Advertisement

Journal of Autism and Developmental Disorders

, Volume 48, Issue 4, pp 1133–1146 | Cite as

Caregiver Burden Varies by Sensory Subtypes and Sensory Dimension Scores of Children with Autism

  • Brittany N. Hand
  • Alison E. Lane
  • Paul De Boeck
  • D. Michele Basso
  • Deborah S. Nichols-Larsen
  • Amy R. Darragh
S.I. : Parenting Children with ASD

Abstract

Understanding characteristics associated with burden in caregivers of children with autism spectrum disorder (ASD) is critical due to negative health consequences. We explored the association between child sensory subtype, sensory dimension scores, and caregiver burden. A national survey of caregivers of children with ASD aged 5–13 years was conducted (n = 367). The relationship between variables of interest and indicators of caregiver burden, including health-related quality of life (HRQOL) and caregiver strain, was examined with canonical correlation analyses. Caregiver strain was, but caregiver HRQOL was not, significantly associated with child sensory subtype and sensory dimension scores. Caregiver age, child age, and household income were also associated with caregiver strain. Potential explanatory mechanisms for these findings, derived from published qualitative studies, are discussed.

Keywords

Pediatrics Autism Caregiver burden Caregiver strain Sensory processing Sensory subtypes 

Introduction

Caregivers of children with autism spectrum disorder (ASD) have higher levels of burden than caregivers of children with attention deficit hyperactive disorder, developmental disabilities or other healthcare needs (Cadman et al. 2012; Dabrowska and Pisula 2010; Estes et al. 2009). Two common ways of quantifying caregiver burden are level of health related quality of life (HRQOL) and caregiver strain (Khanna et al. 2011). HRQOL is a measure of “perceived mental and physical health over time,” and captures ways in which mental and physical health influence quality of life, while caregiver strain reflects “the demands, responsibilities, difficulties, and negative psychic consequences of caring for relatives with special needs” [Centers for Disease Control (CDC) 2016; Brannan et al. 1997, p. 212). Poor HRQOL has been associated with chronic disease and chronic disease risk factors, including high body mass index, physical inactivity, and smoking (CDC 2016). Additionally, high caregiver strain has been linked with negative health behaviors including: decreased physical activity; poor sleep patterns; difficulties with weight maintenance; smoking, and alcohol consumption (Gallant and Connell 1998). Given the long-term implications of poor HRQOL and caregiver strain, it is critical to understand the factors associated with burden in caregivers of children with ASD.

Various parent and child characteristics are associated with the level of HRQOL and strain experienced by caregivers of children with ASD. Parent characteristics, such as higher socioeconomic status, more perceived social support, and adaptive coping strategies, are associated with better HRQOL (Lee et al. 2009; Khanna et al. 2011). Lower levels of social support, negative cognitive appraisal of situations, and passive avoidant coping strategies (i.e., denial, avoidance) have been associated with increased caregiver strain (Stuart and McGrew 2009). Characteristics of the child, such as maladaptive behaviors, physical health problems, and whether or not the child lives in the home, are related to caregiver HRQOL, while severity of ASD symptoms, level of behavioral difficulties, and functional impairment have been linked with caregiver strain (Kring et al. 2008, 2009; Khanna et al. 2012; Stuart and McGrew 2009).

Sensory difficulties, which refer to behaviors associated with difficulties processing and integrating sensory information, are another child characteristic associated with caregiver burden in both qualitative (Schaaf et al. 2011) and quantitative studies (Kirby et al. 2015). Quantitative findings reveal that increased frequency of sensory difficulties predict elevated levels of caregiver strain (Kirby et al. 2015). Individuals with ASD and sensory difficulties often: (a) demonstrate hyper- or hypo-reactivity to sensory stimuli; (b) have unusual sensory interests; (c) inaccurately perceive sensory stimuli; and/or (d) have difficulties managing multiple concurrent sensory stimuli (Schaaf and Lane 2015a). Recent work indicates that increased levels of hyper- and hypo-responsiveness are associated with increased caregiver objective burden, and hyper-responsiveness is also significantly associated with increased caregiver subjective burden (Kirby et al. 2015). A detailed analysis, however, of the impact of patterns of sensory difficulties, rather than frequency of symptoms, on caregiver burden is missing from the literature.

Shared patterns of sensory difficulties have been used in research to classify children with ASD into groups called sensory subtypes (Ausderau et al. 2014; Lane et al. 2010, 2011, 2014). Subtyping provides an opportunity to reduce the heterogeneity of the clinical features associated with ASD so that interventions for children and families can be better targeted. As such, the identification of clinically meaningful subtypes is a priority research area for the field (The Interagency Autism (IACC) Coordinating Committee 2014). Sensory difficulties are promising characteristics upon which subtypes can be based, as they are prevalent in up to 92% of all individuals with ASD (Schaaf and Lane 2015a).

The earliest documented sensory subtypes are those proposed by Lane and colleagues (Lane et al. 2010, 2011, 2014). Lane et al. (2014) propose a classification schema of sensory subtypes of children with ASD based on parent responses to the Short Sensory Profile (SSP; McIntosh et al. 1999). These sensory subtypes were identified using model-based cluster analysis on SSP sub-domain z-scores of children with ASD (Lane et al. 2014). Models reflecting different numbers of subtypes were analyzed and a four-subtype solution was deemed to be the most parsimonious and interpretable after inspection of the Bayesian Information Criteria (BIC). The four sensory subtypes are characterized by the frequency of sensory difficulties and sensory modality (e.g., tactile, taste, smell, auditory) affected:

  1. 1.

    Sensory Adaptive (SA): low frequency of sensory difficulties that are not substantial enough to impair functional performance;

     
  2. 2.

    Taste/Smell Sensitive (TSS): moderate frequency of sensory difficulties, but extreme difficulties in the area of taste/smell sensitivity;

     
  3. 3.

    Postural Inattentive (PI): moderate frequency of sensory difficulties, but extreme difficulties in the area of proprioceptive functioning and postural control; and

     
  4. 4.

    Generalized Sensory Difference (GSD): high frequency of sensory difficulties across all sensory modalities (Lane et al. 2014).

     

Findings from an independent component analysis, conducted by our lab using SSP sub-domain z-scores from two independent samples of children with ASD, indicate that two latent dimensions underlie the four sensory subtypes: (1) sensory reactivity—the intensity of a behavioral response to a sensory stimulus, and (2) multisensory integration—the ability to interpret and respond to multiple sensory stimuli simultaneously (Hand et al. 2017). Figure 1 provides an illustration of the relationship of these dimensions to sensory subtype membership. Children with sensory reactivity difficulties may demonstrate hyper- or hypo-reactivity, which is an impaired ability to modulate the intensity of their response to a sensory stimulus. Children with hyper-reactive responses that are limited to taste and smell stimuli are classified into the TSS subtype, while children who display hyper-reactivity more broadly in movement, tactile, visual, and auditory areas are classified as GSD (Lane et al. 2014). Further, children who exhibit extreme hypo-reactivity across a broad range of sensory modalities are assigned to the GSD subtype. Alternatively, children with multisensory integration difficulties demonstrate behaviors that are characteristic of ineffective integration of multiple concurrent sensory stimuli. For example, difficulties with simultaneously processing input from visual, proprioceptive, and vestibular systems may result in an impaired ability to hold body positions or perform coordinated movements (Izawa et al. 2012; MacNeil and Mostofsky 2012). Children who demonstrate these behaviors (i.e., difficulty with holding body positions and performing coordinated movements) are classified as belonging to the PI or GSD subtypes. In summary, these two dimensions help to explain sensory subtype characteristics whereby: (1) children in the SA subtype do not demonstrate difficulties with sensory reactivity or multisensory integration; (2) children in the TSS and PI subtypes demonstrate difficulty with sensory reactivity and multisensory integration in their respective modalities, and (3) children in the GSD subtype demonstrate behaviors consistent with both sensory reactivity and multisensory integration deficits across sensory modalities.

