Review of Mother-Infant Linkage Algorithms in Health Care Data Sources.

Area of Science:

• Pharmacoepidemiology
• Health Informatics
• Data Linkage

Background:

• Assessing infant health outcomes after maternal medication use requires linking maternal and infant health records.
• Existing linkage methods are varied and lack comprehensive evaluation for research use.

Purpose of the Study:

• To systematically review and describe methods used for linking maternal and infant records in healthcare databases.
• To inform the fitness for use of different linkage strategies in pharmacoepidemiology.

Main Methods:

• Conducted multiple Medline searches and reviewed reference sections of key papers.
• Abstracted data on data sources, linkage fields, methodology, and evaluation results from 114 publications (1990-2025).

Main Results:

• Linkage methods varied globally, with administrative claims data yielding 56%-90% success.
• Incorporating vital records (e.g., birth certificates) or birth records with hospital data increased linkage proportions to over 86% and 90%, respectively.
• Probabilistic methods generally achieved higher linkage rates than deterministic methods, though validation was infrequent.

Conclusions:

• Successful mother-infant data linkage is feasible worldwide, supported by robust healthcare infrastructure and policy.
• Researchers should report linkage methods, success rates, and validity assessments to enhance trust in pharmacoepidemiology findings.