The current state of research on the emergence and maintenance of somatoform symptoms in childhood identifies the influence of both child and parental factors. The aim of the present study is to examine reciprocal relations and stability between somatoform complaints, children’s adaptive emotion regulation (ER) strategies, withdrawal behavior, and overprotective parenting. In total, 97 children (female n = 46, (47%); M age T1 = 9.7 years (SD) = 0.54) and their parents completed questionnaires at two time points. A cross-lagged panel design was used to analyze reciprocal associations and stability effects between the included variables over one year (Kearney, 2017; Selig & Little, 2012). Results show significant autoregressive effects of both the individual child and parental variables. Parental overprotection predicted child emotion regulation, withdrawal behavior, and somatoform complaints over time; with no reciprocal effects. Similarly, only unilateral effects were shown between adaptive ER strategies, and withdrawal behaviors, and somatoform symptoms. The results are discussed in the framework of learning theory and secondary gain.
Opmerkingen
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The current state of research shows a broad consensus that somatoform complaints, like stomachache, nausea, headache, and dizziness, seriously impair children’s psychological development and social life (Beck, 2008; Kelly et al. 2010; van Geelen & Hagquist, 2016; Vesterling & Koglin, 2020; 2021). In general, not only the children, but also their social environment (e.g., in the family and school) are affected by the impact of the complaints. For the children this leads to both a decrease of self-efficacy expectations and age-appropriate behavior, as well as an increase in social withdrawal and school avoidance. Simultaneously, parents are frequently worried about repeatedly occurring symptoms, which may explain an inappropriate parental reinforcement of the children’s avoidance behavior (Malas et al. 2017; van Geelen & Hagquist, 2016). It is important to examine the influence of children’s psychological factors and family interaction factors on the development of somatoform symptoms (Horwitz et al., 2015; Kozlowska, 2013; Landa et al., 2012; Malas et al., 2017; Walker et al., 2006). Although research demonstrates associations with these factors, less attention has been paid to possible reciprocal effects or stability over time. This neglects the notion that high levels of somatoform complaints in children may lead to an increase of parental overprotective behavior which in turn results in a reduction of children’s age-appropriate adaptive emotion regulation (ER) strategies, school attendance, and social contact. Such a reduction, might again lead to an increase or maintenance of somatoform complaints.
Middle childhood as a crucial developmental phase
Middle childhood, with childrens ages ranging from 6–12 years, is described as a period of serious developmental transitions (Collins & National Research Council, 1984). Crucial changes in this developmental stage are based on the expansion of the children’s social world, mainly through school entry. New social interactions occur, both within the peer group and with teachers. These new experiences of interaction and the associated conflicts and social-emotional challenges, which can be accompanied by school entry, have to be mastered as developmental tasks (Collins & National Research Council, 1984). One remarkable and challenging consequence of this social expansion could be changes in the child-parent relationship during this period (Collins et al. 2002; Kerns & Brumariu, 2016; Saarni, 1999; Skinner & Zimmer-Gembeck, 2007). In this context Zarra-Nezhad and colleagues (2022) stated that the predominant number of studies relevant to the topic focus on the effects of parental education on child behavior. Much less is known about the opposite direction, i.e., on the possible effects of child behavior on parental factors.
This indicates that in particular longitudinal studies focusing on the bidirectional relations between children and their parents in middle childhood are of importance and should be carried out more frequently (Zarra-Nezhad et al. 2022). For these reasons, the present study considers the developmental phase of middle childhood.
Interactions Effects of Child Factors
A great number of research projects in the field of child developmental psychology or psychopathology focus on the influences of psychological child factors, particularly on children’s competencies in regulating their emotions (Cicchetti et al., 1995; Eisenberg et al., 2001; Eisenberg & Morris, 2002). Based on Thompson’s (1994) definition an important aspect of ER is the ability to modulate both the expression and the intensity of emotions (Thompson, 1994). Therefore, the internalization and use of suitable strategies highlights a remarkable process in child development (Saarni, 1999; Skinner & Zimmer-Gembeck, 2007). In this context, functional/adaptive and dysfunctional/maladaptive ER strategies or emotion suppression have often been studied separately in regard to their effects on the developmental process (John & Gross, 2004; Schäfer et al., 2017). Regarding the influences of the emergence and maintenance of internal psychological problems, (e. g. social anxiety or withdrawal and somatoform complaints) current research revealed reducing effects of functional or adaptive ER strategies (Aldao & Nolen-Hoeksema, 2011; Aldao & Nolen-Hoeksema, 2012). Dysfunctional or maladaptive ER strategies, as well as emotion suppression and emotional control, increase the risk of internal psychological problems (Aldao et al., 2014; Aldao et al., 2010; Cutuli 2014; John & Gross, 2004). However, the question if preexisting psychological problems may also predict the use of certain ER strategies in children has received less attention (Domaradzka & Fajkowska, 2018). Sfärlea and colleagues report (2019) that adaptive ER strategies can be inversely predicted by depressive problems. Burr and colleagues (2021) report that individuals utilize different ER strategies depending on their levels of anxiety.
