Population-Based Registries in Autism Research
Understanding of the epidemiology of autism spectrum disorder (ASD) is benefited by eclectic approaches, study designs and data sources. Contributions to the ASD epidemiologic literature from studies based on Nordic and other population-based registries have increased in recent years, but the strengths and limitations of the registry approach is best understood in the broader context of epidemiologic studies over all.
Describe the population-based registry approach in the broader context of ASD registries overall and illustrate the kinds of epidemiologic research gaps they can and cannot fill, with reference to published and ongoing studies.
The objectives are achieved in three steps: 1) a general overview of registries, with specific ASD examples, and including a description of key epidemiologic features that distinguish the research utility of individual registries; 2) a description of epidemiologic characteristics of Nordic and other population-based registry systems; and 3) a description of select studies based on the Nordic and other population-based registry-system approach that have contributed to ASD epidemiologic questions, including geographic and temporal prevalence variation, quantitative studies of sources of variation in prevalence, mortality, perinatal risk, medical contact patterns during the years of transition from adolescent to adult care settings and multi-generation family morbidity patterns in association with ASD. Each example is described in terms of its registry-based design, results, strengths and limitations, highlighting both published findings and new findings in ongoing studies.
Registries are widely used in ASD research and service provision but their research utility varies depending on the source of new cases, mechanism of inclusion of new cases and data collection. The limitations of individual registries have been mitigated, in part, in the Nordic and other countries by the creation of registry systems. Multiple, single data bases containing limited individual data are linked together using stable unique identifiers to create research data sets that greatly expand the available data per individual. Rigorous data harmonization and data federation approaches have been applied to further expand the research utility of individual national- or state-level registry systems. Registry-system studies have documented geographic and temporal ASD prevalence trends based on harmonized data analyses; quantified effects of systematic registry data collection changes on observed ASD prevalence trends; and elucidated ASD mortality risks considering comorbidity, fine-grained patterns of common perinatal risks, medical contact perturbations during the transition years, and family-wide morbidity patterns that may reflect heritable ASD risk from underlying liability across other morbidities.
Registry-systems have limited data breadth and complexity, however, their robust sample sizes, population representativeness and time depth make them well suited to derive reliable patterns of risk from routinely collected factors, against which results from smaller studies with more in depth, sophisticated data collection can be calibrated.