Probabilistic Linkage Creates a Novel Database to Study Bronchiolitis Care in the PICU

Area of Science:

• Pediatric critical care medicine
• Health informatics
• Data linkage methodologies

Background:

• Comprehensive databases are crucial for comparative effectiveness research (CER) in pediatric intensive care units (PICUs).
• Existing Pediatric Hospital Information System (PHIS) and Virtual Pediatric Systems (VPS) databases hold valuable data but cannot be linked due to patient identifier restrictions.
• A method is needed to link these datasets for robust CER.

Purpose of the Study:

• To demonstrate that probabilistic linkage can accurately merge PHIS and VPS data without patient identifiers.
• To create a combined database for CER in the PICU, focusing initially on bronchiolitis patients.

Main Methods:

• Probabilistic linkage was employed to connect PHIS and VPS records for patients admitted between July 1, 2017, and June 30, 2019.
• Key metrics included the percentage of matched records, false-positive match rates, and demographic comparisons between linked and unlinked subjects.

Main Results:

• A high linkage rate of 91% (839 of 920 records) was achieved with a low false-positive rate of 0.5% (4 matches).
• No significant differences were observed in age, comorbidities, illness severity, intubation rates, or PICU length of stay between linked and unlinked patients.

Conclusions:

• Probabilistic linkage successfully created an accurate and representative combined database from PHIS and VPS for patients with bronchiolitis.
• This scalable methodology can be applied across 38 contributing hospitals to build a national CER database for various PICU conditions.