Direction of Effects of Parental and Child Characteristics
The impact of parenting on children’s behavioral adjustment (Bornstein et al., 2018; Otterpohl & Wild, 2015), emotional development (Otterpohl & Wild, 2015; Rueter & Conger, 1998) and chronic functional pain (Walker et al., 2006) has been the focus of numerous previous studies. Although the interaction processes between child and parent factors could be bidirectional (Kalomiris & Kiel, 2016; Otterpohl & Wild, 2015; Rueter & Conger, 1998; Sameroff, 2009; Zarra-Nezhad et al. 2022), the focus of previous research has predominantly been on the effects of parental factors on child developmental outcomes. Connections between child factors such as temperament (Klein et al., 2018), emotional competencies (Goagoses, et al., 2022; Morris et al., 2007; Rueter & Conger, 1998; Zarra-Nezhad et al. 2014; Vesterling & Koglin, 2021), behavior problems or psychopathological development outcomes (Vesterling & Koglin, 2021) and parental factors such as child-caregiver attachment (Maunder et al., 2017; Vesterling & Koglin, 2020; Vesterling & Koglin, 2021), family climate and several parenting styles (dimensions or behaviors) (Goagoses et al., 2022; Klein et al., 2018; Morris et al., 2007; Rueter & Conger, 1998; Schiffrin, et al., 2019; Zarra-Nezhad et al. 2014) are of particular interest. For example, insecure child-caregiver attachment (Vesterling & Koglin, 2021), poor family climate, overprotective or controlling parenting (Brenning et al., 2017; Goagoses, et al., 2022; Otterpohl & Wild, 2015; Rueter & Conger, 1998) predict dysfunctional ER strategies in children. The assumption that parental overprotection increases both (1) children’s social withdrawal or anxiety and (2) somatoform complaints in direct or indirect ways is of particular interest for our investigation (Janssens et al., 2009; Schiffrin et al., 2019; Zarra-Nezhad et al. 2022). Although research based on a transactional model of development (Sameroff, 2009) supports the assumption that child factors, such as emotional competencies and behavior, influence parenting behavior in a bidirectional manner, a notable research gap can be identified concerning the potential impact of child factors on parenting behavior (Zarra-Nezhad et al. 2022).
Maintaining Factors and the Framework of Learning Theory
With regard to the maintenance of children’s somatoform symptoms and resulting functional disabilities, the model of operant conditioning can help clarify possible effect mechanism (Walker et al., 2002; Walker & Zeman, 1992; Whitehead et al., 1982). Prior research has described that supportive social responses on children’s presented somatoform complaints (i.e., increased parental attention and/or age-inappropriate support and/or support of avoidance behavior) could be (mis-) interpretated as a reward (according to the principles of positive and negative reinforcement) leading to a reinforcement of the complaints (Walker et al., 2002; Walker & Zeman, 1992; Whitehead et al., 1982). An underlying belief of overprotecting parents is that there are specific reasons to believe that their children are not able to manage difficult situations self-reliantly. In consequence, these parents try to control and relieve their children by excessive (not age-appropriate) supporting behavior, resulting in both (1) a reduction of children’s age-appropriate tasks (i.e., social contact and school visit) and (2) a reduction of the use of children’s own adaptive ER strategies (Luebbe et al., 2014). Based on the principles of positive and negative reinforcement such a parental approach may lead to children’s short-term emotional relief (caused by positive and negative reinforcement). However, in the long-term children’s self-efficacy expectations and self-confidence decreases which in turn could lead to social anxiety, withdrawal and somatoform complaints or school avoidance (Clarke et al., 2013; Janssens et al., 2009).
Secondary Gain Concept
The psychoanalytic construct of secondary gain has also been alleged to maintain somatoform symptoms (Fishbain, 1994; Fishbain et al., 1995; van Egmond, 2003; Walker et al., 2002; Worzer et al., 2009). Concerning the history of the terms “primary and secondary gain” Fishbain (1994) reported the original definitions by Sigmund Freud (1917/1959), writing “To Freud primary gain was a decrease in anxiety brought about through a defensive operation that had resulted in the production of the symptom of the illness. Primary gain was therefore an intrapsychic phenomenon. […] Freud went on to define secondary gain as an interpersonal or social advantage attained by the patient as a consequence of the illness” (Fishbain, 1994, p. 264–265). The author points out, that further definitions have only been slightly amended over time and refers to Barsky’s definition, describing secondary gain as an “Acceptable or legitimate interpersonal advantages that result when one has the symptoms of a physical disease” (Barsky & Klerman, 1983, p. 276). Examples of secondary gains for children include both (1) parental overprotectiveness and (2) the possibility of withdrawal or avoidance from unpleased social, age-appropriate activities like regular school attendance (Fishbain, 1994). It should be noted that there is a broad scientific consensus that gain concepts are evoked unconsciously (Fishbain, 1994; van Egmond, 2003).
