Introduction

Parents of a child with Autism Spectrum Disorder (ASD) are reported to suffer from elevated stress compared to parents of children without ASD (Bouma and Schweitzer 1990; Mugno et al. 2007; Sharpley et al. 1997) or parents of children with other developmental disabilities (SA Hayes and Watson 2013). These data suggest that it may be the nature of ASD itself that is particularly stressful to these parents. Several studies have attempted to identify the ASD-related factors that are most likely to contribute to parental stress, including severity of the child’s ASD symptoms, level of functioning, child’s age, adaptive behaviour and the self-reported stress levels of their parents (Bebko et al. 1987; Davis and Carter 2008; Lecavalier et al. 2006; Rivard et al. 2014). The most commonly used measure of parent stress in this literature has been the Parenting Stress Index (PSI) (Loyd and Abidin 1985), a 36-item self-report scale for parents, which is comprised of three subscales (Parental Distress, Parent-Child Dysfunctional Interaction, and Difficult Child Characteristics). These subscales measure (in turn) such aspects of parenting as: the distressing changes that have occurred to the parent’s life since having a child, whether the parent feels that their child likes them, and the problematic behaviour of their child.

While this research focus is satisfactory for the purpose of identifying stress in these parents, the three areas measured by the PSI subscales do not closely relate to the kind of symptomatology that is associated with clinically-significant anxiety and depression. That is, although stress is unpleasant, clinical levels of anxiety or depression are classified as mental illnesses by the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (APA 2013) and are more serious than stress per se. Anxiety and depression also have unwanted sequalae, such as increases in physical disease, relationship problems and cognitive difficulties (Nutt 2004) as well as elevated risk of suicide (Malone et al. 1995; Zimmerman et al. 2000). Anxiety can be a precursor of low-level illnesses (Fries et al. 2005) and may precede emotional and behavioural problems such as demoralization, hostility and mistrust (Langewitz and Ruddell 1989), as well as elevated risk of coronary heart disease (Bosma et al. 1997) and depletion of the immune system (S. Cohen et al. 1991). Depression is a major contributor to the total disease burden (Ustun et al. 2004) and has greater adverse effects on personal health (Moussavi et al. 2007) and higher costs of care (Langa et al. 2004) than other chronic diseases. It is also associated with suicide in about 15% of all depressed patients (APA 2013) and carries a similar risk for mortality from all causes as smoking does, even when related health factors such as blood pressure, alcohol intake, cholesterol and social status are taken into account (Mykletun et al. 2009). Recent meta-analytic data indicate that people with depression have a relative risk of mortality from all causes that is 1.86 times that for non-depressed individuals and that there are 2.74 million deaths annually from depression (Walker et al. 2015). Clearly, anxiety and depression represent problems of great magnitude and urgency, and the continued exploration of the levels of these disorders, plus their possible correlates, in parents of a child with ASD can provide further information relevant to the delivery of services to these parents and (perhaps) the quality of their interactions with their child who has ASD.

In terms of the possible ASD-related correlates of parental anxiety and depression, it is relevant to define Autism Spectrum Disorder (ASD) by a triad of difficulties in communication, social interaction, and repetitive and restricted behaviour (APA 2013). These three areas of difficulty may vary between individual children with ASD and may be measured via a range of instruments. In this regard, it is pertinent to obtain the parents’ reports of their children’s ASD-related difficulties rather than those from a third party because it is the parents’ perspective on those difficulties that is most likely to be correlated with their own anxiety or depression. Similarly, parents’ self-reports on their own anxiety and depression are also the logical source of information in studies such as these which rely upon the perspective of the person under examination (i.e., the parents).

Therefore, this study focussed upon the association between the parent-reported ASD symptoms of their child and their own self-reports of anxiety and depression. Because they represent the most comprehensive and/or common forms of anxiety and depression, Generalised Anxiety Disorder (GAD) and Major Depressive Disorder (MDD) were chosen as the target variables in this study. To avoid possible confounds due to age or gender, and because previous studies have highlighted the presence of parental stress during their child’s early adolescence (e.g., Compas et al. 1989), the sample was restricted to mothers of boys with ASD who were aged 10 to 15 years.

