Affiliations: CHRU Dijon, Service de Biostatistique et
d'Informatique Médicale (DIM), Dijon, France | Inserm, U866, Univ de Bourgogne, Dijon, France | Health Claim Medical Chair ESC Dijon, Dijon,
France | Centre d'Etudes Périnatales de l'Océan
Indien, CHU de la Réunion, Saint Pierre, La Réunion, France | Institut National d'Etudes Démographiques,
Paris, France
Note: [] Corresponding author: Catherine Quantin, CHRU Dijon, Service de
Biostatistique et d'Informatique Médicale (DIM), Dijon, F-21000, France.
E-mail: [email protected]
Abstract: When we need to match different files, we generally face the problem
of preserving confidentiality and respecting private life, especially if the
files come from various sources. This problem may, on some occasions,
dramatically restrict the utilization of administrative data for social
statistics and research. In order to overcome this problem, some organizations,
statistical institutes and research centers use the Secured Matching
technique. Secured Matching is performed by using anonymous matching keys. One
way to proceed is by using a technique for the irreversible encryption of the
identifiers. The identifying elements common to several files are transformed
by the same encryption process, allowing the matching to be done in an
anonymous way by using the encrypted identifiers, with no possible link to the
original identifiers. As an example, we will talk about the use of Secured Matching in
France and describe some applications in epidemiology. We also show how these
techniques of secured matching can be considered as a potential for innovation
within official French statistics.
Keywords: Record linkage, hash-coding, security, non-reversible encryption, epidemiological survey
