A Cross-Sectional Cohort Study of a Large, Statewide Medicaid Home and Community-Based Services Autism Waiver Program

Autism Spectrum Disorders (ASD) are neurological developmental conditions characterized by marked impairment in social interactions, deficits in communications skills, and restricted and repetitive behaviors and interests. These impairments may have a substantial and long-lasting impact on the child, as symptoms of ASD often result in problematic peer interactions, disciplinary infractions, difficulties with school adjustment, and poor academic performance (e.g., Gresham et al. 2006). In addition to the negative impact of ASD on the child, the condition has also been associated with major stressors on the family unit, often more so than among families with children with other developmental disabilities (e.g., Down syndrome; Brown et al. 2006). Research has consistently demonstrated a link between child behavior problems and decreased family well-being (e.g., Eisenhower and Blacher 2006; Turnbull et al. 2007), as well as between ASD and decreased family productivity (e.g., parental employment; Kogan et al. 2008; Montes and Halterman 2008).

The rise in documented prevalence of ASD over the last couple of decades has dramatically increased the financial burden of caring for individuals with ASD. One of the most costly services for ASD involves residential placement in which the child lives away from the home and receives 24-h care. However, this resource-intensive approach is financially infeasible for most families, as well as for government budgets when spread across a population of children. In an attempt to meet the public demand for more cost efficient services, states in the US have emphasized the central role of the Medicaid program in providing life-long services and supports specifically tailored to individuals with ASD (Mauch et al. 2011). These specialized services are often administered through 1915(c) Medicaid Home and Community-Based Services waiver programs, and there has been increased federal and state interest in expanding these programs to meet the needs of various populations with developmental disabilities, including ASD. State interest is understandable; in 2009, for example, the per person annual cost to Medicaid of serving a person with a disability in an institution was more than four times the cost of serving an individual in the community through a Medicaid Waiver (Lakin et al. 2010; National Council on Disability, n.d.). In a recent report (Eiken and Lelchook 2013), approximately $20.2 billion was spent in the US in 2010 for services tailored to individuals with developmental disabilities, including ASD. Some waiver programs are specific to ASD and not other clinical populations. In 2010, $65 million was spent on these specific ASD waiver programs in the US.

Specifically for ASD, the Maryland waiver program (henceforth referred to as the “Maryland waiver”) is the largest of its kind and includes a set of services identified to meet the needs of a child with ASD and the family unit: (a) service coordination through the local school system; (b) intensive individual support services (i.e., intensive, one-on-one interventions with the child/youth provided by a direct care worker); (c) therapeutic integration (i.e., structured programs focused on expressive therapy and therapeutic recreation provided in a non-residential setting separate from the home); (d) residential habilitation; (e) respite care; (f) environmental accessibility adaptations for the participants’ home (physical changes to the home which are reasonable and medically necessary to assure a safe therapeutic environment, e.g., installing a locked gate in the backyard), (g) family training; and (h) adult life planning services for transition from waiver services to the adult services delivery system. Established in 2001, the Maryland waiver’s slots were filled by 2002. As a result, the state created an autism waiver registry for families that were interested in applying to receive services but were required to wait until a space on the waiver became available and they were identified as eligible to receive services (this group is henceforth referred to as the “registry”). According to the Maryland State Department of Education (MSDE), which implements the Maryland waiver, over 3,400 children were on the registry at the time of the current data collection.

Despite the appeal of waiver services, it remains unclear if this funding is yielding tangible results, as research evidence is limited. In fact, to our knowledge, only two studies—Eskow et al. (2011) and Warfield et al. (2013)—have collected data and published on outcomes of families participating in ASD-specific waiver services. Eskow et al. (2011) compared a group of families receiving Maryland waiver services (n = 229) to a group of families on the registry (n = 627). The study demonstrated that Maryland’s ASD-specific 1915(c) Medicaid waiver program was statistically associated with improved family quality of life and parental employment status, suggesting some tangible benefits. Using a sample of families receiving services via the Massachusetts autism waiver, Warfield et al. (2013) found that strong collaborative partnerships between families and waiver service providers and officials was related to increased family well-being. These studies were an important first step in establishing an evidence base for a costly and complex service delivery model, but further research is needed for several reasons. First, the Eskow et al. (2011) and Warfield et al. (2013) studies only examined a small subset of outcome variables (e.g., family quality of life and employment status), but waiver services often address ASD impairments across a variety of domains (e.g., academic performance, independent living skills), all of which have remained unstudied. Second, there has been limited research evaluating statistical predictors of improvement¹ among families who are receiving waiver services. Third, the Warfield et al. (2013) study did not have a control group, and in Eskow et al. (2011), the waiver and registry (i.e., control) groups were unmatched, which could be problematic for ruling out critical lurking variables. Indeed, there are no requirements for being placed on the registry, meaning there is no assurance that registry families have a child with ASD. That is, there is no diagnostic screening for ASD until the child is deemed ready to be placed on the waiver. In addition, some families pay out-of-pocket for waiver-like services (e.g., one-to-one intensive behavior therapy) while waiting on the registry. Thus, to improve the quality of the group comparisons, the use of statistical group-matching procedures would facilitate the drawing of causal inferences based on non-experimental data with a higher level of confidence.

