Abstract

We examined the association between states' legislative mandates that private insurance cover autism services and the health care–related financial burden reported by families of children with autism. Child and family data were drawn from the National Survey of Children with Special Health Care Needs (N  =  2,082 children with autism). State policy characteristics were taken from public sources. The 3 outcomes were whether a family had any out-of-pocket health care expenditures during the past year for their child with autism, the expenditure amount, and expenditures as a proportion of family income. We modeled the association between states' autism service mandates and families' financial burden, adjusting for child-, family-, and state-level characteristics. Overall, 78% of families with a child with autism reported having any health care expenditures for their child for the prior 12 months. Among these families, 54% reported expenditures of more than $500, with 34% spending more than 3% of their income. Families living in states that enacted legislation mandating coverage of autism services were 28% less likely to report spending more than $500 for their children's health care costs, net of child and family characteristics. Families living in states that enacted parity legislation mandating coverage of autism services were 29% less likely to report spending more than $500 for their children's health care costs, net of child and family characteristics. This study offers preliminary evidence in support of advocates' arguments that requiring private insurers to cover autism services will reduce families' financial burdens associated with their children's health care expenses.

The evidence that children with autism, whose care needs are complex, require greater health care and ancillary services than other children is extensive (Ganz, 2007; Gurney, McPheeters, & Davis; 2006; Kogan et al., 2008; Mandell, Cao, Ittenbach, & Pinto-Martin, 2006). These substantially elevated health care needs translate into high costs borne by the health care system—both the public Medicaid system that is jointly funded by the states and the federal government (Ganz, 2007; Krauss, Gulley, Sciegaj, & Wells, 2003; Mandell et al., 2006) and, to a lesser extent, even, private insurers (Leslie & Martin, 2007). However, families bear high costs for their children with autism, and families' costs of care for their children with autism are disproportionately higher than those of families raising children with other disabilities or health conditions (Fujiura, Roccoforte, & Braddock, 1994; Jarbrink, 2007; Jarbrink, Fombonne, & Knapp, 2003; Kogan et al., 2008; Wang & Leslie, 2010).

Private insurers have borne more limited health care costs associated with treating children with autism because of patterns of exclusion, in which services that are explicitly intended to treat autism are either limited or denied by a diagnostic exclusion for autism. Peele, Lave, and Kelleher (2002) found pervasive exclusion of care among a wide range of private insurers.

To address both gaps in the receipt of adequate health care and the financial burden borne by families, advocates have pushed for state legislation mandating the coverage of health care and ancillary services for children with autism (Autism Speaks, 2007). The intent of these legislative mandates has been to directly tackle the exclusion of autism services. As of June 2011, 26 states had enacted some form of insurance reform legislation. This legislation has taken different forms. Parity laws require private health insurers to provide coverage for services required for autism equal to that provided for other kinds of needs. Other legislative mandates require coverage of certain services (e.g., diagnostic, behavioral support) up to prescribed limits.

Little research has examined the impact of these private insurance mandates for autism services. What research has been conducted to date has examined the effect of autism insurance mandates on private insurance premiums. The few studies along this line have found that legislative mandates thus far have not been associated with the catastrophic costs predicted by opponents of the legislation (Bouder, Spielman, & Mandell, 2009; Reinke, 2008).

What remains wholly unclear is the extent to which these state mandates have been effective in reducing the financial burden borne by families of children with autism. Our aim in this study was therefore to examine the association between state legislative mandates for private insurers and the financial burden of families raising children with autism. We hypothesized that compared with families living in states without such legislative mandates, families of children with autism living in states that passed legislation mandating coverage of autism services would report less financial burden, net of individual and family characteristics, the severity of the child's condition, and the relative wealth of their state of residence. We tested this hypothesis using data from the 2005–2006 National Survey of Children with Special Health Care Needs (NS-CSHCN) and state legislation mandating coverage of autism services.