Fig. 1

Theoretical concepts underlying sensory subtypes.

Figure from Hand et al. (2017)

Distinct non-sensory behavior profiles have also been associated with each sensory subtype. For example, children in the TSS subtype display increased picky eating behaviors and more severe communication impairments, while children in the GSD subtype demonstrate more behavioral difficulties that interfere with everyday activities (Lane et al. 2010, 2011). Although the sensory subtypes differ from one another in clinical presentation, they are not well explained by systematic variation in age, non-verbal IQ, or severity of autistic symptoms including social interaction, communication, restricted and repetitive behaviors (Lane et al. 2014).

As frequency of sensory difficulties as well as personal characteristics, behaviors, and activities differ between subtypes, it is possible that caregivers of children with ASD experience different levels of caregiver burden as a function of their children’s sensory subtype rather than frequency of symptoms alone. Further understanding of the relationship between level of caregiver burden and child sensory subtype may be useful in alerting clinicians working with children with ASD and their families to caregivers who may be in the most need of interventions to alleviate burden. However, to date, the level of caregiver burden experienced as a function of the child’s sensory subtype and the underlying dimensions of sensory reactivity and multisensory integration have not been explored. The objectives of the present study are to examine this relationship by determining:

  1. 1.

    The association of caregiver-reported HRQOL and strain with child sensory subtype; and

     
  2. 2.

    The association of caregiver-reported HRQOL and strain with child’s scores on the two sensory dimensions.

     

Given that increased frequency of sensory difficulties is linked with elevated levels of caregiver burden (Kirby et al. 2015), our hypothesis for objective 1 is that caregiver HRQOL and strain will be associated with frequency of sensory symptoms (i.e., higher levels of sensory symptoms will be associated with higher strain and lower QOL). Our hypothesis for objective 2 is that caregiver HRQOL and strain will be associated with child sensory dimension scores. Specifically, we expect parents to be more burdened by symptoms associated with sensory reactivity than multisensory integration (Kirby et al. 2015).

Methods

Study Design

We conducted a non-experimental, cross-sectional, online survey. The primary variables of interest were: (1) caregiver HRQOL, as measured by the Short Form 12 Health Survey version 2 (Ware and Sherbourne 1992); (2) caregiver strain, as measured by the caregiver strain questionnaire (Brannan et al. 1997); (3) child sensory subtype; and (4) child sensory dimension scores. Sensory subtype classification and dimension scores were derived from the algorithm proposed by Lane et al. (2014) using short sensory profile (McIntosh et al. 1999) data. Age of the caregiver, age of the child, household income, and number of children in the household were included as covariates due to the established influence that these variables have on caregiver burden in other populations (Goldstein et al. 2004; Hastings 2002; Liu et al. 2007; Nabors et al. 2002).

Participants and Procedures

Study participants were primary caregivers of children aged 5–13 years with ASD (n = 367), defined as the person who takes primary responsibility for the child with ASD. To eliminate potential confounding effects on level of caregiver burden, caregivers were excluded from the present study if they reported that their child had any of the following comorbidities: (1) neurological diagnosis (i.e., cerebral palsy, stroke, spinal cord injury); (2) genetic diagnosis (i.e., Rett, Down, or Fragile × syndromes); and/or (3) a significant physical disability (i.e., blindness or deafness).

Participants were recruited through multiple avenues including the Interactive Autism Network (IAN), ResearchMatch, and pediatric clinics in the greater Columbus, Ohio area. The IAN is an online research registry for caregivers of children with ASD, which recently authenticated the parent-report ASD diagnoses of children in their database via medical record review (Daniels et al. 2011), and ResearchMatch is a national online registry for individuals interested in being matched with ongoing studies for which they qualify. A total of 8215 and 255 registrants met our inclusion criteria from the IAN and ResearchMatch, respectively. Participants meeting study criteria from the IAN and ResearchMatch received a brief study description with a link to the survey site via email. Three emails were sent to each potential participant to maximize survey response rate. Additionally, promotional fliers were posted in local pediatric occupational, physical, and speech therapy clinic waiting rooms and reception areas. Participants received a free Redbox DVD code via email for survey completion. Figure 2 provides a schematic representation of the survey response rate and retention throughout online survey completion.

Fig. 2

Survey response rate. SF-12v2 Short Form 12 Health Survey, version 2; SSP short sensory profile, CGSQ caregiver strain questionnaire

Measures

Caregiver Health-Related Quality of Life

The Short Form 12 Health Survey version 2 (SF-12v2), a measure of caregiver HRQOL (Ware et al. 1996), is widely utilized in survey research and has measured HRQOL of caregivers of children with ASD previously (Khanna et al. 2011). This questionnaire is a 12-item, norm-referenced, self-report measure that assesses eight health domains that combine to yield aggregate summary measures for physical component summary (PCS) and mental component summary (MCS), where higher scores indicate better health related quality of life. The PCS and MCS have been shown to have excellent internal consistency in the general population with reliability coefficients (Cronbach’s \(\alpha\)) of 0.92 and 0.88, respectively (Maruish 2012).

Caregiver Strain

Level of caregiver strain was assessed with the caregiver strain questionnaire (CGSQ, Brannan et al. 1997), which is a 21-item questionnaire that measures three types of strain: (1) objective strain—negative happenings or events as a result of the child’s behavioral difficulties; (2) subjective internalized strain—negative feelings experienced by the caregiver and (3) subjective externalized strain—negative feelings that the caregiver has towards the individual for whom they provide care (Kirby et al. 2015). The CGSQ has excellent internal consistency reliability with a Cronbach’s \(\alpha\) of 0.93 for the entire scale (Brannan et al. 1997). Cronbach’s \(\alpha\) for the three subscales were as follows: 0.92 for the objective strain subscale, 0.86 for the subjective internalized strain subscale, and 0.74 for the subjective externalized strain subscale. While the CGSQ was originally developed for caregivers of children with emotional and behavioral difficulties, it has been validated for use in the population of caregivers of children with ASD (Brannan et al. 1997; Khanna et al. 2011; Stuart and McGrew 2009), demonstrating excellent internal consistency reliability (Cronbach’s \(\alpha\) = 0.94), and has been shown to have convergent validity with conceptually similar measures (Khanna et al. 2011, 2012). Further, the three-construct (objective, subjective internalized, and subjective externalized strain) factor structure was validated in a sample of caregivers of children with ASD (Khanna et al. 2012).

Child Sensory Subtype and Sensory Dimension Scores

The Short Sensory Profile was administered for the purpose of determining child sensory subtype and sensory dimension scores (McIntosh et al. 1999). The SSP, a 38-item parent questionnaire, measures behaviors associated with abnormal sensory processing in seven sensory domains: (1) tactile; (2) taste/smell; (3) movement; (4) visual/auditory sensitivity; (5) under-responsive/seeks sensation; (6) auditory filtering; and (7) low energy/weak. Scores for the seven sensory domains are compared to normative data from 1200 typically developing children. Higher scores indicate a more typical performance, while lower scores indicate a probable or definite difference in sensory processing. Internal consistency of overall and subdomain sections of the SSP is moderate to excellent (r = 0.70–0.90), and the SSP is > 95% accurate in differentiating children with and without sensory impairments (McIntosh et al. 1999). While the SSP was originally designed and validated for measuring sensory processing in children aged 3–10 years, it has been widely used in populations of children over 10 years of age (Mangeot et al. 2001; Schoen et al. 2008; Tavassoli et al. 2016; Uljarević et al. 2016). Moreover, the SSP demonstrates convergent validity with conceptually similar measures in individuals with ASD up to 14 years of age (Tavassoli et al. 2016), and in individuals with sensory processing difficulties, but not ASD, up to 16 years of age (Schoen et al. 2008).

