Health Reports
Child functional characteristics explain child and family outcomes better than diagnosis: Population-based study of children with autism or other neurodevelopmental disorders/disabilities

by Anton Miller, Jane Shen and Louise C. Mâsse

Release date: June 15, 2016

Children with neurodevelopmental disorders/disabilities (NDD/D) or “neurodisability”Note 1 are the largest identifiable subpopulation of children with disabilitiesNote 2 and account for 7% to 14 % of all children in developed countries.Note 3Note 4 NDD/D comprise an array of conditions characterized by impairment in posture-mobility, cognitive-adaptive functioning, communication, relating socially, and regulating emotions and behaviour; biological or physical markers of a specific medical condition may or may not be present. Diagnoses under NDD/D include autism spectrum disorders (ASD), intellectual or learning disabilities, attention-deficit/hyperactivity disorder (ADHD), cerebral palsy, and Down and fetal alcohol syndromes.

Children with neurodisability require services and supports that span the health, educational, and family and social services sectors.Note 5 Health conditions, as listed in diagnostic taxonomies such as the International Classification of Diseases (ICD)Note 6 or the Diagnostic and Statistical Manual of Mental Disorders (DSM),Note 7 often predominate in eligibility criteria for services and supports,Note 8Note 9Note 10 or form the basis of “disability categories” on which the provision of educational supports is determined.Note 11Note 12Note 13Note 14 However, this approach has been challenged by calls to emphasize individuals’ functional characteristicsNote 9Note 15Note 16Note 17Note 18 rather than the diagnosis category in which they are placed. These concerns arise in light of advances in the conceptualization of disability, along with recognition of the limitations of categorical diagnostic classification systems in mental and developmental health.Note 19Note 20

The World Health Organization’s 2001 International Classification of Functioning, Disability and Health (ICF)Note 21 has advanced the conceptualization of disability in three ways:

In the ICF, functioning is understood to manifest at the level of body, person, or person-in-society, operationalized respectively as intactness (or impairment) of body structures or physiological functioning, ability (or limitation) to carry out daily activities, and participation (or restriction) in meaningful activities with other people. While the ICF framework and concepts are increasingly being adopted and deployed in clinical and research contexts,Note 22Note 23 empirical research into the dynamics and mechanisms of interactions among diagnosed health conditions and functional characteristics remains limited.

The appropriateness of using categorical diagnosis classifications in mental and developmental health has been questioned because of concern about the validity of conditions so definedNote 19; the clinical heterogeneity of persons grouped under one diagnosis category such as ASDNote 24; and the functional and diagnostic complexity of children with NDD/D.Note 2Note 25 Clinicians observe considerable overlap in the day-to-day functional characteristics of children diagnosed with different NDD/D, despite the distinctness of symptom sets for conditions such as ASD, ADHD, or intellectual disability.

Empirical evidence of the roles of diagnosis versus functional characteristics in child and family outcomes could inform policy on services and supports for children with neurodisability. To this end, a national disability dataset, the 2006 Participation and Activity Limitation Survey, was used to study relationships between a child’s diagnosis status (ASD versus other neurodevelopmental diagnoses), a range of functional characteristics that were largely neurodevelopmental or psychological, and measures of child and family health and well-being. The analysis examined whether the functional characteristics most closely associated with a diagnosis of ASD (communication, learning, and regulation of emotions and behaviour) mediate associations between a child’s diagnosis status and the physical and psychological health of parents, the economic well-being of the family, and the child’s participation in various activities, while controlling for other functional characteristics less closely associated with ASD (hearing, vision, mobility, dexterity and pain) and for demographic variables. It was hypothesized that neurodevelopmental and related functional characteristics would be more informative than a child’s diagnosis in explaining child and family outcomes, and would largely or fully mediate apparent effects of diagnosis. ASD was selected as the “reference diagnosis” because of increases in its prevalence and the impact on public health and services.Note 26Note 27 Also, funding programs have been implemented specifically for children with ASD, although an ASD diagnosis is recognized as being one among many factors relevant to understanding the characteristics and needs of affected children and their families.Note 28Note 29

Data and methods

Data source

The 2006 Participation and Activity Limitation Survey (PALS) is a national post-censal survey of adults and children who had a disability (operationalized as experiencing limitation in everyday activities due to a health condition/problem). PALS participants were selected from the 2006 Census of Canada based on responses to two general “filter” questions about activity limitations, as well as age and geography. For those younger than 15, the Child Questionnaire was used to determine the nature, severity, and impact of disabilities (vision, hearing, communication, mobility, dexterity, learning, developmental and emotional or psychological conditions), and to identify conditions and diagnoses that limit participation. Telephone interviews were conducted during 2006/2007 with 7,072 parents/guardians (“parents”) from a sample of approximately 9,000. The data were weighted following adjustment for patterns of non-response and various child characteristics,Note 30 yielding a weighted sample of 340,340.