The Present Study
The aim of the present study was to examine longitudinal reciprocal relations and stability between somatoform complaints, adaptive ER strategies, social withdrawal behavior, and overprotective parenting. In order to get a reliable prediction of changes in the dependent variables over time (from point of measurement 1 = T1 to point of measurement 2 = T2), the T1 level of the dependent variables need to be controlled (Selig & Little, 2012). To facilitate this, a cross-lagged panel design including two measuring points investigating both (1) reciprocal relations of parent and child factors and (2) interindividual stability of the included variables over a one-year period was conducted (Finkel, 1995). As the theoretical background has shown there is support for both a parent-child factor model and a child-parent factor model (Otterpohl & Wild, 2015; Sameroff, 2009).
Hypotheses
In line with these theoretical assumptions, reciprocal relations between all included factors were hypothesized (see Fig. 1). This may cause an internal cycle of maintenance and increase of somatoform complaints over time. Firslty, it was hypothesized that high levels of parental overprotection at T1 predict (1) lower levels of children’s adaptive ER strategies (2) higher levels of social withdrawal and (3) higher levels of somatoform complaints at T2. Secondly, it was hypothesized that children’s adaptive ER strategies at T1 as well as high levels of social withdrawal and somatoform complaints on T1 predict high levels of parental overprotection at T2. Furthermore, reciprocal relations between the several child factors, in the sense that low levels of adaptive ER strategies in children at T1 predict high levels of social withdrawal and somatoform complaints at T2 and that high levels of social withdrawal and somatoform complaints at T1 predict low levels of adaptive ER strategies in children at T2 are assumed. In addition, positive reciprocal effects between social withdrawal and somatoform complaints at T1 to T2 are expected. Finally, significant stability between the respective constructs over time are assumed. With regard to individual child factors and the child-parental interaction factors, gender differences were identified as relevant influence factors in prior studies (Endendijk et al., 2016; Nolen-Hoeksema & Aldao, 2011). Therefore, gender is included in the present analyses to control for possible confounding effects. Differences in the associations between child and parental factors in the sense that parents overprotect their daughters more than their sons, which in turn results in stronger links within the respective child factors were presumed. Also, stronger positive relations with social withdrawal and somatoform complaints, and stronger negative associations with adaptive ER strategies for girls then for boys were suspected. Since there is a small age range in the present sample (eight to eleven years at T1), age has been considered as an irrelevant factor in regard to relations in present analyses. The hypothesized relations over time are presented in Fig. 1.
Fig. 1
Hypothesized model with parental overprotection, adaptive ER strategies, withdrawal behavior and somatoform complaints over time (T1, T2). Gender as moderating variable. Solid lines indicate expected positive relations. Dashed lines indicate expected negative relations. Dotted lines indicate expected gender influences
×
Methods
Procedure and Participants
Data used in the present study are taken from our longitudinal research project “Impact factors and conditions of somatoform symptoms in childhood within familial and school contexts”. We listed our project on the WHO International Clinical Trials Registry Platform. All variables of interest (children’s somatoform complaints, adaptive ER strategies, withdrawal and parental overprotection), were collected at two timepoints of measurement with a sufficient lapse of time between the two measurement points (Timepoint 1 (T1) in spring 2018 (T11) / 2019 (T12) and Timepoint 2 (T2) in spring 2019 (T21) / 2020 (T22). Inclusion criteria were formulated for all participants: (1) children have to attend the 4th grade in primary school at T1, (2) children and their parents have sufficient language skills in German language to understand and answer the questionnaire items, (3) participating parents have to agree to participate by signing a consent form. In order to recruit participants, we contacted and informed 124 primary schools in Lower Saxony (Germany) about our study project via e-mail. After about two weeks, we asked school directors by phone if their schools wanted to participate. As a result, 29 primary schools took part. Subsequently, we distributed 1797 informed consents to children and their parents. 254 families agree to participate by signing the consent form. 221 families returned at least one completed questionnaire to the first wave. Data collection at T1 took place in all participating primary schools in Lower Saxony (Germany). Pupils were allowed to stay away from lesson for the time of the survey. In the questionnaire session we read each item to the pupils and explained the task. After the survey period, we handed over an envelope with a parent questionnaire, asking them to invite their mother (or their primary caregiver) to complete the parent questionnaire. At T2 (one year later) we sent the questionnaires (children and parents) directly to the families and added a stamped return envelope, so that they could send the documents back to us free of charge. Families response rate to the second wave was 97 (43,9%). These 97 families (responding both waves, T1 and T2) were included in the current study. All of the included participants (female n = 46, (47%); M age T1 = 9.7 years (SD) = 0.54) attended the 4th grade of primary school at T1. The current research project is in accordance with the Ethical Principles for Medical Research Involving Human Subjects, formulated by the World Medical Association Declaration of Helsinki (World Medical Association, 2013). There is a positive vote from the medical institutional review board of the Carl von Ossietzky University of Oldenburg.
Measurement
Several questionnaires were used for data collection. Parents completed both (1) questionnaires assessing somatoform complaints and withdrawal behavior of their children and (2) self-report questionnaires assessing overprotecting behavior. Children estimated their use of different ER strategies in a self-report questionnaire.
