Related Experiment Videos
[Record linkage with cryptographic identification data in a population-based cancer registry. Development,
V Krieg1, H W Hense, M Lehnert
1Epidemiologisches Krebsregister für den Regierungsbezirk Münster. krebsregister@uni-muenster.de
Gesundheitswesen (Bundesverband Der Arzte Des Offentlichen Gesundheitsdienstes (Germany))
|July 27, 2001
Summary
Cryptographic methods offer a feasible approach for cancer registry record linkage, achieving low error rates for case notification. This technique aids data anonymization, though further research is limited.
Area of Science:
- Health Informatics
- Cancer Epidemiology
- Data Security
Background:
- Population-based cancer registries require complete case notification for accuracy.
- German legal regulations mandate written informed consent, hindering data reporting by health professionals.
- Anonymized data linkage using cryptographic methods presents a potential solution.
Purpose of the Study:
- To evaluate the feasibility of cryptographic record linkage in the Münster Cancer Registry (EKR-MS).
- To compare cryptographic methods against traditional personal identifier-based linkage.
Main Methods:
- Generated unique control numbers using two distinct chiffrations for 19 name-related characteristics.
- Employed the semi-automatic computer program AUTOMATCH for record linkage based on control numbers.
- Utilized plausibility rules to minimize manual linkage decisions.
Main Results:
- Compared cryptographic linkage against traditional methods for 27,262 new reports against 101,880 known cases.
- Cryptographic linkage yielded a synonymous error rate of nearly 2% and a homonymous error rate below 0.5%.
Conclusions:
- Cryptographic methods are potentially feasible for cancer registry record linkage.
- Error rates suggest slight, potentially acceptable overestimation of disease rates.
- Data anonymization limits further epidemiologic research, restricting utility to registration under legal constraints.