 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ARTICLE |
a RTI International, Research Triangle Park, North Carolina
b FPG Child Development Institute, University of North Carolina, Chapel Hill, North Carolina
c SRI International, Menlo Park, California
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
METHODS. A nationally representative sample of 
2100 parents completed a 40-minute telephone interview near their child's third birthday. Structural equation modeling examined the relationships between 3 support variables (quality of child services, quality of family services, and family/community support) and 2 outcomes at 36 months (impact on child and impact on family) and determined whether these relationships were mediated by 2 perceptual variables (optimism and confidence in parenting) or moderated by 5 demographic variables (poverty, maternal education, ethnicity, age of initial Individual Family Service Plan, and health at 36 months).
RESULTS. Perceived impact of early intervention on both child and family were significantly related to each other. The quality of child services was related to impact on the child but not on the family. The quality of family services was related to both child and family impact. Informal support was not related to perceived impact on children or families but was strongly related to confidence in parenting and optimism. Neither optimism nor confidence in parenting mediated the relationships between services or supports and perceived impact. Minority families and families of children with poor health reported lower quality of services, but these characteristics did not moderate the relationships between services and perceived impact on the child. However, both poverty status and minority status were associated with perceptions of impact on the family.
CONCLUSIONS. Findings reinforce the role of high-quality services in maximizing perceived impact. They also highlight the important role of informal support in promoting optimism and confidence in parenting. Poverty status, minority status, and poor health of the child are salient factors in predicting lower perceived quality of and benefit from services.
Key Words: early intervention • support systems • structural equation modeling
Abbreviations: NEILS—National Early Intervention Longitudinal Study • SEM—structural equation modeling • IFSP—Individual Family Service Plan • RMSEA—root-mean-square error of approximation • CFI—comparative fit index
Early intervention for children with or at risk for disabilities is grounded in a public health prevention model, with the fundamental assumption that timely identification and individualized services can maximize development, reduce secondary conditions, and promote positive family adaptation.1–3 Research suggests that the early years are formative and that optimal development is influenced by the amount, quality, and timing of early experiences.4–6 High-quality early intervention can significantly influence the development of children with disabilities and benefit families.7–9 Numerous studies have examined specific components of early intervention, ranging from models for service coordination,10 location of services,11 or curriculum approaches12 to more focused studies, such as intense behavioral interventions for children with autism,13 increasing maternal responsivity to enhance early communication development,14 or teaching parents to embed teaching strategies in caregiving.15
However, the outcomes experienced by young children with disabilities and their families will be influenced by a complex interplay of child, family, program, and community variables.1,2,16,17 Specialized services (such as those provided by pediatricians, therapists, or special educators) constitute only one aspect of support, and the benefits of early intervention could be enhanced or reduced depending on other resources and the extent to which they are integrated and coordinated. Of recent interest is the role of informal family and community support. Research consistently shows that families with strong support systems are able to handle challenges more effectively than families with few supports.18 A longitudinal study of children participating in early intervention showed that social support and problem-focused coping were related to changes in both maternal and paternal parent-related stress.19 Thompson et al20 found that family centered early intervention practices had a direct relationship to feelings of empowerment and an indirect relationship where enhanced social support reduced parenting stress, which was related to feelings of empowerment. Similarly, Dunst21 found that family centered practices were related to feelings of personal control, which was related to parenting supports provided by service providers and ultimately to child progress. A second set of analyses showed a path leading from social support to personal well-being, with resultant effects on parent interaction styles and, ultimately, to child development.
Helping families strengthen informal supports may not be viewed by pediatricians as central to their work or even possible, especially in an era of managed care, limited resources for both health care and early intervention services, and a primary policy focus on outcomes for children. But in a recent survey, families of children with disabilities gave the lowest ratings of satisfaction to their physician's ability to understand the impact of disability on the family or to link them with other families,22 despite research showing the benefits of programs such as parent-to-parent support.23 More research is needed to understand the respective roles of formal and informal supports in family and child adaptation.