Emotion regulation
Children’s ER strategies were assessed with the reliable and valid questionnaire of ER strategies in childhood and adolescent, FEEL-KJ (Grob & Smolenski, 2005; 2009). This self-report questionnaire can be applied to assess children’s habitual use of different ER strategies, which occur in response to emotions. More specifically, children, aged 8 to 19 years (Cracco et al., 2015; Czaja et al., 2009) can be questioned about their usual behaviors in case of sadness, anxiety, or anger. Two orthogonal factors, called adaptive and maladaptive ER strategies, were distinguished by explorative factor analyses. In 90 items both adaptive and maladaptive ER strategies can be selected (see Grob & Smolenski, 2005; 2009). The authors Grob and Smolenski (2005; 2009) report good Cronbach’s alpha values for the superordinate scales (maladaptive ER strategies = α = 0.82 and adaptive ER strategies = α = 0.93) in their test manual (Grob & Smolenski, 2005; 2009). Regarding the validity of the questionnaire, the authors report appropriate correlations to both symptoms of anxiety and depression (Grob & Smolenski, 2005; 2009). In the present study we only focus on seven out of 15 scales, which assess adaptive ER strategies including (1) problem solving, (2) distraction, (3) lift the mood, (4) acceptance, (5) forgetting, (6) cognitive problem solving and (7) reappraisal. Children answer the 42 items (14 items for each emotion) concerning adaptive ER strategies on a 5-point Likert-scale (1 = never to 5 = almost always) at both survey waves (α T1 = 0.92, α T2 = 0.96). As an example, an item from the original German-language questionnaires could be translated as follows: ‘When I am angry (sad; afraid) I think about things that make me happy’) (see Grob & Smolenski, 2005; 2009).
Children’s withdrawal behavior and somatoform complaints
To access children’s withdrawal behavior and somatoform complaints the german version of the Child Behavior Checklist CBCL/6-18R questionnaires (reliable and valid parent report) were used (Achenbach & Rescorla, 2001; Achenbach et al., 2008; Döpfner et al., 2014). Overall, the questionnaire assessed eight problem scales (1) Anxious/depressed, (2) Withdrawn/depressed, (3) Somatic Complaints, (4) Social Problems, (5) Thought Problems, (6) Attention Problems, (7) Rule Breaking Behavior, and (8) Aggressive Behavior with a total of 104 Items (Achenbach et al., 2008). From the first three scales mentioned above, the superordinate scale “Internalizing” is summarized. To the second superordinate scale “Externalizing” the above-mentioned scales (7) and (8) are grouped together (Achenbach et al., 2008). Both, “Internalizing” and “Externalizing” show good psychometric properties (α = 0.92, retest reliability = 0.92, Long-term stability = 81) (Achenbach et al., 2008). Parents rated their perceived children’s behavior of the last six month prior to the survey on a 3-point-scale (0 = not true, 1 = sometime/somewhat true, 2 = often true/very true). In the present study, only the 8 items of the scale “withdrawal / depressed” and the 12 items of the scale “somatic complaints” were included in the calculations. The internal consistency values of the scales in this study were α T1= 0.79, α T2= 0.75 for withdrawal and α T1= 0.73, α T2= 0.66 for somatic complaints.
Parental overprotection
In order to collect data concerning the construct of parental overprotection we use a german adaption of the“Parental Overprotection measure; OP” (see Clarke et al., 2013; Edwards et al., 2010). The OP measure contains 19 items on a 5-point Likert-scale from 0 (not at all) to 4 (very much). Authors reported both good reliability (retest and internal consistency α = 0.87) and validity (predictive and construct validity) (see Clarke et al., 2013; Edwards et al., 2010). The parent self-report questionnaire asks about restrictive and controlling parenting behaviors, in situations that parents believe could be threatening to their children (e.g., ‘I do not allow my child to climb trees’). Good internal consistency at both measurement times (α T1 = 0.88, α T2 = 0.90) can also confirmed in the current study.
Data Analyses
To perform descriptive calculations like means, standard deviations, α-coefficients, bivariate correlations SPSS version 27 were used. To verify significant mean value differences to T1 and T2 t-tests for dependent samples were conducted. Both, the hypothesized stability effects and the reciprocal associations between the included variables over time (see Fig. 1), were proofed by cross-lagged panel analysis (Kearney, 2017; Selig & Little, 2012) with AMOS 26.0 (IBM® SPSS®Amos™26, 2019). A non-significant Little-MCAR test (Little, 1988) result (χ2 = 164.23, df = 137, p = 0.056), indicated that the data of interest are missing completely at random in the current study. To verify the MCAR properties and to validate the results of Little’s test, additional logistic regression analyses were conducted (Urban et al., 2016). The results confirm that the relationship between the missing and measured values of all included model variables should be considered as purely random. Due to that missing data were handeled with Full Information Maximum Likelihood (FIML) method. In order to determine the degree of fit between the empirical data of our study and theoretical models, we assess the chi-square (χ2), the Root Mean Square Error of Approximation (RMSEA) and Comparative Fit Index (CFI).
