Using deterministic-dominant approach in Linking Birth Certificate Records with Administrative Data: Florida, 1998-2007

Background and Objective: Policy makers require accurate information to inform decision-making. Vital records linked with hospital discharge data have been found to improve the validity of results based on the two sources. We undertook linkage of birth certificate records for the state of Florida covering the previous decade with hospital discharge records obtained from the Florida Agency for Health Care Administration (AHCA). Methods: We primarily applied a deterministic linkage algorithm to perform the merging procedures: Stage 1- infants within AHCA were linked to their mothers' records in AHCA; Stage 2- maternally linked infants from stage 1 were then linked to their corresponding information in vital records; Stage 3- infants unmatched from stage 1 (designated as stage 1b) were then directly and externally linked to vital records information, and subsequently returned to the AHCA data file for merging with their mothers. This step was necessary because only unique identifiers were permitted for linkage purposes in AHCA. Results: Of the total 1,794,772 singleton live births from AHCA, 1,747,336 were successfully linked to the birth certificate records, achieving a match rate of 97.4%. Of these, 208,711 (12.0%) infants were re-hospitalized with visits ranging from 1 (n=167,187 infants) to 20 (n=1) after one year of life. Examination of maternal records showed that 71,564 (4.1%) mothers were re-hospitalized within the first year. Validation procedures confirmed consistency between prevalences of selected feto-maternal conditions with those reported in the literature. Conclusion: Predominantly deterministic linkage algorithms could enhance accuracy of linkage procedures of administrative data to birth records.