Related Experiment Video
Updated: Jun 6, 2026
Inverse Probability of Treatment Weighting (Propensity Score) using the Military Health System Data Repository and National Death Index
Published on: January 8, 2020
Linking clinical registry data with administrative data using indirect identifiers: implementation and validation in
Sara K Pasquali1, Jeffrey P Jacobs, Gregory J Shook
1Department of Pediatrics, Duke University School of Medicine, Durham, NC; Duke Clinical Research Institute, Duke University Medical Center, Durham, NC 27715, USA. sara.pasquali@duke.edu
Background:
The use of clinical registries and administrative data sets in pediatric cardiovascular research has become increasingly common. However, this approach is limited by relatively few existing datasets, each of which contain limited data, and do not communicate with one another. We describe the implementation and validation of methodology using indirect patient identifiers to link The Society of Thoracic Surgeons Congenital Heart Surgery (STS-CHS) Database to The Pediatric Health Information Systems (PHIS) Database (a pediatric administrative database).
Methods:
Centers submitting data to STS-CHS and PHIS during 2004 to 2008 were included (n=30). Both data sets were limited to patients 0 to 18 years old undergoing cardiac surgery. An exact match was defined as an exact match on each of the following: date of birth, date of admission, date of discharge, sex, and center. Likely matches were defined as an exact match for all variables except ±1 day for one of the date variables.
Results:
Of 45,830 STS-CHS records, 87.4% matched to PHIS using the exact match criteria and 90.3% using the exact or likely match criteria. Validation in a subset of patients revealed that 100% of exact and likely matches were true matches.
Conclusions:
This analysis demonstrates that indirect identifiers can be used to create high-quality link between a clinical registry and administrative data set in the congenital heart surgery population. This methodology, which can also be applied to other data sets, allows researchers to capitalize on the strengths of both types of data and expands the pool of data available to answer important clinical questions.