Sample Size Justification
In order to assess whether our sample size is large enough to identify possible effects, we conducted a priori test power analysis (Faul et al., 2007; Faul et al., 2009). We calculated a linear multiple regression, fixed model, single regression coefficient, with five predictors included. Sample size analysis revealed that the required extent of sample size is N = 74 to detect medium effects (f² = 0.15, α err prob = 0.05, power (1-β err prob) = 0.95).
Results
Preliminary Analyses
Mean values, standard deviations and internal consistencies (α) at both timepoints (T1 and T2) as well as the paired t-test results, are presented in Table 1. As presented in Table 1 all values of internal consistencies were acceptable or even better (0.73 ≤ α ≤ 0.96) with exception of the somatoform complaints scale on T2, (α = 0.66). To evaluate possible mean value changes of variables over time a paired t-test was conducted. Results detected small effects (Cohen, 1988) between parental overprotection at T1 and T2 (t (96) = 0.2.24, p = 0.027, d = 0.22), adaptive ER strategies (T1/T2; t (96) = −0.2.01, p = 0.048, d = −0.20) and withdrawal behavior (T1/T2; t (96) = −0.2.71, p = 0.008, d = −0.28). Although there were significantly lower scores for parental overprotection at T2 (M = 22.9, SD = 10.4) compared to T1 (M = 24.5, SD = 10.6), at the same time both adaptive ER strategies (T1 = M = 126.6, SD = 2.5) and children’s withdrawal behavior (T1 = M = 1.40, SD = 2.09) reported significantly higher scores on T2 (adaptive ER strategies = M = 132.6, SD = 3.0; withdrawal behavior = M = 1.8, SD = 2.24).
Table 1
Means, Standard deviations and Internal Consistencies (α) timepoints T1 and T, N = 97 and paired t-test results
Variables
M
SD
SE mean
α
Paired t-test
t-value
df
Sig (two-tailed)
Cohen’s d
T1 Parental Overprotection
24.50
10.57
1.07
0.88
2.24
96
0.027
0.227
T2 Parental Overprotection
22.92
10.41
1.06
0.90
T1 Adaptive ER Strategies
126.60
25.10
2.55
0.92
−2.01
96
0.048
−0.204
T2 Adaptive ER Strategies
132.63
29.74
3.02
0.96
T1 Withdrawal Behavior
1.40
2.09
0.21
0.79
−0.2.71
96
0.008
−0.276
T2 Withdrawal Behavior
1.88
2.24
0.23
0.75
T1 Somatoform Complaints
2.06
2.34
0.24
0.73
0.94
96
0.352
T2 Somatoform Complaints
1.85
2.08
0.21
0.66
M mean; SD standard deviation; SE mean standard error mean; α Cronbach’s alpha; ER emotion regulation; T1 timepoint 1; T2 timepoint 2
As presented in Table 2, bivariate intercorrelations between included variables at T1 and T2 revealed high stability (0.43 ≤ rstability ≤0.78). As expected, significant positive relations between parental overprotection at T1 and both, withdrawal behavior (T1 = r = 0.21, p < 0.05; T2 = 0.34 < 0.01) and somatoform complaints (T1 = r = 0.25, p < 0.05; T2 = r = 0.33, p < 0.01) were found. Contrary to expectations, positive correlations between parental overprotection at T2 and children’s adaptive ER strategies at T2 (r = 0.21, p < 0.05) and withdrawal behavior at T2 (r = 0.27, p < 0.01) were found. In addition, positive associations were found between withdrawal behavior at T1 and somatoform complaints at T1 (r = 0.36, p < 0.01) as well as between withdrawal behavior at T2 and somatoform complaints at T1 (r = 0.33, p < 0.01) and T2 (r = 0.51, p < 0.01) A significant negative correlations between adaptive ER strategies at T1 and withdrawal behavior at T2 (r = −0.25, p < 0.05) as well as negative correlations between adaptive ER strategies at T2 and somatoform complaints T1 (r = −0.38, p < 0.01) were found. There were no significant correlation between child gender and any of the included variables.
Table 2
Bivariate correlations between the included variables and whose autocorrelations over time
Variables
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(1) gender
–
(2) T1 Parental Overprotection
0.100
–
(3) T2 Parental Overprotection
0.115
0.782**
–
(4) T1 Adaptive ER Strategies
−0.156
−0.017
0.055
–
(5) T2 Adaptive ER Strategies
−0.171
0.152
0.211*
0.427**
–
(6) T1 Withdrawal Behavior
−0.102
0.213*
0.141
−0.066
−0.069
–
(7) T2 Withdrawal Behavior
−0.053
0.339**
0.269**
−0.246*
−0.099
0.674**
–
(8) T1 Somatoform Complaints
0.195
0.248*
0.188
−0.124
−0.375**
0.357**
0.334**
–
(9) T2 Somatoform Complaints
0.107
0.327**
0.293
−0.101
0.046
0.137
0.511**
0.503**
*p < 0.05, **p < 0.01, ***p < 0.001; Coding for gender: girls = 2, boys = 1; ER emotion regulation; T1 timepoint 1; T2 timepoint 2
Model Comparison
Considering all included paths of the hypothesized model (Model 1) (see Fig. 1) cross- lagged panel results revealed several non-significant links. Against the background of the assumption that the less constrained model (fewer parameters but instead more degrees of freedom) should be used (if this is theoretically possible), it was decided to create less constrained models (Model 2 and Model 3) while fixing all paths p > 0.1 to zero. In Model 2 only the concerned regression weights were fixed to zero. In Model 3 the concerned covariances were added. After that a nested model comparison was conducted to verify if there are significant differences between the models. Results presented in Table 3, show no significant χ2 difference values, indicating a statistically, equally good model fits of Model 1, Model 2 and Model 3. Model 3 show the smallest AIC index value indicating a comparatively best model fit on the data; Model 3 was used in present analysis which exhibited an overall good model fit (see Table 3).
