Comparative effectiveness research with administrative health data in rheumatoid arthritis

Comparative effectiveness research (CER) enables the comparison of different treatments in the real-world setting to inform clinical decision-making. Rheumatoid arthritis (RA) has been identified as a high priority area for CER. Administrative health databases, which generally consist of physician billing claims, hospital discharge summaries and prescription drug records, have been widely used to conduct observational research in RA. These data are accurate and complete records of health-care use unaffected by recall bias, and provide large, general population samples and information on long-term follow-up. However, administrative health data pose unique methodological challenges for CER in the field of RA. Here, we discuss the challenges of studying treatment effectiveness with CER (as distinct from harms and costs), in particular, issues relating to the identification and definition of RA cases, timing of disease onset and determination of disease severity. We also discuss an algorithm developed to measure effectiveness outcomes of RA treatments in administrative data, and potential sources of bias that might affect the validity of results. Finally, we explore opportunities for use of administrative data in CER, such as comparisons between reference drugs and biosimilars.