Data Analyses

Participants with missing data rates greater than 15% for the SF-12v2 (n = 2) or the SSP (n = 3) were excluded from analyses. No participants had missing data rates greater than 15% for the CGSQ. Additionally, participants who did not report child age (n = 3) were excluded, as their eligibility for the study could not be confirmed. Lastly, participants who did not complete 50% or more items from any single SSP domain (n = 3) were excluded, due to the impact this may have on subtype determination.

Iterative model-based imputation using robust stepwise regression was used to impute any missing values for the remaining 367 participants with package “VIM” in the statistical software R (Templ et al. 2015). With this imputation approach, in each step of the iteration, one variable is used as a dependent variable and the remaining variables serve as the independent variables in a regression model.

Demographic information was summarized descriptively. PCS and MCS scores were calculated from SF-12v2 data using the SF-12v2 Scoring Software 5.0. Participant factor scores for objective, subjective internalized, and subjective externalized strain were extracted from the CGSQ using the confirmatory factor analysis model published by Khanna et al. (2012) in the statistical software R using package “sem” (Fox et al. 2016). Sensory subtype membership for each child was determined from SSP domain z-scores and results from Lane et al.’s (2014) model-based cluster analysis using an algorithm. The general form of the equation utilized by the algorithm is as follows:
$${C}_{t}={\gamma }_{\text{t}}{e}^{\sum _{s}{{\lambda }_{st}({x}_{s}-{\mu }_{st})}^{2}}$$
$${P}_{t}=\frac{{C}_{t}}{\sum _{t}{C}_{t}}$$
where s indexes SSP domains, t indexes subtypes, \({x}_{s}\) is the child’s standardized score for the subscale (s) of the SSP, and \({\lambda }_{st}\), \({\mu }_{st}\), and \({\gamma }_{t}\) are paramaters derived from the model based cluster analysis reported in Lane et al. (2014). The constants needed to calculate the parameters \({\lambda }_{st}\), \({\mu }_{st}\), and \({\gamma }_{t}\) are found in Supplemental Table 1. Each child was classified into the subtype for which they had the highest probability of membership (\({P}_{t}\)). Continuous scores for each child on the dimensions of sensory reactivity and multisensory integration were calculated by summing the probabilities of belonging to the TSS and GSD subtypes, and summing the probabilities of belonging to the PI and GSD subtypes, respectively (Alison E. Lane, personal communication). This calculation was derived from the relationship of the dimensions to the sensory subtypes (Fig. 1).

Lastly, canonical correlation analysis (CCA), a multivariate generalized linear modeling technique, was conducted to explore the relationship between caregiver burden and person characteristics, including the child’s sensory subtype for objective 1 and sensory dimension scores for objective 2. Canonical correlation analysis can be conceptualized as “a logical extension of multiple regression” (Hair 2006). While multiple regression allows for examination of a single dependent variable and several independent variables, CCA allows for the concurrent examination of several dependent variables and several independent variables. CCA generates a linear combination of each set of variables (independent and dependent) that maximizes the correlation while the variance for each variable that is attributable to the other variables in the set is controlled (Hair 2006; Dardas and Ahmad 2014). These linear combinations, called canonical functions, can be interpreted like multiple regression equations (Sherry and Henson 2005). The number of canonical functions generated for a given model is equal to the number of variables in the smaller of the two variable sets. Each canonical function yields a canonical correlation, which is a measure of the relationship between the independent and dependent variables and can be conceptualized as a Pearson correlation (Sherry and Henson 2005).

CCA provides multiple advantages over univariate statistical analyses. First, CCA limits the probability of Type I error due to the multivariate nature of the analysis (Fish 1988). Additionally, as caregiver HRQOL and strain are complex constructs that are influenced by multiple variables and the correlation of those variables, this technique allows for the detection of important multivariate relationships that may not be captured when using univariate methods.

In the present study, four CCAs were performed. Overall, the CCAs examined the relationships between a set of personal characteristics and either HRQOL or caregiver strain variables. Figure 3a–d provide graphic illustrations of the CCAs conducted in the present study. CCAs were conducted using the statistical software program R with ad-on package “yacca” (R Development Core Team 2011; Butts and Butts 2009). A dummy-coding scheme was used for the categorical variable of sensory subtype, where GSD served as the reference group.

Fig. 3

Schematic representations of canonical correlation analyses for caregiver burden and personal characteristics. a Analysis of caregiver HRQOL and sensory subtypes. *Child sensory subtype variable; SA Sensory adaptive, TSS taste/smell sensitive, PI postural inattentive, GSD generalized sensory difference (reference group), PCS physical composite summary, MCS mental composite summary. b Analysis of caregiver strain and sensory subtypes. *Child sensory subtype variable; SA sensory adaptive, TSS taste/smell sensitive, PI postural inattentive, GSD generalized sensory difference (reference group). c Analysis of caregiver HRQOL and sensory dimension scores. d Analysis of caregiver strain and sensory dimension scores

Canonical correlations were interpreted using a multi-step procedure (Sherry and Henson 2005). First, the overall significance of the full model was evaluated via significance test of Wilk’s lambda (\(\lambda\)). Second, the relative importance of individual functions was evaluated by examining Wilk’s \(\lambda\) and squared canonical correlations \({\text{(R}}_{{\text{c}}}^{2})\), which represent the variance-accounted-for effect size and are analogous to R2 effects in multiple regression. Third, for significant functions, the importance of individual variables were interpreted using the following values: (1) structural correlations, which are bivariate correlations between the observed and synthetic variables; and (2) squared structural correlations, which indicate the amount of variance an observed variable linearly shares with its respective synthetic variable (Sherry and Henson 2005). A cutoff value of 0.30 was used to determine significance of structural correlations (Dardas and Ahmad 2014; Matson et al. 2010). In the present model, a structural correlation of 0.30 or greater for dummy-coded variables indicates that a change in subtype membership is significantly associated with a change in the synthetic variable (Brown 2015). While we also examined standardized canonical coefficients, which are the coefficients used in the linear equations to combine the observed independent and dependent variables into respective synthetic variables (Sherry and Henson 2005), primary interpretations were derived from structural correlations in accordance with standard recommendations for interpretation of CCA results (Cohen and Cohen 1983; Meredith 1964).

Power Analysis

An a-priori power analysis was conducted using G*Power 3.1 for a Pearson’s r with a small (Cohen’s q = 0.30) effect size at 80% power (Faul et al. 2009). The power analysis yielded a necessary sample size of 282.

Results

Usable survey responses totaled 367. Table 1 provides descriptive information about the participants. On average, respondents were 42.46 years of age (SD 6.87 years), primarily female (83.38%), married (82%), college educated (73%) and employed full or part time (69%). The majority of children with ASD for whom the participants provide care were male (93.73%), with an average age of 10.09 years (SD 2.51 years). Descriptive statistics for SF-12v2 and CGSQ for all participants and by subtype are detailed in Table 2.

