Data linkage between the French multiple sclerosis cohort (OFSEP) and the French national health insurance database (SNDS).

Abstract

Background: Linking disease registries to nationwide healthcare administrative databases increases the research opportunities. Recent guidelines emphasize the need of transparency in this process.

Objective: Our aims were to describe the process of record linkage between the French multiple sclerosis (MS) cohort (OFSEP) and the national health insurance database (SNDS) and to evaluate the linkage quality.

Methods: As no unique identifier was available in the two databases, the OFSEP-SNDS data linkage was performed by indirect matching using the following sixteen patient variables to create a unique key: sex, dates of birth and death, of visits to a neurologist, of MS-related hospitalizations, of MRI, and use of disease-modifying therapies. Three indicators were computed to assess the linkage quality.

Results: Among the 52,034 eligible patients in the OFSEP registry, 42,603 (81.9%) were matched with patients in the SNDS database, with good overall quality (robustness=3.19; this is the number of linkage variables that can be removed without losing the uniqueness of the linked pair; 87.8% of common information). Comparison of the linked and unlinked populations revealed no major selection bias regarding age and sex distributions.

Conclusion: The successful linkage of more than 40,000 patients with MS broadens the research perspectives by allowing access to a wide range of clinical and administrative data (e.g., comorbidities, care pathways) over a long mean disease duration (> 15 years).