Challenges for decision-makers when assessing within-class comparative effectiveness: the case of anticoagulation therapy for atrial fibrillation

In September 2020, the UK National Institute of HealthCare Excellence (NICE) published a draft report for consultation on anticoagulation therapy for stroke prevention in people with atrial fibrillation (AF) [1]. This intervention evidence review is part of the process of updating the NICE clinical guideline on management of AF [2] and addressed the question of which nonvitamin K oral anticoagulation (NOAC) therapy is most clinically and cost-effective for stroke prevention in people with AF. Four NOACs were included in the evidence review, each with a low and high dose formulation: apixaban, dabigatran, rivaroxaban and edoxaban, and all of which had been previously evaluated and recommended in NICE single technology appraisals [3–6]. On the basis of a systematic literature review to identify relevant evidence, network meta-analysis (NMA) to address comparative effectiveness, and subsequent cost–effectiveness analysis (CEA), the committee concluded that apixaban and dabigatran had the most favorable cost–effectiveness (CE) results of the four NOACs, at the NHS list prices and this led to a draft recommendation in the report that apixaban and dabigatran should be used as first-line options [1]. A further recommendation was made that patients who are stable on one of the other anticoagulants (rivaroxaban, edoxaban or warfarin) should discuss switching with their physician [1]. This draft decision by the committee, to recommend two NOACs within a class of four, represents an unusually strong recommendation to differentiate drugs within a particular class. This is especially so, given that the committee themselves state that the acquisition costs based on NHS list prices of the four NOACs are similar (page 74, line 28) [1]. This suggests that the opinion reached by the committee was grounded in the results of the NMA produced comparative effectiveness estimates that then formed the basis of the CEA. This manuscript reports on the process of examining the committee’s draft decision to differentiate drugs within the NOAC class on the basis of clinical and CE in light of the evidence considered by the committee. This manuscript was sponsored by the manufacturer of one of the NOACs that was not recommended as a first-line option by the committee (edoxaban, Daiichi Sankyo) in order to help them respond to the consultation process NICE initiated and which closed in November 2020. The aim of this manuscript is to elucidate the methodological challenges involved in any decision made by a reimbursement authority, such as NICE in the UK, and ask an open question regarding the strength of evidence of comparative effectiveness be required in order to differentiate similarly-priced drugs within a single drug class. The manuscript is structured around three further sections. The next section looks at the evidence on comparative effectiveness of seven key clinical outcomes, the need for NMA to generate those estimates of comparative effectiveness and the representation of uncertainty in the results of NMAs. The third section looks specifically at how the results of the NMA are then turned into a CEA that uses quality-adjusted life-years (QALYs) as the metric to synthesize the seven clinical outcomes into a QALY comparison, before considering the inclusion of costs and the representation of uncertainty in the CE results. A final section then discusses the challenges of making recommendations in the context of the uncertainty in clinical and CE both between the NOAC and vitamin-K antagonist (VKA) drug classes and within the NOAC class.