Author + information
The interesting publication by Dangas et al. (1)claims to compare arteriotomy closure devices with manual compression after percutaneous coronary intervention. Unfortunately, this is a comparison only in the fashion that can be ascribed to a retrospective trial with mismatched procedural variables and operator experience, ad hoc recruitment, and broadly applied statistical techniques. The conclusion that closure devices are associated with a higher rate of vascular complications could not possibly be ascertained from this study. In fairness to the investigators, the segment of the interventional cardiology literature to which they have contributed is largely a collection of testimonials and historical comparisons (2). Unfortunately, there is a striking paucity of properly conducted controlled clinical trials in this arena.
Our problems with this study (1)are multiple, some of which were outlined in the accompanying editorial by Tavris et al. (3). First, the investigators claim that “no large report exists on the ‘real world’ application” of these devices. Multiple studies have reported far larger experiences, at least as “real world” as the investigators’ (4–6).
Second, Dangas et al., highly experienced and with an international reputation in interventional cardiology, are largely novices at vascular closure, having apparently declined to adopt them in routine use. They include experience with only 6 Vasoseals, 32 Duetts, and 6 Prostars; the majority of cases used Angioseals and Techstars. The 516 deployments constitute only 8% of their cases, and amounts to fewer than 150 cases per year, spread over multiple operators. In contrast, the learning curve for one Perclose device as published by Balzer et al. (6)appears to level off at approximately 350 cases.
Third, the individual operators may not have used the recommended techniques for device deployment. For example, even though femoral artery angiography “was recommended before arteriotomy closure device application,” it was used only “in the majority of cases” (we are not told the actual number). This is clearly a disadvantage and probably inappropriate handling of the devices, as numerous factors that predict outcome and might demonstrate contraindications to device deployment were unknown to the operators in what is likely a high percentage of the cases. Also, because at least 13% and probably more of these femoral sticks were not in the common femoral artery (regardless of the operators being “very experienced with arterial puncture”), any number of the pseudoaneurysms and retroperiotoneal bleeds could have been due to misplacement of the femoral puncture (7). In the manual compression group, such sins were much more easily masked when sheath pulling took place at an activated clotting time (ACT) <150 s.
Fourth, prior published data have demonstrated that level of anticoagulation, sheath size, physician learning curves, location of puncture site, vessel size and presence of local atherosclerotic disease all influence outcomes of vascular closure. The investigators have demonstrated a mismatch in anticoagulation (ACT 277 vs. <150 s) and learning curves (experienced manual compression technicians versus inexperienced closure device users). They do not have the data regarding location of puncture site, atherosclerotic disease or vessel size, and have failed to inform us regarding sheath sizes other than that ≥10F sheath pulls were excluded. The latter also raises the issue of possibly inappropriate use of closure devices for larger than approved sheath sizes.
Fifth, Dangas and colleagues draw conclusions comparing two sealing methods in a retrospective fashion using statistical methods designed with the assumption that samples were selected at random (8). Thus, the p values are misleading, adding to the problems of this observational study with uncontrolled (retrospective) data acquisition, and ad hoc group assignment. These groups are likely to have varied in ways the investigators did not notice or chose to ignore, and these differences, rather than the treatment modalities, may account for the potentially different outcomes. Further, such studies can be subject to bias in completeness and quality of information recorded in the hospital charts; the investigators reading such charts often must use considerable judgment in assessing the data present. Sizing hematomas, for example, can be difficult even in prospective studies and can be quite subjective even with the most sophisticated measuring tools.
The study by Dangas et al. demonstrates that relatively novice users of vascular closure devices, probably frequently not following manufacturers’ recommendations, and deploying these devices in fully anticoagulated patients, had a higher complication rate than full-time employees trained to do manual compression who were pulling sheaths when the ACT was <150 s. Perhaps the only truly useful conclusion one could draw from the study (1)is that operators should be well into their learning curve, should perform an angiogram on the femoral artery before device deployment and otherwise follow the recommended protocols before deployment in fully anticoagulated patients. Although we use these devices after percutaneous intervention in nearly 100% of our cases, we believe that any recommendation should await the results of prospective randomized studies applying uniform definitions of complications and uniform measurement of end points.
- American College of Cardiology Foundation
- Dangas G,
- Mehran R,
- Kokolis S,
- et al.
- Tavris D,
- Gross T,
- Gallauresi B,
- Kessler L
- Glantz S.A