The dramatic increase in the number of children and adults diagnosed with ASD (Centers for Disease Control and Prevention 2007; Fombonne 2003, 2005), combined with the high cost of their care, has caused many states to consider other alternatives to pay for this care (Shattuck and Grosse 2007). Some states have increased the public contribution to these services. For example, Colorado, Indiana, Kansas, Maine, Maryland, Pennsylvania and Wisconsin, have or are considering Medicaid waivers, which would allow states to use Medicaid funds to pay for services not included in their Medicaid plan, or to cover individuals that otherwise would not be Medicaid eligible. Other states have increased the private contribution to these services. Many private insurance companies severely limit coverage of behavioral health services for individuals with autism or exclude coverage altogether. In response, Arizona, California, Indiana, Kentucky, Louisiana, Mississippi, Pennsylvania, South Carolina, Texas, West Virginia and Wisconsin all recently considered legislation that would mandate private insurance companies to pay for autism-related healthcare services. This legislation has met with varying degrees of success, with legislation not passing in Mississippi, West Virginia or Wisconsin. Successful legislation has varied tremendously in the ages of covered individuals, the types and quantity of covered services, and annual caps on associated expenditures. As of this writing, many other states, including Connecticut, Florida, Illinois, Michigan, Missouri, and Oklahoma, recently have introduced similar legislation.

A major challenge to these insurance mandates has been the concern that they will result in increased premiums for policy holders. Since most policy holders are employers, increases in premiums may be passed onto employees in the form of a net loss of wages or loss of employment, although the effect of mandates on both has been debated in the literature (Collins et al. 2005; Hopkins and Zweifel 2005; Klerman and Goldman 1994; Sommers 2005; Wolaver et al. 2003). Insurance companies therefore have argued that an autism insurance mandate would result in a burdensome rate increase. Autism advocates, on the other hand, have argued that the burden associated with a rate increase would be outweighed by the benefits to individuals with autism and their families.

Sf is number of all children in state, ages 2–20 years. This number was abstracted from US Census data.

F is number of insured children in Commonwealth, ages 2–20 years, with non-ERISA exempt healthcare insurance coverage. This number was derived from the percentage of uninsured children and youth living in Pennsylvania as reported by the Centers for Disease Control and Prevention (Cohen and Martinez 2007) and from the percentage of private-sector enrollees that are enrolled in self-insured plans at business firms offering health insurance in Pennsylvania, as reported in the Medical Expenditure Panel Survey (Medical Expenditure Panel Survey-Insurance Component 2005). Using these sources, the number of children and youth between the ages of 2 and 20 living in Pennsylvania and covered under state-regulated insurance plans, was estimated at 1.37 million.

is mean annual per capita expenditure for autism-related services. This number was estimated from existing studies of healthcare utilization, and ranged from the $2,900 found by (Croen et al. 2006) to the $36,000 cap proposed by the Pennsylvania legislation. Only estimated annual expenditures of $10,000 and higher are presented here. A summary of this literature is presented in the table.

A is assumed load factor for the first year’s administrative and incidental costs associated with the mandate. This number was obtained from insurer comments filed with the Pennsylvania Health Care Cost Containment Council in regard to the implementation of new insurance mandates, and assumed to be 10% of total provider payments attributable to the mandated benefits.

MLR is medical loss ratio. The MLR refers to the proportion of revenues from health insurance premiums spent on medical services covered by the plan. The MLR was calculated by dividing the total medical losses incurred by total premium revenue collected by insurers. The MLR is used to convert increased medical/clinical costs to a revenue requirement needed to encompass both the hard and soft costs attributable to administering health care coverage. An MLR of 85%, considered the industry standard, was used for these calculations (Robinson 1997).

Our analyses imply that the percentage increases in healthcare premiums changed linearly based on changes in assumptions in annual costs and treated prevalence. It is important to note, however that the treated prevalence and the per-child expenditures may interact. That is, as treated prevalence increases, the proportion of children with less intensive medical needs may increase, thereby reducing the average per-child expenditure.

Two study limitations should be noted. First is that these calculations were based on data from existing literature. Dramatic changes in payment available for services to individuals with ASD may affect the treated prevalence and on the provider market. While we attempted to model changes in prevalence up to the accepted community prevalence (Centers for Disease Control and Prevention 2007), we did not model any changes to the provider market. Second, estimated changes to healthcare insurance premiums were based on total costs for children with ASD, not incremental increases, because there is little information on current expenditures for children with ASD not associated with an ASD diagnosis per se (Leslie and Martin 2007; Mandell et al. 2006). Many children with ASD may receive services that are associated with a different diagnosis, as clinicians may assign diagnoses that result in a higher probability of reimbursement. To that extent, the estimates presented here may overestimate actual increases to premiums, given that some healthcare expenditures would remain the same but now would be associated with an ASD diagnosis.