A paper with the Staes lab came out today:
Jeffrey Duncan1, Scott P Narus1,2, Stephen Clyde3, Karen Eilbeck1, Sidney Thornton1,2, Catherine Staes1. J Am Med Inform Assoc doi:10.1136/amiajnl-2014-002774
Abstract
Introduction Identity information is often used to link records within or among information systems in public health and clinical settings. The quality and stability of birth certificate identifiers impacts both the success of linkage efforts and the value of birth certificate registries for identity resolution.
Objective Our objectives were to describe: (1) the frequency and cause of changes to birth certificate identifiers as children age, and (2) the frequency of events (ie, adoptions, paternities, amendments) that may trigger changes and their impact on names.
Methods We obtained two de-identified datasets from the Utah birth certificate registry: (1) change history from 2000 to 2012, and (2) occurrences for adoptions, paternities, and amendments among births in 1987 and 2000. We conducted cohort analyses for births in 1987 and 2000, examining the number, reason, and extent of changes over time. We conducted cross-sectional analyses to assess the patterns of changes between 2000 and 2012.
Results In a cohort of 48 350 individuals born in 2000 in Utah, 3164 (6.5%) experienced a change in identifiers prior to their 13th birthday, with most changes occurring before 2 years of age. Cross-sectional analysis showed that identifiers are stable for individuals over 5 years of age, but patterns of changes fluctuate considerably over time, potentially due to policy and social factors.
Conclusions Identities represented in birth certificates change over time. Specific events that cause changes to birth certificates also fluctuate over time. Understanding these changes can help in the development of automated strategies to improve identity resolution.