Introduction
Approximately 10% of children have a parent with a chronic medical condition (CMC) and seem to be at an increased risk for persistent stress and adjustment difficulties (Sieh et al.
2010a,
b; Verhaeghe et al.
2005; Visser-Meily et al.
2005). A meta-analysis of 19 studies including a total of 1,858 children of parents affected by CMC revealed that they displayed more internalizing problems such as anxious, depressed and withdrawn behavior and somatic complaints than children with healthy parents or norm groups (Sieh et al.
2010a).
Generally, the body of research on CMC includes illness-specific studies and studies using mixed illness samples and or/self-identified caregivers. The majority of studies focusing on specific diagnoses are cancer studies. Children of parents with cancer show less problem behavior than children in studies investigating other CMC’s than cancer (Sieh et al.
2010a). However, several studies still report particular problems. For instance, a study concluded that children (
N = 27) of cancer patients perceived their risk of developing cancer as significantly higher than controls (Harris and Zakowski
2003). A study of more than 200 adolescents revealed elevated stress symptoms in 21% of boys and 35% of girls (Huizinga et al.
2005b). In addition, these adolescents appeared to communicate less openly with their parents (Huizinga et al.
2005a). A review of 52 studies showed that children displayed signs of anxiety and depression in both qualitative and quantitative research (Visser et al.
2005).
The remaining illness-specific studies have mainly focused on HIV and/or hemophilia, multiple sclerosis, Parkinson disease, rheumatoid conditions and stroke. Children whose mothers were infected with HIV developed more externalizing problems than children whose mothers were not infected (Tompkins and Wyatt
2008). A large study of more than 300 children of parents infected with HIV showed that these children had more life stressors, family conflict and lower self-esteem than controls (Wu et al.
2008). Forehand et al. (
1998) compared 87 African American children of parents infected with HIV to controls, stating that children of parents with HIV had more difficulties in all domains of psychosocial adjustment including problem behavior and reading achievement scores. Diareme et al. (
2006) found that children of mothers with multiple sclerosis exhibited increased scores on domains of emotional and behavioral problems in comparison with controls. Children of parents with Parkinson disease (
N = 77) reported a high frequency of daily hassles affecting their personal life (Dufour et al.
2006). Evans et al. (
2007) examined attachment in children of mothers with chronic pain resulting from arthritis, other conditions or no medical condition, concluding that these children had more insecure attachment than controls. Further, children of parents with stroke frequently exhibited daily hassles (Dufour et al.
2006) and elevated levels of problem behavior even several years post-stroke (Visser-Meily et al.
2005).
Studies with mixed illness samples and/or studies of self-identified caregivers are often retrospective or qualitative of nature. From qualitative research, it can be concluded that anxious symptoms are especially pronounced and oftentimes constitute worries about health-related issues (e.g., Dearden and Becker
2000). Banks et al. (
2002) deducted from their interviews of self-identified carers that academic achievement was relatively poor, explaining that young caregivers frequently leave school right after the minimum leaving age. Two quantitative studies have focused on self-identified caregivers of chronically ill parents. A quantitative study on hundred self-identified caregivers aged ten to 25 years found that caregiving and family responsibilities were highly common. In addition, young caregivers scored lower on life satisfaction than non-caregivers. Their reliance on problem solving and social skills was comparatively low (Pakenham and Bursnall
2006; Pakenham et al.
2006). Houck et al. (
2007) administered quantitative measures to 38 adolescents and found clinical levels of posttraumatic symptoms in one-third of the sample.
In conclusion, children with a chronically ill parent seem to have more adverse outcomes in behavioral, psychosocial and academic adjustment than other children. Although the effects are small across studies, it should not be neglected that adjustment difficulties may pose a threat to a healthy development of these children.
This study responds to several boundaries of previous research. First, researchers have not always corrected for the statistical dependence between members from the same family, leading to violation of the assumption of independence. To our knowledge, only Visser-Meily et al. (
2005) have examined the clustering structure of families with a chronically ill parent, using multilevel analysis. Families share the same environment and children from the same family may therefore score similarly on outcome variables (Tabachnik and Fidell
2007). Consequently, data of siblings within families require researchers to be considerate of the between-subject dependence. The family cluster effect is related to multiple factors of genetics and environment. Second, many studies in the field omitted comparison groups and therefore lack the possibility to draw conclusions about specific characteristics and needs of children with a chronically ill parent. In addition, those studies including a comparison group mostly focused on problem behavior as measured with the Child Behavior Checklist or the Youth Self Report (Sieh et al.
2010a). As far as we know, only Pakenham et al. (
2006) have compared specific characteristics (e.g., caregiving context variables) of children with a chronically ill parent compared to other children. However, this sample is composed by caregivers and non-caregivers with a wide age range. Instead, we focus on adolescents with a chronically ill parent (target group) because they appear to be a particularly vulnerable group (Kraaij et al.
2003; Visser et al.
2005). To our knowledge, no study has treated statistical dependencies between adolescents from the same families, using a mixed illness sample. It is therefore largely unknown on which variables and to what extent teenagers in families with a chronically ill parent differ from teenagers with healthy parents in regard of the strong between-subject dependence in families.
Our first aim is to examine whether adolescent outcomes are rather explained by the family cluster effect than by individual characteristics. Our second aim is to identify group differences in a wide variety of outcome variables by comparing the target group with controls. Following the meta-analysis by Sieh et al. (
2010a), we examine internalizing and externalizing problem behavior. In addition, this study compares caregiving variables and stress and coping variables that may constitute specific characteristics of the target group. Further, we investigate parent attachment and academic achievement (e.g., grade point average) because these variables are understudied. Thus, this study analyzes problem behavior, caregiving characteristics, daily hassles, stress and coping behavior, quality of parent attachment and grade point average. We hypothesize that the target group displays more problem behavior and scores higher on caregiving variables (i.e., frequency of household chores, caregiving responsibilities, activity restrictions and feeling of isolation) than controls. We also presume that the target group reports more daily hassles, more global psychological stress and lower levels of active problem solving and social support seeking than controls. Finally, we suppose that the quality of parent attachment is lower and that the grade point average is worse in the target group compared to controls.
Discussion
This study is, to the best of our knowledge, the first to assess key differences between adolescents with a chronically ill parent and adolescents with healthy parents on a wide array of behavioral, psychosocial and academic outcomes, using a multilevel design that accounts for the family cluster effect. The results show that the target group scored higher on internalizing problem behavior, caregiving variables, daily hassles and stress and lower on GPA than controls. The family cluster effect had a major contribution to the variance of many adolescent outcome variables such as internalizing problem behavior, caregiving variables and quality of parent attachment. Contrarily, the variance of stress and coping variables was mainly due to adolescents’ individual characteristics. Active problem solving coping was entirely explained by individual characteristics.
Regarding problem behavior, the target group reported more anxious and depressive behavior and slightly more somatic complaints in comparison with controls, but the groups did not differ in externalizing problems. An explanation for the predominance of internalizing problems is that these problems counteract against externalizing problems, meaning that adolescents who report high levels of anxious or depressed behavior may inherently avoid aggressive and rule-breaking behaviors. Our effect sizes for internalizing problem behavior are slightly higher than the average effect size of
d = .23 in the field (Sieh et al.
2010a). While we found no significant effect for externalizing problem behavior, the meta-analysis from Sieh et al. (
2010a) reported a significant average effect size of
d = .09 with several studies showing a marked group difference (Hough et al.
2003; Tompkins and Wyatt
2008). A possible explanation is that studies with medium to large effects for externalizing problems focused on a different diagnostic group. Those studies were non-cancer studies and mainly included children of parents with HIV in families characterized by a low socio-economic status. Accordingly, studies showing that externalizing problems were less prevalent in the target group than in the comparison group analyzed a mixed illness sample (Barkmann et al.
2007) or cancer in families with high socio-economic status (Visser et al.
2007).
With respect to caregiving variables, the effect sizes were medium except for caregiving responsibilities, demonstrating the importance of taking outcomes into consideration that can be a specific result of the impact of parental illness. Specifically, the target group appears to be characterized by activity restrictions, feeling of isolation and high frequency of household chores. Our effect sizes for activity restrictions and feeling of isolation were medium like in the study of Pakenham et al. (
2006). On the contrary, our effect size for caregiving responsibilities was small versus an effect size of
d = .70 in the study from Pakenham et al. This discrepancy is probably due to sample differences, that is, Pakenham et al. examined caregivers versus non-caregivers. Self-identified caregivers may inherently score higher on caregiving responsibilities.
Moreover, the target group comparatively experienced more daily hassles that negatively affected their personal life during leisure time and school duties. A study found that a higher frequency of daily hassles predicted problem behavior (Meijer et al.
2008), so daily hassles may be a crucial risk factor for problem behavior in the target group. Similarly, the stress level in the target group was pronounced, which is in line with prior research (Houck et al.
2007; Huizinga et al.
2005b; Verhaeghe et al.
2005).
A positive outcome for the target group was that coping skills seemed to be fairly equal between the two groups. This is in contrast to what was expected based on prior research (Pakenham and Bursnall
2006). It is possible that we did not find a significant effect because we used a questionnaire that may be less sensitive to detect caregiving- or illness-related coping. Nonetheless, the Utrecht Coping List is a valid and reliable measure for active problem solving and social support seeking, and our results clearly indicate that the target group exhibited these skills to the same degree as adolescents who were not affected by parental CMC. The quality of parent attachment did not differ between the groups either. This is a valuable finding in the sense that children of parents with CMC could fall back on their parents for support to the same degree as controls. All the same, it may be argued that children need extra coping skills and require highly secure parent attachment to buffer the risk for adjustment difficulties.
