114 CHAPTER 7 Primary outcome After having observed 163 primary outcome events (101 (15.3%) in the NOAC arm and 62 (9.4%) in the VKA arm), this superiority trial, with the hypothesis that switching to NOAC treatment would lead to fewer major and CRNM bleeding, was halted for futility following the advice of the DSMB and in accordance with our prespecified protocol. It was decided to stop inclusion and complete follow-up for all participants in the study. After complete follow-up, the HR for our primary outcome was 1.69 for switching to a NOAC relative to continuing INR guided VKA treatment (95% CI 1.23–2.32; P=0.00112; see Figure 2, see Table 2). The location of bleeding sites differed per treatment arm (see Table 3). Numerically, more gastrointestinal and urogenital bleedings were observed in the intervention arm compared to the control arm: 17 (2.6%) versus 4 (0.6%) gastrointestinal bleedings and 20 (3.0%) versus 11 (1.7%) urogenital bleedings, respectively. Haemorrhagic stroke was seen in 7 patients (1.1%) who switched to a NOAC versus in 6 patients (0.9%) who continued with a VKA. Visual inspection of the cumulative incidence curve revealed the potential of non-proportionality related to the switch period, namely from day 1 to day 100, with lines only diverging after day 100 (in fact, this is the time point when all patients were switched from a VKA to a NOAC in our intervention arm). Following the statistical analysis plan, in such circumstances, a step function using a time-period interaction term should be introduced in the Cox model. This sensitivity analysis showed a HR of 1.17 (95% CI 0.70–1.96) for the first 100 days and a HR of 2.10 (95% CI 1.40–3.16) for days 100 to 365 (see Supplementary file S2). Subgroup analyses yielded no apparent differences in subgroups on the basis of age, sex, GFI score, or renal function (see Figure 3). Some differences were observed in relation to the prescribed NOAC. The HR for our primary outcome was similar for the two most prescribed NOACs in our trial (rivaroxaban (HR 1.95 (95% CI 1.36–2.79) and apixaban (HR 2.17 (95% CI 1.28–3.68)), yet appeared to be lower for edoxaban (HR 1.10 (95% CI 0.57–2.13)). Nevertheless, these analyses should be interpreted with caution because they were post-hoc and non-randomised.
RkJQdWJsaXNoZXIy MTk4NDMw