Table 1

Caregiver characteristics

 

All

SA

TSS

PI

GSD

n

367

20

137

77

133

Mean age (SD)

42.46 (6.87)

43.33 (3.91)

42.95 (7.12)

41.82 (6.42)

42.14 (7.22)

Age range

25.82–66.80

36.20–50.02

28.74–66.80

25.82–61.28

29.08–61.18

% Female

83.38

75.00

83.94

81.82

84.97

Household income*

35.50% 100k+

55.00% 100k+

41.61% 100k+

35.06% 100k+

38.34% 25–50 k

Median number of children (range)

2 (1–9)

2 (1–9)

2 (1–5)

2 (1–6)

2 (1–5)

Highest level of education*

32.79% some graduate school

45.00% associates degree

34.31% some graduate school

45.81% some graduate school

61.65% some college/associates degree

Employment status*

46.07% full time

40.00% part time

51.82% part time

42.85% full time

42.86% full time

Marital status*

82.11% married

100% married

85.40% married

87.01% married

72.18% married

Mean child’s age (SD)

10.09 (2.51)

10.19 (2.57)

9.95 (2.67)

10.29 (2.50)

10.11 (2.35)

Child’s age range

4.95–13.98

5.29–13.76

5.01–13.95

5.35–13.98

4.95–13.88

*Categorical variable, reported as % in most common category. All ages are listed in years. SD Standard deviation, SA sensory adaptive, TSS taste/smell sensitive, PI postural inattentive, GSD generalized sensory difference

Table 2

Descriptions of SF-12v2 and CGSQ scores

  

All

SA

TSS

PI

GSD

Short Form 12, version 2 (Scoring range 0–100)

 PCS

Mean

54.15

55.81

54.93

53.24

53.62

Median

56.46

56.72

58.10

55.10

56.20

Range

21.61–69.27

38.44–67.73

30.28–69.27

25.10–66.29

21.61–66.58

IQR

49.38–59.90

53.64–59.95

51.40–59.94

48.81–58.48

48.69–59.86

 MCS

Mean

41.07

42.81

41.60

41.77

39.87

Median

40.34

41.37

41.60

39.69

39.95

Range

13.76–62.73

27.10–62.38

13.76–62.73

20.05–61.46

15.41–60.54

IQR

33.46–62.73

35.72–62.73

34.82–62.73

32.89–51.18

32.13–48.83

Caregiver strain questionnaire (Scoring range − 3.0 to + 3.0)

 Objective

Mean

0.00

− 0.90

− 0.23

− 0.02

0.40

Median

0.01

− 1.05

− 0.26

− 0.13

0.48

Range

− 1.84 to 1.98

− 1.84 to 0.87

− 1.70 to 1.98

− 1.69 to 1.90

− 1.59 to 1.97

IQR

− 0.75 to 0.75

− 1.39 to (−)0.60

− 0.87 to 0.36

− 0.78 to 064

− 0.25 to 1.13

 Subjective int.

Mean

0.00

− 0.69

− 0.18

0.02

0.28

Median

0.03

− 0.72

− 0.19

− 0.06

0.46

Range

− 1.80 to 1.56

− 1.80 to 0.90

− 1.65 to 1.48

− 1.51 to 1.53

− 1.71 to 1.56

IQR

− 0.60 to 0.60

− 1.20–(−)0.35

− 0.72 to 0.41

− 0.54 to 0.58

− 0.12 to 0.82

 Subjective ext.

Mean

0.00

− 0.34

− 0.13

0.04

0.15

Median

− 0.12

− 0.43

− 0.22

− 0.16

0.06

Range

− 1.01 to 1.84

− 1.01 to 1.84

− 0.98 to 2.13

− 0.85 to 1.67

− 0.95 to 2.09

IQR

− 0.52 to 0.39

− 0.74 to (−)0.15

− 0.57 to 0.21

− 0.46 to 0.41

− 0.37 to 0.49

SA Sensory adaptive, TSS taste/smell sensitive, PI postural inattentive, GSD generalized sensory difference, IQR interquartile range, presented as upper–lower, PCS physical composite summary, MCS mental composite summary, Subjective int. = Subjective internalized strain, Subjective ext. = Subjective externalized strain

All CCA assumptions were examined and confirmed, indicating that the analysis meets criteria for significance testing (Tabachnick and Fidell 2001). The assumption of linearity was met for all variables within and across variable sets, based on visual inspection of scatterplots. All variable pairs within and across variable sets were evaluated via Levene’s test and were found to meet the assumption of homoscedasticity. The assumption of normality was examined descriptively and all variables, with the exception of number of children, were reasonably normally distributed upon visual inspection. Number of children in the household was found to be positively skewed; however, because the impact of non-normality on multivariate analyses effectively diminishes with sample sizes of at least 200, this variable was retained in the model (Hair 2006, p. 86; Tabachnick and Fidell 2001, p. 74–75). Although it is not a requirement for CCA, the absence of multicollinearity (i.e., correlations \(\ge\) |0.90|) was evaluated (Nimon et al. 2010). Examining the bivariate correlations (Supplemental Table 2) within and between the person and participation variable sets revealed an absence of multicollinearity in all variables with the exception of objective and subjective internalized strain (r = 0.94). All within-set correlations were r ≤ 0.54 and remaining between-set correlations were r ≤ 0.25.

Caregiver HRQOL and Strain as a Function of Sensory Subtype

The model of the relationship between caregiver HRQOL and person characteristics (Fig. 3a) was not statistically significant (Wilk’s \(\lambda\) = 0.96, \(\chi\) 2 (14) = 15.76, p = 0.34), indicating that the present analysis did not detect a relationship between PCS and MCS scores from the SF-12v2 and the person characteristics included in this model. As a result, the canonical functions generated are not suitable for further interpretation (Sherry and Henson 2005).

The model of the relationship between caregiver strain and person characteristics (Fig. 3b) was statistically significant (Wilk’s \(\lambda\) = 0.76, \(\chi\) 2(21) = 96.38, p = 1.25e–11). Evaluation of the \({\text{R}}_{{\text{c}}}^{2}\) revealed that the full model explained 24.2% of the variance shared between the two variable sets (\({\text{R}}_{{\text{c}}}^{2}\)  = 0.242). Three canonical functions were generated in this analysis, as the number of canonical functions is equal to the number of variables in the smaller of the two variable sets. Function 1 accounted for over 17% of the variance explained (\({\text{R}}_{{\text{c}}}^{2}\) = 0.177). While Function 2, when tested in isolation, was statistically significant (Wilk’s \(\lambda\) = 0.93, \(\chi\) 2(12) = 26.23, p = 0.01), it accounted for less than 5% of the variance explained (\({\text{R}}_{{\text{c}}}^{2}\) = 0.049), and thus was not clinically significant. As a result, Function 2 was not considered in subsequent interpretation. Function 3 did not explain a significant amount (\({\text{R}}_{{\text{c}}}^{2}\) = 0.022) of shared variance between the two variable sets (Wilk’s \(\lambda\) = 0.98, \(\chi\) 2(5) = 8.09, p = 0.15).

Results of the first canonical function (Table 3) indicate that all caregiver strain variables had significant structural correlations, with the highest effect size for objective (r s 2 = 0.98) and subjective externalized strain (r s 2 = 0.77). The high structural correlation (r s = − 0.58) but low standardized coefficient (coef = 0.08) and comparatively low effect size (r s 2 = 0.34) of subjective internalized strain is explained by the previously noted multicollinearity between this variable and objective strain. The significant variables from the person characteristics set were SA, TSS, caregiver age, child age, and household income. Examination of the squared structural correlations revealed that SA and TSS had the largest effect sizes of the person characteristic variables (r s 2 = 0.32 and 0.22, respectively).

Table 3

Standardized coefficients, structural correlations, and squared structural correlations for the first canonical function between caregiver strain and sensory subtype

 

Coefficient

r s

r s 2 (%)

Caregiver strain

Objective

− 1.41

− 0.99*

98.01

Subjective internalized

0.08

− 0.58*

33.64

Subjective externalized

0.39

− 0.88*

77.44

Person characteristics

Child subtype

 SA

0.76

0.57*

32.49

 TSS

0.79

0.47*

22.09

 PI

0.47

0.06

0.36

Caregiver age

0.10

0.31*

9.61

Child age

0.26

0.31*

9.61

Household income

0.10

0.32*

10.24

Number of children

− 0.14

− 0.16

2.56

Structural correlations \(\ge\) 0.30 are marked with an asterisk (*). r s = structural correlation; r s 2 = squared structural correlation; SA sensory adaptive, TSS taste/smell sensitive, PI postural inattentive, reference group: generalized sensory difference

Findings from this analysis indicate that caregivers who are older (r s = 0.31), with a higher income (r s = 0.32), and older child (r s = 0.31) were associated with less objective (r s = − 0.99), subjective internalized (r s = − 0.58), and subjective externalized strain (r s = − 0.88). Caregivers of children in the sensory adaptive (r s = 0.57) and taste/smell sensitive (r s = 0.47) subtypes, when compared with caregivers of children in the generalized sensory difference subtype, were associated with less strain when variance attributable to caregiver age, child age, and household income was held constant. There was not a significant difference between the level of strain experienced by caregivers in the postural inattentive and generalized sensory difference subtypes (r s = 0.06).