The National Early Intervention Longitudinal Study (NEILS) recruited a nationally representative sample of families shortly after they enrolled in early intervention to describe program participants, services, and outcomes. Previous NEILS publications have provided conceptual models for family outcomes24 and services25; described the children and families participating in early intervention26–28; examined families' initial experiences29; and described family outcomes near the end of early intervention.30 These articles have shown that parents generally are pleased with services and report positive outcomes. However, some families are less satisfied and report poorer outcomes, prompting questions regarding factors associated with outcomes.
To date, our research has focused on child and family variables related to satisfaction and outcomes.29,30 Missing from these analyses is an examination of how formal and informal supports are related to family outcomes and the likely mechanisms of influence. This article uses structural equation modeling (SEM)31 to examine relationships between 3 support variables (quality of child services, quality of family services, and family/community support) and 2 outcomes at 36 months of age (impact on child and impact on family). We then examine whether these relationships were mediated by optimism in the family's future or confidence in parenting (2 key indicators of successful family adaptation24) or moderated by 5 demographic variables (poverty, maternal education, ethnicity, age of initial Individual Family Service Plan [IFSP], and health at 36 months).
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Participants
This article is based on 2166 families with complete data on the variables described below of the 2586 families who were located and agreed to participate in the 36-month interview around the time the children completed early intervention. Data were weighted based on the initial study sample to provide national estimates.
Instrument and Procedure
A telephone interview (www.sri.com/neils) lasting 40 minutes was conducted with the "person able to answer questions about the child and the child's program." This article is based on interview items asking parents about quality of services, informal support, optimism about the future, confidence in parenting, and perceived program impact on child and family.
Quality of Child Services
Parents rated the quality of therapy (physical therapy, occupational therapy, and speech-language therapy) and other early intervention services (eg, early childhood special education) on a 5-point scale (excellent, good, fair, poor, or mixed). These 2 items were combined into a composite item for quality of child services, with 3 levels: high (3; if both therapy and other services were excellent), moderate (2; if 
1 type was good), and low (1; if either were fair, poor, or mixed).
Quality of Family Services
Parents rated the quality of help the family received from early intervention using the same 5-point scale. This item was reduced to 3 levels: high (excellent), moderate (good), and low (fair, poor, or mixed).
Family and Community Support
Three items asked whether families took part in community activities, such as religious, school, or social events; had relatives, friends, or others to help deal with challenges; and were able to work and play normally together as a family. These items were used to create a latent factor (family and community support) with acceptable loadings of 0.52, 0.68, and 0.85, respectively.
Confidence in Parenting
Parents rated the extent to which they knew (1) how to care for their child's basic needs, (2) how to help their child learn and develop, (3) how to work with professionals and advocate for their child's needs, and (4) what to do if they were worried that their child was not getting good services. These items were used to create a latent factor (confidence in parenting) with acceptable loadings of .86, .88, .89, and .70, respectively.
Optimism for the Future
A single item asked parents, "looking toward the future, do you expect that your family's overall life situation will be excellent, good, fair, or poor?" This item was used as a 4-category measure of optimism.
Impact of Services
Parents rated the impact of early intervention on their child (none, some, a lot, or too soon to tell) and on their family (worse off, about the same, somewhat better, or much better). Because of skewed distributions, these items were categorized in 2 levels: positive impact (a lot or much better) and not very much impact.
Family Demographics
We included data on poverty (poor or not poor), maternal education (college versus less than college), and ethnicity (white versus nonwhite).
Child Characteristics
Two child characteristics were included, age the first IFSP was written (to account for length of time in early intervention) and health status at 36 months (poor, fair, good, very good, or excellent) as rated by the respondent to account for the added challenges that poor health poses to positive child and family outcomes.30
Analysis Strategy
Analysis Plan
SEM was used initially to examine the direct relationships between 3 support variables (quality of child services, quality of family services, and family/community support) and 2 outcomes at 36 months of age (impact on child and impact on family). We then examined whether these relationships were mediated by 2 family perceptual variables (optimism in the future and confidence in parenting) or moderated by 5 demographic variables (poverty level, maternal education, ethnicity, age of first IFSP, and health at 36 months).