Method

Participants

A total of 68 mothers, each with a son (M age = 12.6 yr., SD = 1.6 yr., range = 10 to 15 yr) with ASD was recruited from the Gold Coast, Queensland, Australia. Written informed consent to participate was provided by these mothers and the study was approved by a university research ethics committee constituted according to the Helsinki Declaration of 1964 and recent amendments. All of the sons of these mothers had been diagnosed with ASD via clinical interviews with a paediatrician or psychiatrist, and confirmed by a clinical psychologist. Symptoms of ASD were referenced to the developmental history of the participants and the social context in which their ASD symptoms occurred and all diagnoses were confirmed by behavioural observation. These initial diagnoses were conducted some time prior to this study and therefore they were confirmed by administration of the ADOS by a research-reliable staff member during recruitment for this study. All these boys were also found to have an IQ of at least 70 according to the WASI-II during recruitment (M = 94.3, SD = 12.2, range = 74 to 128). None of the mothers reported that they were currently taking medication for GAD or MDD or that their sons had been formally diagnosed with GAD or MDD. On the basis of their IQ, plus reports on their self-management ability and the fact that all the boys were attending mainstream schools, the sample was classified as “high-functioning”.

Instruments

Depression and Anxiety

Depression was assessed via the self-report Patient Health Questionnaire-9 (PHQ9) (Kroenke et al. 2001). The PHQ9 was developed from the diagnostic criteria for Major Depressive Disorder (MDD) from the DSM-IV-TR (APA 2000) (these have not changed in the DSM-5 (APA 2013)), and has been shown to possess excellent validity (Kroenke et al. 2001) as well as specificity and sensitivity above 95%. Possible total scores on the nine items of the PHQ9 range from 0 to 27, with cutoff points of 1 to 4 (signifying a rating of “none”), 5 to 9 (“Mild”) , 10 to 14 (“Moderate”), 15 to 19 (“Moderately severe”) and 20 to 27 (“severe”) (Spitzer, Kroenke, Williams, & Group, 1999). Anxiety was measured by the GAD7 (Spitzer et al. 2006), which was developed from the DSM-IV-TR (APA 2000) diagnostic criteria for Generalised Anxiety Disorder (GAD) (unchanged for the DSM-5 (APA 2013)) and which has specificity and sensitivity of .80. Scores on the seven items of the GAD7 range from 0 to 21. Cutoffs for the GAD7 are: 0–4 = “minimal”, 5–9 = “mild”, 10–14 = “moderate” and 15–21 = “severe” anxiety (Spitzer et al. 2006).

Sons’ ASD Symptoms

Although several measures are commonly used to diagnose ASD (e.g., the Autistic Diagnostic Observation Schedule (ADOS) (Lord et al. 2012) and the Childhood Autism Rating Scale (CARS) (Schopler et al. 1999), they are most commonly used to determine the initial presence of ASD for the purpose of a formal diagnosis of ASD rather than feeding information about the frequency and severity of specific ASD symptoms directly into detailed treatment-planning that is based upon a child’s behaviour within particular contexts. Following the initial diagnosis, clinicians then need to plan for treatment that is based upon the specific symptoms that are exhibited by a child in particular settings and that are derived from the more global DSM-5 diagnostic criteria. The Autism Spectrum Disorder Behaviour Checklist (ASDBC; Bitsika 2000) is a thirty-item rating scale used to identify the presence of a wide range of autism-based symptoms in each participant’s behavioural repertoire as assessed by parental report or clinical interview. Content validity of the ASDBC was established by reference to the three major areas of impairment specified in the DSM-IV (APA 2000) for Autistic Disorder and Aspergers Disorder (i.e., Communication Behaviours, Social Interaction Behaviours and Restricted/Repetitive Behaviours), each of which is measured by a subscale of 10 items on the ASDBC. These symptoms remain relevant for the collapsed diagnosis of ASD in the DSM-5 (APA 2013) although the three areas have been reduced to social communication and interaction, plus restricted and repetitive behaviours in the DSM-5. Each ASDBC item is answered as 0 (“never”), 1 (“rarely”), 2 (“sometimes”) or 3 (“mostly”), so that total scores for each of the three areas of impairment range from 0 to 30 on each of three subscales, and the total ASDBC score ranges from 0 to 90. Criterion validity was shown by a correlation of the total ASDBC score with the scores obtained on the Childhood Autism Rating Scale (r = .705, p < .001) and Adaptive Behaviour Composite scores from the Vineland Adaptive Behavior Scale (r = −.604, p < .001) in a previous study (Bitsika et al. 2008). Reliability of the ASDBC was assessed in that study via Cronbach’s Alpha (.774), considered as satisfactory (Pallant 2007).

