Limitations

The study is a retrospective research, and it has all the limitation that this kind of research should have. For example, missing data, coding bias, loss follow up patients, different inclusion status of patients and different treatment result in the end are all inevitable limitations. On the other hand, it is based on a Taiwan National Health Insurance Research database, which is reflecting only current medical and economic situation under particular situation and in particular time interval. This is a small piece of real-world evidence demonstrated to the world that the early treatment with biologics could cut oral medication in half in just two years. Nevertheless, no other objective evidence could be provided to demonstrate the efficacy of the biologics which is also another limitation in this study. This also affects the statistical analyses of the results. Furthermore, the reimbursement of biologics other than the Etanercept and the Adalimumab as first-line treatment of RA was not available in Rituximab and Tocilizumab, and the Golimumab was not available in Taiwan until the end of 2012. All of which could limit the case numbers treated by the biologics other than the Etanercept and the Adalimumab.
Besides, disease activity, genes and smoking may be involved in RA long term treatment efficacy, which all these three factors cannot be direct evaluated in this study. For example, we only calculated the decrease in treatment in 50% of the days, but there are no cumulative doses in each category of medication. The situation is similar between two groups, which we consider these issues contribute equally to each subgroup and may not affect the final comparison result. Also, by decreasing the use of oral treatment (NSAIDs and glucocorticoid), it could only mean a symptomatic effect and not necessarily have an effect on the activity or accumulated damage of the disease (it is a bias not to have activity measures such as DAS28 or radiographic damage on this national database analysis). Not having measurements of poor prognosis factors such as serology, persistent activity, smoking, extra-articular manifestations and adherence to treatment limit the results in this large nation-wide study.
It is therefore possible that a minor portion of the included patients with RA were misdiagnosed from other types of arthritis, such as seronegative arthritis. However, we have done all the effort to minimize this entire situation by confirm the diagnosis with treatment medication. Unfortunately, the data in the medical records did not include enough information to assess the RA patient functional class and is why we omitted this parameter in the statistical analyses.