Use of Medicare Claims As a Source for Research Prevalence and Utilization of Medical Care Services in Persons with Autism Spectrum Disorder over and Under Age 65

Background:  There is a lack of information on how adults with autism spectrum disorder (ASD) use medical care services, or how their utilization differs from the general population of persons with other chronic conditions. Persons who are fully disabled under Social Security rules become eligible for Medicare.  Medicare claims may be a useful source studying medical services utilization among adults age 18-64 and 65 and over with ASD.

Objectives:  Our purpose was to evaluate the feasibility of using Medicare claims to study adults with ASD by determining the number of identifiable persons with ASD in a three year timeframe and to describe the structure, types, and limitations of various Medicare claims files for research.

Methods:  The Centers for Medicare and Medicaid Services (CMS) national Limited Data Files (LDS) were used to explore Medicare beneficiaries with ASD diagnoses.  LDS files have limited PHI; each file includes 100% of claims for fee-for-services beneficiaries by calendar year except for the Carrier files which are a 5% sample of national professional claims.  Any single claim with an ICD9-CM diagnosis of 299, 2990, 2998, 29980 was counted as an ASD case. We tracked unique persons over time and summarized prevalence by age less than and over age 65.

Results:  In 2008, there were a total of 36.6 million Medicare Fee-for-service (FFS) beneficiaries.  Of these beneficiaries, 66.8% (28.2 million) were 65 and older, and 23.2% (8.5 million) were under the age of 65. There were 3,118 beneficiaries with an ASD diagnosis. Those under age 65 accounted for 90.4% (2,820) of ASD beneficiaries for 2008, while those 65 and older comprised 9.6% (298).  Findings for 2009 & 2010 were similar.  Over 1,000 persons appear in at least two years (See Table).

Conclusions:  The prevalence of persons under 65 with ASD appears adequate to undertake further study.  LDS file population estimates for the nation are feasible by up-weighting Carrier claims by 20.  Individual outcome may be tracked by linking 5% sample carrier beneficiaries across other files.  Cohort effects will have to be considered given the changes in diagnostic criteria over the lifespan of this population.  The small number of claims for those over 65 is consistent with the first description of the condition by Kanner in the 1940’s.  Older persons in this population likely received the diagnosis at an older age.  Limitations:  The prevalence of ASD in the LDS may be underestimated due to the 5% sample of Carrier claims used in this study.  Lack of Medicaid claims means some utilization is not observed, e.g. dual eligibles and persons in long-term-care facilities.  The covered population likely represents a severely disabled subset of the larger ASD population.  Limitations of the LDS files may be circumvented by use of Research Identifiable files, Standard Analytic Files and Medicare/Medicaid linked files which all include 100% of all claim types by state.