Procedures

The mothers of the boys with ASD were given a questionnaire booklet that included the PHQ9 and GAD7 to answer about themselves, and the ASDBC about their son’s symptomatology. These mothers also received an Information Statement and a Consent Form, plus a verbal explanation from a research staff member of the procedures required to complete the questionnaires in confidence and at the same time. These questionnaires were collected by research staff at the end of that week and checked to ensure that all questions had been answered.

Statistical Analyses

SPSS 23 was used to obtain descriptive data. Pearson correlation coefficients were used to test for significant associations between IQ, GAD7, PHQ9 and ASDBC scores, and to explore the relationships between the ASDBC subscales and the mothers’ GAD7 and PHQ9 scores. Hierarchical regression identified the primary ASDBC subscale contributors to GAD7 and PHQ9 scores. MANOVA on subgroups of the mothers divided according to their GAD7 and PHQ9 severity identified those ASDBC items which had significantly different scores across the severity subgroups. Partial eta squared values were used to formulate a graphic map of the associations between the sons’ ASD symptoms and their mothers GAD and MDD severity status.

Results

Table 1 presents the descriptive data for the GAD7, PHQ9, and the ASDBC subscale scores. There was little evidence of skewness or kurtosis, and inspection of the histograms indicated no need for transformation of the data (Tabachnik and Fidell 2013). Application of the cutoff points for the PHQ9 and GAD7 allowed division of the sample into severity categories. There were no severity indicators for the ASDBC. The Pearson correlation between GAD7 and PHQ9 scores was .729, p < .001. The correlations between ASDBC subscale and total scores were between .761 and .857 (all p < .001). Although not a focus of this study, it is always of interest to check if IQ is a confound in the associations between variables such as those examined here. There were no significant correlations between WASI-II IQ and any of the GAD7, PHQ9 or ASDBC scores. The only significant correlation between parents’ GAD7 and PHQ9 scores and their ASDBC ratings of their child was for GAD7 and ASDBC total scores.

Table 1 Descriptive data for PHQ9, GAD7, and ASDBC subscales for 68 mothers of boys with ASD
Full size table

To investigate the possible association between ASD symptoms (as measured by the ASDBC) and parental anxiety and depression, an initial set of correlation coefficients was derived and is shown in Table 2. These results suggested that further investigation of the ASDBC Social Interaction subscale was not justified, and that the ASDBC Restricted & Repetitive subscale had the strongest relationship with the measures of anxiety and depression, followed by the ASDBC Communication subscale.

Table 2 Pearson correlation coefficients (and significance values) for ASDBC subscales and GAD7 and PHQ9 total scores
Full size table

Therefore, to identify the relative amount of variance each of these two aspects of ASD symptomatology contributed the mothers’ anxiety and depression, two hierarchical regressions were performed using GAD7 and PHQ9 scores as the target variables and the two ASDBC subscales as contributor variables. Each of the two regression equations was significant but each also showed that the ASDBC Communication subscale did not significantly add to the contribution made to the variance for each target variable that was made by the ASDBC Restricted & Repetitive Behaviour subscale. Table 3 shows these results.

Table 3 Hierarchical regression results for GAD7 and PHQ9, using ASDBC communication and restricted & repetitive behaviour subscales
Full size table

These findings indicate the relative influence that each of the three groups of ASD symptoms represented by the ASDBC subscales had upon mothers’ GAD and MDD, and that it was their sons’ ASD-related behavioural symptoms that were most powerfully related to the mothers’ GAD and MDD, but they do not identify the specific behaviours exhibited by their sons that were most powerful in exerting that influence upon the severity of mothers’ anxiety or depression. To enable that comparison to be made, MANOVA was conducted on the GAD7 and the PHQ9 separately for the 10 ASDBC Restricted & Repetitive Behaviour subscale items. Two subgroups of GAD and MDD severity were formed by recoding mothers according to the cutoff criteria set down by the authors of the GAD7 and the PHQ9 and described in the Methods. That is, mothers were divided into those who reported “minimal” or “mild” GAD (n = 55) or MDD (n = 45) versus those who reported “moderate” or “severe” GAD (n = 13) or MDD (n = 23). As might be expected within a community sample, the lower severity subgroups were larger than the more severe subgroups. Although there are some drawbacks to performing MANOVA on different sized subsamples due to ambiguity in assigning the overlapping sums of squares to sources, thus producing a nonorthogonal design, Tabachnik and Fidell (2013) argued that use of the Sum of Squares Method 1 in SPSS GLM can allow the procedure to test for MANOVA effects using an unweighted-means approach that treats each cell as being of equal size. In addition, Pillai’s Trace is more robust than Wilks’ Lambda when sample sizes are unequal, and so these procedures were used here. Each of the two MANOVAs had significant main effects with robust effect sizes (partial eta squared): GAD7 F(10,57) = 2.467 (Pillai’s Trace), p = .015, η 2 = .303), PHQ9 F = 2.780 (Pillai’s Trace), p = .007, η 2 = .328. Significant univariate effects were found for both MANOVAs, but there was some difference in the ASDBC items identified in these univariate effects, as shown in Table 4.

Table 4 Univariate results for major effects of ASDBC restricted & repetitive behaviour subscale items on GAD7 and PHQ9 severity categories
Full size table

Table 4 presents those ASDBC items that had effect sizes that were described as being of “medium” strength by Cohen (1988), i.e., with a partial eta squared of .5 or greater, which were examined because of the exploratory nature of this study. One of those ASDBC items for the GAD7 that is shown in Table 4 did not reach traditional statistical significance levels of .05 or less but, as noted by Tabachnik and Fidell (2013), exploratory studies may be more valuable if results that approach traditional significance are reported for the benefit of future confirmation studies. It is apparent (i) that there is one common and powerful contributor to GAD7 and PHQ9 variance in the form of the sons’ intense responses to stressors, but (ii) that the remaining contributors are different for GAD and MDD. These findings are described in graphic format in Fig. 1, using the effect size (partial eta squared) as an indicator of the strength of the relationship between the various ASDBC items and GAD/MDD severity, and provide a clearer presentation of the ways in which various ASD symptoms were associated with GAD and MDD in this sample of mothers and their sons with ASD.

Fig. 1
figure 1

Sons’ ASD symptoms contributing to mothers’ GAD and MDD

Full size image

Discussion

These findings add to the understanding of the ways in which their child’s ASD-related symptoms influence the self-reported anxiety and depressive status of parents, particularly during a period of development that is associated with elevated child-parent tension (i.e., early adolescence) in many families. The limitation of that association to items measured by the Restricted & Repetitive Behaviour subscale of the ASDBC (rather than Communication or Social Interaction subscales) narrows the range of ASD-related behaviours that proved to be distressing to these mothers. Further, the results of the MANOVAs clearly identified that their son’s intense reactions to stressors was the most powerful ASD-related contributor to the GAD and MDD experienced by these mothers. It may be that these intense reactions represented behaviour that the mothers found worrying, frightening, extremely saddening, or otherwise distressing, and the further exploration of the exact relationship between this aspect of their sons’ ASD-related symptoms and their mothers’ GAD and MDD status would be of value in understanding those pathways and for potentially developing strategies for mothers to avoid these unfortunate outcomes. That is, no causal relationships can be inferred from these data and it is necessary to develop alternative research models (e.g., prospective studies) to identify if there is such a causal relationship in effect. Qualitative research methods might also elucidate these findings.

