Related Experiment Videos
A comparison of record linkage yield for health research using different variable sets
Michael S Simon1, Beth A Mueller, Dennis Deapen
1Barbara Ann Karmanos Cancer Institute at Wayne State University School, 4th Floor Hudson Webber Cancer Research Center, 4100 John R, Detroit, MI 48201, USA. simonm@karmanos.org
Breast Cancer Research and Treatment
|February 5, 2005
Summary
Linking cancer registries with birth records using only the mother's Social Security Number (SSN) is highly accurate. This method is valuable for childbearing and survival studies, despite minor discrepancies in a small number of cases.
Area of Science:
- Epidemiology
- Public Health
- Cancer Research
Background:
- Linking cancer registry data with vital records is crucial for childbearing and survival studies.
- Maternal Social Security Number (SSN) was the primary identifier for linkage in Michigan before 1989.
- Additional identifiers became available post-1989, enabling more robust data linkage.
Purpose of the Study:
- To evaluate the accuracy and effectiveness of using maternal Social Security Number (SSN) alone for linking invasive breast cancer patient records with birth certificate data.
- To compare linkage quality between SSN-only matching and a comprehensive multi-variable approach.
Main Methods:
- A deterministic linkage approach was employed, utilizing seven variables across 14 distinct steps.
- Linkage procedures were conducted using data from 1989-1994 to compare SSN-based matches against those derived from multiple identifiers.
- Manual review was performed to verify the accuracy of identified matches.
Main Results:
- Out of 4496 patients, 793 live births were successfully linked.
- Using Social Security Number (SSN) alone identified 98% (780 out of 793) of the confirmed matches.
- A small proportion (5 out of 7) of SSN-identified matches were not validated by manual review, indicating a minor rate of false positives.
Conclusions:
- Social Security Number (SSN) is a highly accurate and valuable identifier for linking cancer registry data with other vital statistics, such as birth records.
- The SSN-based linkage method demonstrates significant utility for epidemiological studies on childbearing and survival among cancer patients.
- While SSN provides a strong basis for linkage, incorporating additional variables can further enhance data accuracy and reduce potential false positives.