dabagratan, again

there have been a slew of articles recently promoting dabigatran and the direct acting thrombin inhibitors, and a remarkable number of articles in the on-line throw-away journals (delete-away??), like MedPage, interviewing leading researchers (many on the pharma dole) extolling the virtues of these medications. recent articles include:

1. a meta-analysis noting that dabigatran, rivaroxaban, apixaban and edoxaban are non-inferior to vitamin K antagonists in stroke prevention for patients with atrial fibrillation, assessed the endpoints of all-cause and vascular mortality along with safety issues (see afib direct acting vs warfarin metaanal j thromb haem 2014 in dropbox, or DOI:10.1111/jth.12651). 

--71,683 patients with nonvalvular atrial fibrillation (from 4 RCTs , with followup of 1.8-2.8 years) found significant declines in all-cause mortality by 11% with NNT=132 (p<0.0001) and vascular mortality by 12%, NNT =189 (p<0.0001)​, with a striking decrease in mortality from bleeding (RR 0.54, p<0.0001, esp intracranial bleeding with RR 0.42, p<0.00001)

2. a meta-analysis was done of the efficacy and treatment of acute venous thromboembolism -- VTE (see dvt heparin warf vs new agents JAMA 2014 in dropbox, or doi:10.1001/jama.2014.10538). they assessed 8 strategies, though this was a "network meta-analysis", mathematically combining different studies, since there were no direct comparisons of them. Strategies included: unfractionated heparin (UFH), low-molecular weight heparin (LMWH) in combo with vitamin K antagonists (VKA), LMWH with dabigatran or edoxaban, rivaroxaban, apixaban, and LMWH alone. 

--45 studies with 45K patients, 22 comparing UFH-VKA with LMWH-VKA. only 6 studies looked at the new agents (2 with rivaroxaban, 2 with dabigatran, one each with edoxaban and apixaban), median followup of 3 months.
--compared with LMWH-VKA, UFH-VKA associated with increased risk of recurrent VTE (HR 1.42), with proportion of patients with recurrent VTE over the 3 months: UFH-VKA 1.84%, LMWH-VKA 1.30%. so, no statistically signif diff for efficacy in any of these strategies, except UFH-VKA (which did badly)
--risk of major bleeding event: rivaroxaban with 0.49%, apixaban 0.28%, LMWH-VKA 0.89%, with apixaban being the only one with statistically significant decrease in bleeding

BUT, a study released by JAMA Internal Medicine assessed retrospectively the Medicare pharmacy and medical claims for 1302 people on dabigatran used in atrial fibrillation vs 8102 on warfarin (see afib dabigatran inc bleeding jama int med 2014, or doi:10.1001/jamainternmed.2014.5398). this was a real-world post-marketing study, assessing major and minor bleeding events (major= intracranial hemorrhage, hemoperitoneum, hosp visits/admits for hematuria, GI or other hemorrhage).

--dabigatran associated with higher risk of bleeding than warfarin-- 32.7% vs 26.5%, with HR 1.30 (CI 1.20-1.41) for any bleeding event, major bleeding 9.0% vs 5.9%, with HR 1.58. GI bleeding​ 17.4% vs 10.0%, with HR 1.58 (CI 1.36-1.83),  though risk of intracranial bleeding lower with dabigatran, 0.6% vs 1.8%, with HR 0.32 (CI 0.20-0.50)​
--the risk of major bleeding with dabigatran was especially high in certain subgroups: African Americans had more bleeding with dabigratan with HR 2.12, and patients with chronic kidney disease had an increase with HR 2.07. the increased rate of intracranial bleeds was only higher in warfarin in white patients older than 75 (no diff in those <75yo and African Americans).(see comments in appended blog below on perceived deficiencies of the RE-LY trial regarding intracranial bleeds)

so, what does this all mean? below are several prior blogs on dabigatran, some showing drug company malfeasance in promoting it, both withholding data and their own sense that levels should be monitored (the big push for this drug was that you didn't need to do INRs, so it was easy for patients and providers). also an article on increase in MI and ACS. and one on postmarketing surveillance and increased bleeding with dabigatran.  the current article on the Medicare population confirms (at least to me) the importance of post-marketing surveillance. all-too-often, there are serious adverse effects of meds (eg COX-2 inhibitors such as vioxx....) that are not found on initial studies, either legitimately through the gaps of statistical analysis or drug company malfeasance (or both). some may be due to study design or to the cloistered setting of the academic medical center and selection biases. in any event, post-marketing analysis provides larger numbers of real-world patients on the med. as noted below, one of the big issues with direct thrombin inhibitor bleeding, unlike warfarin, is that there is no antidote/ability to reverse the effect.