The current study pertains to children aged 5 to 14 who were attending school because data on participation in various activities were available for them. The analyses were limited to children with NDD/D as identified through a method described in previous work with PALS, who accounted for almost three-quarters of 5- to-14-year-olds with disabilities in the dataset.Note 2 This involved classifying children into one of six NDD/D groups that reflect the most widely acknowledged functionally based domains of child development (motor, speech-language, learning/cognition, social, sensory, and psychological). Classification followed a detailed review of all ICD-10 codes available for each child, with consensus between two developmental pediatricians, triangulation of unclear cases with a third reviewer, and elimination of cases deemed unclassifiable.Note 2

The current study was approved by the University of British Columbia’s Research Ethics Board.

Measures

Child’s diagnosis status: ASD diagnosis (yes/no) was ascertained from the presence of relevant ICD-10 codes (F84.0-F84.5, F84.8 and F84.9) based on information provided by parents during the PALS interview. This included their perception of their child’s main/most responsible health conditions, or their response to a question about a number of specific chronic conditions, one of which was professionally diagnosed autism.

Child’s functional status comprised: 1) functional domains most closely associated with a diagnosis of ASD; and 2) other domains. The three most relevant domains for ASD-related functioning available in PALS were speech-communication, learning-cognition, and emotional-behavioural-psychological. The other domains were vision, hearing, mobility, dexterity, and pain. Two sources were used for child functional characteristics: items from the Health Utilities Index (HUI), slightly modified in PALS for the targeted age group,Note 30 and the PALS impairment index (Table 1). Two latent variables that captured the construct of a child’s neurodevelopmental and related functional characteristics were created: 1) ASD-related functional characteristics (“ASD-related functioning”); and 2) non-specific child functional characteristics (“other functioning”). ASD-related functioning consisted of the three HUI attributes and the three PALS impairment index items most strongly associated with the core ASD impairments identified above. It was entered as a mediator in the analyses; other functioning (described below) was used as a covariate.

In accordance with the concept of family quality of life in the context of disability,Note 31 family health and well-being comprised physical health, psychological well-being, and the economic impact of having a child with a disability, as they relate to parents (Table 1). A second-order factor structure with psychological well-being and economic impacts treated as latent constructs and physical health as an indicator variable was supported by a Confirmatory Factor Analysis (CFA) performed in this sample (χ2(df = 64) = 418.56, p < 0.001; RMSEA = 0.048, 90% CI 0.044–0.052, p = 0.768; CFI = 0.933; and WRMR = 2.002; Appendix Figure A). Cronbach’s alphas were 0.79 and 0.74 for psychological well-being and economic impacts, respectively.

Child participation comprised in-school participation and out-of-school participation (Table 1). A CFA performed in this sample supported the factor structure of child participation (χ2(df = 34) = 63.25, p = 0.0017; RMSEA = 0.018, 90% CI 0.011–0.025, p = 1.000; CFI = 0.958; and WRMR = 0.998; Appendix Figure B). Cronbach’s alphas were 0.71 and 0.49 for in-school and out-of-school participation, respectively. The Cronbach’s alpha was not optimal for out-of-school participation, likely because of lack of variance in the data for this construct.

The covariates entered into the analytic models were child age and sex, annual family income, residential location (urban/rural), and child “other functioning,” which regrouped the remaining five HUI attributes and four PALS impairment items (Table 1).

Analyse

Using Structural Equation Modeling (SEM), three models were run. The covariates model examined the relationship among all covariates, and the three outcome variables—family health and well-being, child in-school participation, and child out-of-school participation. The direct effects model tested whether a child’s diagnosis status had a direct effect on the three outcome variables when accounting for the covariates. The indirect effects model tested whether a child’s diagnosis status had an indirect effect on the three outcome variables with ASD-related functioning included in the model, and if so, whether the indirect effects were mediated by ASD-related functioning when accounting for the covariates (Figure 1).

