BackgroundRheumatoid arthritis (RA) is initially treated with methotrexate and other disease-modifying antirheumatic drugs (DMARDs). Active RA patients who fail such treatments can receive tumour necrosis factor inhibitors (TNFis), which are effective but expensive.ObjectiveWe assessed whether or not combination DMARDs (cDMARDs) give equivalent clinical benefits at lower costs in RA patients eligible for TNFis.DesignAn open-label, 12-month, pragmatic, randomised, multicentre, two-arm trial [Tumour necrosis factor inhibitors Against Combination Intensive Therapy (TACIT)] compared these treatment strategies. We then systematically reviewed all comparable published trials.SettingThe TACIT trial involved 24 English rheumatology clinics.ParticipantsActive RA patients eligible for TNFis.InterventionsThe TACIT trial compared cDMARDs with TNFis plus methotrexate or another DMARD; 6-month non-responders received (a) TNFis if in the cDMARD group; and (b) a second TNFi if in the TNFi group.Main outcome measuresThe Heath Assessment Questionnaire (HAQ) was the primary outcome measure. The European Quality of Life-5 Dimensions (EQ-5D), joint damage, Disease Activity Score for 28 Joints (DAS28), withdrawals and adverse effects were secondary outcome measures. Economic evaluation linked costs, HAQ changes and quality-adjusted life-years (QALYs).ResultsIn total, 432 patients were screened; 104 started on cDMARDs and 101 started on TNFis. The initial demographic and disease assessments were similar between the groups. In total, 16 patients were lost to follow-up (nine in the cDMARD group, seven in the TNFi group) and 42 discontinued their intervention but were followed up (23 in the cDMARD group and 19 in the TNFi group). Intention-to-treat analysis with multiple imputation methods used for missing data showed greater 12-month HAQ score reductions with initial cDMARDs than with initial TNFis [adjusted linear regression coefficient 0.15, 95% confidence interval (CI) −0.003 to 0.31;p = 0.046]. Increases in 12-month EQ-5D scores were greater with initial cDMARDs (adjusted linear regression coefficient −0.11, 95% CI −0.18 to −0.03;p = 0.009) whereas 6-month changes in HAQ and EQ-5D scores and 6- and 12-month changes in joint damage were similar between the initial cDMARD group and the initial TNFi group. Longitudinal analyses (adjusted general estimating equations) showed that the DAS28 was lower in the initial TNFi group in the first 6 months (coefficient −0.63, 95% CI −0.93 to −0.34;p < 0.001) but there were no differences between the groups in months 6–12. In total, 36 patients in the initial cDMARD group and 44 in the initial TNFi group achieved DAS28 remission. The onset of remission did not differ between groups (p = 0.085 on log-rank test). In total, 10 patients in the initial cDMARD group and 18 in the initial TNFi group experienced serious adverse events; stopping therapy because of toxicity occurred in 10 and six patients respectively. Economic evaluation showed that the cDMARD group had similar or better QALY outcomes than TNFi with significantly lower costs at 6 and 12 months. In the systematic reviews we identified 32 trials (including 20–1049 patients) on early RA and 19 trials (including 40–982 patients) on established RA that compared (1) cDMARDs with DMARD monotherapy; (2) TNFis/methotrexate with methotrexate monotherapy; and (3) cDMARDs with TNFis/methotrexate. They showed that cDMARDs and TNFis had similar efficacies and toxicities.ConclusionsActive RA patients who have failed methotrexate and another DMARD achieve equivalent clinical benefits at a lower cost from starting cDMARDs or from starting TNFis (reserving TNFis for non-responders). Only a minority of patients achieve sustained remission with cDMARDs or TNFis; new strategies are needed to maximise the frequency of remission.Trial registrationCurrent Control Trials ISRCTN37438295.FundingThis project was funded by the National Institute for Health Research Health Technology Assessment programme and will be published in full inHealth Technology Assessment; Vol. 18, No. 66. See the NIHR Journals Library website for further project information.