To this end, the current investigation collected data on another sample of Maryland families who were enrolled in the waiver program or were on the registry and waiting for enrollment. As a replication of Eskow et al. (2011) and Warfield et al. (2013), our first hypothesis was that waiver status would be associated with more improvement in family quality of life over the last year compared to registry status. For the second hypothesis, we predicted that waiver status would be associated with increased child improvement over the previous year in academic performance, independent living skills, communication skills, peer relationships, and problematic behavior. For the third hypothesis, in the waiver group, we predicted that more improvement over the last year would be associated with increased frequency or use of environmental accessibility adaptations, intensive individual support services, family training, therapeutic integration services, and respite care.

The current study enhanced the Eskow et al. (2011) methodology in several ways. The current study statistically paired families in the waiver and registry groups using propensity score matching. In addition, missing data and inflated error rates were handled using gold-standard techniques (i.e., multiple imputation and false discovery rate correction, respectively). Thus, the current study aims to establish additional and methodologically stronger evidence for the usefulness of the Maryland ASD waiver program, which, in a sense, is an understudied population-level experiment, with the well-being of children with ASD and their families at stake. Evidence for the effectiveness of the Maryland waiver could provide support for this level and type of fiscal support for ASD services.

The current investigation set out to replicate and extend findings on the ASD-specific 1915(c) Medicaid Home and Community-Based Services waiver program in Maryland. Our first hypothesis was confirmed; waiver status was associated with more improvement in family quality of life over the last year compared to registry status. This replicates the results of Eskow et al. (2011) in another sample and using a more rigorous methodology. Unlike Eskow et al. (2011), waiver and registry families were statistically paired using propensity score matching on family income, child age and sex, ASD severity, and the number of years the family had been receiving waiver services or been waiting on the registry. Based on these results, which afford slightly more confidence in causal inference given the matching methodology, there is a medium-sized and positive effect of waiver services on family quality of life. Gardiner and Iarocci (2012) recently inquired about the Eskow et al. (2011) findings, as they were curious about which “services accounted for the observed differences” between the waiver and registry groups (p. 9). The current study provides some data to answer this question. The exploratory analysis of predictors of improvement in the waiver group indicated that family quality of life may be most positively influenced by increased intensive individual support services and therapeutic integration for the child. Ultimately, this is an encouraging finding, given the importance of quality of life of the child with ASD (e.g., it may be a mediator of ASD problem behavior; Garcia-Villamisar et al. 2013), as well as the negative quality of life consequences of parenting a child with ASD (e.g., Johnson et al. 2011).

Our second hypothesis was partially supported. Waiver status was associated with increased perceived child improvement over the previous year in independent living skills, but not academic performance, communication skills, peer relationships, and problematic behavior. Given the age group of the current sample—young adolescence—evidence of improvement in independent living skills is critical. This group is nearing the transition to adulthood, in which independent living is a crucial stepping stone for facilitating productivity of the individual with ASD and his or her family. It is unclear why waiver status was unrelated to child improvement in the other domains, but the age of the sample may also explain the findings. As young adolescents, children in the sample may have started the study having already developed some skills for dealing with problematic behavior, communication, dealing with peers, and academic performance, precluding measurable improvement in these domains over the last year. Indeed, as the children in the waiver group had been receiving waiver services for many years before participating in the study, these other challenges (e.g., academic performance, communication) could have already improved at an early stage in care, when the child was younger. Another non-mutually-exclusive possibility is that waiver services were designed to target independent living skills to a higher degree than these other domains of improvement, especially given the ages of the children with ASD, but data are unavailable to confirm or refute this possibility. One additional possibility is that children require different service elements (to effectively target areas like communication skills) not yet offered by the waiver, suggesting a need for expansion of support. Ultimately, future research that incorporates multi-method assessments of child outcome would help clarify further.

