Securizing data linkage in french public statistics

BMC Medical Informatics and Decision Making

Debate

Eric Benzenine¹  Maxence Guesdon²  Catherine Quantin³  Kamel Gadouche⁴ 

[1] CHRU Dijon, Service de Biostatistique et d’Informatique Médicale (DIM), Université de Bourgogne Franche-Comté, Dijon, France;CHRU Dijon, Service de Biostatistique et d’Informatique Médicale (DIM), Université de Bourgogne Franche-Comté, Dijon, France;INRIA, Institut National de Recherche en Informatique et Automatique, Palaiseau, France;CHRU Dijon, Service de Biostatistique et d’Informatique Médicale (DIM), Université de Bourgogne Franche-Comté, Dijon, France;INSERM, CIC 1432, Dijon University Hospital, Clinical Investigation Center, clinical epidemiology/clinical trials unit, Dijon, France;INSERM UMR 1181 “Biostatistics, Biomathematics, Pharmacoepidemiology and Infectious Diseases” (B2PHI), Univ. Bourgogne Franche-Comté, Dijon, France;Centre d’Accès Sécurisé aux Données (CASD), Malakoff, France;

关键词: Data linkage;    Patient data privacy;    Population statistics;   

DOI  :  10.1186/s12911-016-0366-4

 received in 2016-03-26, accepted in 2016-09-19,  发布年份 2016

来源: Springer

【 摘 要 】

Administrative records in France, especially medical and social records, have huge potential for statistical studies. The NIR (a national identifier) is widely used in medico-social administrations, and this would theoretically provide considerable scope for data matching, on condition that the legislation on such matters was respected.The law, however, forbids the processing of non-anonymized medical data, thus making it difficult to carry out studies that require several sources of social and medical data.We would like to benefit from computer techniques introduced since the 70 s to provide safe linkage of anonymized files, to release the current constraints of such procedures.We propose an organization and a data workflow, based on hashing and cyrptographic techniques, to strongly compartmentalize identifying and not-identifying data.The proposed method offers a strong control over who is in possession of which information, using different hashing keys for each linkage. This allows to prevent unauthorized linkage of data, to protect anonymity, by preventing cumulation of not-identifying data which can become identifying data when linked.Our proposal would make it possible to conduct such studies more easily, more regularly and more precisely while preserving a high enough level of anonymity.The main obstacle to setting up such a system, in our opinion, is not technical, but rather organizational in that it is based on the existence of a Key-Management Authority.

【 授权许可】

CC BY   
© The Author(s) 2016