Method

We took several steps in conducting these analyses. First, we matched individual-level data to state-level data; second, we imputed missing individual-level data using multiple imputation; and third, we subjected the data to multilevel analysis.

Data Sources and Sample

Individual and family data were obtained from the 2005–2006 wave of the NS-CSHCN (Blumberg et al., 2008). The NS-CSHCN was a random-digit-dialed telephone survey representative of the U.S. noninstitutionalized, civilian population of children younger than age 18 conducted by the National Center for Health Statistics. A complete description of the survey methodology and sample are available elsewhere (Blumberg et al., 2008; Maternal and Child Health Bureau, 2007). At least 750 interviews were conducted in every state and the District of Columbia, making population estimates obtained from the NS-CSHCN representative at the state level (Blumberg et al., 2008). The parent or guardian who was interviewed was identified as the person most knowledgeable about the child's health care. The population investigated in this study consisted of children with autism only. On the basis of responses to the question “To the best of your knowledge, does [your child] currently have autism or autism spectrum disorder, that is, ASD?” we determined that 2,123 children with autism were sampled in the NS-CSHCN. The final sample of children with autism, which was adjusted for nonresponse on the dependent variables (n  =  41), included 2,082 families.

Legislation on health insurance was identified from the literature and a Lexis-Nexis search on autism and insurance. The legislation was collected from the offices of state insurance commissioners (Bunce & Prikazsky, 2006; Bunce, Wieske, & Prikazsky, 2006; Bunce, Wieske, & Siedlick, 2007; Crosby, Moore, & Broderick, 2004; National Alliance on Mental Illness, 2007). State median income values for families with children younger than age 18 were obtained from the Annie E. Casey Foundation (2008). The proportion of people in each state living in areas classified as nonmetropolitan was obtained from the U.S. Census Bureau (2006, 2010).

Measures

Dependent variables

We adapted two binary dependent variables from questions in the NS-CSHCN asking families to report the level of out-of-pocket expenses associated with their child's medical care over the 12 months before their interview. A variety of health-related needs specifically for the child with autism were considered. Copayments, medications, special foods, and durable equipment were considered eligible out-of-pocket expenses. However, insurance premiums, deductibles, and reimbursable costs were excluded.

The first dependent variable classified families into two categories: those with any eligible out-of-pocket expenses versus those without (N  =  2,082). The second dependent variable, reported only among those having out-of-pocket expenses, contrasted families spending more than $500 on annual eligible expenses (high) with those spending less than $250 (low). There were 1,280 families with a valid response for this variable, a net loss of 802 families indicating that they did not have any out-of-pocket expenses (n  =  398) or that their level of expense was between $250 and $500 (n  =  404). We compared extremes of expenditures to assess important rather than simply statistical impacts of the legislation.

State independent variables for children with autism

States were classified according to whether they had (a) no legislation; (b) parity legislation, which required coverage of autism services at the same level as other covered health insurance services; or (c) any other type of insurance mandate implemented by 2005. Most states had no legislation; four states (California, Maine, New Hampshire, Virginia) had parity legislation; five states (Iowa, Indiana, Kansas, Kentucky, Tennessee) had other types of mandates. We note that the status of legislation we analyzed was what was in effect in 2005; these mandates have subsequently evolved in some states.

State covariates

To control for confounding effects of the legislation at the state level and reduce state-level variance, we included two state-level covariates. The first was a proxy indicator of state wealth, measured as median income for families with children younger than age 18 in 2005, measured in tens of thousands of dollars. The second was the percentage of families living in nonmetropolitan areas.

