Validating an administrative data-based case definition for identifying children and youth with autism spectrum disorder for surveillance purposes

OBJECTIVES: To evaluate the sensitivity and positive predictive value (PPV) of administrative health and education data for identifying cases of autism spectrum disorder (ASD) in Manitoba, and to recommend a surveillance case definition. METHODS: Four service providers abstracted information on children who had been clinically diagnosed with ASD (sensitivity cohort). That information was linked to Manitoba's administrative health and education data and records were extracted into the study dataset. Records were also included for children who had an administrative diagnosis of ASD but who were not part of the sensitivity cohort. Study packages were mailed to the parents of the latter group in order to verify their diagnostic status. The sensitivity and PPV of various case definitions were calculated. RESULTS: Among the 1728 service provider-reported cases, 1532 had an administrative diagnosis of ASD. A total of 2414 children had an administrative diagnosis, of whom 882 were not part of the sensitivity cohort. The response to the mail-out was very poor (< 3%). Accordingly, we calculated minimum PPVs. Our recommended surveillance case definitions are >= 1 physician claim (ICD-9-CM 299) or >= 1 ASD special education record (2-5 years of age), and >= 2 physician claims or >= 1 ASD special education record (6-14 years of age). The sensitivity ranged from 80% (95% CI: 77-83) to 88% (95% CI: 83-91) and the minimum PPV from 70% (95% CI: 67-73) to 78% (95% CI: 75-81) for these definitions. CONCLUSION: This work advances the goal of establishing a cost-effective national surveillance system for ASD.