In the interpretation of clinical trials or meta-analyses that show
no significant difference between the two comparators, one controversial
issue is the need to differentiate between "no proof of difference"
(inconclusive result) and "proof of no difference" (or demonstrated non-
inferiority). For this purpose, trial-sequential analysis (TSA) is
considered to be an appropriate statistical tool (1-4).
In the meta-analysis by Providencia et al. comparing dabigatran vs
warfarin in patients with atrial fibrillation undergoing catheter ablation
(5), one limitation that the authors themselves have pointed out is that
"some comparisons [were] limited by the low event rates". As shown in
Figure 2 of Providencia's paper, the overall event rate in the two
patient groups was 0.3% (15/4782) for thromboembolic complications and
1.4% (67/4782) for major bleedings.
In the light of these findings, the conclusion that "the rate of
thrombolembolic complications and/or major bleeding in patients on
dabigatran.... is similar to that seen with warfarin" is not justified by
any statistical proof. In fact, if a TSA is applied to determine the
optimal information size required to reach a conclusive result from these
data, the analysis clearly shows that the number of patients actually
enrolled in the available trials (N=4,782) is much lower than the optimal
number required for evaluating thromboembolic complications (N=18,767) or
major bleedings (N=17,461); see Figure 1 at
www.osservatorioinnovazione.net/supplement/dabigatran-ablation.pdf .
Hence, the best interpretation of these results is that no conclusion
can be made on these two end-points.
References
1. Messori A, Fadda V, Maratea D, Trippoli S. Erythropoietin in
patients with acute myocardial infarction: no proof of effectiveness or
proof of no effectiveness? Clin Cardiol. 2013 Aug 8. doi:
10.1002/clc.22187. [Epub ahead of print] PubMed PMID: 23929820.
2. Messori A, Fadda V, Maratea D, Trippoli S. Intra-aortic balloon
pump in high-risk percutaneous coronary interventions without cardiogenic
shock: Trial sequential analysis of outcomes. Int J Cardiol. 2013 Jul 23.
doi:pii:S0167-5273(13)01155-8. 10.1016/j.ijcard.2013.06.098. [Epub ahead
of print] PubMed PMID: 23890904.
3. Messori A, Fadda V, Maratea D, Trippoli S. Erythropoiesis-
stimulating agents in heart failure: no proof of effectiveness or proof of
no effectiveness? Eur J Heart Fail. 2013 Aug;15(8):944-5.
4. Messori A, Fadda V, Maratea D, Trippoli S. ?-3 Fatty acid
supplements for secondary prevention of cardiovascular disease: from "no
proof of effectiveness" to "proof of no effectiveness". JAMA Intern Med.
2013 Jun 17:1-2. doi:10.1001/jamainternmed.2013.6638.
5. Providencia R, Albenque JP, Combes S, Bouzeman A, Casteigt B,
Combes N, et al. Safety and efficacy of dabigatran versus warfarin in
patients undergoing catheter ablation of atrial fibrillation: a systematic
review and meta-analysis. Heart Published Online First: 2013 Jul 22
doi:10.1136/ heartjnl-2013-304386
Conflict of Interest:
None declared
In the interpretation of clinical trials or meta-analyses that show no significant difference between the two comparators, one controversial issue is the need to differentiate between "no proof of difference" (inconclusive result) and "proof of no difference" (or demonstrated non- inferiority). For this purpose, trial-sequential analysis (TSA) is considered to be an appropriate statistical tool (1-4).
In the meta-analysis by Providencia et al. comparing dabigatran vs warfarin in patients with atrial fibrillation undergoing catheter ablation (5), one limitation that the authors themselves have pointed out is that "some comparisons [were] limited by the low event rates". As shown in Figure 2 of Providencia's paper, the overall event rate in the two patient groups was 0.3% (15/4782) for thromboembolic complications and 1.4% (67/4782) for major bleedings.
In the light of these findings, the conclusion that "the rate of thrombolembolic complications and/or major bleeding in patients on dabigatran.... is similar to that seen with warfarin" is not justified by any statistical proof. In fact, if a TSA is applied to determine the optimal information size required to reach a conclusive result from these data, the analysis clearly shows that the number of patients actually enrolled in the available trials (N=4,782) is much lower than the optimal number required for evaluating thromboembolic complications (N=18,767) or major bleedings (N=17,461); see Figure 1 at www.osservatorioinnovazione.net/supplement/dabigatran-ablation.pdf .
Hence, the best interpretation of these results is that no conclusion can be made on these two end-points.
References
1. Messori A, Fadda V, Maratea D, Trippoli S. Erythropoietin in patients with acute myocardial infarction: no proof of effectiveness or proof of no effectiveness? Clin Cardiol. 2013 Aug 8. doi: 10.1002/clc.22187. [Epub ahead of print] PubMed PMID: 23929820.
2. Messori A, Fadda V, Maratea D, Trippoli S. Intra-aortic balloon pump in high-risk percutaneous coronary interventions without cardiogenic shock: Trial sequential analysis of outcomes. Int J Cardiol. 2013 Jul 23. doi:pii:S0167-5273(13)01155-8. 10.1016/j.ijcard.2013.06.098. [Epub ahead of print] PubMed PMID: 23890904.
3. Messori A, Fadda V, Maratea D, Trippoli S. Erythropoiesis- stimulating agents in heart failure: no proof of effectiveness or proof of no effectiveness? Eur J Heart Fail. 2013 Aug;15(8):944-5.
4. Messori A, Fadda V, Maratea D, Trippoli S. ?-3 Fatty acid supplements for secondary prevention of cardiovascular disease: from "no proof of effectiveness" to "proof of no effectiveness". JAMA Intern Med. 2013 Jun 17:1-2. doi:10.1001/jamainternmed.2013.6638.
5. Providencia R, Albenque JP, Combes S, Bouzeman A, Casteigt B, Combes N, et al. Safety and efficacy of dabigatran versus warfarin in patients undergoing catheter ablation of atrial fibrillation: a systematic review and meta-analysis. Heart Published Online First: 2013 Jul 22 doi:10.1136/ heartjnl-2013-304386
Conflict of Interest:
None declared