Individual covariates

To promote an estimate of the policy effect unconfounded by common correlates of the financial burden of families with children with special health care needs, we examined several child and family characteristics for inclusion in the model. Because of the relatively small within-state sample sizes (an average of 33 children per state for any out-of-pocket costs and an average of 18 for absolute costs of more than $500), our final selection of individual-level covariates was parsimonious. Therefore, of the available child and family characteristics obtained from the NS-CSHCN, and on the basis of the literature on financial burdens experienced by families with children with autism, the final models included these four variables: income relative to the federal poverty level (less than 200%); child's minority status (children reported as being Black, Hispanic, Asian, multiracial, Native American, Aleut, or Pacific Islander); the severity of the child's condition (mild or moderate vs. severe); and an indicator of health insurance status (no public or private insurance).

Analytic Strategy: Hierarchical Generalized Linear Models

The two dependent variables were binary and therefore best examined using logistic regression. Because the independent variable, type of legislation, was a state-level variable, with the children in the sample nested within the 50 states and the District of Columbia, we used multilevel regression techniques (Raudenbush & Bryk, 2002). These approaches account for the commonalities between individuals in a state and adjust the variance or standard error of the legislation variable and state covariates accordingly to reflect the higher uncertainty associated with having fewer observations at that level. Multilevel models that enable estimation of parameters for binomial outcomes are referred to as hierarchical generalized linear models (Raudenbush & Bryk, 2002).

Each type of legislation has the potential to explain residual variance between states, once the individual and state covariates are controlled. The proportion of state-level variance explained by legislation was estimated by taking the difference between the state-level variance parameter estimate in the full model and the model with all individual- and state-level covariates but not the policy variables (Table 1; Raudenbush & Bryk, 2002).

Table 1

Multinomial Multilevel Logistic Regression Predicting Measures of Family Financial Burden

Multinomial Multilevel Logistic Regression Predicting Measures of Family Financial Burden
Multinomial Multilevel Logistic Regression Predicting Measures of Family Financial Burden

Missing data

To reduce the potential bias from deletion of records with missing values (ranging from 3 to 136), we imputed data using SAS Proc MI. We combined the individual logistic hierarchical linear modeling results from 15 imputed datasets, using the methods suggested by Rubin (Graham, 2009; Rose & Fraser, 2008; Rubin, 1987; Schafer, 1997).

Weighting and variance adjustment

Mplus modeling program (Muthén & Muthén, 2010) accommodates the adjustment to standard errors from multilevel clustered data and handles the complex sampling weights of the stratified random sample according to the specification in the NS-CSHCN (the U.S. Census estimates for the age, sex, race, and ethnicity of the population; Blumberg et al., 2008; Carle, 2009). Mplus does not allow for further adjustment to the standard error usually conducted on complex survey data, and we identified no software that performed all three adjustments. Because we used a software that fully adjusts for complex survey data but not for the multilevel aspects of the data, the standard errors for state-level variables were not sufficiently inflated to account for the independence of only 51 observations at this level, a critical factor for the validity of inferences in light of the salience of the state-level policy variables to this investigation.

Results

Table 2 contains a description of the sample, including individual and family characteristics, state characteristics, and dependent variables. Among families of children with autism, 78% reported having any health care expenditures for their child. Of families with out-of-pocket costs, 21% had a burden of between $1 and $249, 24% had a burden of between $250 and $500, and 55% had a burden in excess of $500.

Table 2

Description of the Sample of Children with Autism (N  =  2,082) and State-Level Measures of Income and Insurance Eligibility Thresholds

Description of the Sample of Children with Autism (N  =  2,082) and State-Level Measures of Income and Insurance Eligibility Thresholds
Description of the Sample of Children with Autism (N  =  2,082) and State-Level Measures of Income and Insurance Eligibility Thresholds

Table 3 presents the percentage of families of children with autism within each state that reported any expenditures, annual expenditures between $250 and $500, and annual expenditures of more than $500. Table 3 also includes the state rankings for each state on each of these characteristics and the type of legislation in each state. States are listed in order by their ranking on the percentage of their population of families with high (more than $500) annual expenditures for their children with autism.