Using MPlus software (version 7.11, MUTHÉN & MUTHÉN, Los Angeles, CA), the mean and variance-adjusted weighted least-squares method (WLSMV) was employed. Based on published criteria,Note 32Note 33 fit was considered to be acceptable if: Root Mean Square Error of Approximation (RMSEA) with an upper 90% confidence interval (CI) was < 0.08 or a p-value ≥ 0.05; comparative fit index (CFI) was > 0.95; and Weighted Root Mean Square Residual (WRMR) was < 1.0. Because of the strong influence of sample size on the statistical significance of correlation coefficients, a standardized path coefficient (SPC) was considered to have a direct or indirect effect on the outcomes only if it was statistically significant (p < 0.05) and had an absolute magnitude greater than .22 (thereby explaining at least 5% of total variance, approximatelyNote 34). To compare models, the incremental r-square (percentage of additional variance explained by adding variables in a model) was examined, and an additional 5% of variance explained was considered to be meaningful. Finally, the Sobel test was used to test the significance of the mediated effects.Note 35

Results

Characteristics of the children and their families are presented in Table 1. Summary indices for the three SEM analyses are presented in Table 2.

All models adequately fitted the data—the RMSEAs were within acceptable ranges, although the CFI and WRMR were outside ideal range (Table 2). In all cases, examination of the modification indices did not uncover ways to improve model fit. Deleting non-significant paths can increase model fit, but in view of the confirmatory nature of these analyses, non-significant paths were not deleted.

The covariates model explained 11.6% of the variance for family health and well-being, 16.6% of the variance for in-school participation, and 16.4% of the variance for out-of-school participation.

In the direct effects model, child ASD diagnosis status had a significant and meaningful direct effect on family health and well-being (SPC = .29, p < 0.05) and on in-school participation (SPC = .35, p < 0.05) (Figure 2), but not on out-of-school participation (SPC = .10, p = 0.02), as the path coefficient did not meet the criterion for a meaningful magnitude of effect. Inclusion of child diagnosis status in the covariate model explained an additional 8% and 10.9% of the total variance in family health and well-being and in-school participation, respectively.

In the indirect effects model, the direct effects of child diagnosis status on child and family outcomes disappeared when ASD-related functioning was added (Figure 3), with SPCs no longer reaching thresholds for statistical significance or effect size magnitude. ASD-related functioning had a significant and meaningful direct effect on family health and well-being (SPC = .73, p < 0.05), in-school participation (SPC = .54, p < 0.05), and out-of-school participation (SPC = .25, p < 0.05). As anticipated, ASD-related functioning was also significantly related to child diagnosis status (SPC = .39, p < 0.05), meaning that absence of an ASD diagnosis was related to less impairment in the domains of speech and communication, cognition-learning, and emotional-behavioural/psychological functioning (Figure 3). Importantly, ASD-related functioning was a significant mediator between diagnosis status and family health and well-being (z = 7.88, p < 0.05), child in-school participation (z = 7.18, p < 0.05), and child out-of-school participation (z = 3.38, p < 0.05). Including ASD-related functioning in the model explained an additional 43.6%, 24.0%, and 5.2% of the total variance in family health and well-being, in-school participation, and out-of-school participation, respectively (Table 2).

Discussion

Consistent with the hypotheses, for children diagnosed with ASD or other NDD/D, neurodevelopmental functional characteristics more fully explained variance in family and child outcomes than did the child’s diagnosis, and mediated the apparent effects of diagnosis status on these outcomes. To clinicians, these results may seem intuitive; however, the present findings contribute evidence that may be prove relevant to thinking about and planning services and supports for children with neurodisability.Note 36Note 37 They highlight the importance of functional characteristics at a time when a given diagnosis often continues to be an important and sometimes major criterion in determining eligibility.

Previous studies demonstrated links between specific neurodevelopmental diagnoses and child and family outcomes,Note 38Note 39Note 40 but paid little attention to the possible explanatory role of the child’s functional characteristics. One study found parental stress to be higher in families in which a child had been diagnosed with fetal alcohol spectrum disorder than with ASD,Note 38 but did not investigate variations in child functional or behavioural characteristics. Other studies reported an ASD diagnosis to be associated with higher unmet health care needs and more adverse family impact than is found in families of children with special health care needs in general,Note 39 or in families in which children had other developmental disabilities (Down syndrome, cerebral palsy or developmental delay) or mental health conditions (anxiety or behavioural problems).Note 40 Although the latter study took children’s functional status into account, this was ascertained through a single generic question.Note 40 The composite measures in the present analysis were likely more sensitive because they involved ratings of functioning across an array of domains, and revealed that child functional characteristics better explained, and mediated, the apparent effect of diagnosis status. Kogan et al.Note 39 found that various emotional, developmental and behavioural problems, apart from an ASD diagnosis, predicted family impact and unmet needs, and that the child’s functional ability was the strongest and most robust predictor. Despite these findings, the authors emphasized the role of a diagnosis of ASD in their conclusions.Note 39