Furthermore, the target group reported worse grades and failed a school year slightly more often than teenagers from the comparison group. Hedges and Hedberg (
2007) affirm that a small effect size of
d = .20 for academic achievement is of policy interest. In our study, the effect size for GPA was medium, which can be regarded as a large difference in the area of academic achievement. Banks et al. (
2002) concluded that adolescents have educational difficulties as a result of caring. Another explanation is that lower academic achievement is mediated by elevated levels of anxiety which are more common in the target group, so especially children with preoccupations and anxieties concerning their parent’s illness perform worse at school (Duchesne and Ratelle
2010). Noteworthy, academic achievement is the main route into unemployment and adolescents with a chronically ill parent should not be at a disadvantage (Dearden and Becker
2000). The target group followed intermediate vocational education decisively more often and high school less often than controls, while there was no group difference between parents’ educational level. Davis-Kean (
2005) affirmed that parents’ and children’s educational level highly correlate, suggesting that parents’ expectations influence children’s academic choices. In our study, no such correlation was found, so it can be assumed that adolescents with a chronically ill parent are rather influenced by their personal needs or by economic needs of the family than by parents’ education.
This study had some limitations. First, the sample was select considering that the ethnic composition was largely Caucasian, which may limit the generalizability of our results. In addition, only CMC’s combined with impairments were included and some conditions were excluded, for example, cancer and HIV. Consequently, it should be avoided to make general statements about children of specific illness groups. Second, we exclusively examined group differences, hence, no statements can be made about predictors for adjustment difficulties and generally, no causal relationships can be derived from growing up with a parent who has been diagnosed with CMC. Third, the large amount of variables investigated could have resulted in type 1 errors, meaning that the chance to find significant effects augments with the number of variables. Several effect sizes are medium and implicate a lower chance of type 1 errors, while other effect sizes are lower than Cohen’s
d = .3 and should be interpreted with more caution. Notwithstanding, Durlak (
2009) argues that even small effect sizes can be considerate depending on the research area. In addition, we believe that the chance for type 1 errors is unlikely because we found significant effects for the majority of variables. Besides, we categorized several variables within domains (e.g., caregiving) and found effects for most domains in consistence with the literature that demonstrates that parental illness has impact on children’s caregiving outcomes and internalizing problems (e.g., Barkmann et al.
2007; Pakenham et al.
2006; Sieh et al.
2010a). Fourth, our sample had a wide age range. This may be an issue because the developmental variability could have played a large role, so differences within certain age groups could be smaller or larger than what we found for the whole sample. Thus, conclusions with respect to pubertal stages and narrow age groups cannot be drawn. Finally, we were not able to investigate which specific factors accounted for the variance explained by the family cluster effect. A follow-up study should also assess variables at the family level (e.g., family cohesion) to be entered into the analyses. Future research needs to replicate this study with a culturally diverse sample or focus on non-western cultures. It should be noted that illness is not defined as a medical condition in some cultures. Parents with CMC from different cultures with similar symptoms may have distinct illness-related interpretations and coping systems, which in turn may affect child outcomes in different ways (Helman
2007).
In conclusion, this study shows that the family cluster effect explains a large proportion of adolescent outcomes and illustrates distinct features of adolescents with a chronically ill parent. The target group appears to display more internalizing problems compared to adolescents with two healthy parents. They seem to perceive adverse effects of caregiving and experience daily hassles and stress. In addition, their GPA is comparatively low. Hence, growing up with a chronically ill parent appears to pose a risk for behavioral, psychosocial and academic problems of adolescents. Professionals and teachers dealing with children should consider the possibility of parental illness and accordingly, be alert to signs of fears, depressed mood, somatic complaints, isolation and academic underachievement. In congruence with a family-centered approach (Gorter et al.
2010), we recommend professionals to communicate with parents and children about the diagnosis and the short and long term impact of parental illness on the family. Parents should be encouraged to make sure that their children have enough illness-related information and know how to deal with the medical condition. Considering that clinical levels of stress poses a threat to a substantial part of the target group (Houck et al.
2007), clinicians and researchers should collaborate to create evidence-based interventions aiming to reduce stress in this specific group. An intervention that may specifically help the target group is stress management which particularly makes use of strengths and important resources such as coping skills (Murray and Pizzorno
2006). Negative consequences of caregiving responsibilities and illness demands on child development may be a major concern. In this respect, it should be evaluated whether children need assistance for caregiving tasks, leaving sufficient time for their leisure and school activities.