Table 3

Percentage of Families of Children with Autism with Financial Burden by State and State Rankings

Percentage of Families of Children with Autism with Financial Burden by State and State Rankings
Percentage of Families of Children with Autism with Financial Burden by State and State Rankings
Table 3

Continued

Continued
Continued

The proportion of each state's population of families raising children with autism that had any health expenditures for their child ranged from 52% in the District of Columbia to 96% in Nevada. The proportion of families that reported spending $250–$500 ranged considerably (2% in Oklahoma to 35% in Colorado), as did the proportion of families that reported spending more than $500 (18% in Rhode Island to 60% in Massachusetts, Missouri, and Utah). The results of the multilevel logistic regression models are reported in Table 1. In the second column, parity legislation was significantly associated with lowered odds of having out-of-pocket costs among families of children with autism (OR [9+]  =  .72, p < .05). The state insurance legislation variables explained 8% of state-level variance. In the third column, both other mandates (OR  =  .72, p < .05) and parity legislation (OR  =  .71, p < .05) were significantly associated with reduced odds of having out-of-pocket costs in excess of $500 per year. The policy variables explained 7% of the state-level variance.

Discussion

Consistent with research reported elsewhere, we found significant state-level variability in the level of financial burden reported by families of children with autism (Shattuck & Parish, 2008). This state-level variability persisted after controlling for characteristics of the family and the individual child, including impairment and whether the child was insured. The important contribution of this study is that after controlling for these covariates, we found a significant association between state legislative mandates requiring private insurers to cover autism services and families' financial burden. Net of family and individual characteristics, families living in states with legislative mandates were less likely to have any financial costs and were less likely to have high financial costs.

A consideration of the limitations of this study is warranted to fairly assess these findings. First, the analyses are correlational, and we cannot infer causality between state programs and family financial expenditures. As such, the results must be interpreted tentatively. Further study could fruitfully examine whether families' financial burden actually declined after implementation of these state legislative mandates, which would provide more robust evidence of the causal link between mandates and family burden. Second, the ordinal measures of household income and families' expenditures cannot fully capture the level of nuance that would be ideal to inform policymakers. Third, we were unable to model parental employment because the NS-CSHCN did not measure it. However, parental employment is strongly associated with insurance status (DeNavas-Walt, 2006) and financial burden (Parish, Seltzer, Greenberg, & Floyd, 2004). Fourth, the specification of autism mandates varies across states, and modeling the effects of the unique program characteristics in each was beyond the scope of this study. Finally, these findings offer preliminary evidence of the relationship between a novel legislative approach and family financial burden. Further research is necessary to understand the long-term effects of these legislative mandates as well as how specific features reduce or increase family financial burden.

Several important strengths offset the study's limitations. First, the sampling design of the NS-CSHCN resulted in a representative sample of children with autism from each state. Second, the use of multilevel regression enabled us to simultaneously examine both individual-level and state-level public health program characteristics that are correlated with families' out-of-pocket spending for their children with autism. As far as we know, this study is the first to analyze the relationship between state legislative mandates for autism services and the financial burden experienced by families raising children with autism.

We cannot infer causality from this study. However, we speculate that families living in states that have implemented mandates may experience a direct reduction in the health care costs they incur to meet the care needs of their children with autism. The results presented here provide initial tentative evidence that families may be able to share the costs of their child's care with private insurers in states with such legislative mandates. In light of the existing evidence of modest costs of implementing these legislative mandates (Bouder et al., 2009) and high costs to families, policymakers should be encouraged to further support initiatives that reduce the financial burden borne by families of children with autism.

Support for this study was provided by the Maternal and Child Health Bureau of the Health Resources and Services Administration (No. R40MC17158).

