Objectives: Establish 1) the collaborative goals and infrastructure for multi-national registry autism research, including critical data confidentiality, access and management solutions, and 2) partner characteristics, including local autism administrative and registry features that will inform multi-site analytical decisions and interpretation.
Methods: 1) By consensus, prepare structural framework for the design and implementation of collaborative studies, the collection and sharing of data, and other consortium issues; 2) Site-completed queries regarding data access parameters and site characteristics, including local autism referral patterns, diagnosis, treatment and service provision and registry characteristics (population, time coverage, reporting processes).
Results: Memorandum of agreement specified collaborative parameters. Differences in site requirements concerning data confidentiality and access led to development of options for data access and handling. All sites have access to complete birth population data for their respective countries/states from which autism cases are ascertained. There are site similarities in autism referral and diagnostic evaluation practices and accessibility (e.g., public funding), but some administrative changes over time within sites that could affect ASD reporting (e.g., inclusion of in- and/or outpatient data; services provision). All registries are nation/state-wide and government sponsored; all use ICD diagnostic criteria except for Australia and Israel which use the DSM. Based on each site’s available birth years, unrestricted data pooling from 1980 onwards yielded 26,317 ASD cases from 8.2 million births; 10,544 children with autism were ascertained among 3.4 million births in the years 1987-1994 with the greatest site overlap. For 1987-1994, average birth prevalence was 84.3 per 10,000 births in Denmark, 49.8 per 10,000 births in Finland, and 28.1 per 10,000 births in Western Australia; Sweden, Israel and Norway were similar at 14–15 per 10,000 births.
Conclusions: Options for data access and handling provide flexibility that enhances the collaborative and analytic breadth among current iCARE sites and increases iCARE long-term potential. Given site differences in autism identification, registry reporting and birth prevalence overall and through time, it is necessary to give careful consideration to potential biases associated with ascertainment (e.g., registry-specific variation in ascertainment of different ASD diagnostic or phenotypic subtypes), as well as effects of changes in diagnostic criteria, and their impact on case characteristics and associations with risk factors. Data harmonization and analytic approaches to address some of these issues will be key.