Table 3
Nested model comparisons between default model (Model 1) and less constrained models (Model 2 and 3) by excluding not significant paths
Model
χ2
(df)
CFI
REMSA
AIC
∆ model
∆ χ2 (df)
Model 1
1.949
4
1.00
0.000
101.949
1 vs. 2
8.508 (10)
Model 2
10.457
14
1.00
0.000
90.457
2 vs. 3
3.097 (7)
Model 3
13.554
21
1.00
0.000
79.554
1 vs. 3
11.605 (17)
Cross-lagged panel analysis
As presented in Fig. 2, we found significant autoregressive effects of the individual variables from T1 to T2 (0.40 ≤ ßstability ≤ 0.78, p < 0.001). The most stable effect (ß = 0.78, p < 0.001) was seen with parental overprotection. Furthermore, parental overprotection at T1 predicted children’ adaptive emotion regulation (ß = 0.26, p < 0.01), withdrawal behavior (ß = 0.19, p < 0.01) and somatoform complaints (ß = 0.22, p < 0.05) at T2 significantly. Regarding the reverse direction, no included variable at T1 (excepted parental overprotection by itself) predicted parental overprotection at T2. Regarding child factors, adaptive ER strategies at T1 predicted withdrawal behavior at T2 into the expected, inverse direction (ß = −0.18, p < 0.01). Somatoform complaints at T1 predicted adaptive ER strategies at T2 inversely (ß = −0.39, p < 0.001). Significant effects of gender only become apparent on somatoform complaints (ß = 0.21, p < 0.05) at the disadvantage of girls. The final cross-lagged panel model is presented in Fig. 2.
Fig. 2
Final model with parental overprotection, adaptive ER strategies, withdrawal behavior and somatoform complaints over time (T1, T2). Gender as moderating variable, coding: girl = 2, boys = 1. Fine dotted lines indicate links fixed to zero. Bold solid lines indicate significant positive relations. Bold dashed lines indicate significant negative relations; *p < 0.05, **p < 0.01, ***p < 0.001
×
Discussion
The aim of the current study was to evaluate longitudinal reciprocal relations and stability between somatoform complaints, adaptive ER strategies, social withdrawal behavior, and overprotective parenting in middle childhood. Based on theoretical assumptions, reciprocal relations between all included factors were expected, resulting in an internal cycle of maintenance and increase of somatoform complaints over time. Cross-lagged panel results only partially supported the aforementioned hypotheses shown in Fig. 1.
Cross-lagged Panel Results between Child Factors
Differently than expected based on prior research (Janssens et al., 2010; Kopala-Sibley & Klein, 2017), no reciprocal relations between the different child factors could be detected. Instead, only significant unilateral direct paths were identified. Children using more adaptive ER strategies on T1 show less withdrawal behavior on T2, whilst withdrawal behavior on T1 does not negatively predict the use of adaptive ER strategies on T2. This result fits with previous research supporting the assumption that the frequent use of adaptive ER strategies promotes children’s social competencies by enabling them to face and cope with difficult social situations in a self-effective and age-appropriate manner (Blair et al., 2015; Calkins et al., 2001). Children’s experiences to cope with difficult situations independently leads to more self-efficacy and self-confidence (Bandura, 1997) and in turn to a decrease of withdrawal behavior (Rubin et al., 2009). In line with these assumptions, results could indicate that the availability of functional strategies in children decreases the need to use withdrawal behaviors as possible avoidance strategy to escape from difficult social situations over time. According to learning theory assumptions (Bandura & Walters, 1977), positive reinforcement mechanisms lead to children’s empowerment experiences. These, in turn, reinforce children’s use of own adaptive ER strategies, which finally promotes age-appropriate autonomous behavior (Webb et al., 2012; Zimmer-Gembeck & Skinner, 2016).
The study results revealed that high somatoform complaints appear to predict low adaptive ER strategies over time. In light of both the theoretical assumptions (Clarke et al., 2013; Janssens et al., 2009; Rubin et al., 2009) and the high correlations between children’s withdrawal behavior and somatoform symptoms (see Table 2), the inverse correlation between somatoform symptoms at T1 and children’s adaptive ER strategies at T2 could be explained. Similar to withdrawal behavior, symptoms of physical discomfort might be associated with children’s experiences of helplessness and self-efficacy limitations (Ejdemyr et al., 2021; van Geelen et al., 2015), which may increase the need for external (parental) support in emotion regulation. As a possible result, the use of their own age-appropriate strategies may be reduced.
