Objective: This systematic literature review (SLR) had two objectives: to analyse published economic evaluations of biological disease-modifying anti-rheumatic drugs (bDMARDs) for patients with moderate to severe rheumatoid arthritis (RA) previously treated with DMARDs and to assess the quality of those that included sequences of treatments.
Methods: We performed an SLR on PubMed, Central, Cochrane, and French databases from January 2000 to December 2018. The search focused on cost-effectiveness/utility/benefit analyses. We extracted data on treatment sequences, outcomes (e.g. quality-adjusted life year) and choices of economic evaluation methods (e.g. model type, type of analysis, and method of utility estimation). We analysed the improvement of methods by comparing two sub-periods (2000-2009 and 2010-2018). The quality of reporting and the quality of the methods were assessed using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) and a set of eight key aspects for a reference case for economic evaluation of bDMARDs based on the Outcome Measures in Rheumatology (OMERACT) and Drummond checklists. Data extraction and study assessment were performed independently by two health economists.
Results: From the 824 records identified in the initial search, 51 publications were selected. Of these, 31 included sequences. Individual models such as discrete-event simulations were used in over two-fifths (22/51, 43%) of the selected studies. Few studies (7/51, 14%) used utility scores based on generic instruments (e.g. EQ-5D). Estimation of hospitalization costs was described in only approximately one-third of studies (19/51). Loss of quality of life (QoL) related to adverse events such as tuberculosis and pneumonia was included in one-tenth (5/51, 10%) of the studies. It was difficult to compare the results of the economic evaluations (i.e. incremental cost-effectiveness ratios) due to the high heterogeneity of studies in terms of disease stage, data sources, inputs, and methods of health outcome assessment used. For identified studies including sequences, the CHEERS assessment of reporting quality showed insufficient reporting of uncertainty analyses and utility weights in more than a third of the studies (11/31, 35%; 9/25, 36%). An in-depth assessment of the quality of the studies revealed that only seven, mostly conducted during the sub-period 2010-2018, addressed the majority of methodological quality assessment issues such as the simulation of patient sequence pathways, the use of systematic reviews and meta-analyses of comparative effectiveness, the choice of treatment sequence, and rules for switching.
Conclusion: Our SLR identified a lack of high-quality evaluations assessing bDMARD sequences, although some improvements were made in the reporting and modelling of patients' pathways in studies published after 2010. In order to improve economic evaluations of RA, clear health technology assessment guidance on RA health-related QoL instruments must be provided, and data including long-term disease progression must be made available.