A limited body of work has explicitly examined the relative impact of diagnosis status and functional characteristics on caregiver health and service needs. According to Blacher and McIntyre,Note 29 a diagnosis of autism among young adults with cognitive-adaptive disabilities predicted poorer maternal well-being than did undifferentiated intellectual disability, cerebral palsy or Down syndrome. However, in additional analyses, they found the relationship to be almost entirely accounted for by the level of behaviour problems in the child. Other researchers reported that the need for home care supports among families in which a child had intellectual disability or other special needs was most strongly predicted by the child’s degree of impairment in activities of daily living and intellectual disability, not by a diagnosis of intellectual disability per se.Note 10Note 15 Finally, among children with a wide range of chronic conditions and functional difficulties, measures of health services use, personal limitations, and family impact were all predicted by health conditions and by functional difficulties; however, functional difficulties were the stronger predictor for all outcomes except school absences.Note 16

The present study confirms the importance of functional characteristics versus diagnosis status in explaining variations in family impact and child participation.Note 10Note 21Note 41 Demonstration of these relationships in a well-defined population of Canadian children with NDD/D strengthens the relevance of this work to the organization of health, rehabilitative and social services for children with neurodisability. These findings complement evidence of functional, diagnostic, and biological complexity among children with NDD/D,Note 2Note 25Note 42Note 43Note 44Note 45 and of the frequency of co-existing behavioural, emotional or other mental health issues,Note 46Note 47Note 48 which collectively reinforce concerns about the use of DSM or ICD categorical classifications for planning services. This study also illustrates the scope for innovative research on the dynamic inter-relationships among ICF domains.

Limitations

The strengths and significance of these findings should be considered in the context of several limitations. The main variables were ascertained from parents’ reports without corroboration of a child’s diagnosis or functional characteristics by third parties. However, most survey-based research is subject to the same potential limitation. Also, the risk of misunderstanding and bias is reduced in PALS through collection of information by trained, experienced interviewers.

The cross-sectional design of this study reduces the ability to infer that outcomes were the result of differences in children’s diagnoses or functional characteristics. In setting up the model, child ASD-related functioning was considered to be a proximal explanatory variable associated with aspects of child and family outcomes that possibly mediates relationships between diagnosis and “outcomes.” The assumption was that individually measured characteristics tell more about how a child “is” in daily life (with consequences for participation and caregiving), than does knowing the child’s diagnosis category. This assumption is supported by evidence of clinical heterogeneity and functional complexity within diagnosis categories.Note 2Note 24Note 25 In mitigation of this possible limitation, when the models were run with child diagnosis status as mediator, the results and conclusions were unchanged (data not shown).

It might have been preferable to have more fine-grained and investigator-selected items to analyze children’s functional characteristics. Nonetheless, PALS provided consistent and uniform data with which to study diagnosis-functioning relationships among children with neurodisability.

Finally, the filter question used as part of the PALS post-censal strategy may have led to under-representation of people with mental and psychological disability.Note 49 The possible impact of this on child participants, and on the analyses, is unclear.

Conclusion

These findings are most relevant to planning and providing services and supports for children with neurodisability. They are also relevant to clinicians who may focus mainly or exclusively on establishing a diagnosis. An expanded assessment horizon that includes measurement and documentation of individual functional characteristics would enable physicians to partner more fully with providers of ancillary and enabling services.Note 50 Future research might examine predictive interrelationships among diagnosis status, functional characteristics and outcomes for other neurodevelopmental and chronic childhood conditions, and how diagnosis status and child functional characteristics inter-relate in predicting family- and professional-perceived need for services and supports.

Acknowledgements

This work was supported by funds provided to the Canadian Research Data Centre Network (CRDCN) from the Social Science and Humanities research Council (SSHRC), the Canadian Institute for Health Research (CIHR), the Canadian Foundation for Innovation (CFI), and Statistics Canada. Dr. A. Miller and Dr. L.C. Mâsse receive support from the Sunny Hill Foundation for Children and the Child and Family Research Institute, respectively.

The authors thank all staff of the British Columbia Interuniversity Research Data Centre at the University of British Columbia for their support in accessing the data.

Appendix

References
Date modified: