Abstract
Variability within the literature investigating typically-developing siblings of children with an autism spectrum disorder suggests that the quality of sibling outcomes may depend on specific factors. For this study, 42 parents of a child with an autism spectrum disorder and a typically- developing sibling provided data via online questionnaires. Birth order rank of the child with an autism spectrum disorder significantly moderated the relation between externalizing behaviors in children with an autism spectrum disorder and externalizing behaviors in their typically-developing siblings. Children with an autism spectrum disorder and higher levels of behavior problems had typically-developing siblings with higher levels of behavior problems only when the child with an autism spectrum disorder was older. These results provide a hint of clarification about the complex nature of sibling relations, but a great deal more research is needed to further examine outcomes of typically-developing siblings of children with an autism spectrum disorder.
Although the number of typically-developing siblings of children with an autism spectrum disorder (ASD) is likely on the rise along with the increase in prevalence of the disorder (underscoring the importance of considering sibling outcomes), the literature remains mixed: some studies suggest siblings show maladjustment, some show no differences, and others suggest siblings may experience developmental benefits (Meadan et al., 2010). Behavioral similarities within sibling dyads are predicted by both biological theories (e.g. the Broader Autism Phenotype; Bolton et al., 1994) and environmental theories (increased burden on families, especially when problematic associated features are present with ASD, leading to increased risk of parental stress and distress; Fisman et al., 2000). Empirical evidence supporting these theories suggests that typically-developing siblings of children with an ASD—particularly if those children also show elevated levels of behavior problems—may be more at risk of behavior problems themselves (e.g. Hastings, 2003, 2007).
Nevertheless, research examining the impact of behavior problems among children with an ASD on typically-developing siblings remains inconclusive and identification of specific risk factors for maladjustment is needed (Meadan et al., 2010). Meadan et al. indicate that variability in findings may partly be explained by family structure and emphasize a need for additional research to identify specific predictors—and possible moderating variables—of the adjustment of typically- developing siblings. A multitude of variables thought to be related to sibling adjustment have been examined, such as family socioeconomic status (Fisman et al., 2000), family conflict (Mandleco et al., 2003), perceptions of parental favoritism (Wolf et al., 1998), and parental mental health (Fisman et al., 2000), among others; however, the specific relations of these variables to sibling outcomes remain uncertain. Evidence supporting a diathesis–stress model has been found, suggesting that behavior problems of children with an ASD, parental distress, stressful life events, and subclinical symptoms of ASDs in typically-developing siblings (i.e. symptoms related to the Broader Autism Phenotype) likely interact in their relation to maladjustment of typically-developing siblings (Orsmond and Seltzer, 2009; Petalas et al., 2012). In line with this model, it has been shown that behavior problems in typically-developing siblings may fluctuate according to the severity of the behavior problems in the child with an ASD (e.g. Hastings, 2003, 2007; Petalas et al., 2012), and younger siblings appear to exhibit more maladjustment (Petalas et al., 2009; Verté et al., 2003).
Researchers suggest that younger siblings may experience benefits or disadvantages related to their birth order placement (Downey, 2001; Gass et al., 2007). One proposed theory suggests that younger siblings may be at a disadvantage due to a dilution of family resources (Downey, 2001); however, the literature examining this dilution effect remains mixed (Lawson and Mace, 2010), with some studies not fully exploring its potential impact on the relations among family variables. Certainly, dilution of resources for younger children may be more evident in families when the older child has special needs, such as families with a child with an ASD. That is, any negative outcomes may be compounded and benefits reduced for younger siblings of children with an ASD, resulting in less appropriate coping behaviors and more behavior problems (Azmitia and Hesser, 1993). Therefore, birth order rank of children with an ASD (i.e. relative to their siblings) may serve as a possible moderator in the relation between behavior problems in children with an ASD and their siblings. Examination of this potential interaction was the focus of this study. Specifically, it was predicted that birth order would serve as a moderator, such that the relation between externalizing problems in children with an ASD and externalizing problems in their typically-developing siblings would be stronger when children with an ASD were older than the typically-developing siblings.
Method
Participants
Parents of at least one child with an ASD and a fully, biologically related, typically-developing sibling provided data on 42 children with an ASD (33 males and 9 females) aged 8–18 years (M = 12.81, standard deviation (SD) = 3.02) and their typically-developing siblings (17 males and 25 females) aged 6–18 years (M = 12.33, SD = 2.98). The child with an ASD from each sibling pair was independently diagnosed with a pervasive developmental disorder (PDD) (45% with autism, 29% with Asperger’s disorder, and 26% with PDD–not otherwise specified (NOS)) according to the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV). The racial distribution of the current sample was 93% Caucasian, 2% African American, 2% Latino, and 2% Other. Parent ages ranged from 32 to 58 years (M = 43.62, SD = 5.66); 97.6% were mothers.
Measures
Parents completed a Demographic and Diagnostic Questionnaire, which included confirmation of an ASD diagnosis for one child via information on diagnostic classification, age of diagnosis, professional affiliation of person who made the diagnosis (i.e. to rule out self-reported diagnosis), and other related information. Demographic information was gathered on both children, and birth order was coded into a dichotomous variable, where 0 = the child with an ASD was younger than the typically-developing sibling (or the same age in the rare case of twins) and 1 = the child with an ASD was older than the typically-developing sibling. Among the 42 pairs, 20 children with an ASD were younger than their sibling and 22 children with an ASD were older than their sibling. Other demographic variables (e.g. gender of either child) did not relate to typically-developing sibling externalizing behaviors, and therefore, were not used as control variables.