Both considerations based on learning theory (positive reinforcement) and secondary gain concept could provide further explanations in this context. One resulting explanation would be that functional impairment, due to physical complaints, is largely understood and accepted by the social environment. Out of understanding and compassion more offers of support are made in the child’s everyday life, which in turn could lead to a reduction of the usage of adaptive ER strategies.
Stability Results of Parental Overprotection over Time
Since the study results show that parental overprotection is stable over time and that it is not influenced by included child factors, a central relevance of individual parental factors could be assumed. This could explain or condition overprotective behavior. Based on both theoretical assumptions and empirical research, different parenting factors can be discussed.
Consistent with assumptions in personality psychology and previous research (see e. g. Belsky, 1984; Huver et al., 2010) parental behaviors may not primarily represent situationally influenced interactional aspects (states), but rather may themselves represent stable parental personality factors (traits) or at least be fundamentally influenced by other personality traits such as neuroticism e. g. associated the trait anxiety (Belsky, 1984; Huver et al., 2010). Parental mental health factors also play a central role in this context. In particular associations between parental psychological disturbance patterns of anxiety and overprotective parenting behaviors could be identified through current research findings (see e. g. Clarke et al., 2013; Chapman et al., 2022; Jones et al., 2021). Furthermore, factors such as parental traumatic experiences play a central role in the scientific discourse with regard to parenting behavior. In this context for example McWhorter and colleagues (2021) described parental post-traumatic stress and overprotective behavior as possible risk factors in child development. Similarly, as early as 1993 Thomasgard and Metz report associations between parents’ unresolved traumatic experiences and overprotective parenting behaviors (Thomasgard and Metz, 1993). The authors focused on attachment theory explanations (Thomasgard and Metz, 1993). Based on this it could be explained why parents with own inner anxiety often intervene in their children’s daily lives with excessive control and overprotective behavior (for more information, see Thomasgard and Metz, 1993, p.77).
Cross-lagged Panel Results between Parental Overprotection and Child Factors
Contrary to assumptions based on the transactional model of development (Sameroff, 2009) and previous studies (Lansford et al. 2011; Lansford et al. 2018; Newton et al. 2014; Rueter & Conger, 1998; Serbin et al.,2015; Zarra-Nezhad et al. 2022), supporting the idea of reciprocal relations between parental and child factors, the study revealed only significant paths in one direction, namely from parental overprotection to child factors (see Fig. 2). The independence of parental overprotection is further underscored by the high autoregressive results over time. Regarding the associations of parent behavior on child factors, results support the assumption that both children’s withdrawal behavior and somatoform complaints were directly affected by parental overprotection. The hypothesis that high levels of parental overprotection on T1 predict high levels of withdrawal behavior and somatoform complaints on T2 could be confirmed (Bornstein et al., 2018; Janssens et al., 2009; Rubin et al., 2009; Walker et al., 2006). Both, learning theory and secondary gain assumptions can be used as a base for plausible explanations regarding this associations (Bandura & Walters, 1977; Fishbain, 1994). As mentioned above, overprotective parents probably relieve their children by excessive and age inappropriate supporting behavior, leading to children’s short-term emotional reliefs caused by positive and negative reinforcement. The same applies to secondary gains for children taking advantage of ‘Acceptable or legitimate interpersonal advantages that result when one has the symptoms of a physical disease’ (Barsky & Klerman, 1983, p. 276) leading to short-term feelings of relief. However long-term perspective shows that parental overprotection favors the decrease of self-efficacy expectations and self-confidence which in turn leads to an increase of social anxiety and withdrawal, somatoform complaints or school avoidance (Clarke et al., 2013; Janssens et al., 2009; Rubin et al., 2009).
Study results also revealed that parental overprotecting on T1 also predicts children’s adaptive ER strategies on T2 but in an unexpected positive direction. It was hypothesized that parental overprotection over time would inversely predict child adaptive ER strategies, as parents overprotect children in their emotion regulation and do not provide age-appropriate support, thus limiting their own ER competencies. As a methodological explanation for the unexpected results a suppressor effect could be discussed (MacKinnon, et al., 2000). The initially non-significant correlation of parental overprotection at T1 and adaptive ER strategies at T2 (see Table 1) could become significant in the path model because the suppressor variable (adaptive ER strategies) reduces the irrelevant variance of the other predictors included in the model (filtering out “data noise”). Furthermore, a possible theoretically based explanation could be that although children learn a number of adaptive ER strategies based on Banduras social learning theory (Bandura & Walters, 1977), they are often constrained by the overprotective behavior of their parents in using these strategies independently. Additionally, the study results (see Fig. 2) could be influenced by possible indirect effects. Thus, a pathway could emerge through which parental overprotection directly predicts somatoform complaints and somatoform complaints in turn negatively influence adaptive ER strategies. More specifically, high levels of somatoform complaints could thus lead to decreasing levels of adaptive ER strategies, which could weaken the direct positive effect. Further research should therefore investigate somatoform symptoms as a mediator variable in this pathway.
