Use of a Large Administrative Dataset to Examine Health Outcomes in Children with Autism Spectrum Disorders

Thursday, May 17, 2012
Sheraton Hall (Sheraton Centre Toronto)
3:00 PM
D. Spencer1, J. Marshall2, T. Dennen2, G. Yang2, C. J. Newschaffer3, L. J. Lawer4 and A. Jain2, (1)Health Economics and Outcomes Research, Optum Insight, Eden Prairie , MN, (2)The Lewin Group, Falls Church, VA, (3)Drexel University School of Public Health, Philadelphia, PA, (4)University of Pennsylvania, Philadelphia, PA
Background:   Autism spectrum disorders (ASDs) affect a large and heterogeneous group of children and youth.   Most studies of health outcomes in ASD to-date have included only relatively small or narrow clinical samples limiting their generalizability. 

Objectives: To use a large administrative dataset to further our understanding of how children/youth with ASDs differ from those without ASDs in terms of health outcomes and related enrollment and demographic information.

Methods: We conducted a non-concurrent prospective analysis of administrative data maintained by a large private health plan.   We considered medical, behavioral health and pharmacy claims and health plan enrollment information from January 2001 to December 2009.  From this dataset, we identified subjects aged 0-21 with at least one Autism, Asperger’s Syndrome, or Pervasive Developmental Disorder not otherwise specified (PDD-NOS) ICD-9 diagnosis code and a like-aged cohort without these diagnoses.   Subjects with Rett Syndrome and Childhood Disintegrative Disorder diagnoses were excluded.  All included subjects had at least one six-month or longer period of continuous enrollment in the health plan.  Results include two databases, only one of which was potentially linkable to family health information and to an external consumer database of socioeconomic data.  We conducted descriptive analyses characterizing demographics, extent of longitudinal data available, socioeconomic variables when available and frequency of select health outcomes across the cohorts.   

Results: The ASD cohort and comparison cohorts comprised 79,994 and 240,165 children/youth respectively.  For 46,236 children with ASDs and 138, 876 comparison children, sibling and parent health information was also available.  Children with ASDs averaged 39.2 and 41.8 months of continuous enrollment (differed in the two databases) and the comparison group averaged 30.5 and 29.9 months of enrollment.  Thirty nine percent of the cohort of children with ASDs had claims for Autistic Disorder only, 41% for Asperger’s/PDD-NOS only, and 21% for both. As expected 80% of the ASD cohort were male.  In one database, race/ethnicity and household income data were available for ~60% of the ASD cohort and half of the comparison cohort.  Of these, 85% of the ASD and 79% of the comparison cohort were White while 16.5% of the ASD and 24.1% of the comparison cohort had household income <$50K.  Health outcomes were classified using AHRQ’s clinical classification software.  The top ten morbidity areas in the two cohorts overlapped considerably (e.g., common childhood infections, eye and ear conditions) with the expected exception of increased morbidity in developmental, neurologic and behavioral disorders among the children with ASDs.  Morbidity frequency was higher in the children with ASDs for all diagnostic areas, however, some of which may be explained by surveillance bias.   The cohort of children with ASDs averaged 14.9 health care encounters annually compared to 4.4 in the comparison cohort. 

Conclusions: Large cohorts of children with ASDs and comparison children/youth can be identified from private insurance claims.  General descriptive analyses suggest increased morbidity among the children with ASDs but more detailed analyses will reveal if the increases persist after careful consideration of sociodemographic and medical surveillances differences across the two groups.

| More