| Description |
Exchanging patient specific information across heterogeneous information systems is a critical but increasingly complex and expensive challenge. Lacking a universal unique identifier for healthcare, patient records must be linked using combinations of identity attributes such as name, date of birth, and sex. A state's birth certificate registry contains demographic information that is potentially very valuable for identity resolution, but its use for that purpose presents numerous problems. The objectives of this research were to: (1) assess the frequency, extent, reasons, and types of changes on birth certificates; (2) develop and evaluate an ontology describing information used in identity resolution; and (3) use a logical framework to model identity transactions and assess the impact of policy decisions in a cross jurisdictional master person index. To understand birth certificate changes, we obtained de identifified datasets from the Utah birth certifificate registry, including history and reasons for changes from 2000 to 2012. We conducted cohort analyses, examining the number, reason, and extent of changes over time, and cross sectional analyses to assess patterns of changes. We evaluated an ontological approach to overcome heterogeneity between systems exchanging identity information and demonstrated the use of two existing ontologies, the Simple Event Model (SEM) and the Clinical Element Model (CEM), to capture an individual's identity history. We used Discrete Event Calculus to model identity events iv across domains and over time. Models were used to develop contextual rules for releasing minimal information from birth certificate registries for sensitive cases such as adoptions. Our findings demonstrate that the mutability of birth certificates makes them a valuable resource for identity resolution, provided that changes can be captured and modeled in a usable form. An ontology can effectively model identity attributes and the events that cause them to change over time, as well as to overcome syntactic and semantic heterogeneity. Finally, we show that dynamic, contextual rules can be used to govern the flow of identity information between systems, allowing entities to link records in the most difficult cases, avoid costly human review, and avoid the threats to privacy that come from such review. |
