In this study, the NNTs of bDMARDs in patients with RA were 5- to 8-fold lower than those of MTX. However, the mean total actual health care costs were much higher in the bDMARDs group than in the MTX group. Mean costs per NNT based on both DAS28-ESR and CDAI were higher in the bDMARDs group than in the MTX group. This means that although the effectiveness of bDMARDs was much better than that of MTX, the cost-effectiveness of MTX was superior to that of bDMARDs from the standpoint of NNT and total actual health care cost.
Some economic measures can be used to assess the cost-effectiveness of drugs, such as the incremental cost effectiveness ratio, quality-adjusted life-years, and NNT. As mentioned earlier, NNT is not itself a true economic indicator of cost-effectiveness, but an indicator to assess the effectiveness of therapeutic drugs. However, because NNT is easy to understand and calculate, it is used to assess the cost-related effectiveness of drugs. Several studies have examined the cost-effectiveness of bDMARDs in RA using NNT. Batticciotto et al.  investigated NNT of monotherapy with ADA, ETN, certolizumab pegol, or TCZ in patients intolerant to MTX. They reported that TCZ had the lowest NNT and the lowest costs per NNT of all other agents, which was similar to the results of the present study. Maurizio et al.  compared NNT between TCZ plus MTX and ABT plus MTX in patients previously treated with MTX and concluded that TCZ was superior to ABT in terms of NNT. To our knowledge, however, no previous studies have compared bDMARDs with MTX in terms of NNT or cost per NNT.
Although simple comparison is difficult because each country has own health insurance system, several studies have reported the cost-effectiveness of bDMARDs in the treatment of RA. Hyon et al.  compared five monotherapies for the patients with MTX-naïve; ETN, MTX, leflunomide, SASP, and no second line agents. They reported that the costs of combination therapy of ETN plus MTX were 1.5-fold higher than those of MTX monotherapy, and MTX is cost effective choice for MTX-naïve RA patients, which is similar as our findings. Joseph et al.  reported that the 5-year per-patient drug costs were 4- to 8-fold higher in those switched to bDMARDs versus those continued MTX. They also reported that total costs per patient were 3 to 4 times higher when adding or switching to bDMARDs than when continuing MTX. The difference of total costs between bDMARDs with MTX was smaller than our study. MTX group in this study included both oral and subcutaneous MTX, which may lead to relatively higher costs in MTX group. Because both studies calculated the costs based on simulations, they may have underestimated the actual costs. In our study, the mean costs were more than 10-fold higher in the bDMARDs group than in the MTX group. We investigated the total actual costs based on the real-world clinical setting under the Japanese health insurance system, not based on simulations, which might be why the mean costs of bDMARDs were much higher than those of MTX.
Syngle et al.  noted that the mean direct medical cost varied due to several factors such as disease activity, variable number of patients, differences in medication, differences in methods of cost assessment, and differences in health care delivery systems worldwide. In the present study, although the costs of bDMARDs were more than 10-fold higher than those of MTX, the mean costs per NNT of TCZ and ETN were equivalent to that of MTX. Based on our results, TCZ and ETN are cost-effective drugs close to MTX.
The adjusted NNT of bDMARDs was superior to that of MTX based on both DAS28-ESR and CDAI, which means that the effectiveness of bDMARDs was superior to that of MTX when patients’ backgrounds were adjusted to exclude bias. On the other hand, the costs per adjusted NNT in the bDMARDs group were 8-fold higher than those of the MTX group based on DAS28-ESR and 5-fold higher based on CDAI. Given that the total health care cost of the bDMARDs group was more than 10-fold higher than that of the MTX group, the difference in the mean costs per adjusted NNT between the bDMARDs and MTX groups was smaller than in the mean costs per non-adjusted NNT. However, the mean cost per adjusted NNT of MTX were still superior to those of bDMARDs.
The results of this study indicate that although the effectiveness of bDMARDs was better than that of MTX, the cost-effectiveness of MTX was superior in terms of the adjusted NNT and the total costs under Japanese health insurance. Therefore, considering the costs of therapeutic drugs, we recommend MTX for first-line appropriate use in patients with RA who have no complication such as chronic kidney disease, interstitial pneumonia, or liver dysfunction that prevent the use of MTX. If the treatment target is not achieved with MTX, bDMARDs should be added immediately from the economic viewpoint, similar to the EULAR recommendations on the management of RA .
There are some limitations to this study. First, the placebo group should be set as controls in this type of study, however, it was difficult to assign some patients to truly placebo group because this study based on real-world clinical setting, therefore the patients who administered csDMARDs without bDMARDs or MTX was set as the control group in this study. Second, due to the small number of patients involved in this single center study, bias might exist in terms of medication selection. Particularly in the bDMARDs group, two or three specific agents were frequently used as the first bDMARDs. Especially, GLM is used for relatively higher aged patients in our hospital, in addition, 11 out of 16 patients administered with 100mg of GLM, so the mean total health care costs in the patients who administered GLM was much higher than the other patients. Third, there were significant differences in background characteristics between the bDMARDs and MTX groups, such as RA duration, Steinbrocker stage and class, and rate of SASP and BUC administration. However, the NNTs after adjusting for bias from patients’ background factors were lower in the bDMARDs group than in the MTX group. This suggests that the effectiveness of bDMARDs is clearly superior to that of MTX from the standpoint of NNT. There may be the patients who had complications that influenced to the selection of therapeutic drugs, which might lead to higher costs in the bDMARDs group. The rheumatologists should keep in mind not to miss the opportunity of use of bDMARDs for the patients who have higher disease activity or have a complication that prevent the use of MTX. Fourth, because the initial purpose of this study was to compare remission rates at 24 weeks between the bDMARDs and MTX groups, we excluded patients who switched medications within 24 weeks due to an adverse drug reaction or who did not need to change the former medication, which might have led to a higher remission rate than in previous studies. Finally, this study evaluated only the disease activities using DAS28-ESR and CDAI 24 weeks after the introduction of therapeutic drugs. Further investigation is required including the spacing or tapering of bDMARD, or the evaluation of joint destruction to compare between the MTX and bDMARDs groups.
Despite these limitations, we believe that this study is unique and clinically relevant. To our knowledge, no earlier studies have reported on the cost-effectiveness of bDMARDs versus MTX in patients with RA from the standpoint of NNT and total health care costs in the real-world clinical setting.
In conclusion, the NNT of bDMARDs in patients with RA, which is used as an indicator of the effectiveness of therapeutic drugs, was superior to that of MTX. However, the mean total actual health care costs were much higher in the bDMARDs group than in the MTX group. In addition, the mean costs per adjusted NNT based on both DAS28-ESR and CDAI were higher in the bDMARDs group than in the MTX group, suggesting that the cost-effectiveness of MTX is superior to that of bDMARDs from the standpoint of NNT and total actual health care costs under Japanese health insurance system.