Several methodologic issues made the analysis complex, requiring Mplus 3.12.32 The NEILS data were obtained under a complex survey design. Mplus accommodates this by incorporating survey weights. NEILS used balanced repeated replication for variance estimation. To adjust the SEs estimated under a conventional SEM analysis in Mplus for the balanced repeated replication method, we followed steps outlined by Stapleton.33 An analysis was run for each set of replicate weights. The variability in each regression coefficient across these sets was taken to estimate the SE. The average design effect was also computed and used to adjust the overall 
2 goodness-of-fit statistic.
Mplus allows for categorical outcomes. The measurement level for the constructs examined was either binary (ie, "0" and "1") or ordinal (ie, "0," "1," and "2"). Categorizing an unmeasured continuous response segments the underlying distribution. However, not everyone assigned "2" has the same degree of certainty. For categorical outcomes, Mplus developers recommend using the weighted least-squares estimator and the probit link function.34 In probit regression analysis, a normal continuous cumulative distribution function "links" the probability of responding more positively on the outcome to predictors.35 Therefore, in all of the analyses, the categorical outcomes have been transformed such that a positive regression coefficient means that a 1-unit increase in the predictor increases the probability of responding positively (ie, increases the probit), whereas a negative regression coefficient means that a 1-unit increase in the predictor decreases the probability of responding positively (ie, decreases the probit).
Model Description
Separate SEM analyses were run to assess whether perceived services or informal supports were related to 2 impact outcomes: perceived impact on child (Fig 1) and perceived impact on family (Fig 2). For each outcome, 6 mediation hypotheses were simultaneously tested by allowing direct and indirect paths among the predictors, mediators, and outcome. Three hypotheses examined whether confidence in parenting mediated the relationship between quality of child services, quality of family services, or informal support and impact. The second set of hypotheses examined whether optimism mediated the relationship between quality of child services, quality of family services, or informal support and impact.
|
|
2 difference test, which adjusted for the average design effect, was computed.32 When the difference was not significant, there was no evidence of moderation, and the model averaging across the groups was selected.
After the moderation hypotheses were tested, we adjusted the relationships indicated in Figs 1 and 2 for nonmoderating child and family characteristics by allowing direct paths from these covariates to the parent's perception factors and outcome. Model fit was assessed by multiple conventional goodness-of-fit statistics. The 2 overall goodness-of-fit test statistic was adjusted by the average design effect of 2.5. Following Hu and Bentler,36 we selected 2 widely used measures to supplement the 2 statistic, both less sensitive to measurement model misspecification. The root-mean-square error of approximation (RMSEA)37 and the comparative fit index (CFI)38 each determine factor-loading misspecification. The RMSEA is an absolute fit index. Hu and Bentler36 suggest that a value 
.06 represents good fit. The CFI is an incremental fit index that measures the proportion of improvement in fit between the model and a null model; Hu and Bentler39 suggest that a value .95 represents good fit. Furthermore, we checked for large residuals to determine how well the sample covariance structure fit the proposed model's covariance structure.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
y" in this notation, where β indicates the strength of the association. Double-headed arrows indicate a correlation.
|
|
582 equaled 119.5, after the unadjusted statistic of 298.82 was divided by the average design effect of 2.5. The associated P value is <.05; the CFI equaled 0.96, and the RMSEA equaled 0.04. These statistics indicate a tolerable fit, and the residuals were observed to be low (–0.20 to 0.20). The significant regression coefficients and SEs are shown in Fig 1. Table 2 contains paths included in the model but not shown in Fig 1.
The 3 contextual variables (quality of child services, quality of family services, and informal support) were interrelated and possibly mutually interdependent constructs. Also, perceived impact on the child was correlated with perceived impact on the family (r = 0.54; P < .001). Adjusting for all other variables and covariates, a direct and significant path was found between perceived quality of child services and impact of services on the child (β [SE] = .36 [.17]; P < .05). We also found a direct and significant path between the quality of family services and the ratings of child impact (β [SE] = .33 [.07]; P < .05). Quality of family services was significantly related to optimism about the future (although the effect size was small) but not to confidence in parenting competence. Informal support was strongly related to both optimism about the future and confidence in parenting confidence but not to ratings of impact on the child. Confidence in parenting and optimism in the future were not directly related to the probability of perceiving a positive impact on the child. Because the direct paths between the potential mediators and the outcomes were not significant, we did not test the significant of the indirect association between the predictors and outcomes through the hypothesized mediators. Furthermore, the adjusted 2 difference test for the 3 separate multiple-group analyses of poverty, maternal education, and ethnicity indicated no evidence of moderation. Therefore, the final model controls for child and family characteristics by allowing direct paths between the covariates and each construct. Of those paths that were directly associated with an increase in the probability of perceiving a positive child impact, families characterized as white and reporting better child health tended to report more positive quality in child services (respectively, β [SE] = .08 [.04], .06 [.02]; P < .05), as well as family services (respectively, β [SE] = .04 [.02], .12 [.04]; P < .05). However, these characteristics were not directly associated with an increase in the probability of perceived child impact.
