Author + information
- Félix Valencia-Serrano, MD, PhD∗ (, )
- Francisco Moreno, MD,
- María Valencia-Serrano, MD and
- Miguel A. Martínez-González, MD
- ↵∗Cardiology Department, Torrecárdenas University Hospital, Paraje de Torrecárdenas sn, 04009 Almería, Spain
We read with interest the article by Wimmer et al. (1). They propose the use of an unexpected finding in an observational study (that is unlikely to be causally related to the intervention being studied) to highlight the presence of residual confounding. Their strategy assumes that residual confounding accounts for these unexpected, noncausal findings, even despite the application of advanced and complex statistical methods to control for confounding. However, the discovery of unexpected results in observational studies may have other explanations and, consequently, potentially useful applications, such as: the search for possible alternative causal explanations, the generation of new hypotheses, and the design of further studies to evaluate these unexpected findings. These applications may be useful to improve our understanding of disease pathophysiology, expand the frontiers of knowledge, and ultimately improve patient care.
From this point of view, in the present case (1), there could be an alternative explanation for the increased bleeding rate not apparently related to vascular access associated with femoral percutaneous intervention (PCI); for example, the hemorrhagic transformation of clinically unapparent embolic infarctions in thoracoabdominal solid organs, viscera and soft tissues subsidiary of the aortic arch, and the thoracoabdominal descending aorta. The uneven distribution of these events according to the intervention under study (femoral versus radial vascular access PCI) could be justified by the higher probability of catheter contact with the endothelial aortic surface when using the femoral approach (a greater length of aorta is exposed to contact with the catheter from femoral access than from radial access). Furthermore, the arch and descending aorta compared with the ascending aorta are the most common locations for complex atherosclerotic plaques responsible for embolic phenomena (2). Moreover, antiplatelet and anticoagulant regimens may increase the risk of hemorrhagic transformation of unapparent embolic events and favor their clinical presentation as hemorrhagic events. These aspects may justify the design of a specific trial using imaging pre/post studies to assess the incidence of thoracoabdominal embolic events during femoral versus radial vascular access in PCI.
Finally, some methodological aspects should be pointed out. First, bleeding outcomes in the study by Wimmer et al. (1) cannot be considered rare events. Falsification hypothesis has been proposed to validate the identification of rare adverse effects from population data, defined as those occurring in <1 per 1,000 subjects (3). Second, the authors propose a single hypothesis to falsify rather than several as recommended. Falsification hypothesis should be operationalized by testing multiple implausible pre-specified hypotheses with the same methods applied in the primary analysis. When many false relationships are present, caution is warranted in the interpretation of study findings. Finally, the mere absence of evidence together with a low biological plausibility according to the state of mechanistic knowledge at that time (or even less, the opinion of experts) do not seem enough before proposing nonaccess bleeding as a falsification endpoint. There is a need to positively demonstrate from properly designed studies that such an observed effect is unlikely to be true. Otherwise, the generation of knowledge outside the box and scientific progress may be halted.
- 2014 American College of Cardiology Foundation
- Wimmer N.J.,
- Resnic F.S.,
- Mauri L.,
- Matheny M.E.,
- Yeh R.W.
- Khatibzadeh M.,
- Mitusch R.,
- Stierle U.,
- Gromoll B.,
- Sheikhzadeh A.