recent study found that mothers’ ability to respond appropriately and effectively to their children buffered the impact of mothers’ PTSD symptoms on children’s mental health (Greene et al., Reference Greene, McCarthy, Estabrook, Wakschlag and Briggs-Gowan2020). Mothers suffering from traumatic experiences may try to compensate for an impact of their symptoms on their interactional behavior by increasing sensitivity and responsiveness to their children’s needs, especially when children were exposed to trauma themselves (Samuelson et al., Reference Samuelson, Wilson, Padrón, Lee and Gavron2017). In addition, children’s developmental stage seems to be relevant in understanding this link. Whereas most observational studies showing less optimal, for example, less sensitive and more hostile, interactional behaviors of mothers with PTSD or depressive symptoms have been conducted with young children (Lovejoy et al., Reference Lovejoy, Graczyk, O’Hare and Neuman2000; van Ee et al., Reference van Ee, Kleber and Jongmans2016), several studies with older children did not find associations between mothers’ trauma-related psychopathology and the quality of their observed parenting behaviors (Hartzell et al., Reference Hartzell, Stenson, van Rooij, Kim, Vance, Hinrichs, Kaslow, Bradley and Jovanovic2022; Koren-Karie et al., Reference Koren-Karie, Oppenheim and Getzler-Yosef2008; Samuelson et al., Reference Samuelson, Wilson, Padrón, Lee and Gavron2017). Adding to those latter studies, our finding may also suggest that maternal mental health problems manifest themselves stronger in interactions with young children who depend more on the mother’s initiative and engagement in interactions, whereas older children’s increasing capacity to initiate and maintain reciprocal interactions may mitigate interactional deficits among mothers. Future research should examine the factors that mediate or moderate the association between mothers’ psychopathology and their observed parenting.
Children’s own trauma exposure was not related to their EA, which is inconsistent with the finding of a link between higher trauma exposure and lower EA among toddlers in Israel (Cohen & Shulman, Reference Cohen and Shulman2019). However, besides children’s age, an important difference to our study is that the Israeli mothers reported on children’s trauma exposure. In a study conducted in the aftermath of the Gaza war, war-exposed young children without PTSD showed more comfort-seeking behaviors toward their mothers than exposed children with PTSD and non-exposed children (Feldman et al., Reference Feldman, Vengrober, Eidelman-Rothman and Zagoory-Sharon2013). This suggests that, in the absence of PTSD, children’s own experiences might also be related to more responsiveness and involvement toward parents.
From a theoretical perspective, our results support the view that caregiver EA plays a central role in child development (Biringen et al., Reference Biringen, Derscheid, Vliegen, Closson and Easterbrooks2014). They extend recent meta-analytic work on the association between parental sensitivity and child behavioral problems (Cooke et al., Reference Cooke, Deneault, Devereux, Eirich, Fearon and Madigan2022) by considering the role of relationship quality from a multidimensional perspective and in the context of a war refugee camp – a problematic context for caregiver–child relationships that is vastly understudied, despite its prevalence and unfortunate relevance. Our findings relate well to investigations demonstrating the broad applicability of the construct of caregiver EA even in different cultural contexts (Biringen et al., Reference Biringen, Derscheid, Vliegen, Closson and Easterbrooks2014; Putnick et al., Reference Putnick, Bornstein, Suwalsky and Selin2022). One next step is to examine in greater detail how differences in parent–child EA and the individual dimensions relate to child emotional and behavior problems. Given that parent–child EA has been related to the development of emotion regulation (Kammermeier & Paulus, Reference Kammermeier and Paulus2022), social competences and empathic concern for others (Moreno et al., Reference Moreno, Klute and Robinson2008), as well as Theory of Mind (Licata et al., Reference Licata, Kristen and Sodian2016), it would be interesting to include these outcomes in future studies.
Strengths and limitations
Our study is one of the first to assess the quality of the mother–child relationship using a well-established behavioral observation paradigm in a vulnerable, yet understudied population of refugee families living in camps. Moreover, our study adds to the field by linking these observational data with parents’ self-reports of children’s mental health problems obtained through structured interviews. However, some limitations of our study should be noted. First, the cross-sectional nature of the data does not permit to draw conclusions about the causality of the examined associations. Second, mothers with mental health problems may have reported higher levels of children’s symptoms because they might perceive their child’s behaviors and emotional states in a more negative way (Najman et al., Reference Najman, Williams, Nikles, Spence, Bor, O'Callaghan, Le Brocque, Andersen and Shuttlewood2001; Schechter et al., Reference Schechter, Moser, Reliford, McCaw, Coates, Turner, Serpa and Willheim2015). However, we also considered fathers’ perspective on children’s mental health to obtain a more robust measure. Moreover, if the association between mothers and children’s mental health problems were mainly due to more negative maternal representations of their child, this would be likely to play out also in their actual interactional behavior, which was not the case. Third, we had to rely on nonverbal cues for coding the mother–child interactions because background noise in the camps made it difficult to understand the few verbal cues in the interactions. Fourth, the mother–child interaction covered a short timespan and took place outside of the homes of the families including play materials and stimuli that are usually not available in this context. Thus, the interactional paradigm may not have accurately reflected the natural environment of the families. Fifth, we did not consider the role of fathers’ trauma exposure, psychopathology and EA for children’s mental health, although previous studies point to the importance of these paternal factors for child outcomes (Palosaari et al., Reference Palosaari, Punamäki, Qouta and Diab2013; van Ee et al., Reference van Ee, Sleijpen, Kleber and Jongmans2013).
