In this study, we evaluated through a simulation model the expected cost-utility of three anticoagulation options namely SD-W, GT-W and dabigatran 150 mg twice daily, over a 5 years’ time horizon and under a public health care perspective. Our results show that the dabigatran option is the most C/U option if the public health care system accepts to invest CAD $4 800 per additional QALY gained. This amount is lower than the 50 000 per QALY gained commonly proposed as the threshold to adopt an innovation in North-America. Although our study has used a more complete model than several other C/E analyses, its results are consistent with the literature[17–20, 22]. Nevertheless, we don’t know if the results could be generalized to all other NOACs as there is still a lack of data concerning the direct comparison between the different products on the market. Currently, only indirect comparisons exist on the efficacy and safety of the NOACs that show that they are effective compared to warfarin. Yet, when they are compared to each other, there is no difference in efficacy, although some differences in safety might exist[62, 63].
The GT-W option was not cost/effective (if not dominated) compared to the two other options. This is in line with other studies that have compared warfarin treatments with or without genetic testing[23, 25, 26]. However, GT-W could be the most C/U option compared to SD-W and dabigatran150 mg, if the average patient time in the therapeutic range moves from 66.6% to 76.8% i.e. if it is 20% higher than in the SD-W option in the first year. However, this is a very ambitious objective to reach even in clinical trial conditions. This result is close to that of You et al. who showed that the GT-W option could become the most cost-effective option if the patient time in the therapeutic range was > 77%.
This study has some limitations. First, the key input parameters (events probabilities, INR control) comparing dabigatran vs. SD-W or SD-W vs. GT-W were taken from one single randomized controlled clinical trial. This issue could have decreased the ICUR since the effectiveness of dabigatran or GT-W may be overestimated by controlled clinical trials compared to the situation in real life. Nevertheless, we have done extensive sensitivity analyses in order to handle this problem.
The second limitation is the complexity of mapping the reality. Some simplifications and assumptions were inevitable in the modeling approach. For example, we did not consider minor events (bleedings and thromboembolism) that occur with anticoagulation therapy. This issue could have increased the ICURs.
Thirdly, our model did not consider patients’ adherence with medication in order to make a fairly comparison of the dabigatran option with the two others. Indeed, while longtime adherence for warfarin is available, there is still a lack of data on longtime adherence for dabigatran. Taking into account the medication adherence would have led us to make assumptions about the adherence for dabigatran.
Finally, our model is limited by the consideration of only direct costs and one single perspective, i.e. the public healthcare perspective. The addition of the patients’ perspective could increase the ICUR especially in the case of SD-W that might require time and travel expenses for INR control.
Despite these limitations, the results of this study suggest that dabigatran 150 mg twice per day is a C/U alternative to SD-W. Its additional cost per QALY gained is considered as socially acceptable. GT-W is not C/U and should not now be recommended in routine management of warfarin anticoagulotheray in FA patients. However, our results produced in the Quebec/Canadian context (a quasi-exclusive public health care system) remain to be confirmed for other health care jurisdictions especially where the public system is not dominant.