Caregiver HRQOL and Strain as a Function of Sensory Dimension Score

Similar to the model of HRQOL that included child sensory subtype, the model including child sensory dimension scores (Fig. 3c) was not statistically significant (Wilk’s \(\lambda\) = 0.96, \(\chi\) 2(12) = 19.38, p = 0.08). This indicates that the present analysis did not detect a relationship between PCS and MCS scores from the SF-12v2 and the person characteristics included in this model.

The model of the relationship between caregiver strain and person characteristics (Fig. 3d) was statistically significant (Wilk’s \(\lambda\) = 0.77, \(\chi\) 2(18) = 97.85, p = 5.47e–13). The \({\text{R}}_{{\text{c}}}^{2}\) revealed that the full model explained 23.2% of the variance shared between the two variable sets (\({\text{R}}_{{\text{c}}}^{2}\) = 0.232). Three canonical functions were generated in this analysis. Function 1 accounted for over 19% of the variance explained (\({\text{R}}_{{\text{c}}}^{2}\) = 0.194). Functions 2 and 3, when tested in isolation, were not statistically significant (Function 2: Wilk’s \(\lambda\)= 0.95, \(\chi\) 2(10) = 17.24, p = 0.07; Function 3: Wilk’s \(\lambda\)= 0.98, \(\chi\) 2(4) = 6.96, p = 0.14). Results of Function 1 (Table 4) revealed that there was a significant association between the child characteristics and indicators of caregiver strain and that increased child difficulties with sensory reactivity (r s = 0.37) and multisensory integration (r s = 0.76) scores were associated with increased objective (r s = 0.97), subjective internalized (r s = 0.58), and subjective externalized strain (r s = 0.84). Examination of the squared structural correlations revealed that multisensory integration had the largest effect size of the person characteristic variables (r s 2 = 0.61).

Table 4

Standardized coefficients, structural correlations, and squared structural correlations for the first canonical function between caregiver strain and sensory dimension

 

Coefficient

r s

r s 2 (%)

Caregiver strain

Objective

− 1.44

− 0.98*

96.04

Subjective internalized

0.40

− 0.56*

31.36

Subjective externalized

0.13

− 0.87*

75.69

Person characteristics

Child dimension score

 Sensory reactivity

− 0.54

− 0.39*

15.21

 Multisensory integration

− 0.85

− 0.78*

60.84

Caregiver age

0.11

0.26

6.76

Child age

0.25

0.30*

9.00

Household income

0.06

0.31*

9.61

Number of children

− 0.15

− 0.14

1.96

Structural correlations \(\ge\) 0.30 are marked with an asterisk (*). r s = structural correlation; r s 2 = squared structural correlation

Discussion

The objectives of the present study were to determine the relationship between caregiver burden and child sensory subtype, as well as the relationship between caregiver burden and child sensory dimension scores. Results indicate that child sensory subtype membership and sensory dimension scores were not significantly associated with caregiver physical and mental HRQOL, as measured by the SF-12v2, when controlling for demographic variables. Findings from studies comparing SF-12v2 scores of caregivers of children with and without ASD, however, have consistently demonstrated that caregivers of children with ASD have poorer mental and physical health than caregivers of typically developing children and children with other developmental disabilities (Khanna et al. 2011; Allik et al. 2006; Eapen and Guan 2016). This may suggest that caregivers of children with ASD in all four sensory subtypes would benefit from routine screening for level of caregiver burden and proactive measures to improve HRQOL.

In the present study, child sensory subtype membership was significantly associated with levels of caregiver reported objective, subjective internalized, and subjective externalized strain, while controlling for caregiver age, child age, and household income. The dummy coding scheme used in the analysis allows for comparisons of strain reported by caregivers of children in the GSD subtype with each of the other three subtypes. Specifically, as we hypothesized, when compared to child membership in the GSD subtype, child membership in the SA subtype was most strongly associated (r s = 0.57) with lower levels of caregiver strain (objective strain r s = − 0.99, subjective internalized strain r s = − 0.58, subjective externalized strain r s = − 0.88), followed by child membership in the TSS subtype (r s = 0.47). Contrary to our hypothesis, however, comparison of child membership in the PI and GSD subtypes was not found to be significantly associated (r s = 0.06) with differences in level of caregiver reported strain.

One possible explanation for this finding is that there was less power for the comparison of the PI (n = 77) and GSD (n = 133) subtypes than the TSS (n = 137) and GSD subtypes, which did have a significant (i.e., r s > 0.30) result. Visual inspection of mean CGSQ scores (Table 2) reveals that caregivers of children in the TSS subtype report lower levels of strain than caregivers of children in the PI subtype, although the difference between these subtypes is not as large as other subtype-to-subtype comparisons (e.g., SA vs. GSD). In a post-hoc analysis to facilitate comparison of caregiver reported strain for caregivers of children in the TSS and PI subtypes, the dummy coding scheme was revised such that TSS served as the reference group and the CCA was re-run. Results of this post-hoc analysis revealed that, when compared to child membership in the TSS subtype, child membership in the PI subtype was not significantly associated (r s = 0.06) with a change in caregiver reported strain. Findings of this post-hoc analysis support the notion that caregivers of children in the TSS and PI subtypes may experience similar levels of strain, and the comparison between the PI and GSD subtypes did not detect a difference due to a smaller number of children in the PI subtype.

A second possible explanation for the lack of a significant association of child membership in the PI versus GSD subtypes with a difference in caregiver reported strain relates to the purported latent dimensions that underlie the four sensory subtypes (i.e., sensory reactivity and multisensory integration). This theory asserts that children in the TSS subtype demonstrate difficulties with sensory reactivity, while children in the PI subtype demonstrate behaviors indicative of deficits in multisensory integration (Hand et al. 2017). Our current findings suggest that deficits in either sensory reactivity or multisensory integration are associated with increased caregiver strain; difficulties in multisensory integration, however, are associated with a higher level of caregiver strain than are deficits in sensory reactivity. Further support for this notion was obtained from analyses of sensory dimensions that revealed, based on magnitude of structural correlations, multisensory integration (r s = 0.76) was more strongly associated with elevated levels of caregiver strain than sensory reactivity (r s = 0.37).

Previous reports regarding sensory difficulties and caregiving demonstrate broad congruence with these results. First, qualitative studies (Schaaf et al. 2011; Bagby et al. 2012) suggest that the child’s sensory difficulties influence many aspects of family life, including three areas that are captured by items in the objective strain section of the CGSQ: (1) disruption of family routines; (2) disruption of family social activities; and (3) decreased time spent with other family members. For example, caregivers report having to take two cars on family outings because the child with ASD often needs to leave the event before the rest of the family due to an inability to cope with the multisensory nature of the experience (Schaaf et al. 2011). When the child’s sensory difficulties disrupt family outings, caregivers report increased feelings of incompetence, frustration, disappointment, and worry, which also has implications for subjective internalized and externalized strain (Bagby et al. 2012). In addition, caregivers describe avoidance of social situations that will be multisensory in nature (e.g., bowling alleys, outdoor weddings) or in unfamiliar spaces (e.g., another family’s home) as a result of the sensory difficulties experienced by the child with ASD (Bagby et al. 2012; Schaaf et al. 2011). For example, one caregiver reported that her son’s behaviors in response to sensory stimuli made the family “not [want to] attend as many social functions as [they] normally would have” (Bagby et al. 2012, p. 82). As a lack of perceived social support is known to be predictive of increased levels of strain, these findings may suggest a feed-forward mechanism, whereby sensory difficulties of the child with ASD reduce socialization, which in turn increases caregiver strain (Eapen and Guan 2016).