References

References
Annie
E
.
Casey Foundation
. (
2008
).
Median family (with child) income: 2005
.
KIDS COUNT 2008 data book online
. .
Autism Speaks
. (
2007
).
Arguments in support of private insurance coverage of autism-related services
. .
Blumberg
,
S. J
.,
Welch
,
E. M
.,
Chowdhury
,
S. R
.,
Upchurch
,
H. L
.,
Parker
,
E. K
.,
&
Skalland
,
B. J
.
(
2008
).
Design and operation of the National Survey of Children with Special Health Care Needs, 2005–2006
.
Vital Health Statistics
,
1
((
45
)),
1
188
.
Bouder
,
J. N
.,
Spielman
,
S
.,
&
Mandell
,
D. S
.
(
2009
).
Brief report: Quantifying the impact of autism coverage on private insurance premiums
.
Journal of Autism and Developmental Disorders
,
39
,
953
957
.
Bunce
,
V
.,
&
Prikazsky
,
V
.
(
2006
).
Trends in state mandated benefits: 2006
.
Alexandria, VA
:
Council for Affordable Health Insurance
.
Bunce
,
V
.,
Wieske
,
J
.,
&
Prikazsky
,
V
.
(
2006
).
Health insurance mandates in the states
.
Alexandria, VA
:
Council for Affordable Health Insurance
.
Bunce
,
V
.,
Wieske
,
J
.,
&
Siedlick
,
L
.
(
2007
).
Health insurance mandates in the states
.
Alexandria, VA
:
Council for Affordable Health Insurance
.
Carle
,
A. C
.
(
2009
).
Fitting multilevel models in complex survey data with design weights: Recommendations
.
BMC Medical Research Methodology
,
9
,
49
.
Crosby
,
M
.,
Moore
,
N
.,
&
Broderick
,
C
.
(
2004
).
State parity update
.
Washington, DC
:
American Academy of Child and Adolescent Psychiatry
.
DeNavas-Walt
,
C
.
(
2006
).
Income, poverty, and health insurance coverage in the United States: 2005 (Report No. P60-231)
.
Washington, DC
:
U.S. Census Bureau
.
Fujiura
,
G. T
.,
Roccoforte
,
J. A
.,
&
Braddock
,
D
.
(
1994
).
Costs of family care for adults with mental retardation and related developmental disabilities
.
American Journal of Mental Retardation
,
99
,
250
261
.
Ganz
,
M. L
.
(
2007
).
The lifetime distribution of the incremental societal costs of autism
.
Archives of Pediatric and Adolescent Medicine
,
161
,
343
349
.
Graham
,
J. W
.
(
2009
).
Missing data analysis: Making it work in the real world
.
Annual Review of Psychology
,
60
,
549
576
.
Gurney
,
J. G
.,
McPheeters
,
M. L
.,
&
Davis
,
M. M
.
(
2006
).
Parental report of health conditions and health care use among children with and without autism: National Survey of Children's Health
.
Archives of Pediatric and Adolescent Medicine
,
160
,
825
830
.
Jarbrink
,
K
.
(
2007
).
The economic consequences of autistic spectrum disorder among children in a Swedish municipality
.
Autism
,
11
,
453
463
.
Jarbrink
,
K
.,
Fombonne
,
E
.,
&
Knapp
,
M
.
(
2003
).
Measuring the parental, service and cost impacts of children with autistic spectrum disorder: A pilot study
.
Journal of Autism Developmental Disorders
,
33
,
395
402
.
Kogan
,
M. D
.,
Strickland
,
B. B
.,
Blumberg
,
S. J
.,
Singh
,
G. K
.,
Perrin
,
J. M
.,
&
van Dyck
,
P. C
.
(
2008
).
A national profile of the health care experiences and family impact of autism spectrum disorder among children in the United States, 2005–2006
.
Pediatrics
,
122
,
e1149
e1158
.
Krauss
,
M. W
.,
Gulley
,
S
.,
Sciegaj
,
M
.,
&
Wells
,
N
.
(
2003
).
Access to specialty medical care for children with mental retardation, autism, and other special health care needs
