To the Editor:
We read with interest the recently-published article by Whitehouse et al. reporting on the use of apixaban as an alternative to warfarin for anticoagulation in patients with HeartMate 3 (HM3) ventricular assist device.1 The authors noted that the adverse events of bleeding, stroke, and death were similar in HM3 patients receiving warfarin or apixaban. The authors calculated p values for the outcomes they report, even though their subjects were not randomized. Even if the patients had been randomized, the number in this study was too small to detect anything but a dramatic difference in the outcomes of difference. The calculation of 95% confidence intervals would be more appropriate and give readers a sense of the range of uncertainty the results reported.
This was a timely study given that, according to the authors, it is unclear what the risks of thromboembolism and life-threatening bleeding associated with direct oral anticoagulants are in patients with HM3 LVADs. The most surprising aspect of this study was not the results but rather the choices that the authors made in the context of research ethics. The authors note that institutional review board (IRB) approval was obtained and “the informed consent and HIPAA requirements were waived.” Apparently, patients were offered either the conventional anticoagulation with warfarin or apixaban. The authors state, “The risks, benefits, and alternatives to therapy of apixaban use were explained, and patients individually made the choice to switch to apixaban after a thorough discussion with the treating physicians.” We find it hard to understand how this study could be considered exempt from research informed consent by patients.
Without question, IRB oversight increases the effort required and the complexity of any study. We wonder how the authors could have offered something that was not standard of care without clearly explaining that patients were entering a clinical trial or receiving what could be described as “experimental” therapy. The fact that the analysis was performed retrospectively does not, in our opinion, preclude the critical oversight of the IRB for a study that clearly involved unknown risks to patients. It is hard to imagine that the authors did not intend to write up their results when they began this practice. We have no doubt that the authors did fully explain the known risks to the patients, but in the context of unknown risks (the very nature of clinical investigation), the patients deserved the oversight of the informed consent process and patient safety that an IRB could have provided. Did the authors explain to patients that the only trial of a similar approach was stopped because of complications from the DOAC (indirectly referenced in ref 25)?
Modern IRB oversight, flawed though it can sometimes be, is intended to inform patients of the risks they might be undertaking when they participate in clinical trials, and to protect them from being subjected to substantial risks without their consent. Although clinical investigators might be tempted to undertake before-after quality improvement programs as a way to circumvent IRB oversight, this is widely recognized to be problematic. This study is closer to a clinical trial than a quality improvement project. We worry that publishing such a study in a high-quality journal will only serve to undermine the role of IRBs in the protection of research subjects. When patients are also research subjects, they are in a particularly vulnerable position for which additional protections by an IRB are essential.
1. Whitehouse KR, Avula D, Kahlon T, et al.: Apixaban: Alternative anticoagulation for heartmate 3 ventricular assist device. AISAIO J, doi: 10.1097/MAT.000000000000001650.