Moreover, caregivers report that they are able to spend less time with their other child(ren) because the sensory difficulties of the child with ASD require increased attention from one or both parents (Schaaf et al. 2011). Caregivers describe increased feelings of guilt about this disparity in time and attention spent with their typically developing child(ren), which may also influence subjective internalized and externalized strain. These findings from qualitative literature, which directly link sensory difficulties of the child to effects on daily life that are measured by the objective strain section of the CGSQ, may explain why objective strain had the highest effect size (r s 2 = 0.98) of the caregiver strain variables.

Second, findings from qualitative literature also suggest that one of the biggest barriers to completion of daily routines is the multisensory aspect of the environment or activity (Bagby et al. 2012; Schaaf et al. 2011); this may explain why, in the present study, caregivers of children in the two subtypes with difficulties in multisensory integration (PI and GSD) reported the highest levels of caregiver strain. Caregivers describe increased feelings of stress and fear of child elopement or challenging behaviors when taking the child with ASD into multisensory environments in the community, such as a department store (Schaaf et al. 2011). Caregivers describe the influence of their child’s difficulties with managing multisensory environments on a wide array of daily activities, including: social activities; family leisure outings; family events (e.g., weddings); travel; and shopping. This may, in part, explain why the comparison of children in the PI and GSD subtypes, which both demonstrate difficulties with multisensory integration although they are distinct subtypes, was not significantly associated with a difference in the level of caregiver strain (Bagby et al. 2012; Schaaf et al. 2011). In contrast, caregiving issues for children in the TSS subtype, which is characterized by a primary impairment in sensory reactivity, may center around hyper- or hypo-reactivity to discrete sensory stimuli e.g. tastes, smells, or sounds. It is possible that these specific sensory concerns are easier to accommodate on an individual basis and do not disrupt social engagement to the same degree as difficulties in managing the multisensory nature of many social environments. Further, behaviors associated with hypo-reactivity, while restrictive in terms of learning and peer relationships, do not necessarily prohibit family participation in social contexts and are less likely to result in challenging behavioral outbursts that are stressful for caregivers to manage. The contrasting behavioral manifestations of each sensory subtype, in addition to overall frequency of sensory symptoms, may explain why child membership in the TSS subtype, when compared to the GSD subtype, was associated with significantly less caregiver strain.

Methodologic Considerations and Future Directions

The findings of the present study should be considered in light of some limitations; first, we relied on caregiver report of ASD diagnosis. For participants recruited through the IAN, we can be confident that the child did receive an independent ASD diagnosis due to the recent validation of diagnoses in the registry. However, there is less certainty with regard to participants who responded to waiting room fliers and ResearchMatch emails. Therefore, future replication of this study in a sample of children with independently confirmed ASD diagnoses is advisable. Moreover, it is possible that other variables (e.g., intellectual ability, expressive language ability, parental perceived social support, parental coping strategies) could offer additional explanation for the observed variation in caregiver strain. These variables were not collected in the present study due to attempts to minimize respondent burden and maximize survey response rate, but should be considered in future research.

Additionally, the number of participants belonging to the SA subtype was disproportionately small in the present sample when compared with other independent samples of children with ASD (Lane et al. 2014). We hypothesize that this could be due to two primary reasons. First, there may have been sampling bias as a result of recruitment materials that mentioned an interest in “sensory processing” and caregiver burden. While recruitment materials did state that we were looking for caregivers of children with ASD with and without sensory difficulties, mention of interest in sensory processing may have decreased the likelihood of response by caregivers of children in the SA subtype as these children do not have clinically significant sensory difficulties. Second, the use of clinical recruitment sites may have been another source of sampling bias, as it is possible that children in the SA subtype may not receive outpatient occupational, physical, or speech therapy services. Thus, caregivers of children in the SA subtype may not have been presented with the same opportunities for participation as caregivers of children in other subtypes. Future studies seeking to replicate these findings should carefully consider these aspects of recruitment in seeking to recruit more caregivers of children in the SA subtype.

Finally, there are some limitations associated with the statistical analysis. As CCA is associational in nature, we are unable to assert any causal relationships between sensory subtypes, sensory dimensions, and level of caregiver burden. We can, however, conclude that child sensory subtype membership and sensory dimension scores are associated with different levels of caregiver perceived strain. Moreover, findings from qualitative literature provide valuable insight as to how sensory difficulties and caregiver strain may be related. Additionally, it is noteworthy that CCA reflects the variance shared by the canonical variates, which are the linear composites of observed variables, rather than the observed variables themselves. Therefore, it is recommended that the present study be replicated with a larger sample size that would sufficiently power alternate statistical methodologies. Lastly, there may be some concern that the present study was overpowered, as our power analysis indicated that a sample size of 282 would be sufficient to detect a small effect at 80% power. In order to test the possible effect of a large sample size, the CCA was re-run with a random selection of 75% (n = 275) of the data. This analysis revealed that while individual values (i.e., structural correlations, canonical coefficients) differed from the results presented here, the relationships between all variables, and therefore the substantive interpretation, remained the same; this replication of findings lends credibility to the robustness of the results presented here.

Summary and Conclusion

Level of caregiver objective, subjective internalized, and subjective externalized strain were associated with the child’s sensory subtype and sensory dimension scores, while controlling for caregiver age, child age, and household income. When compared with children in the GSD subtype, child membership in the SA subtype was associated the lowest levels of caregiver strain, followed by child membership in the TSS subtype. Child membership in the PI and GSD subtypes was not significantly associated with a difference in level of caregiver strain. Findings suggest that child multisensory integration difficulties are associated with higher levels of caregiver strain that difficulties in sensory reactivity. Increasing caregiver and child age, as well as higher household income, were associated with less caregiver strain. The present study did not detect an association between child sensory subtypes or sensory dimension scores and level of caregiver perceived mental and physical health. This may suggest that, while caring for a child with ASD is known to be associated with poorer caregiver physical and mental HRQOL, differences in the level of HRQOL reported by caregivers of children in distinct sensory subtypes may be small.

Findings may assist in alleviating burden of caregivers of children with ASD by alerting clinicians to aspects of caregiver burden associated with the child’s sensory subtype to target for intervention. These findings, when combined with existing literature, have the capability to support decision making for clinicians working with children with ASD and their families by linking each subtype with intervention targets for the child and caregiver. Ultimately, this may lead to improved treatment outcomes for children with ASD and their families.

Notes

Funding

This work was funded by the M. Rosita Schiller Research Award and Alumni Grant for Graduate Research and Scholarship from The Ohio State University. Additionally, support was received from the Center for Clinical and Translational Science at The Ohio State University and the Interactive Autism Network.

Author Contributions

BH participated in study design, conducted statistical analyses, and drafted the manuscript; AL participated in study design, results interpretation, and manuscript revisions; PDB oversaw statistical analyses and participated in manuscript revisions; DMB participated in study design and manuscript revisions; DNL participated in study design and manuscript revisions; AD oversaw study design and participated in manuscript revisions. All authors read and approved the final manuscript.

Compliance with Ethical Standards

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Conflict of interest

The authors declared that they have no conflict of interest.

Informed Consent

Informed consent was obtained from all individual participants included in the study.