Impact on Family
When examining family impact, the unadjusted overall goodness-of-fit 582 was 306.72. Divided by the average design effect of 2.5, the adjusted 582 equaled 122.7 (P < .05); the CFI equaled 0.96, and the RMSEA equaled 0.04. This provides evidence for a moderately good fit, and the residuals were low, ranging between –0.20 and 0.20. The significant regression coefficients and SEs are shown in Fig 2. Table 2 contains paths that were fitted in the model but not shown in Fig 2.
Quality of family services (but not child services) was directly associated with impact on family, after adjusting for all of the other variables and covariates in the model (β [SE] = .42 [.07]; P < .001). There was no evidence for a direct or mediating effect of confidence in parenting, optimism in the future, or informal supports on predicting the probability of perceiving positive family impact. There was no evidence of moderation using the adjusted 2-difference test for the 3 separate multiple-group analyses of poverty, maternal education, and ethnicity. The final model controls for child and family characteristics by allowing direct paths between the covariates and each construct. Of those paths that were directly associated with an increase in the probability of perceiving a positive child impact, families characterized as white and reporting better child health tended to report more positive perceptions of quality in family services (β [SE] = .04 [.02], .12 [.04]; P < .05). However, after adjusting for perception of quality of family services, we found that families characterized by poverty tend to show an even greater decrease in the probability of perceiving a positive impact on their family in comparison with nonpoor families (β [SE] = –.43 [.09]; P < .001), and families characterized as white tended to be associated with a further increase in the probability of perceiving a positive impact on their family (β [SE] = .47 [.09]; P < .001).
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Second, the cross-sectional design of the study does not allow us to make definitive statements about causal relationships. We proposed and tested a plausible model based on the research literature, but we did not compare this model with other possible models nor did we examine change over time. There was a longitudinal component to the study (in that data were collected at the end of early intervention, and families were asked to rate the quality of services over the entire period of time that they were in early intervention), but the data were still collected at a single point in time, and, thus, we cannot provide a definitive test of directionality in the model.
Third, the dependent measure was not a direct measure of child or family change but rather family perceptions of the impact of early intervention on the child and on the family. The lack of a direct statistical relationship between informal support systems and child or family impact, thus, should not be interpreted as suggesting that informal support has no bearing on family or child outcomes; rather, it means that informal support is not related to how families perceive early intervention working to help them. Indeed, the fact that informal support is related to both optimism and confidence in parenting suggests that both informal support and early intervention services are important to families but that they operate in different ways with different effects.
Summary of Findings
Quality of Child Services
Quality of child services was directly related to perceived impact of early intervention on the child. The survey forced parents to factor multiple dimensions of quality into a single rating, and, thus, we cannot say what aspects of quality were salient. However, the data show that, although the vast majority of families rated child services high in quality, we still were able to detect a statistically significant relationship between quality and impact.
Interestingly, the quality of child services was not related to optimism about the future, confidence in parenting, or ratings of overall impact on family. This finding suggests that professionals should not assume that doing a good job (indeed, a very good job) of providing quality services to children will necessarily result in improved outcomes for families. Families differentiate child and family services when asked to rate quality of services, and they do not perceive quality of child services to be related to family impact.
Although families characterized as white and having children with better health outcomes were more likely to report positive perceptions of the quality of child services, this finding did not moderate the relationship between perceived quality and outcomes for children. In other words, quality of child services mattered in the same way for all of the children and families, at least when considering the variables examined in this study.