Parents completed the Child Behavior Checklist for ages 6–18 (CBCL/6–18; Achenbach and Rescorla, 2001), a broadband measure of child psychopathology, for the child with an ASD and for the sibling. A license was purchased that allowed its online use. For this study, T-scores for the Externalizing Problems Composite for both children (ASD and typically-developing sibling) in each pair were the variables of interest. The CBCL Externalizing Problems Composite has previously shown high test–retest reliability, internal consistency, and validity (Achenbach and Rescorla, 2001). High internal consistency held true for the current sample (for the whole sample, α = 0.88, and for the children with ASD and typically-developing siblings separately, α = 0.86 and 0.91, respectively). Parents also completed the Children’s Social Behavior Questionnaire (CSBQ), a measure of autism symptom severity (Luteijn et al., 2000; current sample α = 0.91). Children with an ASD in this study obtained an average CSBQ total score of 37.38 (SD = 15.48), which is consistent with previous clinical samples (means ranging from 33.64 to 47.22; SD = 14.51–15.94); Hartman et al., 2006), providing evidence that their symptoms were in a clinically significant range. This scale did not relate significantly to typically-developing sibling externalizing behaviors, and therefore, was not used as a control variable. Parents also completed a measure of parenting stress and a measure of parenting practices to determine if either of these parenting constructs should be controlled in the analysis; however, the constructs did not relate to typically-developing sibling externalizing behaviors and were not included in the analysis.
Procedure
Approval from the University Institutional Review Board was obtained before any data were collected. Parents were recruited via email to various autism support groups’ LISTSERVs, as well as through a snowball sampling technique, and had been participants in previous ASD studies in the lab. Once parents agreed to participate, they were sent an email with instructions and their own unique link to a secure website containing the measures. Families with multiple children within the study’s inclusionary criteria were asked to choose the typically-developing sibling closest in age to their child with an ASD. Electronic consent was obtained before completion of any questionnaires.
Results
Zero-order correlations were conducted between possible control variables and the outcome variable, and as noted earlier, none were significant and no controls were deemed necessary. The hypothesis of this study was tested using a moderated multiple regression analysis (Baron and Kenny, 1986). To reduce multicollinearity and to aid with the interpretation of post hoc plots, the continuous predictor was centered before creating the interaction term with the dichotomous moderator (Holmbeck, 2002). An interaction term was created by multiplying externalizing problems of children with an ASD (continuous variable) and birth order rank of children with an ASD (dichotomous variable) and was entered on the Step 2 after the main effects were entered on Step 1. Although Model 1 was not significant, R2 = 0.11, p = 0.11, birth order rank for children with an ASD did emerge as a significant unique main effect of sibling externalizing problems, β = 0.32, t = 2.08, p = 0.045 (the positive β weight indicated that a higher birth order rank (i.e. coded 1, which is when the child with an ASD was older than the typically-developing sibling) was associated with higher levels of sibling externalizing behavior problems). Externalizing problems of children with an ASD were not a significant unique predictor of the siblings’ externalizing problems, β = 0.05, t = 0.30, p = 0.76. Model 2 was significant, ΔR2 = 0.10, p = 0.04, showing that the interaction term was a significant predictor of sibling externalizing problems, β = 0.38, t = 2.13, p = 0.04. More specifically, as shown by the post hoc plot of the interaction in Figure 1, children with an ASD who had higher levels of behavior problems had typically-developing siblings with higher levels of behavior problems only when the child with an ASD was older.
Interaction between the externalizing behavior problems and birth order rank of children with an ASD and the externalizing problems of their typically-developing siblings.
Discussion
Results supported the hypothesis; birth order rank of children with an ASD moderated the relation between externalizing behaviors in children with an ASD and externalizing behaviors in their typically-developing siblings. Typically-developing children with an older sibling with an ASD who exhibits high levels of externalizing behaviors appear to be at greater risk of exhibiting high levels of externalizing behaviors themselves. This finding is consistent with theories of social learning and modeling found in literature exploring sibling relations among typically-developing children (Azmitia and Hesser, 1993) and with other research suggesting that typically-developing children born after their sibling with an ASD may be at increased risk for maladjustment (Petalas et al., 2009). Interestingly, this study builds on previous literature by showing not only that birth order rank of the child with an ASD is a predictor of the sibling’s externalizing behavior but also that birth order rank serves as a moderator between the two siblings’ externalizing behavior. Such information can be used to aid in the identification of siblings of children with an ASD who may be at particular risk for maladjustment and to improve resources for families to ameliorate such maladjustment for siblings. When this dynamic is present within a family, clinicians may find it beneficial to include the typically-developing sibling in treatment planning. While addressing such needs may lead to more positive outcomes for the whole family, further research on sibling outcomes is clearly needed (e.g. Meadan et al., 2010).
A number of limitations should be considered. The sample was small and homogenous, and data were collected cross-sectionally via a sampling strategy (email on LISTSERVs, snowball sampling) that could lead to a sampling bias. A larger, more diverse sample using a longitudinal design would provide additional understanding of sibling outcomes related to birth order. Furthermore, all data were collected online from a single informant, and ASD diagnoses were not independently verified. Future research may be strengthened by using multiple informants (including sibling self-report) from multiple settings and by directly assessing the children with ASD. Despite these limitations, the results of this study provide a snippet of important information about the complex nature of typically-developing sibling outcomes when an ASD is present. Typically-developing children born after siblings with an ASD who exhibit high levels of externalizing behaviors may be at a greater risk for behavior problems themselves. Still, future research is needed to further examine the role for birth order rank and other possible factors that influence typically-developing sibling outcomes. Indeed, given the mixed findings of sibling outcomes across research, more sibling studies examining possible moderators are needed to better develop our understanding of potentially higher risk subgroups of siblings. Likewise, more sibling studies examining possible mediators are needed to identify possible points of intervention once those high-risk siblings are identified.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.