Supplementary material

10803_2017_3348_MOESM1_ESM.docx (79 kb)
Supplementary material 1 (DOCX 78 KB)
10803_2017_3348_MOESM2_ESM.docx (63 kb)
Supplementary material 2 (DOCX 62 KB)

References

  1. Allik, H., Larsson, J.-O., & Smedje, H. (2006). Health-related quality of life in parents of school-age children with Asperger syndrome or high-functioning autism. Health and Quality of Life Outcomes, 4(1), 1. doi: 10.1186/1477-7525-4-1.CrossRefPubMedPubMedCentralGoogle Scholar
  2. Ashburner, J., Rodger, S., Ziviani, J., & Jones, J. (2014). Occupational therapy services for people with autism spectrum disorders: Current state of play, use of evidence and future learning priorities. Australian Occupational Therapy Journal, 61(2), 110–120. doi: 10.1111/1440-1630.12083.CrossRefPubMedGoogle Scholar
  3. Ausderau, K. K., Furlong, M., Sideris, J., Bulluck, J., Little, L. M., Watson, L. R., … Baranek, G. T. (2014). Sensory subtypes in children with autism spectrum disorder: Latent profile transition analysis using a national survey of sensory features. Journal of Child Psychology and Psychiatry, 55(8), 935–944. doi: 10.1111/jcpp.12219.CrossRefPubMedPubMedCentralGoogle Scholar
  4. Bagby, M. S., Dickie, V. A., & Baranek, G. T. (2012). How sensory experiences of children with and without autism affect family occupations. American Journal of Occupational Therapy, 66(1), 78–86. doi: 10.5014/ajot.2012.000604.CrossRefPubMedPubMedCentralGoogle Scholar
  5. Brannan, A. M., Heflinger, C. A., & Bickman, L. (1997). The caregiver strain questionnaire measuring the impact on the family of living with a child with serious emotional disturbance. Journal of Emotional and Behavioral Disorders, 5(4), 212–222. doi: 10.1177/106342669700500404.CrossRefGoogle Scholar
  6. Brown, T. A. (2015). Confirmatory factor analysis for applied research. New York: Guilford Publications.Google Scholar
  7. Butts, C. T., & Butts, M. C. T. (2009). Package “yacca.” Retrieved from http://w.download.idg.pl/CRAN/web/packages/yacca/yacca.pdf.
  8. Cadman, T., Eklund, H., Howley, D., Hayward, H., Clarke, H., Findon, J., … Glaser, K. (2012). Caregiver burden as people with autism spectrum disorder and attention-deficit/hyperactivity disorder transition into adolescence and adulthood in the United Kingdom. Journal of the American Academy of Child & Adolescent Psychiatry, 51(9), 879–888. doi: 10.1016/j.jaac.2012.06.017.CrossRefGoogle Scholar
  9. Centers for Disease Control (CDC). (2016). Health-related quality of life (HRQOL). Retrieved from https://www.cdc.gov/hrqol/index.htm.
  10. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.Google Scholar
  11. Dabrowska, A., & Pisula, E. (2010). Parenting stress and coping styles in mothers and fathers of pre-school children with autism and Down syndrome. Journal of Intellectual Disability Research, 54(3), 266–280. doi: 10.1111/j.1365-2788.2010.01258.x.CrossRefPubMedGoogle Scholar
  12. Daniels, A. M., Rosenberg, R. E., Anderson, C., Law, J. K., Marvin, A. R., & Law, P. A. (2011). Verification of parent-report of child autism spectrum disorder diagnosis to a web-based autism registry. Journal of Autism and Developmental Disorders, 42(2), 257–265. doi: 10.1007/s10803-011-1236-7.CrossRefGoogle Scholar
  13. Dardas, L. A., & Ahmad, M. M. (2014). Psychosocial correlates of parenting a child with autistic disorder. Journal of Nursing Research, 22(3), 183–191. doi: 10.1097/jnr.0000000000000023.CrossRefPubMedGoogle Scholar
  14. Eapen, V., & Guan, J. (2016). Parental quality of life in autism spectrum disorder: Current status and future directions. Acta Psychopathologica, 2(1), 1–14.CrossRefGoogle Scholar
  15. Estes, A., Munson, J., Dawson, G., Koehler, E., Zhou, X.-H., & Abbott, R. (2009). Parenting stress and psychological functioning among mothers of preschool children with autism and developmental delay. Autism: The International Journal of Research and Practice, 13(4), 375–387. doi: 10.1177/1362361309105658.CrossRefGoogle Scholar
  16. Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149–1160. doi: 10.3758/BRM.41.4.1149.CrossRefPubMedGoogle Scholar
  17. Fish, L. J. (1988). Why multivariate methods are usually vital. Measurement and Evaluation in Counseling and Development, 21(3), 130–137.Google Scholar
  18. Fox, J., Nie, Z., Byrnes, J., Culbertson, M., DebRoy, S., Friendly, M., … others. (2016). Package “sem.” Retrieved from http://mirror.mdx.ac.uk/R/web/packages/sem/sem.pdf.
  19. Gallant, M. P., & Connell, C. M. (1998). The stress process among dementia spouse caregivers are caregivers at risk for negative health behavior change? Research on Aging, 20(3), 267–297. doi: 10.1177/0164027598203001.CrossRefGoogle Scholar
  20. Goldstein, N. E., Concato, J., Fried, T. R., Kasl, S. V., Johnson-Hurzeler, R., & Bradley, E. H. (2004). Factors associated with caregiver burden among caregivers of terminally ill patients with cancer. Journal of Palliative Care, 20(1), 38–43.PubMedGoogle Scholar
  21. Hair, J. (2006). Multivariate Data Analysis (6th edn.). Upper Saddle River, NJ: Pearson Prentice Hall.Google Scholar
  22. Hand, B. N., Dennis, S., & Lane, A. E. (2017). Latent constructs underlying sensory subtypes in children with autism: A preliminary study. Autism Research, 10(8), 1364–1371.Google Scholar
  23. Hastings, R. P. (2002). Parental stress and behaviour problems of children with developmental disability. Journal of Intellectual & Developmental Disability, 27(3), 149–160. doi: 10.1080/1366825021000008657.CrossRefGoogle Scholar
  24. Izawa, J., Pekny, S. E., Marko, M. K., Haswell, C. C., Shadmehr, R., & Mostofsky, S. H. (2012). Motor learning relies on integrated sensory inputs in ADHD, but over-selectively on proprioception in autism spectrum conditions. Autism Research, 5(2), 124–136. doi: 10.1002/aur.1222.CrossRefPubMedPubMedCentralGoogle Scholar
  25. Khanna, R., Madhavan, S. S., Smith, M. J., Patrick, J. H., Tworek, C., & Becker-Cottrill, B. (2011). Assessment of health-related quality of life among primary caregivers of children with autism spectrum disorders. Journal of Autism and Developmental Disorders, 41(9), 1214–1227. doi: 10.1007/s10803-010-1140-6.CrossRefPubMedGoogle Scholar
  26. Khanna, R., Madhavan, S. S., Smith, M. J., Tworek, C., Patrick, J. H., & Becker-Cottrill, B. (2012). Psychometric properties of the caregiver strain questionnaire (CGSQ) among caregivers of children with autism. Autism: The International Journal of Research and Practice, 16(2), 179–199. doi: 10.1177/1362361311406143.CrossRefGoogle Scholar
  27. Kirby, A. V., White, T. J., & Baranek, G. T. (2015). Caregiver strain and sensory features in children with autism spectrum disorder and other developmental disabilities. American Journal on Intellectual and Developmental Disabilities, 120(1), 32–45. doi: 10.1352/1944-7558-120.1.32.CrossRefPubMedPubMedCentralGoogle Scholar
  28. Kring, S. R., Greenberg, J. S., & Seltzer, M. M. (2008). Adolescents and adults with autism with and without co-morbid psychiatric disorders: Differences in maternal Well-Being. Journal of Mental Health Research in Intellectual Disabilities, 1(2), 53–74. doi: 10.1080/19315860801988228.CrossRefPubMedPubMedCentralGoogle Scholar