Based on exploratory regression analyses with the waiver group, we also predicted that more improvement over the last year would be associated with increased frequency or use of environmental accessibility adaptations, intensive individual support services, family training, therapeutic integration services, and respite care. This hypothesis was partially supported; more intensive individual support services for the child predicted more improvement in academic performance and independent living skills over the last year, and higher frequency of therapeutic integration for the child was associated with improvement in academic performance. These results highlight the promise of intensive individual support services and therapeutic integration to improve the academic outcome of children on the waiver. It is unclear why increased respite care was associated with less improvement in academic performance among those on the waiver. This finding requires further investigation, although it is possible that increased use of respite care indicates fewer hours that a family member or treatment provider is directly helping a child with his or her academic work. Although no data are available on the details of typical respite care delivery, it is unlikely that respite care providers were tasked with aiding children in the completion of academic work.

A consistent finding in the exploratory regression analysis concerned ASD severity. A higher level of severity was a negative predictor of perceived improvement over the last year in all domains (except problematic behavior, in which there was no relation). This is not surprising—severe ASD is more difficult to treat and may be more refractory to common services. Future research with bigger sample sizes would benefit from exploring ASD severity as a moderator of the relationship between frequency of service use and child improvement. Nonetheless, the findings of the current study suggest that Maryland might benefit from expanded waiver resources for dealing with the more severe segment of children with ASD. Expanded resources might facilitate the inclusion of more evidence-based techniques into the list of approved waiver services, training of more practitioners, and creation of schools and treatment settings designed to meet the specific needs of children with more severe ASD.

The current study has limitations that require acknowledgment. The cross-sectional design limits causal inference regarding the influence of waiver services on child improvement. Despite the strong diagnostics indicators supporting the success of the propensity score matching—which enhances confidence in drawing causal conclusions with non-experimental designs—it is possible that the waiver and registry families differed in some important way not specified in the propensity score model. For example, with respect to non-waiver services utilized in the last year, there were too many to include as statistical covariates and matching variables. However, univariate analyses indicate that the waiver and registry groups did not statistically differ in terms of the frequency of usage in the last year with a wide range of non-waiver services (e.g., early intervention, special education, occupational therapy, speech therapy). Nonetheless, future research would benefit from a prospective design that tracks matched waiver and registry families over time. One lurking variable that was not addressable given the data collection was group differences in cognitive functioning. Future research would benefit from matching on this individual differences variable. In addition, children on the waiver must receive a comprehensive evaluation to confirm ASD diagnosis, but children on the registry do not receive the same evaluation, suggesting that some children on the registry may not formally meet criteria for ASD. Nonetheless, to address this concern, the propensity score matching procedure successfully matched participants on ASD severity, which was rated at a high level for both the waiver and registry groups (see Table 1 for descriptive data). Last, common to community-based survey research, a large percentage of families opted not to participate in the survey. This could result in biased data, as families who volunteered and completed the survey may have specific characteristics that moderate survey responses.

There are some limitations to the survey instrument. The MASS-R contains many sections that have not been psychometrically validated, although considerable care was taken during the construction of the previous version by using focus groups of professionals and families with children with ASD. Furthermore, the patterns of MASS-R findings in the current study (direction and size of effects; adequate Cronbach’s α for the ASD severity total score) were generally consistent with expectations. Future revisions of the MASS-R instrument would also benefit from expanding the Likert scale on the items that measure child improvement to include more variability in participant responses. Another concern with the MASS-R is that measurement of improvement is entirely based on the perspective of a parent or other guardian (e.g., grandmother), which can be affected by factors like parental stress (e.g., Stokes et al. 2011). Seemingly straightforward ratings, like frequency of past service use, can be difficult to recall and report accurately by parents. Future research would benefit from evaluating this research question using more sophisticated assessment techniques, and frequency of service use could be corroborated with administrative data from treatment providers or government agencies.

Maryland maintains a comprehensive and thus expensive ASD-specific 1915(c) Medicaid Home and Community-Based Services waiver program. Eskow et al. (2011) provided preliminary evidence for the benefits of the waiver program to families, and the current study has reinforced this conclusion and extended it to provide additional evidence that the Maryland waiver might be associated with family and child improvement. Future research should focus on prospectively following waiver participants to learn whether the waiver services appear to be reaching an expressed goal of the program, which is preventing out-of-home placement.

In conclusion, while data on child outcomes post-waiver are lacking, findings from the current study are encouraging and expand the growing research base on ASD waiver programs (e.g., Timberlake et al. 2014; Warfield et al. 2013). The waiver might be associated with improvements in a child’s independent living skills and family quality of life. Other than Maryland, which maintains the largest autism waiver, nine other states have adopted a 1915(c) waiver for autism, and seven more have reported interest in doing so (Merryman et al. 2014). Thus, this promising Maryland approach could serve as a useful model for other regions who wish to adopt ASD services to meet the heavy demands of a large public health burden.