in Hospital Vital Status After Interventions for Congenital Heart Diseases: Completeness of Case Ascertainment Using the National Death Index

Background: With today's advancements in surgical or transcatheter interventions about 90% of infants with congenital heart diseases (CHD) are expected to reach adulthood. However, the long term altered history of patients surviving interventions for CHD remains largely unknown. We aim to study mortality patterns in the cohort registered in the Pediatric Cardiac Care Consortium (PCCC) after undergoing interventions for CHD between 1982 and 2003. The PCCC is the longest registry for pediatric cardiac interventions in the world containing data from surgical and transcatheter procedures from 27 U.S. states. Methods: We linked the PCCC dataset with the National Death Index (NDI) using direct identifiers to determine vital status of individuals with repaired or palliated CHD. The identifiers included first name, middle name, last name, day, month and year of birth, state residence at birth and sex. A total of 59,324 subjects registered in the PCCC as having undergone an intervention for CHD up to April 2003 (date of implementation of HIPAA rules) had sufficient identifiers and were submitted for linkage with the NDI. This number includes also 4,655 inhospital deaths recorded in the PCCC registry. After linking with the NDI, these records are assigned by NDI a probabilistic score based on the NDI algorithm and are identified as true or false matches. We tested for completeness of case ascertainment of subjects with known vital status based on their last admission in a PCCC center. Results and Conclusions: Of the 4,655 known deaths submitted to the NDI, 3,906 were identified as “true” matches, 381 as undetermined matches and 368 deaths had no match. Based on these numbers the sensitivity of the PCCC-NDI match is within the range of 82.8% to 85.0% [95% confidence interval]. There was no difference in sensitivity across 48 congenital cardiovascular conditions after adjustment for age, sex, year and state of death. An additional 3,722 deaths were identified through 12/31/2012 that occurred outside the PCCC. These events provide a conservative estimate of deaths that forms the basis for estimating the long term survival of this cohort.