  29. Kring, S. R., Greenberg, J. S., & Seltzer, M. M. (2009). The impact of health problems on behavior problems in adolescents and adults with autism spectrum disorders: Implications for maternal burden. Social Work in Mental Health, 8(1), 54–71. doi: 10.1080/15332980902932441.CrossRefGoogle Scholar
  30. Lane, A. E., Dennis, S. J., & Geraghty, M. E. (2011). Brief report: Further evidence of sensory subtypes in autism. Journal of Autism and Developmental Disorders, 41(6), 826–831. doi: 10.1007/s10803-010-1103-y.CrossRefPubMedGoogle Scholar
  31. Lane, A. E., Molloy, C. A., & Bishop, S. L. (2014). Classification of children with autism spectrum disorder by sensory subtype: A case for sensory-based phenotypes: Sensory phenotypes in autism. Autism Research, 7(3), 322–333. doi: 10.1002/aur.1368.CrossRefPubMedGoogle Scholar
  32. Lane, A. E., Young, R. L., Baker, A. E. Z., & Angley, M. T. (2010). Sensory processing subtypes in autism: Association with adaptive behavior. Journal of Autism and Developmental Disorders, 40(1), 112–122. doi: 10.1007/s10803-009-0840-2.CrossRefPubMedGoogle Scholar
  33. Lee, G. K., Lopata, C., Volker, M. A., Thomeer, M. L., Nida, R. E., Toomey, J. A., … Smerbeck, A. M. (2009). Health-related quality of life of parents of children with high-functioning autism spectrum disorders. Focus on Autism and Other Developmental Disabilities, 24(4), 227–239. doi: 10.1177/1088357609347371.CrossRefGoogle Scholar
  34. Liu, M., Lambert, C. E., & Lambert, V. A. (2007). Caregiver burden and coping patterns of Chinese parents of a child with a mental illness. International Journal of Mental Health Nursing, 16(2), 86–95. doi: 10.1111/j.1447-0349.2007.00451.x.CrossRefPubMedGoogle Scholar
  35. MacNeil, L. K., & Mostofsky, S. H. (2012). Specificity of dyspraxia in children with autism. Neuropsychology, 26(2), 165–171. doi: 10.1037/a0026955.CrossRefPubMedPubMedCentralGoogle Scholar
  36. Mangeot, S. D., Miller, L. J., McIntosh, D. N., McGrath-Clarke, J., Simon, J., Hagerman, R. J., & Goldson, E. (2001). Sensory modulation dysfunction in children with attention-deficit–hyperactivity disorder. Developmental Medicine and Child Neurology, 43(6), 399–406. doi: 10.1017/S0012162201000743.CrossRefPubMedGoogle Scholar
  37. Maruish, M. E. (Ed.). (2012). User’s manual for the SF-12v2 Health Survey (3rd edn.). Lincoln, RI: QualityMetric Incorporated.Google Scholar
  38. Matson, J. L., Neal, D., Fodstad, J. C., & Hess, J. A. (2010). The relation of social behaviours and challenging behaviours in infants and toddlers with Autism Spectrum Disorders. Developmental Neurorehabilitation, 13(3), 164–169. doi: 10.3109/17518420903270683.CrossRefPubMedGoogle Scholar
  39. McIntosh, D. N., Miller, L. J., Shyu, V., & Dunn, W. (1999). Overview of the Short Sensory Profile (SSP). In The sensory profile: Examiner’s manual (pp. 59–73). San Antonio, TX: Psychological Corporation.Google Scholar
  40. Meredith, W. (1964). Canonical correlations with fallible data. Psychometrika, 29(1), 55–65. doi: 10.1007/BF02289567.CrossRefGoogle Scholar
  41. Nabors, N., Seacat, J., & Rosenthal, M. (2002). Predictors of caregiver burden following traumatic brain injury. Brain Injury, 16(12), 1039–1050. doi: 10.1080/02699050210155285.CrossRefPubMedGoogle Scholar
  42. Nimon, K., Henson, R. K., & Gates, M. S. (2010). Revisiting interpretation of canonical correlation analysis: A tutorial and demonstration of canonical commonality analysis. Multivariate Behavioral Research, 45(4), 702–724. doi: 10.1080/00273171.2010.498293.CrossRefPubMedGoogle Scholar
  43. R Development Core Team. (2011). R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. Retrieved from http://www.R-project.org.
  44. Schaaf, R. C., & Lane, A. E. (2015a). Toward a best-practice protocol for assessment of sensory features in ASD. Journal of Autism and Developmental Disorders, 45(5), 1380–1395. doi: 10.1007/s10803-014-2299-z.CrossRefPubMedGoogle Scholar
  45. Schaaf, R. C., Toth-Cohen, S., Johnson, S. L., Outten, G., & Benevides, T. W. (2011). The everyday routines of families of children with autism: Examining the impact of sensory processing difficulties on the family. Autism: The International Journal of Research and Practice, 15(2), 1–7. doi: 10.1177/1362361310386505.Google Scholar
  46. Schoen, S. A., Miller, L. J., & Green, K. E. (2008). Pilot study of the sensory over-responsivity scales: Assessment and inventory. American Journal of Occupational Therapy, 62(4), 393–406. doi: 10.5014/ajot.62.4.393.CrossRefPubMedGoogle Scholar
  47. Sherry, A., & Henson, R. K. (2005). Conducting and interpreting canonical correlation analysis in personality research: A user-friendly primer. Journal of Personality Assessment, 84(1), 37–48.CrossRefPubMedGoogle Scholar
  48. Stuart, M., & McGrew, J. H. (2009). Caregiver burden after receiving a diagnosis of an autism spectrum disorder. Research in Autism Spectrum Disorders, 3(1), 86–97. doi: 10.1016/j.rasd.2008.04.006.CrossRefGoogle Scholar
  49. Tabachnick, B., & Fidell, L. (2001). Using multivariate statistics (4th edn.). Boston, MA: Allyn and Bacon.Google Scholar
  50. Tavassoli, T., Bellesheim, K., Siper, P. M., Wang, A. T., Halpern, D., Gorenstein, M., … Buxbaum, J. D. (2016). Measuring sensory reactivity in autism spectrum disorder: Application and simplification of a clinician-administered sensory observation scale. Journal of Autism and Developmental Disorders, 46(1), 287–293. doi: 10.1007/s10803-015-2578-3.CrossRefPubMedGoogle Scholar
  51. Templ, M., Alfons, A., Kowarik, A., Prantner, B., & Templ, M. M. (2015). Package “VIM”. Retrieved from http://alvarestech.com/pub/plan/R/web/packages/VIM/VIM.pdf.
  52. The Interagency Autism (IACC) Coordinating Committee. (2014). Strategic plan for autism spectrum disorder (ASD) research: 2013 update. Retrieved from the U.S. Department of Health and Human Services Interagency Autism Coordinating Committee website http://iacc.hhs.gov/strategic-plan/2013/index.shtml.
  53. Uljarević, M., Lane, A., Kelly, A., & Leekam, S. (2016). Sensory subtypes and anxiety in older children and adolescents with autism spectrum disorder. Autism Research, 9, 1073–1078. doi: 10.1002/aur.1602.CrossRefPubMedGoogle Scholar
  54. Ware, J., Kosinski, M. M., & Keller, S. (1996). A 12-Item short-form health survey: Construction of scales and preliminary tests of reliability and validity. Medical Care, 34(3), 220–233.CrossRefPubMedGoogle Scholar
  55. Ware, J. E., & Sherbourne, C. D. (1992). The MOS 36-item short-form health survey (SF-36): I. conceptual framework and item selection. Medical Care, 30(6), 473–483.CrossRefPubMedGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2017

Authors and Affiliations

  1. 1.College of Health ProfessionsMedical University of South CarolinaCharlestonUSA
  2. 2.Priority Research Centre Grow Up Well®The University of NewcastleNewcastleAustralia
  3. 3.Department of PsychologyThe Ohio State UniversityColumbusUSA
  4. 4.School of Health and Rehabilitation SciencesThe Ohio State UniversityColumbusUSA

Personalised recommendations