.
Mental Retardation
,
41
,
329
339
.
Leslie
,
D. L
.,
&
Martin
,
A
.
(
2007
).
Health care expenditures associated with autism spectrum disorders
.
Archives of Pediatric and Adolescent Medicine
,
161
,
350
355
.
Mandell
,
D. S
.,
Cao
,
J
.,
Ittenbach
,
R
.,
&
Pinto-Martin
,
J
.
(
2006
).
Medicaid expenditures for children with autistic spectrum disorders: 1994 to 1999
.
Journal of Autism and Developmental Disorders
,
36
,
475
485
.
Maternal and Child Health Bureau
. (
2007
).
The national survey of children with special health care needs: Chartbook 2005–2006
.
Rockville, MD
:
Author
.
Muthén
,
L. K
.,
&
Muthén
,
B. O
.
(
2010
).
Mplus (6th ed.) [Modeling program]
.
Los Angeles, CA
:
Author
. .
National Alliance on Mental Illness
. (
2007
).
State mental health parity laws 2007
. .
Parish
,
S. L
.,
Seltzer
,
M. M
.,
Greenberg
,
J. S
.,
&
Floyd
,
F
.
(
2004
).
Economic implications of caregiving at midlife: Comparing parents with and without children who have developmental disabilities
.
Mental Retardation
,
42
,
413
426
.
Peele
,
P. B
.,
Lave
,
J. R
.,
&
Kelleher
,
K. J
.
(
2002
)
Exclusions and limitations in children's behavioral health care coverage
.
Psychiatric Services
,
53
,
591
594
.
Raudenbush
,
S
.,
&
Bryk
,
A
.
(
2002
).
Hierarchical linear models: Applications and data analysis methods (2nd ed.)
.
Thousand Oaks, CA
:
Sage
.
Reinke
,
T
.
2008
.
States increasingly mandate special autism services. Insurers are being handed some treatment responsibility for this complex disorder, but some say that it is an educational, not medical, task
.
Managed Care, 17
,
35–36
,
39
.
Rose
,
R
.,
&
Fraser
,
M
.
(
2008
).
A simplified framework for using multiple imputation in social work research
.
Social Work Research
,
32
,
171
178
.
Rubin
,
D. B
.
(
1987
).
Multiple imputation for nonresponse in surveys
.
New York, NY
:
Wiley
.
Schafer
,
J
.
(
1997
).
Analysis of incomplete multivariate data
.
Boca Raton, FL
:
Chapman & Hall/CRC
.
Shattuck
,
P. T
.,
&
Parish
,
S. L
.
(
2008
).
Multilevel analysis of financial burden in families of children with special health care needs
.
Pediatrics
,
122
,
13
18
doi:10.1542/10.1542/peds.2006-3308
.
U.S. Census Bureau Population Division
. (
2006
).
Annual estimates of the population for counties: April 1, 2000 to July 1, 2005 (CO-EST2005-01-04)
.
Washington, DC
:
Author
.
U.S. Census Bureau Population Division
. (
2010
).
Metropolitan and micropolitan statistical areas and components, December 2009, with codes
.
Washington, DC
:
U.S Census Bureau
.
Wang
,
L
.,
&
Leslie
,
D. L
.
(
2010
).
Health care expenditure for children with autism spectrum disorders in Medicaid
.
Journal of the American Academy of Child and Adolescent Psychiatry
,
49
,
1165
1171
.

Author notes

Editor-in-Charge: Glenn Fujiura

Authors:

Susan L. Parish (e-mail: slp@brandeis.edu), Lurie Institute for Disability Policy, Heller School for Social Policy and Management, Brandeis University, 415 South Street, MS 035, Waltham, MA 02454, USA; Kathleen Thomas, Cecil G. Sheps Center for Health Services Research; Roderick Rose, University of North Carolina; Mona Kilany and Robert McConville, University of North Carolina at Chapel Hill.