University of California, Davis
Basic & Clinical
This study aims to: 1) estimate national annual prevalence costs of all US persons combined with autism for 2010 and for different groups separately: only ages 0-9; 10-19; 20-64; and ages 65+; males; females; 2) forecast costs for the same groups as in Aim 1 for years 2016 and 2020; and 3) generate transparent methods to allow annual updates by other researchers. The investigators will take the cost-of-illness approach that estimates annual total cost of existing disease. Costs are divided into direct and indirect categories. Direct costs include, for example, adult care programs, physician and therapist visits, hospitalizations, medications, paid caregivers, and special education. Indirect costs use the Human Capital method and include morbidity and mortality categories that measure lost wages, lost fringe benefits, and lost home production. Morbidity costs apply to both persons living with disease and unpaid (family) caregivers. Mortality costs apply to persons who prematurely die from causes that can be reasonably attributed to autism. Taking the societal perspective, the study applies a 3% discount rate and uses the prevalence method as opposed to incidence. The incidence method has data difficulties and requires heroic assumptions; moreover, incidence-based cost estimates cannot be compared to the great majority of cost estimates for other diseases such as cancer or diabetes because the latter use the prevalence method.
Broadly speaking, the approach multiplies estimates of prevalence of autism with average per-person costs of autism for 8 demographic groups and more than 30 categories of costs. The study is based on secondary data and figures from the literature to generate these estimates, followed by the significant innovation of analyzing cost from data from roughly 32,000 children and 6,000 adults with autism in the California Department of Developmental Services (CDDS) databank. The CDDS data are especially important for addressing the Autism Speaks priority area involving "factors that influence quality of life for adults with autism" because data are available for costs of adult care at home and at licensed facilities. Forecasts with be generated with regression models using 10 most recent years of data and projections of prevalence from the literature. Because of the lack of primary data, considerable attention will focus on a sensitivity analysis that allows for varying parameter estimates.