Quality of Family Services
The quality of family services was related to perceived impact on the family. Despite the relatively brief interactions that early intervention professionals have with families (typically 1–2 hours/week), families perceive these interactions as high in quality and effective, benefiting their family. The quality of family services was also related to optimism about the future, suggesting that early intervention can play an important role in promoting hope, a key component of positive family adaptation.
Interestingly, however, the quality of family services was not related to parents' ratings of confidence in parenting. A long-standing goal of early intervention from its inception has been to enhance the capacity of families to meet the special needs of their infants and toddlers, and much has been written about how this can be accomplished. Although families generally rated themselves as having high confidence, their ratings did not vary relative to perceived quality of family services. However, quality of family services was directly related to parents' perceptions of impact on their child. This finding supports long-held assumptions that working with families cannot only benefit the family but also can benefit the child.
Ethnicity and child health were associated with quality of services and with the perceived impact on families. Nonwhite families and families of children in poor health were likely to report lower quality of family services and lower family impact.
Family and Community Support
Informal support from family and community was not related to perceived impact of services on either the child or the family. In some ways this might be expected, because these are supports not provided by early intervention, and, thus, families would not factor informal support into their ratings of the impact of early intervention. However, informal support was related to both optimism for the future and confidence in parenting.
Ethnicity and child health were related to family and community support; families of children from ethnic minority families and children with poor health were more likely to report lower levels of family and community support. Poverty was associated with confidence in parenting; low-income families were more likely to report lower levels of confidence in parenting. Poor child health was related to lower optimism about the future. However, the relationships among informal support, optimism, and parenting confidence were not moderated by these characteristics.
Quality of child and family services was related to social support, but quality of services and social support probably interact in complicated ways not fully explained in this study. For example, Thompson et al20 showed that quality (as measured by family centered practices) was related to the enhancement of social support systems that reduced stress and ultimately led to feelings of empowerment. Dunst21 reported that building social supports led to personal well-being, improved parent interaction styles, and, ultimately, to improved child development. Both of these studies incorporated relatively small sample sizes in defined geographical areas and did not fully examine the moderating effects of child and family variables but were able to assess constructs more thoroughly than was possible in the NEILS. Research that combines larger and more representative samples with a comprehensive and independent assessment of constructs is needed to more fully understand the dynamics of the relationship between provision of services and informal sources of support.
Interrelationships Between Child and Family Services and Outcomes
Perceived quality of child and family services was significantly related. Likewise, perceived impact on the child was related to perceived impact on families. These findings reinforce the interrelated nature of both services and outcomes.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Although additional research is needed to more fully explicate the relationships among these and other service variables, the findings suggest that professionals should pay more attention to the family consequences of disability and to helping families build and access informal support systems. Recent pediatric literature continues to reinforce the importance of partnerships between medical professionals and families of children with special health care needs,42 the benefits of high-quality communication and informal interactions between pediatricians and families,43 and the importance of physicians recognizing the impact of disability on the family and the salience of informal support systems in long-term family adaptation.22 Also, early intervention and medical professionals need to deliver services that families consider high quality, not just because families will be more satisfied, but because these are the services that families see as making the greatest impact on both their child and their family. This is true for all families, but may be especially important for those most vulnerable (eg, minority families and those where the child has ongoing health care needs). Health ratings were associated with quality of child services, quality of family services, family and community support, and optimism about the future. This finding reinforces the need for pediatricians to be actively involved in early intervention, especially for children with poor health, showing that the ramifications of poor health for young children with delays and disabilities extend well beyond the health arena.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Address correspondence to Donald B. Bailey, Jr, PhD, RTI International, 3040 Cornwallis Rd, Research Triangle Park, NC 27709-2194. E-mail: dbailey{at}rti.org
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. B. Bailey Jr, F. D. Armstrong, A. R. Kemper, D. Skinner, and S. F. Warren Supporting Family Adaptation to Presymptomatic and "Untreatable" Conditions in an Era of Expanded Newborn Screening J. Pediatr. Psychol., March 30, 2008; (2008) jsn032v1. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2007 American Academy of Pediatrics. All rights reserved. |
||
|
||