Implications for research and practice
Our findings provide initial evidence from a humanitarian setting suggesting that mothers’ trauma exposure confers a risk to their children’s mental health through higher maternal psychopathology and partly through lower maternal EA. Future studies with war-affected families in low-income settings are needed to corroborate this evidence, using longitudinal designs and contextually as well as culturally appropriate observational measures of parent–child interaction quality. In doing so, future research should strive to elucidate the possible mediating and moderating mechanisms underlying the intergenerational transmission of mental health risk in this population. Furthermore, it is important to strengthen the validity of findings by considering multiple informants on children’s mental health, for example, children, parents, and teachers, by including a broader spectrum of indicators of children’s well-being, for example, prosocial behavior and academic functioning, and by using observational and experimental approaches to operationalize child outcomes. Lastly, future research should also shed light on the role of fathers in contributing to displaced children’s mental health risk and resilience.
The findings of this study can inform practitioners in humanitarian settings who aim to improve the mental health of children and adolescents. In general, the findings advocate for preventative mental health approaches with caregivers affected by war and armed conflicts to contribute to disrupting the intergenerational transmission of mental health problems. The association between mothers and children’s mental health suggests that treating mothers’ trauma-related psychopathology could reduce children’s mental health problems. There are some intervention models that have demonstrated effectiveness in reducing trauma-related disorders among adult refugees living in camps, for example Narrative Exposure Therapy for PTSD symptoms (Neuner et al., Reference Neuner, Onyut, Ertl, Odenwald, Schauer and Elbert2008) and Problem Management Plus for depressive symptoms (Bryant et al., Reference Bryant, Bawaneh, Awwad, Al-Hayek, Giardinelli, Whitney, Jordans, Cuijpers, Sijbrandij, Ventevogel, Dawson and Akhtar2022). The findings of a meta-analysis by Cuijpers et al. (Reference Cuijpers, Weitz, Karyotaki, Garber and Andersson2015) indicate that psychotherapy for maternal depression also has a positive effect on children’s mental health. However, rigorous studies testing the effects of maternal mental health interventions on children in the context of refugee camps are needed. In addition to maternal mental health, our findings suggest that programs that aim to foster mother–child EA (Saunders et al., Reference Saunders, Kraus, Barone and Biringen2015; Yousafzai et al., Reference Yousafzai, Rasheed, Rizvi, Armstrong and Bhutta2015) may be promising to improve child mental health. We are currently not aware of any such program evaluated for families living in refugee camps, where interventions should consider the scarcity of professional providers and technical resources. However, focusing more globally on the parent–child relationship, a family skills training intervention delivered by lay counselors for displaced families in Thailand showed positive effects on parent–child relationship quality as reported by children and parents (Puffer et al., Reference Puffer, Annan, Sim, Salhi and Betancourt2017).
Conclusions
This study provides unique insights into the role of mother–child interactions in the intergenerational transmission of mental health risk in a high-risk population of refugee families in an active humanitarian setting. The findings suggest that mothers’ trauma exposure increases their risk of developing psychopathology, which may contribute to children’s mental health problems. However, mothers’ exposure to trauma also appears to undermine their ability to be emotionally available to their children independently of their mental health, and this lower EA can adversely affect children’s mental health. Practitioners working with families in humanitarian contexts should give attention to mothers’ history of trauma, maternal mental health and the interactional quality of the mother–child relationship.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S0954579423000846
Acknowledgements
We thank all the families that participated in this study. Special thanks go to Plan International Tanzania and International Rescue Committee Tanzania for providing space and resources for data collection. We are extremely grateful to our highly motivated research team, to Frank Neuner, Department of Psychology, Bielefeld University, as well as to Markus Landolt and Andreas Maercker, both Department of Psychology, University of Zurich, for their continuous support. Special thanks go to Antonia Traub for her support with coding the behavioral observations. The positions of Florian Scharpf and Tobias Hecker at Bielefeld University are funded by the Emmy-Noether Program of the German Research Foundation.
Funding statement
This research was supported by the North-South Cooperation at Zurich University [grant number: (F-63212-13-01)].
Competing interests
The author(s) declare none.