Practical Implications
There are two central aspects that emerge from the study results, on the basis of which possible practical implications can be presented. The first aspect focusses on both the interactions among children’s adaptive ER strategies at T1 and withdrawal behavior at T2 and the association between children’s somatoform complaints at T1 and children’s adaptive ER strategies at T2. Therefore, preventive measures and psychotherapeutic interventions for the treatment of somatoform complaints and the resulting impairments in children’s everyday life should include both emotional and social skills training. On one hand, it is important for children to improve emotional competencies and to increase the use of adaptive ER strategies but on the other hand it is equally important to promote social age-appropriate behavior (Izard et al., 2008). More social-competencies in children may contribute and foster, for example a fearless and stress-reduced climate in e. g. social school situations (Kaeppler et al., 2017). If such preventive or therapeutic interventions succeed, an avoidance strategy such as withdrawal behavior can be reduced over time.
The second aspect focusses on the relationships between parental behavior and child factors. An important step is to educate parents about the negative effects of overprotective parenting behavior. For this, parents need to know how much support can be considered age-appropriate at which stage of the child’s development and how much support is functional and necessary (Ungar, 2009). In addition, situations in which overprotective behavior occurs could be identified and reflected upon in parent training sessions. From the insights gained, more appropriate strategies can then be developed and practiced in a further step.
Regarding the aforementioned connection between overprotection and parental factors, it might also be of central importance to discuss and reflect on the parents’ problematic part in the parent-child interaction. Possibly the concept of tertiary gains may play an important role. Based on the definition of Dansak (1973) and Fishbain (1994) tertiary gains are described as ‘gains sought or attained from a patient’s illness by someone other than the patient, usually a family member’ (Fishbain, 1994, p. 265). Based on this assumption it would be important to investigate, whether parents may have any unconscious emotional reliefs when their children avoid social activities and stay at home instead. For example, Fishbain (1994) described one type of tertiary gain as follows: ‘Decrease family tension and keep family together’. For this purpose, possible parental personality trait factors, mental illness or traumatic experiences which influence the construct of overprotective behavior should be identified, reflected, and treated therapeutically if necessary (Chapman et al., 2022).
Limitations
Finally, a number of limitations should be pointed out which weaken the conclusiveness of the reported results. The results can therefore only be considered preliminary. With regard to the reported effect sizes, the partially unsatisfying internal consistency values of the used questionnaire scales have to be criticized. Due to the resulting high proportion of measurement error in the data, there may be a bias in the explained variance. Overlaps in the construct operationalizations of the variables may additionally influence the criteria of selectivity and lead to biased results. Furthermore, there are limitations based on the study design. In further studies, individual and socio-demographic factors of both children and parents should be taken into account in order to control for possible confounders and thus obtain more valid results. For example, it would be of further interest to include children’s temperamental and parental personality factors as well as mental health and/or physical disease symptoms. Another limitation of the study stems from the underlying ad hoc sample, which contradicts the generalization of the study results. Additionally, the sample size should be increased in further research. This need is particularly evident from the low response rate of participating subjects and the high dropout rates from T1 to T2 for the underlying data collection. A priori power analyses yielded a required sample size of N = 543 to determined small effects (f² = 0.02, α err prob = 0.05, power (1-β err prob) = 0.95) reliably (Faul et al., 2007; Faul et al., 2009). Furthermore, limitations due to the two-wave design in cross-lagged panel analysis have to be mentioned (Cole & Maxwell, 2003). In particular, to answer any questions about possible mediations effects between the included variables over time and in turn of possible underlying circling impact structures, at least three measurement times are required. Further studies should also focus on possible intraindividual aspects in addition to the interindividual aspects in children and their parents over time, possibly examined by using latent growth models. Lastly, further restrictions in the current study result from the one-sided questioning of children and parents perceptions by using only self-report questionnaires. The combination of various survey methods in further research would help prevent such methodological limitations.
Conclusion
The study results provide a first insight of the child-parent-interaction processes regarding somatoform symptoms in middle childhood. The results provide an important indication that parental factors, in relation to individual child factors, should be investigated more intensively in further longitudinal studies. As discussed earlier, learning theory principles and the secondary gain concept appear to play a significant role in the development and maintenance of somatoform symptoms in childhood. Therefore, these approaches should be given special consideration in the diagnosis and intervention process of somatoform complaints in childhood, which also means that in addition to child-related interventions, parent-child interaction interventions must also play a central role. However, in order to gain a deeper understanding of both behavioral and analytically based mechanisms of action related to the maintenance of somatoform symptoms, more research should be conducted in this direction.
Finally, all of these further considerations and investigations may be used to develop and implement optimal and evidence-based support and treatment services for affected children and their families.
Compliance with ethical standards
Conflict of interest
The author declares no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Met BSL Psychologie Totaal blijf je als professional steeds op de hoogte van de nieuwste ontwikkelingen binnen jouw vak. Met het online abonnement heb je toegang tot een groot aantal boeken, protocollen, vaktijdschriften en e-learnings op het gebied van psychologie en psychiatrie. Zo kun je op je gemak en wanneer het jou het beste uitkomt verdiepen in jouw vakgebied.