National reimbursement databases: Use and limitations for rheumatologic studies

The aim was to review and analyze the current knowledge on the use and limitations of National Reimbursement Databases (NRD) in rheumatology. Three main categories of NRDs were identified, according to the data available in these databases: NRDs without details from clinical practice (such as the French and Quebec NRDs), NRDs with diagnosis details from clinical practice (such as the British NRD), and NRDs linkable to clinical databases (such as the Swedish NRD). NRDs allow the construction of cohorts with prospective data collection and with important statistical power, and therefore enable better knowledge of the rheumatic diseases' epidemiology, notably about the risk-benefit balance of available therapies as well as about the economic burden of these diseases and treatments to Society. This may have an important impact on public health decision-making as well as on development or adjustment of management guidelines. The main limitation of NRDs is the lack of exhaustive medical information, e.g., measure of disease activity, work-up results, ascertainment of diagnostic codes, etc. Other issues are related to the size and complexity of these databases and the difficulty to get access to data extractions. In conclusion, NRDs represent an important opportunity for medical research in rheumatology and the need to create scientific interactions between rheumatologists, data managers and biostatisticians. A major issue remains the lack of exhaustive clinical and paraclinical data in those databases. The latter should be addressed in the years to come by the linkage with clinical registers.