An additional similarity in the sons’ ASD-related behaviour across GAD and MDD was the symptoms of the boys being unable to anticipate the consequences of their actions (contributing to GAD variance) and their discrepant reactions across different contexts. Both of these are linked with the lack of social insights that people with ASD sometimes exhibit because both refer to the mothers’ anxiety and depression at being unable to predict how their sons will respond in different situations. As indicated by the correlation between GAD7 and PHQ9 scores, accounting for over 56% of the variance, anxiety and depression share some common symptoms (Zinbarg et al. 1994) and are often comorbid (APA 2013). These data support that connection between these two disorders in that the presence of both disorders in some of these mothers was significantly correlated with two of the same ASD-related symptoms in their sons.

Although the correlation between the GAD7 and PHQ9 was significant and robust, it did not explain all the variance, and so it is reasonable to focus upon these aspects of ASD-related behaviour that were not consistent correlates of both GAD and MDD. The different ASD-related symptoms that were associated with mothers’ MDD, but not GAD, are indicated by the two ASDBC items measuring a focus upon only certain aspects of tasks or objects, and a lack of consistency in reactions and skills in different settings. These might be loosely grouped into a more generalised repetitive response behavioural repertoire. That is, although some ASD-related behaviours measured by the ASDBC were common correlates of both GAD and MDD, presence of the latter disorder at moderate or severe levels in these mothers was also characterised by the presence of two specific rigid and restrictive behaviour patterns by their sons.

This difference in contributors (apart from the boys’ intense reactions to stressors and inability to understand the social consequences of their actions or to maintain consistency in those reactions across situations) is consistent with the different content of the GAD and MDD symptomatologies themselves. That is, GAD is largely defined by uncontrollable worry, plus physiological reactivity (APA 2013), whereas MDD is defined by sadness and/or anhedonia, plus some physiological and cognitive symptoms (APA 2013), reflecting a lack of responsivity (i.e., behavioural withdrawal), as noted by Bolling et al. (1999) and others (Dougher and Hackbert 1994; Ferster 1973; Kanter et al. 2004). Thus, GAD and MDD represent different sets of responses to long-term stressors, linked by physiological pathways (Sharpley and Bitsika 2010). Elevated GAD has been shown to increase the risk of developing MDD (Muris et al. 2001), and it may be that their sons’ inability to anticipate the consequences of their own actions/remain consistent across settings might trigger mothers’ GAD but also could contribute to their later development of MDD, during which time their sons’ rigid response pattern symptoms become more distressing for these women. These comments are conjectural at this stage and require longitudinal data to test their validity, but they do provide a possible pathway between the ASD-related symptoms shown by their sons and the GAD/MDD status of these mothers.

Therefore, although further data are needed before firm recommendations can be made for assisting mothers of teenage boys with ASD to avoid GAD and MDD, there are some initial suggestions that might be of possible use when developing training programmes for these parents. For example, helping parents to become aware that their child’s intense reactions to stressors (Corbett et al. 2006), inability to anticipate the consequences of their own actions (Fabbri-Destro et al. 2009), and different reactions across contexts and rigidity (D'Cruz et al. 2013) are aspects of ASD (APA 2013) rather than intentional misbehaviour, and could be a first step in helping them understand why their children present with these behaviours. Secondly, by developing an ability to functionally analyse (Sturmey 1996) some of these behaviours (particularly those associated with intense reactions to stressors, plus rigidity) and see them as possible methods of the child reducing the degree of anxiety that they may be experiencing in relation to a social environment that they do not comprehend themselves (Bellini 2004, 2006), parents might reduce their own anxiety or depression by accepting that these behaviours are aspects of their child’s diagnosis in the same way that (for example) inability to see dangers when walking might be an aspects of the diagnosis of a child who is visually impaired: such acceptance-focussed approaches have been shown to ameliorate depression (S Hayes et al. 2006). Third, although rigidity in their sons’ behaviour was significantly associated with depression in the mothers studied here, that might be ameliorated by applying reframing techniques (Swoboda et al. 1990), focussing instead upon the positive outcomes of such rigidity and recognising that it is a trait that can produce a consistency in life experience that brings calm to the person exhibiting it. As well as their parents, the adolescents themselves might also use this reframing technique, which has been shown to improve interpersonal communication (Teti et al. 2016). Thus, by reconceptualising the ASD-related behaviours shown to be associated with elevated maternal anxiety and depression in this sample, and seeing them as disability-related rather than wilful misbehaviour, plus focussing upon the possible valuable outcomes that these behaviours have for the child who is exhibiting them, the levels of anxiety and/or depression that parents feel might be reduced. None of these suggestions is intended to replace the real value of relaxation-focussed strategies, mindfulness training, biofeedback for parents’ anxiety and depression (Da Paz and Wallander 2017) or even respite time (Harper et al. 2013), and should best be incorporated in a comprehensive parent-mental health model rather than seen as ‘solutions’ in their own right. Parents’ GAD and MDD are complex phenomena and require multi-focussed solutions that incorporate personal well-being as well as strategies for understanding the behaviour of their child with ASD. Trials of such comprehensive models of parent-training will be required to test the various aspects suggested here.

