French health insurance databases are organized since 2003 into a huge digital data warehouse, the Système national d'information inter-régime de l'assurance maladie (SNIIR-AM). It covers the entire French population (65 million inhabitants). In order to facilitate studies on more frequent conditions, a random sample of 1/97th of national health system beneficiaries has been built since 2005, called the échantillon généraliste des bénéficiaires (EGB). The aim of this article is to describe the main characteristics of the SNIIR-AM and the EGB, to detail their accessibility according to French law, and to present their strengths and limits. It is illustrated with the most recent studies conducted in these databases. These databases include demographic, out-hospital reimbursement (including drug dispensing), medical (costly long-term diseases, occupational diseases, sick-leaves…), and in-hospital data. All these data are prospectively recorded, individualized, made anonymous and linkable. Consequently, the SNIIR-AM is a very useful data source for epidemiological, pharmacoepidemiological and health economics studies, particularly for rare diseases. The EGB is appropriate for long-term research on more frequent diseases.

Background Various observations have suggested that the course of COVID-19 might be less favourable in patients with inflammatory rheumatic and musculoskeletal diseases receiving rituximab compared with those not receiving rituximab. We aimed to investigate whether treatment with rituximab is associated with severe COVID-19 outcomes in patients with inflammatory rheumatic and musculoskeletal diseases.Methods In this cohort study, we analysed data from the French RMD COVID-19 cohort, which included patients aged 18 years or older with inflammatory rheumatic and musculoskeletal diseases and highly suspected or confirmed COVID-19. The primary endpoint was the severity of COVID-19 in patients treated with rituximab (rituximab group) compared with patients who did not receive rituximab (no rituximab group). Severe disease was defined as that requiring admission to an intensive care unit or leading to death. Secondary objectives were to analyse deaths and duration of hospital stay. The inverse probability of treatment weighting propensity score method was used to adjust for potential confounding factors (age, sex, arterial hypertension, diabetes, smoking status, body-mass index, interstitial lung disease, cardiovascular diseases, cancer, corticosteroid use, chronic renal failure, and the underlying disease [rheumatoid arthritis vs others]). Odds ratios and hazard ratios and their 95% CIs were calculated as effect size, by dividing the two population mean differences by their SD. This study is registered with ClinicalTrials.gov, NCT04353609.