Trial sequential analysis of efficacy and safety of direct oral anticoagulants and vitamin K antagonists against left ventricular thrombus

Figures

• 

  Figure 1. Schema of four examples of trial sequential analysis (TSA). (A) False positive: the cumulative Z-curve crosses the traditional boundary (Z = ± 1.96, p = 0.05), but does not cross the superiority boundary. (B) True positive: the cumulative Z-curve crosses the superiority boundary before the required sample size is reached. (C) False negative: the cumulative Z-curve crosses neither the traditional boundary nor the futility boundary. (D) True negative: the required sample size is not reached but the cumulative Z-curve crosses the futility boundary.

• 

  Figure 2. Forest plot of direct comparison in six outcomes for direct oral anticoagulants (DOACs) and vitamin K antagonists (VKAs). (A) Left ventricular thrombus resolution, (B) stroke, (C) any thromboembolism, (D) major bleeding, (E) any bleeding, (F) all-cause death. CI confidence interval, DOACs direct oral anticoagulants, OR odds ratio, VKAs vitamin K antagonists.

• 

  Figure 3. TSA of DOACs and VKAs in left ventricular (LV) thrombus resolution using power of 80%, 5% significance, to detect a 6% absolute difference (DOACs: 74%, VKAs: 68%). The required sample size was 1796 (vertical red line). The cumulative Z-curve (blue line with small black squares representing each trial) crossed both the traditional boundary (horizontal red line) and the superiority boundary (trial sequential monitoring boundary) (concave red line), indicating evidence is sufficient (true positive).

• 

  Figure 4. TSA of DOACs and VKAs in stroke using power of 80%, 5% significance, to detect a 3% absolute difference (DOACs: 7%, VKAs: 10%). The required sample size was 2714 (vertical red line). The cumulative Z-curve (blue line with small black squares representing each trial) crossed both the traditional boundary (horizontal red line) and the superiority boundary (trial sequential monitoring boundary) (concave red line), indicating evidence is sufficient (true positive).

• 

  Figure 5. TSA of DOACs and VKAs in any thromboembolism using power of 80%, 5% significance, to detect a 5% absolute difference (DOACs: 16%, VKAs: 21%). The required sample size was 4955 (vertical red line). The cumulative Z-curve (blue line with small black squares representing each trial) crossed the futility boundary (convex red line), indicating evidence is sufficient (true negative).

• 

  Figure 6. TSA of DOACs and VKAs in major bleeding using power of 80%, and 5% significance, to detect a 2% absolute difference (DOACs:2%, VKAs: 4%). The required sample size was 3749 (vertical red line). The cumulative Z-curve (blue line with small black squares representing each trial) did not cross the traditional boundary (horizontal red line), the superiority boundary (trial sequential monitoring boundary) (concave red line), or the futility boundary (convex red line), indicating a lack of evidence (false negative).

• 

  Figure 7. TSA of DOACs and VKAs in any bleeding using power of 80% and 5% significance, to detect a 2% absolute difference (DOACs: 8%, VKAs: 10%). The required sample size was 6429 (vertical red line). The cumulative Z-curve (blue line with small black squares representing each trial) crossed the traditional boundary (horizontal red line), but did not cross the superiority boundary (trial sequential monitoring boundary) (concave red line), indicating a lack of evidence (false positive).

• 

  Figure 8. TSA of DOACs and VKAs in all-cause death using power of 80% and 5% significance, to detect a 4% absolute difference (DOACs:6%, VKAs: 10%). The required sample size was 2124 (vertical red line). The cumulative Z-curve (blue line with small black squares representing each trial) crossed both the traditional boundary (horizontal red line) and the superiority boundary (trial sequential monitoring boundary) (concave red lines), indicating evidence is sufficient (true positive).