There are some limitations to the generalisability of these results. First, although the current study was restricted to mothers and sons, the ratio of boys to girls with diagnosed ASD remains high, and it is still the case that child care in the home is more commonly provided by mothers than fathers. Extension of this research to fathers and to girls with ASD is needed before a comprehensive model of the relationships between child ASD-related symptoms and parental anxiety and depression can be formulated. Similarly, these data were collected via a ‘snapshot’ methodology, and it may be valuable to collect longitudinal data to determine any variability in the parental GAD and MDD observed here. Only mothers of high-functioning boys were included here and the behavioural repertoires of low-functioning children with ASD and the effects of these upon parental mental health need further investigation. It should be emphasised that the term “high functioning” was applied by the authors and does not have a definition in the ASD diagnostic literature. The ASDBC is not widely used but is closely based upon DSM-5 diagnostic criteria and possesses satisfactory validity and reliability. As such, it is a direct measure of the specific ASD-related behaviours exhibited by children with ASD and therefore potentially of value in studies such as this where the association between parental mental health and their child’s specific behaviour is the major focus of research. In terms of the data themselves, although there is a possibility that the mothers who were most anxious due to other factors than their child with ASD could also have rated their son’s ASD behaviour more severely because of their (innate) anxiety, that potential confound is always present when data are collected from a single source, as is the case for the overwhelming majority of studies about children with ASD and their parents’ mental health status. For example, all four of the major studies mentioned in the Introduction that examined parental stress and their child’s ASD features (i.e., Bebko et al. 1987; Davis and Carter 2008; Lecavalier et al. 2006; Rivard et al. 2014) did so via parental self-reports on their own stress and on their child’s ASD features. This possible confound must be accepted as a potential limitation in this study as it is in other similar studies, but does not prevent data such as those presented here and in the four previous studies making a valuable contribution to the literature, provided this caveat is kept in mind. As mentioned in the Results, the current sample was divided into subsamples of different sizes and, although recommended procedures were adopted within the MANOVA calculations to cater for this, replication with larger and more equal samples would be valuable. It was decided not to collect a psychiatry history from these parents (a pilot study indicated that this would be a sufficient reason for parental non-participation), but such information would be of value in determining the exact impact that child vs parent issues had upon parental states. Finally, although the triangulation of these data with those from clinical interviews would help clarify these findings, there is a strong argument for self-report measures in studies such as these because it is the perspective of the parents that is the key vantage point under investigation.

Keeping those limitations in mind, these data provide some greater detail than was previously available regarding the associations between the ASD-related behaviour of boys and the GAD and MDD experienced by their mothers and as reported by those mothers. As mentioned in the Introduction, parental stress has been examined quite extensively, and remains a valuable outcome measure in studies such as this one. The inclusion of formal measures of GAD and MDD extends those previous findings regarding parental stress and also provides information regarding the more serious sequalae of the ASD-related behaviour shown by their children upon the mental health of parents.